Harmonising Rock Engineering and The Environment

HARMONISING ROCK ENGINEERING AND THE ENVIRONMENT
© 2011 by Taylor & Francis Group, LLC

PROCEEDINGS OF THE 12TH ISRM INTERNATIONAL CONGRESS ON ROCK MECHANICS,
BEIJING, CHINA, 18–21 OCTOBER 2011
Harmonising Rock Engineering and

the Environment
Qihu Qian & Yingxin Zhou

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20120201
International Standard Book Number-13: 978-0-203-13525-9 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to publish reliable data
and information, but the author and publisher cannot assume responsibility for the validity of all materials or the consequences of their use. The
authors and publishers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to copyright hold-
ers if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let us know
so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any elec-
tronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or con-
tact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that
provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system
of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation
without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com

CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business
© 2012 Taylor & Francis Group, London, UK
Typeset by MPS Limited, a Macmillan Company, Chennai, India
Printed and bound in Great Britain by Antony Rowe (A CPI Group Company),
Chippenham, Wiltshire
All rights reserved. No part of this publication or the information contained herein may be
reproduced, stored in a retrieval system, or transmitted in any form or by any means,
electronic, mechanical, by photocopying, recording or otherwise, without written prior
permission from the publisher.
Although all care is taken to ensure integrity and the quality of this publication and the
information herein, no responsibility is assumed by the publishers nor the author for any
damage to the property or persons as a result of operation or use of this publication
and/or the information contained herein.
Published by: CRC Press/Balkema

P.O. Box 447, 2300 AK Leiden, The Netherlands
e-mail: Pub.NL@taylorandfrancis.com
www.crcpress.com – www.taylorandfrancis.co.uk – www.balkema.nl
ISBN: 978-0-415-80444-8 (Hbk+CD-ROM)

ISBN: 978-0-203-13525-9 (ebook)

Harmonising Rock Engineering and the Environment – Qian & Zhou (eds)
© 2012 Taylor & Francis Group, London, ISBN 978-0-415-80444-8
Table of Contents
Preface XXIII
Organisation XXV
Müller lecture
From empiricism, through theory, to problem solving in rock engineering 3

N.R. Barton
Rocha Medal lecture
Reduction of blast-induced vibration in tunnelling using barrier holes and air-deck 17

D. Park
Invited lectures on ISRM 50th anniversary
Fifty years of the ISRM and associated progress in rock mechanics 29

E.T. Brown
The next 50 years of the ISRM and anticipated future progress in rock mechanics 47
J.A. Hudson
New developments of rock engineering and technology in China 57
Q.H. Qian
Keynote lectures
New rock mechanics developments in China 71

X.-T. Feng
Rock strength and failure: Some common and uncommon issues 81
A.J. Gonzalez-Garcia
Discontinuum models for dam foundation failure analysis 91
J.V. Lemos
Discontinuous deformation analysis: Advances and challenges 99
G.W. Ma
Synthetic rock mass applications in mass mining 109
M.E. Pierce & C. Fairhurst
How to select rock mass parameters for analysis and design 115
T. Ramamurthy & A. Ghazvinian
Risk oriented design and construction of tunnels 127
W. Schubert
Support of excavations subjected to dynamic (rockburst) loading 137
T.R. Stacey
Advances in rock dynamics modelling, testing and engineering 147
J. Zhao, Y.X. Zhou & K.W. Xia

Analysis and design methods
Mine-by experiment performed in the Callovo-Oxfordian claystone at the Meuse Haute Marne underground
research laboratory (France) 157
G. Armand, A. Noiret, M. Cruchaudet & N. Conil
Burgers rock creep around axisymmetric tunnels 159
S.D. Athanasopoulos, P.P. Nomikos & A.I. Sofianos
Multi-parameter responses of soft rocks during deformation and fracturing and their implications in
geomechanics and geoengineering 161
Ö. Aydan, Y. Ohta & H. Tano
Fractal approach to determine rock mass strength and deformation 162
M.N. Bagde, A.K. Raina, P.B. Choudhury & A. Sinha
A theoretical analysis of core extrusion and squeezing behaviour in tunnelling 163
L. Cantieni, G. Anagnostou & R. Hug
Applications of strain based damage criterion in geotechnical engineering 165
Y. Chang
Solving some problems of geomechanics on the base of defining relations of post-limit deformation of rocks 167
A.I. Chanyshev, I.M. Abdulin & O.E. Belousova
The structural geology contribution to rock mechanics modelling and rock engineering design 169
J.W. Cosgrove & J.A. Hudson
Rock mass instability caused by incipient block rotation 171
A.V. Dyskin & E. Pasternak
Attempt to advanced observational construction considering predictive analysis of long-term deformation 173
T. Fukuda, T. Takahashi, H. Yamada & Y. Jiang
Vector sum method: A new method for anti-sliding stability analysis 174
X.R. Ge
Design of a railway tunnel parallel to an existing tunnel 176
R.K. Goel, A. Swarup, R.K. Dayal & V. Bhardwaj
Shear failure mechanism in oil wells due to reservoir compaction “A case study in Gachsaran formation, Iran” 177
M. Gorjian, M. Moosavi, H. Memarian & H. Ajamzadeh
Fractal analysis on the fracture development of sandstone using AE measurement 179
Z.Y. Yang, T.H. Huang, Y.X. Zhao & M.C. Tsai
Static and dynamic response analyses of rock mass considering joint distribution and its applicability 181
N. Iwata, T. Sasaki, & R. Yoshinaka
Numerical study on zonal disintegration of rock mass around deep underground openings 182
P. Jia, C.A. Tang & Y.B. Zhang
The secondary lining design of subsea tunnel due to water inflow 184
K.-S. Joo, Y.-H. Suh, K.-H. Park, D.-H. Huh, E.-J. Hong, S.-J. Moon & S.-C. Lee
An assessment of rock pillar considering excavation damaged rock zone 186
W.-B. Kim, T. Ha & H. Yang
Response of granite under strain controlled loading and effect of support system on behaviour of
large underground cavern 187
Rakesh Kumar, K.G. Sharma & A. Varadarajan
Determination of backfill requirements in multi-reef mining using numerical modelling 188
K.B. Le Bron & R.A. Johnson
Comparisons between two empirical yield criteria for rock masses 190
A.J. Li & R.S. Merifield
Buckling failure of highly-stressed hard rocks surrounding deep underground openings and confining effects of
backfilling material 191
D.Y. Li, X.B. Li, Z.X. Liu, C.D. Ma & L.N.Y. Wong
Bemposta II powerhouse shaft: Geotechnical characterization, design and construction 193
C. Lima, E. Resende, C. Esteves & J. Neves
VI

Initiation and evolution of failure zones in the vicinity of heterogeneities in rock masses:
Seismology and deep oil well stability problems 194
L.A. Nazarova, L.A. Nazarov, M.I. Epov & I.N. Eltsov
Monitoring of cracks opening in the walls of deep pits 196
V. Oparin & V. Vostrikov
Mine plug design for the San Genaro Ramp 198
A.A. Samaniego & R.L. Cabrera
A case study of alteration effects in granitic rocks on rock cavern stability at Jinzhou of NE China 200
Y.J. Shang & Y.Y. Shi
Design of tunnels in jointed rock on the basis of mechanical models and classification systems 202
R. Sommer & W. Wittke
A study on mechanical behaviors of concrete lining and rock caused by shaft sinking at the
Horonobe underground research laboratory 204
K. Tsusaka, D. Inagaki, M. Koike, Y. Ijiri & Y. Hatsuyama
Use of rock mass classifications for design: Recommendations and suggestions 205
C. Vibert & P. Vaskou
Analysis to the design factors of the diversion tunnel plug based on additional rock elastic resistance 207
K. Wang, M.J. Zhao & X. Sun
The influence of bedding angle on failure mechanisms in sandstone 209
P.L.P. Wasantha, P.G. Ranjith, D.R. Viete, A. Haque, J. Kodikara & A. Bouazza
Estimating the porosity-depth relation of sedimentary rocks from an effective stress/stress history-dependent
porosity model 211
W.-J. Wu, J.-J. Dong, J.-H. Wu & S.-S. Lin
Experimental study on constitutive models of rocks under loading with active confining pressure 213
J.Y. Xu, J.Z. Liu, X.C.C. Lv, Z.D. Wang & E.L. Bai
Global stability analytical method of cavern group in salt rock based on energy criterion 214
Q. Yang, J.Q. Deng, Q.C. Lu & Y.R. Liu
Analytical study on underground continuous impervious curtain for shaft lining treatment at eastern
Chinese coal mines 216
Q. Yu, H. Shimada, T. Sasaoka & K. Matsui
Back-analysis of schistose rock mass strength parameters based on displacement monitoring data 217
X.P. Zhang, L.N.Y. Wong & S.J. Wang
Different analytical models and methods to study wave propagation across rock joints 219
J.B. Zhu, W. Wu & J. Zhao
Numerical modelling
Coupled flow-deformation simulation for mine scale analysis of cave initiation and propagation 223
D.A. Beck & G. Putzar
A numerical analysis of a four-way coal mine intersection with primary and secondary supports 224
Y.P. Chugh & B. Abbasi
3-D modeling and parameters research of irregular shape particles for DEM based on computerized tomography 225
X. Du, Y.W. Zeng, R. Gao & J. Yan
Three-dimensional thermomechanical modelling of backfilling of old mining rooms in the Morsleben repository 227
S. Fahland, R. Eickemeier & St. Heusermann
Development of a non-contact survey method of tunnel excavation face for DEM modelling 229
A.M. Ferrero, G.F. Forlani, R. Migliazza & R. Roncella
Numerical simulation of crack propagation in rock by clumped particle model 231
T. Funatsu & N. Shimizu
Stability modelling of isolated rock blocks at the surface of underground excavations taking into
account in-situ stresses 232
R. Ghazal, F. Hadj-Hassen & M. Tijani
VII

Back analysis of tunnel response using the Differential Evolution Genetic Algorithm (DEGA) 233
M. Gutierrez, S. Vardakos & C. Xia
The assessment of the hydro-jacking risk of the pressurized water tunnel by theoretical and numerical method 235
T.W. Ha, Y.W. Shin, H.G. Lee & H.S. Yang
Dynamic analysis of rock tunnnel failure by using 3-D Numerical Manifold Method 237
L. He, X. Huang & G.W. Ma
Thermomechanical modelling of the Gorleben exploration site to analyse the integrity of the salt barrier 239
St. Heusermann, P. Vogel, R. Eickemeier & H.-K. Nipp
Response of nuclear waste repository host rock to thermo-mechanical loads 241
H. Hökmark & M. Lönnqvist
Mixed complementary formulation of discontinuous deformation analysis and solution 243
W. Jiang & H. Zheng
X-ray CT technology-based finite element simulation of coal samples under uniaxial compression 245
Y.D. Jiang, W.J. Wang, Y.X. Zhao, L. Zhang & J.L. Han
Pumping simulations using 3D analysis on multi-pumping wells 247
T. Kitaoka, H. Kusumi & M. Nakamura
Abutment stability assessment of the Hongrin arch dam using 3D distinct element method 248
A. Koliji, T. Bussard, A. Wohnlich & J. Zhao
Development of numerical model for grout injection and its application to the real in-situ injection tests 249
T. Koyama, Y. Ohnishi, K. Takahashi, A. Kobayashi & T. Katayama
Modelling the evolution of fractures using a combined FEM-DEM numerical method 250
J.-P. Latham, L. Guo, X. Wang & J. Xiang
A constitutive contact law for discrete element modeling of sandstone 252
L. Li
Numerical modeling of rock cutting 254
J.-S. Lin, J.A. Mendoza, M.C. Jaime, Y. Zhou, J. Brown, I.K. Gamwo & W. Zhang
Strength symmetry with respect to the stress state imposed by Mogi’s theory against long-prevailing observations 255
X.D. Ma & R.J. Rodriguez
Scale effects in a synthetic rock mass 257
C.D. Martin, Y. Lu & H. Lan
2D-/3D-stress analysis of deep-seated gold mines in South Africa 259
F. Menschik, K. Thuro, H. Käsling & M. Bayerl
Numerical modeling for the evaluation of grout penetration in fractured rock masses 260
Y. Mito, M. Ashraf & T. Yamamoto
Numerical method for determining contact areas of a rock joint under shear load 261
J.-W. Park, Y.-K. Lee & J.-J. Song
Approximate methods in numerical rock mechanics modelling 263
J.J. Peura, J.P. Pöllä & I. Konstantas
Hydromechanical modeling of an underground excavation with an elastoviscoplastic behavior law and regularization 265
R. Plassart, R. Fernandes, F. Laigle, A. Giraud & D. Hoxha
Implementation of ANN-based rock failure criteria in numerical simulations 266
H. Rafiai & A. Jafari
Stochastic model generation for discontinuous rock mass media and numerical analysis using the
NSCD method (case study: Saint-Béat gallery in the southwest of France) 268
A. Rafiee, M. Gasc-Barbier & M. Vinches
Rock fracture propagation from mechanical, thermal and hydraulic loadings 269
M. Rinne, B. Shen & O. Stephansson
Meshless simulation of rock mechanics problems by Element Free Galerkin Method 271
H. Salari-Rad & M. Rahimi Dizadji
Applications of two dimensional discontinuous deformation analysis 273
G.H. Shi
VIII

Stochastic boundary element method for fracture reliability analysis 275
K.J. Shou & H.Y. Yu
Numerical anisotropic fracture mechanics modelling in crystalline rock 276
T. Siren, K. Kemppainen, B. Shen & M. Rinne
Contact description in numerical simulation for rock mechanics 278
L. Sun, J. Zhao & G.F. Zhao
Numerical investigation of the potential applicability of surface deformation measurement by InSAR
technology for monitoring CO2 behavior in deep underground 280
H. Yamamoto, T. Onuma, T. Kumagai & J. Rutqvist
An introduction to Distinct Lattice Spring Model (DLSM) 281
G.F. Zhao, N. Khalili & J. Zhao
A discussion about boundary conditions in numerical simulation and its application in mining-induced
rock mass movement 283
H.J. Zhao, F.S. Ma, Y.M. Zhang, J. Guo & A.H. Wei
Theoretical and experimental modeling of material hardening and softening by compression tests 285
V.M. Zhigalkin, V.N. Semyonov, O.M. Usoltseva, P.A. Tsoi, A.I. Chanyshev & I.L. Abdulin
Multiple scale numerical modeling of growth process of multiple cracks in rock mass 287
X.P. Zhou, H.Q. Yang & Y.X. Zhang
Laboratory testing and rock properties
Mechanical behavior of Callovo-Oxfordian claystone-steel interfaces at high levels of normal stress in

“dry” and “wet” conditions 291
M. Boulon, M. Keshavarz, G. Armand, N. Conil & F.L. Pellet
Microcrack growth in granite during fatigue process 292
Y. Chen, A. Yamazaki, H. Kusuda, E. Kusaka & M. Mabuchi
An experimental study on deformation and strength anisotropy of transversely isotropic rocks in Korea 293
J.-W. Cho, H. Kim, S. Jeon & K.-B. Min
Mechanical rock properties prediction: Deriving rock strength and compressibility from petrophysical properties 295
B. Crawford, B. Alramahi, P. Gaillot, P. Sanz & N. DeDontney
Behaviour of carbonated rocks under true triaxial compression 296
F. Descamps, J.-P. Tshibangu, M. Ramos da Silva, C. Schroeder & J.-C. Verbrugge
Mechanical rehearsals in special slate schists to potentiate constructive solutions of structures with great slenderness 297
L.M. Ferreira Gomes, C.R.V. Batalha, P. Gabriel de Almeida, L.J. Andrade Pais & E. Mendes
An appraisal and modification of Ramamurthy criterion on anisotropic rocks 299
A. Ghazvinian, M.R. Hadei, H.R. Nejati & V. Sarfarazi
Effects of fracture size and normal stress on the shear behavior of a fracture as estimated by mortar replica 301
A.A. Giwelli, K. Matsuki, K. Sakaguchi, A. Kizaki, H. Sekino & K. Okatsu
Study of the short and long-term brittle behavior of a porous rock with continuous monitoring of
(P- and S-) wave velocities under different hydrous conditions 303
D. Grgic, D. Hoxha & J. Eslami
The effect of certain chemical solutions on some material properties of clay bearing rocks 305
P. Hagan, S. Saydam, E. Elias, L. Deramore Denver & B. Hebblewhite
Creep test of tuff for over thirteen years and its know-how 306
K. Hashiba, S. Okubo & K. Fukui
Macro-fracture characteristics of granite in different boundary conditions 307
M.C. He, W. Nie, Z.Y. Zhao & C. Cheng
Experimental analysis of deformation mechanisms induced by desiccation in Tournemire argillites using
DIC method 309
S. Hedan, P. Cosenza, V. Valle, A.L. Fauchille, P. Dudoignon & J. Cabrera
Radiology of Canarian volcanic rocks 310
L.E. Hernández, A. Eff-Darwich, R. Viñas & J.A. Rodríguez-Losada
IX

A new laboratory test for shear fracture formation and its permeability measurement 312
T. Ito, T. Shono, K. Sekine & K. Yamamoto
Direct shear and tensile test on cemented healed joints from El Teniente mine, Chile 313
L. Jacobsson, M. Flansbjer, A. Brzovic & C. de los Santos
Rheological behavior of sandstone under unloading 315
Y.Z. Jiang & R.H. Wang
Strength evaluation of discontinuous rock mass by multi-scale limit load analysis 317
J. Jung, J. Kato, K. Terada & T. Kyoya
Estimation of fracture toughness of anisotropic rocks by SCB test and visualization of fracture by
means of X-ray CT 318
M. Kataoka, Y. Obara & M. Kuruppu
Laboratory measurement of poroelastic and hydraulic constants of sedimentary rocks focused on
pore water discharge from specimen with its deformation 319
M. Kato, H. Park, M. Takahashi & K. Kaneko
Relationship between point load strength and uniaxial compressive strength of hydrothermally
altered soft rocks 320
M. Kohno & H. Maeda
Study on the dilatancy angle of rocks in the pre-failure domain 321
M. Kwaśniewski & P. Rodríguez-Oitabén
The thermal and mechanical properties around an underground research tunnel in granite body 323
S. Kwon, J.S. Kim, C.S. Lee & W.J. Cho
Shear behavior model of rock joints applicable to CNL and CNS conditions 325
Y.-K. Lee, J.-W. Park & J.-J. Song
Evaluation of shear velocity dependency of rock fractures by using repeated shear tests 327
B. Li, Y. Jiang & G. Wang
An elastic-plastic-brittle constitutive model of rocks and its numerical validation 328
Q. Li, D.G. Wang, J.H. Xiong & L. Song
Drilling geological strength index in altered gneiss 330
D.M. Lin, Y.J. Shang & W. Zhang
Investigation on mechanical parameters of water-rock interaction 332
Y. Lu & T. Lu
Surface roughness measurements using three-dimensional laser imaging 334
J. Mah, C. Samson & S.D. McKinnon
Understanding failure mode in uniaxial and triaxial compression for a hard brittle rock 336
V.B. Maji
The mechanical properties of CO2 -saturated coal specimens 338
M. Masoudian-Saadabad, D.W. Airey, A. Gainey, T. Morris & J. Berger
Analysis of uniaxial tensile fracture of monomineral polycrystalline rock based on intergranular cracking 339
K. Matsuki, Y. Karino, K. Sakaguchi & A. Kizaki
Determination of a correlation between intact rock unconfined compressive strength and index parameters 341
H.J. Mendieta
Experimental research on the electrical parameters of coal mass under load 343
L. Meng, M.J. Liu, Y.D. Jiang, Y.X. Zhao & Y.G. Wang
Creep property of artificial methane-hydrate-bearing rock 345
K. Miyazaki, Y. Sakamoto, K. Aoki, N. Tenma, T. Yamaguchi & S. Okubo
An investigation of brittle rocks failure mechanism based on the damage mechanics 346
H. Molladavoodi & A. Mortazavi
Influence of humidity and rock fabric on subcritical crack growth in rock 348
Y. Nara, K. Kaneko, T. Yoneda, N. Hiroyoshi & K. Morimoto
Estimating the basic friction angle of a very strong meta-sandstone for rockfill strength prediction 349
D.J.M. Ngan-Tillard & A. Mulder

Application of pattern recognition analysis of rock acoustic emission for determination of Kaiser Effect 350
M. Nikkhah, M. Ahmadi & A. Ghazvinian
Acoustic emission of Dionysos marble specimens in uniaxial compression 352
P.P. Nomikos, K.M. Sakkas & A.I. Sofianos
An elaborative evaluation of various methods for simultaneous estimation of uniaxial
compressive strength and elasticity modulus of intact rocks 354
R. Noorani & M. Hesami
Drying-induced deformation and consequent change in mechanical properties of sedimentary rocks 356
M. Osada, N.M. Adikaram & T.N. Illankoon
Assessment of clay minerals around fault using slake durability test and reflectance spectroscopy 357
C.-S. Park, C.-U. Hyun & H.-D. Park
Relation between mechanical properties of rock and sedimentary structure 359
H. Park & M. Takahashi
Estimation of uniaxial compressive strength of weak rocks using needle penetrometer 360
Y.-H. Park, Y. Obara, & S.-S. Kang
Testing of fiber reinforced sprayed concrete – influences on the energy absorption capacity 362
G. Pittino & Y. Bian
A comparative study of mi , the Hoek-Brown constant for intact rock material 364
S.A.L. Read & L. Richards
Inconsistencies in the determination of the compressive strength of intact rock 366
M. Romana
Effect of heat on the mechanical properties of selected rock types – a laboratory study 368
C. Saiang & K. Miskovsky
Weathering mechanisms and mechanical property changes of Oya tuff 369
T. Seiki, A. Satoh & K. Kikuchi
Relationship between the rupture modulus and the flexural strength of building and decorative stones of Iran 370
K. Shahriar, M.S. Dehkordi & H.A. Lazemi
Shear behavior of rock under different normal stiffness 371
A.K. Shrivastava, K.S. Rao & G.W. Rathod
On the reliability of the uniaxial compressive strength obtained from the point load index 372
L.O. Suarez-Burgoa
Preliminary laboratory thermo-hydro-mechanical characterization of Opalinus clay 373
A.M. Tang & Y.J. Cui
Experimental evidences of dramatic embrittlement of hard rocks due to rising confining pressure
at triaxial compression 374
B.G. Tarasov
Method of determination of orthotropy ratio and principal material direction for orthotropic
rock material with diametrical compression test 375
T. Tsutsumi
Influence of confinement dependent failure processes on rock mass strength at depth 376
B. Valley, B.-H. Kim, F.T. Suorineni, N. Bahrani, R.P. Bewick & P.K. Kaiser
Nondestructive evaluation of granite building stones and relation with key mechanical parameters 378
G. Vasconcelos, C.A.S. Alves & T. Miranda
Relation between tensile and compressive engineering properties of granites 380
G. Vasconcelos, P.B. Lourenço, C.A.S. Alves, J. Pamplona & T. Miranda
Modeling the deformation behavior of sandstone using a bonded-particle model 382
M.-C. Weng, H.-H. Lee, S.-S. Zhang & F.-S. Jeng
Initiation and propagation of tensile wing cracks and anti-wing cracks from a pre-existing open
flaw under compression 384
L.N.Y. Wong & H.Q. Li
Physico-mechanical and thermo-mechanical characterisation of sandstone 386
G. Wu, Y. Wang, G. Swift & J. Chen
XI

Development of experimental technique on seismic response of rock joints 388
W. Wu, J.B. Zhu, Q.B. Zhang, J.F. Mathier & J.P. Dudt
Mechanical properties of freeze-thaw and damage propagation law for rock 389
G.S. Yang & H.M. Zhang
Experimental study on crack coalescence mechanism of granite specimens containing two surface cracks 391
P. Yin, R.H.C. Wong & K.T. Chau
Simulation of laboratory tested bolted rock beams 392
P. Yiouta-Mitra, N. Halakatevakis & A.I. Sofianos
Empirical failure criterion for biaxially loaded granite 394
X. Yun, H.S. Mitri & X. Yang
Experimental and numerical research on crack propagation in rock under uniaxial compression 396
K. Zhang, Q. Yang, J.C. Jiang & S.B. Tian
Investigation on the failure features of coal under uniaxial loading condition by Micro
X-ray Computed Tomography 398
Y.X. Zhao, Y.D. Jiang, W.F. Zhao & T. Wang
Field measurements and site investigations
A new method for visualizing stresses in rock support measures by using a light emitting sensor 403
S. Akutagawa, Y. Minami, H. Yamachi, T. Haba, T. Hiroshima & T. Nozawa
Microseismic monitoring strategy as a key component of post-mining risk management:
Review and feedback experience over the past decade 404
P. Bigarré, M. Bennani, I. Contrucci, E. Klein, H. Baroudi, R. Hadadou, I. Vuidart & J.P. Josien
Factors dominating deterioration of clastic sedimentary rocks 406
P.-C. Chan, S.-J. Lyu, T.-T. Wang, F.-S. Jeng & T.-H. Ueng
Image analysis system for areal measurement to preserve historic masonry wall, Japan 408
H. Chikahisa, R. Ohta, T. Kawahara, N. Kitayama & Y. Okamoto
Property evaluation of beachrock based on field investigation in Okinawa Island, Japan 409
T. Danjo & S. Kawasaki
An engineering view of the seismic prediction during tunneling 410
T. Dickmann & F. Pittard
Application of acoustic emission measurement on assessment of loosened zone around underground cavern 411
L. Fan, H.M. Zhou, Y.H. Zhang & Z.W. Zhong
Investigation of 3D terrestrial laser scanning techniques for potential application to rock mechanics 412
Q. Feng, G. Wang & K. Röshoff
Interpretation of rock mass deformability measurements using large flat jack tests 413
B. Figueiredo, F. Bernardo, L. Lamas & J. Muralha
Abandoned roadways aged up to 50 years observed in Kushiro Coal Mine, Japan 415
Y. Fujii, T. Sugawara, J. Kodama, Y. Ishijima, T. Kiyama, M. Takada, Y. Ichihara, S. Kumakura,
T. Narita, M. Sawada & E. Nakata
The role of instrumentation and monitoring system on the stability of a slope 416
M. Gharouni Nik
Determination of mechanical properties of a large deformation zone 417
R. Glamheden, I. Olofsson & A. Fredriksson
Spalling in a field experiment of heated boreholes in hard rock 419
R. Glamheden, B. Fälth & R. Christiansson
About the determination of the joint set intensity 421
N.F. Grossmann
Determination of in-situ stress orientation by thermally induced spalling 423
E. Hakami & R. Christiansson
Site investigation in volcanic rock mass for building projects 425
L.E. Hernández, M.J. Pomares, J.A. Rodríguez-Losada, A. Eff-Darwich & C. Olalla
XII

Prediction of rock mass along tunnels by geostatistical method 427
K. Kaewkongkaew, N. Phien-wej & D. Kham-ai
Developing in situ stresses estimation algorithm using a hybrid numerical method based
on local stress measurement-case study: Seymare dam 428
K. Shahriar, M. Gharouni Nik & Z. Khademian
In situ stress determination by hydro jacking tests on fractured rock mass 430
M.A. Kanji
A case study on in situ testing of rock mass deformability 432
B. Kavur, N. Stambuk Cvitanovic & I. Vrkljan
In situ rock stress-strength comparison: Posiva’s Olkiluoto Spalling Experiment (POSE) 434
K. Kemppainen, M. Hakala, E. Johansson, H. Kuula & J.A. Hudson
Comparison of monitoring items for hazardous ground in subsea tunnels 437
H. Kim & S. Jeon
Determination of principal stress orientations from wireline logging caliper data and borehole images
in deep drilling 439
W. Lin
Experimental and numerical analysis of rock block stability using a remotely positioned laser
Doppler vibrometer 440
G.C. Ma, K. Sawada, A. Yashima, F. Uehan, O. Murata, H. Saito & Y. Ootsuka
Investigation of spalling failure at Nayriz Marble Mine (Iran) 441
M. Moosavi, H.M. Doghozlo, M. Goodarzi, N. Ghavami & F. Moosavi
Geotechnical and monitoring challenges during excavation of a large urban tunnel in soft ground
and low overburden 442
B. Moritz, R. Matt & R. Heissenberger
A study of the compound evaluation for geophysical explorations by self-organizing maps 444
M. Nakamura, H. Kusumi, T. Yamamoto & T. Tsuji
Measurement of exterior deformation of an earth-filled dam using GPS displacement monitoring system 445
S. Nakashima, H. Kawasaki, N. Shimizu, S. Kubota & T. Nakano
An innovative method to determine the indirect deformation modulus of rock masses in Asmari formation of Iran 446
A. Noorzad, K. Ahangari & A. Haghnejad
Measurement of rock stress change by Cross-sectional Borehole Deformation Method (CBDM) 447
Y. Obara, Y. Fukushima, T. Yoshinaga, T. Shin, M. Ujihara, S. Kimura & T. Yokoyama
Rock investigations for the construction of railway tunnel through Slovenian Karst 448
B. Petkovšek, J. Prestor & A. Geršak
In-situ test program for fault zone characterization 450
T. Pilgerstorfer & W. Schubert
Geological origin and stability analysis of the Suoduoxi accumulation body in the upper reaches of the Jinsha River 452
A.W. Ren, X.G. Wang, Z.X. Jia, Y.J. Wang & Q.W. Duan
In situ rock stress measurement using an improved Downward Compact Conical-ended Borehole Overcoring technique 453
K. Sakaguchi, A. Kizaki & K. Matsuki
Laser scanning images analysis for tunnel inspection 454
F. Sandrone & R. Wissler
Numerical simulations of hydraulic fracture intersecting an interbed of sandstone 455
M. Sarmadivaleh, V. Rasouli & W. Ramses
GPS displacement monitoring system for the precise measuring of rock movements 457
N. Shimizu, T. Masunari & T. Iwasaki
A study on prediction of squeezing and rockburst based on surface exploration for selection of
HLW disposal site in Japan 458
K. Shin, M. Sawada, Y. Inohara, T. Shidahara, T. Hatano, S. Arai & T. Miwa
Evaluation of stresses based on overburden and in-situ measurements 459
R. Singh
XIII

Stress parameter analysis in hydraulic fracturing with open fracture pattern 460
J.-H. Synn, Chan Park, Chulwhan Park & K.-S. Kim
Nanoseismic monitoring of shallow caverns for detection of incipient failures 462
M. Tsesarsky & G.H. Wust-Bloch
The stability of an underground congress center in soft tuffs through an integrated in-situ monitoring,
experimental, analytical and numerical methods (Cappadocia, Turkey) 463
R. Ulusay, Ö. Aydan, M. Genis˛ & H. Tano
Evaluation study of nozzle design selection for hydraulic fracturing test on concrete,
limestone and andesite with triaxial loading at the laboratory 465
R.K. Wattimena, S. Kramadibrata, N.P. Widodo & R.K. Uripto
Seismic imaging of stressed pillars in abandoned coal mines: Case studies from Australia 466
R.J. Whiteley
Dynamics and blasting
Damping analysis with depth and influence of geology on the transmission of vibrations generated by blasts 471
M.I. Alvarez-Fernandez, C. Gonzalez-Nicieza & A.E. Alvarez-Vigil
Comparison of blast-induced ground vibration predictors in Seyitomer coal mine 472
E. Arpaz, Ö. Uysal, Y. Tola, K. Görgülü & M. Çavuş
A new testing station for rock impact on full-scale structures 473
M. Bost, S. Loock, L. Lenti, F. Rocher-Lacoste, J. Gineys & L. Dubois
Mechanical behaviour of rock discontinuities under seismic conditions 474
L. Canelli, A.M. Ferrero, M. Migliazza & A. Segalini
Dynamic responses of tunnels subjected to Rayleigh wave 476
C.-H. Chen, T.-T. Wang & T.-H. Huang
Dynamic tensile-splitting tests of rocks 478
S.H. Cho, J.R. Ahn, M.S. Kang, Y. Nakamura, Y. Ogata & K. Kaneko
Characterization of dynamic rock fracture parameters using Notched Semi-circular Bend (NSCB)
method and Cracked Chevron Notched Brazilian Disc (CCNBD) method 480
F. Dai & K. Xia
Numerical simulation of train-induced vibrations in rock masses 482
A. Eitzenberger, P. Zhang & E. Nordlund
Effect of micro-defects and macro-joints on stress wave propagation in rock mass 484
L.F. Fan, F. Ren & G.W. Ma
Progress in study of rockburst experiments and control countermeasures 486
M.C. He, X.N. Jia, W.L. Gong, G.J. Liu & J.T. Zhang
Evaluation of dynamic behaviors of bedrock foundation based on FEM and DEM simulations 488
Y. Jiang, L. Yang, Y. Tanabashi, B. Li & S. Li
Application of discrete element method for validity evaluation of rock dynamic fracture
toughness measured by semi-circular bend technique 489
T. Kazerani & J. Zhao
The mechanism behind seismic damage to railway mountain tunnels and assessment of their aseismic performance 490
Y. Kojima, K. Yashiro, T. Funahashi, H. Miyabayashi, T. Iura & T. Asakura
Destress rock blasting as a rockburst control technique 491
P. Konicek, P. Konecny & J. Ptacek
Dynamic test of a high energy-absorbing rock bolt 492
C.C. Li
A 1D equivalent medium method for wave propagation across parallel jointed rock mass 494
J.C. Li, H.B. Li, G.W. Ma & J. Zhao
Effects of decoupled charge blasting on rock fragmentation efficiency 496
W. Liang, H.Y. Liu, X. Yang & D.J. Williams
Modelling earthquake-induced rock failures using NMM and DDA 497
Y.J. Ning, X.M. An, Z.Y. Zhao & G.W. Ma
XIV

Prediction of rockburst based on an accident database 499
A. Peixoto, L. Ribeiro e Sousa, R. Leal e Sousa, X.-T. Feng, T. Miranda & F. Martins
An investigation on stress wave propagation in micromechanical models of rock 501
R. Resende, L.N. Lamas, J.V. Lemos & R. Calçada
A comparison of empirical methods and evolutionary programming to predict blast-induced ground vibration 503
A. Sadeghee & H. Khoshrou
Effective explosive energy utilization for engineering blasting – initial results of an inventive stemming plug, SPARSH 505
Md. Sazid, M.R. Saharan & T.N. Singh
Rock support subjected to dynamic loading: Field testing of ground support using simulated rockburst 507
S. Shirzadegan, E. Nordlund, U. Nyberg, P. Zhang & L. Malmgren
Effects of loading rate on compressive strength of Maha Sarakham salt 509
T. Sriapai, P. Samsri & K. Fuenkajorn
Effect of artificial discontinuities on blast induced vibrations in open pit mines 511
Ö. Uysal, E. Arpaz, M. Cavus, S. Beyhan, Y. Tola & S. Yuvka
Wave propagation in inhomogeneous orthotropic rocks 513
C.D. Wang, D.S. Wu & J.J. Liao
Experimental investigation on dynamic mechanical performances of granite 515
W. Wang, Y. Hu, X. Ren & Y. Xiong
The application of smooth blasting for freezing rock in vertical shaft 516
X.G. Wang, E.X. Gao, Z. Wu, L.L. Zhang & L. Li
Fluid and gas flow
Experimental and numerical study on the behavior of supercritical CO2 injected to low permeable rocks 521
A. Arsyad, Y. Mitani, H. Ikemi, K. Kuze & S. Oura
A fully coupled thermoporoelastic model for drilling in HPHT formations 522
A. Diek, L. White, J.-C. Roegiers & D. Blankenship
Hydraulic stimulation of geothermal wells: Modeling of the hydro-mechanical behavior of a stimulated
fractured rock mass 524
S. Gentier, X. Rachez, M. Peter-Borie & A. Blaisonneau
Investigation of thermal effect on stability of deep wellbores in low and high permeable formations 526
Z. Hou & X. Luo
Modelling and optimization of multiple fracturing along horizontal wellbores in tight gas reservoirs 528
Z. Hou & L. Zhou
Dependence of fracture geometry and spatial variation in pressure on hydraulic conductivity in rock fractures 530
K. Kishida, T. Hosoda, A. Sawada, H. Sato, S. Nakashima & H. Yasuhara
Analytical solutions for poroelasticity within a finite two-dimensional domain 531
P.C. Li & D.T. Lu
Mechanical effect of the presence of gas on faults modeled as a sandwiched Cam-Clay material 532
V. Maury, J.-M. Piau & D. Fitzenz
Impact of fluids and formation anisotropy on acoustic, deformation and failure characteristics of
reservoir shales and pure clay minerals 534
A.N. Tutuncu & A.I. Mese
Change of joint aperture during formation of ice-ring around underground LNG storage cavern 535
E.-S. Park, S.-K. Chung, J.-H. Synn, W.-C. Jeong, T.-K. Kim & S.-H. Bae
A parametric study to investigate the effect of injection and production well operation on the CO2
sequestration process in deep unmineable coal seams 537
M.S.A. Perera, P.G. Ranjith & S.K. Choi
Development of the confined pressure type of multi-purpose rock flow system and its application
to the evaluation of permeability of various geomaterials 538
A. Sato, H. Yoshitake, T. Shiote & K.Sasa
Near-field application of aperture back calibrated distinct fracture network 539
M. Sharifzadeh & M. Javadi
XV

A combined Bayesian-Wavelet-Data fusion workflow for breakout zone identification in oil and gas wells 541
H. Soroush & V. Rasouli
Hydraulic stress effect on 3 dimensional pore network and permeabillity change 542
M. Takahashi, C. Ahn & M. Kato
Algorithm of coupled seepage and normal stress processes in fractured rock mass by composite element method 543
L.L. Xue, S.H. Chen & I. Shahrour
Evolution of fracture permeability in granite under temperature, confining pressure and pH conditions prescribed 545
H. Yasuhara, N. Kinoshita, S. Nakashima & K. Kishida
Physical clogging model using stick rate of calcite on CO2 mineral trapping 546
S.Y. Yoo, Y. Mito, T. Matsuoka, Y. Kuroda & A. Ueda
Risks and hazards
The SEALEX in situ experiments: Performance tests of repository seals 549

J.D. Barnichon, P. Dick & C. Bauer
Post-closure hydromechanical behaviour of a backfilled cavity 550
N. Dufour, H. Wong & F. Deleruyelle
Risk assessment of carbon dioxide storage in carboniferous formations 552
H. Manchao, R. Leal e Sousa, A. Gomes, L. Ribeiro e Sousa, E. Vargas Jr. & Zhang Na
Optimisation of a borehole-tunnel concept design for HLW disposal in granite using a 3D coupled THM modelling 554
H. Kunz, H.-G. Zhao, H. Shao, J. Hesser, J. Wang & V. Bräuer
Design and development of a large-scale THMC experiment of compacted bentonite for geological disposal
of high level radioactive waste in China 556
Y.M. Liu, J. Wang, X.G. Zhao, D. Ke, J.L. Xie, S.F. Cao, L.K. Ma, W.J. Jiang & L. Chen
Is hydraulic jacking a threat to the long-term safety of a nuclear waste repository? 558
M. Lönnqvist & H. Hökmark
Modelling of rockfall impact using spring-dashpot-slider system 560
T. Nishimura
Micro-mechanical study on the failure process of clay formations during excavation of galleries 561
S. You, G.F. Zhao & V. Labiouse
Criterion and classification for strain mode rockbursts based on five-factor comprehensive method 562
J.J. Zhang, B.J. Fu, Z.K. Li, S.W. Song & Y.J. Shang
Numerical simulation of micro-earthquakes induced by reservoir stimulation in the deep heat mining project Basel 564
L. Zhou & M.Z. Hou
Mining and rock support
New grouting materials using calcium phosphate compounds 569

M. Akiyama & S. Kawasaki
Design of room & pillar exploitations of non-expensive minerals and ornamental rocks in Spain 570
L.R. Alejano, F. García-Bastante, J. Taboada & R. Migliazza
Double shearing of rebar steel and cable bolts for effective strata reinforcement 572
N. Aziz, J. Nemcik & H. Jalalifar
Numerical evaluation of stress distribution along grouted rockbolts 573
S.A. Boulamatsi, P.P. Nomikos & A.I. Sofianos
Rock support for deep tunnels in highly stressed rocks 575
M. Cai & P.K. Kaiser
Analytical derivations of bolt profile shear strength 577
C. Cao, J. Nemcik & N. Aziz
Layout assessment of underground tunnel for the stability of gangway in non-coal mine 578
S.O. Choi
XVI

Longer-term performance of grouted bolts in different lithologies in the Illinois coal basin mines 580
Y.P. Chugh, H. Gurley & W. Bell
A new approach to simulate crush pillar behaviour in a tabular layout 581
M. du Plessis & D.F. Malan
Experiences with tunnel support in a South African platinum mine 583
L.J. Gardner
A new method on extraction of two thick seams simultaneously on extremely difficult ground condition – a case study 584
C.N. Ghosh, Prashant, P.K. Mandal, John Loui P. & A. Sinha
Interpretation of pull-out tests on instrumented rock bolts by means of 3D numerical modelling 585
R. Giot, C. Auvray, A. Giraud, B. Gatmiri & A. Noiret
Estimation of support requirement for large diameter ventilation shaft at Chuquicamata underground mine in Chile 586
E. Hormazabal, C. Soto, A. Russo & C. Carranza-Torres
Load distribution in inner anchor segment of prestressed cable with time 587
F. Jing, W.S. Li, C.H. Chen & Y.K. Liu
Reinforcement technique and its application in complicated roadways in underground coal mines 589
H.P. Kang, J.H. Wang & J. Lin
Determining the dimensions of combined supports in roadways 591
P. Konečný & R. Šňupárek
Leachate characteristic of sulphidic coal mine waste rock due to various water pouring interval 592
G.J. Kusuma, K. Hiroto, H. Shimada, T. Sasaoka, K. Matsui, R.S. Gautama & B. Sulistianto
Stresses and stability of a mine paste fill 593
C. Laín Huerta, P. Ramírez Oyanguren, R. Laín Huerta, F. Carnero, C. Santana & L. Alejano Monge
Establishing mining limits for partial undermining of surface infrastructure at Union Mine 595
Recent developments in the monitoring of excavation stability in South African mines 597
D.F. Malan
The influence of shear stress and weak contacts on pillar behaviour 599
J.A. Maritz & D.F. Malan
Performance of polymer skin spray-on liner in coal mines 601
J. Nemcik, I. Porter & E. Baafi
Rock slope reinforcement in open cut mining using grouted rock pins 602
G. Ren & S. Kerris
Geotechnical design of the sublevel stoping method for the Poder and Trabajo Veins in the San Genaro Mine – Peru 603
A. Samaniego & H.I. Pehovaz
Numerical study on disturbance of longwall mining to stress state in the rock mass 605
M. Shabanimashcool & C.C. Li
Ground support system at Leeville underground mine 606
C. Sun
Surface settlement due to deep mining with the Numerical Manifold Method 608
Y. Tal & Y.H. Hatzor
Numerical analysis of rock support interaction in tunnels in seismic regions 609
T. Tshering, R. Bhasin, R. Olsson & V. Maupin
Investigation into the strength of iron ore pillars in the Lorraine basin 611
J.N. van der Merwe, C. Didier & F. Poulard
On the fractal characterization of mining-induced crack network in overburden rock strata 613
Z. Wang, H. Xie & H. Zhou
Safety evaluation of an underground excavation for crusher operation 615
H.-S. Yang, M.-H. Jang, W.-B. Kim, H.-D. Jang, J.-G. Kim & S.-J. Kim
Experimental study on reinforcing effect of bolt on jointed rock masses under uniaxial tension 616
N. Zhang, S.C. Li, A.Z. Lu, M.T. Li & L. Yang
XVII

Tunnelling, rock caverns, and underground space
For a correct interpretation of ground reaction curves 621

F. Amberg
Creep closure rate of a shallow salt cavern 622
P. Bérest, J.F. Béraud, V. de Greef, B. Brouard, E. Hertz & C. Lheur
Convergences and core extrusions during construction of the Gotthard Base Tunnel 623
L. Cantieni, G. Anagnostou & R. Hug
Investigation of the tunnel penetration to the interaction of soil/rock-machine and its indices of cuttability 625
L.-H. Chen, Y.-C. Chen, C.-N. Chen & Z.-X. Yang
On the generalized thrust system for underground tunneling 626
L.-H. Chen, Y.-C. Chen, K.-J. Shou & G.-L. Lin
Numerical simulation of ice-ring formation in the jointed rock mass around underground
LNG storage cavern using thermo-hydraulic model 627
S.-K. Chung, E.-S. Park, J.-H. Synn, W.-C. Jeong & T.-K. Kim
Underground storage of natural gas and CO2 in salt caverns in deep and ultra-deep water offshore Brazil 629
A.M. da Costa, C.S. Amaral, E. Poiate, A.M.B. Pereira, L.F. Martha, M. Gattass & D. Roehl
Qualitative evaluation of the influence of blocky grounds on TBMs performances in deep Alpine tunnels 631
A. Delisio & J. Zhao
Impact assessment of water head on stability of a tunnel in India 632
R.D. Dwivedi & R.K. Goel
Development of moisture curing system for tunnel concrete lining 633
Y. Furukawa, M. Suzuki & Y. Mitani
The response of an argillaceous rock to ventilation: Process identification and analysis of an in situ experiment 634
B. Garitte & A. Gens
Prediction of convergences in rock tunnels excavated by conventional methods 636
J.A. González del Álamo & R. Jiménez
Reliable method for controlling tunnel convergence based on analytical results and monitoring results
in squeezing rocks 638
H. Hayashi, K. Aoki, H. Yoshioka, K. Yanagawa & N. Kitou
Probabilistic prediction of squeezing in tunneling under high-stress conditions 639
R. Jiménez & D. Recio
Shaft sinking and lining design for a deep potash shaft in squeezing ground 641
G. du T. Judeel, G.J. Keyter & N.D. Harte
Feasibility to control the permeability of clay barriers using geophysical methods 643
K. Kadiri, C. Schroeder & J.C. Verbrugge
Evaluation of the stability of a rock septum between two large derivation tunnels under high hydrostatic pressure 644
M.A. Kanji
Geological assessment of underground storage cavern sites 646
G. Kannan, R. Rath, A. Nanda & P. Vaskou
Development of low speed self-rotating nozzle system for drilling 648
A. Kizaki, K. Yokoi, K. Sakaguchi & K. Matsuki
Challenges and progress designing deep shafts and wide-span caverns 649
C. Laughton
Mechanism and key controlling technology of water inrush in tunnel construction 650
L.P. Li, S.C. Li, Z.H. Xu, B. Sui, S.H. Shi, Y. Zhou, C. Hu, Z.Q. Zhou & S.G. Song
Key technology to treat defective geological data-based comprehensive prediction during tunnel
construction in karst areas 652
S.C. Li, B. Liu, Q.S. Zhang, S.C. Li & L.C. Nie
Geomechanical model test of underground caverns in salt rock 654
Y.R. Liu, B. Li, F.H. Guan & Q. Yang
XVIII

4D seismic monitoring of small CO2 injection – Results from the Ketzin pilot site (Germany) 656
S. Lüth, A. Ivanova, C. Juhlin, A. Kashubin, N. Juhojuntti, P. Bergmann & J. Götz
The effect of pick orientation on rock fragmentation 658
S.S. Mostafavi, Q. Yao, L.C. Zhang, J. Lunn & C. Melmeth
The soundness diagnosis system of plane concrete lining for railway tunnel in Japan 659
N. Okano, K. Tsuno & Y. Kojima
Case study of huge squeezing behavior in expressway tunnel in service 660
Y. Okui, T. Tsuruhara, H. Ohta, M. Tada, S. Sakuma, C. Nakata & M. Sueoka
San José tunnels: Reparation of a ground collapse 661
C.S. Oteo, J.M. Maya & J. Vicente
Observational construction management by field measurement of large scale underground
railway station by urban NATM – Railway Bosphorus tube crossing, tunnels and stations 663
I. Otsuka, H. Taki, T. Aoki, M. Shimo, T. Kaneko, M. Iwano & S. Sakurai
Comparison of field and numerical modelling data for the steel pipe umbrella arch applications 664
S.C. Ozer, V. Ozacar & C.O. Aksoy
Assessment on stress induced instability in a tunnel project of the Himalaya 666
K.K. Panthi
Integrated modelling and design of reinforced concrete linings for tunnels 668
K.P. Paschalidis & S.C. Bandis
System behaviour in weak ground: comparison of yielding elements 670
N. Radončić & W. Schubert
Gypsum crystal growth in tunnels 672
A. Ramon & E.E. Alonso
Geomechanical aspects of CO2 sequestration and modeling 674
J. Rutqvist
The integrated prediction system for geological conditions ahead of tunnel faces 675
S. Shirasagi, T. Yamamoto, Y. Mito & M. Ashraf
Stability analysis of concrete plugs in a pilot cavern for compressed air energy storage 676
W.K. Song, D.W. Ryu & Y.K. Lee
Salt and rock mechanic challenges at the verification of drift seal systems at the Morsleben repository, Germany 678
J. Stahlmann, R. Mauke, M. Mohlfeld & C. Missal
TBM vs. D&B – a difficult choice in mountain terrain – some geotechnical guidelines 679
M. Verman, T.G. Carter & L. Babendererde
Underground research laboratory network 681
J.S.Y. Wang, P.H. Smeallie, X.-T. Feng & J.A. Hudson
The state-of-the-art and research advances on the underground natural gas storage in bedded salt rocks in China 682
C.H. Yang, Y.P. Li & D.A. Qu
Probabilistic assessment of stability of underground rock caverns and cavern shape optimization 684
W.G. Zhang, A.T.C. Goh & J.Y.K. Wong
Deformation and failure mechanism of surrounding rock mass around underground caverns in
Jinping hydropower station I 685
X.P. Zhou & Q.H. Qian
Rock slopes and foundations
Importance of anisotropy in dam foundation, estimated by in-situ dilatometer tests 689

B. Abrah, M. Karami & L. Faramarzi
Application of rockfall risk assessment techniques in two aggregate quarries 691
L.R. Alejano, M. Veiga, I. Gómez-Márquez & H. Dellero
Seismic stability analysis of rock slopes using the numerical manifold method 693
X.M. An, Y.J. Ning, G.W. Ma & Z.Y. Zhao
XIX

Stabilization of a landslide in Valgrisenche, Italy 695
G. Barla, F. Antolini & M. Barla
Development and characteristics of a deep-seated slope movement 697
Y.-C. Chiu, C.-H. Lee, T.-T. Wang & T.-H. Huang
3DEC modelling of p-y behaviour of laterally loaded piles in jointed rock 699
W.L. Chong, A. Haque, P.G. Ranjith & A. Shahinuzzaman
Field test and interpretation of the side shear resistance of a drilled shaft in rock 700
D. del Olmo, C. Olalla, S. Melentijevic & L. Prieto
What modern rock mechanics owe to the Malpasset arch dam failure 701
P. Duffaut
Prediction of ultimate bearing capacity of discontinuous rock foundations on the basis of limit analysis 702
A. Fahimifar & M. Imani
Simulating the failure mechanism of rock slopes (due to kinked and secondary cracks
propagation) by a higher order displacement discontinuity method 703
M. Fatehi Marji & M. Eghbal
The assessment of stability of grand Dallagkaus landslide in reservoir of Zaramag hydro power development, Russia 704
E.G. Gaziev & V.V. Rechitski
Slope stability analysis using a hybrid Finite-Discrete Element method code (FEMDEM) 706
G. Grasselli, A. Lisjak, O.K. Mahabadi & B.S.A. Tatone
Evaluation of the stability of a bullnose within Coedmore Quarry, Durban, South Africa 708
A.J. Greet, K. Naidoo & E.D.C. Hingston
Centrifuge model test on earthquake-induced failure behaviour of slope in discontinuous rock mass 709
M. Ishimaru & T. Kawai
Evaluation of slope hazards stability and failure probability in a mountainous area along a highway 710
N. Jia, Y. Mitani, I. Djamaluddin & H. Ikemi
Seismic instability assessment of rock slopes in a large area based on planar sliding mode 711
J.-C. Jiang, T. Yamagami & Q. Yang
Deterministic stability analyses of a rehabilitated pit wall slope at Coedmore Quarry, Durban 712
K. Naidoo, B.R. Jones & E.D.C. Hingston
Developing a slope stability curve of open pit coal mine by using dimensional analysis method 713
S. Kramadibrata, S. Saptono, R.K. Wattimena, G.M. Simangunsong & B. Sulistianto
Rock slope risk analysis based on non-linear failure criterion 714
A.J. Li, M.J. Cassidy & A.V. Lyamin
Control and evaluation for residual subgrade settlement of high speed railway 715
X.H. Li, Z. Chen, X.Y. Zhao, H.L. Sun, J.H. Guo & T. Zhao
The role of rock mechanics investigations in Bakhtiary dam and hydropower plant project 717
A. Mehinrad, Kh. Binazadeh, A. Gheshmipour, H. Hamzehpour, H. Chehreh & M. Haftani
Design of spread foundations on rock masses according to Eurocode 7 719
T. Miranda, F. Martins & N. Araújo
Unloading scheme to control sliding mass at Angouran open pit mine, Iran 721
P. Moarefvand, M. Ahmadi & M. Afifipour
A heterogeneous multi-velocity model for the location of microseismic events in rock slopes 722
C. Occhiena, M. Pirulli & C. Scavia
Deep foundations on weathered rocks 724
C.S. Oteo
Failure modes and countermeasures of large-scale rock slope composed of Phyllite in Korea 726
B.-S. Park, H. Cho, S.-H. Cha & D.-I. Park
Foundations of gravity dams: Rock mechanics requirements 728
M. Romana
XX

Use of ground-based synthetic aperture radar to investigate the complex 3-D kinematics of a large open pit slope 730
J. Severin, E. Eberhardt, L. Leoni & S. Fortin
Rock slope stability analysis using the discontinuous deformation analysis 732
J.P. Sun & Z.Y. Zhao
Mechanism of rock avalanche induced by earthquake – Insight from the discontinuous numerical modeling approach 734
X. Tu, G. Zhao, F. Dai & J. Zhao
Stability analysis of jointed rock slope evaluated by 3DEC 736
A. Varughese, G.W. Rathod & K.S. Rao
Estreito HPP – Geotechnical and geomechanical characterization of the Sandstone foundation rock mass 738
E. Wunder, M.C. Meireles, N. Borchardt, A.M. Calcina & M.A. Camargo
Mechanism and dynamics of dip-slope failure revealed by LiDAR data and Discrete Element method 740
C.-H. Yeh, M.-L. Lin, Y.-C. Chan & K.-J. Chang
Interdisciplinary
A comprehensive approach in predicting excessive sand production 743

M. Bai, C. Santana & X. Shen
The theory of underground structures design has gained strength 744
N.S. Bulychev, N.N. Fotieva & P.V. Deev
Simulation of steel fiber reinforced shotcrete behavior using fracture mechanics 746
A. Fahimifar & B. Mehrzad
Rock mechanics aspects of drill bit rock interaction 747
S.A.B. da Fontoura, N. Inoue, I.M.R. Martinez, C. Cogollo & D.A. Curry
The impact of structural-tectonic of the rock mass on the formation and development of geo-deformation processes 749
E. Freiberg, E. Bellendir, V. Golitsyn, N. Ablyamitov, E. Cherkez, E. Tchujko & G. Bich
Simulation of arenaceous weak rocks by means of cemented sands 751
V.R. Galván Liévano & M.A. Kanji
3D geological documentation system based on digital photogrammetry 753
L.Q. Guo, J.B. Liao, J.J. Chen, X.Yu & J. Tao
Uncertainty and sources of error in rock engineering 755
J. Hadjigeorgiou & J.P. Harrison
Analysis of the infrastructure system and core factors for environment-friendly urban regeneration 757
K.C. Han, D.W. Ryu, H.M. Kim & T.H. Kim
Microwave assisted mechanical rock breaking 759
F. Hassani, P.M. Nekoovaght, P. Radziszewski & K.E. Waters
Continuously variable frequency electrochemical characteristics of rock 761
J.C. Hu, F. Wang, J.J. Cui & M.C. He
Workflow design and software customization to enable integration of lidar data with active
drill-and-blast tunnel construction projects 763
M.J. Lato, M. Vöge, E. Dale & E. Morgan
Parametric sensitivity investigation: Analysis of wellbore stability 765
G. Li & M. Bai
Study and application of visualized model test technologies for gas storage cluster in salt rock 766
L.J. Cheng, Z.K. Li & B. Xu
Real-time microseismic navigation for CBM directional drilling 768
X. Luo
On the minimum size of representative volume element: An n-point probability approach 769
D. Łydżba & A. Różański
Estimation of rock mass deformation modulus based on GSI system 771
A. Majdi, A. Bashari & M. Beiki
XXI

New physical modeling of flexural toppling failure with titling machine 773
A. Majdi, Ö. Aydan & M. Amini
Fusion of drill monitoring data with geological borehole assays 775
A. Melkumyan, P. Hatherly & H. Zhou
Parametric study of formation stability using a hollow cylinder model 776
P.A. Nawrocki, Z. Qi & D. Wang
Rock mass characterization using attenuation of coda waves 777
R.A.O. Pedroza, J.C.A. Vargas & P.C.E. Rodríguez
Rock mass classification based on seismic measurements using Support Vector Machines 779
A.J. Rechlin, S. Lüth & R. Giese
A rock mass classification to estimate excavation rate, support and rock properties in a borehole survey 781
M. Sawada, K. Shin, Y. Inohara, T. Shidahara, T. Hatano & T. Miwa
Charts for estimating active earth pressures on retaining walls assuming Hoek & Brown non-lineal strength criterion 782
A. Serrano, A. Perucho & M. Conde
A new classification system for wet rocks (MiC System) 784
H. Soroush & A. Fahimifar
The life-cycle information management practice of Shanghai Yangtze River Tunnel and Bridge Project 785
H.Y. Tian & Q.W. Liu
In-situ experimental studies on improvement of deformability of rock masses by grout treatment 787
S. Utsuki, T. Asakura & K. Kikuchi
Intelligent recognition of rock mechanical models 788
C.X. Yang, X.T. Feng & B.R. Chen
Verification of reinforcing effects of a tunnel face reinforcement method by centrifuge model
tests and numerical analysis 790
Y. Yokota, K. Date & T. Yamamoto
Author index 791
XXII

Preface
This proceedings volume contains papers of ISRM 2011 Beijing – the 12th International Congress on Rock Mechanics of the
International Society for Rock Mechanics (ISRM), held at the China National Convention Centre in Beijing, China, from 17–21
October 2011. It contains the full papers of the Müller lecture by Dr Nick Barton, the Rocha Medal lecture by Dr Dohyun Park,
13 keynote papers and the extended abstracts of 384 technical papers. A CD-ROM contains all the full papers.
The ISRM will be 50 years old in 2012. This congress had an added significance because the ISRM Board decided to kick
off the 50th Anniversary celebrations at this congress. For the congress proceedings, we have three special keynote lectures
marking this major milestone in the ISRM. Two keynote lectures by Prof Ted Brown and Prof John Hudson deliberated on the
progress in rock mechanics of the past 50 years, and how rock mechanics will develop in the next 50 years. The third keynote
lecture by Prof. Qihu Qian provided an overview on developments of rock engineering technology in China.
The theme of the ISRM 2011 Congress is Harmonising Rock Engineering and the Environment, emphasing the critical
role of rock mechanics in meeting rock engineering requirements for development and the need to protect and preserve our
environment. The Congress received a total of 384 contributed papers from National Groups, other regions, and individual
members of the ISRM. Abstracts and full papers were reviewed by the National Groups. Papers submitted by individual members
were received and reviewed by the Scientific and Publications Committee.
The ISRM 2011 Congress was jointly organised by the Chinese Society for Rock Mechanics and Engineering and the Society
for Rock Mechanics & Engineering Geology Singapore, with support from the Korean Society for Rock Mechanics.
On behalf of Organising Committee, the editors express their appreciation for the contribution of all supporters, sponsors,
ISRM National Groups, authors, and staff of secretariat from both China and Singapore, which has made this congress a success.
Finally, we would like to thank Cheng Hoon Khoo and Zaiyi Guo for their excellent work in managing the paper submission.
The Editors
Qihu Qian, China

Yingxin Zhou, Singapore
XXIII

Organisation
Congress Chairs:
Qihu Qian, Jian Zhao
Congress Co-Chairs:
Xia-ting Feng, Chung-In Lee, Yingxin Zhou
Organising Committee
Chairman: Qihu Qian
Co-chairman: Xia-ting Feng, Yingxin Zhou
Committee Members (China)

Mei-Feng Cai Hai-Bo Li Chun-An Tang Wen-Li Xu
Bing-Jun Fu Shu-Cai Li Leslie George Tham Qiang Yang
Man-Chao He Zhong-Kui Li Jin-An Wang Zhi-Fa Yang
Li-Xing Huang Yan-Jun Shang Ju Wang Bing-Xu Zheng
Yu-Yong Jiao Qian Sheng Fa-Quan Wu He-Hua Zhu
Committee Members (Singapore)

Jungang Cai Chun Fai Leung Guowei Ma Louis Ngai Yuen Wong
Zhiye Zhao Jian Zhao
Secretariat Members
Ting Feng Wei Hu Chun-Guang Li Wen-Lin Lu
Zhong Ren Yan-Jun Shang Xiao-Jie Yang Shuang-Liang Zheng
Yan Zhou
National Advisory Board (China)

Xuecheng Dong Guangding Liu Guangzhong Sun Xuefu Yu
Qiuling Ha Tongyou Liu Mengshu Wang Chaoran Zhang
Dingcheng Huang Jin Ma Wulin Wang Jingjian Zhang
Jiongjun Liang Xizhuo Niu Yongjia Wang Simeng Zhou
Junde Lin Minggao Qian Dingyi Xie Weiyuan Zhou
Yunmei Lin Yongjin Song Liangying Xie Keshan Zhu
Baochen Liu Zhenqi Song Linde Yang
XXV

International Scientific Committee
Giovanni Barla, Italy Jia-Zheng Pan, China
Nick Barton, Norway Frederic Pellet, France
Pierre Bérest, France Shunsuke Sakurai, Japan
Barry Brady, Australia Wulf Schubert, Austria
Ted Brown, Australia Krishan Gopal Sharma, India
Zu-Yu Chen, China Gen-Hua Shi, USA
Eda Freitas de Quadros, Brazil Luís Ribeiro e Sousa, Portugal
Erik Eberhardt, Canada Richard Stacey, South Africa
H.H. Einstein, USA Ove Stephansson, Sweden
Charles Fairhurst, USA Jun Sun, China
Xiu-Run Ge, China Joong-Ho Synn, Korea
Jing-Cai Gu, China Resat Ulusay, Turkey
John Harrison, United Kingdom Alain Van Cotthem, Belgium
Yossef Hatzor, Israel Nielen van der Merwe, South Africa
Seokwon Jeon, Korea Si-Jing Wang, China
Lan-Ru Jing, Sweden Walter Wittke, Germany
Peter K. Kaiser, Canada He-Ping Xie, China
Suseno Kramadibrata, Indonesia Jiro Yamatomi, Japan
Marek Kwasniewski, Poland Kwet Yew Yong, Singapore
C.F. Lee, China Thierry You, France
Abbas Majdi, Iran Ying-Ren Zheng, China
Yuzo Ohnishi, Japan Robert Zimmerman, United Kingdom
International Advisory Board

John A. Hudson, United Kingdom C. Derek Martin, Canada
ISRM President ISRM Vice President for North America
Daniel Francois Malan, South Africa Alvaro Gonzalez Garcia, Colombia
ISRM Vice President for Africa ISRM Vice President for South America
Abdolhadi Ghazvinian, Iran Claus Erichsen, Germany
ISRM Vice President for Asia ISRM Vice President At Large
Anthony Meyers, Australia Xia-Ting Feng, China
ISRM Vice President for Australasia ISRM Vice President At Large
Nuno Grossmann, Portugal Luis Nolasco Lamas, Portugal
ISRM Vice President for Europe ISRM Secretary General
XXVI

Müller lecture

From empiricism, through theory, to problem solving in rock engineering
N.R.Barton
Nick Barton and Associates, Oslo, Norway
ABSTRACT: The behaviour of the jointed-and-faulted-anisotropic-water-bearing media that we call rock masses, was an
abiding pre-occupation of Leopold Müller. The author has been similarly pre-occupied. So starting with modest developments
from tension-fractured physical models, and progressing via simple UDEC models to the real jointed and three-dimensional
world in due course, a few of the lessons learned and subsequently applied in rock engineering practice will be described.
The failure of jointed rock masses involves mobilization of the intact rock strength at smaller strain than that for the jointing.
This was also discussed by Müller in 1966. Continuum modelling with conventional strength criteria is questioned. A simple
solution is proposed. Excavation disturbed zones caused by joint shearing are compared with log-spiral failures. The reality of
jointing as affecting the deformation of a large cavern, as affecting NMT pre-injection design in rock masses, and as affecting
the compaction and subsidence behaviour of a reservoir are each illustrated. Discontinuous behaviour provides rich experiences
for those who value reality, even when reality has to be simplified by some empiricism.
Subject: General
Keywords: Rock joints, rock masses, physical modelling, empiricism, site characterization, tunnelling, rock failure
1 INTRODUCTION the use of empiricism both in rock engineering and in rock

mechanics.
The deformation resistance of the material bridges takes effect David Hume (1711–1776) nevertheless claimed that there
at much smaller deformations than the joint friction: this joint is no certainty that the future will resemble the past, which
friction makes partly up for lost strength. (Müller, 1966). This presents a potential challenge to our rock classification abil-
opinion remains highly relevant, just as it was at the time of ities, and fuel for the few Austrian, Australian, Italian and
the first ISRM congress in Lisbon. Norwegian critics of empiricism. New developments repre-
Why are so many young people, even some older, being sent a threat to ‘the establishment’ in all subjects, but even
so attracted by the ease of producing colourful appendices of critique that is blind may be a basis for advances and fine-
questionable relevance to reality? Is simplicity a good enough tuning by those who are making developments, so should be
reason, when Müller is no longer guiding us in more chal- welcomed, even if reluctantly.
lenging directions? Is rock mechanics supposed to be ‘easy’? A fundamental part of the scientific method is that all
Why are we so confident that ‘c plus σn tan ϕ’should be a guid- hypotheses and theories must be tested against observations
ing principle of shear failure in rock masses? In fact ‘c then of the natural world, rather than resting solely on a priori
σn tan ϕ’or degraded cohesion as friction is mobilized, is much reasoning or intuition. On this basis the representation of an
more logical. obviously anisotropic rock mass in a continuum model can be
Of course rock masses are very complex media, so some seriously questioned, despite its widespread practice.
level of simplification is needed. The question is whether the Leopold Müller was one of our discipline founders, and
real effects of joints and faults can be ‘lifted’ into a contin- one who constantly emphasized the importance of joints and
uum analysis. Empiricism is used in this process, but later the discontinuities. His analysis of the Vajont rock slide (Müller,
rock mass loses its identity. Are layers of algebra a justified 1964) was a direct source of inspiration for one who started
next step to the empiricism? Popularity suggests yes. Common to study rock mechanics in 1966. The dramatic discussion
sense suggests no. Where is the anisotropy? of shear strength and elevated joint water pressure at Vajont,
It would seem to be an advantage if classical British empiri- from Müller’s landmark publication, certainly had important
cists could influence the more recent empiricism in rock educational influence on the author’s life-time pre-occupation
engineering. In the late 17th century John Locke (1632–1704) with the importance of discontinua. This was a strong fea-
proposed in An Essay Concerning Human Understanding ture of early Imperial College and subsequent NGI work, with
(1689) the influential view that the only knowledge humans numerous valued colleagues. Discontinuous behaviour is also
can have is a posteriori, i.e., based upon experience. Accord- interesting.
ing to Locke, the ultimate building blocks of knowledge are Leonardo da Vinci (1452–1519) once gave advice that was
based on simple ideas. George Berkeley (1685–1753), a min- distinctly helpful to one starting out in a relatively undeveloped
ister, considered that any order humans may see in nature is the field: ‘If you find from your own experience that something is
language or handwriting of God, and also insisted that expe- a fact and it contradicts what some authority has written, then
rience is the source of most knowledge. The role of experience you must abandon the authority and base your reasoning on
and evidence in the formation of new ideas, should be guiding your own findings’.

Figure 1. Confronted with this potentially unstable jointed rock
slope, multiple reasons for the over-break and instability suggest Figure 2. A study in contrasts: physical modelling using ten-
themselves, at least to one involved with quantitative empirically- sion-fracture generation, and numerical modelling, using µDEC:
based methods. There are clearly adverse values of JRC, JCS, and example with a friction angle for the joints of ϕ = 20◦ throughout.
ϕr , and there are also adverse ratings of Jn, Jr, Ja (and Jw on
occasion). However dip and dip direction must be the fundamental
geometric/gravitational drivers in this apparent stability problem.
Q × σc /100, with the uniaxial compression strength expressed
in units of MPa, thereby normalizing Q.
As a result of the strong non-linearity of the peak shear The fact that both the Q-parameter development and their
strength envelopes for the tension fractures to be illustrated ratings were strongly steered by case records concerning suf-
in Figure 2, the writer has followed a life-long habit of never ficient shotcrete (compensating for lack of cohesion), and by
using cohesion ‘c’ for representing the shear strength of rock sufficient bolting (compensating for lack of friction), may add
joints. Shear tests on non-planar rock joints also demonstrate some slight credibility to this unexpected finding.
non-linear strength envelopes. Since ‘c’ may really be non-
existent, convention should be questioned.
2 TWO-DIMENSIONAL ROCK MASSES: FICTIONAL
According to a MacGraw Hill dictionary definition: The
BUT INFORMATIVE
collection of empirical data may be guided largely by prelim-
inary theoretical exploration of what to expect. The empirical
The desire to model the behaviour of jointed rock slopes in
method is necessary when entering hitherto completely unex-
late nineteen sixties post-graduate studies at Imperial Col-
plored fields, and becomes less purely empirical as the
lege, (consistently and innocently called Empirical College
acquired mastery of the field increases. Successful use of
in several countries in the Far East), led to tension-fracture
an exclusively empirical method demands a high degree of
models by the writer, and numerical modelling developments
intuitive ability in the practitioner.
(pre-µDEC) in the case of student colleague Peter Cundall.
The above opinions seem to have some application to devel-
The relative inflexibility and flexibility of the two approaches
opments such as the Q-system and the Barton-Bandis model
is readily imagined from Figure 2.
for rock joint behaviour, that will be used in various contexts
The single numerical slope model shown in Figure 2,
in the field examples presented in this paper.
demonstrated the influence of changed friction angles, and
The development and application of these simple
was reported in Cundall et al., 1977, but performed some years
empirically-based parameters will inevitably form a part of
earlier, following Cundall, 1971 developments. The writer has
this paper, as they have proved very useful for describing con-
optimistically assumed that the demonstrable inflexibility of
ditions in the form of readily understood numbers, rather than
his physical-models, contributed to Cundall’s later develop-
adjectives, as in the distant past.
ments of UDEC and 3DEC, which are so much appreciated
Historically the discovery of empirical relationships has
by those desiring a closer approach to modelling the complex
been important as a stepping stone to the later discovery of
reality of excavating in jointed rock.
theoretical relationships, and on occasion, what was thought
Despite the shortcomings of physical tension-fracture mod-
to be an empirical factor may later be considered a funda-
els, the writer was nevertheless excited to discover that the
mental physical constant (or variable). One may claim that
peak shear strength of these rough and clearly unweathered
the rock joint parameters JRC, JCS and ϕr introduced in the
tension fractures could be described by a simple relation
next section, have resemblance to physical variables of merit.
involving the uniaxial compression strength (σc ) of the model
However the form of the equation linking these parameters to
material (Barton, 1971):
the peak shear strength has yet to be explained on theoretical
grounds, so its empiricism remains.
Likewise the approximation of the (Jr/Ja) part of the Q-
formula to inter-block friction coefficient, shows close resem-
blance to this important physical variable, though the overall This equation, and simple links to peak dilation angle, proved
empirical combination in the form a/b × Jr/Ja × e/f anchors to be the ‘end-member’of the Barton and Choubey, 1977 equa-
the Q-formula firmly in the realm of empiricism. tion for the peak shear strength of rock joints, since unweath-
A more recent discovery that Qc ≈ c × tan ϕ for the rock ered tension-fractures clearly maximise surface roughness,
mass, may seem to defy rational explanation. However, the wall strength and residual friction angle:
fact remains that a strong resemblance to c and tan ϕ is indeed
present in two halves of the formula for Qc , which equates to

Figure 3. Physical models each of 40,000 blocks created for rock
slope analyses in Barton, 1971, were followed some years later, by
physical model studies of caverns for underground nuclear power
plants, studying the generic effect of joint-set (fracture) orientation,
anisotropy, and horizontal stress in models with 20,000 blocks. This
research was performed in NGI, Oslo. Barton and Hansteen, 1979
also compared the physical results with FEM.
where the joint roughness coefficient (JRC), the joint wall

compression strength (JCS) and the residual friction angle
(ϕr ) can each assume variable magnitudes, caused by variable
roughness and variable weathering. The first equation was
based on direct shear tests of more than 200 artificial samples,
while the second equation was based on DST of 130 rock joint
samples, some of them weathered.
Note that the angle ϕr may be several degrees less than ϕb
(used earlier for flat unweathered surfaces), and JCS may be
significantly lower than σc for the same reasons. JRC should Figure 4. The contrasting effects of k0 = 1, or σH > σV on deforma-
be calculated on this basis, if back-analysed from testing, as tion vectors, were recorded in these pre-UDEC-BB cavern models
emphasized by Barton and Choubey, 1977. In addition there (top two figures). These effects seemed to be accentuated by frac-
are block-size scale effects, to be mentioned later. ture-set orientations and by fracture-set anisotropy (set 1 was always
dominant, set 2 was always ‘stepped’ where crossing set 1, due to the
As will be shown in the next figure, the performance method of generation seen in Figure 2). The sensitivity to ‘joint-set’
of UDEC-MC models with varied block sizes, nevertheless orientation led to speculation 10 years later, when checking designs
showed a significant effect of block size, due to the greater for the 62 m span Gjøvik Olympic cavern, that this huge rock arch
freedom for block rotations when blocks were very small. could ‘heave’ or ‘subside’. The reality, based on surface optical mon-
Something resembling kink-band rotation could occur. itoring, and seven MPBX installed from the surface, integrated with
Although with 2D limitations, and highly idealized, Figure three MPBX installed from the top-heading, was a 7 to 8 mm down-
6 demonstrates the potential for a much deeper EDZ in the case ward movement. The UDEC-BB models shown above, gave a good
of severely faulted or crushed rock, if the assumption is made (Class A) prediction of 7 to 9 mm downward movement. (Barton
that rock in the neighbourhood of faults is often crushed or et al., 1994).
displaying smaller block sizes, even additional jointing. Core
is often crushed and appears to show Jn = 15 (four sets) or release, and frequent blocking of the cutter head by the falling
worse in such locations. larger blocks. Barton, 2008.
With the possible addition of clay coatings or thin fillings of
clay, or even a core of thick clay somewhere in the midst of a
fault zone, one may expect Q-parameters in the neighbourhood 3 MODELLING FAILURE THROUGH
of Q = 10/15 × 1/6 × 0.33/2.5 = 0.01. The severity of the rock MASSIVE ROCK
mass characteristics, and effects on e.g. TBM progress, may
be heavily dependent on the presence (or absence) of water in The ‘45◦ sectors’ EDZ seen in the previous figure, that are
such cases. Outwash of clay may trap TBM, due to rock-block a feature of (idealized) jointed rock with two equal sets and

Figure 5. Biaxially loaded two-dimensional physical models with
respectively 250, 1000 or 4000 discrete blocks also helped to give
clues about scale effects caused by different block sizes. In the case of
‘rock masses’ formed with rough tension fractures, the shear strength
of the mass tended to be highest with smaller block sizes (due to
the tendency for block rotations), as also observed by Bhasin and
Barton, 1997 in UDEC-BB models. UDEC-MC (Mohr-Coulomb)
models did not show this block rotation effect with the same block
sizes. These physical models were described by Barton and Hansteen, Figure 6. UDEC-MC models with successively halved block sizes
1979, therefore pre-dating UDEC-BB. but 45◦ /45◦ jointing, demonstrate a strong ‘mass scale effect’, that
appears to depend on the greater degrees of freedom for block rota-
tions, as block size reduces. The central plots show widely different
penetration of inter-block shearing. Mohr-Coulomb based analytical
any conjugate orientation (45◦ , 60◦ , 75◦ , 90◦ ) of the jointing,
solutions, also given by Shen and Barton, 1997, confirm the gen-
forms a strong contrast to the development of stress-induced eral trends. UDEC-BB models with 0◦ /90◦ block orientations, also
failure in the case of massive rock or intact model mate- demonstrate joint shearing in the same ≈45◦ sectors, confirming the
rial. Joint-industry borehole stability studies performed for penetration of the EDZ into these zones.
oil companies at NGI in the late eighties, provided strong
evidence of shear failure development in the form of log-
spiral surfaces, with physically measured shear displacements.
This log-spiral form has recently been seen when inspect- radial strain. In reality it may be a tangential strain-related
ing TBM headrace tunnels in massive sections of marble and failure phenomenon.
schists, where the estimated theoretical maximum tangential The very important findings of Hajiabdolmajid et al., 2000
stress (σθ = 3σ1 –σ3 ) was exceeding the ‘limit’ 0.4 × UCS, are summarised briefly by means of the first six figures
i.e. increased SRF in the Q-system. assembled in Figure 8. The demonstrated shortcomings of
The borehole stability studies involved drilling into a continuum modelling with ‘c plus σn tan ϕ’ shear strength
0.5 × 0.5 × 0.5 m polyaxial cell full of anisotropically loaded assumptions, should have alerted our profession for change
model sandstones. Drilling could be performed in various already ten years ago, but deep-seated beliefs or habits are
directions in relation to the three principal stress directions. traditionally hard to change.
Loading was provided by flat-jacks. Figure 6 shows some Rock masses actually follow an even more complex pro-
of the phenomena observed by ‘slicing’ the models after gression to failure, as suggested in Barton and Pandey, 2011,
low pressure saturation with epoxy, in order to preserve the who recently demonstrated the application of a similar ‘c then
drilling-induced (and therefore stress-induced) failures. tan ϕ’ modelling approach, but applied it in FLAC 3D, for
Attempts to model ‘break-out’ phenomena such as those investigating the behaviour of multiple mine-stopes in India.
illustrated are not especially successful with standard Mohr A further break with convention was the application of peak
Coulomb or Hoek Brown failure criteria, because the actual ‘c’ and peak ‘ϕ’ estimates that were derived directly from
phenomena are not following our long-standing concept of mine-logged Q-parameters, using the CC and FC parameters
‘c plus σ n tan ϕ’. The reality is degradation of cohesion at suggested in Barton, 2002. For this method, an estimate of
small strain and mobilization of friction (first towards peak, UCS is required, as CC (cohesive component) and FC (fric-
then towards residual) which occur at larger strain. We register tional component) are derived from separate ‘halves’ of the
closure or squeezing, and also can measure it, as an apparent formula for Qc = Q × σc /100.

Figure 7. Top: log-spiral shearing as a result of drilling a 45◦ Figure 8. Top: The Canadian URL mine-by break-out that devel-
inclined borehole in the direction of σH into a stressed block of model oped when excavating by line-drilling, in response to the obliquely
sandstone, with principal stress ratios σv = 1, σH = 0.8, σh = 0.4. acting anisotropic stresses. This is followed by an important demon-
Addis et al., 1990. The theoretical tangential and radial stresses acting stration of unsuccessful modelling by ‘classical methods’ given by
on log-spiral shear surfaces are from Bray 1966 lectures, but these Hajiabdolmajid et al., 2000. They followed this with a more real-
might be suppressed by dilation if rock strength is too high. istic degradation of cohesion and mobilization of friction, which
was applied in FLAC (their 6/02/1999 modelling date removed
for clearer presentation). The bottom figure is a FRACOD fracture
The two or three classes of discontinuities (natural and mechanics approach to modelling of stress-fracturing in a TBM tun-
induced) involved in pre-peak and post-peak rock mass failure, nel (pers. comm, Baotang Shen, 2004). The project is now delayed,
due to irreparable damage to the TBM from a rock burst. Advised
will also have quite different sets of shear strength proper-
drill-and-blast is finally occurring.
ties. For instance the new stress-induced failure surfaces, if
described with JRC, JCS and ϕr , might have respective num-
bers (at small scale) like 18 to 22, 100 to 150 MPa and 30◦ models, making them even poorer representations of the
to 32◦ (i.e. rough and unweathered and strongly dilatant), strain-and-process-sensitive reality?
compared to perhaps 4 to 8, 50 to 100 MPa and 27◦ to 29◦ Input data obtained via Hoek and Brown and GSI formu-
for potential joint sets, or perhaps Jr/Ja = (1 to 2)/4 for any lations that obviously ignore such complexity, nevertheless
clay coated discontinuities, that might also be involved in consist of remarkably complex algebra (e.g. Table 1) in com-
post-peak behaviour of the rock mass. parison to the more transparent formulae for discontinuum
The dilatancy obviously reduces strongly between these codes, where JRC0 , JCS0 , ϕr , L0 and Ln and use of Barton-
three groups of discontinuities. Furthermore, each of the Bandis scaling equations, are sufficient to develop the key
above are features that begin to resist shearing at consid- joint strength and joint stiffness estimates. (L0 and Ln are the
erably larger strains/deformations than is the case for the lab-scale and in situ scale block sizes).
also strongly dilatant failure of the ‘intact bridges’. Why A demonstration of the simpler, even simplistic ‘cohesive
therefore are we adding ‘c and σn tan ϕ’ in ‘continuum’ component’CC and the ‘frictional component’FC for a variety

of rock mass characteristics is given in Table 2. These param- stope face area and perimeter (the ‘hydraulic radius’) to the
eter estimates have the advantage of not requiring software rockmass stability number N, which is a direct product of the
for their calculation – they already exist in the Q-parameter factors: RQD/Jn (‘block-size’) and Jr/Ja (frictional strength)
logging data, and the effect of changed conditions such as between the rock block surfaces. Three other factors are added
clay-fillings, can be visualized easily. in the N’ method, described by Hutchinson and Diederichs.
The relatively new parameters CC and FC, were ‘hiding’ in One of them is a necessary stress term to allow for the strength-
the Q-formula for 25 years, and needed the simple addition of to-induced-stress ratio, as traditionally used when estimating
a normalising (Q) × σc /100 for their final recognition. SRF for the Q-value.
It should be emphasised that all the Q-parameter ratings
developed long ago by Barton et al. 1974, were succes-
sively improved by 6 months of trial-and-error fit to required 4 Q AND SINGLE-SHELL NMT FOR METRO
rock bolting needs (bolting is compensating for low frictional TUNNELS
strength), and to required shotcreting needs (shotcrete is com-
pensating for low cohesive strength). Numerous tunnels and Chance or fate bought the writer to Norwegian line-where-
caverns spanning an order of magnitude of dimensions (3 to needed, bolt-where-needed ‘nominally unlined’ hydropower
30 m), and four orders of magnitude of rock quality (0.001 to tunnel territory in 1971. This region of Scandinavia eventually
10) provided the guiding details of support needs. excavated about 3,500 km of such tunnels, 1,500 km of road
The fact that Qc strongly resembles c x tan ϕ (different from and rail tunnels, and 200 underground power houses. In 1973
c + tan ϕ), perhaps emphasizes the inherent utility of this num- the Norwegian State Power Board (subsequently Statkraft)
ber as a multiple-orders-of-magnitude descriptor of rock mass had requested a report from NGI, concerning a technical
conditions. The simple peak ‘c’ and peak ‘ϕ’ estimates, as explanation as to why Norwegian hydropower caverns were
illustrated below, have potential limitations, but they are trans- displaying widely different magnitudes of deformation. This
parent, and can be rejected (with due care), if really considered agency, now the owner of the most electrical generating capac-
unrealistic. In this connection it is considered that rock mass ity in the world, was apparently not hurt by waiting more than
cohesive strength seems to be systematically under-estimated six months for their report, which could not actually be written
in numerical modelling. until a rock-mass classification method had been developed.
There may be some additional advantages of this simple The nature of the question meant that rock-mass quality,
semi-empirical approach, because the pairs of parameters rock support needs (shotcrete) and rock reinforcement needs
RQD/Jn and Jr/Ja are already being logged at a lot of mines, (bolts and anchors) for different sized openings, situated at
following Potvin and Matthews method, and the subsequent widely different depths, needed to be linked to the different
Modified Stability Graph, now in common use in Canadian deformations recorded. This was a different and apparently
and Australian mining for preliminary stope dimensioning.
The original technique has been well documented in the
literature. The method is based on a graph of the ratio of
Table 1. The remarkable complexity of the algebra for estimating

c and φ with Hoek-Brown based formulations (equations 4 and 6)
are contrasted with the simplicity of equations 3 and 5, derived by
‘splitting’ the existing Qc formula into two parts, as described in
Barton, 2002. (Qc = Q · σc /100, with σci expressed in MPa).
Figure 9. The integration of rock mass quality and seismic veloc-

ity, with adjustment (+ve) for depth (or stress level) and porosity
(−ve), and rock strength. Estimates of static deformation modulus
(right-hand column) should also be depth dependent. See example in
Figure 4c that slightly pre-dates the above method of Barton, 1995.
Table 2. Illustration of parameters CC (MPa) and FC◦ for a declining sequence of rock mass qualities, with simultaneously reducing σc
(MPa). Estimates of VP (km/s) and Em (GPa) are from Figure 9, whose derivation was described in Barton, 2002.
RQD Jn Jr Ja Jw SRF Q σc QC FC ◦ CC Vp Em
100 2 2 1 1 1 100 100 100 63 50 5.5 46

90 9 1 1 1 1 10 100 10 45 10 4.5 22
60 12 1.5 2 0.66 1 2.5 50 1.25 26 2.5 3.6 11
30 15 1 4 0.66 2.5 0.1 33 0.04 9 0.3 2.1 3.5

more challenging problem than addressed when Beniawski in construction time and give millions of city dwellers’ earlier
developed RMR (rock mass rating) one year earlier. Why the relief from traffic jams, due to five-year budgets extending to
Q-system (and Bieniawski’s RMR) were not developed long new projects.
before, may perhaps relate to the increasing use of more eco-
nomic single-shell solutions. These are epitomised the world
over in >2000 big caverns of 20 to 60 m span, for hydropower 5 IMPROVING ROCK MASSES BY PRE-GROUTING
and other purposes.
The six orders-of-magnitude Q-value scale that was grad- An important insight into the benefits of pre-grouting for
ually developed during six months of trial and error, proved tunnels was provided some years ago in Brazil, and involved
capable of answering the question posed, and has since been what one might call pre-grouting for dams, since a grout cur-
found to have simple links to joint, discontinuity and rock mass tain is also a form of ‘pre-grouting’. Using boreholes spaced
shear strength, deformation modulus, seismic velocity, seis- 4 to 8 m apart, Quadros and Correa Filho, 1995, measured the
mic attenuation, as well as tunnel and cavern support needs, three anisotropic permeability tensors of the local rock mass.
at depths from the surface to about 3 km depth. The guiding The same three holes were then used for local cement grout-
philosophy of support class selection has been to characterize ing, using just industrial cement and a small percentage of
the rock mass round by round, update prognoses, and use bolt- bentonite, as was normal in Brazil.
ing and shotcrete of the best quality, or occasional concrete This was followed by re-drilling of the three holes in nearly
linings (but mostly unlined in the case of headrace and deep- the same locations, and a new round of 3D permeability mea-
pressure tunnels), to ensure faster and cheaper construction. surements, using the three holes simultaneously as before,
Maximum water pressure head reached 1,000 m about 20 years supplied with multiple piezometers, and a movable quadruple-
ago, where minimum rock stress justified this more sophis- packer assembly for pumping out water. These were used in
ticated approach to rock engineering, than pressure-resistant rotation. The result was reduction of two permeability tensors
linings with steel. (to 1/18th and 1/11th of the magnitudes), and rotation of all
In contrast to single-shell tunnelling, the cost and schedule permeability tensors, implying joint-set sealing and reduced
of metro construction that suffers from the ‘short escalator anisotropy.
syndrome’, may cause just the top heading of the neces- As a result of these experiences, the writer argued
sary double-shell NATM tunnels, to average no better than (Barton, 2002) for the likelihood that most of the Q-parameters
10 m/week, due to all the unnecessary mixed face conditions. in a successfully pre-grouted rock mass, could change ‘for the
Near-surface construction may also cause settlement damage better’. In other words, due to partial or complete local seal-
to hundreds or thousands of houses, even far from the tunnel, ing of one or more joint sets, logically starting with the most
if pre-grouting is unsuccessful, or not even performed. One permeable set, there could be effects mostly better than the
may contrast the serious over-break seen in Figure 10, with very conservative ones shown here:
the lack of over-break seen in Figure 11, where systematic
pre-injection was performed throughout, resulting in
improved Q-values.
Mixed-face construction can be avoided by developing sta-
tions from underground in better quality rock, or at least in the Pre-grouting at 5 to 10 MPa pressures is necessary to get a
best geology within reasonable reach. This, of course, means good result, like penetration of the typical 1 to 5 litres of grout
London clay in London. The best available tunnelling horizon per m3 of rock in an assumed (mean) 5 to 6 m thick annulus
need not include much saprolite if the will is there to reach of pre-grouted rock around a 100 m2 tunnel, or approximately
an ‘economic’ geologic horizon. Escalator advertising makes 500–1000 litres of grout per meter of tunnel. The logarithmic
a 45 seconds stair journey more satisfying than 15 seconds; or linear pressure decay for Newtonian fluids, which should be
and this deeper ‘rock engineering’ approach may save years even stronger for Bingham fluids, guarantees about 50% loss
of pressure only 1 m from a given injection hole, while grout
is still flowing in the joints. When flow has stopped, pressure
Figure 10. Over-break of 5 m and near break-through into the sapro- Figure 11. Systematic pre-injection and NMT permanent support
lite, following lack of pre-injection, and due to a cost-driving shallow (one final S(fr) layer is needed), gives an average 20 m per week for
metro choice by owners that took many years to complete. the completed tunnel, due to Q-improved rock mass properties.

Figure 12. Pre-injection boreholes (40 to 60 in number, 25 to 30 m
long) intersect numerous joint planes. Injection pressure must be high
as pressure-drop is rapid away from each injection hole.
must be reduced, otherwise hydraulic jacking will not be well

controlled.
When an obstinate consultant prevents a contractor from
using high enough grouting pressures, and limits use of micro-
cement and additives, in a tight rock mass that nevertheless
needs pre-grouting, due to overlying clay and risk of settlement
damage, the result can be several of the following: aban-
doned settlement-damaged houses, leaking bolt holes, leaking
shotcrete, at least 100,000,000 US $ of post-grouting, and one
year of delays. There are examples with most of the above, in
several different countries.
The equation that can be derived from Snow’s cubic network
model, using non-American units, and the basis for Figure 14,
is as follows:
Figure 13. Application of a modified Snow, 1968 cubic network
approximation to grout-take estimation at a dam site. This method
(Barton et al., 1985) makes adjustment for larger physical apertures
where e = mean hydraulic aperture (mm), L = Lugeons, and (E) than conducting apertures (e), using small-scale joint rough-
S = mean spacing of water conductors (mm) (Barton, 2004). ness JRC0 . So empiricism and theory are combined to estimate
not only grout volume, but also appropriate size of grout particles,
The estimation of aperture (e) from Figure 14 and equation 7,
using another empirical-theoretical relation: E ≥ 3dmax , (theoretical)
depends on prior estimation of an average Lugeon value from or E ≥ 4d95 (empirical). The latter has also been shown to govern
a series of water injection tests (using constant packer spac- flow of blocks down ore passes in mines, where the mean physical
ing and continuous test stages. The all-important estimation joint aperture E is replaced by shaft diameter D.
of (S) is based on an assumed Poisson distribution of water
conductors in relation to the percentage of zero flow stages. A
final stage of the grout-take and grout-selection process is to
estimate the equivalent physical joint apertures (E) using the
JRC0 -based empirical relation from Figure 13, and compare
it with the ≥4 × d95 criterion for particle entry into the joints.
As a rough, and slightly conservative set of rules-of-thumb,

the mean physical apertures may need to be ≥0.05, 0.1 and
0.4 mm, for respectively ultra-fine, micro-, and industrial
cements. The approximate JRC0 values can be estimated from
the approximation: JRC0 ≈ 7Jr-3. This was derived from 1000 Figure 14. The estimates of mean hydraulic apertures (e) and
m of recent rail-tunnel core logging. mean spacing (S) of water-conducting joints are obtained from
Snow’s cubic network model, (Snow, 1968), in which permeability
is assumed to be provided by two of the three joint sets, or by weaker
6 DEFORMATION COMPONENTS FROM contributions from the three sets when the flow gradient rotates.
BARTON-BANDIS MODEL
JRC and JCS, Bandis made a remarkably thorough appli-
On the basis of the development of the empirical Barton- cation and extension of these parameters, during studies of
Choubey (1977) shear strength criterion (equation 2) involv- scale-effects on shear strength (using joint replicas), and also
ing the development of the joint characterization parameters applied these parameters to numerous normal-loading tests on
10

Figure 15. Blocks flowing down ore-passes, and micro-particles
trying to penetrate into tight joints ahead of tunnel faces, need more Figure 17. NGI’s UDEC-BB modelling of the three contrasting
space for more efficient flow, which can be created by reducing par- ‘rock masses’ (Figure 16). The ‘linearity’ of the load-deformation
ticle sizes and/or by increasing apertures locally, using high pressure for ‘horizontally’ loaded columnar basalt, matches the linearity of an
injection. Fortunately joint aperture E is greater than e, which makes in situ block test performed in the basalt Waste Isolation Project in
pre-injection easier than just considering the 3D network hydraulic Hanford, Washington, USA in the eighties. This model also assisted
aperture (e) trends shown in Figure 14. in recent major dam-abutment stiffness studies in China.
Figure 18. Compaction of the jointed-chalk reservoir of 300 m

thickness at 3 km depth caused a rapidly increasing sea-bed sub-
sidence that amounted to about 5 m when investigations began in
earnest in 1985–1987. Following simultaneous raising of all plat-
forms by 6 m in the late eighties, and provision of a 100 m diameter
protective wall, and final re-location of operations away from the cen-
tre of the 9 × 14 km field, compaction has at least doubled, despite
extensive sea-water injection for pressure maintenance (that also
caused inevitable weakening of the chalk).
Figure 16. Joint deformation components N and S from Bandis

et al., 1981 and 1983 applied to rock mass load-deformation under-
standing by Barton, 1986.
rock joints. His studies, and some of the author’s multi-block Figure 19. NGI and Itasca UDEC-BB ‘uniaxial-strain’ M-H mod-
studies, are summarised in Bandis et al., 1981 and 1983. Fig- elling of a vertical 1.5 × 1.5 m 2D-element of jointed Ekofisk chalk,
ure 16 shows a graphic summary of the joint closure (N) and which had a porosity of 40%. Input data was obtained from JRC and
shear components (S) of deformation, which were utilised in JCS characterization of numerous joint samples that operator Philips
Barton, 1986 to explain the measured concave, linear and recovered as part of the extensive exploration process. A vertical total
convex load-deformation trends of three different styles of stress of 62 MPa was applied, and internal pore pressure was reduced
rock mass. The dominance or combination of the N and S from 48 to 28 MPa, causing shrinkage of the porous matrix, and con-
components are shown in the lower half of Figure 16, and sequent space for down-dip joint shearing, despite one-dimensional
confirmatory numerical modelling with UDEC-BB is shown strain. Barton et al., 1986.
in Figure 17.
from this giant off-shore field projected to last for 80 years,
there was naturally concern when a supply-boat captain appar-
7 MODELLING A COMPACTING SUB-SEA ently reported that a reduced number of wave-baffle holes (to
RESERVOIR protect the temporary oil storage tank) appeared to be above
sea levels in 1985, than at the beginning of production. These
The Ekofisk reservoir in the North Sea was discovered dur- holes are spaced at 2 m c/c. (See effect of 13 years time interval
ing the very last attempts to find oil in this ‘unpromising’ in Figure 18).
petroleum province, at the end of the 1960’s. With reservoir The down-dip shearing that can occur despite one-
development from about 1971, and large-scale production dimensional strain, is a fundamental necessity for the
11

Figure 20. Rough tension fractures sheared as measured, give an
exaggerated clue to the possibilities of maintained production despite
compaction, in a reservoir with too low matrix permeability. For
shearing to occur, steep dip is of course required. Note voids and
contacts, and their contrary rotation. Barton, 1971.
Figure 22. Wide-azimuth VSP exploration with P-wave sources,

with conversion to pseudo-shear wave components. (Stenin et al.,
2002). The assumed major-stress-aligned micro-cracks or single set
of reservoir joints may however be closed by too high effective normal
stress, and insufficient JRC and JCS (Barton, 2006), unless the more
likely scenario of steeply dipping sets under shear is invoked, as for
Figure 21. Left: An example of the steeply dipping conjugate joint- instance in Figures 19 and 23.
ing and its significant roughness. Right: One of NGI’s axi-symmetric
10 km × 3 km subsidence studies with a ‘bedded’ UDEC model,
showing a 4.8 m/5.5 m S/C ratio. Shearing events occur close to the
reservoir and in the ‘over-burden’, and confirm one of the reasons for
casing damage. High ratios of S/C, such as 0.85 or more, apparently
matched estimates of S/C, that Philips obtained using radioactive
targets fired into casing perforations.
continued conductivity of the joints, as the matrix is of low

permeability, and cannot otherwise be drained. Of course
‘rubble-ization’of the highest porosity chalk, caused by water-
weakening and over-loading, is a further contributor to the
excellent productivity of this huge reservoir.
There are several interesting observations to be added
here, concerning the foregoing examples of discontinuum
modelling at widely different scales. During exploration,
slickensided joints in the chalk were not observed. According
to Philip’s geologist (H. Farrell, pers. comm.), slickensiding
Figure 23. A scenario involving maintenance of permeability
was observed in some much later cored holes connected with through shearing and slight dilation (exaggerated as in Figure 20)
the water-injection operations, after 1985. Production was is more consistent with analysis of deep-well conductors and
causing joint shearing, and is presently a seemingly ignored non-conductors, as emphasised by Zoback and co-workers at
part of 4D interpretation (Barton 2006). Stanford.
NGI’s discontinuum approach to the modelling of Ekofisk
compaction and subsidence, produced much higher ratios of petroleum and service companies, and a need to move beyond
S/C (subsidence divided by compaction) than that achieved the belief that geomechanics is something that stopped devel-
by another team, who used 3D FEM continuum modelling oping after Mohr Coulomb. A case in point is geophysics
with the large Dynaflow code. However, low values of shear expectation of stress-aligned reservoir fractures (micro-cracks
stiffness appropriate to large scale bedding-planes, such as or one set of joints) that are the assumed source of shear-wave
0.01 MPa/mm were used (e.g. Barton et al., 1985), in con- splitting into fast and slow components. (See Figure 22 exam-
trast to typical ‘lab values’ of 1.0 MPa/mm. It is understood ple). There may in reality need to be two joint sets involved,
that artificially low values of shear modulus were needed to each under shear stress. 4D temporal rotation of fast S-wave
achieve better match to measurements, when modelling as a axes, suggest this.
3D continuum with Dynaflow. It is easy to show that unless joints in the reservoir are held
open by mineral fillings, or unless for instance, a distinctly
8 DISCIPLINE DISCREPANCIES IN EXPLORATION strong limestone with rough joints is involved, there will be
insufficient apertures for production, at reservoir effective
While performing a wide-reaching literature search on geo- normal stress levels of tens of MPa. Barton-Bandis mod-
physics reservoir-and-exploration topics (Barton, 2006), the elling with input data assumptions such as JCS0 = 25 MPa
writer discovered a seemingly serious discrepancy between and JRC0 = 5, with moderate in situ block sizes, suggest
geophysicist beliefs and the presumed realities of rock closed stress-closure curves, and indicate that permeability
mechanics joint behaviour. The following example perhaps can only be generated (or maintained) if shearing is occur-
emphasises the need for real cross-disciplinary teams in ring, as in Figures 19 and 22. If on the other hand, JRC0 = 10
12

10 CONCLUSIONS
Jointed and anisotropic water-bearing rock masses most

frequently represent the reality of engineering in rock. The
variability of weathering, the presence of clay, and stress mag-
nitudes and permeabilities that usually vary spatially and with
azimuth, represent challenges of quantification that require
insight, some use of empirical methods like the Q-histogram
method, and an absence of decimal places, which give a false
sense of actual levels of understanding.
Those who restrict their rock mechanics experience to con-
tinuum modeling, inevitably miss many exciting insights in the
rewarding field of rock engineering. It surely makes no real
sense to collect data from discontinuous anisotropic media,
and then convert it into isotropic continuum format.
Coloured appendices with their exaggerated ‘plastic zones’
Figure 24. Critical state line for intact rock at extreme stress lev- may be misleading many clients, in view of the fact that shear
els (Barton, 1976), suggested the limiting condition σ1 = 3σ3 . Singh strength is seldom ‘c plus σn tan ϕ’, whether in linear or non-
et al., 2011 show that most often, the critical value of σ3 ≈ σc . This linear form. So many consultant offices in so many countries
seems to be logical, and helps to define a strength-envelope curvature are producing coloured appendices that one can wonder if
which correctly matches test data at all confining pressure levels, and the colour is considered the most important sign of good-
improves upon all existing strength criteria. quality analyses. It may be wise for more young people to be
educated for discontinuum modelling. This does require more
understanding of rock mechanics.
REFERENCES
Addis, M.A., Barton, N., Bandis, S.C. & Henry, J.P. 1990. Laboratory
studies on the stability of vertical and deviated boreholes. 65thAnn.
Tech. Conf. & Exhibition of the Society of Petroleum Engineers,
New Orleans.
Bandis, S., Lumsden, A. & Barton, N. 1981. Experimental studies
of scale effects on the shear behaviour of rock joints. Int. J. Rock
Figure 25. The tilt test result can be extrapolated from 0.001 to Mech. & Min. Sci. 18, 1–21.
10 MPa, and can be performed on samples of 10 cm to 1 m in size. Bandis, S., Lumsden, A.C. & Barton, N. 1983. Fundamentals of rock
The same method has also been used on 5 m long as-built rockfills. joint deformation. Int. J. Rock Mech. & Min. Sci. 20: 6: 249–268.
Barton, N. 1971. A model study of the behaviour of steep excavated
rock slopes. Ph.D. Thesis, Univ. of London.
Barton, N., Lien, R. & Lunde, J. 1974. Engineering classification of
and JCS0 = 50 MPa, some aperture can be maintained even rock masses for the design of tunnel support. Rock Mechanics, 6,
without the benefits of shearing. 4: 189–236.
Barton, N. 1976. The shear strength of rock and rock joints. Int. Jour.
Rock Mech. Min. Sci. 13, 9: 255–279.
Barton, N. & Choubey, V. 1977. The shear strength of rock joints in
9 SHEAR STRENGTH AT EXTREME STRESS LEVELS theory and practice. Rock Mechanics, Vienna: Springer; 1/2:1–54.
Barton, N. & Hansteen, H. 1979. Very large span openings at shal-
In 1976 the writer proposed a ‘critical state’ concept for the low depth: Deformation magnitudes from jointed models and
shear strength of intact rock at high stress, which involved F.E. analysis, 4th Rapid Excavation and Tunnelling Conference,
both the expectation and the actual horizontal orientation of RETC; Atlanta Georgia, Vol. 2: 1131–1353. A.C. Maevis and
W.A. Hustrulid (Eds), American Inst. of Min. & Metall, &
the Mohr strength envelope. This condition was shown or sug- Petroleum Engineers, Inc. New York, New York, 1979.
gested by tectonophysicists’ triaxial tests of various rocks at Barton, N. & Bandis, S. 1982. Effects of block size on the shear
extreme confinement. Recently, this concept has been applied behaviour of jointed rock. Keynote, 23rd US Symposium on Rock
to better define the curvature of intact rock strength envelopes. Mechanics, Berkeley, California.
A few tests at low confining pressures provide all the data Barton, N., Bandis, S. & Bakhtar, K. 1985. Strength, deformation
needed for extrapolation to high levels of confinement. The and conductivity coupling of rock joints. Int. J. Rock Mech. &
elegant Singh et al., 2011 criterion heralds a new era in rock Min. Sci.. 22, 3, 121–140.
mechanics understanding. Barton, N. 1986. Deformation phenomena in jointed rock. 8th Laurits
Since a blend of theory and empiricism has been promised Bjerrum Memorial Lecture, Oslo. Publ. in Geotechnique, Vol. 36,
in the title, with possible application to problem solving in 2: 147–167.
Barton, N., Hårvik, L., Christiansson, M., Bandis, S., Makurat, A.,
rock engineering, the final figure to be presented will be of Chryssanthakis, P. & Vik, G. 1986. Rock mechanics modelling
the tilt testing concept. This is at the other end of an extreme of the Ekofisk reservoir subsidence. Proc. 27th US Rock Mech.
stress range.Tilt tests on joints of 10 cm and up to 1 m length, at Symp., Univ. of Alabama.
normal-stress-at-failure of 0.001 to 0.01 MPa, have provided Barton, N. & Grimstad, E. 1994. The Q-system following twenty
JRC0 and JRCn values for defining strength envelopes three years of application in NMT support selection. 43rd Geomechanic
to four orders of magnitude higher. Colloquy, Salzburg. Felsbau, 6/94. pp. 428–436.
13

Barton, N. 1995. The influence of joint properties in modelling Cundall, P.A., Voegele, M.D. and Fairhurst, C. 1977. Computer-
jointed rock masses. Keynote, 8th ISRM Congress, Tokyo, 3: ized design of rock slopes using interactive graphics for the
1023–1032, Balkema, Rotterdam. input and output of geometrical data. Proc. of 16th US Symp. on
Barton, N. 2002. Some new Q-value correlations to assist in site Rock Mechanics, Minnesota 1975, Vol. 1, 5–14, C. Fairhurst and
characterization and tunnel design. Int. J. Rock Mech. & Min. Sci., S.L. Crouch (Eds). American Soc. of Civil Engineers, New York.
39: 185–216. Hajiabdolmajid,V., C. D. Martin and P. K. Kaiser,. Modelling brittle
Barton, N. 2004. The theory behind high pressure grouting. failure. Proc. 4th North American Rock Mechanics Symposium,
Tunnels & Tunnelling, Sept. 28–30, Oct. 33–35. NARMS 2000 Seattle J. Girard, M. Liebman, C. Breeds and T. Doe
Barton, N. 2006. Rock Quality, Seismic Velocity, Attenuation and (Eds), 991–998. A.A. Balkema, Rotterdam.
Anisotropy. Taylor & Francis, UK & Netherlands, 729 p. Müller, L. 1964. The rock slide in the Vajont Valley. Felsmechanik
Barton, N. 2008.TBM tunnel construction in difficult ground. Invited und Ing. Geol., 2, 148–212.
lecture. 1st South American Symposium on Rock Excavations- Müller, L. 1966. The progressive failure in jointed media. Proc. of
ISSAER- Bogota, Colombia- Bulletin of the Colombian Geotech- ISRM Cong., Lisbon, 3.74, 679–686.
nical Society No. 12, 17–32, October, 2008 Quadros, E. and Correa Filho, D. 1995. Grouting efficiency using
Barton, N. and S.K.Pandey, 2011. Numerical modelling of two stop- directional (3-D) hydraulic tests in Pirapora Dam, Brazil. Proc. of
ing methods in two Indian mines using degradation of c and 8th ISRM Cong., Tokyo, Fujii (Ed), Balkema, 823–826.
mobilization of ϕ based on Q-parameters. Int. J. Rock Mech & Shen, B. & Barton, N. 1997. The disturbed zone around tunnels in
Min. Sci. D-10-00157R1. jointed rock masses. Technical Note, Int. J. Rock Mech. Min. Sci,
Bhasin, R.K. & Barton, N. 1997. A comparison of Barton-Bandis 34, 1: 117–125.
joint constitutive model with the Mohr-Coulomb model using Singh, M., A. Raj & B. Singh, 2011. Modified Mohr-Coulomb Crite-
UDEC. Proc. of 1st Asian Rock Mech. Symp. ARMS ’97, Seoul, rion for Non-linear Triaxial and Polyaxial Strength of Intact Rocks.
Korea, pp. 413–420. Int. J. Rock Mech. & Min. Sci. -D-10-00044R1.
Cundall, P.A. 1971. A computer model for simulating progressive, Stenin, V.P., Kasimov, A.N. & Tikhonov, A.A. 2002. Cracked layers
large scale movements in blocky rock systems. Proc. of ISRM characterization using far offset VSP. 10th Int.Workshop of Seismic
Symp., ‘Rock Fractures’, Nancy, Vol. 1, Paper II-8. Anisotropy, Tutzing, Germany.
14

Rocha Medal lecture

Reduction of blast-induced vibration in tunnelling using barrier holes and air-deck
Dohyun Park
Underground Space Department, Korea Institute of Geoscience and Mineral Resources, Daejeon, South Korea
ABSTRACT: This paper presents two different kinds of blasting techniques for reducing blast-induced vibration in tunnelling,
which are applicable according to the location of nearby structures: line-drilling; and bottom-air-deck. Numerical investiga-
tions through a non-linear hydrocode, which can numerically model chemical explosives, were carried out in order to obtain
quantitative information on the effectiveness of both methods, and then the numerical results were verified through field
experiments. The numerical and experimental results agreed well with each other, and indicated that both methods effectively
reduced blast-induced vibration according to the location of structures. From the present numerical and experimental study,
it was revealed that when both methods are applied to tunnelling sites, the vibration-reduction effects, that were predicted
through the numerical study, can be expected, and thus based on the numerical results, design conditions of both methods for
the reduction of blast-induced vibration in tunnelling were proposed by introducing the concept of factor of safety.
Subject: Rock breakage and excavation techniques
Keywords: Blasting, Tunnelling, Dynamics, Numerical modelling, Field measurements
1 INTRODUCTION
Since a tunnel is excavated in one direction, it is important

to reduce blast-induced vibration, considering the location of
nearby structures. For example, structures can be located in the
front of and in the rear of a vertical plane, including the tunnel
face, as shown in Figure 1(a). Also, at a point in time, the struc-
ture in Figure 1(b) is located in the front of S-1 plane, but as we
excavate the tunnel, the structure is located in the rear of S-2
plane. Therefore, effective vibration-reduction methods appli-
cable according to nearby structures are needed. However, up
to now, there have not been many researches on the reduction
of blast vibration, considering the location of structures. The
present study proposed line-drilling method and an air-deck
method for reducing blast vibration in the transverse direction
of a tunnel, i.e. in the rear of a vertical plane including the
tunnel face, as shown in Fig. 1(a)) and in the direction of tun-
neling, i.e. in the front of the vertical plane, respectively. The
air-deck method (hereafter referred as the ‘bottom-air-deck
method’) involves the use of a thin paper-tube at the bottom
of a blasthole.
A quantitative assessment of the effectiveness of both
methods was numerically and experimentally carried out. The
parameters of line-drilling were the spacing and diameter of Figure 1. Reduction of blast-induced vibration in tunnelling con-
drill holes, the distance between blasthole and line-drilling, sidering the location of nearby structures.
and the number of rows of drill holes, including their arrange-
ment, with the parameter of bottom-air-deck, being the length
of the air-deck at the bottoms of blastholes.
In the numerical investigations, a non-linear hydrocode,
AUTODYN, which can numerically model explosives, was From the comparison, it was concluded that when both
used to realistically simulate blasting problems. From the methods are applied to tunnelling sites, the vibration-
numerical investigations of both methods, correlation equa- reduction effects, which were predicted through the numerical
tions of vibration-reduction effect, which give quantitative study, can be expected. On the basis of the conclusion, design
information on the drilling of barrier holes and on the conditions for obtaining vibration-reduction effects of 10–
bottom-air-deck ratio, were derived. In order to verify the 20% and 10–25% in the transverse direction of a tunnel and
numerical results, field experiments were carried out, and in the direction of tunnelling, respectively, were proposed by
then good agreement was found from a comparison between introducing the concept of ‘factor of safety’ for the results that
the numerical and experimental results. were predicted through the numerical investigations.
17

2 BASIC CONSIDERATIONS
2.1 Screening of blast-induced waves for reducing ground

vibration in the transverse direction of a tunnel
Screening methods can be divided into two categories, as
shown in Figure 2: (1) active isolation; and (2) passive isola-
tion (Woods 1968). With active isolation, a barrier is located
at points close to or surrounding the source of vibration. Con-
versely, with passive isolation, a barrier is located at points
remote from the source of vibration, but near a site where the
Figure 2. Active and passive isolations.
amplitude of vibration must be reduced.
Since passive isolation is constructed in the vicinity of a
structure, the construction work can give rise to additional
noise and vibration prior to the primary construction oper-
ations. The noise and vibration can also cause secondary
pollution from both environmental and structural views. In
addition, when passive isolation is installed in the vicinity of
a structure during blasting work, it is necessary for the original
ground state to be restored, following completion of construc-
tion, as the isolation will no longer be required. In conclusion,
it is believed that compared with active isolation, passive iso-
lation is not practically effective in the screening of blast
induced waves from various constructional features. There- Figure 3. (a) Charge pattern of bottom-air-deck method, and (b)
fore, only the problem of active isolation was investigated in paper-tube used for bottom-air-deck.
the present study.
There are three kinds of screening methods for reducing same blasthole, can significantly reduce the charge weight per
blast-induced vibration in the transverse direction of a tun- delay. However, since the deck-charge requires two or more
nel: (1) a barrier wall; (2) pre-splitting; and (3) barrier holes. detonators in the same blasthole, it is difficult to load blast-
According to Park et al. (2009), a barrier wall (trench) is the holes and a detonator with an accurate firing time, such as an
most effective method, in that it is capable of relatively com- electronic detonator, is needed. In addition, due to the separa-
plete isolation, but does requires great expense. In addition, the tion of explosives in the same blasthole, dead pressing is most
excavation of a barrier wall at tunnelling sites is not practical likely to occur. Therefore, the deck-charge is not usually suit-
from a constructional point of view. Pre-splitting can reduce able for reducing blast-induced vibration in the direction of
the amount of drilling, when compared with the other meth- tunnelling. Finally, since low-velocity explosives have a low
ods, but in the case that the amount of explosive for pre-split rate of reaction and thus low pressures, the low-velocity explo-
blasting is not accurate, cracks for screening blast-induced sives are more suitable for quarrying rather than tunnelling
vibration cannot be generated or the amplitude of vibration, where free faces are made by cut blasting.
induced by pre-splitting, can be greater than that induced by In the present study, a bottom-air-deck method was newly
primary blasting. Barrier holes, i.e. line-drilling can eliminate investigated. According to Marchenko (1982), when a charge
the problems of pre-splitting, because line-drilled holes are is blasted with air gaps, the products of explosion can no
not loaded and can reduce the amount of drilling, compared longer generate a powerful shock wave in the medium, because
with a trench. In addition, if there is quantitative information they can expand into the air gap after the charge detonates,
on the drilling of barrier holes, line-drilling is the most prac- as a result of which their pressure decreases. On the basis
tical method, compared with the other methods. Therefore, of such a principle, the present study has assumed that the
the present study proposed line-drilling method for reducing bottom-air-deck can suppress blast vibration in the direction
blast-induced vibration in the transverse direction of a tunnel. of tunnelling. The essence of this idea is that by installing the
bottom-air-deck, the initial pressure of the products of explo-
sion, which is transmitted to the surrounding ground through
2.2 Suppression of blast-induced waves for reducing ground the bottoms of blastholes, decreases. Further, the distance
vibration in the direction of tunnelling from the blasting source to the point of monitoring increases,
There are several conventional methods for suppressing blast- and thus one can reduce blast vibration in the direction of
induced vibration in the direction of tunnelling: reduction of tunnelling. Figure 3 shows the charge pattern of the bottom-
charge weight per delay by delay detonators or by reducing air-deck method and the paper-tube for bottom-air-deck, of
the length of advance per round; creation of a new internal which the inner diameter and thickness were 31.8 mm and
free face through a hole of large bore in a cut area; a deck- 1.1 mm, respectively.
charge; and use of low-velocity explosives. The reduction of
the charge weight per delay, especially through the reduction
3 PARAMETRIC NUMERICAL STUDY
in the advance per round, necessarily causes time and cost
overruns in construction. Further, special equipment, such as
3.1 Numerical modeling using the non-linear hydrocode
a large-bore drilling machine, is needed for creating a new
internal free face in a cut area. Therefore, these methods are The present study used the non-linear hydrocode, AUTODYN
neither practical nor cost-effective from a constructional point which is suited for simulating a high-velocity impact, espe-
of view. The deck-charge, whereby explosives are separated cially an explosion. The severe distortion of Lagrange grids in
from other explosives by inert material or air cushions in the the finite element method (FEM) gives rise to errors as well
18

as minuscule time-steps. On the other hand, SPH (smoothed
particle hydrodynamics), which is a mesh-free Lagrangian
technique, has the advantage of computing a large deformation
problem, owing to the absence of a real mesh. In consid-
eration of the above characteristics, the present study used
SPH for modeling the materials in large-deformation regions
and modeled low-deformation materials using elements of
Lagrangian FEM.
The RHT (Riedel, Hiermaier, and Thoma) material model
(Riedel et al. 1999, Riedel 2000), which is particularly use-
ful for modeling the dynamic loading of brittle materials, was
used for modeling the ground. Since the model is formulated
in such a manner that input can be scaled with the uncon-
fined compressive strength, the ground can be modeled by its
unconfined compressive strength. According to Bieniawski
(1979), the unconfined compressive strength under the fair-
ground condition is within the range 50 to 100 MPa, and thus
the present study applied an unconfined compressive strength
of 75 MPa to the RHT model for modeling the fair ground.
The present study used TNT (trinitrotoluene) as an explo-
sive. The JWL (Jones-Wilkins-Lee) equation of state (EOS)
was used for modeling TNT (Eq. (1)), and the parameters
proposed by Lee et al. (1973) were used.
Figure 4. (a) Schematic of numerical analysis, and (b) configuration

of the numerical model. G is the gauge.
where P = pressure; V = relative volume; e = internal energy;

and A, B, R1 , R2 , and ω = constants determined by the cylin-
drical expansion test.
In order to model the stemming material, the present study
used the mechanical properties of non-cohesive and loose sand
that are generally used in tunnel blasting. The N-value of the
loose sand, which is obtained from the standard penetration
test, was assumed to be 4–10 (Dunham 1954, Hunt 1984).
Accordingly, the density, bulk and shear moduli, and angle of
internal friction of the loose sand were 1,800 kg/m3 , 33.3 MPa,
11.1 MPa, and 25◦ , respectively.
3.2 Numerical investigations of the effectiveness of

line-drilling method
The vibration-reduction effect (amplitude reduction factor)
was assessed by comparing the values of the peak vector sum
(PVS) between two cases, the one with a vibration-reduction Figure 5. Relationship between the vibration-reduction effect and
method and the other without (Eq. (2)). D2 /(S − D).
vibration-reduction effect and drilling condition was derived

by regression analysis (Fig. 5).
where ARF (%) = amplitude reduction factor; Vno-red and

Vred = PVS values (mm/s) of the two cases (without and
with a vibration-reduction method, respectively); and D
(m) = distance from blasting source to monitoring point. where ARF (%) = amplitude reduction factor; D (mm) =
diameter of drill holes; and S (mm) = spacing of drill holes.
3.2.1 Effect of the spacing and the diameter of drill holes
This analysis focused on assessing the effect of the spacing 3.2.2 Effect of the number of rows of drill holes and their
and the diameter of drill holes on vibration-reduction. The arrangement
applied diameters of drill holes were 50, 100, and 150 mm, This analysis focused on assessing the effect of the number
and the net spacings for each diameter were 50, 150, 250, of rows of drill holes and their arrangement on vibration-
and 350 mm. The distance between line-drilling and blast- reduction. The region of this analysis was the same as that
hole was 300 mm. Figure 4 shows a schematic of numerical applied in Subsection 3.2.1. A double row and triple row of
analysis, and the configuration of the numerical model. From drill holes of 50 mm in diameter were applied and the rows
the analysis, the following correlation equation between the were arranged in a parallel and zigzag manner. The spacing
19

Table 1. Summary of vibration-reduction effects for the number of blasthole of 300 mm and Equation (5) was obtained by com-
rows of drill holes and their arrangement. bining the two equations. This equation was used for verifying
the results of field experiments and for proposing the design
Case LD-200-DP LD-200-DZ LD-200-TZ LD-100 conditions of line-drilling for the reduction of blast-induced
ARF (%) 27.78 30.61 33.33 33.51
vibration in the transverse direction of a tunnel.
where ARF (%) = amplitude reduction factor; D (mm) =

diameter of drill holes; S (mm) = spacing of drill holes; and
DLD−BH (mm) = distance between line-drilling and blasthole.
3.3 Numerical investigations of the effectiveness of

bottom-air-deck method
3.3.1 Effect of bottom-air-deck ratio on vibration-reduction
The present study applied the following conditions, which
are commonly practiced for tunnel blasting under fair-ground
conditions: the diameter and length of a blasthole were 45 mm
and 2.2 m, respectively; and the amount of emulsion-type
explosives was 1.0 kg. An empirical equation of the relative
weight strength (RWS) of explosives, which was proposed by
Figure 6. Relationship between the vibration-reduction effect and Jimeno et al. (1995), was used to convert the amount of the
the distance between line-drilling and blasthole. emulsion-type explosive (1.0 kg) to that of theTNT (0.782 kg),
and thus the length and diameter of 0.782 kg TNT were 1.0 m
of the drill holes and the rows was 200 mm and the distance and 25 mm, respectively.
between line-drilling and blasthole was 300 mm.Table 1 shows Figure 7 shows a schematic diagram of numerical analysis
the vibration-reduction effects for the number of rows of drill for a single blasthole under axis-symmetry condition and the
holes and their arrangement. The vibration-reduction effect configuration of the numerical model. Figure 7(a) shows the
of a double row of drill holes with a zigzag configuration half of the cross-section cut through the diameter of the cylin-
(LD-200-DZ) was about nine-tenths of that of a triple row der. The applied lengths of bottom-air-deck were 100, 200,
of drill holes with a zigzag configuration (LD-200-TZ). The 300, 400, and 500 mm, and thus the bottom-air-deck ratios
vibration-reduction effect of a triple row of drill holes was were 0.1, 0.2, 0.3, 0.4, and 0.5, respectively. The mechanical
almost the same as that of a single row of drill holes (LD-100) strength of the paper-tube used for bottom-air-deck is negligi-
with half of the spacing of a triple row of drill holes. ble when compared with the detonation pressure of the TNT
and the strength of the ground. Thus, the bottom-air-deck was
3.2.3 Effect of the distance between line-drilling and modeled by leaving it an empty space, as shown in Figure 7.
blasthole From the analysis, the vibration-reduction effect was found
This analysis was carried out to investigate the effect of to increase as the bottom-air-deck ratio increased. The follow-
the distance between line-drilling and blasthole on vibration- ing equation was obtained using regression analysis (Fig. 8).
reduction. A double row of drill holes was applied and the
spacing of rows was 200 mm. The diameter and spacing of drill
holes were 50 mm and 200 mm, respectively. Figure 6 shows
the relationship between the vibration-reduction effect and the
distance between line-drilling and blasthole. From the analy- where ARF (%) = amplitude reduction factor; and Ra =
sis, the vibration-reduction effect was found to decrease with bottom-air-deck ratio.
increasing the distance between line-drilling and blasthole.
Consequently, similar to the result of Woods (1968), active 3.3.2 Effect of the length of bottom-air-deck on excavation
isolation was more effective than passive isolation. From the efficiency
analysis, the following correlation equation was derived. Figure 9 shows the configuration of numerical model. The
modeling conditions of the blasthole and the TNT were
the same as those applied in Subsection 3.3.1. According
to Jimeno et al. (1995), the specific charge, required for
excavating medium-strength rocks in bench blasting, was
where ARF (%) = amplitude reduction factor; DLD−BH (m) = 0.3–0.6 kg/m3 . Thus, a specific charge of 0.45 kg/m3 , which
distance between line-drilling and blasthole. is the mid-point value of the required specific charge, was
3.2.4 Quantitative information on the drilling of barrier used. Correspondingly, the burden, used in the numerical
holes analysis, was 500 mm. The excavation efficiency was assessed
The two correlation equations, Equations (3) and (4) can be by Equation (7).
combined to yield quantitative information on the drilling of
barrier holes. In order to combine the two equations, Equation
(4) was normalized for a distance between line-drilling and
20

Figure 9. Configuration of the numerical model.
Figure 10. Advances in the cases with and without the bottom-air-
deck. (a) length of the bottom-air-deck = 0 mm, and (b) length of the
bottom-air-deck = 300 mm.
where EE (%) = excavation efficiency through blasting; and

Ladvance and Ldrilling = advance (m) and the drilling length (m)
of blastholes, respectively.
In real blasting practice, damaged zone, induced by blasting,
is usually removed by ‘scaling’, and consequently the advance
includes this portion of length. However, in the numerical
model, the scaling process could not be technically carried
out, since there is no numerical criterion to handle this prob-
lem. Thus, for a conservative estimation, the advance was
estimated by measuring the shortest distance between the top
of a blasthole and the ground finally left after blasting. Figure
10 presents the advances in the two cases with and without the
bottom-air-deck. In the case when the length of bottom-air-
deck was 300 mm, the reduction of the advance was calculated
Figure 7. (a) Schematic of numerical analysis, and (b) configuration to be 286 mm (=2,114 mm – 1,828 mm). In the same way, the
of the numerical model. G is the gauge. reductions of the advance for bottom-air-deck at lengths of
100, 200, 400, and 500 mm were 70, 221, 404, and 497 mm,
respectively. From the results, the reduction in the advance
was found to be approximately proportional to the length of
the bottom-air-deck.
4 FIELD EXPERIMENTAL STUDY
The vibration-reduction effect was assessed by comparing the

values of PVS between two cases without and with a vibration-
reduction method, as calculated through ground-attenuation
equations of the respective cases. It is important to note that
since a ground-attenuation equation is obtained from regres-
sion analysis, it is only valid within the range of the sample
data. Therefore, by using Equation (8), the vibration-reduction
effect was assessed within the range of overlap of the sample
data for the two cases.
Figure 8. Relationship between the vibration-reduction effect and where ARF (%) = amplitude reduction factor; Vno-red and
bottom-air-deck ratio. Vred = PVS values (mm/s) of the two cases (without and
21

Figure 11. Schematic of experiment layout in the field.
with a vibration-reduction method, respectively); and SD1 and

SD2 = minimum and maximum scaled distance for assessing
the vibration-reduction effect, respectively. Figure 12. Peak vector sum vs. scaled distance.
4.1 Field experiments of line-drilling method

4.1.1 Single-hole experiment
Single-hole experiment on a single row of drill holes with a
300 mm spacing was carried out atYeongwol limestone quarry
in Korea. Figure 11 shows a schematic of experiment layout.
The diameter and length of drill holes and blastholes were
45 mm and 2.2 m, respectively. An emulsion-type explosive of
0.625 kg/hole was used. A total of 12 single-hole experiments
were carried out: six experiments without line-drilling (LD-
NO); and six experiments with the line-drilling (LD-300).
Figure 12 shows the ground vibration results of the two cases
and the derived ground-attenuation equations (Eqs. (9)–(10)).
Ground attenuation equations:
Figure 13. Plan view of the field experiment site.
where VLD-NO and VLD-300 = PVS values (mm/s) of the two

cases; and SD2 = square-root scaled distance (m/kg1/2 ).
The vibration-reduction effect was calculated to be 16.5%.
The dark area indicated in Figure 12 denotes the range of
the scaled distance for investigating the vibration-reduction
effect. The vibration-reduction effect, predicted by Equation
(5), was 5.7% under the conditions of this field experiment.
The vibration-reduction effect of the field experiment was
greater than that predicted through the numerical analysis.
This difference was believed to be due to the effect of discon-
tinuities in the rock mass and the irregularity in the profile of
burden of the two cases.
4.1.2 Full-scale experiment Figure 14. Peak vector sum vs. scaled distance.
Full-scale experiment on a double row of drill holes with a
200 mm spacing was carried out at Duhak railway tunnel in Ground attenuation equations:
Korea. Figure 13 shows a plan view of the field experiment
site. The ground condition, based on the RMR (rock mass rat-
ing), was fair. The diameter of blastholes and drill holes was
45 mm, with the lengths of blastholes and drill holes being 1.1–
2.2 m and 10.0 m, respectively. A maximum charge weight per
delay was 0.4–0.9 kg/delay. A total of three full-scale experi- where VLD-NO and VLD-200-DZ = PVS values (mm/s) of the
ments were carried out. Figure 14 shows the ground vibration cases without line-drilling and on a double row of drill holes
results of the two cases and the derived ground-attenuation with a 200 mm spacing in a zigzag manner, respectively; and
equations (Eqs. (11)–(12)). SD2 = square-root scaled distance (m/kg1/2 ).
22

Figure 15. Schematic of experimental layout in the field.
The vibration-reduction effect in the case on the double

row of drill holes was calculated to be 46.0%. The vibration-
reduction effect, predicted by Equation (5), was 14.6% under
the conditions of this field experiment. The result of the field
experiment was greater than that predicted through the numer-
ical study. This difference was believed to be due to the effect
of discontinuities in the rock mass as well as the difference of
the ground condition between the two cases; the topography
showed that the soil layer in the regions with line-drilling was
thicker than that in the regions without the line-drilling.
4.2 Field experiments of bottom-air-deck method

4.2.1 Effect of bottom-air-deck ratio on vibration-reduction
4.2.1.1 Single-hole experiment
Single-hole experiment was carried out at a granite quarry Figure 16. Peak vector sum vs. scaled distance. Ra is the bottom-
in Korea. Figure 15 shows a schematic of the experimental air-deck ratio.
layout for bottom-air-deck at the site. A total of three single-
The vibration-reduction effects for bottom-air-deck ratios
hole experiments on bottom-air-decks at lengths of 100, 200,
of 0.169, 0.339, and 0.508 were calculated to be 28.8, 38.4,
and 300 mm were carried out.
and 59.8%, respectively. The corresponding effects, predicted
The blastholes were drilled in the horizontal direction.
by Equation (6), were 17.0, 27.2, and 35.8%. The vibration-
The spacing and the burden of blastholes were 1.0–1.5 m
reduction effects calculated in this experiment were greater
and 1.0 m, respectively, with the blastholes being 2.2 m in
than those predicted in the numerical study. The difference
length and 45 mm in diameter. An emulsion-type explosive
between the numerical and experimental studies was caused
of 0.5 kg/hole (the explosive length = 590 mm) was used, and
by using the existing ground-attenuation equation, which was
thus the bottom-air-deck ratios for the three bottom-air-deck
derived under blasting conditions that were different from the
lengths were 0.169, 0.339, and 0.508, respectively.
cases with bottom-air-decks. However, similar to the results
The vibration-reduction effect was assessed by using the
of the numerical study, as the bottom-air-deck ratio increased,
existing ground-attenuation equation (Eq. (13)), which was
the vibration-reduction effect tended to increase.
obtained during production blasting at the site. Figure 16
In addition, the normalized vibration-reduction effects in
shows the ground-vibration results and the derived ground-
the cases of bottom-air-deck ratios of 0.169, 0.339, and 0.508
attenuation equations (Eqs. (14)–(16)).
were 17.0, 22.6, and 35.2%, respectively. The normaliza-
Case without bottom-air-deck:
tion assumed that for a bottom-air-deck ratio of 0.169, the
vibration-reduction effect obtained from the field experiment
is equal to 17.0%, the value that is predicted by Equation (6).
These normalized vibration-reduction effects were close to
Case with bottom-air-deck at a length of 100 mm: those predicted by Equation (6), and then the numerical and
experimental results were found to agree well with each other.
4.2.1.2 Full-scale experiment
Case with bottom-air-deck at a length of 200 mm: Full-scale experiment was carried out at the tunnel site of
Geondong limestone mine in Korea. Figure 17 shows a plan
view of the field experiment site. The tunnel had a cover depth
of approximately 90 m over the crown. A total of six full-
scale experiments were carried out: three experiments without
Case with bottom-air-deck at a length of 300 mm: bottom-air-deck and three experiments with bottom-air-deck
at a length of 200 mm.
The ground condition, based on the result of face mapping,
belonged to fair. The diameter of blastholes was 45 mm and
In the above, V is the PVS (mm/s) and SD2 is the square-root their drilling length was 3.0 m in the direction of tunnelling.
scaled distance (m/kg1/2 ). Powdery ANFO (ammonium nitrate fuel oil) explosive was
23

Figure 20. Measurement of the station of the tunnel face for

investigating the advance.
Figure 19 shows a plan view of the field experiment site. The

experiment was carried out at two sites of Songhak tunnel that
was connected to the inclined shaft No.1. A total of three full-
scale experiments on bottom-air-deck with a length of 200 mm
Figure 18. Peak vector sum vs. scaled distance. Ra is the bottom- were carried out: two experiments at site 1; and one experiment
air-deck ratio.
at site 2. All the blastholes, except for the contour blastholes,
used as the column explosive with one cartridge explosive had the bottom-air-deck. The ground condition based on the
of the emulsion-type being used as a primer. The amount of RMR was fair. The diameter of blastholes was 45 mm and
explosives per blasthole was 2.2–3.0 kg/hole and the average their designed drilling length was 4.0 m in the direction of
explosive length was approximately 2.5 m. Thus, the average tunnelling. An emulsion-type explosive of 2.8 kg/hole was
bottom-air-deck ratio was 0.08. Figure 18 shows the ground- used.
vibration results and the derived ground-attenuation equations In order to investigate the advance after blasting, the stations
(Eqs. (17)–(18)). of the tunnel face at the four points, MP 1 through MP 4, shown
Case without bottom-air-deck: in Figure 20, were surveyed after blasting. The advance was
the difference between the averages of the stations at the four
points before and after blasting.
The excavation efficiency was assessed by Equation (7).
Case with bottom-air-deck at a length of 200 mm: The excavation efficiency in the case without bottom-air-deck
was investigated on the basis of daily report available at the
site. The report indicated that the average drilling length was
4.0 m which is exactly the same with the designed length. The
In the above, V is the PVS (mm/s) and SD2 (m/kg1/2 ) is the advance in the case without bottom-air-deck was calculated
square-root scaled distance. from the daily report on the measurement of stations at each
The average vibration-reduction effect at a bottom-air-deck blasting round.
ratio of 0.08 was calculated to be 9%. The corresponding Excavation efficiencies in the absence of bottom-air-deck
effect, predicted by Equation (6), was 10.2%. Thus, good and in the case of bottom-air-deck with a length of 200 mm
agreement between the numerical and experimental results were 88.8% and 84.5%, respectively. Thus, the reduction in
was found. the advance was calculated to be 172 mm. The reduction in
the field was found to be smaller than the numerically pre-
4.2.2 Effect of bottom-air-deck ratio on excavation dicted value of 221 mm. The difference is attributable to not
efficiency considering damaged region around the blasthole in the cal-
Full-scale experiment was carried out at the railway tunnel culation of the advance in the numerical study, as stated in
of Jecheon-Ssangyong Construction Section No. 1 in Korea. Subsection 3.3.2.
24

Table 2. Design conditions of line-drilling method.
Vibration-reduction effect
Ground condition Design conditions Predicted Designed
Fair D (mm) 45 Single row 13.19% 10%

S (mm) 200 (10.99%∗ )
DLD-BH 300
D (mm) 45 Double row (zigzag) 20.87% 15%
S (mm) 250 (Spacing of rows = (17.39%∗ )
DLD-BH 300 200 mm)
D (mm) 45 Double row (zigzag) 25.97% 20%
S (mm) 200 (Spacing of rows = (21.64%∗ )
DLD-BH 300 200 mm)
D and S = diameter and spacing of drill holes, respectively; DLD-BH = distance between line-drilling and blasthole
∗
The value was determined by applying a factor of safety of 1.2 to the numerically predicted effect.
5 DESIGN CONDITIONS FOR PRACTICAL

APPLICATION
From a comparison between the numerical and experimen-

tal results of both line-drilling and bottom-air-deck methods,
it was revealed that the vibration-reduction effects, predicted
through the numerical study, can be expected when both meth-
ods are applied to real tunnelling practices. Since the present
study has detailed information to design both methods through
experimental and numerical works, a factor of safety of 1.2
was used to propose their design conditions.
5.1 Design conditions of line-drilling method

Table 2 shows the design conditions of line-drilling method. In
order to propose the design conditions, Equation (5) derived
from the numerical study was used, and the following drill
conditions were applied in consideration of constructability Figure 21. Comparison of the vibration-reduction effects in terms
of drilling works: diameter of drill holes = 45 mm; spacing of the amplitude reduction factor when reducing the advance per
of those = min. 200 mm; and the number of rows of drill round (ARF1 ) and when using the bottom-air-deck method (ARF2 ).
holes = 1 to 2. n is the decay constant in a ground attenuation equation.
5.2 Design conditions of bottom-air-deck method

the blasthole was 2.2 m and the charge was 1.0 kg of emulsion-
Since bottom-air-deck was found to reduce the excavation type explosive (the explosive length = 1.18 m). The distance
efficiency, the vibration-reduction effects between bottom- from blasting source to monitoring point in the absence of the
air-deck method and the conventional method of reducing reduction in the advance per round was 30.0 m. The applied
the advance per round should be compared. Based on a lengths of the bottom-air-deck and the reduction in the advance
ground-attenuation equation derived in the field, the vibration- per round were 100, 200, 300, 400, and 500 mm, and thus the
reduction effect of the method of reducing the advance per bottom-air-deck ratios were 0.085, 0.169, 0.254, 0.339, and
round can be calculated by the following equation: 0.424, respectively.
Figure 21 shows the results of comparison of the vibration-
reduction effects between both methods. The vibration-
reduction effects of bottom-air-deck, used in Figure 21, were
calculated by using Equation (6). From the comparison, the
vibration-reduction effect of bottom-air-deck method was
where ARF (%) = amplitude reduction factor; SD1 and found to be greater than that of the method of reducing
SD2 = square-root scaled distances (m/kg1/2 ) in both cases the advance per round. However, when n was −2.5 and the
(without and with the reduction in the advance per round, bottom-air-deck ratio was 0.424 (500 mm in length of bottom-
respectively); and n = decay constant in a ground attenuation air-deck), the vibration-reduction effects of both methods
equation. were almost the same, as shown in Figure 21. Thus, the
The comparison of the vibration-reduction effects between present study proposed design conditions under the condi-
both methods was carried out under the following condition of bottom-air-deck ratio ≤0.4. Table 3 shows the design
tions. The range of −2.5 to −1.0 was used for n of the conditions of bottom-air-deck method. Since the vibration-
ground-attenuation equation in consideration of the range of reduction effects, used for proposing the design conditions,
the constant, n, reported by New (1982). The drilling length of were based on the results of numerical analyses for a single
25

Table 3. Design conditions of bottom-air-deck method.
Design conditions
Vibration reduction effect
Drilling length & Bottom-air-
Ground condition blast initiation deck ratio Predicted Designed
Fair 2.2 m & blast 0.1 11.8% (9.9%∗ ) 10%

initiation employing 0.2 19.0% (15.8%∗ ) 15%
one hole per delay 0.3 25.0% (20.8%∗ ) 20%
0.4 30.4% (25.3%∗ ) 25%
∗
The value was determined by applying a factor of safety of 1.2 to the numerically predicted effect.
blasthole, the blast initiation that employs one hole per delay conditions of the ground and blasting are needed to extend the
was proposed. engineering application.
6 CONCLUSIONS AND DISCUSSION

REFERENCES
In the present study, a quantitative assessment on the effective-
ness of line-drilling and bottom-air-deck methods was carried Bieniawski, Z.T. 1979. The geomechanics classification in rock
out through numerical and experimental investigations. The engineering applications. Proceedings of the fourth international
results obtained from the numerical and experimental study congress on rock mechanics, Montreaux, 2–8 September 1979:
41–48. Rotterdam: Balkema.
were as follows. Dunham, J.M. 1954. Pile foundations for buildings. Journal of the
From the numerical study, correlation equations of Soil Mechanics and Foundations Division 80(1): 1–21.
vibration-reduction effect, which give quantitative informa- Hunt, R.E. 1984. Geotechnical engineering investigation manual:
tion on the drilling of barrier holes and on the bottom-air-deck 187–207. New York: McGraw-Hill.
ratio, were derived. The numerical results of bottom air-deck Jimeno, C.L., Jimeno, E.L. & Carcedo, F.J.A. 1995. Drilling and
method indicated that the reduction in excavation efficiency blasting of rocks. Rotterdam: Balkema.
occurred due to the bottom-air-deck. The numerical results Lee, E., Finger, M. & Collins, W. 1973. JWL equation of state
were verified through field experiments, and then good agree- coefficients for high explosives. Livermore: Lawrence Livermore
ment was found between the numerical and experimental National Laboratory.
results. Marchenko, L.N. 1982. Raising the efficiency of a blast in rock
crushing. Journal of Mining Science 18(5): 395–399.
From the present numerical and experimental study, it New, B.M. 1982.Vibration caused by underground construction.Tun-
was revealed that when both methods are applied to tunnelling 82, Proceedings of the third international sysmposium,
nelling sites, the vibration-reduction effects that were pre- Brighton, 7–11 June 1982: 217–229. London: IMM.
dicted through the numerical study can be expected, and thus Park, D., Jeon, B. & Jeon, S. 2009. A numerical study on the screen-
based on the numerical results, design conditions of both ing of blast-induced waves for reducing ground vibration. Rock
methods were proposed by introducing the concept of fac- Mechanics & Rock Engineering 42(3): 449–473.
tor of safety. It is believed that vibration-reduction effects of Riedel, W., Thoma, K., Hiermaier, S. & Schmolinske, E. 1999. Pene-
10–20% and 10–25% in the transverse direction of a tunnel tration of reinforced concrete by BETA-B-500, numerical analysis
and in the direction of tunnelling, respectively, can be expected using a new macroscopic concrete model for hydrocodes. Pro-
under the design conditions proposed in the present study. ceedings of the ninth international symposium on interaction of
the effects of munitions with structures, Berlin, 3–7 May 1999:
In conclusion, the numerical analyses and the field obser- 315–322.
vations suggested that both methods reduce blast-induced Riedel, W. 2000. Beton unter dynamischen lasten meso- und
vibration in tunnelling according to the location of nearby makromechanische modelle und ihre parameter. Freiburg: Ernst-
structures. However, the present experimental study was car- Mach-Institut (EMI).
ried out in only a few sites. Thus, in the future, further field Woods, R.D. 1968. Screening of surface waves in soils. Journal of
evaluations of the effectiveness of both methods under various the Soil Mechanics and Foundations Division 94(4): 951–979.
26

Invited lectures on ISRM 50th anniversary

Fifty years of the ISRM and associated progress in rock mechanics
E.T. Brown
Golder Associates Pty Ltd, Brisbane, Australia
ABSTRACT: The 12th Congress of the International Society for Rock Mechanics (ISRM) is being held in the fiftieth year of
the life of the Society which was established formally on 24 May 1962 in Salzburg, Austria. This paper discusses the emergence
of rock mechanics as a distinctive engineering and scientific discipline, the state and achievements of the discipline at the time
the ISRM was founded in 1962, the events leading up to the formation of the Society, the development and achievements of
the Society in the 50 years since 1962, and the progress made in the discipline of rock mechanics and rock engineering since
that time.
Subject: Fifty years of the ISRM and rock mechanics
Keywords: General
1 INTRODUCTION 2 THE WORLD IN 1962
This 12th Congress of the International Society for Rock To begin, it is worth recalling what the world was like in 1962.
Mechanics (ISRM) falls in the fiftieth year of the life of the In world politics, the Cold War was in full swing. The Pre-
Society which was established formally on 24 May 1962 in mier of the then U.S.S.R, Nikita Kruschev, and the young
Salzburg, Austria (Fairhurst 2010). This Congress therefore President of the U.S.A., John F. Kennedy, were exercised over
provides an important opportunity for the Society to recall its what became known as the Cuban missile crisis. Elsewhere,
foundation, reflect on its achievements and on the progress Harold Macmillan was the Prime Minster of the U.K., Charles
made in our discipline in the 50 years since its formation, and de Gaulle was still the President of France, Jawaharlal Nehru
to look forward to the next 50 years. was the Prime Minister of India and Chairman Mao Zedong
This paper attempts the rather large and complex task of was the leader of China. It has to be said that, at that time, the
reviewing and synthesising the emergence of rock mechanics People’s Republic of China was not at all well known to those
as a distinctive engineering and scientific discipline, the state living in what is usually referred to as “the West”.
and achievements of the discipline at the time the ISRM was In culture – popular and otherwise – the American writer,
founded in 1962, the events leading up to the formation of the John Steinbeck, won the Nobel Prize for Literature. There was
Society, the development and achievements of the Society in an unhappy link to one of the life-long interests of the ISRM’s
the 50 years since 1962, and finally, the progress made in the Founder and First President, Professor Leopold Müller, when
discipline since that time. Although too young to have been the world-famous Austrian violinist, Fritz Kreisler, died on
involved in the work of the Society in its earliest years before 29 January. In popular music, The Beatles released their first
the First Congress held in Lisbon in 1966, the writer has fol- recording and The Rolling Stones made their public debut.
lowed the developments discussed here with keen interest for In the writer’s favourite form of music – jazz – the pianist
about 48 of these 50 years. He is honoured and delighted to Herbie Hancock’s tune Watermelon Man topped not only the
have been invited by the Society’s current President, Professor jazz charts but the popular music charts as well for several
John Hudson, to prepare and present this paper. In a compan- weeks in 1962.
ion paper and lecture, Professor Hudson will look forward to In sport, Brazil won the World Cup in football; the NewYork
the next 50 years of the Society and the discipline (Hudson Yankees won the World Series in baseball; theAmerican golfer,
2011). Arnold Palmer, won his second consecutive British Open
Throughout this paper, the term rock mechanics will be used and his third U.S. Masters titles; and an Australian, Rod
in the sense given in the ISRM Statutes: “The field of rock Laver, won the Grand Slam in tennis.
mechanics is taken to include all studies relative to the phys- In science and technology, the first observation of laser
ical and mechanical behaviour of rocks and rock masses and activity was reported; the first practical light emitting diode
the applications of this knowledge for the better understand- was produced; the world’s first active telecommunications
ing of geological processes and in the fields of engineering”. satellite, Telstar 1, was launched on 10 July; the felt-tipped
Thus, the term rock mechanics will be generally taken to pen came into being; and, although digital computing was in
include rock engineering, although on occasion, the two may its infancy and the personal computer was still some years
be expressed independently as in the now commonly-used off, in a sign of things to come, the first computer game was
term, rock mechanics and rock engineering. developed for a PDP-1 computer.
29

on Rock Pressure and Ground Support was held in Liège,
Belgium. In 1958, the International Bureau of Rock Mechan-
ics was established at the third International Strata Control
Congress held in Leipzig in the then East Germany. From
the early 1950s, systematic research in rock mechanics began
in the U.S.A., and in 1956 the first of what soon became
annual U.S. symposia on rock mechanics was held at the Col-
orado School of Mines. In 1962, the eighth ISRM President,
Professor Charles Fairhurst, organised the 5th U.S. Sympo-
sium at the University of Minnesota (Fairhurst 1963). In that
year, the first Canadian Symposium on Rock Mechanics was
held (Anon 1963). And as well as there being publications
in dedicated and more broadly-based journals and in the var-
ious conference proceedings, by the early 1960s books on
rock mechanics had started to appear, notably Talobre’s La
Figure 1. The late Professor Leopold Müller, 1908–1988, founding Mécanique des Roches, published in Paris in 1957 (Talobre
President of the ISRM. 1957). A conference on the subject was also held in Paris in
that year.
These few facts show clearly that, by 1962, the disci-
3 THE EMERGENCE OF ROCK MECHANICS
pline of rock mechanics was identifiable and reasonably well
AS AN IDENTIFIABLE DISCIPLINE
established, at least in several parts of the world.
The writer has argued previously that, by the early 1960s, the
subject of rock mechanics, if it wasn’t yet fully established, 4 THE STATE AND ACHIEVEMENTS OF ROCK
was well on its way to becoming established as an identifi- MECHANICS IN 1962
able scientific and engineering discipline (Brown 1999, 2002,
Hood & Brown 1999). The state of knowledge in the disci- The next question that might reasonably be asked is what were
pline in 1962 will be summarised in the next section. Here, the concerns, the state of knowledge, and the achievements
it will be demonstrated that it was at least well on the way of the discipline of rock mechanics and rock engineering at
to becoming established through the appearance of specialist that time? The short answer is that, with one vital deficiency,
journals, conferences and societies. they were perhaps more advanced than might now be first
From the mid-nineteenth century, publications on what we supposed.
would now recognize as rock mechanics and rock engineering By 1962, many of the major concerns of the discipline
appeared in mining and engineering journals, in the publica- had been clearly identified and were being investigated by
tions of professional societies, and in reports of university and both researchers and practitioners. For example, what is often
other research institutes in a number of countries (Obert & referred to as the Austrian School of Rock Mechanics was
Duvall 1967). In discussing what they refer to as the first well established. According to Müller (1979), as early as
stage of the development of rock mechanics, the Russian writ- 1905, Albert Heim had recognized the distinction between
ers, Turchaninov et al. (1974) concluded that “by the end of rock material and the rock mass. The use of stereonets had
the 19th century, the basic patterns of deformation and frac- been introduced by Schmidt (1925); methods of studying
ture of rock in underground workings and on the surface had joint systems in rock masses had been developed by Stini
been studied, instrumented observations of the deformation in the 1920s (Stini 1922) and subsequently by Müller him-
of the surface had been made for the most important western self (Müller 1933, 1950, 1979); laboratory and large-scale
European coal deposits, and the first practical rules had been field testing techniques were being used (e.g. Blanks &
suggested regarding the leaving of blocks of ore (safety pillars) McHenry 1945, Golder & Akroyd 1954, John 1962, Rocha
and delineating the area of dangerous surface subsidence”. et al. 1955); ground-support interaction concepts had been
The first use of the term “rock mechanics” in English that developed (Pacher 1964, Rabcewicz 1969); and limiting equi-
the writer is aware of was in a paper published by Appleton librium methods were used to analyse the stability of blocks of
(1944) in South Africa. The first issue of the first special- rock in slopes and foundations (John 1962, Terzaghi 1962b).
ist journal devoted to rock mechanics and rock engineering, Experimental investigations of the engineering properties
Geologie und Bauwesen, edited by Professor Josef Stini, was of rocks were well underway in many parts of the world (Mayer
published in Vienna in 1929. Professor Leopold Müller who 1953, Obert et al. 1946). In fact, the mechanical testing of rock
soon was to become the First President of the ISRM (Fig. 1) can be traced back at least to Coulomb and others in Europe
succeeded Stini as editor in 1958, and in 1962 this jour- in the mid-18th century (Coulomb 1776, Heyman 1972). From
nal changed its name to Felsmechanik und Ingenieurgeologie a perhaps more scientific, geological or geophysical rather
(Rock Mechanics and Engineering Geology). Further name than a rock engineering perspective, high pressure compres-
changes to Rock Mechanics (Müller 1969) and to the cur- sion testing of rock material had been underway since the
rent Rock Mechanics and Rock Engineering were made in pioneering work of von Kármán (1911) and King (1912), fol-
1969 and 1983, respectively. The other major journal in the lowed by that of Griggs and Handin in the USA from the
field, the International Journal of Rock Mechanics and Mining mid-1930s (Griggs 1936, Handin 1953). In Japan, Professor
Sciences, was founded by Albert Roberts in the UK in 1963 Kiyoo Mogi’s large body of experimental work in this area
with a primarily mining orientation (Roberts 1963) which has was well underway by 1960 (Mogi 1959, 2007) as was that of
disappeared with the efflux of time. Dr Mervyn Paterson in Australia (Paterson 1958, 1978). Also
Since 1950, annual colloquia on rock mechanics had been at a fundamental level, Professor John Jaeger’s initial semi-
held in Austria. In 1951, the First International Conference nal work on the friction of rock joints and surfaces had been
30

published (Jaeger 1959, 1960). At the same time, large-scale work was the experimental and theoretical study of the brittle
field shear testing of rock discontinuities was being practiced fracture of rock using, among other things, modified Griffith
in many parts of the world (John 1962, Rocha et al. 1955). crack theory (Cook 1965, Cook et al. 1966, Hoek 1964).
Rock mechanics studies formed an important part of the Importantly, as Hoek (2007) has argued, “the early 1960s
investigation and design of dams and of tunnels for hydroelec- were very important in the general development of rock engi-
tric developments and other purposes (Jaeger 1955, 1972). As neering world-wide because a number of catastrophic failures
early as 1951, Dr Charles Jaeger had submitted a proposal to occurred which clearly demonstrated that, in rock as well as in
the International Commission on Large Dams (ICOLD) “to soil, ‘we were over-stepping the limits of our ability to predict
create a sub-committee on rock mechanics” (Jaeger 1972). the consequences of our actions’(Terzaghi andVoight, 1979)”.
Dr Klaus John later remarked that at the 7th ICOLD meet- Reference to just three of these failures will serve to make the
ing held in Rome in 1961, “an increased emphasis on rock point. In December 1959, the foundation of the Malpasset
mechanics could be observed” (John 1962). The ISRM’s concrete arch dam in France failed with the resulting flood
second President, Professor Manuel Rocha, and his National killing about 450 people (Londe 1987). A few weeks later in
Civil Engineering Laboratory in Lisbon, Portugal, was a world January 1960, the pillars at a coal mine at Coalbrook in South
leader in that endeavour (Rocha 1964, Rocha et al. 1955). And Africa collapsed with the loss of 432 lives (Bryan et al. 1964).
in the writer’s home country ofAustralia, from 1949, a team led And in October 1963, about 2500 people in the Italian town
by Tom Lang working on the monumental Snowy Mountains of Longarone were killed as a result of a landslide-generated
Hydro-electric Scheme, made significant contributions to the wave that overtopped the Vaiont Dam (Müller 1964). These
development of rock mechanics and rock engineering both three failures led to detailed investigations, much debate in
nationally and internationally (e.g. Alexander 1960, Brown the literature, the establishment of research programs, and the
1999, Lang 1961, Moye 1959, Pinkerton et al. 1961). development and application of new methods of rock engi-
At that time, elastic stress analyses of underground exca- neering investigation and analysis (see, for example, Bernaix
vations in rock were carried out using closed-form solutions 1969, Jaeger 1972, Londe 1987, Londe et al. 1969, Müller
such as those published by Terzaghi & Richart (1952). Photo- 1968, Müller-Salzberg 1987, Salamon & Munro 1967).
elasticity was used for more complex excavation shapes (e.g. The examples and references given in the preceding para-
Hoek 1963, Pinkerton et al. 1961), and an electric analogue graphs will make the essential points that by the early 1960s
solution was being developed in South Africa (Cook et al. many of the major concerns of our discipline had been iden-
1966, Salamon et al. 1964). Digital computers and numerical tified, many of the techniques that we use were in at least the
methods of stress analysis such as the finite element method early stages of their development, and significant advances
existed (Clough 1960), but their serious application to rock had been made in the understanding of at least some areas of
engineering stress analyses was still a few years away (e.g. rock mechanics and in its application in rock engineering. But,
Goodman 1966, Zienkiewicz & Cheung 1964). as is still the case today, many significant questions remained
As part of this general effort, there was wide-spread inter- unanswered. The most important of these, at least in the con-
est in measuring the stresses in the Earth’s crust (Judd 1964, text of the ISRM, was Müller’s implicit question, “what is the
Terzaghi 1962a). The Society’s fourth President, Professor strength of a rock mass?” (Müller 1967a). The background to
Pierre Habib of France, was involved in the development and this question and a masterly account of the progress made in
application of the flat-jack method as early as 1950 (Habib answering it are given by Fairhurst (2010).
1950, Habib & Marchand 1952, Mayer et al. 1951). This
method was also being used to measure the in situ mod-
uli of rock masses (Habib 1950), as were dynamic methods 5 THE FOUNDING OF THE ISRM
(Brown & Robertshaw 1953, Evison 1953). In 1958, one of the
more important papers in the history of rock stress measure- Following a meeting of 16 men held in Professor Müller’s
ment, or estimation, as we now prefer to call it in the ISRM home in Salzburg, Austria, in 1951, an Internationalen
Suggested Methods, was published by Nils Hast in Sweden Arbeitsgemeinschaft für Geomechanik or International Work-
(Hast 1958). By the early 1960s, a wide range of methods of ing Group for Geomechanics, was established to study rock
rock stress measurement had been investigated and/or devel- mechanics problems, essentially using the Austrian School
oped (see, for example, the papers in the proceedings of a 1963 approaches outlined above (Müller 1967b). This group which
conference edited by Judd (1964)). And, as has been discussed was also known as the Salzburger Kreis or Salzburg Circle,
by Obert & Duvall (1967) and Hood & Brown (1999) in the remained largely Austrian until Professor Müller invited Pro-
context of mining rock mechanics, during the 1950s signif- fessor Fairhurst to join at the 3rd U.S. Rock Mechanics
icant advances were made in the development of a range of Symposium held at the Colorado School of Mines in 1958
instruments for other forms of field monitoring for rock engi- (Fairhurst 2010). On 24 May 1962, Professor Müller offi-
neering applications in both civil and mining engineering (e.g. cially registered the name of the Internationale Gesellschaft
Potts 1957). für Felsmechanik (International Society for Rock Mechanics)
A significant program of research on rock mechanics as in Salzburg. The minutes of the Constitutional Meeting of
applied to deep-level hard rock mining and the associated the Society held in Salzburg on the following day lists 46 par-
problem of rock bursts, had been established in South Africa ticipants from Austria, Germany, Italy, Poland, U.K., U.S.A.
in 1953 (Hill 1954, 1966). At around the same time, a similar and Yugoslavia. Under Professor Müller’s dynamic leadership
problem was under investigation at the Kolar goldfield, India and with growing support from a number of other countries,
(Taylor 1962–63). A number of towering figures in the history particularly in Europe, the Society soon grew to become truly
of our discipline, including two Müller Lecturers, Drs Neville international.
Cook and Evert Hoek, made their initial theoretical and exper- As noted at the 11th Congress held in Lisbon in July 2007,
imental contributions to the development of rock mechanics the historic First Congress held in Lisbon in September 1966
as part of the South African program. A central feature of that was a major milestone and highlight in the early life of the
31

Society and in the writer’s own life (Brown 2007). A total of on activities being carried out by the rock mechanics
814 delegates from 40 countries attended that First Congress. community;
In his Address given at the majestic Opening Ceremony, • operate Commissions for studying scientific and technical
Professor Müller made this important statement (translated matters of concern to the Society;
from the original German into English): • award the Rocha Medal for an outstanding doctoral thesis,
“Many experts agree with me that discontinuity and every year, and the Müller Award in recognition of distin-
anisotropy are the most characteristic properties of the mate- guished service to the profession of rock mechanics and
rial rock and that the properties of jointed media depend much rock engineering once every four years; and
more upon the joints of the unit rock block system than upon • cooperate with other international scientific associations.
the rock material. Therefore, any theoretical investigation of
The writer may be accused of bias, but he concludes that,
that material has to go its own ways, in the same way as
over an extended period of time, the Society has been very
the construction material of soils years ago suggested to soil
successful in pursuing its objectives and purposes through
mechanics its own methods, which differ greatly from the way
these various activities, all of which have been carried out
of thinking of technical (or continuum) mechanics” (Müller
very effectively. In terms of the objectives and purposes, it is
1967a).
suggested that the Society has been demonstrably successful
Thus, rock mechanics became recognized as being largely
in pursuing at least the first two objectives. It might be argued
the study of the mechanics of discontinua as the writer’s own
that the third has been pursued and achieved perhaps more at
PhD supervisor at that time, Professor Hugh Trollope, put
an implicit, rather than at an explicit, level.
it (Trollope 1968, Trollope & Brown 1965). That does not
A number of the Society’s activities have been very success-
mean to say that theoretical continuum mechanics in the fields
ful over the years. The excellent ISRM web-site maintained
of both elasticity and plasticity did not, and does not, have
by the Secretariat indicates that beginning with a Sympo-
application to rock mechanics and rock engineering. But it
sium held in Madrid in October 1968, to date the Society has
does mean that answering Professor Müller’s implicit question
sponsored a total of 33 International Symposia, 68 Regional
about the strength of a rock mass requires consideration of the
Symposia and two Specialised Conferences. The work of the
mechanics of a discontinuum.
various Commissions of the Society established since the time
An important development in the early life of the Soci-
of the First Congress has brought great credit to the Soci-
ety occurred in 1966 following the election of Manuel Rocha
ety. Without wishing to minimise the contributions made by
as our second President, when the Secretariat was estab-
the other Commissions in the past, and those being made by
lished at the National Civil Engineering Laboratory (LNEC)
the current Commissions, in the writer’s opinion, one of the
in Lisbon, Portugal, with generous financial support provided
most impressive and lasting achievements of the Society has
by the Calouste Guggenheim Foundation. Since that time,
been the work of what is now the Commission on Testing
our Society has been well-served by a succession of seven
Methods. That Commission began its life at the time of the
Secretaries-General provided by LNEC.
First Congress in 1966 as the Commission on Standardiza-
So by 1966, our discipline had been established, our Inter-
tion of Laboratory and Field Tests under the Chairmanship of
national Society had been well and truly founded with a
Dr Don Deere. It became the Commission on Testing Meth-
well-funded Secretariat in Lisbon, and our highly successful
ods at the 4th Congress held in Montreux, Switzerland, in
First International Congress had been held.
1979.
A great many members of the Society have been associated
6 THE DEVELOPMENT AND ACHIEVEMENTS OF with the quite detailed work of the Commission and its Work-
THE ISRM ing Groups over the last 45 years. However, it is believed that
they will all agree that the contributions of two people have
The Statutes and By-laws by which the Society is governed towered above those of all others – Dr John Franklin, who
have changed periodically over time since 1962, but the was Commission President from 1974 until 1987 when he
Society’s essential objectives and purposes have remained: became the Society’s seventh President, and our current Pres-
ident, Professor John Hudson, who was Commission President
• to encourage international collaboration and the exchange from 1987 until 2006. The lasting contribution made by the
of ideas and information between rock mechanics practi- Commission in developing a wide range of Suggested Meth-
tioners; ods for laboratory and field testing is reflected in what we call
• to encourage teaching, research, and the advancement of the ‘Blue Book” edited by the current Commission President,
knowledge in rock mechanics; and Resat Ulusay, and John Hudson (Ulusay & Hudson 2007).
• to promote high standards for professional practice among Other achievements of the Society which address the objec-
rock engineers so that civil, mining and petroleum engineer- tive of advancing teaching and the advancement of knowledge
ing works might be safer, more economic and less disruptive in rock mechanics, have been the production of the ISRM
to the environment. slide collection, the introduction in recent years of the ISRM
The main activities carried out by the Society in pursuit of Lectures and the Annual Technical and Cultural Field Trips,
these objectives are to: and the recent publication of the outcomes of the work of
the Commission on Rock Engineering Design Methodology
• hold International Congresses at intervals of four years; (Feng & Hudson 2011).
• sponsor a co-ordinated program of International Symposia, Many of the ISRM Lectures are now available on-line
Regional Symposia and Specialised Conferences on topics through the Society’s web site. And the on-line Digital Library
in rock mechanics and rock engineering, organized by the developed by the Secretariat “went live” in 2010.
National Groups of the Society; Perhaps one disappointing feature of the Society’s history
• publish a News Journal to provide information about tech- is the relative stagnation in individual membership numbers,
nology related to rock mechanics and up-to-date news both overall and in some of our six Regions of Africa, Asia,
32

Australasia, Europe, North America and South America, and attended the 11th Congress held in Lisbon in 2007. Of course,
in some countries within those Regions. As shown in Figure 2, a great many more Chinese delegates are attending this 12th
the total individual membership was only 383 at the time of Congress!
the First Congress in 1966, but quickly grew to reach more In the 1983 to 1987 ISRM term of office, the Society had its
than 4000 in 1970. It first passed 5000 in 1982 and 6000 in first Chinese Vice-President in Professor Tan Tjong Kie. The
1988, reaching a peak of 6470 in 1989. It then declined until first ISRM International Symposium to be held in China was
about the year 2000 and then increased steadily to 6312 in held in Beijing in November, 1986. Interestingly, in a paper
2010. Over the same period, the number of National Groups given to that Symposium, Chen et al. (1986) reported that
increased more steadily from 12 in 1966 to 48 or 49 in recent a program of in situ stress measurement at the Three Gorges
years. Dam area had been initiated by Professor Tan as early as 1958.
In many ways, Europe has always been the numerical and Now the Society is to have its first Chinese President, Profes-
financial mainstay of the Society. Figure 3 taken from the 2010 sor Xia-Ting Feng, who will assume the Presidency for the
annual review shows the distribution in individual member- 2011 to 2015 term at the end of this Congress.
ship by Region from 2003 to 2010. As well as the high and The European, and particularly the English-speaking,
increasing European membership shown by the yellow bars, influence in our Society is illustrated in a number of other
this slide illustrates the increasing Asian membership shown ways, including by the list of Presidents shown in Table 1.
in pink, mainly as a result of increases in membership in China Of the 12 ISRM Presidents to date, seven have been from
and India. European countries. Furthermore, another two of our Past
In fact, the growth of rock mechanics in China has been a Presidents, John Franklin and Charles Fairhurst, although
feature of the development of our Society and of our discipline North American when they became President, were both
over the last 25 to 30 years. The first 11 Chinese individ- originally British.
ual members joined the Society in 1979. China is now the The Society’s major award, the Müller Award, has been
Society’s largest National Group with 558 members in 2010, awarded every four years since the first Award was made to Dr
followed closely by India. The writer’s records and recollec- Evert Hoek at the Seventh Congress held in Aachen, Germany,
tion suggest that China didn’t really begin to participate in the in 1991. The list of Müller Award recipients given in Table 2
work of the Society until about the time of the 5th Congress
held in Melbourne, Australia, in 1983 when 18 Chinese del-
egates attended. The numbers were small for the next two
Congresses but then grew steadily until 54 Chinese delegates Table 1. ISRM Presidents from 1962 to date.
Year President (Country)
1962–1966 Leopold Müller (Austria)

1966–1970 Manuel Rocha (Portugal)
1970–1974 Leonard Obert (USA)
1974–1979 Pierre Habib (France)
1979–1983 Walter Wittke (Germany)
1983–1987 Edwin T. Brown (UK)
1987–1991 John A. Franklin (Canada)
1991–1995 Charles Fairhurst (USA)
1995–1999 Shunsuke Sakari (Japan)
1999–2003 Marc Panet (France)
2003–2007 Nielen van der Merwe (South Africa)
2007–2011 John A. Hudson (UK)
2011–2015 Xia-Ting Feng (China)
Figure 2. Total ISRM individual membership, 1963–2010.
Figure 3. Individual ISRM membership by Region, 2003–2010.
33

shows a similar European and English-speaking influence to 7 DEVELOPMENTS AND PROGRESS
the list of Presidents. It should be understood that all four IN ROCK MECHANICS
of the North American recipients were born and educated in
other countries. Furthermore, the 2007 recipient, although an 7.1 Scope
Australian, actually worked in the United Kingdom for 12
The task of making assessments of the main developments
years when he was most closely involved in rock mechanics
and the progress made in rock mechanics and rock engi-
and in the work of the Society.
neering in the 50 years since the ISRM was established is a
Annually since 1982, the Society has awarded the Rocha
most challenging one. It is inevitable that any account such as
Medal for an outstanding doctoral thesis to honour the mem-
that attempted here will reflect the writer’s personal interests,
ory of its distinguished second President, Manuel Rocha. As
knowledge and experience, and so, to at least some extent,
might be expected, the 30 Rocha Medal recipients to date
will be idiosyncratic. It is also inevitable that, for much the
have been spread more evenly among the Society’s Regions
same reasons, but also because of limitations of space, some
and National Groups than have been the Presidents and the
important contributions made by some individuals and in some
Müller Award recipients with five of the Society’s six Regions
countries will be overlooked. In such cases, the writer offers
and 17 different National Groups being represented. Impor-
his apologies. The references given for each of the advances
tantly, a number of the Rocha Medal recipients have gone on
listed under sub-headings 7.2 to 7.6 below are intended to
to become senior and important figures in the discipline and
provide examples of the significant contributions made to
in the Society.
the various topics or techniques being discussed and are not
It can be concluded, therefore, that during the first 50 years
intended to be either exhaustive or definitive. The references
of its life, the ISRM has pursued its stated objectives and
given are, quite simply, those known and available to the writer
purposes conscientiously by carrying out all of its main activ-
at the time of writing.
ities at least satisfactorily, with some being carried out to
In making an assessment such as that reported here, it is first
considerable effect.
necessary to establish a framework within which the assess-
ment will be made and the results summarised. The framework
used here is the simplified general rock engineering process
Table 2. Müller Award recipients, 1991–2011. used by a number of authors (e.g. Brady & Brown 2004,
Hudson & Feng 2007, Read & Stacey 2009) as represented
Year Award Recipient by the headings of sub-sections 7.2 to 7.6. This generalised
process is necessarily simplified so that it can be adapted to
1991 Evert Hoek (Canada) apply to a wide range of rock engineering activities and appli-
1995 Neville G. W. Cook (USA)
cations. It should be emphasised that, in practice, the process
1999 Herbert H. Einstein (USA)
2003 Charles Fairhurst (USA) is not linear but includes parallel activities and multiple feed-
2007 Edwin T. Brown (Australia) back loops. Figure 4 shows a rock mechanics modelling and
2011 Nicholas R. Barton (Norway) rock engineering design flowchart which provides more detail
of the design analysis stage of the overall process.
Figure 4. Flowchart of rock mechanics modelling and rock engineering design approaches (Feng & Hudson 2004).
34

7.2 Site investigation and site characterisation • the development and refinement of empirical methods
of estimating rock material and rock mass strength and
Adequate site investigation and geological, geotechnical and
deformation properties (Hoek & Brown 1980, Hoek &
hydrogeological site characterisation remains at the core of
Diederichs 2006, Suorineni et al. 2009);
any successful rock engineering project. As was suggested
• the introduction of fracture mechanics concepts and the
in Section 4, the fundamentals and basic techniques of site
development of related theory and methods of testing and
investigation and site characterisation were established by
analysis (Atkinson 1987, Ulusay & Hudson 2007, Zhou
the time the ISRM was founded. However, since that time,
et al. 1986);
significant advances have been made in a number of relevant
• the development of improved methods of direct shear test-
areas such as:
ing of rock discontinuities under constant normal load
• drilling and borehole logging technologies, including core (CNL) and constant normal stiffness (CNS) conditions and
orientation methods and a range of geophysical logging of improved understandings of the influence of surface
techniques such as the Acoustic Televiewer (ATV) (Brown roughness and infilling materials on the shear strengths
2007a, Takahashi et al. 2006); and stiffnesses of artificial and natural discontinuities
• geophysical methods of site investigation, including (Barla et al. 2007, Barton & Bandis 1990, Goodman 1989,
seismic reflection, refraction and cross-hole tomography Grasselli & Egger 2003, Indraratna et al. 2010);
(Barton 2006, Takahashi 2004, Takahashi et al. 2006); • the numerical modelling of the progressive fracture and
• methods of discontinuity data collection using photogram- break-down of rocks and rock masses using, in particular,
metric, remote sensing, laser scanning and 3D digital Discrete Element Methods and bonded particle modelling
imaging technologies (Gaich et al. 2007, Grobler et al. 2003, (Potyondy & Cundall 2004). A most significant develop-
Read & Stacey 2009, Slob et al. 2007); ment in this area has been the Synthetic Rock Mass model
• methods of carrying out and interpreting hydrogeological approach which has the potential to answer Müller’s implicit
characterisation tests (Elsworth & Mase 1993, Louis 1974); question, “what is the strength of a rock mass?” (Fairhurst
• modern rock mass classification schemes developed from 2010, Mas Ivars et al. 2011, Pierce et al. 2009); and
the early 1970s (Barton 2006, Barton et al. 1974, Bieniawski • improved fundamental understandings of fluid flow in
1976, 1989, Hoek 1994, Marinos et al. 2007); single discontinuities and in discontinuous rock masses
• understanding the statistics of discontinuity systems (Barton & de Quadros 1997, Long & Witherspoon 1985,
(Billaux et al. 1989, Dershowitz & Einstein 1988, Priest Louis 1974, Witherspoon et al. 1980).
2004, Priest & Hudson 1981, Zhang et al. 2002) and the
application of this knowledge in tools such as Discrete Frac- Despite the advances that have been made and the sophisti-
ture Network (DFN) simulations (Dershowitz 1995, Rogers cated approaches being used for rock and rock mass property
et al. 2010); estimation on some rock engineering projects, and the wide-
• the 3D representation of site geology, structures and spread availability of the ISRM Suggested Methods, the
geotechnical domains using engineering design and mine writer’s experience has been that some practicing engineering
planning software (Beer 2010); and geologists and geotechnical engineers do not have an adequate
• methods of in situ stress estimation, although this process understanding of the purposes, complexities and requirements
remains fraught with difficulty (Amadei & Stephansson of good testing and rock property estimation programs. All
1997, Fairhurst 2003, Hudson et al. 2003). too often, empirical approaches such as the Hoek-Brown rock
mass strength criterion are treated as “black boxes” without
For a range of reasons, one area of site investigation that has adequate thought being given to the problem at hand (Brown
not developed to any great extent in the period since the early 2008, Kaiser & Kim 2008). At the research level, some of the
1960s is in situ testing. Despite the range of tools now available work being published, particularly in conference proceedings,
to assist in the overall site investigation and site characterisa- appears to add little of value to the corpus of rock mechan-
tion process, the writer’s recent experience on some projects ics knowledge and to represent little more than “turning the
has been that basic core logging skills and the training pro- handle”.
vided for that important task sometimes leave much to be
desired.
7.4 Design analysis
7.3 Rock and rock mass properties Methods of rock engineering design analysis have continued to
Key developments in the general area of estimating the develop during the last 50 years with an emphasis on numer-
mechanical properties of rocks and rock masses in the last ical methods of stress and deformation analysis, but not to
50 years include: the exclusion of other approaches. For present purposes, it
is convenient to adopt the classification of rock mechanics
• the study of the progressive breakdown of rock in labora- modelling approaches used by Feng & Hudson (2004, 2010)
tory tests and the development and use of servo-controlled and illustrated in Figure 4.
testing techniques (Hudson et al. 1971, Martin 1997, In this classification, Level 1 methods are direct methods
Wawersik & Fairhurst 1970); in which an explicit attempt is made to represent the geome-
• the augmentation of these studies by the use of acoustic try and mechanisms of the problem directly. They include the
monitoring (Martin 1997); use of pre-existing standard methods, analytical methods, and
• the development and dissemination by the ISRM of a set of basic and extended numerical methods. Level 2 methods are
Suggested Methods for laboratory and field rock property indirect methods in which the problem geometry and mech-
tests (Brown 1981, Ulusay & Hudson 2007); anisms are not represented explicitly. They include precedent
• the use of computerised methods of test control and auto- type analysis, the use of rock mass classification schemes,
matic test data collection and analysis (Barla et al. 2007); basic systems methods and integrated systems approaches,
35

possibly internet-based (Feng & Hudson 2004, Hudson & (Goodman & Shi 1985), DEM formulations for particle
Feng 2007). systems including bonded particle systems (Potyondy &
Of these eight approaches, pre-existing standard methods, Cundall 2004), and quasi-static and dynamic lattice net-
analytical methods, precedent type analysis (but not of a for- work models (Cundall 2011, Cundall & Damjanac 2009,
mal type) and some early rock mass classification methods Oñederra et al. 2009);
(but not the main methods used since the 1970s), were used • hybrid or linked methods of a number of types (FEM/BEM,
before the early 1960s. Since that time, new and widely- DEM/BEM, DEM/FEM – see, for example, Beer & Watson
used rock mass classification-based design methods have 1992, Brady 1987, Elsworth 1986, Lorig & Brady 1982);
been introduced (Barton et al. 1974, Bieniawski 1976, 1989, • DFN-based methods which may be combined with a num-
Hoek 1994, Marinos et al. 2007), and a wide range of basic ber of the other methods (e.g. Beck et al. 2009, Pine et al.
and extended numerical methods have been developed and 2006, Rogers et al. 2010);
applied. There has been some further development of pre- • coupled hydro-mechanical (Beck et al. 2010) and thermo-
viously existing analytical (e.g. Diederichs & Kaiser 1999, hydro-mechanical models (Detournay 1995, Hudson et al.
Sofianos et al. 1999) and empirical (e.g. Galvin et al. 1999, 2001, Stephansson et al. 1996); and
Hedley & Grant 1972, Salamon & Munro 1967) methods. In • inverse solution methods as used in back analysis in rock
relatively recent years, artificial intelligence, expert systems, engineering (see Section 7.6 below).
integrated intelligent, internet-based, neural network and
Despite the significant advances that have been made, it
other systems approaches have been developed and applied
must be recognised that the successful application of numer-
(e.g. Dershowitz & Einstein 1984, Feng & An 2004, Feng &
ical methods in rock engineering design analyses depends to
Hudson 2004, 2010).
a great extent on the geotechnical models, the constitutive
As noted in Section 4 above, the application of numeri-
models and boundary conditions developed from the site char-
cal methods, specifically the finite element method, to rock
acterisation data. Because of the difficulty of defining some
engineering began in the early to mid-1960s. Since that time,
of the input data deterministically, probabilistic or stochastic
the development and application of numerical methods of
methods are often used to represent the rock mass geometry,
analysis has been a major feature of rock mechanics and rock
the mechanical properties of rocks and rock masses, and in the
engineering research and practice. A comprehensive review
analyses themselves. When formal probabilistic or stochas-
of the formulation of numerical methods and their applica-
tic methods are not used, a range of input data may be used
tion in rock mechanics and rock engineering was presented
in sensitivity studies. As Starfield & Cundall (1988) pointed
by Jing (2003). Jing’s paper contained no less than 774 refer-
out, rock mechanics problems are data-limited problems that
ences to the published literature in the field to that date, but
cannot be modelled unambiguously.
obviously does not deal with more recent developments. Here,
As the numerical methods of design analysis outlined here
it will be possible to refer to only a very small number of the
were developed, high levels of expertise in the numerical meth-
great many important contributions made internationally to
ods themselves, and in their application in rock mechanics and
numerical modelling in rock mechanics and rock engineering.
rock engineering, were developed by a number of individuals
The main numerical methods were developed originally for
and groups internationally. However, it has been the writer’s
continuum applications but, from the 1960s, were adapted
experience that, despite the vast range of knowledge and expe-
to allow for the discontinuous nature of rock masses (e.g.
rience that is now available in this field, the application of these
Goodman et al. 1968, Wittke 1977), often treating them as
methods in engineering practice often suffers because some
equivalent continua. The fifth ISRM President, Professor
analysts regard the computer codes used as “black boxes” and
Walter Wittke of Germany, made significant contributions to
pay insufficient attention to the mechanics of the problems
the rock engineering applications of the finite element method.
concerned, the input data and to the meaning or “believabil-
Importantly, specific discontinuum methods of numerical
ity” of the results obtained. Furthermore, there is a tendency to
analysis have been developed. In rock mechanics and rock
disregard features of a problem that are not catered for specifi-
engineering, these various methods have been applied mainly
cally in the software selected or available for use. Although the
to stress and deformation analyses, but they have also been
paper was written more than 20 years ago, the writer considers
applied to the modelling of fracture processes and of fluid
that many of those seeking to use modern numerical methods
flow and heat transfer in rock masses. Following Jing (2003),
in rock engineering design analyses should pay greater atten-
the numerical methods that have been developed or adapted
tion to the guidance provided by Starfield & Cundall (1988),
for rock mechanics and rock engineering applications may be
especially the warning that numerical modelling is an aid to
classified as:
thought rather than a substitute for thinking.
• Finite Element (FEM) and related methods, including
meshless methods (Beck et al. 2009, 2010, Goodman et al.
7.5 Excavation and support and reinforcement
1968, Wittke 1977, 1990, Zienkiewicz 1977);
• Finite Difference Methods (FDM) including the Finite Vol- 7.5.1 Excavation
ume (FVM) approach, applied perhaps most notably in the In the last 50 years, significant advances have been made
well-known FLAC series of codes (Detournay & Hart 1999, in understanding the underlying mechanics and in develop-
Hart et al., 2008, Itasca 2011, Sainsbury, D.P. et al. 2011); ing the techniques and technologies associated with the two
• Boundary Element Methods (BEM) using direct and indi- generic methods of rock excavation – drilling-and-blasting
rect formulations (Beer & Watson 1992, Brady 1979, 1987, and mechanical excavation. In the 1950s and 1960s there was
Crouch & Starfield 1983); considerable research interest in the mechanics of percussive
• Discrete Element Methods (DEM) including the explicit drilling, rock fragmentation by blasting and the mechanics
or Distinct Element Method (Cundall 1971, 1987, Itasca of rock cutting, so much so that sessions on what came to
2011), the implicit or Discontinuous Deformation Analysis be called rock dynamics were usually included in the pro-
Method (Shi & Goodman 1985), key block theory grams of general rock mechanics symposia and conferences
36

(e.g. Fairhurst 1963). While these fundamental issues remain An essential concept in the mechanics of the support and
of concern to mainstream rock mechanics, it is the writer’s reinforcement of underground excavations is ground-support
impression that they have now become something of a special interaction as represented in ground-support interaction dia-
interest with their own specialist conferences, interest groups grams. This concept had been developed by the Austrian
and societies having been established. School (Pacher 1964, Rabcewicz 1969) but has since been
West (1988) traced the development of drilling and blasting extended through the convergence-confinement method to
and mechanical excavation technologies in the international which the 10th ISRM President, Marc Panet of France, made
tunnelling industry from the 1850s to the 1980s. The last 50 important contributions (Panet 1993, 1995), and a wide range
years have seen a range of developments in drilling and blast- of linear and non-linear closed-form and numerical solutions
ing technologies, many of them originating in Sweden. They for ground response curves and the characteristic curves for
include: support and reinforcing elements and systems (e.g. Carranza-
Torres & Fairhurst 1999, Daemen 1977). Despite the advances
• integrated drilling systems with electronic controls and a that have been made in incorporating ground support and its
range of drill support software for drill pattern design, effects into numerical analysis codes (e.g. Itasca 2011), as
navigation, automatic drilling and measurement while noted by Jing (2003), the numerical modelling of all aspects
drilling (MWD) (Schunnesson 2009); of support and reinforcement elements and systems, includ-
• large diameter blast hole drilling for both surface and ing their interfaces with the rock mass, still provides some
underground applications; challenges.
• new and improved explosive types for a variety of Considerable advances have been made in the develop-
applications; ment of support and reinforcement techniques and technology,
• electronic detonators; including several new rock bolt types, for example, those with
• blast monitoring systems to improve blasting operations yielding or dynamic capabilities for use in rock burst prone
(McKenzie 1987); environments (Falmagne & Simser 2004, Ortlepp 2007); cable
• blast vibration monitoring to assess environmental and bolts, particularly, but not exclusively, for application in the
other impacts (Dowding 1985); and mining industry (Hutchinson & Diederichs 1996, Windsor
• improved understanding of drilling and blasting processes 2004); shotcrete, particularly fibre-reinforced shotcrete and
through a range of industrial and research-level experi- the wide-spread use of the wet mix process (Bernard 2010);
mental, theoretical and computational studies (e.g. Furtney mesh and thin sprayed liners (Potvin et al. 2004); the corro-
et al. 2009, Minchinton & Dare-Bryan 2005, Ouchterlony & sion protection of rock bolts, cable bolts and ground anchors
Moser 2006). (Windsor 2004); the static and dynamic testing of support and
reinforcement elements (Player et al. 2008); and understand-
Building on a range of previous studies, Hustrulid (2010) ing the corrosion process in rock bolts and cables in corrosive
provides a detailed account of current blast design practice environments (Villaescusa et al. 2008). A range of design
for development drifts in hard rock underground mining, while methods for support and reinforcement systems have now been
Williams et al. (2009) give an account of blasting practices, developed (e.g. Barrett & McCreath 1995, Barton et al. 1994,
including controlled blasting, in large open pit mines. Hoek et al. 2008, Hutchinson & Diederichs 1995, Pells 2002)
In the area of mechanical excavation, new, larger and more to replace, or be used in conjunction with, the largely empir-
powerful machines of a range of types have been developed for ical methods that were used earlier in the 50-year period and
civil engineering and for soft rock mining, including coal min- still find wide-spread use today.
ing, but not as yet with any great success for hard rock mining.
The development of full-face tunnel boring machines (TBMs)
can be traced back to the mid-19th century but successful hard 7.6 Monitoring and back analysis
rock TBM tunnelling dates back only to the 1950s. The largest
Monitoring the performance of mining and civil excavations in
full-face tunnelling machines are now some 15 m in diameter
rock had been carried out for many years before the establish-
and can excavate in a range of ground conditions including
ment of the ISRM in 1962, and had become an integral part of
very strong rocks (Zhao & Gong 2006). Rock mechanics-
rock engineering practice through the observational method.
based data and methodologies are used in the selection of
Early monitoring used mechanical and optical, and then elec-
modern TBMs and in computer simulations of their perfor-
trical, electro-optical, electronic and microseismic techniques
mance (Barton 2000, Gong & Zhao 2009, Rojek et al. 2010,
(see, for example, Brady & Brown 2004, Dunnicliff 1988,
Zhao & Gong 2006). As with excavation by drilling and blast-
Franklin & Denton 1973, Kovari et al. 1979, Windsor 1993).
ing, improvements in the performance, automation and control
In open pit mining, for example, GPS, photogrammetry, laser
of TBMs and other types of mechanical excavators such as
scanning (LiDAR), radar and satellite imaging techniques
road headers and longwall coal shearers, are being achieved
and systems are now also used to monitor slope movements,
through industrial and research-level experimental, theoreti-
in particular (Girard & McHugh 2001, Hawley et al. 2009,
cal and computational studies (e.g. Rojek et al. 2010, Pichler
Herrera et al. 2010, Sakurai et al. 2009).
et al. 2010).
Subsequently, the on-line acquisition, storage, processing,
7.5.2 Support and reinforcement management, interpretation and reporting of rock engineer-
Following Windsor & Thompson (1993), a distinction will ing monitoring data followed the introduction of computer
be made between support and reinforcement, although the and digital technologies (Gilby & Socol 2010, Kimmance
two are often treated together as in the term ground support. 1999). Figure 5 illustrates the components of a current instru-
Support is the application of a reactive force to the surface of mentation data management and monitoring system. The
the excavation while reinforcement is a means of improving formalisation and power of these processes today and their
the overall rock mass properties from within the rock mass by potential to support well-informed decision making and risk
techniques such as rock and cable bolts. management systems (e.g. Akutagawa 2010, Dewynter et al.
37

Figure 5. Geotechnical instrumentation data interpretation and evaluation (GIDE) system components (Gilby & Socol 2010).
2010, Hawley et al. 2009, Schubert 2006) represents a great tunnels Sakurai (1997, 1999). Hoek (2001) discusses a simi-
advance on the comparable processes available 50 years ago, lar approach based on the normalised radial displacement at
although the purposes and principles of good monitoring the tunnel boundary.
programs remain essentially unchanged (e.g. Franklin 1977,
Kovari & Amstad 1993). There can be little doubt that modern
7.7 Rock engineering applications
monitoring systems have contributed significantly to the rock
engineering achievements to be outlined in Section 7.7 below. Over the last 50 years, the pre-existing rock mechanics knowl-
One of the common objectives of monitoring formalised edge outlined in Section 4, and the further development of that
by the seventh ISRM President, Dr John Franklin of Canada knowledge outlined thus far in Section 7, have contributed
(Franklin 1977), is to check the validity of the assumptions, to some remarkable rock engineering achievements in the
conceptual models and values of rock mass properties used general area of earth resources engineering (Fairhurst 2011),
in design calculations. Originally, and often still today, these including civil, mining, petroleum, energy resources and envi-
checks were made, and values of rock mass properties and the ronmental engineering. Space limitations preclude reference
in situ stresses estimated, using a process of trial and error or to all but a very few of these achievements.
curve fitting using forward analyses based on the theory of In civil engineering, the scale of rock engineering projects
elasticity. However, in the late 1970s and early 1980s, it was has steadily increased in terms of project dimensions, costs
realised that it was possible to improve this process by devel- and environmental significance (Feng & Hudson 2004). We
oping and applying more formal back analysis techniques. might think, for example, of achievements such as the 62 m
These techniques used essentially two approaches, the inverse span cavern for the Olympic Ice Hockey stadium at Gjøvik,
and the identification or calibration methods (Gioda 1980). Norway (Barton et al. 1994); the helical underground car
In the earliest use of the inverse method in rock mechanics park for the Sydney Opera House, Australia, possibly the
known to the author, Kirsten (1976) used measured displace- widest shallow-cover rock cavern in the world (Pells 2002,
ments in a slope and displacements calculated using the FEM Pells et al. 1991); the Three Gorges Dam and the associ-
to back calculate values of the elastic moduli. ated ship locks and hydro-electric developments in China
The ninth ISRM President, Professor Shunsuke Sakurai of (Feng & Hudson 2004, Liu et al. 2003a,b); and the long
Japan, was heavily involved in the development of back anal- European alpine tunnels such as the Lötschberg and Gotthard
ysis methods early in their history (Sakurai & Takeuchi 1983) base tunnels excavated under difficult conditions including
and subsequently (Sakurai & Akutagawa 1995, Sakurai et al. squeezing and faulted ground and brittle rock under high
2009). Although formal back analysis methods were applied stress at great depth (Hagedorn et al. 2007, Rojat et al.
originally to 2D linear elastic problems, the theory was soon 2008).
extended to the three dimensional and non-linear elasto-plastic In mining engineering, greater numbers of larger, deeper
problems encountered in rock engineering. A wide range of and more productive open pit mines with optimised slope
innovative techniques for solving the sets of non-linear equa- angles have been designed and operated effectively using a
tions involved have since been developed (Feng et al. 2004, range of advances in rock mechanics and rock slope engi-
Gioda & Sakurai 1987, Sakurai 1993). neering dating back to the seminal work of Hoek & Bray
Professor Sakurai has also emphasised the importance of (1974) and earlier. Read & Stacey (2009) provide a detailed
monitoring the performance of excavations in rock for pur- account of the state of this particular art. In underground
poses other than the use of the measurement data in back metalliferous mining over the last 50 years, rock mechanics
analyses. For example, in order to assess the stability of tun- has contributed significantly to the mining of highly stressed,
nels from measured displacements, he proposed the concept brittle, tabular orebodies in the deep-level gold mines of
of critical direct and shear strains, and on the basis of the South Africa (Cook et al. 1966, Ryder & Jager 2002), and
critical strain, proposed hazard warning levels for the mea- to the development and implementation firstly of cut-and-
sured displacements as well as for the measured strains in fill and then of long hole open stoping and bench stoping
38

methods (Brady & Brown 2004, Villaescusa 1996, 2008). ACKNOWLEDGEMENTS
Most importantly, the last 20 years has seen an increased
emphasis on underground mass mining methods using highly The writer wishes to record his appreciation of the support
productive and cost effective sub-level and block and panel and assistance of the following members of the international
caving methods at greater depths and in stronger rocks (Brown rock mechanics community in providing him with infor-
2007a, b, Chitombo 2010). Caving processes and cave propa- mation and material used in the preparation of this paper:
gation are now being modelled successfully (Sainsbury, B.L. Nick Barton, David Beck, Pierre Berest, Maria de Lurdes
et al. 2011). In underground coal mining, larger and more pro- Eusébio, Charles Fairhurst, Xia-Ting Feng, John Gilby, Nuno
ductive longwall faces have been introduced as has longwall Grossmann, Pierre Habib, Evert Hoek, John Hudson, Luís
top coal caving to mine thicker seams, especially in China Lamas, Italo Oñederra, Matt Pierce, Eda Quadros, Bre-Anne
(Alehossein & Paulsen 2010, Vakili & Hebblewhite 2011). Sainsbury, David Sainsbury, Shun Sakurai, Penny Stewart
Rock mechanics has become increasingly important in and Walter Wittke. He also wishes to thank Rob Morphet
petroleum engineering since the 1970s as production has and the management of Golder Associates, Brisbane, for
taken place from deeper and more difficult geological set- their support during the preparation of this paper and Jillian
tings (Roegiers 1999). Rock mechanics has contributed to the Roche for her skilful assistance in readying the paper for
achievements of the petroleum and gas industry in areas such publication.
as geological and fracture modelling, including DFN mod-
elling; seismic tomography and other geophysical techniques;
borehole mechanics and wellbore stability; the measure- REFERENCES
ment of in situ stresses; hydraulic fracturing and methods of
enhanced oil recovery; poromechanics and coupled thermo- Akutagawa, S. 2010. On site visualisation as a new paradigm
hydro-mechanical processes; the control of solids (generally for field measurement in rock engineering. In K.G. Sharma,
sand) production; smart drilling including drilling deviated T. Ramamurthy, V.K. Kanjlia & A.C. Gupta (eds), Advances in
Rock Engineering, Proc. ISRM Int. Symp. 2010 & 6th Asian Rock
holes and drilling in deep water; and reservoir geomechanics Mech. Symp., New Delhi, 23–27 October, KN 34–KN 45. New
(Detournay 1995, Fjaer et al. 2008, Roegiers 1999, Schutjens Delhi: Central Board of Irrigation & Power.
2009). Rock mechanics has also contributed to a number of Alehossein, H. & Paulsen, B.A. 2010. Stress analysis of longwall
other achievements in the exploitation of a range of energy top caving. International Journal of Rock Mechanics & Mining
resources, including hot, dry rock geothermal energy at depth Sciences, 47(1): 30–41.
(Thorsteinsson et al. 2008), the underground storage of oil Alexander, L.G. 1960. Field and laboratory tests in rock mechan-
and gas (You et al. 2006), and underground radioactive waste ics. Proc. 3rd Australia-New Zealand Conf. Soil Mechanics &
repository planning and design (Fairhurst 2004, Hudson 2010, Foundation Engineering, Sydney, 161–168. Sydney: Institution
Hudson et al. 2001), and the deep injection of solid and liquid of Engineers, Australia.
wastes (Dusseault 2010). Conversely, research in these areas, Amadei, B. & Stephansson, O. 1997. Rock Stress and its Measure-
ment, 512 p. Dordrecht: Kluwer.
including through field trials, has contributed to basic and Anon, 1963. Proceedings, Rock Mechanics Symposium, McGill
applied rock mechanics knowledge. University, Montreal, 7–8 September, 1962, 218 p. Ottawa: Mines
Branch, Department of Mines & Technical Surveys.
Appleton, E. 1944. Scientific research: fundamentals of its appli-
8 CONCLUSIONS cation to rock mechanics. South African Mining & Engineering
Journal, 55(1): 519–520.
Atkinson, B.K. (ed.) 1987. Fracture Mechanics of Rock, 534 p.
This necessarily brief overview of the development and London: Academic Press.
achievements of the ISRM and of the discipline of rock Barla, G., Barla, M., Camusso, M. & Martinotti, M.E. 2007. Setting
mechanics and rock engineering in the 50 years since the up a new direct shear testing apparatus. In L. Ribeiro e Sousa,
ISRM was founded in 1962 shows clearly that much of sig- C. Olalla & N.F. Grossmann (eds), Proc. 11th Congr., Int. Soc.
nificance has been achieved. New basic understandings and Rock Mech., Lisbon, 9–13 July, 1: 415–418. Leiden: Taylor &
engineering tools have been developed to support an increas- Francis.
ingly wide range of rock engineering applications. Many of Barrett, S.V.L. & McCreath, D.R. 1995. Shotcrete support design in
the advances made in rock mechanics research, in the tools and blocky ground: towards a deterministic approach. Tunnelling &
techniques now available for data collection and analysis, and Underground Space Technology, 10(1): 79–89.
in the application of rock mechanics in engineering practice, Barton, N. 2000. TBM Tunnelling in Jointed and Faulted Rock, 172
p. Rotterdam: Balkema.
have been enabled by modern digital computing, informa- Barton, N., 2006. Rock Quality, Seismic Velocity, Attenuation and
tion and communications technologies. Others have drawn on Anisotropy, 729 p. London: Taylor & Francis.
advances made in other branches of science and technology. Barton, N. & Bandis, S.C. 1990. Review of predictive capabilities
And during this first 50 years of its life, the ISRM has pursued of JRC-JCS model in engineering practice. In N.R. Barton &
its stated objectives and purposes conscientiously by carrying O. Stephansson (eds), Rock Joints, Proc. Int. Symp. on Rock Joints,
out all of its main activities at least satisfactorily, with some Loen, 4–6 June, 603–610. Rotterdam: Balkema.
being carried out to considerable effect. Barton, N., By, T.L., Chryssanthakis, P., Tunbridge, L., Kristiansen, J.,
For those involved, the last 50 years has been an exciting, Løset, F., Bhasin, R.K., Westerdahl, H. & Vick, G. 1994. Pre-
satisfying and possibly fortuitous time in which to have been dicted and measured performance of the 62 m span Norwegian
involved in rock mechanics and the ISRM. However, increases Olympic Ice Hockey Cavern at Gjøvik. International Journal of
Rock Mechanics & Mining Sciences, 31(6): 617–641.
in knowledge in our and other disciplines, and change and Barton, N. & de Quadros, E.F. 1997. Joint aperture and roughness in
further development of our institutions and in society at large, the prediction of flow and groutability of rock masses. In K. Kim
are inevitable. From a rock mechanics and rock engineering (ed.), Linking Science to Rock Engineering, Proc. 36th U.S. Rock
perspective, the future remains full of promise, challenge and Mech. Symp. & ISRM Int. Symp., New York, 29 June–2 July, 2:
excitement. 907–916. New York: Columbia University.
39

Barton, N., Lien, R. & Lunde, J. 1974. Engineering classification Bryan, A., Bryan, J.G. & Fouché, J. 1964. Some problems of strata
of rock masses for the design of tunnel support. Rock Mechanics, control and support in pillar workings. The Mining Engineer,
6(4): 189–236. 123(41): 238–266.
Beck, D.A., Fillery, B. & Reusch, F. 2010. 3d hydro-mechanical Carranza-Torres, C. & Fairhurst, C. 1999. The elasto-plastic response
simulation of faulted open pit slopes. Proc. 44th U.S. Rock Mech. of underground excavations in rock masses that satisfy the
Symp. & 5th U.S.-Canada Rock Mech. Symp., Salt Lake City, Hoek-Brown failure criterion. International Journal of Rock
27–30 June, Paper ARMA-425, 6p. (on CD-ROM). Mechanics & Mining Sciences, 36(6): 777–809.
Beck, D.A., Pfitzner, M.J., Arndt, S.M. & Fillery, B. 2009. Esti- Chen, Y., Liu, Y., Zhang, M. & Chen, R. 1986. A preliminary study
mating rock mass properties and seismic response using higher of in-situ stress field in the Three-Gorge Dam area. Proc. Int.
order, discontinuous, Finite Element models. In M. Diederichs & Symp. on Engineering in Complex Rock Formations, Beijing,
G. Grasselli (eds), Rock Engineering in Difficult Conditions, 3–7 November, 149–162. Beijing: Science Press.
Proc. 3rd Canada-U.S. Rock Mech. Symp. & 20th Canadian Rock Chitombo, G.P. 2010. Cave mining – 16 years after Laubscher’s 1994
Mech. Symp., Toronto, 9–15 May, 12 p. (on CD-ROM). paper “cave mining – state of the art”. In Y. Potvin (ed.), Caving
Beer, G. (ed.) 2010. Technology Innovation in Underground Con- 2010, Proc. 2nd Int. Symp. on Block & Sublevel Caving, Perth,
struction, 504 p. Leiden: CRC Press/Balkema. 20–22 April, 45–61. Perth: Australian Centre for Geomechanics.
Beer, G. & Watson, J.O. 1992. Introduction to Finite and Boundary Clough, R.W. 1960. The finite element method in plane stress analy-
Element Methods for Engineers, 509 p. Chichester: John Wiley. sis. Proc. 2nd ASCE Conf. on Electronic Computation, Pittsburgh,
Bernaix, J. 1969. New laboratory methods of studying the mechanical September, 345–378.
properties of rock. International Journal of Rock Mechanics & Cook, N.G.W. 1965. The failure of rock. International Journal of
Mining Sciences, 6(1): 43–90. Rock Mechanics & Mining Sciences, 2(4): 389–403.
Bernard, S. (ed.) 2010. Shotcrete: Elements of a System, Proc. 3rd Int. Cook, N.G.W., Hoek, E., Pretorius, J.P.G., Ortlepp, W.D. & Salamon,
Conf. on Engineering Developments in Shotcrete, Queenstown, M.D.G. 1966. Rock mechanics applied to the study of rockbursts.
15–17 March, 310 p. Leiden: CRC Press/Balkema. Journal, South African Institute of Mining & Metallurgy, 66(10):
Bieniawski, Z.T. 1976. Rock mass classifications in rock engineer- 436–528.
ing. In Z.T. Bieniawski (ed.), Proc. Symp. on Exploration for Coulomb, C.A. 1776. Essai sur une application des régles de
Rock Engineering, Johannesburg, 1–5 November, 1: 97–106. Cape maximis et minimis à quelques problèmes de statique, rela-
Town: Balkema. tives à l’architecture. Mémoires de Mathématique et de Physique,
Bieniawski, Z.T. 1989. Engineering Rock Mass Classification, 251 l’Académie Royale des Sciences, 7: 343–382.
p. New York: John Wiley. Crouch, S.L. & Starfield, A.M. 1983. Boundary Element Methods in
Billaux, D., Chiles, J.P., Hestir, K. & Long, J. 1989. Three- Solid Mechanics, 322 p. London: Allen & Unwin.
dimensional statistical modelling of a fractured rock mass – an Cundall, P.A. 1971. A computer model for simulating progressive,
example from the Fanay-Augères Mine. International Journal of large-scale movements in blocky rock systems. In Rock Fracture,
Rock Mechanics & Mining Sciences, 26(3–4): 281–299. Proc. Int. Symp. Rock Mech., Nancy, 4–6 October, 1: Paper II-8.
Blanks, R.F. & McHenry, D. 1945. Large triaxial testing machine built Cundall, P.A. 1987. Distinct element models of rock and soil struc-
by Bureau of Reclamation. Engineering News Record, 135(6): ture. In E.T. Brown (ed.), Analytical & Computational Methods
171–172. in Engineering Rock Mechanics, 129–163. London: Allen &
Brady, B.H.G. 1979. A direct formulation of the boundary element Unwin.
method of stress analysis for complete plane strain. Interna- Cundall, P.A. 2011. Lattice method for modelling brittle, jointed rock.
tional Journal of Rock Mechanics & Mining Sciences, 16(4): 235– In D.P. Sainsbury, R.D. Hart, C.L. Detournay & M.J. Nelson (eds),
244. Continuum and Distinct Element Numerical Modeling in Geome-
Brady, B.H.G. 1987. Boundary element and linked methods for chanics – 2011, Proc. 2nd Int. FLAC/DEM Symp., Melbourne,
underground excavation design. In E.T. Brown (ed.), Analyti- 14–16 February, 11–19. Minneapolis: Itasca International Inc.
cal & Computational Methods in Engineering Rock Mechanics, Cundall, P.A. & Damjanac, B. 2009. A comprehensive 3D model for
164–204. London: Allen & Unwin. rock slopes based on micromechanics. In Slope Stability 2009,
Brady, B.H.G. & Brown, E.T. 2004. Rock Mechanics for Under- Proc. 3rd Int. Symp. on Rock Slope Stability, Santiago, 9–11
ground Mining, 3rd edn, 628 p. Dordrecht: Kluwer. November, 10 p (on CD-ROM).
Brown, E.T. (ed.) 1981. Rock Characterization, Testing and Daemen, J.J.K. 1977. Problems in tunnel support mechanics. Under-
Monitoring – ISRM Suggested Methods, 211 p. Oxford: Pergamon ground Space, 1(3): 163–172.
Press. Dershowitz, W. 1995. Interpretation and synthesis of discrete fracture
Brown, E.T. 1999. Rock mechanics and the Snowy Mountains orientation, size, shape, spatial structure and hydrologic data by
Scheme. In The Spirit of the Snowy Fifty Years On, Proc. 1999 forward modelling. In L.R. Myer, N.G.W. Cook, R.E. Goodman &
Invitation Symp., Australian Academy of Technological Sciences C-F. Tsang (eds), Proc. Conf. on Fractured & Jointed Rock Masses,
& Engineering, Cooma, 23–24 November, 89–101. Melbourne: Lake Tahoe, 3–5 June 1992, 579–586. Rotterdam: Balkema.
Australian Academy of Technological Sciences & Engineering. Dershowitz, W.S. & Einstein, H.H. 1984. Applications of artificial
Brown, E.T. 2002. Rock mechanics in Australia. International intelligence to problems of rock mechanics. In C.H. Dowding &
Journal of Rock Mechanics & Mining Sciences, 39(5): 529–538. M.M. Singh (eds), Rock Mechanics in Productivity & Protection,
Brown, E.T. 2007a. Block Caving Geomechanics, 2nd edn, 696 p. Proc. 25th U.S. Symp. Rock Mech., Evanston, 25–27 June, 483–
Brisbane: JKMRC. 494. New York: A.I.M.E.
Brown, E.T. 2007b. Rock mechanics – the basic mining science: Dershowitz, W.S. & Einstein, H.H. 1988. Characterizing rock joint
challenges in underground mass mining. In L. Ribeiro e Sousa, geometry with joint system models. Rock Mechanics & Rock
C. Olalla & N.F. Grossmann (eds), Proc. 11th Congr., Int. Soc. Engineering, 21(1): 21–51.
Rock Mech., Lisbon, 9–13 July, 3: 1335–1347. Leiden: Taylor & Desseault, M.B. 2010. Deep injection disposal: environmental
Francis. and petroleum geomechanics. In K.G. Sharma, T. Ramamurthy,
Brown, E.T. 2008. Estimating the mechanical properties of rock V.K. Kanjlia & A.C. Gupta (eds), Advances in Rock Engineer-
masses. In Y. Potvin, J. Carter, A. Dyskin & R. Jeffrey (eds), Proc. ing, Proc. ISRM Int. Symp. 2010 & 6th Asian Rock Mech. Symp.,
1st Southern Hemisphere Int. Rock Mech. Symp., Perth, 16–19 New Delhi, 23–27 October, KN 21–KN 33. New Delhi: Central
September, 1: 3–22. Perth: Australian Centre for Geomechanics. Board of Irrigation & Power.
Brown, P.D. & Robertshaw, J. 1953. The in-situ measurement of Detournay, C. & Hart, R. (eds) 1999. FLAC and Numerical Model-
Young’s Modulus for rock by a dynamic method. Géotechnique, ing in Geomechanics, Proc. Int. FLAC Symp., Minneapolis, 1–3
3(7): 283–286. September, 528 p. Rotterdam: Balkema.
40

Detournay, E. 1995. Coupled thermo-hydro-mechanical processes in intelligent methodology. International Journal for Numerical &
rock mechanics, with applications to the petroleum industry. In T. Analytical Methods in Geomechanics, 28(11): 1141–1165.
Fujii (ed.), Proc. 8th Congr., Int. Soc. Rock Mech., Tokyo, 25–29 Fjaer, E., Holt, R.M., Horsrud, P., Raaen, A.M. & Risnes, R. 2008.
September, 3: 1061–1068. Rotterdam: Balkema. Petroleum Related Rock Mechanics, 2nd edn, 514 p. Amsterdam:
Dewynter, V., Magne, S., Rougeault, S., Ferdinand, P., Vallon, F., Elsevier.
Avallone, L., Vacher, E., de Broissia, M., Canepa, C. & Poulain, A. Franklin, J.A. 1977. Monitoring of structures in rock. International
2009. Optical fiber sensing cable for underground settlement Journal of Rock Mechanics & Mining Sciences, 14(4): 163–192.
monitoring during tunnelling. In G. Beer (ed.), Technology Inno- Franklin, J.A. & Denton, P.E. 1973. The monitoring of rock slopes.
vation in Underground Construction, 163–188. Leiden: CRC Quarterly Journal of Engineering Geology, 6(3–4): 259–286.
Press/Balkema. Furtney, J.K., Cundall, P.A. & Chitombo, G.P. 2009. Developments in
Diederichs, M.S. & Kaiser, P.K. 1999. Stability of large excavations in numerical modeling of blast induced fragmentation: updates from
laminated hard rock masses: the voussoir analogue revisited. Inter- the HSBM project. In J.A. Sanchidrián (ed.), Rock Fragmentation
national Journal of Rock Mechanics & Mining Sciences, 36(1): by Blasting – FragBlast 9, Proc. 9th Int. Symp. on Rock Fragmen-
97–117. tation by Blasting, Granada, 13–17 September, 335–342. Leiden:
Dowding, C.H. 1985. Blast Vibration Monitoring and Control, 297 p. CRC Press/Balkema.
Englewood Cliffs: Prentice-Hall. Gaich, A., Potsch, M. & Schubert, W. 2007. Rock mass characteri-
Dunnicliff, J. 1988. Geotechnical Instrumentation for Monitoring zation for tunneling and mining using 3D images. In L. Ribeiro
Field Performance, 577 p. New York: John Wiley. e Sousa, C. Olalla & N.F. Grossmann (eds), Proc. 11th Congr.,
Elsworth, D. 1986. A hybrid boundary element-finite element anal- Int. Soc. Rock Mech., Lisbon, 9–13 July, 2: 1019–1024. Leiden:
ysis procedure for fluid flow simulation in fractured rock masses. Taylor & Francis.
International Journal for Numerical & Analytical Methods in Galvin, J.M., Hebblewhite, B.K. & Salamon, M.D.G. 1999. Uni-
Geomechanics, 10(6): 569–584. versity of New South Wales pillar strength determinations for
Elsworth, D. & Mase, C.R. 1993. Groundwater in rock engineer- Australian and South African mining conditions. In C. Mark,
ing. In J.A. Hudson, E.T. Brown, C. Fairhurst & E. Hoek (eds), K.A. Heasley, A.T. Iannocchione & R.J. Tuchman (eds), Proc. 2nd
Comprehensive Rock Engineering, 1: 201–226. Oxford: Pergamon Int. Workshop on Coal Pillar Mechanics & Design, Vail, 6 June,
Press. NIOSH IC 9448, 63–71. Pittsburgh: U.S. Department of Health &
Evison, F.F. 1953. The seismic determination ofYoung’s modulus and Human Services.
Poisson’s ratio for rocks in situ. Géotechnique, 6(3): 118–123. Gilby, J. & Socol, M. 2010. Geotechnical instrumentation data
Fairhurst, C. (ed.) 1963. Proceedings, Fifth Symposium on Rock management. In K.G. Sharma, T. Ramamurthy, V.K. Kanjlia &
Mechanics, University of Minnesota, Minneapolis, May 1962, 726 A.C. Gupta (eds), Advances in Rock Engineering, Proc. ISRM Int.
p. Oxford: Pergamon Press. Symp. 2010 & 6th Asian Rock Mech. Symp., New Delhi, 23–27
Fairhurst, C. 2003. Stress estimation in rock: a brief history and October, Paper 124, 7p. (on CD-ROM). New Delhi: Central Board
review. International Journal of Rock Mechanics & Mining of Irrigation & Power.
Sciences, 40(7–8): 957–973. Gioda, G. 1980. Indirect identification of the average elastic char-
Fairhurst, C. 2004. Nuclear waste disposal and rock mechan- acteristics of rock masses. In P.J.N. Pells (ed.), Structural Foun-
ics: contributions of the Underground Research Laboratory dations on Rock, Proc. Int. Symp., Sydney, 7–9 May, 65–73.
(URL), Pinawa, Manitoba, Canada. International Journal of Rock Rotterdam: Balkema.
Mechanics & Mining Sciences, 41(8): 1221–1227. Gioda, G. & Sakurai, S. 1987. Back analysis procedures for the inter-
Fairhurst, C. 2010. First Vienna-Leopold-Müller Lecture: What is pretation of field measurements in geomechanics. International
the strength of a rock mass? Progress in answering Müller’s Journal for Numerical & Analytical Methods in Geomechanics,
(implicit) question. Proc. 5th Colloquium, Rock Mechanics – 11(6): 555–583.
Theory & Practice, Vienna, 26–27 November 2009, Mitteilungen Girard, J.M. & McHugh, E.L. 2001. Emerging technologies and the
für Ingenieurgeologie und Geomechanik, Band 9: 87–110. future of geotechnical instrumentation. In D. Elsworth, J.P. Tinucci
Fairhurst, C. 2011. Grand challenges in Earth Resources & K.A. Heasley (eds), Rock Mechanics in the National Interest,
Engineering – and some implications for rock mechanics and Proc. 38th U.S. Symp. Rock Mech., Washington D.C., 5–7 July, 1:
rock engineering. In D.P. Sainsbury, R.D. Hart, C.L. Detournay & 445–450. Lisse: Balkema.
M.J. Nelson (eds), Continuum and Distinct Element Numerical Golder, H.Q. & Akroyd, T.N.W. 1954. An apparatus for triaxial-
Modeling in Geomechanics – 2011, Proc. 2nd Int. FLAC/DEM compression tests at high pressures. Géotechnique, 4(4): 131–136.
Symp., Melbourne, 14–16 February, 3–10. Minneapolis: Itasca Gong, Q.M. & Zhao, J. 2009. Development of rock mass charac-
International Inc. teristics model for TBM penetration rate prediction. International
Falmagne, V. & Simser, B.P. 2004. Performance of rockburst support Journal of Rock Mechanics & Mining Sciences, 46(1): 8–18.
systems in Canadian mines. In E. Villaescusa & Y. Potvin (eds), Goodman, R.E. 1966. On the distribution of stresses around cir-
Ground Support in Mining & Underground Construction, Proc. cular tunnels in non-homogeneous rocks. Proc. 1st Congr., Int.
5th Int. Symp. on Ground Support, Perth, 28–30 September, 313– Soc. Rock Mech., Lisbon, 25 September – 1 October, 2: 249–255.
318. Leiden: Balkema. Lisbon: LNEC.
Feng, X-T. & An, H. 2004. Hybrid intelligent method optimization Goodman, R.E. 1989. Introduction to Rock Mechanics, 2nd edn, 562
of a soft rock replacement scheme for a large cavern excavated in p. New York: John Wiley.
alternate hard and soft rock strata. International Journal of Rock Goodman, R.E. & Shi, G-H. 1985. Block Theory and its Application
Mechanics & Mining Sciences, 41(4): 655–667. to Rock Engineering, 338p. Englewood Cliffs: Prentice-Hall.
Feng, X-T. & Hudson, J.A. 2004.The ways ahead for rock engineering Goodman, R.E., Taylor, R.L. & Brekke, T.L. 1968. A model for the
design methodologies. International Journal of Rock Mechanics mechanics of jointed rock. Journal, Soil Mechanics & Foundations
& Mining Sciences, 41(2): 255–273. Division, A.S.C.E., 94(SM3): 637–659.
Feng, X-T. & Hudson, J.A. 2010. Specifying the information required Grasselli, G. & Egger, P. 2003. Constitutive law for the shear strength
for rock mechanics modelling and rock engineering design. of rock joints based on three-dimensional surface parameters.
International Journal of Rock Mechanics & Mining Sciences, International Journal of Rock Mechanics & Mining Sciences,
47(2): 179–194. 40(1): 25–40.
Feng, X-T. & Hudson, J.A. 2011. Rock Engineering Design, 464 p. Griggs, D.T. 1936. Deformation of rocks under high confining
Leiden: CRC Pres/Balkema. pressures. Journal of Geology, 44: 541–577.
Feng, X-T., Zhao, H. & Li, S. 2004.A new displacement back analysis Grobler, H.P., Poropat, G. & Guest, A.R. 2003. Photogramme-
to identify mechanical geo-material parameters using integrated try for structural mapping in mining. Technology Roadmap
41

for Rock Mechanics, Proc. 10th Congr., Int. Soc. Rock Mech., Hood, M. & Brown, E.T. 1999. Mining rock mechanics, yesterday,
Johannesburg, 8–12 September, 1: 431–436. Johannesburg: South- today and tomorrow. In G. Vouille and P. Berest (eds), Proc. 9th
ern African Institute of Mining & Metallurgy. Congr., Int. Soc. Rock Mech., Paris, 25–28 August, 3: 1551–1576.
Habib, P. 1950. Détermination du module d’élasticité des roches en Lisse: Balkema.
place. Annales de l’Institut Technique du Bâtiment et des Travaux Hudson, J.A. 2010. Underground radioactive waste disposal: the
Publics, No 145, September: 27–35. rock mechanics contributions. In K.G. Sharma, T. Ramamurthy,
Habib, P. & Marchand, R. 1952. Mesures des pressions de terrains V.K. Kanjlia & A.C. Gupta (eds), Advances in Rock Engineer-
par l’essai de vérin plat. Suppléments aux Annales de l’Institut ing, Proc. ISRM Int. Symp. 2010 & 6th Asian Rock Mech. Symp.,
Technique du Bâtiment et des Travaux Publics, Série Sols et New Delhi, 23–27 October, KN 3–KN 20. New Delhi: Central
Foundations, No 58, October: 967–971. Board of Irrigation & Power.
Hagedorn, H., Rehbock-Sander, M. & Stadelmann, R. 2007. Hudson, J.A. 2011. The next 50 years of the ISRM and anticipated
Gotthard Base tunnel: rock burst phenomenon during construc- future progress in rock mechanics. In Q. Qian & Y.X. Zhou (eds),
tion of a multifunctional section in a fault zone area. In L. Ribeiro Harmonising Rock Engineering and the Environment, Proc. 12th
e Sousa, C. Olalla & N.F. Grossmann (eds), Proc. 11th Congr., Int. Congr., Int. Soc. Rock Mech., Beijing, 16–21 October, 1: (this
Soc. Rock Mech., Lisbon, 9–13 July, 2: 917–920. Leiden: Taylor & volume). Leiden: CRC Press/Balkema.
Francis. Hudson, J.A., Brown, E.T. & Fairhurst, C. 1971. Optimizing the
Handin, J. 1953. An application of high pressure geophysics: control of rock failure in servo-controlled laboratory tests. Rock
experimental rock mechanics. Transactions, American Society of Mechanics, 3(4): 217–224.
Mechanical Engineers, 75: 315–324. Hudson, J.A., Brown, E.T., Fairhurst, C. & Hoek, E. (eds) 1993.
Hart, R.D., Detournay, C.L. & Cundall, P.A. (eds), 2008. Con- Comprehensive Rock Engineering, 5 Vols. Oxford: Pergamon
tinuum and Distinct Element Modeling in Geo-Engineering, Press.
Proc. 1st FLAC/DEM Symp., Minneapolis, 25–27 August, 698 Hudson, J.A., Cornet, F.H. & Christiansson, R. 2003. ISRM Sug-
p. Minneapolis: Itasca Consulting Group Inc. gested methods for rock stress estimation – Part 1: Strategy for
Hast, N. 1958. The measurement of rock pressure in mines. Sveriges rock stress estimation. International Journal of Rock Mechanics &
Geologiska Undersokning, Series C, 52(3). Mining Sciences, 40(7–8): 991–998.
Hawley, M., Marisett, S., Beale, G. & Stacey, P. 2009. Performance Hudson, J.A. & Feng, X-T. 2007. Updated flow charts for rock
assessment and monitoring. In J. Read & P. Stacey (eds), Guide- mechanics modelling and rock engineering design. International
lines for Open Pit Slope Design, 327–379. Melbourne: CSIRO Journal of Rock Mechanics & Mining Sciences, 44(2): 174–195.
Publishing. Hudson, J.A., Stephansson, O., Andersson, J., Tsang, C-F. & Jing, L.
Hedley, D.G.F. & Grant, F. 1972. Stope and pillar design at the Elliot 2001. Coupled T-H-M issues relating to radioactive waste repos-
Lake uranium mines. Canadian Institute of Mining & Metallurgy itory design and performance. International Journal of Rock
Bulletin, 65(723): 37–44. Mechanics & Mining Sciences, 38(1): 143–161.
Herrera, G., Tomás, R., Vicente, F., Lopez-Sanchez, J.M., Mallorquí, Hustrulid, W. 2010. Some comments regarding development drifting
J.J. & Mulas, J. 2010. Mapping ground movements in open pit practices with special emphasis on caving applications. InY. Potvin
mining areas using differential SAR interferometry. International (ed.), Caving 2010, Proc 2nd Int. Symp. on Block & Sublevel
Journal of Rock Mechanics & Mining Sciences, 47(7): 1114– Caving, Perth, 20–22 April, 3–43. Perth: Australian Centre for
1125. Geomechanics.
Heyman, J. 1972. Coulomb’s Memoir on Statics, 212p. Cambridge: Hutchinson, D.J. & Diederichs, M.S. 1996. Cablebolting in Under-
Cambridge University Press. ground Mines, 406 p. Richmond, B.C.: Bitech Publishers.
Hill, F.G. 1954. An investigation of the problem of rockbursts; an Indraratna, B., Oliveira, D.A.F. & Brown, E.T. 2010. A shear-
operational research project. Part I. The approach to the problem displacement criterion for soil-infilled rock discontinuities.
and analyses of the rockbursts that have occurred on the ERPM Géotechnique, 60(8): 623–633.
during the years 1948–1953. Journal, Chemical, Metallurgical & Itasca, 2011. FLAC Ver 6.0, FLAC3D Ver 4.0, FLAC/Slope Ver 6.0,
Mining Society of South Africa, 55(October): 63–83. UDEC Ver 5.0 & 3DEC Ver 4.1 Minneapolis: Itasca Consulting
Hill, F.G. 1966. Preface to “Rock Mechanics applied to the study Group Inc.(See www.itascacg.com/home.php.)
of rockbursts”. Journal, South African Institute of Mining & Jaeger, C. 1955. Present trends in the design of pressure tunnels
Metallurgy, 66(10): 435. and shafts for underground hydro-electric power stations. Proc.,
Hoek, E. 1963. Experimental study of rock-stress problems in deep Institution of Civil Engineers, 4(2): 116–174.
level mining. Experimental Mechanics, 3: 177–194. Jaeger, C. 1972. Rock Mechanics and Engineering, 417 p.
Hoek, E. 1964. Fracture of anisotropic rock. Journal, South African Cambridge: Cambridge University Press.
Institute of Mining & Metallurgy, 64(10): 501–518. Jaeger, J.C. 1959. The frictional properties of joints in rock. Geofisica
Hoek, E. 1994. Strength of rock and rock masses. ISRM News Pura e Applicata, 43(Part 2): 148–158.
Journal, 2(2): 4–16. Jaeger, J.C. 1960. Shear fracture of anisotropic rocks. Geological
Hoek, E. 2001. Big tunnels in bad rock. The 36th Karl Terzaghi Magazine, 97: 65–72.
Lecture. Journal of Geotechnical & Geoenvironmental Engineer- Jaeger, J.C. 1971. Friction of rocks and stability of rock slopes.
ing, A.S.C.E., 127(9): 726–740. Géotechnique, 21(2): 97–134.
Hoek, E. 2007. The development of rock engineering. In Practical Jing, L. 2003. A review of techniques, advances and outstand-
Rock Engineering. Available at http://Rocscience.com/hoek/pdf/ ing issues in numerical modelling for rock mechanics and rock
1_The_development_of_rock_engineering.pdf. engineering. International Journal of Rock Mechanics & Mining
Hoek, E. & Bray, J.W. 1974. Rock Slope Engineering, 309 p. London: Sciences, 40(3): 283–353.
Institution of Mining & Metallurgy. John, K.W. 1962. An approach to rock mechanics. Journal, Soil
Hoek, E. & Brown, E.T. 1980. Underground Excavations in Rock, Mechanics & Foundations Division, A.S.C.E., 88(SM4): 1–30.
527 p. London: Institution of Mining & Metallurgy. Judd, W.R. (ed.) 1964. State of Stress in the Earth’s Crust, Proc. Int.
Hoek, E., Carranza-Torres, C., Diederichs, M.S. & Corkum, B. Conf., Santa Monica, 13–14 June 1963, 732 p. NewYork:American
2008. Integration of geotechnical and structural design in tun- Elsevier.
nelling. Proc., University of Minnesota 56th Annual Geotechnical Kaiser, P.K. & Kim, B-H. 2008. Rock mechanics challenges in
Engineering Conf., Minneapolis, 29 February, 1–53. underground construction and mining. In Y. Potvin, J. Carter, A.
Hoek, E. & Diederichs, M.S. 2006. Empirical estimation of rock Dyskin & R. Jeffrey (eds), Proc. 1st Southern Hemisphere Int.
mass modulus. International Journal of Rock Mechanics & Mining Rock Mech. Symp., Perth, 16–19 September, 1: 23–38. Perth:
Sciences, 43(2): 203–215. Australian Centre for Geomechanics.
42

Kármán, T. von, 1911. Festigkeitsversuche unter allseitigem Druck. Mayer, A., Habib, P. & Marchand, R. 1951. Mesure en place des
Zeitschrift Verein Deutscher Ingenieure, 55: 1749–1757. pressions de terrains. Proc. Conf. Int. sur les Pressions de Terrains
Kimmance, J.P. 1999. Geographical information system applica- et le Soutènement dans les Chantiers d’Exploration, Liège, 24–28
tion to construction and geotechnical data management on large April, 217–221.
projects. In C.F. Leung, S.A. Tan & K.K. Phoon (eds), Proc. 5th Minchinton, A. & Dare-Bryan, P. 2005. The application of computer
Int. Symp. on Field Measurements in Geomechanics, Singapore, modelling for blasting and flow in sublevel caving operations.
1–3 December, 151–156. Rotterdam: Balkema. Proc. 9th AusIMM Underground Operators’ Conf., Perth, 7–9
King, L.V. 1912. On the limiting strength of rocks under conditions March, 65–73. Melbourne: Australasian Institute of Mining &
of stress existing in the earth’s interior. Journal of Geology, 20: Metallurgy.
119–138. Mogi, K. 1959. Experimental study of deformation and fracture of
Kirsten, H.A.D. 1976. Determination of rock mass elastic moduli by marble (1). On the fluctuation of compressive strength of marble
back analysis of deformation measurements. In Z.T. Bieniawski and relation to the rate of stress application. Bulletin, Earthquake
(ed.), Proc. Symp. Exploration for Rock Engineering, Johannes- Research Institute, University of Tokyo, 37: 155–170.
burg, 1–5 November, 1: 165–172. Cape Town: Balkema. Mogi, K. 2007. Experimental Rock Mechanics, 361 p. London:
Kovari, K. & Amstad, C. 1993. Decision making in tunnelling based Taylor & Francis.
on field measurements. In J.A. Hudson, E.T. Brown, C. Fairhurst Moye, D.G. 1959. Rock mechanics in the investigation and con-
& E. Hoek (eds), Comprehensive Rock Engineering, 4: 571–606. struction of T.1 underground power station, Snowy Mountains,
Oxford: Pergamon Press. Australia. Geological Society of America Engineering Geology
Kovari, K., Amstad, C. & Köppel, J. 1979. New developments in Case Histories, No 3: 13–44.
the instrumentation of underground openings. In A.C. Maevis & Müller, L. 1933. Untersuchungen über statistische Kluftmessungen.
W.A. Hustrulid (eds), Proc. 1979 Rapid Excavation & Tunneling Geologie und Bauwesen, 5(4): 185–255.
Conf., Atlanta, 18–21 June, 1: 817–837. New York: A.I.M.E. Müller, L. 1950. Der Kluftkörper. Geologie und Bauwesen, 18(1):
Lang, T.A. 1961. Theory and practice of rockbolting. Transactions, 52–60.
Society of Mining Engineers of A.I.M.E., 220: 333–348. Müller, L. 1964. The rock slide in the Vajont Valley. Rock Mechanics
Liu, J., Feng, X-T., Ding, X-L., Zhang, J. & Yue, D-M. 2003a. Sta- & Engineering Geology, 2: 148–212.
bility assessment of the Three Gorges Dam foundation, China, Müller, L. 1966. Der progressive bruch in geklufteten median. Proc.
using physical and numerical modeling – Part I: physical model 1st Congr., Int. Soc. Rock Mech., Lisbon, 25 September–1 October,
tests. International Journal of Rock Mechanics & Mining Sciences, 1: 679–686. Lisbon: LNEC.
40(5): 609–631. Müller, L. 1967a. Address to the Opening Session. Proc. 1st Congr.,
Liu, J., Feng, X-T. & Ding, X-L. 2003b. Stability assessment of the Int. Soc. Rock Mech., Lisbon, 25 September–1 October, 1966, 3:
Three Gorges Dam foundation, China, using physical and numeri- 80–83. Lisbon: LNEC.
cal modeling – Part II: numerical modeling. International Journal Müller, L. 1967b. Address to the Closing Session. Proc. 1st Congr.,
of Rock Mechanics & Mining Sciences, 40(5): 633–652. Int. Soc. Rock Mech., Lisbon, 25 September–1 October, 1966, 3:
Londe, P. 1987. The Malpasset Dam failure. Engineering Geology, 90–92. Lisbon: LNEC.
24(1–4): 295–329. Müller, L. 1968. New considerations of the Vajont slide. Rock
Londe, P., Vigier, G. & Vormeringer, R. 1969. Stability of rock slopes, Mechanics & Rock Engineering, 6(1): 1–91.
a three-dimensional study. Journal, Soil Mechanics & Foundations Müller, L. 1969. New editorial system. Rock Mechanics, 1(1):
Division, A.S.C.E., 95(SM1): 235–262. 1–4.
Long, J. & Witherspoon, P. 1985. The relationship of the degree of Müller, L. 1979. Josef Stini: contributions to engineering geology
interconnection to permeability in fracture networks. Journal of and slope movement investigations. In B. Voight (ed.), Rockslides
Geophysical Research, 90(B4): 3087–3098. and Avalanches, 2: 95–109. Amsterdam: Elsevier.
Lorig, L.J. & Brady, B.H.G. 1982. A hybrid discrete element- Müller-Salzburg, L. 1987. The Vajont catastrophe - a personal view.
boundary element method of stress analysis. In R.E. Goodman & Engineering Geology, 24(1–4): 423–444.
F.E. Heuze (eds), Issues in Rock Mechanics, Proc. 23rd U.S. Obert, L. & Duvall, W.I. 1967. Rock Mechanics and the Design of
Symp. Rock Mech, Berkeley, 25–27 August, 628–636. New York: Structures in Rock, 650 p. New York: John Wiley.
A.I.M.E. Obert, L., Windes, S.L. & Duvall, W.I. 1946. Standardized tests for
Louis, C. 1974. Rock hydraulics. In Müller, L. (ed.), Rock Mechanics, determining the physical properties of mine rock. U.S. Bureau of
(Course held at CISM, Udine, Italy), 300–387. Vienna: Springer- Mines Report of Investigation 3891.
Verlag. Oñederra, I., Chitombo, G.P., Cundall, P.A. & Furtney, J.K. 2009.
McKenzie, C.K. 1987. Blasting in hard rock: techniques for diagnosis Towards a complete validation of the lattice scheme in the Hybrid
and modeling for damage and fragmentation. In G. Herget & S. Stress Blasting Model (HSBM). In J.A. Sanchidrián (ed.), Rock
Vongpaisal (eds), Proc. 6th Congr., Int. Soc. Rock Mech., Montreal, Fragmentation by Blasting – FragBlast 9, Proc. 9th Int. Symp.
30 August – 3 September, 3: 1425–1431. Rotterdam: Balkema. on Rock Fragmentation by Blasting, Granada, 13–17 September,
Marinos, P.G., Marinos,V. & Hoek, E. 2007. The Geological Strength 343–351. Leiden: CRC Press/Balkema.
Index (GSI): a characterization tool for assessing engineering Ortlepp, W.D. 2007. Dynamic capacity in cable anchors and rock-
properties of rock masses. In C. Mark, R. Pakalnis & R.J. Tuchman bolts. In Cave Mining, 1st Int. Symp. on Block & Sub-level Caving,
(eds), Proc. Int. Workshop on Rock Mass Classification for Under- Cape Town, 8–10 October, 405–419. Johannesburg: Southern
ground Mining, Vancouver, 31 May, NIOSH IC 9498, 87–94. African Institute of Mining & Metallurgy.
Pittsburgh: U.S. Department of Health & Human Services. Ouchterlony, F. & Moser, P. 2006. Likenesses and differences in the
Martin, C.D. 1997. Seventeenth Canadian Geotechnical Colloquium: fragmentation of full-scale and model-scale blasts. Fragblast-8,
the effect of cohesion loss and stress path on brittle rock strength. Proc. 8th Int. Symp. on Rock Fragmentation by Blasting, Santiago,
Canadian Geotechnical Journal, 34(5): 698–725. 7–11 May, 207–220. Santiago: Editec.
Mas Ivars, D., Pierce, M.E., Darcel, C., Reyes-Montes, J., Potyondy, Pacher, F. 1964. Deformationsmessungen im Versuchstollen als
D., Young, R.P. & Cundall, P.A. 2011. The synthetic rock mass Mittel zur Erforschung des Gebirgsverhaltens und zur Bemessung
approach for jointed rock mass modelling. International Journal des Ausbaues. Rock Mechanics & Engineering Geology, Suppl IV,
of Rock Mechanics & Mining Sciences, 48(2): 219–244. 149–161.
Mayer, A. 1953. Les propértiés mécaniques des roches. Géotech- Panet, M. 1993. Understanding deformations in tunnels. In J.A.
nique, 3(8): 329–338. Hudson, E.T. Brown, C. Fairhurst & E. Hoek (eds), Com-
Mayer, A. 1963. Recent work in rock mechanics. Géotechnique, prehensive Rock Engineering, 1: 663–690. Oxford: Pergamon
13(2): 99–118. Press.
43

Panet, M. 1995. Calcul des Tunnels par la Methode de Convergence- Sublevel Caving, Perth, 20–22 April, 425–439. Perth: Australian
Confinement. Paris: Press de l’Ecole Nationale des Ponts et Centre for Geomechanics.
Chausses. Rojat, F., Labiouse, V., Kaiser, P.K. & Descoeudres, F. 2008. Brittle
Paterson, M.S. 1958. Experimental deformation and faulting in rock failure in the Steg lateral adit of the Lötschberg base tunnel.
Wombeyan marble. Bulletin, Geological Society of America, 69: Rock Mechanics & Rock Engineering, 42(2): 341–359.
465–476. Rojek, J., Oñate, E., Labra, C., Kargl, H. & Akerman, J. 2010. Opti-
Paterson, M.S. 1978. Experimental Rock Deformation – The Brittle mizing rock cutting through computer simulation. In G. Beer (ed.),
Field, 254 p. Berlin: Springer-Verlag. Technology Innovation in Underground Construction, 299–314.
Pells, P.J.N. 2002. Developments in the design of tunnels and cav- Leiden: CRC Press/Balkema.
erns in the Triassic rocks of the Sydney region. International Ryder, J.A. & Jager, A.J. (eds) 2002. Rock Mechanics for Tabular
Journal of Rock Mechanics & Mining Sciences, 39(5): 569– Hard Rock Mines, 489 p. Johannesburg: SIMRAC.
587. Sainsbury, B.L., Sainsbury, D.P. & Pierce, M.E. 2011. A histor-
Pells, P.J.N., Poulos, H.G. & Best, R.J. 1991. Rock reinforcement and ical review of the development of numerical cave propagation
design for a shallow large span cavern. In W. Wittke (ed.), Proc. simulations. In D.P. Sainsbury, R.D. Hart, C.L. Detournay &
7th Congr., Int. Soc. Rock Mech, Aachen, 16–20 September, 2: M.J. Nelson (eds), Continuum and Distinct Element Numerical
1193–1198. Rotterdam: Balkema. Modeling in Geomechanics – 2011, Proc. 2nd Int. FLAC/DEM
Pichler, J., Lammer, E. & Bertignoll, H. 2010. Innovative roadheader Symp., Melbourne, 14–16 February, 23–36. Minneapolis: Itasca
technology for safe and economic tunnelling. In G. Beer (ed.), International Inc.
Technology Innovation in Underground Construction, 315–333. Sainsbury, D.P., Hart, R.D., Detournay, C.L. & Nelson, M.J. (eds)
Leiden: CRC Press/Balkema. 2011. Continuum and Distinct Element Numerical Modeling
Pierce, M., Gaida, M. & DeGagne, D. 2009. Estimation of rock block in Geomechanics – 2011, Proc. 2nd Int. FLAC/DEM Symp.,
strength. In M. Diederichs & G. Grasselli (eds), Rock Engineering Melbourne, 14–16 February, 852 p. Minneapolis: Itasca Interna-
in Difficult Conditions, Proc. 3rd Canada-U.S. Rock Mech. Symp, tional Inc.
Toronto, 9–15 May, Paper 4360 (on CD-ROM). Sakurai, S. 1993. Back analysis in rock engineering. In J.A. Hudson,
Pine, R.J., Coggan, J.S., Flynn, Z.N. & Elmo, D. 2006. The devel- E.T. Brown, C. Fairhurst & E. Hoek (eds), Comprehensive Rock
opment of a new numerical modelling approach for naturally Engineering, 4: 543–569. Oxford: Pergamon Press.
fractured rock masses. Rock Mechanics & Rock Engineering, Sakurai, S. 1997. Lessons learned from field measurements in
39(5): 395–419. tunnelling. Tunnelling & Underground Space Technology, 12(4):
Pinkerton, I.L., Andrews, K.E., Bray, A.N.G. & Frost, A.C.H. 1961. 453–460.
The design, construction and commissioning of Tumut 1 Power Sakurai, S. 1999. Interpretation of field measurements in tunnelling
Station. Journal, Institution of Engineers, Australia, 33(7–8): 235– practice. In G. Vouille & P. Berest (eds), Proc. 9th Congr., Int. Soc.
252. Rock Mech., Paris, 25–28 August, 3: 1517–1523. Lisse: Balkema.
Player, J., Thompson, A. & Villaescusa, E. 2008. Dynamic testing Sakurai, S. & Akutagawa, S. 1995. Some aspects of back analy-
of reinforcement systems. In T.R. Stacey & D.F. Malan (eds), sis in geotechnical engineering. In L. Ribeiro e Sousa & N.F.
Proc. 6th Int. Symp. on Ground Support in Mining & Civil Engi- Grossmann (eds), Eurock ’93 - Safety & Environmental Issues
neering Construction, Cape Town, 30 March – 3 April, 597–622. in Rock Mechanics, Proc, Int, Symp., Lisbon, 21–24 June 1993,
Johannesburg: Southern African Institute of Mining & Metallurgy. 1133–1140. Rotterdam: Balkema.
Potts, E.L.J. 1957. Underground instrumentation. Quarterly, Col- Sakurai, S., Farazmand, A. & Adachi, K. 2009. Assessment of the
orado School of Mines, 52(3): 135–182. stability of slopes from surface displacements measured by GPS
Potvin, Y., Stacey, D. & Hadjigeorgiou (eds) 2004. Surface Support in an open pit mine. In G. Deák & Z. G. Agioutantis (eds), Sus-
in Mining, 416 p. Perth: Australian Centre for Geomechanics. tainable Exploitation of Natural Resources, Proc. 3rd Int. Seminar
Potyondy, D.O. & Cundall, P.A. 2004. A bonded-particle model for ECOMINING – Europe in 21st Century, Milos Island, Greece, 4–5
rock. International Journal of Rock Mechanics & Mining Sciences, September, 239–248.
41(8): 1329–1364. Sakurai, S. & Takeuchi, K. 1983. Back analysis of measured dis-
Priest, S.D. 2004. Determination of discontinuity size distributions placement of tunnels. Rock Mechanics & Rock Engineering, 16(3):
from scanline data. Rock Mechanics & Rock Engineering, 37(5): 173–180.
347–368. Salamon, M.D.G. & Munro,A.H. 1967.A study of the strength of coal
Priest, S.D. & Hudson, J.A. 1981. Estimation of discontinuity spacing pillars. Journal, South African Institute of Mining & Metallurgy,
and trace length using scanline surveys. International Journal of 68(2): 55–67.
Rock Mechanics & Mining Sciences, 18(3): 183–197. Salamon, M.D.G., Ryder, J.A. & Ortlepp, W.D. 1964. An analogue
Rabcewicz, L. 1969. Stability of tunnels under rock load. Water solution for determining the elastic response of strata surround-
Power, 21(6–8): 225–229, 266–273, 297–304. ing tabular mine excavations. Journal, South African Institute of
Read, J. & Stacey, P. (eds) 2009. Guidelines for Open Pit Slope Mining & Metallurgy, 65(2): 115–137.
Design, 496 p. Melbourne: CSIRO Publishing. Schmidt, W. 1925. Gefügestatistik. Tschermaks Mineralogische und
Roberts, A. 1963. Editorial. International Journal of Rock Mechanics Petrographische Mitteilungen, 38: 392–423.
& Mining Sciences, 1(1): 3–4. Schubert, W. 2006. The role of on-site engineering in underground
Rocha, M. 1964. Mechanical behaviour of rock foundations in con- projects. In C.F. Leung & Y.X. Zhou (eds), Rock Mechanics in
crete dams. Transactions, 8th Int. Congr. Large Dams, Edinburgh, Underground Construction, Proc., ISRM Int. Symp. 2006, 4th
1: 785–832. Asian Rock Mech. Symp., Singapore, 8–10 November, 71–82.
Rocha, M., Serafim, J.L., Silveira, A. & Neto, J.R. 1955. Deforma- Singapore: World Scientific Publishing Co.
bility of foundation rocks. Proc. 5th Congr. Large Dams, Paris, Schunnesson, H. 2009. Modern tools and equipment for tunnelling
R75, 3: 531–559. Paris: Editions Science et Industrie. and drifting. In P. Dight (ed.), Proc. 1st Int. Seminar on Safe
Roegiers, J-C. 1999. The importance of rock mechanics to the & Rapid Development Mining, Perth, 6–7 May, 139–148. Perth:
petroleum industry. In G. Vouille & P. Berest (eds), Proc. 9th Australian Centre for Geomechanics.
Congr., Int. Soc. Rock Mech., Paris, 25–28 August, 3: 1525–1549. Schutjens, P. 2009. Book review: Petroleum Related Rock Mechan-
Lisse: Balkema. ics, 2nd edn, 2008. International Journal of Rock Mechanics &
Rogers, S., Elmo, D., Webb, G. & Catalan,A. 2010.A discrete fracture Mining Sciences, 46(8): 1398–1399.
network based approach to defining in situ, primary and secondary Shi, G-H. & Goodman, R.E. 1985. Two dimensional discontinu-
fragmentation distributions for the Cadia East panel cave project. ous deformation analysis. International Journal for Numerical &
In Y. Potvin (ed.), Caving 2010, Proc. 2nd Int. Symp. Block & Analytical Methods in Geomechanics, 9(6): 541–556.
44

Slob, S., Hack, H.R.G.K., Feng, Q., Röshoff, K. & Turner, A.K. Mining, Luleå, 9–11 June, 341–352. Luleå: Luleå University of
2007. Fracture mapping using 3D laser scanning techniques. In Technology.
L. Ribeiro e Sousa, C. Olalla & N.F. Grossmann (eds), Proc. Villaescusa, E., Hassell, R. & Thompson, A.G. 2008. Develop-
11th Congr., Int. Soc. Rock Mech., Lisbon, 9–13 July, 1: 299–302. ment of a corrosivity classification for cement grounted cable.
Leiden: Balkema. In T.R. Stacey & D.F. Malan (eds), Proc. 6th Int. Symp. on Ground
Starfield, A.M. & Cundall, P.A. 1988. Towards a methodology Support in Mining & Civil Engineering Construction, Cape Town,
for rock mechanics modelling. International Journal of Rock 30 March – 3 April, 461–476. Johannesburg: Southern African
Mechanics & Mining Sciences, 25(3): 99–106. Institute of Mining & Metallurgy.
Stephansson, O., Jing, L. & Tsang, C-F. (eds) 1996. Coupled Wawersik, W.R. & Fairhurst, C. 1970. A study of brittle rock frac-
Thermo-Hydro-Mechanical Processes of Fractured Media, 596 p. ture in laboratory compression tests. International Journal of Rock
Amsterdam: Elsevier. Mechanics & Mining Sciences, 7(5): 561–575.
Stini, J. 1922. Technische Geologie, 789 p. Stuttgart: Ferdinand Enke. West, G. 1988. Innovation and the Rise of the Tunnelling Industry,
Suorineni, F.T., Chinnasane, D.R. & Kaiser, P.K. 2009. A procedure 355 p. Cambridge: Cambridge University Press.
for determining rock-type specific Hoek-Brown brittle parameter Williams, P., Floyd, J., Chitombo, G. & Maton, T. 2010. Design imple-
s. Rock Mechanics & Rock Engineering, 42(6): 849–881. mentation. In J. Read & P. Stacey (eds), Guidelines for Open Pit
Takahashi, T. 2004. ISRM Suggested Methods for land geophysics Mine Design, 265–326. Melbourne: CSIRO Publishing.
in rock engineering. International Journal of Rock Mechanics & Windsor, C.R. 1993. Measuring stress and deformation in rock
Mining Sciences, 41(6): 885–914. masses. In T. Szwedzicki (ed.), Proc. Australian Conf. on Geotech-
Takahashi, T., Takeuchi, T. & Sassa, K. 2006. Suggested Methods for nical Instrumentation and Monitoring in Open Pit and Under-
borehole geophysics in rock engineering. International Journal of ground Mining, Kalgoorlie, 21–23 June, 33–52. Rotterdam:
Rock Mechanics & Mining Sciences, 43(3): 337–368. Balkema.
Talobre, J. 1957. La Mécanique des Roches, 444 p. Paris: Dunod. Windsor, C.R. 2004. A review of long, high capacity reinforcing
Terzaghi, K. 1962a. Measurement of stresses in rock. Géotechnique, systems used in rock engineering. In E. Villaescusa & Y. Potvin
12(2): 105–124. (eds), Ground Support in Mining & Underground Construction,
Terzaghi, K. 1962b. Stability of steep slopes on hard unweathered Proc. 5th Int. Symp. on Ground Support, Perth, 28–30 September,
rock. Géotechnique, 12(4): 251–270. 17–41. Leiden: Balkema.
Terzaghi, K. & Richart, F.E. 1952. Stresses in rock about cavities. Windsor, C.R. & Thompson, A.G. 1993. Rock reinforcement – tech-
Géotechnique, 3(2): 57–90. nology, testing, design and evaluation. In J.A. Hudson, E.T. Brown,
Terzaghi, R.D. & Voight, B. 1979. Karl Terzaghi on rock slides: the C. Fairhurst & E. Hoek (eds), Comprehensive Rock Engineering,
perspective of a half-century. In B. Voight (ed.), Rockslides and 4: 451–484. Oxford: Pergamon Press.
Avalanches, 2: 111–133. Amsterdam: Elsevier. Witherspoon, P.A., Wang, J.S.Y., Iwai, K. & Gale, J.E. 1980. Validity
Thorsteinsson, H., Augustine, C., Anderson, B.J., Moore, M.C. & of cubic law for fliud flow in a deformable rock fracture. Water
Tester, J.W. 2008. The impacts of drilling and reservoir technology Resources Research, 16(6): 1016–1024.
advances on EGS exploitation. Proc., 33rd Workshop on Geother- Wittke, W. 1977. Static analysis for underground openings in jointed
mal Reservoir Engineering, Stanford University, 28–30 January, rock. In C.S. Desai & J.T. Christian (eds), Numerical Methods in
Stanford Geothermal Program Publication SGP-TR-2008, 14 p. Geotechnical Engineering, 589–638. New York: McGraw-Hill.
Tincelin, E. 1952. Mesures des pressions de terrains dans les mines Wittke, W. 1990. Rock Mechanics: Theory and Applications with
de fer de l’est. Annales de l’Institut Technique du Bâtiment et des Case Histories, 1075 p. Translated from German by R. Sykes in
Travaux Publics, Série Sols et Foundations, No 58, October: 972– co-operation with S. Semprich & B. Plischke. Berlin: Springer.
990. You, T., Gatelier, N. & Laurent, S. 2006. Design methodology for
Trollope, D.H. 1968.The mechanics of discontinua or clastic mechan- hydrocarbon caverns, influence of in-situ stresses on large sec-
ics in rock problems. In K.G. Stagg & O.C. Zienkiewicz (eds), tions. In C.F. Leung & Y.X. Zhou (eds), Rock Mechanics in
Rock Mechanics in Engineering Practice, 275–320. London: John Underground Construction, Proc., ISRM Int. Symp. 2006 & 4th
Wiley. Asian Rock Mech. Symp., Singapore, 8–10 November, Paper 63,
Trollope, D.H. & Brown, E.T. 1965. Pressure distributions in some 9 p. (on CD-ROM). Singapore: World Scientific Publ. Co.
discontinua. Water Power, 17(8): 310–313. Zhang, L., Einstein, H.H. & Dershowitz, W.S. 2002. Stereological
Turchaninov, I.A., Iofis, M.A. & Kasparyan, E.V. 1979. Principles of relationship between trace length and size distribution of elliptical
Rock Mechanics, 493 p. Translated from Russian by A.L. Peabody; discontinuities. Géotechnique, 52(6): 419–433.
translation edited by W.A. Hustrulid. Rockville, MD: Terraspace. Zhao, J. & Gong, Q.M. 2006. Rock mechanics and excavation by
Ulusay, R. & Hudson, J.A. (eds) 2007. The Complete ISRM Suggested tunnel boring machine. In C.F. Leung & Y.X. Zhou (eds), Rock
Methods for Rock Characterization, Testing and Monitoring: Mechanics in Underground Construction, Proc., ISRM Int. Symp.
1974–2006, 628 p. Ankara: ISRM Turkish National Group. 2006 & 4th Asian Rock Mech. Symp., Singapore, 8–10 November,
Vakili, A. & Hebblewhite, B.K. 2010. A new cavability assessment 83–96 Singapore: World Scientific Publ. Co.
criterion for Longwall Top Coal Caving. International Journal of Zhou, W.,Yang, R. & Zhou, L. 1986. An analysis of jointed rock using
Rock Mechanics & Mining Sciences, 47(8): 1317–1329. fracture mechanics for stability of arch dam abutment. Proc. Int.
Villaescusa, E. 1996. Excavation design for bench stoping at Mount Symp. on Engineering in Complex Rock Formations, Beijing, 3–7
Isa mine, Queensland, Australia. Transactions, Institution of Min- November, 470–478. Biejing: Science Press.
ing & Metallurgy, Section A: Mining Industry, 105: A1–10. Zienkiewicz, O.C. 1977. The Finite Element Method, 3rd edn, 787 p.
Villaescusa, E. 2008. Geotechnical considerations for planning London: McGraw-Hill.
and design of open stopes. In H. Schunnesson & E. Nordlund Zienkiewicz, O.C. & Cheung, Y.K. 1964. Buttress dams on complex
(eds), MassMin 2008, Proc. 5th Int. Conf & Exhibition on Mass rock foundations. Water Power, 16(5): 193–217.
45

The next 50 years of the ISRM and anticipated future progress in rock mechanics
J.A. Hudson
Department of Earth Sciences and Engineering, Imperial College of Science, Technology and Medicine,
London, UK
President International Society for Rock Mechanics (2007–2011)
ABSTRACT: This paper on the 50-year anticipated future of the ISRM and rock mechanics forms part of the ISRM 50-year
anniversary celebrations and complements the preceding paper by E T Brown in these proceedings on the previous 50 years of the
ISRM. The current status of rock mechanics and the unsolved problems are summarised—so that the results of extrapolating our
current capabilities can be considered. The problems are noted under the headings of geology, rock stress, intact rock, fractures,
water flow, modelling and design. The anticipated future developments are then highlighted under the headings of information
access, site investigation, subject integration, international co-operation, ‘intelligent’ computer programs, increased excavation
speed and larger excavations, environmental aspects, the future of the ISRM itself, and the implications of future computing
power. It is predicted that the greatest changes to rock mechanics and the ISRM itself will result from the continuing growth
in computing power.
Subject: Rock mechanics prediction
Keywords: Rock failure, rock joints, lab testing, field measurements, numerical modelling, neural networks
1 INTRODUCTION
This paper forms part of the 50-year ISRM celebrations and

is the companion contribution to Professor Brown’s preceding
paper on “50 Years of the ISRM and Associated Progress in
Rock Mechanics”. Also, this paper is an updated and extended
version of an article previously presented at the ARMS5
symposium held in Iran (Hudson, 2008).
It is important for our subject of rock mechanics and its
application to rock engineering that we not only consider the
developments over the last 50 years, but that we also anticipate
the directions that are likely to be taken in the future, with this
ISRM anniversary period and the 2011 ISRM Congress in
Beijing being a suitable opportunity for such speculation
The method for predicting the future through extrapola- Figure 1. Hippocrates (460–377 BC).
tion of the present is mentioned as a prelude to the generic
and specific predictions for rock mechanics and rock engi- individual and group interactions, the storage and dissemina-
neering, although some ‘blue skies’ thinking is also included. tion of corporate knowledge, and the impact of the continuing
Implicit in the paper’s title is consideration of the ISRM entity increase in computing power.
itself—and so, in the light of the predicted technological and
developments, a later section discusses whether the ISRM will
continue to exist for another 50 years. 2 THE CURRENT ROCK MECHANICS KNOWLEDGE
Hippocrates (Fig. 1), the Greek physician, used a method AND CAPABILITIES
for predicting the future which is summarised as:
The rock mechanics knowledge and capabilities in 1993 have
“Consider the past, been summarised in encyclopaedic form via the 4407 page
diagnose the present, compendium “Comprehensive Rock Engineering” which was
foretell the future.” produced by Elsevier (Hudson, 1993). The five volumes
comprising the compendium cover the subjects of:
So, in this paper, and based on what has been achieved in the
1. Fundamentals
past, let us identify some of the major problems that have not
2. Analysis and Design Methods
yet been solved.
3. Rock Testing and Site Characterisation
This leads naturally to consideration of which technological
4. Excavation, Support and Monitoring
developments are likely in the future and hence whether these
5. Surface and Underground Case Histories.
will enable the remaining problems to be solved. Consider-
ation of the far future for the ISRM is interesting because Although it is now 18 years since this compendium was
it encompasses relevant questions relating to the nature of published and some advances have been made in many areas,
47

Figure 2. Similarities in the roles of structural geology and rock
Figure 3. Orthogonal fracturing in the Coal Measures strata, South
mechanics for engineering.
Wales, UK (∼1 m wide sample).
the essence of the state-of-the-art remains similar, so let us

now consider the current unsolved problems.
3 UNSOLVED PROBLEMS IN ROCK MECHANICS
Despite the major progress that has been made in rock mechan-
ics and rock engineering over the last 50 years as described
by Brown (2011), there are still outstanding problems. In fact,
Leopold Müller’s motivation for founding the ISRM in 1962
was encapsulated in his May 1962 comment, “We do not know
the rock mass strength, and that is why we need an Interna-
tional Society”, yet we still have problems estimating the rock
mass strength!
In this Section, some of the major unsolved problems in
rock mechanics will be outlined. These are presented under
the subject headings of
Figure 4. More irregular fracturing in the Coal Measures strata,
• Geology South Wales, UK (∼1 m wide sample).
• Rock stress
• Intact rock
• Fractures
• Water flow
• Modelling
• Design
At the end of each of these secondary headings, italicised text
is included concerning a key point related to that particular
subject.
3.1 Geology
The roles of geology, particularly structural geology, and rock
mechanics for engineering are highlighted in Figure 2.
In both cases, structural geology and rock mechanics/
engineering, we are interested in deciphering the mechanisms
that led to the current circumstances and then in predicting Figure 5. Even more irregular fracturing in the Coal Measures
what will happen in the future. In fact, the quintessence of strata, South Wales, UK (∼1 m wide sample).
rock engineering is predicting the future. We wish to predict
what will happen in the future so that rock engineering designs
are appropriate; if such prediction is not possible, there can stretched without expert geological advice in obtaining the
be no coherent design. For example, we might wish to predict correct data for modelling and identifying the hazards that may
what will occur if a tunnel is driven in a certain direction and accompany certain geological formations. For example, many
depth as opposed to other directions and depths. rock mechanics modelling exercises require knowledge of the
In this context and because the engineered structure will rock fractures, but this may be difficult to obtain. The three
be founded on or contained within a rock mass, it is essential photographs in Figures 3–5 illustrate three different types of
to have an adequate understanding of the geological circum- fracturing—all observed in close proximity within the same
stances relating to the host rock, yet the engineer will be sorely Lower Middle Coal Measures strata in South Wales, UK.
48

Needless to say, significant help can be given to the rock
mechanics modeller and designer in dealing with the types
of fracture networks illustrated in Figures 3–5 by a geologist
who is familiar with the host rock circumstances. Illustrations
of the value of structural geology interpretations can be seen
in Price & Cosgrove (1990), Hudson & Cosgrove (1997) and
Cosgrove & Hudson (2011).
However, the major unsolved problem in the current context
is that often either a geologist is not consulted or the geologist
is not intimately involved with the rock mechanics studies, so
that the advice is only received remotely, e.g., through some
form of standardised site investigation.
This is a management problem but one that we must address
and one that ought to be easy to solve.
3.2 Rock stress

Figure 6. Complete stress–strain curve for marble (courtesy EPFL
Unlike other forms of engineering, in underground rock engi- Laboratory, Switzerland).
neering the primary activity is to remove material from a
pre-loaded rock mass, i.e., rather than fabricating components
and then assembling them, as in other forms of engineering.
This means that we need to estimate the natural stress state
existing in the rock mass—both for basic considerations of
the intensity of loading and as boundary condition input to
numerical modelling.
Rock stress estimation is one of the most intractable
problems that we have to face: stress is a tensor quantity
which requires six independent pieces of information;
measuring/estimating rock stress is difficult; and the rock
stress may vary considerably at all scales, mainly through the
various types of fracturing inherent in the different rock types.
A Special Issue of the International Journal of Rock Mechan-
ics and Mining Sciences (Vol. 40, Issues 7–8, 2003) is devoted
to rock stress estimation. This Special Issue contains the four Figure 7. The concentrated principal stress component around the
new ISRM Suggested Methods on rock stress estimation and periphery of an underground opening is parallel to the periphery but
the anisotropy induced by foliation is in one main direction, meaning
17 supporting papers describing various aspects of rock stress
that the rock failure will not necessarily begin at the location of the
and rock stress estimation campaigns. A summary of the con- highest stress value.
tents of these 17 papers in eight words is: “Rock stress estima-
tion is not an easy task.” The actual measurement is a problem
and the stress itself generally varies throughout the rock mass.
However, recent developments in understanding, presenting 3.3 Intact rock
and modelling in situ rock stress have improved the situation. Because it is easy to obtain and test samples of intact rock,
For example, Lee (2006) has shown that plotting each prin- this subject has been studied in greater detail than any other
cipal stress versus the first stress invariant (i.e., the sum of in rock mechanics, and we now have a good understanding of
the principal stresses) provides a more coherent picture than intact rock behaviour and the relation between the type of rock
plotting the principal stresses versus depth. A recent book and its mechanical characteristics. Also, through the advent
(Zang & Stephansson, 2010) provides a clear understanding of high speed, servo-controlled testing machines in the 1970s,
of the distribution of rock stress and the stress measurement the full mechanical behaviour, from initial loading through
procedures. And numerically modelling the stress distribution to complete disintegration of the specimen, can be studied
through fractured rock masses using computer programs such in detail. An example of a complete stress–strain curve for
as 3DEC can provide a significantly enhanced understanding marble is shown in Figure 6. Through the use of a polyaxial
of the role of fractures in perturbing a regional stress state. testing machine with appropriate servo-control, the behaviour
The two major unsolved problems in the context of rock of intact rock can be studied in circumstances only limited
stress are: by the imagination. Nevertheless, there are still aspects of the
• developing a method of rapidly and reliably estimating the intact rock requiring further study, particularly in relation to
six components of the rock stress tensor at a given location; the inhomogeneity, anisotropy and time dependency.
and The illustration in Figure 7 shows how studies of the poten-
• understanding the variations in the stress state due tial failure of rock around the periphery of an excavation is
to the presence of discontinuities, inhomogeneities and a function of both the concentrated rock stress around the
anisotropies at various scales. opening and the rock strength in different directions. For this
problem, a knowledge of the rock strength as a function of the
The first problem above is intractable and not likely to be anisotropy is required. In the same way, the influence of the
solved quickly. The second problem is amenable to numerical inhomogeneity of the rock and its time dependency also need
modelling studies and will be solved relatively soon. to be understood.
49

Figure 9. The fracture properties recommended for measurement
by the ISRM.
Figure 8. The types of deformation and fracturing that can develop

in rock masses, from tensile fractures through to a pervasive ductile
fabric.
Thus, although a great deal of work has been conducted

on the mechanical behaviour of intact rock (i.e., without Dis-
continuities), there are still unsolved problems relating to its
IANE nature (Inhomogeneous, Anisotropic and Not Elastic).
As noted in the previous section, computer modelling can
improve our understanding of in situ stress variability and,
in a similar way, the failure of intact rock in a wide variety
of loading conditions can also be studied via computer mod-
elling (Tang & Hudson, 2010). However, the subject of time
dependency remains intractable.
Perhaps the most surprising omission in relation to intact Figure 10. The limited scales at which one can obtain rock fracture
rock is the general use of a failure criterion which incorpo- information (from Follin (2008) of SKB, Sweden).
rates the three principal stresses. Despite the nature of rock
stress as a tensor with its three principal stresses, the failure possible from a borehole core or television image of the bore-
criteria most commonly used in modelling and in practice for hole wall. The spacing between the fractures can be readily
intact rock and rock masses are the Mohr–Coulomb and Hoek– estimated, but many aspects, such as the fracture tracelength
Brown criteria—but both of these only contain the major and (persistence), cannot.
minor principal stresses: the intermediate principal stress is In Figure 10, Follin (2008) has illustrated the limited scales
not included. At the time of writing this paper, the ISRM Com- at which one can obtain fracture information. Given the D
mission on Testing Methods is developing advice on the use in the DIANE nature of the rock, the Discontinuousness, the
of rock failure criteria which will be published in the Interna- ability to characterise the geometry and mechanical behaviour
tional Journal of Rock Mechanics and Mining Sciences and of the fractures is essential for input to our modelling for rock
in the new 2012 ISRM ‘Orange Book’ which will follow the engineering design—and to achieve Leopold Müller’s original
style of the ISRM ‘Blue Book’ (Ulusay & Hudson, 2007). ISRM ambition of estimating the rock mass strength.
In terms of solving these problems, the inhomogeneity and Thus, developing the ability to adequately establish and
anisotropy characteristics of intact rock are easily approached characterise the geometry and mechanical properties of rock
via testing and numerical modelling but the issue of time fractures at a given site is the major unsolved problem in
dependency is more intractable. The general use of a failure relation to the subject of rock fractures.
criterion incorporating the three principal stresses is immi- Although solving the problem of reliably obtaining frac-
nent because candidate criteria have already been suggested ture information (geometry and mechanical properties) may
in the literature. be considered easier than obtaining reliable stress informa-
tion for a given rock mass, there is still a long way to go
3.4 Fractures because of the multi-parametric nature of rock fracturing and
the uncertainty related to reducing the mass of fracture data
Figures 3–5 have indicated some types of fracture variation to the quantity that can be input to numerical models. There is
that can occur and the importance of including a structural also the problem of including ‘stochastic’ fractures in models.
geologist in rock mechanics studies has been emphasised. In
fact, it is useful to consider the spectrum of rock fractures that
3.5 Water flow/permeability
can occur from a geological viewpoint, as shown in Figure 8
(Hudson et al., 2011). Following on from the need to characterise the rock fractures
The fracture properties recommended for measurement by is the subject of modelling water flow and establishing the
the ISRM are shown in Figure 9. It is evident that these fracture rock mass permeability—another tensor with three principal
properties are variable in terms of whether the measurement is components. Note the white arrow in Figure 11 indicating
50

Figure 11. Idiosyncratic water flow in a fractured rock mass (the
Mountsorrel granodiorite, UK). Figure 12. The eight main methods of rock mechanics modelling
(within the emboldened dashed box). The four columns, Methods
A–D, represent increasing complexity from A–D; whilst the two
the highly localised nature of water flow in the rock mass rows, Levels 1 & 2, represent 1:1 mapping and non-1:1 mapping
illustrated. respectively. In 1:1 mapping there is an explicit attempt to model the
For rock masses where the intact rock is relatively imper- engineering geometry; in non-1:1 mapping there is not.
meable, water flow and the associated hydraulic conductivity
tensor will be dominated by the network of fractures (their
geometry, connectivity, aperture) plus the rock stress state
(Franciss, 2010). Not only is the water in the heavily fractured
rock mass in Figure 11 only exiting through one fracture but it
is exiting at specific points along the fracture, indicating chan-
nel flow within the fractures. Additionally, this channel flow
can change if there are any significant fracture displacements
due to engineering activities.
Considerable advances have been made in incorporating
fracture geometries into discrete fracture networks and estab-
lishing the flow characteristics of the associated rock masses.
However, the problem is obtaining sufficiently accurate data
on the fracture characteristics.This is the major unsolved prob-
lem relating to rock mass hydraulic conductivity/permeability. Figure 13. The technical auditing required for numerical modelling
The solution to this problem is also intractable because used to support rock engineering design.
not only are the geometrical fracture data required but also
the hydrogeological data, especially the equivalent fracture
apertures. Currently, the separate models are mostly constructed within
separate sub-disciplines, e.g., mechanical and hydraulic, and
3.6 Modelling these have to be ‘stitched together’—which cannot always
be done smoothly. However, considerable progress has been
In the early years of rock mechanics, physical models played
made recently in developing coupled models.
an important role in supporting rock engineering design,
A complementary necessity is the development of methods
especially for dams located on rock masses. Also today,
to technically audit such a fully-coupled numerical model, and
such physical models can play a useful role in establishing
indeed existing sub-models, both concurrently during their use
mechanisms and verifying numerical models. However, the
and subsequently when they are used for back analysis after
emphasis has gradually changed (Fig. 12) to rock mass clas-
observation of the construction perturbation. The validation
sification methods (Barton, 2011) and computer numerical
of numerical model inputs, operation and outputs is a partic-
modelling, especially using discrete element methods (Jing
ular problem in rock mechanics because of the scale of rock
and Stephansson, 2007).
masses. We can test small samples in the laboratory and we
The major unsolved problems in the numerical modelling
can conduct tests in underground laboratories at specific sites
area begin with the need to develop one model which captures
(Wang et al., 2011), but the overall problem of technically
all the required variables, mechanisms and parameters, i.e.,
auditing computer models supporting rock engineering design
the models relating to the separate sub-disciplines giving way
in large rock masses is an intractable problem, although a start
to the required all-encompassing coupled model. It is antici-
has been made in Feng & Hudson (2011), see Figure 13.
pated that such a model will include, inter alia, the following
It is unlikely that the actual validation problem will be
aspects.
solved quickly (i.e., does the model actually represent the
Geological: site geometry, lithology, fractures rock behaviour) because of the need to conduct large experi-
Thermal: heat loads, heat flow ments. However, projects such as DUSEL (Deep Underground
Hydrological: water pressures, water flow Science and Engineering Laboratory) presently being devel-
Mechanical: rock stress, stiffness, strength oped at the Homestake Mine in South Dakota, USA (e.g., (e.g.,
Chemical: rock and water chemistry www.dusel.org) will eventually lead to full validation checks.
Engineering: effects of excavation Basic technical auditing can be implemented now, but the
51

analyse projects on a similar basis to human experience and
its subsequent scrutiny of new situations.
Rock engineering design will continue to be enhanced by the
use of feedback from the construction itself, both directly and
via ‘intelligent’ computer procedures that retain a memory
not only of the current project but of all previous projects they
have experienced.
4 THE NEXT 50 YEARS
In this Section, the future for rock mechanics and the ISRM
is predicted.
4.1 Improved methods of obtaining/accessing/collating

information
There are many aspects of information processing that are
developing rapidly—from laser capture of rock surface geom-
etry, to seismic tomographic methods of rock mass properties,
to internet databases of rock properties. All these are functions
of the current rapid technological and computer advances and
Figure 14. Rock engineering design flowchart. there is every reason to expect that this will continue.
4.2 More emphasis on geophysical methods in site

final auditing capability with actual validation checks will investigation
take much longer. As mentioned earlier, there is a fundamental need to obtain
more rock property information, especially on the geome-
3.7 Rock engineering design try and mechanical properties of rock fractures. The ideal
method of non-destructive rock mass interrogation is through
An ISRM Commission on Rock Engineering Design has been the also rapidly developing seismic techniques, especially
operative in the years 2007–2011, culminating in the book by tomography, and the associated advanced 3-D visualisation
Feng and Hudson (2011). The modelling for rock engineering methods.
has been summarised in Figure 12 and the design components
are summarised in Figure 14.
With regard to the immediate future of rock engineering 4.3 More integration of subjects (e.g., fully-coupled
design, we anticipate that there will be much more use of numerical modelling)
‘intelligent’ methods, i.e., computer processes in which the There are initiatives underway to enhance the degree of
previous behaviour of the rock is learnt—thus enabling more coupling in numerical models which will eventually lead
accurate predictions of its future behaviour. For example, to a Geo-Thermo-Hydro-Mechanical-Chemical-Biological
computer learning from the initial displacements induced by model. Currently, the numerical models tend to be developed
excavation of the top level of a cavern enables better predic- separately within specific subject areas, such as geology, rock
tions of the subsequent displacements as the successive layers mechanics, petroleum engineering, soil mechanics, hydro-
are excavated downwards. geology, etc. The future must lie in more expansive models
Artificial neural networks have already been used suc- firmly based on thermodynamic principles with a range of pri-
cessfully for many subjects in rock mechanics modelling, as mary state variables and incorporating the associated coupled
described in Jing (2003). The principle, copying the operation mechanisms that hence allow more coherent and auditable
of the brain, is that the model consists of a series of intercon- modelling to support rock engineering and the other fields of
nected nodes and is trained to match a set of known inputs and endeavour.
outputs by altering weighting functions at the nodes. When a
person catches a ball, they do not solve a series of mathemati-
4.4 More international co-operation
cal equations to do so: they just catch the ball—using a neural
network that has spent a lifetime learning and memorising how On studying the early issues of journals, i.e., those published in
objects move around and how to catch them. Only the inputs the 1960s, there were often papers written by a single author—
and outputs of a neural network have a direct physical mean- because it was possible for a single person to make significant
ing: the values at the internal nodes cannot be directly related contributions. Although this single person endeavour is still
to specific geometrical or mechanical parameters. Indeed, this possible, it is certainly more stimulating and probably more
is the way that we humans analyse what we see and perceive: efficient for a group of people to tackle a given problem.
as the British painter David Hockney has said, “We always see Indeed, on looking at the contents of the April 2011 issue of
with memory”. the International Journal of Rock Mechanics and Mining Sci-
This computer learning and memorising concept should be ences, one notices that all 21 of the contributions are written by
extended, not only for the learning process in one rock engi- two or more authors. This trend will definitely continue as the
neering project, but so that the computer retains a memory of facilities for national and international co-operation continue
its analyses of all projects, i.e., the computer then begins to to improve.
52

excavation with significantly improved penetration rates will
be developed, as represented by the wider horizontal band
in Figure 15. This may be by some combination of continu-
ous ANFO blasting with the TBM continuously removing the
debris or by some entirely new procedure. The point is that
the energy potential is there.
4.8 Emphasis on environmental aspects

Figure 15. Energy input during blasting and the use of a TBM. It may appear at first sight that the construction of under-
ground excavations in rock is intrinsically environmentally
friendly—being out of sight—but it must be remembered
4.5 More use of neural network ‘intelligent’ computer that any engineering activity results in an increase in entropy.
programs Whatever, the order created in the engineered structure, there
Currently, the researcher/engineer inputs the required values will be a greater degree of disorder induced in the surround-
into the computer, runs the program, and evaluates the result. ing environment. This is often visibly manifested in mining
As mentioned in Section 3.7, there is considerable scope for operations but it is also true for all civil engineering projects.
augmenting this procedure through the increased use of more Of course, the disturbance can be reduced to some extent, but
‘intelligent’ computer programs, ranging from further devel- from thermodynamic principles it cannot be avoided.
opment of neural network architectures to the inclusion of This leads to the question, “Is the exploitation of under-
‘perception’ into the programs. Even in the near future, it is ground space compatible with the concept of sustainable
not beyond the realms of possibility to relay television images development?” The Bruntland (1987) definition of sustain-
of underground excavation surfaces, e.g., the roof, and arrange able development is
for the computer to automatically alert the user to developing “Development that meets the needs of the present genera-
roof fall hazards and to conduct continuing re-analyses based tion without compromising the ability of future generations
on new automated displacement values. to meet their own needs.”
4.6 Larger, deeper and longer excavations Thus, when planning a new rock engineering project, we
should ask the question, “Given the existing system (whether
As the demand for materials and underground space continues a virgin rock mass or a rock mass already containing devel-
to increase, it is anticipated that there will be larger, deeper oped space), is it acceptable to locate a new facility in the
and longer excavations. Currently, one of the largest caverns rock mass, i.e., within the context of sustainable development
open to the public has a span of 61 m: the Gjøvik ice hockey as defined above? The whole subject of sustainable develop-
stadium in Norway. But how large can a cavern be? The Kirsch ment in rock engineering has not yet been properly tackled in
solution for the stress concentrations around a circular hole the above context and the subject urgently needs attention.
includes neither the absolute size of the hole nor the elastic
properties of the rock. In a rock mass free of fractures, such 4.9 The future of the ISRM
as can be found in the Canadian Shield, it should be possible
to construct caverns with spans considerably larger than 61 m. For the ISRM to have any purpose, it must provide a bene-
How far can we go? A 100 m span? A 1000 m span? No doubt fit to the members. Accordingly, the ISRM must continually
the record will soon be broken by some entrepreneurial group develop to provide the required benefits in our changing world,
of engineers. especially in relation to ‘immediate’ access to information.
The current Board of the ISRM, operating from 2007–2011,
has been engaged in a modernisation process to ensure that
4.7 Increased rate of mechanised excavation
the necessary benefits are provided to members. This has
It is difficult to imagine how rock support can be improved. included a survey of ISRM Members to establish what they
The rock blocks can be held together with rockbolts so that want, increasing the downloadable material from the ISRM
a fractured rock mass behaves literally more like a mono- website, the availability of papers from all the previous ISRM
lithic continuum. A resisting force can be induced through the symposia in pdf form via the OnePetro website, emailing a
use of introduced supports. Shotcrete can be used to provide digital newsletter to members on a regular basis, increasing
additional resistance and to combat deterioration of the rock the appreciation of the ISRM members’ achievements, ratio-
surface induced by weathering. nalising the style of the ISRM meetings, and organising lecture
Conversely, overall energetic principles indicate that mech- tours and field trips. There are many further possibilities for
anised excavation can be greatly increased. In Figure 15, the the future, including electronic conferences.
two ways that energy is input during current methods of tun- The advantages of interacting with other researchers and
nel construction are illustrated. When blasting, large amounts engineers has already been mentioned in previous sections
of energy are input for a few seconds at widely spaced inter- on coupled models and international co-operation. The same
vals, e.g., eight hours. This is represented in Figure 15 by the applies to increasing the co-operation with other Societies.
narrow vertical white bands. However, when a tunnel boring Additionally, cross-promotion of websites and events is being
machine (TBM) is cutting, a lower level of energy is contin- developed, together with interaction between similar com-
uously input, represented by the lower horizontal white band. missions and interest groups. However, some new method of
It is extraordinary that we have only these two excavation interaction needs to be developed. As is evident from the fact
methods characterised by the two extreme conditions. that the vast majority of papers published on rock mechanics
Since there is generally no problem in providing energy and rock engineering are by more than one author, interac-
to the tunnel construction, it is likely that new methods of tion is to everyone’s advantage. Perhaps, the current method
53

of ad hoc interaction is sufficient, but one feels that some 4.12 Extra-terrestrial rock mechanics and engineering
form of organised interaction would be more beneficial to the
For its first 50 years, the ISRM has studied rock mechanics on
subject itself.
Earth—but do all the fundamentals that we have developed
also apply on the planets? At the moment, we need a theo-
4.10 Future computing power retician to go through the fundamentals of rock mechanics
and establish whether they all apply ‘out of this world’. The
The greatest changes in rock mechanics and in the ISRM will information will be required in the longer term future for the
be caused by the continuing increase in computer power. It is facilities which will be constructed on and beneath the surface
not easy to predict totally new future technological develop- of asteroids and planets.
ments: no one predicted the emergence of the WorldWideWeb
at the time of the start of the ISRM in 1962. However, we
can certainly consider the extrapolation of current trends. For 5 THE LONG-TERM FUTURE FOR THE ISRM
example, at the current time, for US$1000, we can purchase
somewhere between the computing power of an insect brain Given the rapid advances in computing power and communi-
and a mouse brain. However, at some time between 2020 cation systems which are expected to continue, the long-term
and 2040, we will be able to buy the computing power of a future for the ISRM is not easy to predict.
human brain for $1000 and before 2062, when the ISRM will • On the one hand, one could predict that in a world with an
have its centenary anniversary, it is predicted that the $1000 increasing information overload, the ISRM will be increas-
will purchase the computing power of all human brains in the ingly needed to provide clear and focused guidance to its
world. What could/should we do with such massive comput- members on rock mechanics advances and new capabilities
ing power and how would it assist rock mechanics and rock as they develop.
engineering? What will the developments be? • On the other hand, one could predict that the type of user
Already most published papers are available online. profiling pioneered by Google will be increasingly effective
Microsoft and Google are currently enabling books from and it will only be necessary to be individually profiled
major libraries to be available on the Internet. In the long- to receive all the information that one needs in whatever
term, all publications will be available online. (Last year, discipline and sub-discipline one happened to be, and hence
the digital content of the Web was more than one zettabyte: the ISRM would not then be needed.
a million, million gigabytes). It will be possible for com-
puter robots to automatically search all literature relevant to No action can be taken now to anticipate this longer term
a given profile. Computers could then compile papers about future—because either of the two possibilities above could
the cumulative advances in rock mechanics via all the papers come to pass, probably that in the first bullet point followed
and information generated each year. If computers become by that in the second bullet point.
authors, how should they be identified, or identify themselves?
Will we be able to tell if the author is human or a computer?
(Currently, the semantic web is being developed with which 6 CONCLUSIONS
computers can analyse the content to aid their and human
interpretations.) The intention of this paper has been to outline the future for
Fully-coupled models will be able to be run virtually instan- rock mechanics and the ISRM. This has been done through
taneously and adjusted for feedback from the actual rock the Hippocratic method of considering the past, diagnosing
behaviour in real time. Computers will have memory and the present, and foretelling the future, but also with ‘blue skies’
perception. They will use the ‘intelligent’ methods that are thinking added.
currently being developed. The computer models will be
linked to the internet and continuously updated for all new 6.1 Problems that will be addressed in the next 50 years
developments. If computers are monitoring construction expe-
riences and already have installed or can themselves write It was noted that the past developments have been summarised
fully-coupled programs, will the computers be able to design in the 1993 compendium “Comprehensive Rock Engineering”
rock engineering projects on their own? It is tempting to say but that there are still many unsolved problems needing to be
that human input will always be required but will this actually tackled. The ones mentioned in this paper are as follows.
be so?—certainly in the near-future but not necessarily in the • Ensuring that geological knowledge is incorporated into
long-term future. rock mechanics studies, both in the technical and manage-
ment senses.
• Developing better methods of measuring and estimating
4.11 Conferences
the rock stress in a given rock mass.
In the near future, it will be an option whether we attend • Implementing the general use of a failure criterion that
a conference physically or electronically. Some people will includes all three principal stresses.
continue to wish to attend physically but, with more and more • Developing better methods of establishing and characteris-
pressure on everyone’s time, the electronic attendance will ing the geometry and mechanical properties of the fractures
be preferred by others. Electronic conferences have several in a given rock mass.
advantages, not least of which is that one could attend many • And the related problem of improving discrete fracture
more conferences, e.g., one every Monday morning. The Pow- network models for water flow predictions.
erpoint presentations and pdfs of papers would be instantly • Constructing a fully-coupled numerical model that captures
available. One could talk through the computer to any other all the required variables, parameters and mechanisms.
registrants. The registration fee would be much less than now • Developing a method for auditing rock mechanics mod-
and travel and accommodation expenses would be eliminated. elling and rock engineering design.
54

• Incorporating ‘intelligent’ computer methods, which have Berry, A. 1974. The Next Ten Thousand Years: A Vision of man’s
memory and perception, into rock engineering design. Future in the Universe. London: Jonathan Cape Ltd., 224 p.
Berry, A. 1995. The Next 500 Years: Life in the Coming Millennium.
6.2 Future trends London: Headline Book publishing, 338 p.
Brown, E.T. 2011. 50 years of the ISRM and associated progress
The predictions for the near future in rock mechanics and the in rock mechanics. Proc. 12th ISRM Int. Congress on Rock
ISRM were outlined through the anticipated developments and Mechanics: Harmonising Rock Mechanics and the Environment.
extrapolation of the current ISRM modernisation initiative. London: Taylor & Francis.
The anticipated developments mentioned in the paper are as Bruntland, G. (ed.) 1987. Our Common Future: The World Commis-
follows. sion on Environment and Development. Oxford, Oxford University
Press.
• Improved methods of accessing/collating information. Cosgrove, J.W. & Hudson, J.A. 2011. The structural geology contri-
• More emphasis on geophysical methods in site investigation. bution to rock mechanics modelling and rock engineering design.
• More integration of subjects (e.g., fully-coupled numerical Proc. 12th ISRM Int. Congress on Rock Mechanics: Harmonising
modelling). Rock Mechanics and the Environment. London: Taylor & Francis.
• More international co-operation. Feng, X.T. & Hudson, J.A. 2011. Rock Engineering Design. London:
• More use of neural network ‘intelligent’ computer pro- CRC Press, Taylor & Francis, 460 p.
grams. Follin, S. 2008. Personal communication.
• Larger, deeper and longer excavations. Franciss, F.O. 2010. Fractured Rock Hydraulics. London: CRC Press,
Taylor & Francis, 188 p.
• Emphasis on environmental aspects.
Hudson, J.A. (ed.) 1993. Comprehensive Rock Engineering. Oxford:
• Increased rates of mechanised excavation. Pergamon Press, Elsevier. 5 vols, 4407 p.
In terms of the events beyond the near future, the increase Hudson, J.A. 2008. The future for rock mechanics and the ISRM. In
in computing power was highlighted, together with A. Majdi & A. Ghasvinian (eds.), Proc. ARMS5 New Horizons in
Rock Mechanics. Tehran: Iranian Soc. Rock Mech., 105–118.
• Computers as authors, Hudson, J.A. & Cosgrove, J.W. 1997. Integrated structural geology
• Computers as rock engineering designers, and and engineering rock mechanics approach to site characterisation.
• Extra-terrestrial rock mechanics. Int. J. Rock Mech. Min. Sci., 34, 3–4, paper 136.
Hudson, J.A., Cosgrove, J.W., Kemppainen, K. & Johansson, E. 2011.
Two possibilities have been envisaged for the future of the Faults in crystalline rock and the estimation of their mechanical
ISRM in the longer term (more integration of science and properties at the Olkiluoto site, western Finland. Eng. Geol., 117,
engineering or less necessity to be integrated), but no action 246–258.
is required now because either of these two possibilities could Jing, L. 2003. A review of techniques, advances and outstand-
prevail. We will have to await the outcome. ing issues in numerical modelling for rock mechanics and rock
engineering Int. J. Rock Mech. Min. Sci. 40, 3, 283–353.
Jing, L. & Stephansson, O. 2007. Fundamentals of Discrete Element
7 POSTSCRIPT Methods for Rock Engineering: Theory and Applications. Oxford:
Elsevier, 545 p.
The theme of this paper has been the prediction of rock Lee, M.F. 2006. In situ rock stress measurements in Western
mechanics and rock engineering over the next 50 years. Those Australia’s Yilgarn Craton. In M. Lu, C.C. Li, H. Kjørholt &
interested in general technological developments and their H. Dahle (eds.), In Situ Rock Stress: Measurement, Interpretation
implications over a much longer period are referred to the and Application: 35–42 London: Taylor & Francis.
books by Adrian Berry: “The Next 500 Years” and “The Next Price, N.J. & Cosgrove, J.W. 1990. Analysis of Geological Structures.
10,000 Years”, see the reference section. Cambridge University Press, 502 p.
Tang, C.A. & Hudson, J.A. 2010. Rock Failure Mechanisms. London:
Taylor & Francis, 322 p.
ACKNOWLEDGEMENTS Ulusay, R. & Hudson, J.A. (eds.) 2007. The Complete ISRM Sug-
gested Methods for Rock Characterization, Testing and Monitor-
For their many suggestions, which have supported the ISRM ing: 1974–2006. The compilation prepared by the ISRM Turkish
modernisation programme in 2007–2011, the author is grate- National group on behalf of the ISRM, 2007.
ful to the ISRM Board Members: Claus Erichsen, Xia-Ting Wang, J.S.Y., Smeallie, P.H., Feng, X.T. & Hudson, J.A. 2011.
Feng, Abdolhadi Ghazvinian, Alvaro Gonzalez-Garcia, Nuno Underground research laboratory network. Proc. 12th ISRM Int.
Grossmann, Luis Lamas, Francois Malan Derek Martin, and Congress on Rock Mechanics: Harmonising Rock Mechanics and
Tony Meyers. the Environment. London: Taylor & Francis.
Zang, A. & Stephansson, O. 2010. Stress Field of the Earth’s Crust.
Springer, 322 p.
REFERENCES
Barton, N. 2011. From empiricism through theory to problem solving

in rock engineering. ISRM Müller Lecture. Proc. 12th ISRM Int.
Congress on Rock Mechanics: Harmonising Rock Mechanics and
the Environment. London: Taylor & Francis.
55

New developments of rock engineering and technology in China
Q.H. Qian
Chinese Society for Rock Mechanics and Engineering
ABSTRACT: In terms of rock engineering and technology in hydropower construction, the slope stability and monitoring
techniques for high slopes of Three Gorges Project, the stability and support technology for high slopes of hydropower projects
in deep river valleys, the stabilization techniques for underground cavern group with large span and high side walls are
introduced in this paper. As for rock engineering and technology in highway and railway construction, the Qinghai-Tibet
Railway – new construction techniques in permafrost, the support techniques for large squeezing deformation in Wuqiaoling
Tunnel, the construction techniques for tunnels in alpine and high-altitude region, the geological prediction techniques for
tunnels in karst region, the microseismic monitoring and early warning techniques for rockbursts in deep and long tunnels are
presented. For rock engineering and technology in mining engineering, the innovative techniques for roadway support in mines,
the simultaneous extraction technique of pillarless coal and gas in coal seams with low permeability, the safe and efficient
deep open mining technology, advances in monitoring, early warning and treatment of mine dynamic disasters are presented.
In addition, the new anchorage techniques and precision blasting technique in rock engineering are introduced.
Keywords: Rock engineering; Hydropower engineering; Tunnel engineering; Mining engineering; Anchorage technique;
Blasting technique
1 INTRODUCTION
Rock engineering in China, including large-scale water con-

servancy and hydropower engineering, railway and road tunnel
engineering, mining and energy resources engineering and
subway engineering, no matter on the ground surface or under-
ground, has been at the forefront of the world in terms of
large scale, great difficulty, great number and high construc-
tion speed. A considerable portion of these projects are located
in rugged and complex geological conditions and subjected to
high geological hazard risks. The technical problems encoun-
tered in these projects are unprecedented. In recent years,
Chinese scientists and engineers in rock mechanics and engi- Figure 1. Typical reinforcement section of shiplock high slopes.
neering have devoted a lot of efforts in solving these technical
problems and a number of innovative achievements have been
made.
great height, complex form, extensive coverage, with fully
released stresses, and exhibits obvious unloading and inho-
mogeneous characteristics. b) In order to ensure normal and
2 ROCK ENGINEERING AND TECHNOLOGY IN
smooth operation of the golden watercourse and the shiplock,
HYDROPOWER CONSTRUCTION
the requirements on slope stability and deformation were very
stringent. c) The slope construction was challenging. It was
2.1 Slope stability and monitoring techniques for high
very difficult to construct the narrow, deep and steep ver-
slopes of Three Gorges Project
tical side walls of the shiplock. In addition, the side walls
One of the major achievements in the construction of the Three were constructed simultaneously with underground tunnels
Gorges Project is the stability and monitoring techniques for and shafts. How to solve the interaction between excava-
shiplock high slopes. The permanent shiplock is located in the tion blasting, to minimize the damage to rock masses and
mountain on the left bank of the dam (see Figure 1). to ensure construction safety were the challenging problems
The high rock slope was artificially excavated, with the during construction.
maximum excavated height of 173 m. A 50–60 m division The technical problems involve engineering geological
pier was retained between the two channels of the permanent study on high slopes, mechanical properties of the rock
shiplock. The maximum gross head in the shiplock is 113 m, masses, underground seepage and drainage measures, opti-
and the permanent shiplock is the one with the largest scale, mization of construction methods and excavation and blasting
the most steps and the highest total design head in the world. techniques, anchoring techniques for high slopes, stability
Compared to normal high slopes, the high slopes for the analysis of high slopes and etc. The research results provided
permanent shiplock has the following features: a) As the slope a large amount of qualitative and quantitative information on
is a steep high slope by deep cutting in the mountain, it is of stability of the steep high slopes of the Three Gorges shiplock.
57

A variety of two-dimensional and three-dimensional numeri- The maximum accumulative displacements at the south and
cal modeling was performed. The results provided a reliable north sides of the division pier were 23.16 mm and 31.50 mm,
basis for design, construction and scientific evaluation of respectively. After the shiplock was put into operation, the
slope stability. displacement at the lock head was not more than 0.5 mm
The technical achievements for stability evaluation and during filling in the shiplock chamber, which can meet the
monitoring for the high slopes of the Three Gorges shiplock requirement of normal operation of the gate and was within
are manifold. In the following section, the comprehensive the designed range.
anchoring technology is introduced.
About 36,000 ordinary steel rock bolts, 100,000 high-
strength structural bolts, 229 anchoring cables with 1000 kN
2.2 Stability and support technology for high slopes of
prestress and 3975 anchoring cables with 3000 kN prestress
hydropower projects in deep river valley
were installed and 1054 rock blocks were supported in the high
slope of the Three Gorges shiplock. The anchored volume and With the economic development and great demand on clear
the quantity of reinforced blocks are the largest in the world. energy in China, the development of hydropower in south-
In order to fully utilize the performance of anchor cables west China has shown rapid momentum. A number of large
with different structural forms, different types of cables were (giant) scale hydropower projects are being constructed or
employed at different parts of the shiplock, according to the in the survey and design stage. For example, on the main
structural features of the shiplock slopes. stream of the Jinsha River, a tributary of Yangtze River, 20
The anchorage cables used in the high slope of the shiplock cascade hydropower stations have been planned, among which
included: end-anchored cables with 1000 kN or 3000 kN pre- 17 hydropower stations have capacity over 1000 MW and four
stress and cross-anchored cables with 3000 kN prestress. have capacity over 5000 MW. 21 cascade hydropower stations
Except 113 cables for monitoring and 121 cables for reinforce- have been planned on the main stream ofYalong River, among
ment of the concrete structure at the lock head were unbonded, which 10 stations have capacity of more than 1000 MW.
the rest were fully bonded anchor cables. The total installed capacity for Jinping I Hydropower Station
For the 1000 kN end-anchored cables, the designed pre- and Jinping II Hydropower Station is 8400 MW. 22 cascade
stress was 1000 kN and the over tension was 1150 kN. 7 hydropower stations have been planned on the main stream
steel wire strands with diameter of 15.24 mm and strength of of Dadu River, among which 7 station have capacity of more
1860 MPa were used. The designed strength utilization coef- than 1000 MW. The capacity of Pubugou Hydropower Station
ficient was 0.55, and the strength utilization coefficient for is 3600 MW. In addition, a number of hydropower stations
over tension was 0.63. The boreholes were 30–40 m in length will be constructed on the Lancang River and Hongshuihe
and 115 mm in diameter. The internal anchored section was River.
5 m in length and in a date-pit shape. The rapid uplift of the Qinghai-Tibet Plateau led to the
For the 3000 kN end-anchored cables, the boreholes were formation of a series of major rivers and deep river valleys
30–60 m in length and 165 mm in diameter. The internal with elevation difference of 2000–3000 m, as well as the well-
anchored section was 8 m in length. The designed prestress known major fault zones and seismic zones in west China.
was 3000 kN and the over tension was 3450 kN. 19 steel wire As these hydropower stations are located in the deep river
strands with diameter of 15.24 mm and strength of 1860 MPa valleys, the geological conditions are very complicated. Dur-
were used. The designed strength utilization coefficient was ing construction, complex high slope problems are frequently
0.606 and 0.697 for over tension. The other parameters were encountered in the large-scale engineering projects. This type
the same as the 1000 kN anchor cables. of high slopes has the following prominent features: a) The
The 3000 kN cross-anchored cables were mainly used for natural slope is very steep. The slope angle is generally more
the division pier, the systematic reinforcement of the south than 45◦ degrees, with most above 70◦ –90◦ . The valley depth
and north slopes, cross anchor of the south and north slopes is usually more than 1000 m; b) The height for the cut slope
and the deep drainage tunnels and cross anchor of the divi- is large, usually more than 300 m, with the maximum more
sion pier. Anchor piers were constructed at each end of the than 600 m; c) The geological conditions of the slope are
cables. The cables were prestressed and unbonded. The bore- complicated, featured by faults, high geo-stresses, intensive
holes for anchor cables were 40–60 m in length and 165 mm in unloading and large unloading depth. Jinping I Hydropower
diameter. The other parameters were the same as the 3000 kN Station is a typical example.
end-anchored cables. Specially designed unbonded anchor The project area for Jinping I Hydropower Station is a typi-
cables were used for the concrete structure at the lock head to cal deep V-shaped valley, with relative elevation difference of
cater for the possible deformation between concrete and rock 1500–1700 m. The left bank is a counter slope. Below eleva-
masses. In view of the alternating load by the gate at the lock tion 1820–1900 m is marble and the slope angles are 55◦ –70◦ .
head, the outer anchor head was peeled for 8 m to form bonded Above is sandy slate and the slope angles are 40◦ –50◦ . The
section. In this way, the outer anchor end was double secured local topographic feature is characterized by alternative ridges
by both anchorage devices and 8 m bonded section. and gullies.
In addition, according to the characteristics of hydropower The arch dam of Jinping I Hydropower Station is the high-
engineering, the durability of anchor cables was studied sys- est in the world. The excavated height of the left abutment
tematically. Corresponding measures were taken to enhance slope reached 530–540 m. The geological condition was very
the durability and practicability. complex. The design and implementation were extremely
As of 20 July 2007, the maximum accumulative displace- difficult.
ments along the axial direction of the shiplock on the south Chengdu Hydroelectric Investigation & Design Institute,
and north slopes were 72.07 mm and 52.96 mm, respectively. together with a number of research institutes and universities
The maximum displacements in the vertical wall of the south in China, conducted in-depth and detailed studies on the dam
and north slopes were 36.93 mm and 30.08 mm, respectively. site. The details are shown in Figure 2.
58

Figure 2. Methodology and main contents for study on high slopes.
The reinforcement and treatment techniques for high slopes The shear-resistance tunnels were constructed on eleva-
mainly included: tions 1834 m, 1860 m and 1885 m. The cross section was
9 m×10 m. The fault and shear zones were removed. They
1) Water cut-off and drainage system: Cut-off ditches around
were then back filled by concrete after the surrounding rock
the slope, surface drainage ditches, underground drainage
masses were grouted.
holes and drainage adits were constructed so as to avoid
5) Grouting: The unloaded fractures and unloaded rock
water seepage into the excavated slope, which may induce
masses were grouted by utilizing the available anchor
slope failure.
holes. Grouting was carried out in the slope by using the
2) Reinforcement by rock bolts and anchor bars: In order to
hole for anchor bars in the slope berm and the pre-grouting
ensure rock mass stability and construction safety, system-
holes between berms.
atic rock bolts and anchor bars were employed to reinforce
the loose rock masses in the shallow excavated slope. A monitoring system for the slope was established. The
3) Layout of anchor cables: For the unloaded sandy slate monitoring program included: (1) horizontal and vertical
above elevation 1820 m and the rock mass at the outcrop displacements by geodetic survey; (2) monitoring of the dis-
of fault f42-9 near elevation 1800 m, anchor cables with placement in deep boreholes; (3) monitoring of displacement
prestress of 2000 kN were adopted (prestress of 3000 kN in deep rocks by the graphite rod convergence method and
may be applied when necessary). The spacing of anchor the distance measurement pier method; (4) monitoring of the
cables was 4 m×4 m and the length varied between 40– reinforcement effect, including stress in rock bolt and load
80 m. The unloaded marble rock masses were reinforced in anchor cable; (5) inspection monitoring; (6) monitoring of
by randomly distributed anchor cables with prestress of opening and closing of cracks, shear and dislocation of faults;
2000 kN–1000 kN. The shallow rock masses near the slope (7) monitoring of seepage pressure. More details are listed in
open line were reinforced by anchor cables with prestress Table 1.
of 2000 kN and spacing of 5 m×5 m. The layout of monitoring instruments is shown in Figure 3.
4) Replacement by shear-resistance tunnel: Fault f42-9 was In order to effectively analyze and manage the vast amount
the potential slip surface for overall slope failure. Shear- of monitoring data, a high slope stability monitoring system
resistance tunnels were constructed for reinforcement. for hydropower engineering was developed. With this system,
59

Table 1. Layout of monitoring instruments in the left abutment slope.
Structural Monitoring Monitoring item

system object and method Layout of monitoring instruments
Dip slope Toppling deformation at Surface displacement, 10 observation points along the external border of the top
structure prone the back slope of the shallow rock excavation line, 3 monitoring sections in the excavated
to toppling cableway platform above deformation, load in slope (A-C), 15 external observation points, 13 multi-point
elevation 2000 m anchor cable (rock bolt) extensometers, 13 rock bolt dynamometers, 74 anchor line
dynamometers based on 5% of the number of anchor cables
Wedge-shaped Rock mass with Surface displacement, 8 monitoring sections below elevation 1960 m, 55
structure with tensile fracture (large relative displacement external observation points, 20 multi-point extensometers
double sliding block between elevation of structural planes passing through fault f42-9 and lamprophyre dyke (X) and
surface 1800–2020 m) 54 rock bolt dynamometers
Block-shaped Small local blocks Surface displacement, 8 monitoring section below elevation 1960 m, 41
slope structure outside the rock mass displacement in shallow multi-point extensometers between elevation 1885–1960 m,
with tensile fracture rock mass, load in anchor 82 cable line dynamometers, 27 rock bolt dynamometers,
and shallow rock masses cable (rock bolt) 3 monitoring points for groundwater, 4 boreholes for
inclinometers. Monitoring points for surface displacement
of other blocks
Deep fractures Five-layer monitoring Horizontal and vertical Exploratory pits PD44, PD42 at elevation 1930 m, drainage
pits displacements in the tunnel at elevation 1915 m and curtain grouting pits at
directions perpendicular elevation 1785 m, 1829 m and 1885 m as monitoring pits,
to the river 1–2 measurement lines for horizontal displacement
perpendicular to the river, 9 convergence graphite rods,
10 measurement piers, 3 sliding micrometers
Shear-resistance Three-layer shear- Deformation in the 25 monitoring sections in 8 caverns, 24 three-point
replacement resistance tunnel for surrounding rock mass, extensometers, 8 single-point rock bolt dynamometers,
tunnel fault f42-9 , two-layer deformation in faults and 16 three-point rock bolt dynamometers, 28 dislocation
shear-resistance tunnel dykes, stress and strain in meters, 18 joint meters, 16 rebar strain meters, 12 osmometers,
for fault f5 , three-layer steel-reinforced concrete 10 sets of 5-D strain gauges, 10 non-stress meters
replacement tunnel for
lamprophyre dyke (X)
Resisting force Four-layer monitoring Horizontal and vertical 12 monitoring sections in certain grouting pits at elevations
body in the pits displacements 1829 m, 1785 m, 1730 m and 1670 m, 3–4 measurement lines
left bank for horizontal displacement perpendicular to the river,
27 multi-point extensometers, 27 rock bolt dynamometers,
36 rebar strain meters, 4 joint meters, 12 convergence sections
management and integration of monitoring data, construction walls exists not only during construction, but also during the
information, geological information, hydrological and meteo- long-term operation period.
rological information can be implemented. The system has the The main caverns for a large-scale underground hydropower
functions of data mining and analysis, prediction and graphic station in the powerhouse, the transformer chamber, the surge
output of analysis results. The system can realize comprehen- chamber or tailrace gate chamber, as shown in Figure 4.
sive and dynamic management of distributed network for field Different excavation sequences for underground cavern
monitoring data. group would result in different stress paths. Research on
With the above slope reinforcement measures, systematic the optimum construction sequence is one of the key issues
monitoring indicated that the slope deformation gradually in the study on the stability and support measures for
converged within a few years, which proves that the reinforce- large-scale underground cavern groups. Currently, for the
ment measures were effective. Successful implementation of study on excavation sequence of underground cavern groups,
the slope construction and monitoring indicates that the slope the analysis method has been developed from planar anal-
engineering in China has achieved a new leap and the level of ysis to quantitative analysis reflecting three-dimensional
engineering geological investigation, stability analysis, rein- nonlinear characteristics of large-scale underground cavern
forcement techniques, monitoring system for excavation and groups. In combination with the optimization of construc-
construction has been advanced. tion organization, the mechanical analysis and construction
system engineering can be integrated closely for quantitative
2.3 Stabilization techniques for underground cavern group assessment.
with large span and high side walls By summarizing the construction conditions of the under-
ground cavern groups which have been completed or are under
In recent years, a great number of underground caverns construction, the principle of excavation sequence is proposed
(groups) have been constructed for energy resources, trans- as “arch first and then side walls, from top to bottom, excava-
portation, mining and defense industries. Taking the under- tion in layers, support installation in layers, multiple processes
ground hydropower station as an example, the span, height in the horizontal direction, multiple layers in the vertical
and scale have been in the front ranks of the world. direction, cross operation of multiple work faces”. Figure 5
The stability problem of the surrounding rock masses of shows a sketch of stratified excavation of the powerhouse and
underground caverns (groups) with large span and high side transformer chamber.
60

Figure 3. The major geological structures and layout of monitoring instruments in the slope.
Figure 4. Sketch of cavern group for a large-scale underground Figure 5. Sketch of stratified excavation of powerhouse and trans-
powerhouse. former chamber (unit: cm).
61

As rock mass is a fuzzy system containing various complex
factors and constitutive relations, in order to compensate the
gap between the feasibility study, design and the engineer-
ing construction, after the construction scheme is selected,
rapid monitoring and feedback analysis system shall be estab-
lished during construction. The output information (stress
and deformation, etc.) under certain engineering measures
(excavation sequence, blasting parameters, support measures)
needs to be investigated. The structure and support parame-
ters in the construction scheme shall be continuously modified
and optimized so as to assure the economy and safety of the
underground cavern group construction.
The underground cavern group of Shuibuya Hydropower
Station on Qingjiang River in Hubei Province, China, passes
through interbedded hard and soft rock strata. The strength
of soft rock is relatively low. A great amount of shear zones
exist within rock strata or at the interfaces between hard and
soft rock strata, where shear slip or creep squeezing are likely
to occur under excavation unloading. In view of the com-
plex geological condition of the Shuibuya underground cavern
group, the researchers fromYangtze River Scientific Research
Institute conducted systematic rock mechanics studies and
proposed reinforcement treatment measures for the interbed-
ded soft and hard surrounding rock masses, which mainly Figure 6. Arrangement of support for underground powerhouse.
included:
(1) Replacement of soft rock below crane beam plan, a highway network with 7 rays, 9 longitudinal lines and
As the crane beam is a load-bearing structure, it has strict 18 transversal lines with a total mileage of about 108,000 km
requirements on deformation control. The rock layer in will be constructed before year 2015.
the upper side wall of the powerhouse was replaced by By the end of year 2007, the railway traffic mileage in
concrete. Excavation of the powerhouse started only after China has been 78,000 km, ranked no. 1 in Asia. According to
concrete replacement and consolidation grouting. plan, the railway traffic mileage will reach 100,000 km before
(2) Shotcrete-rock bolt support for the powerhouse 2020, including the high-speed railway network. Railway con-
The roof and side walls of the caverns and the tailrace struction is challenging in southwest China. For instance,
tunnel were supported by systematic steel fiber- Yunnan-Tibet Railway is 1594 km in total length, passing
reinforced shotcrete and mortar-grouted rock bolts. through the Hengduan Mountain in northwest Yunnan, the
(3) Reinforcement and protection of soft rock below the high mountains and deep gorges in southeast Tibet and the val-
powerhouse leys in south Tibet. The tectonic movements are very intensive
In order to control the deformation in the soft rock along the railway. According to the available data, the average
below the powerhouse, comprehensive treatment mea- annual uplift is 12 mm, and it is 12–50 mm in the Jolmolungma
sures were adopted, including anchor piles, grouting and peak region. Along Yunnan-Tibet Railway, the section with
steel-reinforcement of surface concrete. fortification intensity ≥IX is about 127 km in length, occupy-
(4) The structural planes such as interbedded shear zones and ing 8% of the total length. There are 392 extremely large, large
faults were reinforced by prestress anchor cables so as to or heavy-duty bridges with a total length of 108.9 km. There
restrain the dislocation of interbedded shear zones. are 419 tunnels with a total length of 491.8 km, among which
(5) Excavation and support of tunnel intersection the longest one is 12.59 km. The technical difficulties encoun-
At the intersections between the caverns and the power- tered during construction of Yunnan-Tibet Railway were rare
house, the excavation and interlocking support were com- in the world.
pleted before the powerhouse excavation face reached the
corresponding elevation. In this way, the stress distribu-
tion at the intersections can be enhanced and the defor- 3.1 Qinghai-Tibet Railway – new construction techniques in
mation can be reduced. Figure 6 shows the arrangement permafrost
of support for the underground powerhouse. The most prominent achievement in railway construction in
Excavation of the powerhouse started in 2002. Up to the first recent years is the completion of Qinghai-Tibet Railway. The
half of year 2005, excavation and support of the powerhouse total mileage from Golmud to Lhasa is 1142 km, among
were completed. As of July 2007, the first power generation which 960 km railway is above elevation 4000 m. The highest
unit started to operate. The monitoring results indicated that elevation along the railway is 5072 m.
the surrounding rock masses were overall stable. The construction team for Qinghai-Tibet Railway have
successfully overcome the three world-wide engineering
challenges, namely, permafrost, alpine hypoxia, ecological
3 ROCK ENGINEERING AND TECHNOLOGY IN fragility and realized a number of technical innovations. As
HIGHWAY AND RAILWAY PROJECTS for the key engineering techniques on permafrost, researchers
from China Railway First Survey and Design Institute con-
The total mileage of roads in China has reached 2.3 million km ducted systematic studies on the impacts of climate change.
by 2010, and will reach 3 million km by 2020. According to National technical standards for site investigation, design and
62

construction of railways in plateau permafrost were developed. blasting parameters so as to minimize disturbance to the
The design ideas of initiative cooling, foundation cooling surrounding rock;
and protection of frozen soil and key parameters were iden- 2) To optimize construction parameters. The extra-short sub-
tified. Air-cooling rubble embankment, subgrade with ther- step excavation method was adopted so that the close loop
mosyphons and bridges across special harmful frozen sections can be formed as early as possible for the surrounding rock
were adopted innovatively. A complete set of construction and primary linings.
techniques for subgrade in permafrost were successfully 3) The principle of “multiple support” was adopted for pri-
developed, for instance, tunneling lining against frost-heave. mary support. Shotcrete (with a thickness of 20 cm) was
The natural conditions along Qinghai-Tibet Railway are applied together with systematic rock bolts immediately
harsh. In view of the large group of people, high altitude after excavation. The deformation in the surrounding rock
and long operation time during construction, innovative tech- can be in line with that in the supporting system. Part of
niques were developed for integrated treatment of altitude the displacement was released before installation of steel
sickness and the monitoring system for plague prevention support and re-application of shotcrete (with a thickness
was established. Zero-death rate due to altitude sickness and of 15 cm). By this way, the squeezing deformation can be
zero plague infection among human beings were achieved. partially restricted.
The ecological environment is very fragile in Tibet Plateau 4) Proper amount of preset deformation and suitable time for
and the railway passes through the natural reserves such as erection of secondary linings were determined according
Source of Three Rivers. The builders conducted compre- to the actual deformation curve.
hensive researches on and developed innovative practices in 5) The field monitoring was enhanced and the large squeezing
protection of wildlife, alpine vegetation and river sources, deformation was dynamically controlled. Importance was
etc. During such large-scale engineering construction, the attached to geological forecast.
ecological environment was well protected and the har-
mony between the construction and natural environment was
realized. 3.2.2 Construction techniques for tunnels in alpine and
high-altitude region
The construction techniques for tunnels in alpine and high-
3.2 Advances in tunnel construction techniques altitude regions with complex geological conditions in China
12,000 tunnels of different types have been put into use in are world-leading. For instance, during construction of the
road and railway industries, extending over 7000 km mileage. Fenghuoshan Tunnel of Qinghai-Tibet Railway, a series of
Among them, more than 7000 tunnels are railway tunnels with techniques were developed, including tunnel drainage thermal
total mileage of over 4000 km; over 4000 tunnels are road insulation construction technique in plateau permafrost, low-
tunnels with total mileage of more than 2000 km. A number temperature early-strength concrete construction techniques,
of tunnels under river beds have been constructed, including smooth blasting technique for tunnels in plateau permafrost,
3 tunnels crossing the Yangtze River and 8 tunnels crossing health security and environment protection during tunnel
the Huangpu River, 2 tunnels crossing the Yellow River for construction in permafrost.
the South-North Water Diversion Project and tunnels in other During construction of the Erlangshan Tunnel and the
cities crossing the Zhujiang River and the Yongjiang River. Zhegushan Tunnel of Sichuan-Tibet Highway, three key tech-
Some tunnels are constructed below lakes in cities, such as nical difficulties, brought by large-scale road tunnel con-
the Xuanwu Lake in Nanjing, the West Lake in Hangzhou, struction in high-altitude region with complex geological
the Dusu Lake in Suzhou, the Qingshan Lake in Nanchang. conditions, harsh climate conditions and special operation
A few submarine tunnels have been constructed, including conditions, were solved. A series of innovative comprehensive
the Xiamen-Xiang’an tunnel, the Jiaozhou Bay estuary tun- technical achievements were made, including:
nel, the Guangzhou Biological Island-University City tunnel 1) Semi-transverse and longitudinal parallel heading ventila-
and the Shiziyang railway tunnel. Tunnels to be constructed tion mode in single-hole two-way extra-long road tunnels
include the Qiongzhou Strait tunnel, the Bohai Bay (Dalian- was adopted first in the world. The relevant key theoretical
Penglai) tunnel, the Hangzhou Bay (Shanghai-Ningbo) tunnel and technical problems were solved.
and Dalian Bay tunnel. 2) A technical system of freeze-resistance design for extra-
long road tunnels constructed in seasonal freeze-thaw
3.2.1 Support techniques for large squeezing deformation in
regions was established.
Wuqiaoling Tunnel
3) The altitude coefficient in the range of 400–5000 m was
The Wuqiaoling Tunnel of Lanzhou-Xinjiang Railway has a
first tested with consideration of smoke and CO, which
total length of 20.05 km and overburden depth of 500–1100 m.
filled the blank in “Design standard of ventilation and
The tunnel passes through “squeezing zones” formed by a
lighting in rock tunnels” in China.
number of large regional faults. The large squeezing defor-
4) An integrated technical system of intelligent monitor-
mation was significant under high geo-stresses in soft rocks.
ing, maintenance and management of electromechanical
The deformation of was more than 1 m in the crown, and over
equipments in tunnels was established based on GIS. Intel-
60 cm in the side walls. Researchers from China Railway First
ligentization, networking, configuration and integration of
Survey and Design Institute adopted the principles of “short
monitoring and management of electromechanical system
excavation footage, rapid closure, strong support, early for-
were realized for large-scale road tunnels in alpine regions.
mation of loop and timely installation of secondary lining”,
5) According to the cross sectional features of road tunnels,
and successfully solved the technical difficulties over large
some anti-shock measures were proposed for rock tun-
squeezing deformation. The measures included:
nels in high-intensity earthquake zones, such as addition
1) To adopt auxiliary measures for advanced support of of damping layers and use of polymer steel or steel fiber
soft rock passing through faults. To strictly control the reinforced concrete lining.
63

The Erlangshan tunnel has stood the test of the Wenchuan
earthquake and it is almost intact after the earthquake, which
indicates that the construction techniques are correct and
effective.
3.2.3 Geological prediction techniques for tunnels in karst

region
In recent years, great progress has been made in predic-
tion of poor geological conditions during tunnel construction.
Researchers in Shandong University made comprehensive
geological prediction for over 20 tunnels in karst regions
along Shanghai-Chengdu-Xichang Highway, with prediction
accuracy of 70%. The main technical advances include:
1) Based on the analysis of a large amount of geological
investigation data, the main controlling factors for karst
development, vertical and horizontal distribution features
were analyzed. A qualitative assessment method for hazard
risk of karst tunnels was established. The corresponding Figure 7. The monitoring and analysis system for rockbursts during
grading standards of hazard risk for tunnel construction TBM tunneling for Jinping II Hydropower Station.
were proposed.
2) Aiming at the multiple solutions of geophysical investiga-
tion, the complexity of geological conditions, the features
and application scope for different forecast methods were
studied. Transient electromagnetic device for water inrush
and post-processing software for geological radar were
developed.The principle and techniques for comprehensive
geological forecast were proposed. The research results
were successfully applied in the geological forecast prac-
tices, which helped to avoid water and mud inrush and
ensured construction safety.
3) The principles and standards for treatment of water inrush
hazard in high-risk karst tunnels were studied. Based on the
analysis of different inrush types and features, the treatment
principle of “exploration before blocking, comprehensive
treatment” was established for karst water. With the use of
different grouting schemes and grouting materials, grout-
ing and blocking tests for different types of karst water were
successfully carried out at site. Figure 8. The cumulative microseismic events within 30 days
4) A four-color early-warning mechanism for hazards in karst before a very strong rockburst.
tunnels and a corresponding contingency plan were pro-
posed. The basis for release of early-warning information
and flowchart for information dissemination were identi-
for long and large tunnels, the geological structures, lithol-
fied. Emergency rescue scheme and evacuation route for
ogy, mechanical parameters of rock masses can be explored
high-risk karst tunnels were established.
and the occurrence region of rockburst is related to the excava-
tion progress and highly repeatable. Therefore, compared with
3.2.4 Rockbursts in deep and long tunnels and microseismic natural earthquakes, rockburst prediction is more feasible.
monitoring and early warning In 2009, researchers from Dalian University of Technology
In recent years, rockbursts have occurred frequently in many established a movable microseismic monitoring system which
deep tunnels during construction in the hydropower and trans- can advance together with TBM in tunnel construction for
portation engineering in China and have become a technical Jinping II Hydropower Station. With continuous data acquisi-
bottleneck in deep underground engineering. tion by the microseismic data acquisition system, remote data
Rockburst prediction is a commonly recognized challenge transmission (Dalian Mechsoft Server), data processing and
in the world. In our opinion, in terms of mechanism, excava- analysis and self-developed Chinese visualization software
tion breaks the initial stress balance and leads to stress increase MMS-view, 24-hour continuous monitoring and analysis of
and further energy concentration in the local surrounding rock microseismic activities during tunnel excavation was realized.
masses, which further results in localized deformation and Continuous acquisition and collective analysis of the seismic
induces microfracturing in rock masses. The static balance monitoring data were achieved, which provided an important
state in rock masses develops towards dynamic instability and platform for studies on rockburst monitoring and prediction
finally rockburst is induced. Compared with natural earth- during TBM tunneling (Figure 7).
quakes featured by long period, low occurrence rate and deep Figures 8 and 9 show the cumulative microseismic events
seismic source, the seismic source zone for rockburst is acces- within 30 days before a very strong rockburst and the varia-
sible (such as the excavation face) and rockbursts may occur tion of density nephogram of microseismic events before the
in a short period and multiple times. It is more important that, rockburst, respectively.
64

Figure 10. The structure of constant-resistance and large-
deformation rock bolt.
(2) In terms of distance, some failure precursors can be

detected by the microseismic sensors for rockbursts tens
of or more than 100 meters away. The approximate range
of strong rockburst can be located by the microseismic
monitoring system.
(3) In terms of time, some precursors usually appeared a few
days before the rockburst event. As the occurrence of
rockburst is related to the excavation progress, the exact
time of rockburst can hardly be predicted although the
location of rockburst may be determined in advance.
(4) In terms of magnitude, at present, only the relative mag-
nitude can be obtained by using analogy according to the
site experiences.
(5) The sophisticated RFPA simulation showed that stress
concentration was evident near structural planes (such
as faults or joints), which shall be the focus of rockburst
monitoring. By integrating the microseismic monitoring
and the sophisticated RFPA simulation, the feasibility of
rockburst prediction is expected to be enhanced.
For effective treatment of rockbursts during tunnel con-
struction for Jinping II Hydropower Station, in addition
to the microseismic prediction and early warning system,
researchers from China University of Mining and Technol-
ogy (Beijing) developed the constant-resistance and large-
deformation rock bolt. The structure is shown in Figure 10
and the main mechanical properties are shown in Figure 11.
(1) Static tensile stress-strain curve
The results of static tensile tests indicate that the max-
imum deformation of the constant-resistance and large-
deformation rock bolt ranges between 300–1000 mm and
the average constant resistance is 120 kN–130 kN.
(2) Dynamic stress-strain curve
The dynamic impact test indicates that the constant-
resistance large-deformation rock bolt can withstand
a number of impacts. Under the constant resistance,
the maximum cumulative impact deformation ranges
between 500 mm–1000 mm.
1600 sets of such rock bolts were installed at necessary loca-
tions in the hydraulic tunnels for Jinping II Hydropower
Station.
Figure 9. Variation of density nephogram of microseismic events
before a rockburst (Rockburst occurred in the tunnel section below 4 ROCK ENGINEERING AND TECHNOLOGY IN
point B). MINING ENGINEERING
Mining engineering is the largest rock engineering field

The following conclusions can be made from the successful among various industries in national economy of China.
practices of rockburst prediction for tunnel construction in
Jinping II Hydropower Station: 4.1 Innovative techniques for roadway support in mines
(1) Precursory microcracking exists in prior to most rock- Research staffs from China Coal Research Institute devel-
bursts, which can be captured by the microseismic oped a series of innovative technique systems for roadway
monitoring system. support. These techniques can solve complex and difficult
65

Figure 11a. Static tensile stress-strain curve.
Figure 11b. Dynamic stress-strain curve of MG-12-1 rock bolt
(500 mm impact height).
roadway support problems, speed up construction of high-
yield high-efficiency mines, substantially increase coal output 4.2 Simultaneous extraction technique of pillarless coal and
and efficiency and significantly improve safety in roadways. gas in coal seams with low permeability
The main innovative achievements are: More than 70% of coal resources in China are located in
1) “Primary support” shall be strengthened. Systematic rock geological conditions with low permeability, high gas con-
bolts of high strength and high stiffness should be installed. tent, high adsorption and soft coal seams. Gas accidents
Support shall be completed in one time so as to control occur frequently, which have been one of the prominent
deformation in the surrounding rock masses. issues in safe mining. Research staffs from Huainan Mining
2) A dynamic information design method for rock bolt sup- Group have successfully developed simultaneous extraction
port was proposed. The design is a dynamic process. The techniques of pillarless coal and gas in coal seams with low-
method takes full advantages of information provided by permeability, which opened a new direction for safe mining in
each process, collects real-time information collection, coal seams with low permeability and high gas content. The
analyzes and feedback information. main achievements included:
3) Combinational reinforcement techniques were developed 1) Relief mining was adopted to enhance the permeability
with grouted rock bolts, grouted anchor cables and inte- of the coal seams. Roadways and boreholes were laid in
grated drilling-bolting-grouting method. advance in the mining area to extract gases. Gas extraction
4) A set of equipments for monitoring mine pressure and by relief mining and simultaneous extraction of coal and
safety of roadways were developed, including, rock bolt gas were realized.
dynamometer, cable line dynamometer, roof separation 2) Aiming at soft coal rock, easy separation and problems in
detector and multi-point extensometer, etc. The instru- controlling the surrounding rock in the relief mining area,
ments have been widely applied in the mines. comprehensive measures and combinational rock bolts,
5) A series of boring machines for bolting were developed. anchor cables and wire meshes and grouting techniques
The boring machines for bolting have been continuously were adopted.
improved according to the mine roadway conditions in 3) The traditional U-ventilation was changed to Y-ventilation
China. The performance indicators have reached interna- with goaf roadways (replacing the pre-arranged lanes spe-
tional advanced level. cially for gas extraction). The gas extraction rate at the
The results have been applied to 58 large or medium-scale working face can reach 70%, which assured that safe and
mines in 20 provinces in China. efficient mining in coal seams with high gas content.
66

4.3 Safe and efficient deep open mining technology “Stress transfer” includes stress relief technique in deep
large-diameter boreholes in coal seam, pre-fracturing of
More than 80% of iron ore is produced by open mining. At
hard roof, floor breaking technique and energy release
present, many large or medium-sized open mines have been
by blasting. In engineering practices, the technology and
transformed from open slope mining to deep pit mining. With
parameters need to be designed according to the specific
the increase of mining depth, the slope becomes higher and
mine conditions.
steeper. Increase of slope angle can reduce the volume of exca-
vated rock and lower the production cost. For example, for With the coordinated treatment technique system of “strong
an open mine with annual capacity of 10 million tons, if the stress relief, strong support and strengthened monitoring”,
slope angle is increased by 1◦ , the volume of excavated rock effective treatment of coal bumps was realized. The specific
is reduced by 10 million tons. The economic benefit can reach technical contents include:
hundred million RMB. However, we have to optimize slope
(1) According to the results of “coupled multi-factor eval-
design and comprehensively monitor and control slope stabil-
uation technique for dynamic disasters”, for different
ity, so as to maximize the slope angle, reduce the excavated
rock zones, various techniques were adopted to effec-
rock volume, lower the cost and increase efficiency, under the
tively transfer of bump stresses, including stress relief
premise of safe mining.
technique in deep large-diameter boreholes in coal seam,
Research staffs from Beijing University of Science and
pre-fracturing of hard roof, floor breaking technique and
Technology, for the first time, adopted high-precision numer-
energy release by blasting
ical analysis and three-dimensional limit equilibrium analysis
(2) According to the idea of “protection of burst-prone rock
based on GIS for slope stability analysis and design optimiza-
masses by highly intensive support”, intensive chained
tion. The average slope angle was increased by 3◦ –4◦ . At the
support system with strong support function was designed
same time, GPS, the Total Station and other devices were used.
to resist mine tremors and rockburst.
Together with the network theory, a displacement monitoring
(3) The idea of real-time simultaneous monitoring and early
network was established for dynamic analyses and prediction
warning of “vibration field – dynamic stress field” was
on slope stability. It provided the basis for taking necessary
proposed. The relevant monitoring and early warning
measures in time and ensuring slope stability. At the same
system was developed.
time, a series of technical measures were formulated to ensure
slope stability at great depth and strengthen mining safety. The The key to dynamic disaster monitoring and early warn-
measures included: According to investigation and prediction ing is “time, location and intensity”. Therefore, real-time
of the potential slope failure mode, the corresponding control online monitoring, high-precision microseismic monitoring
measures for slope stability were identified; For seismicity- and dynamic stress monitoring were adopted to realize
prone zones, earthquake disaster mitigation techniques for the requirements on time, location and intensity, respec-
steep high slopes were developed; For slopes adjacent to tively. A high-precision microseismic monitoring and a real-
rivers, monitoring and treatment measures for steep high slope time online monitoring and early warning system for coal
failure induced by seepage were identified. bump (dynamic stress) have been developed and successfully
applied. Together with the monitoring instruments and tech-
4.4 Advances in monitoring, early warning and treatment of niques, a real-time monitoring and early warning system for
mine dynamic disasters dynamic disasters was established.
With the rapid increase of mining depth, more severe dynamic
disasters (mine tremor, coal bump, coal and gas outburst,
rockburst, water inrush, etc.) occur in mines. Monitoring and 5 NEW ANCHORAGE TECHNIQUES IN ROCK
treatment of dynamic disasters have become a technical chal- ENGINEERING
lenge in mining engineering. Important progress in theories,
techniques and equipments for monitoring, early warning and The main achievements in comprehensive rock anchorage
treatment dynamic has been made by research staffs in Beijing techniques made by the Chinese rock engineering field,
University of Technology and Science. represented by Central Research Institute of Building and
1) The “coupled multi-factor evaluation technique for Construction Co. Ltd., include.
dynamic disasters” was proposed to predict the bump-
prone area and the degree of bump.
5.1 Development of reusable high-pressure grouted rock
For prediction of the bump-prone area, an interaction
bolts and jet grouted expansion rock bolts
relationship between the spatial structural movement of
the overlying rock strata and the bearing stress was pro- In view of the need in rock bolts with high bearing capacity
posed. Before mining at the working face, the dynamic in complex strata, reusable high-pressure grouted rock bolts
formation process of the spatial structure of the overlying and jet grouted expansion rock bolts were developed.
rock strata during mining was predicted. In combination The key technique of reusable high-pressure grouted rock
with the structural distribution, rock stratum composition bolts is that mortar grouting for the cylindrical anchor sec-
and mining boundary conditions, the “coupled multi-factor tion is performed by one or multiple high-pressure fracture
evaluation technique for dynamic disasters” was proposed. grouting with a unique casing and grouting gun.
The purpose was to evaluate the bump-prone area and the For the jet grouted expansion rock bolts, high-pressure jet
degree of bump so as to provide the approximate range for is employed to expand the anchor hole in the anchor section
dynamic disaster treatment. by hydraulic cutting and cement mortar is then injected for
2) A set of techniques of “stress transfer” and coordinated replacement. A cylindrical expanded head is formed, with
treatment technique system of “strong stress relief, strong diameter being 0.7-0.8 m and the maximum of 1.2 m. The
support and strengthened monitoring” were proposed. bearing capacity of rock bolt can be greatly improved.
67

5.2 Development of load-dispersive anchorage (single-hole blasting operation can be safe, reliable, environment-friendly
composite anchorage) system and economical. The significant difference between precision
blasting and conventional control blasting lies in that by using
The anchorage system consists of a number of rock bolts at a
quantitative blasting design theory, method and experimental
certain spacing in one borehole. Each rock bolt has indepen-
methods, the blasting parameters can be optimized so as to
dent free section and anchored section and bears the same load.
precisely control the blasting effects and the adverse effects.
The bond stress evenly distributed along the anchored sec-
Precision blasting focuses on the mechanical properties of the
tion of the load-dispersive rock bolt and the bearing capacity
blasting objects, blasting conditions and engineering require-
increases linearly with the anchored length.
ments. It relies on good blasting devices and advanced and
The load-dispersive anchorage system composed of
reliable ignition technique, complemented by careful oper-
unbonded steel strands bent around the U-shaped rock bolt
ation and strict management so that the precise control of
element can greatly improve the durability of the rock bolts as
the entire blasting process can be realized. The core idea is
the grouting mortar is unlikely to crack under compression, in
“quantitative design, precise construction, real-time monitor-
addition to the role of double-layer anti-corrosion.
ing, scientific management”. Precision blasting is the third
milestone of blasting technique in China.
5.3 Establishment of evaluation model for long-term During construction of the abutment trench of Xiluodu
performance and safety of rock bolts Hydropower Station, the precision blasting concept was
Research staffs from Central Research Institute of Building implemented. The average over or under excavation rate, the
and Construction Co. Ltd. proposed a safety control index, evenness and the blast hole utilization rate along the normal
covering the variation of lock-off load of rock bolt (initial pre- direction of the working face were 97.2%, 98.8% and 99.8%,
stress), the reduction rate of the existing bearing capacity of respectively. The average blasting affected depth was within
rock bolt, the deformation rate of anchored geomass and struc- 1.0 m, measured by the borehole acoustic method. All the
tures and the corrosion degree of rock bolt. A safety evaluation results indicated good quality of blasting.
model was established, which consisted of a critical technical
indicator covering identification of risk sources, long-term
7 CONCLUSIONS
performance testing, monitoring subjects and methods and
safety evaluation, and the treatment measures for insufficient
From the above sections, it can be seen that China ranks
anchorage.
no. 1 in the world in term of rock engineering, with the
large amount of projects, complex geological conditions and
5.4 Development of standard system for anchorage unprecedented technical difficulties. In recent years, a series
techniques in China of innovative achievements have been made in rock engineer-
A relatively complete standard system for anchorage tech- ing and technology. Rich experiences have been accumulated
niques in geotechnical engineering has been established in in investigation, design, construction and monitoring. Of
China. National standard “Specifications for bolt- shotcrete course, lessons were also learned. They are precious treasures
support” (GB50086-2001) and the standard of China Asso- in rock engineering field. We are willing to share the experi-
ciation for Engineering Construction Standardization “Tech- ences with researchers and engineers in rock mechanics and
nical specifications for ground anchors” (CECS22:2005) and rock engineering over the world. Let’s work together towards
other technical specifications have clearly defined the design, the development of rock mechanics and engineering and a new
material, anti-corrosion requirements, construction, testing, era in rock engineering.
monitoring and inspection of ground anchors.
ACKNOWLEDGEMENTS
6 PRECISION BLASTING TECHIQUE IN ROCK
ENGINEERING The author would like to sincerely thank Professors H.C. Dai,
S.W. Song, S.C. Li, M.C. He, L.K. Cheng,A.Q. Wu, F.X. Jiang,
Precision blasting is realized by precise control of energy C.A. Tang, C. He, M.F. Cai and H.P. Kang for providing
release and explosion process of explosives by quantitative relevant data. The author also would like to thank Professor
blasting design and careful blasting implementation. The Z.L. Fang for data collation and summarization.
adverse effects of blasting can be precisely controlled and the
68

Keynote lectures

New rock mechanics developments in China
Xia-Ting Feng
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics,
Chinese Academy of Sciences, Wuhan, China
ABSTRACT: This paper reviews new rock mechanics developments in China in the last ten years. A series of devices for
laboratory tests, in situ tests and monitoring have been developed. Also, strength criteria based on energy and for hard rocks,
progressive and anisotropic damage models, coupled THM models, new creep models, the vector sum method, reinforcement
theory, and a multi-scale and non-linear thermodynamic theory were established. There have been major developments in
studying zonal disintegration, rock dynamics, fractals in rock mechanics, rock fracturing process simulation methods, and
intelligent rock mechanics methodology. The 3-D limit equilibrium methods, strength reduction method, FEM, BEM, NMM,
Meshless method, and DDA were updated. There have been considerable efforts on mechanism understanding through lab
and real time field monitoring, prediction and the mitigation of rockbursts. The need for rock engineering projects has largely
promoted the development of rock mechanics; and rock mechanics has provided a practical tool to support complicated rock
engineering problems. Further development of rock mechanics in China is anticipated and more applications of rock mechanics
to rock engineering are expected.
Subject: Modelling and numerical methods
Keywords: Lab tests, field measurements, back analysis, neural network, numerical modelling, dynamics
1 INTRODUCTION This paper just provides a brief review of the new devel-
opments in rock mechanics in China over the latest ten years
Rock Mechanics is a discipline having strong practical appli- and covers lab and field tests, mechanical models and crite-
cations. The construction of large-scale rock engineering ria, analysis methods for rock samples, slopes, tunnels, and
projects in China, including high/steep slopes, large cavern cavern groups, rock dynamics and zonal disintegration of a
groups, mining stopes, tunnels, oil wells, and foundations, deep rock mass, and applications of rock mechanics to rock
has promoted the development of rock mechanics in China. engineering. In the years ahead, further development of rock
Oil and gas underground storage, nuclear waste disposal, mechanics in China is anticipated and more applications of
CO2 sequestration, and geothermal development have pro- rock mechanics in rock engineering are expected.
moted the development of coupling THMC processes in rocks.
Sun & Wang (2000) have summarised the development and
the state-of-the-art of rock mechanics in China before 2000. 2 LABORATORY & FIELD TESTING TECHNIQUES
The ‘century achievement’ in rock mechanics in China has
been summarised in the book published by Hohai University 2.1 Laboratory testing techniques
Press (Wang, 2004).
In memory of Tan Tjong Kie, who initiated the Interna- In order to enhance understanding of the rock fracturing
tional Society for Rock Mechanics (ISRM) China National process under complicated environmental conditions, test-
Group in 1979, was the first President of the Chinese Soci- ing systems were developed using a medical or industrial
ety for Rock Mechanics and Engineering (CSRME) and was CT scanner with a loading system and a microscope with
the ISRM Vice-President for Asia during 1983–1987, the Tan CCD cameras. Ge (2004) developed firstly a CT real-time
Tjong Kie Lecture series was initiated in 2008. Three lectures testing technique for triaxial testing and study of the meso-
have been given by: damage evolution law. Through CT scanning, clear CT images
can be obtained which include the micro-hole (microcrack)
• Sun Jun (2007), ISRM Vice-President at Large 1995–1999
compressed ⇒ the microcrack growth ⇒ the microcrack
and President of CSRME 1994–1998; bifurcation ⇒ the microcrack development ⇒ crack frac-
• Wang Sijing (2009), President of CSRME 1999–2003; and
ture ⇒ rock sample destroyed ⇒ unloading. This technique
• Qian Qihu (2009), ISRM Vice-President at Large 2003–
was further developed by Xie Heping’s group using an indus-
2007 and President of CSRME 2003–present. trial CT scanner. Chen Houqun led a group in developing a
The 4th Tan Tjong Kie Lecture will be given by Ge Xiuren portable real-time dynamic loading apparatus with the sup-
this year on the measurement of in situ stress and its estima- port of the medical CT (Tian et al., 2010). The apparatus is
tion. Following the development of Chinese Journals, such as available for performing tests such as dynamic tension, com-
the Chinese Journal of Rock Mechanics, the Chinese Jour- pression test with impact, triangular and sine wave with the
nal of Underground Space and Engineering, Rock and Soil greatest load output being 100 kN, and frequency of 5 Hz.
Mechanics, a new English language journal, the Journal of Under dynamic loading, the whole process of crack propaga-
Rock Mechanics and Geotechnical Engineering, was launched tion and coalescence, CT images are obtained. It is laying the
in 2009. foundation for dynamic CT testing of brittle materials.
71

A loading and microscopic observation apparatus for deformation cable (He et al., 2010). It has been installed in 10
the meso-mechanical experiments on rock was developed areas, and 158 monitoring points, in fields such as slopes at
by Ge et al. (2000). The testing devices were further open-pits, mountains, gas pipelines and active faults.
developed to investigate the coupling process of chemical
flow/permeability–stress by using a microscope, two CCD
cameras, and a servo-control system (Feng & Ding, 2007). 3 MECHANICAL MODELS AND FAILURE CRITERIA
The following testing systems have also been developed or
updated for different purposes. Much effort has been put into developing strength criteria for
rock and rock masses, as follows.
• Remote sensing techniques have been used to test the frac-
turing process of rock samples, and ‘remote sensing rock • A Unified Strength criterion for rock material (Yu et al.,
mechanics’ was developed (Wu et al., 2004). With the 2002).
techniques, precursors for rock fracturing and failure were • A criterion for strength and structural failure of rocks based
found (Wu et al., 2006a, b). on energy dissipation and energy release principles (Xie
• A testing machine for coupled seepage and triaxial stress et al., 2009).
measurements in rocks (Liu, 2007). • A triple shear energy yield criterion (Gao et al., 2007).
• The RMT testing machine developed by Ge has become one • A strain criterion for ductile shear failure based on the
of the main rock mechanics testing systems in institutes maximum principal shear strain (Gao et al., 2007).
and universities in China. Using this machine, a series • A new three shear strength criterion for hard rock (Chen &
of fatigue, uniaxial/triaxial compressive and shearing tests Feng, 2007).
have been conducted (Ge, 2008). • An anisotropic strength criterion for jointed rock masses
• A testing device for rock subjected to coupled static and (Chen et al., 2008).
dynamic loads (Li et al., 2008). • A generalized polyaxial strain energy (GPSE) strength
• A 20 MN servo-controlled rock triaxial testing system with criterion (Huang et al., 2008).
high temperature/ pressure (Zhao et al., 2008). • A true-triaxial strength criterion for rock (You, 2009).
• A Storage-variable transient pulse testing device (Li et al., • Mechanical characteristics of the exponential strength cri-
2008). terion under conventional triaxial stresses (You, 2010 a, b).
• A testing apparatus for simultaneously measuring adsorp- Some typical mechanical models for intact rock and rock
tion, deformation and permeability of coal (Fang et al., masses have been developed. Examples are given below.
2009).
• A coupled shear-seepage test system for rock joints (Xia • Analysis of the strain softening size effect for rock speci-
et al., 2008). mens based on shear strain gradient plasticity theory (Pan
• A rockburst simulation experimental system for loading in et al., 2002),
three directions independently and suddenly unloading on • A localised progressive damage model for fractured rock-
one side of the sample.A rockburst criterion was established like materials (Zhang et al., 2006),
(He et al., 2010). • A coupled model for anisotropic damage and permeability
variation (Zhou et al., 2007),
• A new Cosserat-like constitutive model for bedded salt
2.2 Field test and monitoring technique rocks (Li et al., 2009), and
In situ testing techniques have been significantly developed in • A new model to describe the unloading behaviour of rock
China. Ge proposed a new method for stress measurement, i.e., (Li et al., 2010).
the Borehole Wall Stress Relief Method, BWSRM. This has
been successfully used, in particular in the deep tunnels of the 4 ANALYSIS METHODS AND THEORIES
Jinping hydropower station having an overburden of 2430 m.
A full temperature compensation technique was invented to 4.1 Methods for slope stability analysis
eliminate measuring errors caused by temperature changes
in overcoring stress measurement devices which use strain Ge (2008) developed a vector sum method for slope stability
gauges as the sensing elements. An interpretation method analysis, believing that associated with a given failure surface
for rock stress was developed using the measurement strain there exists a direction, referred to as the critical slip direction
caused by stress relief considering the practical behaviour of s, the factor of safety F can be calculated as
a rock mass, such as non-linearity, anisotropy, discontinuity
and non-homogeneity (Cai et al., 2009). These techniques
make the measuring reliability and accuracy of stress relief by
overcoring technique much better. New hydraulic fracturing
equipment with high pressure capacity was developed which
surmounted the shortcoming of the traditional hydraulic frac-
Here, S denotes the failure surface; τ is the resultant shear
turing technique which cannot be used at high depth if there is
stress vector acting on dS and [τ] is the vector of shear stress
insufficient hydraulic pressure capacity. With the newly devel-
strength in the same direction as τ.
oped equipment, stress measurement with hydraulic fracturing
The following methods have also been developed for
technique has been successfully completed at up to 2800 m
analysis of slope stability.
depth (Cai et al., 2004).
A new remote real-time monitoring warning system for • A 3-D slope stability analysis method using the upper bound
landslide disaster (SPRM) has been established based on theorem (Chen et al., 2001a and b).
the mechanical principle of interaction between the land- • A generalised solution for tetrahedral rock wedge stability
slide body, landslide bed and a constant-resistance large analysis (Chen, 2004).
72

• The rigorous and quasi-rigorous limit equilibrium solutions thermal-mechanical coupling effects of sandstones show that,
to 3-D slope stability, which satisfy six or five equilibrium with the temperature increase, thermal cracking exhibits
conditions respectively, are derived (Zhu & Qian, 2007). a fractal property. A segmented strength model for sand-
• The finite element strength reduction method has been stone considering the coupled thermal-mechanical process is
developed—such as, the definition and computation established (Zuo et al., 2010).
method for double safety factor (double reduction factor),
the structure instability and failure criteria in computation, 4.3 Study on rock reinforcement and thermodynamic theory
and the numerical methods based on strength reduction for
3-D slope analysis (Zheng, 2007). Yang et al. (2007) proposed a reinforcement theory that the
• A 3-D rigorous method that reduces the 3-D analysis of reinforcement force required by a geotechnical structure is just
slope stability to an algebraic eigenvalue problem and thus the residual unbalanced force obtained through elasto-plastic
overcomes numerical problems inherent in the existing 3-D FEM analysis. A principle of minimum plastic complemen-
methods (Zheng, 2009). tary energy is established, which requires that a structure
• A Cauchy problem for 3-D critical slip surfaces based on the always trends to the state of minimum reinforcement force
stress field in slopes at the limit equilibrium state (Zheng and maximum self-bearing force for given loads. Because
et al. 2011). Once the Cauchy problem is solved, the 3-D every load state characterised by an overloading factor K cor-
critical slip surface of the slope can be determined without responds to a minimum plastic complementary energy Emin ,
specifying the shape and the location. the K − Emin curve can be used to evaluate the global sta-
• The meshless method and shortest path algorithm for rock bility of an arch dam. The theory has been widely used in the
slope stability analysis (Zhuang et al., 2008). reinforcement design of the highest arch dams and rock slopes
• Four criteria for stability analysis and assessment of high in China, e.g. Xiaowan arch dam (292 m high), Jinping I arch
rock slopes (Li & Qian, 2010). dam (305 m high) and Xiluodu arch dam (278 m high), and was
adopted in the latest version of the “Handbook of Arch Dam
Design” released by the Chinese Hydroelectrical Authority.
4.2 Study of fractals in rock mechanics The multi-scale and non-linear thermodynamic theory of
solids by J.R. Rice, has been substantially developed by Yang
The geometry and configuration of the discontinuous geo- et al. (2005). The evolution laws of second-order fabric tensors
metry and irregular sub-structures, such as joints, pores and were derived within the extended normality structure. Based
faults, of rock masses are investigated and characterised by on Eshelby’s equivalent inclusion method, the solution of a 3-D
means of 3D laser scanning, the CT imaging process, statis- penny-shaped inhomogeneity surrounded by an unbounded
tics and fractal theories. The morphology and the distribution isotropic matrix in a uniform stress field was developed as a
laws of joints and pores are extracted and analysed (Zhou et al., much more versatile defect model. This model has been used
2003, 2004; Ju et al., 2007, 2008). The novel 3-D reconstruc- for tunnels and dams, back analysis (Xiang et al., Computers &
tion approaches are developed based on the investigation, plus Structures, 80, 1429–1440, 2002), and dynamic damage
fractals and statistical algorithms, to reconstruct those sub- (Zhang et al., Int. J. Eng. Sci., 41, 917–929, 2003). By extend-
structures that govern the apparent physical responses of rocks ing the strain gradient theory and introducing an intrinsic
(Xie et al., 2001; Zhou et al., 2003; Ju et al., 2007, 2008). The material length scale into the constitutive law, the damage
mechanical performance of rock masses subjected to static and model is extended to model damage localisation analysis and
impact loads, as well as the influences of sub-structures on the applied to arch dam analysis by Zhou et al. (2002).
properties, are discussed based on reconstructed model simu-
lations and experimental investigations (Xie et al., 2008; Zhou
4.4 Study of intelligent rock mechanics methodology
et al., 2003; Ju et al., 2007, 2008, 2009, 2010). The intrinsic
micro- or meso-scale processes that damage and invalidate the The intelligent rock mechanics methodology has been fur-
bearing capability of rocks are established (Xie et al., 2000, ther developed. The fracturing and rockburst processes in the
2008; Ju et al., 2007, 2008, 2009, 2010). This study opens a surrounding rock in deep TBM and D&B tunnels having an
new window for understanding and quantifying the effects of overburden of 1900–2500 m was understood through a series
complex sub-structures on rock performance. of in situ monitorings of deformation, wave velocity, crack-
Theoretical and experimental studies indicate that energy ing seen by digital borehole camera, acoustic emission, and
plays a key role in causing rock deformation and failure. The micro-seismicity conducted at the Jinping hydropower sta-
relations among energy dissipation, energy release, strength tion (Li et al., 2010; Chen et al., 2011). New indices, such
and catastrophic breakdown of rock systems during defor- as RVI, failure approach index (FAI) and local energy release
mation are discussed. It is shown that the energy dissipated rate (LERR), were proposed for prediction of rockbursts and
induces damage and irreversible deformations and leads to rock failure in the Jinping headrace tunnels.
reduction in rock strength. The releasable strain energy results A hybrid evolutionary algorithm has been established to
in catastrophic failure of the entire rock system. The criteria recognise the structure of non-linear mechanical models and
for strength reduction and catastrophic breakdown of rocks parameters (Feng et al., 2006). An intelligent back analy-
have been established on the basis of the energy principles. sis method was proposed to recognise mechanical rock mass
The experiments on marbles and layered limestones subjected parameters from the monitored displacement and excavation
to uniaxial and biaxial loads illustrate that the proposed energy damaged zone (EDZ).
criteria are in good agreement with the experimental results An elasto-plastic cellular automaton, EPCA2D and EPCA3D ,
(Xie et al., 2009; Ju et al., 2007, 2010). utilising deformation and strength parameters evolving with
The physical properties of rocks under high tempera- damage, a failure approach index, FAI, and the local energy
ture circumstances is one of the primary concerns in deep release rate, LERR, have been developed for numerical anal-
mine excavation. A series of laboratory tests on inves- ysis of the rock fracturing process and has been used for
tigating the mechanisms at the meso-scale level and for the simulation of the coupled THM process in the EDZ, the
73

thermal effect on a rock shaft at the Swedish Äspö Hard Rock • The boundary meshfree method based on the modified
Laboratory in the DECOVALEX International Co-operative MLS and boundary integral equation to analyse a beam
Project (Pan et al., 2009). on a non-linear foundation (Li et al., 2008),
An intelligent rock design methodology has been developed • The distinct/discrete element method to analyse a jointed
and applied to stability analysis and design optimisation of the rock mass, slope stability and influence of the structural
excavation and support systems for underground powerhouses surface in geotechnical blasting engineering (Zhou et al.,
at Jinping II, Laxiwa and Shuibuya power stations, the slopes 2006, Han et al., 2010),
at Longtan and Nuozhadu hydro-power stations, and the head- • Numerical manifold method and DDA to improve the
race tunnels at the Jinping II power station (Feng & Hudson, quality of the stiffness matrix (Jiang et al., 2002; Peng &
2011). The results were adopted for the design of the projects Ge, 2004; Lin et al., 2006), simplify the integration scheme
and validated in practice. (Lin, 2005), and obtain high order manifold element
(Zhang & Peng, 2000),
• The augmented Lagrange multiplier method to reduce the
4.5 Numerical analysis methods
sensitivity of the spring stiffness parameters (Cai et al.,
A rock failure process analysis code, RFPA2D and RFPA3D 2004),
has been further developed by Tang’s group (Tang & Hudson, • The complementary discontinuous deformation method to
2010). The codes have been used to simulate rock failure in avoid the introduction of the virtual spring and the ‘open-
uniaxial/indirect tension, uniaxial/triaxial compression, and close’ iteration process (Zheng & Jiang, 2009),
failure induced by thermal stress, cutting, time dependency, • Numerical estimation of the REV and permeability ten-
dynamic loading, and longwall coal mining, plus around tun- sor for fractured rock masses by the composite element
nels in jointed rock, gas outbursts in coal mines, and the method (Chen et al., 2008), and
coalescence of fractures. • Parallel FEM based on Jacobi condition and domain decom-
A coupled THM model for analysis of multiphase flow, position (Zhang et al., 2004).
thermal transport and stress/deformation in fractured porous
media was developed and validated by the CEA Mock-up There are developments in methods for stability analysis of
test and the FEBEX in situ experiment (Chen et al., 2009). underground engineering projects, for example as below.
Stress/strain-dependent hydraulic conductivity tensor models
• The safety factor analysis method for stability estimation
were proposed to describe the effects of post-peak mechan-
ical behaviour, e.g. shear dilatancy and strain softening of of tunnels (Zheng et al., 2010).
• Monitoring, mechanism understanding and prediction of
fractures, on the macroscopic permeability of fractured rocks
(Zhou et al., 2008). A parabolic variational inequality method rockburst and mining seismicity (Cai et al. 2005; Tang et al.,
of Signorini’s condition, combined with the sub-structure 2010; Wu et al., 2010).
• Analysis of rock structure stability in coal mines (Pan & Li,
technique, was proposed to solve non-steady seepage flow
problems with complex drainage systems (Chen et al., 2010). 2005).
• An automatic analysis platform for tunnels based on expert
The FEM, DEM, DDA and numerical manifold methods
have been developed further, as listed below. experiences, numerical test results, and the neural network
method (Li et al., 2006).
• Viscous boundary of DDA for modelling stress wave • A study on sidewall displacement prediction and stability
propagation in jointed rock (Jiao, et al., 2007), evaluations for large underground power station caverns
• A general composite element concept for a fully-grouted (Zhu et al., 2010; Sun & Li, 2010).
rock bolted element model in FEM analysis (Chen et al., • Study on mechanical behaviors of mining rock mass and
2003), its related engineering technological innovation progress
• A 3-D boundary element method, dual boundary control (Miao, 2010).
technique of material parameter identification, and the • Deformation and failure mechanism analysis for the under-
infinite element for ‘infinite problems’ in geomechanics ground powerhouse of Jinping I hydropower project using
(Liu et al., 2008), artificial intelligent back analysis method, extended finite
• Determining relative movements on boundaries, analysing element method, and unloading method (Li et al., 2010;
fluid–structure interaction and simulating the static and Zhou & Qian, 2011; Huang et al., 2011).
dynamic responses of discontinuous rock tunnel excava- • Elasto-plastic analysis and analytical stress solutions for a
tion based on discontinuous boundary element, DDA and circular tunnel (Lu et al., 2010; Lu et al., 2011).
deformation block distinct element method (Zhang et al.,
2001; Jin et al., 2001), There have also been considerable efforts in rock mechan-
• Viscoelastic BEM, natural element method coupling with ics studies for environmental and energy issues in China, as
FEM for stability analysis of the surrounding rocks of indicated below.
rock-bolted and shotcreted tunnels (Xu, 2008; Pang, 2004),
• A meshless method to analyse the joint rock structures • The geological survey study via deep boreholes and rock
(Zhang et al., 2001; Cai & Zhu, 2003), mechanics tests and THMC coupling behaviour for the
• Combined MLS with hybrid displacement variational simulation of granite for the Baishan site, a potential nuclear
principle, the hybrid boundary node method and dual waste disposal site (Wang, 2010).
reciprocity hybrid boundary node method to the analysis • The related rock mechanics studies for CO2 underground
of geomechanics problems (Zhang et al., 2002; Miao et al., storage to investigate the interaction mechanism of CO2
2008), and coal, sandstone, etc.
• Element-Free Galerkin method for joint rock and slope • The behaviour of layered rock salt caverns within or without
excavation simulation (Zhang & Lu, 2000; Zhang & Peng, gas or oil storage, which are used as a guide for the design
2000; Pang et al., 2000; Lu et al., 2008), of understand oil/gas storage caverns.
74

5 ROCK DYNAMICS
The formation mechanism of the structural hierarchy of frac-

turing in a rock mass was discovered. Based on the relations
between the deformation and fracture temporal scales and the
structural size scales of a rock mass at different structural
levels, the principle of equal density of work, the principle
of equal energy flow and the principle of equal power of the
fracture of rock mass are derived. These principles can serve
as a tool for the study of the deformation and fracture of a
rock mass at different structural levels (Qian, 2009).
The physical mechanism of strength–strain rate sensitiv-
ity in different strain rate regions and the transition between
them was discovered. A unified model of the co-existence
and competition between the thermo-activational mecha-
nism and phonon viscosity mechanism is proposed, the
relation between dynamic strength, strain rate, plastic defor-
mation and stress state is determined, and the structural
aspects of the mechanism o deformation and fracture are
clarified.
According to the measured wave forms in a comminuted Figure 1. An experiment result of circular failure of a rock sample
region under penetration and explosion in geomaterials and (He et al., 2008).
concrete, by using ‘short wave’ and ‘weak wave’ theories, it
is clarified that the internal friction stress state change deter- • The rheological rock characteristics at high temperature
mines the deformation properties, and the generalised model (Liu & Chen, 2003; Deng & Wang, 2004; Xi et al., 2008)
with internal friction and compression is proposed. and under water conditions.
The resistance acting upon targets is obtained, the dis- • A unified rheological model theory which could describe
tribution of pressure from the elasto-plastic state to the the most complicated rheological characteristics based on
hydrodynamic state is obtained. Based on the criterion for conventional component models (Sun, 2007; Xia et al.,
propagation of a crack, the similarity laws for penetration and 2008).
perforation are obtained (Qian, 2009). • A new seven-component non-linear visco-elasto-plastic
Based on analysis of the measured dynamic constitutive rheological model of rock by connecting the non-linear
curves, and combining a statistical damage model and a vis- viscoplastic body (NVPB) model and five-component
coelastic model, a time-dependent damage model of rock visco-elastic model in series (Xu et al., 2006).
under dynamic loading is established (Shan et al., 2003). • Intelligent method for rheological model coupling recogni-
Analytical simulation of the dynamic compressive strength tion and their parameters (Chen et al., 2006).
of a granite using the sliding crack model was also developed • A robust statistical estimation theory for parameter iden-
(Li et al., 2001). Bifurcation of a collinear crack system under tification by using a double quadratic iteratively weighted
dynamic compression was also studied (Zhou et al., 2010). least squares algorithm (Han & Wang, 2008).
• The effect of water content on shear creep behaviour of a
weak structural plane of sandstone (Li et al., 2008),
6 RHEOLOGICAL ROCK MECHANICS • Unified approach for modelling elastic–plastic and visco-
plastic behaviour plus induced damage over a long time
The study on rheological rock mechanics has achieved great scale in quasi-brittle rocks (Zhou et al., 2008).
development since it was initiated by Tan Tjong Kie. Due • The long-term strength of rocks for evaluating long-term
to the needs of many rock engineering projects, various lab- stability in rock engineering (Cui & Fu, 2006; Li et al.,
oratory and in situ rheological tests were conducted, e.g., 2010; Wang et al., 2010; Li et al., 2010).
creep and relaxation tests under uniaxial compression, split
tension, shear, tension-shear, triaxial compression of grading
loading–unloading, and three point bending, etc. It is inter-
7 ZONAL DISINTEGRATION OF DEEP ROCK MASS
esting to investigate the rheological behaviour of hard and
soft rocks such as marble, sandstone, granite, basalt, tripoli
The temporal process of zonal disintegration is studied by
(siliceous material), tuff, coal, salt rock, shale, mudstone,
physical experimentation (Fig. 1, He et al., 2008), field test-
green schist, diabase, gypsum breccias, rigid joints, and soft
ing, and on the basis of irreversible thermal dynamics, and the
rock strata. Based on these experiments and theoretical analy-
theory of continuous phase transitions and elasticity–plasticity
ses, the rheological models and their parameters are identified
theories. The spatial and temporal evolution equations of
which enables interpretation of the tests results, via non-linear
zonal disintegration are derived and the analytical solution
rheological behaviour, rheological damage and the fracture
is obtained (Qi et al., 2009).
mechanism of specific rocks (Sun, 2007). Some typical results
The unified description of the temporal evolution of zonal
are listed as follows.
disintegration in different surrounding rock masses is inves-
• A formula developed by introducing the principle of dam- tigated using the theory of continuous phase transitions. This
age mechanics to calculate the rheological cracking of points out that the evolution equation of order parameter
jointed rocks is derived under the action of both compres- K = r/d (the ratio of the mean length, r, of cracks to the
sion and shear (Xiao et al., 2000). mean distance, d, between cracks may describe the zonal
75

disintegration phenomenon, deformation wave and ‘drastic’ Qihu, Wang Sijing, Ge Xiurun, and Zheng Yinren provided
fracture phenomenon. helpful advice. Professors Cai Meifeng, He Manchao, Wang
Mingyang, Yang Qiang, Xu Dongjun, Li Haibo, Li Xiaochun,
Zheng Hong, Ju Yang, Huang Lixing, Zhou Hui, Pan Pengzhi,
Yan Fei, Qi Chenzhi, Dr. Zhao Ying, Dr. Liuyaqun, Dr. Jing
Feng, Dr. Xia Qiang, and Mr. Zhu Yong supplied infor-
Based on the laws of the spatial and temporal evolution mation and references. Professor John A. Hudson offered
of zonal disintegration, some supporting methods are sug- some suggestions and polished the English text. The work
gested for the strengthening of the tunnels in the case of the received financial support from the China National Basic
occurrence of such zonal disintegration. Research Program under Grant no. 2010CB7320006. They
The theory of continuous phase transition is employed to are all acknowledged.
study the incremental strain sign change effect, and the gov-
erning equation and its solution are derived. The numerical
results obtained demonstrate that this approach provides an REFERENCES
adequate description of the sign-change effect for strain incre-
ments in rock samples under compression (Wang et al., 2010). Cai, M.F. 2004. Techniques for in-situ stress measurement at great
depth. Journal of University of Science and Technology Beijing
11(6): 486–488.
8 CONCLUSIONS AND DISCUSSION Cai, M.F. Ji, H.G. & Wang, J.A. 2005. Study of the time-space-
strength relation for mining seismicity at Laohutai coal mine
There have been significant developments in rock mechanics and its prediction. International Journal of Rock Mechanics and
Mining Sciences 42(1): 145–151.
studies in China over the last ten years due to needs of large Cai, M.F. Peng, H. Ma, X. Jiang, J. 2009. Evolution of the in situ rock
rock engineering projects and through the support of adequate stain observed at Shandan monitoring station during the M8.0
funding. Several points relating to these developments can be earthquake in Wenchuan, China. International Journal of Rock
mentioned as follows. Mechanics and Mining Sciences 46(5): 952–955.
Cai, Y.C. & Zhu, H.H. 2003. Meshless method for numerical cal-
1) Investigation and the mechanistic understanding of the
culation of geotechnical engineering and its auto-arrangement of
fracturing process of rocks and rock masses have been discrete nodes. Rock and Soil Mechanics 24(1): 21–24.
enhanced through the development of new lab testing Chen, B.R. Feng, X.T. Zeng, X.H. Xiao, Y.X. Zhang, Z.T.
systems, field monitoring and observations by different Ming, H.J. & Feng, G.L. 2011. Real-time microseismic moni-
methods of the excavation process, plus numerical simu- toring and its characteristic analysis during TBM tunneling in
lation techniques. The CT microscope, nuclear magnetic deep-buried tunnel. Chinese Journal of Rock Mechanics and
resonance, digital borehole, and remote sensing tech- Engineering 30(2):275–283.
niques, etc., play important roles in developing new test- Chen, J.T. & Feng, X.T. 2007. Study on constitutive model for hard
ing systems. The post-peak and unloading behaviour in rock under high geostresses. Rock and Soil Mechanics 28(11):
rock failure and the unloading effect of excavation have 2271–2278.
Chen, S.H. Feng, X.M. & Isam, S. 2008. Numerical estimation of
been emphasised. The monitored displacement, micro-
REV and permeability tensor for fractured rock masses by com-
seismicity, cracking, wave velocity etc. have been studied to posite element method. International Journal for Numerical and
obtain comprehensive information reflecting the behaviour Analytical Methods in Geomechanics 32:1459–77.
of a rock mass under complicated conditions. Chen, S.H. Qiang, S. & Chen, S.F. 2003. Study on the three-
2) There has been much effort in investigating the effect of the dimensional composite element model of bolted rock masses.
environment, such as high stress, temperature, permeabil- Chinese Journal of Rock and Engineering 22(1): 1–8.
ity, chemical erosion and their coupling processes, on the Chen W.S. Zheng, H. Cheng, Y.M. Ge X.R. 2004. Detection of 3D
fracturing behaviour of rocks by developing new testing rock block contacts by penetration edges. Chinese Journal of Rock
systems, new models and codes. Mechanics and Engineering, 23(4): 565–571.
3) Analysis methods which consider inhomogeneous, non- Chen, X. Yang, Q. Qiu, K.B. & Feng, J.L. 2008. An anisotropic
strength criterion for jointed rock masses and its application in
linear, discontinuous, and anisotropic behaviour of rocks
wellbore stability analyses. International Journal for Numerical
and rock masses have been developed further. and Analytical Methods in Geomechanics 32:607–31.
4) Greater application of rock mechanics models, theories and Chen, Y.F. Zhou, C.B. & Jing, L.R. 2009. Modeling coupled THM
methods in rock engineering are emphasised. processes of geological porous media with multiphase flow: the-
5) More powerful methods, such as fully-coupled models, ory and validation against laboratory and field scale experiments.
internet-based methods, methods based memory, inte- Computers and Geotechnics 36(8): 1308–1329.
grated methods and virtual rock mechanics laboratories Chen, Y.F. Zhou, C.B. & Jing, L.R. 2010. Numerical modeling of
can be developed for new applications. coupled thermo-mechanical response of a rock pillar. Journal of
6) It is a good idea to establish deep underground rock Rock Mechanics and Geotechnical Engineering 2(3): 262–273.
mechanical laboratories—which can be used for in situ Chen, Z. 2004. A generalized solution for tetrahedral rock wedge
stability analysis. International Journal of Rock Mechanics and
tests for studying the fracturing process in the surround-
Mining Sciences 41(4): 613–628.
ing rocks as induced by excavation, plus time dependency, Chen, Z. Wang, X. Haberfield, C. Yin, J. & Wang, Y. 2001a. A three-
and control through reinforcement, and for nuclear waste dimensional slope stability analysis method using the upper bound
disposal and other projects. theorem : Part I: theory and methods. International Journal of Rock
Mechanics and Mining Sciences 38(3): 369–378.
Chen, Z. Wang, J. Wang, Y. Yin, J. & Haberfield, C. 2001b. A
ACKNOWLEDGEMENTS three-dimensional slope stability analysis method using the upper
bound theorem Part II: numerical approaches, applications and
The Chinese Society for Rock Mechanics and Engineering extensions. International Journal of Rock Mechanics and Mining
gave me the opportunity to write this paper. Professors Qian Sciences 38(3): 379–397.
76

Cui, X.H. & Fu, Z.L. 2006. Experimental study on rheology prop- Li, H.B. Zhao, J. Li, T.J. & Yuan, J.X. 2001. Analytical simulation
erties and long-term strength of rocks. Chinese Journal of Rock of the dynamic compressive strength of a granite using the sliding
Mechanics and Engineering 25(5):1021–1024. crack model. International Journal for Numerical and Analytical
Fang, Z.M., Li, X.C. & Bai, B. 2009. Study of method for simulta- Methods in Geomechanics 25:853–69.
neously measuring adsorption deformation permeability of coal. Li, J.L. Wang, L.H. Wang, X.X. Wang, R.H. Cheng, Z. & Dang, L.
Chinese Journal of Rock Mechanics and Engineering 28(9): 2010. Research on unloading nonlinear mechanical character-
1828–1833. istics of jointed rock masses. Journal of Rock Mechanics and
Feng, X.T., Chen, B.R., Yang, C.X., Zhou, H. & Ding, X.L. 2006. Geotechnical Engineering 2(4): 357–364.
Identification of visco-elastic models for rocks using genetic pro- Li, L.Q. Xu, W.Y. Wang, W. Guo,Y.Q. & Zhao, H.B. 2010. Estimation
gramming coupled with the modified particle swarm optimization of long-term strength for Xiangjiaba sandstone based on creep
algorithm. International Journal of Rock Mechanics and Mining tests. Engineering Mechanics 27(11): 127–136.
Sciences. 43(5):789–801 Li, N. Luo, J.Z. Chang, B. & Zhang, Z.Q. 2006. New method for
Feng, X.T. & Ding, W.X. 2007. Experimental study of limestone design and analysis of underground caverns. Chinese Journal of
micro-fracturing under a coupled stress, fluid flow and changing Rock Mechanics and Engineering 25(10):2155–2159.
chemical environment. International Journal of Rock Mechanics Li, N. & Qian, Q.H. 2010. Four criteria of stability analysis and
and Mining Sciences 44 (3): 437–448. assessment of high rock slope. Chinese Journal of Rock Mechan-
Feng, X.T. & Hudson, J.A. 2011. Rock engineering design. Taylor & ics and Engineering 29(9): 1754–1759.
Francis, London. Li, P. Liu, J. Zhu J.B. & He, H.J. 2008. Research on effects of
Gao, H. Zheng, Y.R. & Feng, X.T. 2007a. Study on energy yield water content on shear creep behavior of weak structural plane
criterion of geomaterials. Chinese Journal of Rock Mechanics and of sandstone. Rock and Soil Mechanics 29(7): 1865–1871.
Engineering 26(12): 2437–2443. Li, S.C. Li, S.C. Sui, B. & Zhu, W.S. 2008. The meshless method for
Gao, H. Zheng, Y.R. & Feng, X.T. 2007b. Deduction of failure analyzing seepage of rockmass. Rock and Soil Mechanics 29(1):
criterion for geomaterials based on maximum principal shear 256–260.
strain. Chinese Journal of Rock Mechanics and Engineering 26(3): Li, S.J. Feng, X.T. Zhang, C.S. Li, Z.H. Zhou, H. Hou, J. &
518–524. Chu, W.J. 2010. Testing on formation and evolution of TBM
Ge, X.R., Li, T.J., Zhang, M.Y., Liu, J.G. 2000. Loading apparatus for excavation damaged zone in deep-buried tunnel based on digital
the meso mechanical experiment rock. Rock and Soil Mechanics, panoramic borehole camera technique. Chinese Journal of Rock
21(3):289–293. Mechanics and Engineering 29(6):1106–1112.
Ge, X.R., Ren, J.X., Pu, Y.B., Ma, W. & Sun H. 2004. Macro-meso Li, X. Zhou, Z. Lok, T. Hong, L. & Yin, T. 2008. Innovative testing
testing of rock and soil damage mechanics. Beijing: Science Press. technique of rock subjected to coupled static and dynamic loads.
Ge, X.R. 2008. Deformation control law of rock fatigue failure, real- International Journal of Rock Mechanics and Mining Sciences
time X-ray CT scan of geotechnical testing, and new method of 45(5): 739–748.
stability analysis of slopes and dam foundations. Chinese Journal Li, X.C. Wang, Y. Wei, N. 2008. Research on measuring
of Geotechnical Engineering 30(1): 1–20. method of permeability by using storage-variable transient pulse
Han, Y.C. Liu X.Y. & Li S.H. 2010. Discrete element model- method. Chinese Journal of Rock Mechanics and Engineering
ing of brittle failure process of rock materials. Mechanics in 27(12):2482–2487.
Engineering.32(3):50–56. Li, Y.P. Yang, C.H. Daemen, JJK. Yin, X.Y. & Chen, F. 2009.
He, Y.N. Jiang, B.S. Han, L.J. Shao, P. Zhang, H.Q. 2008. Study A new Cosserat-like constitutive model for bedded salt rocks.
of Intermit tent Zonal Fr acturing of Surrounding Rock in Deep International Journal for Numerical and Analytical Methods in
Roadways. Journal of China University of Mining & Technology Geomechanics 33:1691–720.
37(3): 300–304. Li, Z.K. Zhou, Z. Xu, Q.J. Liao, C.G. Yin, H.L., Zhang, Z.Z. Cheng,
He, M.C. Miao, J.L. & Feng, J.L. 2010. Rock burst process of lime- L.J. Liu, Z.G. & Guo, K. 2010. Implementation and development
stone and its acoustic emission characteristics under true-triaxial of AI back analysis for the underground powerhouse of Jinping-1
unloading conditions International Journal of Rock Mechanics hydropower project. Journal of Hydroelectric Engineering. 29(3):
and Mining Sciences 47(2): 286–298. 177–183.
He, M.C. Wang,Y. & Tao, Z.G. 2010. A new early-warning prediction Liu, C.L. Zhang, L.Z. Zhao, Q.Q. & Zheng, Y.Z. 2008. Devel-
system for monitoring shear force of fault plane in an active fault. opment of three-dimensional boundary element numerical sys-
Journal of Rock Mechanics and Geotechnical Engineering 2(3): tem FSM.DDM with rock hole and anchor bolt line elements.
223–231. Chinese Journal of Rock Mechanics and Engineering 27(12):
Huang S.L. Feng, X.T. & Zhang, C.Q. 2008. A new generalized 2528–2534.
polyaxial strain energy strength criterion of brittle rock and Liu, G.T. Ye, Y.X. & Xu, Z.H. 2007. Development of testing machine
polyaxial test validation. Chinese Journal of Rock Mechanics and for coupled seepage and triaxial stress measurements in rocks.
Engineering 27(1): 124–134. Journal of Tsinghua University(Science and Technology) 47(3):
Huang R.Q. Huang, D. Duan, S.H. & Wu, Q. 2011. Geomechan- 323–326.
ics mechanism and characteristics of surrounding rock mass Lin, S.Z. 2005. Recursive formula for simplex integration. Journal
deformation failure in construction phase for underground of Yangtze River Scientific Research Institute 22(3): 32–34.
powerhouse of Jinping I hydropower station. Chinese Journal of Lin, S.Z. Qi, Y.F. & Su, H.D. 2006. Improved local cover func-
Rock Mechanics and Engineering, 30(1):223–235 tion of numerical manifold method and its application. Journal of
Jiang, Q.H. Young, M.R. & Feng, D.X. 2002. Three-dimension dis- Yangtze River Scientific Research Institute 23(6): 55–58.
continuous deformation analysis for forward modeling. Chinese Liu, J. & Li, Z.K. 2004. Choice of some control constants in DDA
Journal of Rock Mechanics and Engineering 21(12): 1765–1769. method. Journal of Chengdu University of Technology(Science &
Jiao, Y.Y. Zhang, X.L. Zhao, J. & Liu, Q.S. 2007. Viscous boundary Technology Edition) 31(5): 522–526.
of DDA for modeling stress wave propagation in jointed rock. Liu, Q.S. & Chen, G. 2003. Further verification of time—temperature
International Journal of Rock Mechanics and Mining Sciences equivalence for granite. Chinese Journal of Rock Mechanics and
44(7): 1070–1076. Engineering 22(10):1601–1606.
Jin, F., Jia, W.W. & Wang, G.L. 2001. Coupling model of distinct Lu, A. Xu, G. Sun, F. & Sun, W. 2010. Elasto-plastic analysis of
element-boundary element. Journal of Hydraulic Engineering 1: a circular tunnel including the effect of the axial in situ stress.
23–27. International Journal of Rock Mechanics and Mining Sciences
Ju, Y. Sudak, L. & Xie, H. 2007. Study on stress wave propagation in 47(1): 50–59.
fractured rocks with fractal joint surfaces. International Journal Lu, A. Zhang, L. & Zhang, N. 2011. Analytic stress solutions for
of Solids and Structures 44: 4256–4271 a circular pressure tunnel at pressure and great depth including
77

support delay. International Journal of Rock Mechanics and Wang, M.Y. Qi, C.Z. & Qian Q.H. 2010. Modeling of the incremental
Mining Sciences 48(3): 514–519. strain sign-change effect in rocks under compression. Journal of
Lu, B. Ding, X.L. & Wu, A.Q. 2008. Modeling of rock discontinuity Mining Science 46(4) :359–366.
with meshless method. Chinese Journal of Rock Mechanics and Wang, S.J. Editor. 2004. Century achievements in Chinese Rock
Engineering 27(10): 2108–2117. mechanics and Rock Engineering. Nanjing: Hohai University
Miao X.X. 2010. Review of research on mechanical behaviors of min- Press.
ing rock mass and its related engineering technological innovation Wang, S.J. 2009. Geological nature of rock and its deduction for rock
progress. Chinese Journal of Rock Mechanics and Engineering, mechanics. Chinese Journal of Rock Mechanics and Engineering
29(10):1988–1998. 28(3): 433–450.
Miao, Y. Yan, F. Wang, Y.H. & Wan, Y.D. 2008. Dual reciprocity Wu, F.Q. Wu, J. & Qi S.W. 2010. Theoretical analysis on mechanism
hybrid boundary node method in geotechnical engineering. Rock of rock burst of brittle rock mass. Journal of Engineering Geology
and Soil Mechanics 29(7): 1872–1876. 18(5):589–595.
Pan, P.Z. Feng, X.T. Huang, X.H. Cui, Q. & Zhou, H. 2009. Coupled Wu, L.X. Liu, S.J. Wu, Y.H. & Wang, J.Z. 2004. From qualitative to
THM processes in EDZ of crystalline rocks using an elasto-plastic quantitative information: the development of remote sensing rock
cellular automaton. Environmental Geology 57(6):1299–1311. mechanics (RSRM). International Journal of Rock Mechanics and
Pan, Y. & Li, Z. 2005. Analysis of rock structure stability in Mining Sciences 41(3): 415–416.
coal mines. International Journal for Numerical and Analytical Wu, L.X. Liu, S.J. Wu, Y.H. & Wang, J.Z. 2006a. Precursors for rock
Methods in Geomechanics 29(10): 1045–1063. fracturing and failure–Part I: IRR image abnormalities. Interna-
Pan, Y. Wang, X. & Li, Z. 2002. Analysis of the strain softening size tional Journal of Rock Mechanics and Mining Sciences 43(3):
effect for rock specimens based on shear strain gradient plastic- 473–482.
ity theory. International Journal of Rock Mechanics and Mining Wu, L.X. Liu, S.J. Wu, Y.H. & Wang, J.Z. 2006b. Precursors for
Sciences 39(6): 801–805. rock fracturing and failure–Part II: IRR T-Curve abnormalities.
Pang, Z.H. Ge, X.R. & Wang, S.L. 2000. Simulation discontinu- International Journal of Rock Mechanics and Mining Sciences
ity with element free Galerkin method. Journal of Engineering 43(3): 483–493.
Geology 8(3): 364–368. Xi, B.P. Zhao, Y.S. Wan, Z.J. Dong, F.K. Feng, Z.J. & Zhang,
Peng, Z.Q. & Ge, X.R. 2004. Suggestion to choice of physical cover N. 2008. Rheological experimental investigation on surround-
functions in numerical manifold method. Rock and Soil Mechanics ing rock mass of granite under high temperature and hydrostatic
25(4): 624–627. stress. Chinese Journal of Rock Mechanics and Engineering 27(8):
Qi, C.Z., Qian, Q.H. & Wang M.Y. 2009. Constitutive relations of 1659–1666.
porous elasto-plastic media. Journal of Mining Science 45(4): Xia, C.C. Wang, W. & Wang, X.R. 2008. Development of coupling
339–346. shear-seepage test system for rock joints. Chinese Journal of Rock
Qian, Q.H. 2009. Some advances in rock blasting dynamics. Chinese Mechanics and Engineering 27(6): 1285–1291.
Journal of Rock Mechanics and Engineering 28(10): 1945–1968. Xia, C.C. Wang, X.D. Xu, C.B. & Zhang, C.S. 2008. Method to iden-
Qian, Q.H. Qi, C.Z. & Wang, M.Y. 2009. Dynamic strength of rocks tify rheological models by unified rheological model theory and
and physical nature of rock strength. Journal of Rock Mechanics case study. Chinese Journal of Rock Mechanics and Engineering
and Geotechnical Engineering 2(1):1–10. 27(8): 1594–1600.
Qian, Q.H. & Zhou, X.P. 2011. Non-Euclidean continuum model of Xiao, H.T. Zhou, W.Y. & Yang, R.Q. 2000. A new model of rheolog-
the zonal disintegration of surrounding rocks around a deep cir- ical damage for slope stability analysis. China Civil Engineering
cular tunnel in a non-hydrostatic pressure state. Journal of Mining Journal 33(6): 94–98.
Science 47(1): 42–52 Xie, H. & Gao, F. 2000. The mechanics of cracks and a statistical
Shan, R.L. Xue, Y.S & Zhang, Q. 2003. Time dependent damage strength theory for rocks. International Journal of Rock Mechanics
model of rock under dynamic loading. Chinese Journal of Rock and Mining Sciences 37: 477–488
Mechanics and Engineering 22(11): 1771–1776. Xie, H. Li, L.Y. Peng, R.D. & Ju,Y. 2009. Energy analysis and criteria
Sun, A.H. & Li, S.C. 2010. A study on sidewall displacement predic- for structural failure of rocks. J Rock Mechanics and Geotechnical
tion and stability evaluations for large underground power station Engineering 1 (1): 11–20
caverns. International Journal of Rock Mechanics and Mining Xie, H. Sun, H. Ju,Y. & Feng, Z.G. 2001. Study on generation of
Sciences 47(7): 1055–1062. rock fracture surfaces by using fractal interpolation. Int J Solids
Sun, J. 2007. Rock rheological mechanics and its advance in engi- & Structures 38: 5765–5787
neering applications. Chinese Journal of Rock Mechanics and Xie, H. & Zhou, H.W. 2008. Application of fractal theory to top coal
Engineering 26(6): 1081–1106. caving. Chaos, Solitons and Fractals 36: 797–807
Sun, J. & Wang, S.J. 2000. Rock mechanics and rock engineering Xu, W.Y. Yang, S.Q. & Chu, W.J. 2006. Nonlinear viscoelasto-plastic
in China: developments and current state-of-the-art. Interna- rheological model (hohai model) of rock and its engineering
tional Journal of Rock Mechanics and Mining Sciences 37(3): application. Chinese Journal of Rock Mechanics and Engineering
447–465. 25(3): 433–447.
Tang C.A. & Hudson, J.A. 2010. Rock Failure Mechanisms – Xu, Z.H. Liu, G.T. Ye, Y.X. & Hu, Y. 2006. Research on Effect of
Explained and Illustrated. Taylor & Francis, London, 322p. Temperature on Permeability of Soft Rock. Journal of China Three
Tang, C.A. Wang, J.M. & Zhang, J.J. 2010. Preliminary engineer- Gorges University (Natural Sciences) 28(4): 301–311.
ing application of microseismic monitoring technique to rockburst Yang, Q. Chen, Y.R. & Liu, Y.R. 2007. On monolithic stability and
prediction in tunneling of Jinping II project. Journal of Rock reinforcement analysis of high arch dams. Science in China Series
Mechanics and Geotechnical Engineering 2 (3): 193–208. E 50: 90–97.
Tian, W. Dang, F.N. Ding, W.H. Liang, X.Y. & Chen, H.Q. 2010. Yang, Q. Tham, L.G. & Swoboda, G. 2005. Normality structures
Development of a dynamic loading apparatus for CT exper- with homogeneous kinetic rate laws. ASME Journal of Applied
iment and its application, Rock and Soil Mechanics 31(1): Mechanics 72: 322–329.
309–313. You, M.Q. 2009. True-triaxial strength criteria for rock. Interna-
Wang, J. 2010. High-level radioactive waste disposal in China: update tional Journal of Rock Mechanics and Mining Sciences 46(1):
2010. Journal of Rock Mechanics and Geotechnical Engineering 115–127.
2 (1): 1–11. You, M.Q. 2010a. Mechanical characteristics of the exponential
Wang, L.G. Li, L. Liu, X. & Wang, J.G. 2010. Life expectation of an strength criterion under conventional triaxial stresses. Interna-
rock engineered slope based on the long-term stability. Chinese tional Journal of Rock Mechanics and Mining Sciences 47(2):
Journal of Geological Hazard and Control 21(3): 97–100. 195–204.
78

You, M.Q. 2010b. Three independent parameters to describe con- Zhou, C.B. Sharma, R.S. Chen, Y.F. & Rong, G. 2008. Flow-Stress
ventional triaxial compressive strength of intact rocks. Journal of Coupled Permeability Tensor for Fractured Rock Masses. Int J
Rock Mechanics and Geotechnical Engineering 2(4): 350–356. for Numerical and Analytical Methods in Geomechanics 32(11):
Yu, M. Zan,Y. Zhao, J. &Yoshimine, M. 2002. A Unified Strength cri- 1289–1309.
terion for rock material. International Journal of Rock Mechanics Zhou, C.B. Sharma, R.S. Chen, Y.F. & Rong, G. 2010. A
and Mining Sciences 39(8): 975–989. New Parabolic Variational Inequality Formulation of Signorini’s
Zhang, G.X. & Peng, J. 2000. Second-order manifold method in Condition for Non-steady Seepage Problems with Complex Seep-
structure failure analysis. Acta Mechanica Sinica 34(2): 261–269 age Control Systems. International Journal for Numerical and
Zhang, J.M. Yao, Z.H. & Li, H. 2002. A Hybrid boundary node Analytical Methods in Geomechanics DOI: 10.1002/nag.944.
method. International Journal for Numerical Methods in Engi- Zhou, H. Jia, Y. Shao, & J.F. 2008. A unified elastic-plastic and
neering 53: 751–63. viscoplastic damage model for quasi-brittle rocks. International
Zhang, P. Li, N. & He, R.L. 2006. Research on localized progressive Journal of Rock Mechanics & Mining Sciences 45: 1237–1251.
damage model for fractured rocklike materials. Chinese Journal Zhou, H.W. & Xie, H. 2003. Direct estimation of the fractal dimen-
of Rock Mechanics and Engineering 25(10): 2043–2050. sions of a fracture surface of rock. Surface Review and Letters
Zhang X. & Lu, M.W. 2000. A 2-D meshless model for joint 10(5): 751–762
rock structures. International Journal for Numerical method in Zhou, H.W. & Xie, H. 2004. Anisotropic characterization of rock
Engineering 47(10): 1649–1661. fracture surfaces subjected to profile analysis. Physics Letters A
Zhang, X. Lu, X.H. & Song, K.Z. 2001. Least-squares collocation 325(5–6): 355–362
meshless method. International Journal for Numerical Method in Zhou, J.J. Zhou, H. & Shao, J.F. 2007. Coupled micromechanical
Engineering 51: 1089–1100. model for anisotropic damage and permeability variation in brittle
Zhang, X. Song, K.Z. & Lu, M.W. 2000. Meshless methods based on rocks. Chinese Journal of Rock Mechanics and Engineering 26(2):
collocation with radial basis function. Computational Mechanics 368–373.
26(4): 333–343. Zhou, T. Zhang, J.S. & Chen, X.B. 2006. Two dimensional elasto-
Zhang, X.S. Ye, T.Q. & Ge, S.L. 2001. Several problems associ- plastic distinct element analysis of rock slope in navigation route
ated with discontinuous boundary elements. Chinese Journal of of longtan hydropower station. Journal of Railway Science and
Computational Mechanics 18(3): 331–334. Engineering 3(4): 63–67.
Zhang, Y.L. Feng, X.T. & Ru, Z.L. 2004. Large-scale high perfor- Zhou, X.P. Xie, W.T. & Qian, Q.H. 2010. Bifurcation of collinear
mance parallel finite element system based on domain decom- crack system under dynamic compression.Theoretical andApplied
position method in geomechanics. Chinese Journal of Rock and Fracture Mechanics 54:166–171.
Engineering 23(21): 3636–3641. Zhou, X.P. & Qian, Q.H. 2011. Deformation and failure Mechanism
Zhao, Y.S. Wan, Z.J. Zhang, Y. Qu, F. Xie, G.Y. Wei, X.J. & Ma, W. of surrounding rock mass around underground caverns in Jinping
2008. Research and development of 20MN servo-controlled rock I hydropower station. Proceeding of the 12th ISRM International
triaxial testing system with high temperature and high pressure. Congress on Rock Mechanics, Taylor & Francis, London.
Chinese Journal of Rock Mechanics and Engineering 27(1): 1–8. Zhou, W.Y. Zhao, J.D. Liu,Y.G. &Yang, Q. 2002. Simulation of local-
Zheng, H. 2009. Eigenvalue Problem from the Stability Analy- ization failure with strain-gradient-enhanced damage mechanics.
sis of Slopes. Journal of Geotechnical and Geoenvironmental International Journal for Numerical and Analytical Methods in
Engineering, ASCE 135(5): 647–656. Geomechanics 26: 793–813.
Zheng, H. & Jiang, W. 2009. Complementary theory based on discon- Zhu, D.Y. & Qian, Q.H. 2007. Rigorous and quasi-rigorous limit
tinuous deformation analysis. Science in China E Series 39(10): equilibrium solutions of 3d slope stability and application to engi-
1702–1708. neering. Chinese Journal of Rock Mechanics and Engineering
Zheng, H. Sun, G. H. & Li, C.G. 2011. Cauchy problem of three- 26(8): 1513–1528.
dimensional critical slip surfaces of slopes. International Journal Zhu, W.S. Li, X.J. Zhang, Q.B. Zheng, W.H. Xin, X.L. Sun, A.H. &
for Numerical & Analytical Methods in Geomechanics 35(4): Li, S.C. 2010. A study on sidewall displacement prediction and
519–527. stability evaluations for large underground power station caverns.
Zheng, Y.R. 2007. Discussion on yield and failure of geomateri- International Journal of Rock Mechanics and Mining Sciences
als and stability analysis methods of slope/landslide—communion 47(7): 1055–1062.
and discussion summary of special topic forum on geologic dis- Zhuang, X.Y. Cai, Y.C. Zhu, H.H. & Zhou, D.H. 2008. Rock slope
asters in the three gorges project region. Chinese Journal of Rock stability analysis based on meshless method and shortest path
Mechanics and Engineering. 26(4): 649–661. algorithm. Chinese Quarterly of Mechanics 29(4): 538–543.
Zheng, Y.R. Xu, H. Wang C. & Xiao Q. 2010. Failure mecha- Zuo, J.P. Xie, H. Zhou, H.W. & Peng, S.P. 2010. SEM in-situ
nism of tunnel and dividing line standard between shallow and investigation on thermal cracking behavior of Pingdingshan sand-
deep bury. Journal of Zhejiang University (Engineering Science), stone at elevated temperatures. Geophysical Journal International
44(10):1851–1856. 181(2):593–603.
79

Rock strength and failure: Some common and uncommon issues
A.J. Gonzalez-Garcia
School of Engineering, National University of Colombia, Bogota D.C
ABSTRACT: In this paper two issues about rock strength are reviewed by the Author: a common one (a) rock strength and
brittleness and a supposedly uncommon one (b) intact rock, its definition and methods to obtain its value. For the first issue,
a proposal is presented for a new simple definition for brittle-ductile transition behaviour. For the second issue an alternative
laboratory method to obtain intact rock strength is proposed, instead of the recommended triaxial tests; also a refined definition
of intact rock is recommended, which leads to briefly examine the topic of rock sample disturbance.
Subject: Rock material and rock mass property testing (laboratory and in situ)
Keywords: lab testing, rock mass, rock properties
1 SHEAR STRENGTH AND ITS COMPONENTS mainly from density and angularity of particles in granular
soils and from roughness in rock discontinuities. It is usually
A characteristic of ceramic materials, such as earthen materials represented by ν = tan i.
(soils and rocks) is that they are very weak in tension and Solid friction, or surface resistance, is a very complex phe-
strong in compression, as compared, for instance, with metals, nomenon. It is totally ductile and plastic, and it is not possible
which can be more or less equally strong under tension and to eliminate it, unless contact between surfaces is lost. It is the
compression, all of these being very well known facts. last component of strength in all solid materials and in earthen
Since solids only can fail under tension or shear stresses, materials it can be considered a property of minerals and it is
other types of failure being really an indirect result of these represented by µm = tan φm
stresses (compression and torsion failures are really shear Usually, the components of dilatancy v and friction µm
failures, flexure failure is a combination of tension failure are combined into the so-called effective friction µ =
and shear failure, etc), shear strength is the preferred strength tan(φm + i) = tan φ = dτ/dσ , and cohesion is represented by
parameter for geotechnical materials. the cohesion intercept c (τ for σ = 0). However, in curved
Shear strength can be postulated as composed of three envelopes, such as those of coarse granular materials and rock
main parts: cohesion c , dilatancy ν and friction µ , with the discontinuities (in which σt = 0) or rock materials and masses
apostrophe to denote effective values (Figure 1) (in which σt > 0), c tends to lose its meaning, that comes from
Cohesion, or internal strength, is given by crystal molecu- the Mohr-Coulomb linear equation (Eq. 1).
lar bonds, electrochemical surface forces or cements, and it
really comes from tensile strength σt . In preconsolidated clays,
residual soils and rock materials it is the main component of 2 STRENGTH CRITERIA
strength: it is usually rigid, somewhat elastic but very brittle,
so it is quickly lost with deformation. Prior to its maximum In this paper the following well-known criteria will be used:
value it provides elasticity to the earthen material. Mohr-Coulomb as a general one, Hoek-Brown (Hoek et al,
Dilatancy, or resistance to volume change, is brittle, not 2002) for rock materials and masses and Barton-Bandis
very rigid, elasto-plastic and can be positive or negative. It is (Barton&Bandis, 1982, 1990) for rock discontinuities.
one of the main strength components of granular materials and 2.1 Mohr-Coulomb Strength Criterion
rock discontinuities. It varies with normal stress and comes
where
2.2 Hoek-Brown (H-B) Strength Criterion
where
Figure 1. Shear strength components.
81

2.3 Barton-Bandis (B-B) Strength Criterion
where
3 AN OVERALL VIEW OF ROCK

STRENGTH-BRITTLENESS AND
RESIDUAL FRICTION
Figure 2. Mohr-Coulomb diagram for quartz, steel, intact quartzitic
sandstone and quartz mineral surface friction.
3.1 Variability of rock strength
Rocks cover a wide range of strength, from the lowest min- Table 1. Hoek-Brown strength parameters – Quartzitic Sandstone
eral/mineral friction to the highest pure mineral shear strength Rock Masses
which are the limits for rock strength behaviour. Curve Qtz.S/stone Rock Mass GSI D m s
As an example, the common mineral quartz is chosen and
a sedimentary rock such as quartzitic sandstone is analysed. 2 Intact qtz. sandstone 100 0.0 20.00 1.000
For the analyses, in Mohr-Coulomb diagrams of Fig- 3 Average core sample 90 0.0 13.99 0.329
ure 2, and the two other than follow, for dry conditions, 4 Good rock mass 70 0.5 4.79 0.018
a maximum normal stress σmax = 100 MPa has been cho- 5 Poor rock mass 30 0.5 0.71 8.8E-05
sen, which roughly corresponds (asuming rock mass unit
weight γ = 25 kN/m3 ) to a vertical total stress of a rock
mass column 4,000 meters deep. This near extreme situation Table 2. Barton-Bandis strength parameters – Quartzitic Sandstone
Discontinuities
arises, for example, in deep oil wells (i.e. in Siberia: Z-11
well, 11,300 m), gold mines in South Africa (i.e. TauTona Qtz. Sandstone φr JRC JCS
mine–3,900 m), etc. Also, τ − σ curves of materials with Curve Rock Discontinuity (◦ ) (◦ ) (MPa)
Hoek-Brown parameters were derived using the algorithm
developed by J.Bray (Hoek, 1983) 6 Discontinuity – Lab 25.00 15.00 60.00
Quartz has estimated Mohr-Coulomb parameters c = 7 Discontinuity – Field–L n = 20 m 25.00 3.06 5.53
950 MPa and φ = 13.25◦ (Skempton, 1961) and friction 8 Residual friction φr ≈ φb 25.00 0.00 60.00
9 Quartz/quartz dry 9.09 0.00
for quartz/quartz dry surfaces can be taken as µm = 0.16
10 Brittle-ductile 33.06 0.00 Included
(φm = 9.09◦ ) (Horn & Deere, 1962). Quartzitic sandstone typ- transition (Mogi) criterion
ically can have an intact compressive strength σci = 200 MPa,
and for intact quartzitic sandstone, Hoek-Brown parameters
are GSI = 100, m = mi = 20 (Hoek et al, 2002), s = 1.0 and stress (σ = 1 MPa) to 60 for high normal stress
D = 0.0 (Figure 2) (σ = 100 MPa)
Figure 2 also includes, as a comparison, a derived theoret- c) The shear strength relationship (quartz mineral)/(intact
ical shear strength envelope (Curve A) for a common steel quartzitic sandstone) varies from 32.9 for low normal stress
(fy = 420 MPa) for which, using von Mises theoretical
√ pure (σ = 1 MPa) to 6.3 for high normal stress (σ = 100 MPa)
shear yield criterion (σ1 = −σ3 = τp = fy / 3 = 0.577fy ), d) The shear strength relationship: (intact quartzitic sand-
gives estimates of c = 245.5 MPa and φ = 8.9◦ . stone)/(dry quartz surface) varies from 180.8 for low
From this Figure 2 some interesting points can be seen: normal stress (σ = 1 MPa) to 9.6 for high normal stress
(σ = 100 MPa)
a) Quartz mineral is, on average, 3.8 times stronger than steel,
a fact usually seen when some small quartz particles remain Mainly in this last vast range is where quarzitic sandstone
embedded into the surface of a field hammer after striking strength could vary, and to explore this variation, a series
a quartzite or a sandstone. of hypothetical strength envelopes are examined for clean
b) The shear strength relationship (quartz mineral)/(dry rock discontinuities and rock masses, with strength parameters
quartz surface) varies from about 5,000 for low normal indicated in Tables 1 and 2, and curves shown in Figure 3.
82

Figure 3. Mohr-Coulomb diagram for quartzitic sandstone rock masses and discontinuities.
Figure 4. Mohr-Coulomb diagram for quartzitic sandstone rock masses and discontinuities–close-up of Figure 3.
Examining these quartzitic sandstone envelopes: curves 2 3.2 Rock strength and brittleness
(intact rock), 3 (average core sample) and 4 (good rock mass),
Considering brittleness and using the common Mogi crite-
all show a clear behaviour, brittle condition and descending
rion for brittle-ductile transition σ1 /σ3 = 3.4 (Mogi, 1966 in
strength.
Mogi, 2007), equivalent to τ/σ = tan(33.06◦ ), and examining
However, curves 5 (poor rock mass), 6 (discontinuity-
Figure 4, the following facts emerge:
lab), 7 (discontinuity-field) and 8 (discontinuity-residual
friction) seem intermingled and without a clear distinctive a) Quartzitic sandstone poor rock mass strength becomes
behaviour, so a closer look is taken at these last four curves ductile for normal stress σ > 40.7 MPa (upper circle)
(Figure 4). beyond which H-B strength criterion is no longer valid.
83

b) Quartzitic sandstone laboratory discontinuity strength In the same paper, Barton (1976) has discussed at length his
becomes ductile for normal stress σ > 17.4 MPa (lower criterion for rock discontinuities (equation 6) and concluded,
circle) among other things:
c) Quartzitic sandstone field discontinuity strength becomes “3. At high levels of effective normal stress the frictional
ductile for normal stress σ > 0.0129 MPa strength is related to the fracture strength by means of the
d) Quartzitic sandstone discontinuity residual friction confined compression strength which is represented by the dif-
strength and quartz/quartz dry friction strength are all ferential stress (σ1 − σ3 ) at fracture. The dimensionless ratio
ductile, as friction alone is always ductile. (σ1 − σ3 )/σn (of which σc /σn is a special case, i.e. σ3 = 0)
varies relatively little over a wide range of σ3 and results in
Considering residual friction angle φr , other facts appear: a limited range of frictional strength at high stress levels . . .
compared to the wide range of frictional strength exhibited at
e) Poor rock quartzitic sandstone rock mass strength becomes low stress levels . . .”
less than that given by residual friction angle φr for normal
stress σ > 98.4 MPa (upper triangle)
f) Quartzitic sandstone laboratory discontinuity strength 3.3 Some conclusions about rock strength and
becomes less than that given by residual friction angle brittleness
φr for normal stress σ > 60.0 MPa = lab JCS (middle
triangle) All these analyses and considerations led the Author to the
g) Quartzitic sandstone field discontinuity strength becomes following conclusions:
lower than that given by residual friction angle φr A- Solid friction by itself is always ductile, as postulated
for normal stress σ > 5.53 MPa = field JCS (lower in Part 1, and therefore for an earthen material to have
triangle) brittleness, either cohesion or dilatancy or both must be
present.
From these points the following questions arise: B- Although there are average criteria, each rock should have
its own brittle-ductile transition criterion, corresponding
1) Is common Mogi brittle-ductile criterion valid for all rock mainly to its mineralogical nature.
materials, masses and discontinuities? C- Therefore the following criterion is proposed by the
2) Can rock strength be lower than that given by residual Author, to use in normal rock mechanics works:
friction angle φr ? BRITTLE-DUCTILE TRANSITION FOR ROCK
MATERIALS, ROCK DISCONTINUITIES AND ROCK
Mogí criterion was developed from a series of tests in MASSES IS GIVEN BYTHE RESIDUAL FRICTIONAL
mainly igneous and metamorphic rock materials, although he EFFECTIVE STRENGTH (φ’r ) OF THE MATERIALS
used also data from other sources for other type of rocks. The INVOLVED IN THE FAILURE MECHANISM
common Mogi criterion really applies to silicate rocks whereas D- And a second criterion proposed is: THE RESIDUAL
the criterion for carbonate rocks is somewhat different, spe- FRICTION STRENGTH GIVEN BY:
cially for σ < 200 MPa. In Mogi’s derivation of his criterion,
no mention is done of rock masses or discontinuities. (Mogi,
2007). Besides this, the criterion is an average line which sep- IS THE LOWER LIMIT OF ROCK STRENGTH FOR
arates the brittle and ductile behaviors, but there is dispersion. ENGINEERING PRACTICAL CASES.
Therefore the Author concludes that common Mogi’s criterion Residual friction angle (φr ) values for rocks can be esti-
was derived only for rock materials and that its formulation mated by means of:
can be interpreted as an average lower limiting friction angle
(φlimit = 33.06◦ ). a) Laboratory shear tests done on sand-blasted flat sawn
Barton (1976), referring to the brittle-ductile criterion for surfaces.
rock materials by Byerlee (1968), states: “A single curve for b) Using the methods described by Barton & Choubey (1977)
all rocks is then inadmissable as a general transition bound- mainly tilt tests, Schmidt hammer tests and direct shear test.
ary between brittle and ductile behaviour”, a concept shared This reference gives a formula:
by the Author. In this same paper, Barton proposes a “crit-
ical state line” for rocks given by σ1 /σ3 = 3.0, equivalent to
τ /σ = tan 30.0◦ , very similar to Mogi’s criterion. where
Also Hoek (1983) states: “Examination of the results plot- r = Schmidt rebound on weathered joint surface
ted in figure 9, and of similar results plotted by Mogi, shows R = Schmidt rebound on unweathered rock surface
that there is room for a wide variety of interpretations of the φb = basic friction angle obtained from tilt tests
critical principal stress ratio, depending upon the curve fitting c) Tables or figures from literature (Barton & Choubey, 1977,
procedure employed and the choice of the actual brittle-ductile also give a table of φ’r for several rocks).
transition point. The range of possible values of σ1 /σ3 appears
Some final remarks are given on these proposed criteria:
to lie between 3 and 5”
In dealing with rock discontinuities, points f) and g) of 1) Although φr (≤φb ) is proposed to give the lower limit for
discontinuity strength becoming less than that given by resid- rock strength, it refers to solid friction. If gouge is gener-
ual strength, they result from the Barton-Bandis criterion ated during the shearing process, for instance in granular
itself (equation 6), because when σ = JCS, then τd = σ tan φr , rocks, like sandstone or quarztzite, the resultant particles
roughness effects disappear and Hoek (2007a), referring to could generate a rolling bearing effect and then rolling
this criterion states: “An upper limit for σn is given by friction appears, which is lower than static friction, and
σn = JCS”. proposed criterion 2 will not be longer valid.
84

Hoek & Brown (1980b) remark: “A value of σci is required
as a scaling factor to determine the strength of a particular
sample of rock”. (σc in the original text and equation–Author’s
note)
Rearranging equation 11, gives
Equation 11a is a linear equation of the form Y = AX + B

if Y = (σ1 − σ3 )2 , X = σ3 , A = mσci and B = sσci2
To determine σci , Hoek (2007) uses equation 12 with s = 1
and m = mi and recommends: “The relationship between the
principal stresses at failure for a given rock is defined by
two constants, the uniaxial compressive strength σci and a
constant mi . Wherever possible the values of these constants
should be determined by statistical analysis of the results of a
set of triaxial tests on carefully prepared core samples.” . . .
“When laboratory tests are not possible, Table 2 and Table
Figure 5. Limitations of the Hoek-Brown criterion (taken from 3 can be used to obtain estimates of σci and m i Table 2,
Hoek, 2007). referred to σci , lists a descending series of grades for rock
materials (R6 to R1) according to the unconfined compres-
sive strength UCS = σc of rock materials, from very strong
2) In some rocks φr could vary with mosture content, particle (R6: σc > 250 MPa) to very weak (R1: σc = 0.25 to 1.0 MPa)
orientation (like in micaceous or foliated rocks), stress rate, and gives field estimates criteria for this strength. It must be
temperature, etc. Therefore, proper field conditions should noted that constants A and B related to equation 11a involve
be taken into account. three variables for intact rock, namely: mi , σci and s and linear
3) In weak rock masses and beyond the brittle-ductile tran- regression only gives two constants. Therefore one of the three
sition, ductile linear strength predominates, greater than variables has to be assumed in order to deduce the other two.
Hoek-Brown envelope, a fact that is not always checked in Hoek (2007) chooses s = 1.0, assuming that the rock samples
common enginnering works. corresponds to intact rock, but one of the other two unknowns
could be chosen as well.
4 HOEK-BROWN STRENGTH CRITERION – INTACT
4.2 Alternative method to find H-B intact rock parameters
ROCK DEFINITION – TESTS TO GET ITS STRENGTH
In several countries, especially those belonging to the devel-
4.1 Hoek-Brown strength criterion and intact rock oping, so named Third World, it is not easy to carry out triaxial
tests on rock samples (as an example, in Colombia there are
The well known and widely used Hoek-Brown strength cri-
only two facilities) and therefore the Author has been using
terion (Hoek & Brown, 1980a, b; Hoek, 1983) (equation 2),
an alternative method to deduce the intact rock parameters.
has been an essential landmark in rock mechanics, since for
The method employs unconfined compression and Brazil-
the first time it linked into a single equation the strength of
ian indirect tension tests (i.e., ISRM, 1978, Part 2), two tests
rock materials and of fractured rock masses. The subsequent
that are more readily available than the triaxial test. The indi-
developments, described in detail by Hoek & Marinos (2007),
rect Brazilian tensile strength σtb was originally developed
had added and improved its uselfulness for a varied number
for concrete, and knowing that σ3 = −σtb and that σ1 = 3σtb ,
of practical problems.
using the Hoek-Brown criterion, σtb is given by:
However, it has limitations, as all empirical based criteria
have of which the main ones are (Hoek, 1983),:
a) It only can be used for homogeneous materials (Figure 5)
b) It only can be used for isotropic materials (Figure 5) and also, from equation 2, with σ3 = 0 and s = 1.0
c) It was developed only for brittle materials, according to
Mogi criterion (σ1 /σ3 ≤ 3.4)
And the Author proposes to add another limitation: replacing s from equation 13 into equation 12 gives m as:
d) It is not a direct criterion but a relative one based on intact
rock strength σci .
When the criterion was developed (Hoek & Brown, with the limitation that σc > 4σtb , so that m > 0.
1980a,b), the concept of intact rock was: “intact rock mate- With two equations and three unknown parameters, also
rial making up the specimen”, that is, the rock material of a here one of the three has to be adopted. If s = 1.0 is adopted,
sample, as compared with a fractured rock mass. then σci = σc , σtbi = σtb .
Furthermore, equation 2 was stated and worked in dimen- This is a simple formulation, for which the apostrophes
sionless terms, using σci as the nomalizing parameter and for effective stresses had been dropped. However, it lacks
with a = 0.5, value that will be used for this discussion: the averaging behaviour of the triaxial test, for which Hoek
(2007) recommends at least five points, recommendation
which implies five samples of the same rock material.
85

Table 3. Test results in dry silty sandstone for s = 1.0 In normal civil engineering works, as the one from which the
data for the example were taken, rock drilling and sampling is
Depth U.Wgth σc = σci σtb = σtbi GSI not always a very careful operation and therefore some damage
No (m) (kN/m3 ) (MPa) (MPa) mi avg mi to the samples is to be expected. Besides that, in very fractured
1 5.75 23.63 18.05 3.30 2.55 69
or very stressed rock masses, sampling is difficult and in this
2 8.25 24.57 21.97 3.25 4.40 84 situation a value for σci is not easyly defined.
3 16.00 23.92 18.73 1.28 13.73 116 Also, the value for σci depends upon the definition of “intact
4 18.35 24.83 16.57 1.17 13.08 115 rock”. If it is only defined as “..the unfractured blocks between
5 22.50 23.59 19.91 2.33 6.70 96 discontinuities in a typical rock mass.” (i.e. Zhang, 2005), in
6 23.95 25.37 23.74 3.10 5.57 91 opinion of the Author, this definition really applies to the rock
Avg ⇒ 24.32 19.83 2.40 7.67 95.4 material and this was the definition originally applied when
c.v. (%) ⇒ 2.96 13.31 40.92 60.61 19.01 the Hoek-Brown criterion was developed. Much earlier on,
Deere & Miller (1966), for his carefully series of tests in many
types of rocks, stated a definition for intact rock, which refines
Table 4. Test results in dry silty sandstone for σci = σc max the abovementioned common one as follows:
“Intact rock is defined herein as a polycrystalline solid.,
Depth σc σtb GSI
No (m) (MPa) (MPa) s from s mi consisting of a natural aggregate of minerals, the properties
of which depend upon:
1 5.75 18.05 3.30 0.578 95 2.32 i) the physical properties of the constituents, and
2 8.25 21.97 3.25 0.857 99 4.28
ii) the type of bonding of these constituents to one another”
3 16.00 18.73 1.28 0.623 96 12.61
4 18.35 16.57 1.17 0.488 94 11.50 And in describing their tests, these authors say, with underlin-
5 22.50 19.91 2.33 0.704 97 6.29 ings by the present Author:
6 23.95 23.74 3.10 1.000 100 5.57 “All tests reported herein were performed on intact, cylin-
Avg ⇒ 19.83 2.40 0.708 96.6 7.10 drical rock specimens, macroscopically homogeneous and
c.v. (%) ⇒ 13.31 40.92 26.80 2.45 57.62
free from fractures, joints, and seams. The specimens were
prepared from diamond-drill cores from field projects, or from
The use of the method is illustrated with an example of tests core drilled in the laboratory from rough, quarry blocks. All
in NW dry core samples of a silty sandstone from the Lower cores were standard NX-size, having a nominal diameter of
Tertiary Guaduas formation near Bogota, Colombia. 2 1/8 inches”
From this Table 3, the design values could be taken as the Here, three important factors emerge to obtain σci
average underlined values: σci = 19.83 MPa and mi = 7.67. a) The need of homogeneous samples
Also in Table 3, the values of GSI for the samples are back- b) The features that make a sample “non intact”, such as
calculated with equation 3, using [mi/(avg mi)] and D = 0.0 fractures, joints and seams.
as the calculation parameters. It is noted in Table 3 that for c) The size of the samples
two samples the GSI values are greater than 100.
These anomalous values result from the statistical treat- Point b) complies with the normal definition of “intact”
ment of the data (which is intrinsically done in the case of (i.e. Concise Oxford Dictionary, 2008): “not damaged or
the triaxial tests), but they contradict the definition of GSI, impaired.”
which should be GSI ≤ 100. Therefore another criteria, dif- Point c) was stressed by Hoek & Brown (1980a, b), by
ferent from s = 1.0 was adopted by the Author, in order to setting the popular relationship for unjointed rock samples:
seek coherence for the method and to avoid contradictions
with the postulates for the parameter GSI. Instead of using
s = 1.0, a value for σci is adopted.
In order to have GSI ≤ 100 for the series of data, this value in which: σc = UCS for a sample of diameter d (mm)
of σci must be σci ≥ σc max . As a first approximation, the calcu- σc50 = UCS for a 50 mm diameter sample
lations are done for σci = σc max , D = 0.0 and using equation After all this discussion, it seems clear that the simple com-
13 to obtain s, equation 4 to backcalculate GSI from s, equa- mon definitions of intact rock and intact rock strength in the
tion 3 to obtain (m/mi ) and equation 14 to obtain m, then geotechnical literature, equivalent to rock material and rock
mi = m/(m/mi ). Results are in Table 4. material strength, are somewhat misleading, and that a refined
From Table 4, the design values could be taken as the definition could be such as the following, that the Author
underlined values: σci = σc max = 23.74 MPa and mi = 7.10. proposes:
However, some designer may think that the values from the Intact rock: rock material free from fractures, joints and
sample with maximum σc are the ones to be used and then seams, that has minimal disturbance by the sampling process.
mi = 5.57, but this would lack representativeness for the data. Intact rock strength: the unconfined compressive strength
Table 4 also shows that with this procedure, GSI ≤ 100 of intact rock tested in samples with diameter or dimension
and dispersion for mi and GSI, denoted by the coefficients equal to 50 ± 10 mm.
of variation c.v., reduces in comparison with Table 3.
4.4 Effect of sample disturbance on rock strength
4.3 Discussion about the definition of σci .
With these proposed definitions, the influence of the quality of
The values of GSI for the samples also raises another issue: rock samples becomes an essential factor, and although it was
ARE COMMON CORE SAMPLES, REALLY OF INTACT taken into account during the development of the Hoek-Brown
ROCK, IN WHICH GSI SHOULD BE 100?. criterion and has proper consideration in several geotechnical
86

Table 5. Test results in silty sandstone for σci = 2σc max
Depth σc σtb GSI

No (m) (MPa) (MPa) s from s mi
1 5.75 18.05 3.30 0.145 83 1.81

2 8.25 21.97 3.25 0.214 86 3.34
3 16.00 18.73 1.28 0.156 83 9.85
4 18.35 16.57 1.17 0.122 81 8.98
5 22.50 19.91 2.33 0.176 84 4.91
6 23.95 23.74 3.10 0.250 88 4.35
Avg ⇒ 19.83 2.40 0.177 84.2 5.54
c.v. (%) ⇒ 13.31 40.92 26.80 2.82 57.62
Figure 6. Relationship between GSI, damage factor D and ampli-

fication factor Fa of σc max for example data.
works around the world, it is not commonly stressed in

many common works, and is considered less important in
many developing countries.
If it is accepted that rock samples have disturbance, then
an examination of the effects of disturbances is presented for
the example data of σc and σtb of Table 3.
The analysis is done varying D of the H-B criterion and
the value of σci , so as to have a desired value of average GSI.
Then, the obtained value of σci is divided by σc max to get the
intact rock amplification factor Fa for σc max
Figure 7. Hoek-Brown rock strength envelopes for example data-
core samples – equally scaled axes.
In Figure 6 the variation of Fa with the damage parameter
D and average GSI is presented.
This Figure 6 indicates that:
a) For constant sample damage D, as Fa increases, average
GSI for the sample data decreases
b) For constant Fa, average GSI for the sample data increases
as damage D increases
c) For constant average GSI for the sample data, Fa increases
as damage D increases
It must be stressed that this figure 6 is applicable only to
the example data presented and it is not a general criterion,
although similar trends are expected with other data.
4.5 Comparison of rock strength envelopes deduced with

different values of intact rock strength
Figure 8. Hoek-Brown rock strength envelopes for example data-
If estimated damage to the samples described in tables 3 core samples – distorted σ3 axis.
and 4, is, for instance D = 0.0, and an average core value
of GSI = 84 is estimated, then Fa = 2.0 can be adopted. The
deduced parameters are indicated in Table 5.
From Figures 7 and 8 the following things can be seen:
Also, from shear strength tests done on discontinuities of
this same material of the example, an average value of residual a) The minimum rock core strength is given by the average
friction angle φr = 26.6◦ was obtained, which gives, with the σci deduced with s = 1.0 (curve 1)
proposed new criterion (see 3.3), a brittle-ductile transition b) The rock core strength calculated with σci = σc max
σ1 /σ3 = 2.51 (curve 2) is, on average, only 3.3% higher than the one
In Figures 7 and 8, deduced rock sample strength envelopes calculated with average σci deduced with s = 1.0
are presented for the example core sample data, the first one c) The rock core strength calculated with σci = 2σc max
with equally scaled σ1 vs σ3 (S1 vs S3) axes and the second (curve 3) is, on average, 19.6% higher than the one
one with distorted σ3 (S3) axis. calculated with average σci deduced with s = 1.0
87

Figure 9. Hoek-Brown rock strength envelopes for example – good Figure 11. Hoek-Brown rock strength envelopes for example – poor
rock mass (GSI = 70) – equally scaled axes. rock mass (GSI = 30) – equally scaled axes.
Figure 10. Hoek-Brown rock strength envelopes for example – Figure 12. Hoek-Brown rock strength envelopes for example – poor
good rock mass (GSI = 70) – distorted σ3 axis. rock mass (GSI = 30) – close-up with distorted σ3 axis.
From Figures 11 and 12 the following things can be seen:

d) For the proposed brittleness criterion with φr , all envelopes
are brittle for σ3 < 30 MPa, whereas for Mogi criterion, a) For the new brittleness criterion with φr , brittle behavior
this brittlenes ends at σ3 ≈ 28 MPa for curve 1 and at ends at σ3 ≈ 0.60 MPa for curve 1, at σ3 ≈ 0.65 MPa for
σ3 ≈ 30 MPa for curve 2. curve 2 and at σ3 ≈ 1.05 MPa for curve 3, whereas for Mogi
criterion these limits are, respectively, σ3 ≈ 0.17 MPa,
For good rock mass with GSI = 70, Figures 9 and 10 are σ3 ≈ 0.20 MPa and σ3 ≈ 0.40 MPa
presented and the following things can be observed: b) Due to these very low limits for brittle behavior, residual
a) The minimum rock mass strength is again given by the strength dominates rock mass behavior for σ3 > 1.0 MPa
average σci deduced with s = 1.0 (curve 1) and for this range the unapplicable Hoek-Brown criterion
b) The rock mass strength calculated with σci = σc max (curve greatly underestimates rock mass strength, a fact that is not
2) is, on average, only 4.6% higher than the one calculated always taken into account in common engineering designs.
with average σci deduced with s = 1.0. However, for the brittle range (0.0 < σ3 < 1.0 MPa):
c) The rock mass strength calculated with σci = 2σc max (curve
3) is, on average, 25.3% higher than the one calculated with c) The minimum rock mass strength is given, as before, by
average σci deduced with s = 1.0 the average σci deduced with s = 1.0 (curve 1)
d) For the new brittleness criterion with φr , brittle behav- d) The rock mass strength calculated with σci = σc máx (curve
ior ends at σ3 ≈ 16.5 MPa for curve 1 and 2, and at 2) is, on average, only 3.4% higher than the one calculated
σ3 ≈ 28 MPa for curve 3, whereas for Mogi criterion these with average σci deduced with s = 1.0.
limits are, respectively, σ3 ≈ 6.5 Mpa, σ3 ≈ 7.5 MPa and e) The rock mass strength calculated with σci = 2σc máx
σ3 ≈ 11.5 MPa (curve 3) is, on average, 21.6% higher than the one
calculated with average σci deduced with s = 1.0
For poor rock masses with GSI = 30, Figures 11 and 12 are
presented:
4.6 Some conclusions about intact rock strength
For this poor rock mass it is seen that brittle behavior ends at
very low confining stress, so Figure 12 is a close-up of Figure A- The Hoek-Brown rock strength criterion is a relative
11, with distorted σ3 axis. criterion based on intact rock strength σci
88

B- The original definition of intact rock was equivalent to the 5 FINAL REMARKS AND ACKNOWLEDGEMENTS
defintion fo rock material i.e.: “. . . rock material making
up the specimen. . .” (Hoek & Brown, 1980 a,b) The Author hopes that with this brief discussions on rock
C- Refined definitions are proposed by the Author: strength and the proposed criteria and methods, there could
be some contribution to our improved application of rock
Intact rock: rock material free from fractures, joints and
mechanics to the practical problems encountered in many
seams, that has minimal disturbance by the sampling
engineering works.
process.
The Author wishes to thank Professor Evert Hoek for hav-
Intact rock strength: the unconfined compressive strength
ing initiated him at Imperial College, many years ago, into
of intact rock tested in samples with diameter or dimension these fascinating rock mechanics topics.
equal to 50 ± 10 mm.
D- The recommended procedure to obtain intact rock strength REFERENCES
σci is (Hoek, 2007): “The relationship between the prin-
cipal stresses at failure for a given rock is defined by two Barton, N. 1976. The shear strength of rocks and rock joints. Int. J.
constants, the uniaxial compressive strength σci and a Rock Mech. Min. Sci. & Geomech. Abstr. Vol 13, pp. 255–279.
constant mi . Wherever possible the values of these con- Barton, N. & Bandis, S. 1982, Effects of block size on the shear
stants should be determined by statistical analysis of the behaviour of jointed rock, Keynote Lecture, Proc. 23rd U.S. Symp.
results of a set of triaxial tests on carefully prepared core on Rock Mechanics., pp. 739–760.
samples.”. . . ”When laboratory tests are not possible, Barton, N. & Bandis, S. 1990, Review of predictive capabilities of
Table 2 and Table 3 can be used to obtain estimates of JRC-JCS model in engineering practice, Proc. Int. Symp. On Rock
σci and mi ”. The Author notes that for triaxial data the Joints, Norway, pp. 603–610.
condition s = 1 must be adopted in order to obtain σci and Barton, N.R. & Choubey, V. 1977. The shear strength of rock joints
in theory and practice, Rock Mechanics 10(1–2), 1–54.
mi from regression analyses.
Byerlee, J. D. 1968. Brittle-ductile transition in rocks. J. Geophys.
E- The Author proposes an alternative method to obtain σci Res. 73, 4741–4750 (1968).
and mi by using unconfined compression and Brazilian Deere, D.U. & Miller, R.P. 1966. Engineering classification and index
indirect tension tests in a group of at least 5 samples of the properties for intact rock- Technical Report AFWL-TR-65-116.
same rock material. 327pp. December, 1966
F- Care should be taken with Brazilian indirect tensile Hoek, E. 1983. Strength of jointed rock masses, 23rd. Rankine Lec-
strength and unconfined compression strength tests in ture. Géotechnique 33(3), 187–223. Also at www.rocscience.com/
anisotropic rocks, so as to procure that in both tests the education/hoeks_corner.
load is applied approximately at the same angle in relation Hoek, E. 2007. Practical Rock Engineering – An Ongoing Set of
to the direction of anisotropy. Notes, available on the Rocscience website, www.rocscience.com/
education/hoeks_corner
G- To avoid getting GSI > 100, initially it was proposed to
Hoek, E. & Brown, E.T. 1980a. Underground excavations in rock.
adopt intact rock strength equal to the maximum UCS of London: Instn Min. Metall.
the sample group, i.e., σci = σcmax Hoek, E. & Brown, E.T. 1980b. Empirical strength criterion for
H- After examining data and accepting that there is sample rock masses. Journal Geotechnical Engineering Division, ASCE
disturbance, the Author proposes to adopt 106(GT9), 1013–1035.
Hoek, E., Carranza-Torres, C. & Corkum, B. 2002. Hoek-
Brown Failure Criterion – 2002 Edition. 5th North American
Rock Mechanics Symposium and 17th Tunneling Association of
Canada Conference: NARMS-TAC, 2002, pp. 267–271. Also at
with 1.0 ≤ Fa ≤ 2.0, Fa being a factor that can be adopted
www.rocscience.com/education/hoeks_corner
by the engineer, depending on sample and rock mass dis- Hoek, E. & Marinos, P. 2007.A brief history of the development of the
turbance and which shall be examined for each particular Hoek-Brown failure criterion- Soils and Rocks, No. 2. November
case 2007. Also at www.rocscience.com/education/hoeks_corner
I- With this proposed value for σci , and for a typical case, Horn, H.M. & Deere, D.U. 1962. Frictional characteristics of
it was found that for Fa = 1.0 the strength envelope, both minerals. Géotechnique 12(4), 319–335.
for rock cores and rock masses, is less than 5% higher ISRM (1978) – Suggested methods for determining tensile strength
than that deduced with s = 1. For Fa = 2.0 the strength of rock materials- Int. J. Rock Mech. Min. Sci. & Geomech. Abstr.
envelope is about 25% to 30% higher than that deduced Vol. 15, pp. 99–103. Also available at www.isrm.net
with s = 1. Mogi, K. 2007. Experimental rock mechanics. 358pp. Taylor &
Francis/Balkema – London.
J- The detailed study of the effects of sample disturbance on
Oxford Concise Dictionary (2008). Oxford University Press. http://
rock strength envelopes deserve further research. oxforddictionaries.com/
K- Care should be taken, especially for weak rock masses, Skempton, A.W. 1961. Effective stress in soil, concrete and rocks.
to identify the point of brittle-ductile transition, beyond Pore pressure and suction in soils, pp. 4–16. Butterworths, London
which Hoek-Brown criterion is not applicable and which Zhang, L. 2005. Engineering Properties of Rocks. Elsevier
could greatly underestimate rock mass strength. Geo-Engineering Books Series Volume 4. Elsevier.
89

Discontinuum models for dam foundation failure analysis
J.V. Lemos
LNEC – Laboratório Nacional de Engenharia Civil, Lisbon, Portugal
ABSTRACT: The safety assessment of concrete dam foundations entails the examination of the potential failure mechanisms,
typically defined by natural rock discontinuities or the concrete-rock interface. Numerical models which represent the rock
mass as a discontinuous medium, in particular discrete element models, are particularly adequate for the analysis of these
failure scenarios, given their ability to represent the geologic structure of the rock mass, as well as the concrete structure.
The application of discrete element deformable block models to the safety assessment of arch and gravity dam foundations is
discussed with reference to specific examples. The issues involved in model generation, such as the representation of the rock
discontinuities or the application of joint water pressures, are examined, as well as the procedures for safety factor evaluation.
Subject: Modeling and numerical methods
Keywords: Rock slopes and foundations, Stability analysis, Numerical modeling
1 INTRODUCTION foundation failure analysis. It is not intended to represent in

detail the rock mass jointing, but instead employs a relatively
The design of new concrete dams, or the safety assessment coarse block structure effectively directed at the assessment
of existing dams, requires particular attention to be paid to of safety with respect to specific deformation and collapse
the behavior of the foundation rock mass. Surveys promoted modes.
by ICOLD and other institutions have shown that many of The paper outline is the following. The essential concepts
the incidents or deficiencies experienced by concrete dams of discontinuum modeling are briefly addressed in the next
are linked to the foundation. The accident of the Malpasset section. Then, the main issues involved in the application
arch dam is the best known example of a structural collapse of DE block models to arch dam foundations are examined,
caused by sliding on rock discontinuities (e.g. Londe 1987). namely model generation, representation of jointing, water
In fact, this event motivated a substantial body of research pressure fields, and safety factor calculations. The discussion
on rock foundations issues, namely on the hydro-mechanical of these items will be illustrated by several examples of dam
behavior, and stimulated the development of new methods of studies performed with the code 3DEC (Itasca 2006). Finally,
safety assessment. Analytical or graphical methods, such as some topics related to gravity dam failure analysis will be
those proposed by Londe (1973), became standard tools in addressed.
arch dam design. With these techniques, it became possible
to analyze the potential failure of rock blocks defined by the
rock mass discontinuities, considering the installed water pres- 2 DISCONTINUUM MODELING
sures and the dam loads. More recently, Goodman & Powell
(2003) applied Shi and Goodman’s Block Theory to identify An engineering model is necessarily a simplification of the
moveable blocks in concrete dam foundations. All these tech- physical reality, more often intended to answer a specific ques-
niques, based on simple block mechanics, remained important tion, e.g. about safety or performance of a proposed design
for safety evaluation, while the finite element models became or an existing structure, rather than to provide a meticulous
the preferred tools to analyze dam foundations under operat- description of nature.The amount of detail to be included in the
ing conditions and to predict stresses and displacements (e.g. model is dictated by the purpose of the analysis, e.g. stability
Wittke 1990). assessment or interpretation of monitoring data in operating
Discrete element (DE) models are widely used in rock conditions, and is always limited by the experimental data
mechanics. As will be discussed in the following section, they available.
can be used either at the micro or meso-mechanical scale, e.g. Two fundamental options exist for the representation of a
to investigate fracture phenomena, or at the engineering scale, jointed rock mass: (i) the equivalent continuum approach, in
e.g. to analyze a full arch dam foundation. In this paper, the which a continuum constitutive model is employed to repre-
latter perspective is adopted. In its simplest form, a DE rigid sent in an average manner the effects of the discontinuities;
block model may be viewed as a numerical tool that performs (ii) the discontinuum approach, in which the discontinuities
the same stability analysis as Londe’s method. However, not are explicitly represented individually. Both of these idealiza-
just statics are involved, but a full mechanical analysis is under- tions have their fields of application, and often a combination
taken. A DE deformable block model, with internal meshes of the two is advisable: a number of key discontinuities are
in the blocks, is capable of stress and displacement analy- modeled explicitly, while the others are lumped into the block
sis as a finite element model, while retaining the ability to behavior.
simulate in a straightforward manner failure modes defined The finite element method is the most common tool for
by the rock discontinuities. This is the type of model that equivalent continuum analysis, but it is also capable of
will be examined in this paper as an engineering tool for dam addressing discontinuous models by means of joint, interface
91

the large displacement range, with automatic update of block
connectivity.
Safety requirements for dams cover both operating and
extreme conditions (e.g. Pedro 1995). The monitoring of con-
crete dam foundations over the years has produced extensive
databases to validate and calibrate numerical models for con-
ditions of normal operation. It is much more difficult to be
confident about the models’ ability to evaluate failure sce-
narios. Back analysis of accidents is an important test, but
laboratory experiments with physical models remain a valu-
able source of information (e.g. Gomes 2006, Fei et al. 2009).
Benchmark comparisons between different codes are also very
helpful, namely to examine numerical and implementation
Figure 1. DE block model of Baixo Sabor arch dam. issues.
or gap elements (e.g. Alonso et al. 1994). The development 3 ARCH DAM FOUNDATION ANALYSIS
of the distinct element method by Cundall in 1971 was aimed
at a direct representation of a discontinuum. Presently, the 3.1 The case of Baixo Sabor dam
designation of “discrete elements” (DE) covers a wide family
of numerical methods (distinct elements, discrete finite ele- The main topics involved in the application of DE models to
ments, DDA, etc.), all sharing the concept of representing a the analysis of failure mechanisms in arch dam foundations
discontinuous medium as an assembly of blocks or particles. will be discussed resorting to a few examples, in particular to
These idealizations are applicable at various scales of analy- Baixo Sabor dam, presently under construction. The Hydro-
sis. At one end, we have the micromechanical models based on electric Project of Baixo Sabor, owned by EDP, is located in
many rigid particles or blocks employed, for example, in frac- the north-east of Portugal in the lower branch of the Sabor
ture analysis at lab test scale. At the other end, the deformable river, a tributary of the right bank of the Douro river. It is
block models of large civil and mining engineering works. composed of two dams, a 123 m high arch dam upstream, and
Although the increase of computer power is expanding the a 45m high gravity dam downstream. Both powerhouses will
range of the micro-models (Cundall 2001), it is the latter type have reversible units to enable pumping from the Douro river
of DE model that is most suitable for dam foundation analysis to the reservoir created by the upstream dam with a capacity of
(Fig. 1). 1.1 × 106 m3 . The design was performed by EDP (Matos et al.
The analysis of collapse mechanisms in dam foundations 2007). The arch dam has a crest length of 505 m and a total
involves the representation of the discontinuities where sliding concrete volume of 670000 m3. The 3DEC model for foun-
may take place. In this field of application, the block structure dation safety assessment, shown in Figure 1, is described in
thus defined is better replicated in a numerical model by means more detail by Lemos & Antunes (2011). The full model com-
of deformable blocks. In this way, a more realistic simulation prises about 2300 deformable blocks with 26100 grid-points
of the distribution of structural loads is obtained, influenced by (the internal block mesh is not shown in the figure).
the foundation properties and their spatial variation, even with
a fairly coarse block system. In the code 3DEC (Itasca 2006),
3.2 Model generation
deformable blocks are obtained by internal discretization into
a finite element mesh of tetrahedra. For dam foundation stud- Various techniques and software are now available for the gen-
ies, these rock blocks are typically assumed elastic, with all eration of numerical models, with most analysis codes capable
the nonlinear behavior concentrated on the joints. For arch of using a combination of built-in tools and data input from
dams, the correct bending behavior is more easily achieved external geometric modelers. For dam foundation analysis
with higher order elements, thus 3DEC allows 20-node bricks with the code 3DEC, the model generation typically starts by
to be used for the concrete vault. In this type of model, the reading a FE mesh of the dam. The mesh of the Baixo Sabor
vertical contraction joints and the concrete-rock interface are dam model is plotted in Figure 2, with the contraction joints
also discontinuities which may be assigned general constitu- separating the structure into the cantilever blocks. This mesh
tive models. The Mohr-Coulomb model is the most widely consists of higher order brick elements (in the plot the faces
used, but many other rock joint models exist. are triangulated). More than one element across the thickness
The variety of numerical techniques presently available is normally recommended to provide a finer discretization of
often brings the question of what are their real differences, the foundation surface. Block interaction mechanics in most
e.g. how does a DE deformable block model differs from a FE DE codes is based on sets of point contacts, rather than inter-
model with joint elements. If both models share the constitu- face elements, which allows the contact between blocks with
tive assumptions regarding block material and joint behavior, unmatched meshes, as happens in the present case along the
then their response should be not be dissimilar. It is mainly foundation surface (e.g. Lemos 2008).
the numerical approach that sets them apart. FE models rep- The global model geometry may be created, in the simplest
resent block interaction by means of joint elements, while DE case, by extending the model upstream and downstream from
models typically use point contacts. FE packages favor matrix the dam-rock interface, assuming a cylindrical valley shape,
and implicit solvers, while DE codes obtain static solutions as shown at the top of Figure 3. Alternatively, the actual sur-
by dynamic relaxation. The same explicit algorithm, but with face topography may be input, as a new layer of blocks placed
real values of damping, is employed in time domain dynamic above, as shown at the bottom of the same figure. The model
analysis. Most standard FE analyses assume geometric linear- in the top figure may be used for the analysis of sliding mecha-
ity, while DE codes are designed to extend the solution into nisms on the dam-rock interface, in the simplified scenario of
92

Figure 2. FE mesh of concrete arch.
Figure 4. Location of main faults and dykes.
Figure 3. Model geometry: simplified valley shape (top) and model

with terrain topography (bottom).
Figure 5. DE block model of Baixo Sabor arch dam. Views of right

a continuous foundation, but considering the cantilever block and left bank halves of model.
structure of Figure 2.
In the case of Baixo Sabor dam, 3 main sets were identified
in the granitic rock mass, one sub-horizontal and two sub-
3.3 Representation of rock mass discontinuities
vertical. The model in Figure 1 includes the main faults found
3.3.1 Selection of discontinuities at the site, mostly sub-vertical, whose traces are shown in
The models in Figure 3 simply define the rock mass geometry Figure 4. Only a few joints were selected to represent each set.
in terms of a set of rigidly attached or joined blocks. The next Figure 5 shows, separately, the right ad left banks of the same
phase in the model generation is the representation of the rock model, where the slight upstream dip of the sub-horizontal set
discontinuities. It is at this stage that critical decisions need to is visible.
be made about what features to include and how to represent It should be pointed out that in most figures in this paper, the
them in a necessarily simplified manner. Typically, there are different rock blocks in the assembly are represented by dif-
several faults or other major features identified at the site (e.g. ferent colors. These polyhedral blocks are formed by joining
Fig. 4), which may be inserted at their known locations with convex sub-blocks, with the construction lines also visible.
given orientations. Then, each of the most significant joint sets The mesh of tetrahedral elements inside each deformable
is represented by a few selected joint planes. The purpose of block is not depicted, for clarity.
the analysis, stability assessment, directs the selection of the It may be seen in these figures that the blocky structure only
number and location of joints, as the intention is not to recreate exists in part of the model. Below and to the sides, beyond the
in detail the joint structure, but to identify the possible failure reach of the possible failure modes, large deformable blocks
modes and their likelihood. In practice, a small number of were used. Given the relatively small overall dimensions of
joints from each set is normally sufficient to define the most the model, the boundary conditions applied at the vertical
relevant mechanisms. Valuable lessons can be drawn from the upstream and downstream boundaries have to be carefully
clear rationale manifest in the classical papers (e.g. Londe chosen. It is important to use a stress boundary, applying the
1973). in situ stress state in the rock mass, instead of the common
93

displacement boundary conditions, to avoid an artificial con-
straint of upstream-downstream movements. It is enough to
fix displacements at the base and sides of the model.
In addition, the discontinuities were also not extended
upstream. Besides the computational savings, this is a con-
servative simplification which prevents the upstream rock to
constrain block movements, as often tensile stresses develop
at the dam upstream heel. In fact, in this model, a vertical
joint was placed along the upstream edge of the concrete-rock
interface, which accounts for the expected rock joint opening
in this tensioned area. In the case of this dam, a sub-vertical
joint set normal to the river axis is actually present, so this
simplifying assumption is entirely justified. This upstream
joint is assigned rock joint properties, thus no tensile strength,
and water pressures given by the full reservoir conditions are
applied. Therefore, part of this joint may separate if the normal
stresses become tensile, effectively decoupling the top layer
of the upstream rock mass.
3.3.2 Use of multiple models
The need to avoid excessive complexity of numerical repre-
sentations has been often stressed (e.g. Starfield & Cundall
1988). The computational advantages of limiting the num-
ber of blocks are today less critical, except perhaps for time
domain dynamic analyses. However, the time and labor sav-
ings in model generation, model verification and, particularly,
in the interpretation of results may be significant.
For dam foundation studies, it is frequently a good option
to build several models of the same rock foundation to check
different failure modes than to try to include every aspect of Figure 6. DE block models of Alto Ceira dam, for analysis of right
behavior into a single complicated representation. For exam- and left bank failure modes.
ple, it may be possible, and actually more instructive, to study
failure modes on each abutment separately. This was done for
Alto Ceira dam, a 41m high arch dam under construction by
EDP, where the jointing was complex and with different orien-
tation in each valley side (Lemos & Antunes 2011). In fact, 3
different models were built for the right bank, each one com-
bining 2 joint sets capable of forming potential failure wedges
under the dam. Figure 6 shows one of these models at the top.
At the bottom, the left bank model, in which 3 joint sets were
considered. As the critical failure wedges were in the vicinity
of the dam, joints were only inserted in a limited region.
Fairly detailed representations of jointing are possible
whenever necessary, as shown in Figure 7, in the model devel-
oped by N.S. Leitão to study the stability of the left abutment of
FozTua arch dam (Matos et al. 2011). In this figure, the vertical
discontinuity that limits upstream the blocky representation is
visible.
3.3.3 Persistence of discontinuities
The representation of non-persistent joint sets in stability
calculations still poses some difficulties. Work on joint gen-
eration packages has been mostly directed towards fluid flow
problems, where network connectivity is more important than Figure 7. Detail of block model for the analysis of failure of the left
partition into blocks. In failure studies, the most conservative abutment of Foz Tua dam (dam and upstream rock hidden).
option is to disregard the shear strength of the rock bridges,
assuming the joints to be continuous. If an acceptable safety
margin can be ensured, then there is no need of more elaborate joint plane. In this way, the potential the failure of rock bridges
models. If this is not the case, the non-persistence of the joint and the coalescence of the cracked sections may be contem-
planes has to be taken into account. If blocks are assumed plated, but joint constitutive models with consistent fracture
elastic, then a joint generator that creates non-persistent joint criteria need to be employed (e.g. Resende et al. 2004). For
patterns may be unconservative, as small rock bridges may complex fracture patterns, DE bonded particle models become
prevent the development of a failure mode. A simpler alterna- the most powerful approach, also able to allow fracturing to
tive is to create through-going planar cuts, and simulate rock proceed through the rock blocks, for example, the “synthetic
bridges by assigning cohesive strength to some sections of the rock mass” concept presented by Pierce et al. (2007).
94

3.4 Block and joint deformability
Modeling the foundation with deformable blocks allows
stresses to be evaluated in the rock blocks, but, more impor-
tantly, provides a better approximation of the distribution of
loads applied by the dam in cases of asymmetrical or hetero-
geneous rock mass moduli. The patterns of load redistribution
after incipient slip are also better judged.
In rigid block models, the overall rock mass deformability
is governed only by the joint stiffness. In a deformable block
model, both block moduli and joint stiffness are specified.
If the real joint spacing were used, then the actual stiffness
parameters would be appropriate. However, in a large model
of a dam foundation only a few joints may be included, as
already discussed. Therefore, the effect of joint stiffness on
the overall deformation is small, unless, for example, a thick
gouge fault is present. The block moduli need to be selected to
provide the correct rock mass deformability for each region.
In those cases where the blocky structure only exists in part
of the model (e.g. Fig. 6), the effect of the selected joint stiff-
nesses has to be estimated, so that the moduli assigned to
the blocks compensate any imbalance due to the numerical
options. As is to be expected, in these stability analyses joint
strength parameters are the most decisive properties. Figure 8. Detail of 3DEC model of Alqueva dam for analysis of
water flow (Farinha et al. 2011).
3.5 Joint water pressures
The water pressures in the discontinuities are a critical factor A deformable block formulation implies the generation of an
in stability analysis. In the assessment of existing dams, mon- internal element mesh in each rock block. For polyhedral block
itored piezometric data is very helpful in the calibration of shapes the simpler option is to use tetrahedra, as 3DEC does,
numerical models. For new dams, however, simplified water which can easily be automatically generated. DE codes typi-
pressure distributions are usually assumed, according to stan- cally have routines to detect and update contacts during a large
dard design practices. It is possible to carry out fluid flow displacement analysis, and these are also invoked to identify
analysis with DE models, and several studies have been pre- the initial contact between the blocks, without user interven-
sented for gravity dams (e.g. Lemos 1999, Barla et al. 2004, tion. Contact formulations vary among the different codes, but
Gimenes & Fernández 2006). The code 3DEC also allows the most common option is to create contact points at every
an analysis of fluid flow in the rock joints to be performed vertex-to-face or edge-edge interaction (e.g. Lemos 2008).
(Damjanac & Fairhurst 2000), but there is often not enough When a block has a fine tetrahedral mesh, grid-points exist
information to undertake such studies at design stage. In addi- within the original rigid block faces, which are also treated as
tion, a fracture flow analysis requires a network with many new vertices.
more joint planes than those that are necessary for a failure Material properties must then be assigned to concrete and
analysis. Moreover, the grout curtain and the drainage sys- rock blocks and joint properties prescribed for all discontinu-
tem complicate the flow patterns in the vicinity of the dam. ities. For systems with complex joint patterns, it is necessary
A more practical alternative is to perform an equivalent con- to verify very carefully if the correct constitutive assumptions
tinuum flow analysis to calculate the field of water pressures and properties were assigned to each joint set, as this is the
throughout the rock mass. 3DEC may also be employed in such critical factor in the study.
equivalent continuum flow analysis, as in a model of Alqueva
dam (Fig. 8), where the rock permeabilities were calibrated
by monitored pressure and drainage data for various reservoir 3.6.2 Modeling steps
levels (Farinha et al. 2011). The water pressures obtained in The analysis procedure comprises a sequence of modeling
such analyses may then be applied in the joints of the block steps, which should as much as possible follow the physical
model used in the mechanical failure study. path. The first step corresponds to the in situ condition, before
In the model of Baixo Sabor dam, simplified water pressure dam construction. In situ stress measurements may provide
distributions were considered. Along the concrete-rock inter- an estimate of horizontal stresses, but any prescribed initial
face, uplift pressures were prescribed according to the usual stress field must then be brought to equilibrium under gravity,
design criterion, a bilinear diagram with 1/3 of the reservoir ending in a state compatible with the valley shape.
head at the drain location. In the rock discontinuities, the full The second modeling step is the simulation of dam con-
reservoir head was considered upstream whereas downstream struction, in which gravity is applied to the cantilever blocks.
a simplified pressure field was prescribed, defined in terms The most realistic procedure is typically to assume indepen-
of a water table compatible with the valley slopes. dent cantilevers, and, at the end, impose the closure of the
contraction joints, zeroing joint displacements. The next step
3.6 Modeling sequence is the reservoir filling, with the application of the hydro-
static pressure to the dam upstream face. The water pressures
3.6.1 DE model setup in the discontinuities are also introduced, as discussed in
The definition of the model geometry and the introduction of the previous section. Then, the safety assessment procedure
the rock discontinuities, as discussed, create the block system. follows.
95

Figure 10. Baixo Sabor model. Evolution of displacement indica-
tors with rock joint friction reduction factor.
Figure 9. Baixo Sabor dam model. Displacement field for a strength
reduction factor of 2.
3.6.3 Safety evaluation

The methodology adopted for safety evaluation must satisfy
regulatory requirements, a comprehensive survey and discus-
sion being given in ICOLD European Club (2004). In the
studies presented in this paper, safety factors were calculated
by means of a strength reduction procedure. The shear strength
of the discontinuities, where nonlinear behavior is concen-
trated, is divided by progressively larger factors until collapse
takes place or displacement magnitudes reach unacceptable
levels. In the rock joints and faults, assumed to be purely fric-
tional, the reduction factor was applied to the tangent of their
friction angles. The treatment of concrete-rock interface varies
with national regulations. For example, the Portuguese code
requires that a condition of no cohesion or tensile strength be Figure 11. Alto Ceira model. Evolution of displacement indicators
checked for failure scenarios. with rock joint friction reduction factor.
In the case of Baixo Sabor dam (Figs. 1–5), shear tests
of rock joints led to friction angles of 37◦ to 39◦ for the 3
main sets. Faults were assigned a friction angle of 35◦ , and advisable to expect a complete collapse of the block model,
the rock-concrete interface a value of 45◦ . The orientation as it is possible that equilibrium is reached with displace-
of the discontinuities provided a comfortable safety margin. ments unacceptable for the integrity of the concrete structure,
The development of a failure wedge under the right abutment, or for the continued effectiveness of the grout and drainage
shown in Figure 9, was only obtained for a friction reduction curtains.
factor of 2.
Figure 10 contrasts the evolution of 3 displacement indica- 4 CONCRETE GRAVITY DAM ANALYSIS
tors during the friction reduction process. The displacement
of a grid-point in a rock block under the right abutment (solid 4.1 Sliding failure modes
line) starts increasing at a reduction factor of about 1.6, a
trend displayed more clearly by the curve corresponding to For concrete gravity dams, the most common failure scenarios
the maximum shear displacement of any rock discontinuity to be evaluated involve sliding on the rock-concrete interface
in the model (short dash line). The lower curve (dashed line) or on shallow sub-horizontal rock joints. The common practice
represents the maximum shear displacement at the concrete- is to analyze the stability of individual dam blocks, assumed to
rock interface, assumed cohesionless in this run, also follows move in the upstream-downstream direction, but 3D analysis
the trend, but less expressively. These results show that slip on may sometimes be advisable, as discussed below. At the valley
the rock wedge surfaces may not be immediately noticeable slopes, 3D collapse modes, as considered for arch dams, may
on the structural displacements. be possible. DE models allow the evaluation of all these failure
The Alto Ceira right bank model (Fig. 6), also displays a scenarios, whether for static or seismic loads. They may also
significant safety margin, but a different type of response be employed to check the safety against sliding on the dam
is illustrated by the corresponding curves. The displacement horizontal lift joints, also a concern for concrete gravity dams.
indicators in Figure 11 show that the joint slip starts affect- In all of these calculations, the water pressure distribution has
ing dam displacements when the reduction factor exceeds a crucial influence on the outcome.
1.4, and a clear acceleration can be seen above 1.7. In con-
clusion, for a correct interpretation of the numerical model
4.2 2D vs. 3D analysis
output, it is important to look at multiple indicators, namely
the block movements and joint slip at several points, to detect The analysis of sliding failure modes of gravity dams is usu-
the onset and progression of the failure modes. Often, it is not ally performed in 2D, assuming the dam blocks to behave
96

independently, thus neglecting the possible contribution of
shear keys. This is sometimes even done for dams with a
slight curvature in plant, in which the extra safety provided
by the arch is not taken into account. The 3D effect, how-
ever, is clearly present in the case of narrow valleys, even for
straight axis dams. In studies of safety re-evaluation of older
dams, if the shear keys are capable of providing a monolithic
behavior, this extra contribution may be valuable (e.g. ICOLD
European Club 2004).
When sliding on rock joints is also analyzed, the 3D nature
of rock mass structure always exerts some influence. Even
if the movement of a single dam block can only take place
in the upstream-downstream direction, the water flow pat-
tern, for example, is less well captured by the plane analysis
assumptions.
Analytical methods, based on limit equilibrium techniques,
have also been proposed for 3D gravity dam analysis (e.g.
Lombardi 2007; Sun et al. 2010). These simplified techniques
are more often applied to the study of sliding on the dam-
foundation interface. DE numerical models not only provide
a full mechanical analysis tool, but are also more versatile if
the rock mass joints need to be included.
4.3 Seismic analysis

Seismic action is always a major concern in sliding failure
scenarios for gravity dams. The shortage of historical data on
the response to large seismic events makes experimental tools
of great value in the validation of numerical models. Figure 12
show a model of a concrete dam monolith tested on LNEC’s
shaking table (Gomes 2006). A blocky foundation was created
by 2 joint sets, allowing several mechanisms to be simulated,
by means of changing the friction on the various joints, and Figure 12. Gravity dam model on blocky foundation for shaking
locking or not the joint representing the dam-rock interface. table test of (top); permanent joint displacements after test (bottom)
The test results were then analyzed by a 3DEC model with (Gomes 2006).
deformable blocks.
Often, the seismic stability of a gravity dam is analyzed by acceptable levels of displacement, and their potential effects
very simple 2-block models, in which the dam is represented on structural safety (Alliard & Léger 2008).
by a rigid block and the rock mass by another with a prescribed
acceleration record. It should be emphasized that this is a very
4.4 Modeling and numerical issues
crude model, acceptable only as a first check on the likelihood
of slip under the earthquake action. If this model indicates that The application of DE deformable block models to the fail-
slip is possible, then a detailed numerical model needs to be ure analysis of gravity dam foundations follows, in essence,
built, with proper boundary conditions for seismic analysis the methodology outlined for arch dams. A 2D model is much
(Lemos 1999, 2008). The dam rocking modes need to be cor- easier to create, and faster to run, particularly in dynamic anal-
rectly represented, as well as the foundation deformability and ysis. Therefore it is almost always a good starting point, even if
joint water pressure distribution, which greatly influence the a 3D model becomes necessary for more rigorous predictions.
sliding behavior. For sliding on the dam-rock interface, planar models are
The evolution of the joint water pressures during a dynamic usually sufficient, and the key role is played by uplift pres-
event is another important topic, and experimental results such sures. When assessing existing dams, piezometric readings
as those obtained by Javanmardi et al. (2005) are very valu- provide essential information. In 2D models, the analysis of
able. Dynamic numerical simulations (e.g. Lemos 1999) have water flow in the fractures poses no computational difficulty,
indicated that keeping the steady-state pressure field in the unlike the case of arch dams previously discussed. Various
joints, thus not taking into account the dynamic water pres- studies with DE codes have been published (e.g. Lemos 1999,
sure variations, may be a conservative assumption, leading Barla et al. 2004, Gimenes & Fernández 2006). Of course,
to larger slip predictions. This issue needs experimental con- the questions about the suitability of the fracture networks
firmation, but it would facilitate seismic analysis, especially remain, as well as those on the reliability of in situ joint con-
since the effective water stiffness depends on joint apertures, ductivity data. It is known from various numerical studies
always difficult to characterize. that flow simulations are much more dependent on the chosen
For large design earthquakes, the simplifying assumptions jointing patterns, and exhibit more scattering of results, than
imposed by some regulations, such as a cohesionless dam- mechanical stability calculations.
rock interface, lead to significant permanent displacements. The explicit solutions algorithms employed by most DE
A consistent estimate of these displacements is required, and codes have proved fairly robust for strongly nonlinear condi-
further investigation is needed on criteria for the definition of tions, as experienced during failure analysis of blocky media.
97

For dynamic analysis, these algorithms call for small time Habib & De Gennaro (eds) La sécurité des grands ouvrages,
steps, making large 3D systems still a computational chal- Paris:Presses Ponts et Chaussées, 5–16.
lenge. An experienced user, however, aware of the specific Farinha, M.L.B., Lemos, J.V. & Maranha das Neves, E. 2011. Numer-
numerical stability constraints implemented in the code, is ical modelling of borehole water-inflow tests in the foundation of
the Alqueva arch dam. Can. Geotech. J., 48(1): 72–88.
often capable of building the model in a way that achieves
Fei, W.-P., Zhang, L., Zhang, R. 2009. Experimental study on a geo-
substantially reduced run times. mechanical model of a high arch dam. Int. J. Rock Mech. Min. Sci.,
The comments made above regarding the need for con- 47:299–306.
sistent fracture criteria in the analysis of rock bridges also Gimenes, E. & Fernandez, G. 2006. Hydromechanical analysis of
apply to the study of the concrete-rock contact. A good deal of flow behavior in concrete gravity dam foundations. Can. Geotech.
research has been published on this subject, either using frac- J., 43:244–259.
ture mechanics approaches or slip-weakening joint models, Gomes, J.M.N.P. 2006. Experimental analysis of failure scenarios of
namely at ICOLD’s numerical benchmark workshops. concrete dam foundations – Static and dynamic tests. PhD Thesis,
Lisbon:LNEC (in Portuguese).
Goodman, R.E. & Powell, C. 2003. Investigations of blocks in foun-
5 CONCLUSIONS dations and abutments of concrete dams. Journal of Geotechnical
and Geoenvironmental Engineering, ASCE, 129(2):105–116.
The numerical models presented provide a powerful tool to ICOLD European Club 2004. Working group on sliding safety of
address the safety assessment of structural foundations on existing dams. G. Ruggeri (ed.), Final Report.
Itasca 2006. 3DEC – Three-Dimensional Distinct Element Code,
rock. The current DE models based on deformable block for-
Version 4.0, User’s Manual. Itasca Consulting Group, Minneapolis,
mulations have all the features required to analyze arch or USA.
gravity concrete dam foundations, as discussed in the previous Javanmardi, F., Léger, P. & Tinawi, R. 2005. Seismic structural stabil-
sections. At a first level of analysis, they may substitute tra- ity of concrete gravity dams considering transient uplift pressures
ditional techniques of limit equilibrium, overcoming specific in cracks. Engineering Structures, 27:616–628.
restrictive assumptions. For this purpose, the natural tendency Lemos, J.V. 1999. Discrete element analysis of dam foundations, In
to overelaborate a numerical representation should be resisted. V.M. Sharma, K.R. Saxena, R.D. Woods (eds) Distinct Element
For more complex applications, in order to take advantage of Modelling in Geomechanics. Rotterdam: Balkema, 89–115.
all the capabilities of these numerical models, it is necessary Lemos, J.V. 2008. Block modelling of rock masses – Concepts and
to have enough reliable experimental data. application to dam foundations. European Journal of Environmen-
tal and Civil Engineering, 12(7-8): 915–949.
An effective use of these models in safety assessment
Lemos, J.V. & Antunes, N.S. 2011. Modelling of arch dam foundation
requires continued investigation on a range of issues, extend- failure scenarios – Case studies of Baixo Sabor and Alto Ceira
ing from the strictly numerical aspects, such as reducing the dams. In C. Pina, E. Portela & J.P. Gomes (eds), Proc. 6th Int.
computational cost of large systems, to the more essential Conf. on Dam Engineering, Lisbon: LNEC.
matters of representing rock and joint behavior in ultimate Lombardi, G. 2007. 3D analysis of gravity dams. Hydropower &
scenarios. Among those issues, a few may remain central in Dams, Issue 1:98–102.
dam foundation failure studies: practical joint generation tech- Londe, P. 1973. Analysis of the stability of rock slopes. Q. Jl Engng
niques specifically intended for stability analysis; a consistent Geol. 6:93–127.
and dependable treatment of rock bridge or dam-rock inter- Londe, P. 1987. The Malpasset Dam failure. Proc. Int. Workshop on
face fracture; estimation of joint water pressure distributions Dam Failures, Purdue, Engineering Geology, 24:295–329.
Matos, D.S., Paixão, J. & Antunes, N. 2007. Arch design of the Baixo
providing a reasonable envelope for specific in situ conditions.
Sabor upstream scheme. In C. Pina, E. Portela & J.P. Gomes (eds)
Proc. 5th Int. Conf. on Dam Engineering, Lisbon:LNEC.
ACKNOWLEDGEMENT Matos, D.S., Paixão, J. & Lemos, J.V. 2011. The Foz Tua arch dam
design. In C. Pina, E. Portela & J.P. Gomes (eds), Proc. 6th Int.
Conf. on Dam Engineering, Lisbon: LNEC.
The author acknowledges permission granted by EDP, Ener- Pedro, J.O. 1995. Arch dams – Safety and performance evaluation.
gias de Portugal, to present the dam analysis data. Udine:International Centre for Mechanical Sciences.
Pierce, M., Cundall, P., Potyondy, D. & Mas Ivars, D. 2007. A
Synthetic Rock Mass Model for Jointed Rock. In E. Eberhardt
REFERENCES et al. (eds) Rock Mechanics: Meeting Society’s Challenges and
Demands (1st Canada-U.S. Rock Mech. Symp., Vancouver, May
Alliard, P.-M. & Léger, P. 2008. Earthquake safety evaluation of grav- 2007), vol. 1, London: Taylor & Francis, 341–349.
ity dams considering aftershocks and reduced drainage efficiency. Resende, R., Lemos, J.V. & Dinis, P.B. 2004. Application of a discon-
Journal of Engineering Mechanics – ASCE, 134(1):12–22. tinuity model with softening to the analysis of dam foundations
Alonso, E.E., Gens, A., Carol, I., Prat, P. & Herrero, E. 1994. Three- using the discrete element method. In H. Konietzky (ed.) Num.
dimensional failure mechanisms in arch dam abutments – A safety Modelling of Discrete Materials in Geotechnical Eng., Rotterdam:
study. Proc. 18th ICOLD Congress, Durban, vol. 1, 471–484. Balkema, 249–255.
Barla, G., Bonini, M. & Cammarata, G. 2004. Stress and seepage Starfield, A.M. & Cundall, P.A. 1988. Towards a methodology for
analyses for a gravity dam on a jointed granitic rock mass. In rock mechanics modelling. Int. J. Rock Mech. Min. Sci., 25(3):
H. Konietzky (ed.) Numerical Modeling of Discrete Materials in 93–106.
Geotechnical Engineering, 263–268, Rotterdam:Balkema. Sun, G.-H., Zheng, H. & Liu, D.-F. 2010. A three-dimensional proce-
Cundall, P.A. 2001. A discontinuous future for numerical mod- dure for evaluating the stability of gravity dams against deep slide
elling in geomechanics? Proc. Inst. Civil Engineers, Geotechnical in the foundation. Int. J. Rock Mech. Min. Sci., 48:421–426.
Engineering, 149(1): 41–47. Wittke, W. 1990. Rock Mechanics.Theory andApplications with Case
Damjanac, B. & Fairhurst, C. 2000. Ecoulement tri-dimensionnel Histories. Berlin: Springer-Verlag.
d’eau sous pressions dans les millieux fracturés. In Delage,
98

Discontinuous deformation analysis: Advances and challenges
G.W. Ma
School of Civil and Resource Engineering, University of Western Australia, Crawley WA, Australia
ABSTRACT: State-of-the-art of analysis methods of discontinuous deformation is summarized. In addition to comparing

different numerical methods in simulation of discontinuous deformations in rock mechanics and engineering, the main focus
of this paper is to introduce the numerical manifold method including its advances and challenges in rock mass modeling.
Recent development of the numerical manifold method on rock failure and slope stability analysis is highlighted. A 3-D rock
mass geological tool has been developed to generate the rock mass structures. A 3-D program of numerical manifold method
has also been developed and applied to simulate rock slope collapse. Demand of a more realistic rock mass modeling method
is discussed. A few suggestions are made toward realistic rock mass modeling. These require development of a more realistic
and probability based geological rock mass model, adopting realistic constitutive relations to intact rock and rock joints, use of
realistic numerical model to technically handle numerical difficulties, and achievement of realistic and affordable computational
cost, etc.
Subject: Modeling and Numerical Methods
Keywords: Numerical modeling; Rock mass; Stability analysis; Rock failure; Rock joints.
1 STATE OF THE ART OF NUMERICAL METHODS IN nonlinearity, thus is more suitable for solving problems in
DISCONTINUOUS DEFORMATION ANALYSIS homogeneous and linearly elastic bodies.
The finite element method (FEM) has been the most popu-
Rock mass, as a typical natural geological material, are com- lar numerical method in rock mechanics and rock engineering
plex with different extent discontinuities including bedding due to its flexibility in handling material heterogeneity and
planes, faults, shear zones, joints and fractures. To explic- anisotropy, complex boundary conditions and dynamic prob-
itly characterize such mechanical discontinuities, various lems, and efficiency in dealing with complex constitutive
numerical methods have been developed. These methods are models and fractures. Representation of fractures in the FEM
generally classified into three categorizes, i.e., continuum has been motivated since the late 1960s. Various joint element
based methods (e.g., the FDM, the BEM, the FEM, the mesh- or interface element models such as ‘Goodman joint element’
less methods, etc), discontinuum based methods (e.g., the (Goodman et al., 1968) have been developed. Despite these
distinct element method (DEM), the discontinuous deforma- efforts, explicitly modeling fractures and their growth remains
tion analysis (DDA)), and combined continuum-discontinuum limited in the FEM. In the FEM, the finite element mesh must
methods. be constructed in such a way that element edges/faces coin-
The finite difference method (FDM) is the earliest mem- cide with the crack surfaces and nodes must be placed on each
ber in the family of numerical methods. The implementation side of the crack to allow material separation. The construction
of the FDM is simple in both two-dimensional and three- of such meshes becomes more difficult, especially in three-
dimensional cases. However, the conventional FDM using dimensional cases. When discontinuity evolution needs to be
regular grid system suffers from inflexibility in dealing with modeled, remeshing is inevitable, which makes the simulation
material heterogeneity, fractures, and complex boundary con- tedious and time-consuming. Moreover, during remeshing
ditions, which restricts its application in rock mechanics. process, variables such as displacement, stress and strain need
In order to overcome such shortcomings, the finite volume to map to a new set of nodes and quadrature points, which may
method (FVM) was developed, which allows for the use of result in additional inaccuracies.
unstructured meshes, and is flexible in non-linear material In order to overcome such inconveniences in meshing and
models and boundary condition enforcement. Explicit repre- remeshing processes, a variety of modifications to the conven-
sentation of fractures is possible in the FVM by incorporating tional FEM have been made within the framework of partition
special ‘fracture elements’ as in the FEM, such as reported in of unity (PU). One typical example is the extended finite ele-
Granet et al. (2001). ment method (XFEM). It was first realized in Belytschko and
The boundary element method (BEM) requires only dis- Black (1999) by enriching the nodes of the finite elements near
cretization at the boundary of the solution domain, thus the crack tips and along the crack surfaces with the asymp-
reduces the problem dimensions by one and results in fast totic crack tip functions. Later, Moes et al. (1999) adopted
computing speed and easy mesh generation. One typical BEM the generalized Heaviside function H(x) (H(x) = 1 when x
is the displacement discontinuity method (DDM), which has above crack surface, and −1 when x below the crack surface)
been widely applied to simulate rock fracture propagation instead of asymptotic crack tip functions to introduce a dis-
problems by incorporating single fracture elements (Wen and continuous field across the crack surface away from the crack
Wang, 1991). However, in general, the BEM is not effi- tips. The finite element mesh is allowed to be constructed
cient as the FEM in dealing with material heterogeneity and totally independent of the cracks and remeshing is completely
99

avoided for crack growth problems. The function H(x), how- (Jiang and Yeung, 2004), more comprehensive representation
ever, can only describe the displacement discontinuity across of joints (Zhang and Lu, 1998), and the implementation of
a single crack surface. In order to account for branched and viscous boundaries (Gu and Zhao, 2009).
intersecting cracks, Daux et al. (2000) introduced another The DEMs have been widely applied in rock mechanics and
discontinuous function termed junction function J(x), which rock engineering due to their explicit and effective represen-
takes three different values, 0, +41, and −1, for the three tation of joints and fractures. They are, however, inefficient in
sides of a branched crack. Though great success has been modeling intact rock and pre-failure behavior of rock.
achieved, the XFEM encounters difficulties in dealing with The physical rock mass failure process involves crack initi-
arbitrarily branched cracks because of the complex enrich- ation, propagation, coalescence, formation of complete failure
ment functions and enrichment procedures. Another example surfaces, lastly large movement of discrete blocks, which is a
is the generalized finite element method (GFEM) developed typical combined continuum-discontinuum problem, neither
based on partition of unity. The GFEM and the XFEM use the continuum-based methods nor the disocntinuum-based
exactly the same technique, but the GFEM focuses on solv- methods can conveniently simulate the whole process, thus
ing problems of complex geometry with less error and less better to be modeled by a combined continuum-discontinuum
computer resources by enriching the standard finite element method. Combined continuum-discontinuum methods take
space with high-order terms or handbook functions, while the advantages of the strength of both methods while avoid-
XFEM mainly concentrates on crack propagation problems. ing their disadvantages. To develop continuum-discontinuum
In recent years, a large family of meshless methods with coupled methods, most researchers tend to couple the FEM
the aim of getting rid of meshing constraints has been devel- with the DEM. Recently, a hybrid finite-/discrete element code
oped. These methods require only generation and distribution ELFEN was developed and successfully applied to simulate
of discrete nodes, thus advantaged in sharp reduced demand the rock slope failures (Stead et al., 2006). In the ELFEN,
for meshing compared with the conventional FEMs for both two methods are proposed to insert the discrete crack. Intra-
continuous and fractured bodies. A large number of different element fracturing inserts the discrete crack in the actual
meshless formulations have been developed over the years. direction, but requires remeshing to achieve an acceptable
The element-free Galerkin (EFG) method, as a representative, element topology. In contrast, the inter-element fracturing
has been widely used for fracture problems. The meshless snaps the discrete crack to the most favorably oriented exist-
methods look promising, but are not sufficient to replace the ing element side, which does not require remeshing any more,
FEM because of their difficulties in numerical integration however, it requires very fine meshes to capture the fracture
of weak form, enforcement of essential boundary conditions, orientation accurately.
stability problems and high computational cost. The numerical manifold method (NMM) (Shi, 1991) is
The fundamental continuum assumption in continuum- another type of combined continuum-discontinuum meth-
based methods make them not suitable for block rotations, ods, which combines the continuum-based FEM and the
complete detachment and large-scale opening which are most discontinuum-based DDA in a unified framework. The NMM
concerned issues in rock mechanics. is advanced in simulation of jointed rock mass. Different from
Another class of numerical methods called discrete element the FEM, the approximation in the NMM is based on finite
method (DEM). The key concept of the DEM is that the prob- covers. By constructing mathematical covers, physical covers
lem domain is treated as an assemblage of rigid or deformable and manifold elements, the NMM is able to simulate continua,
blocks and the contacts between them need to be identified and transition from continua to discontinua as well as discontinua
continuously updated during the entire deformation/motion in a single framework. The mathematical covers do not need
process and represented by proper constitutive models. The to conform to neither the external boundaries nor the inter-
DEM was invented for solving rock mechanics problems and is nal discontinuities. Meshing task is greatly released and the
especially suitable for simulation of large-scale displacements remeshing is totally avoided for discontinuity evolution. A
of individual blocks, block rotations, and complete detach- failure criterion is implemented to realize the fracturing pro-
ment. According to the solution method used, DEM methods cess and the transition from continua to discontinua. Lastly,
can be further divided into two groups: explicit ones and the large movements of discrete blocks are effectively modeled
implicit ones. The most representative DEM method is the dis- by its robust contact detection and modeling algorithms inher-
tinct element codes UDEC (Itasca, 2004) and 3DEC (Itasca, ited from the DDA. This paper focuses on the NMM. The basic
2007) for two-dimensional and three-dimensional problems concepts of the NMM are briefly introduced in Section 2. The
in rock mechanics, respectively. Making use of particles to state of art of the NMM is presented in Section 3. The NMM
simulate granular materials is another development direction is compared with other discontinuity modeling methods in
of the distinct element method with the representative code Section 4. Recent developments done by the author’s team on
termed the particle flow code (PFC) (Itasca, 1999). the NMM are summarized in Section 5. The challenges and
The implicit version of the DEM is represented by the suggestions for discontinuous deformation analysis are drawn
discontinuous deformation analysis (DDA) (Shi, 1988). The in Section 6.
DDA is similar to the FEM, but can effectively deal with
the interaction of individual blocks in rock mass. The for-
mulations of the DDA can be derived using the principle 2 INTRODUCTION OF NUMERICAL MANIFOLD
of minimum potential energy. The applications of DDA are METHOD
mainly on rock engineering problems, such as landslides, tun-
neling, fracturing and fragmentation processes of geological The NMM is based on three fundamental concepts, namely
and structural materials, and earthquake effects. Recent devel- mathematical cover (MC), physical cover (PC) and manifold
opments of the DDA include discretizing the DDA block by element (ME). MCs are used to build the field approximation
finite elements (Shyu, 1993), coupled stress-flow simulation and allowed totally independent of the problem domain. PCs
(Jing et al., 2001), extension to three-dimensional analysis are the intersection of MCs and the problem domain and used
100

j
Using the PU ϕI , all the local approximations uI are com-
bined together to give the global approximation over each ME
e as
In Figure 1, two blocks can move away or completely detach

from each other. But, they need to satisfy the requirement of
no tension and no penetration along their interface. Such a
constraint is attributed to a contact problem in physics. In
the NMM, two springs, namely a normal spring and a shear
spring, are applied to each contact position. When friction
and cohesion along the block boundary are considered, the
NMM uses the Coulomb’s slip law to judge its dislocation
movements. There are three possible states for each contact:
(1) when the contact force N is tensile, the contact is open
and no springs are needed; (2) when N is compressive and
Figure 1. Illustration of basic concepts of the NMM. F ≤ N tan φ + c (where F is the shear force, φ is the friction
angle, and c is cohesion), a normal spring and a shear spring
are applied to lock the position; (3) when N is compressive and
F > N tan φ + c, sliding occur and the frictional force is taken
into account. Only a normal spring is applied to allow sliding.
3 STATE OF THE ART OF THE NMM
Since the initiation of the NMM in 1991, various develop-

ments and applications have been achieved during the last
20 years. Most of the publications are included in the series
of proceedings of the International Conference on Analysis
Figure 2. Generating NMM model from a structured triangular
of Discontinuous Deformation (ICADD) symposia. The main
mesh. developments and applications of the NMM are summarized
below.
Improvement of the approximation accuracy. The origi-
nal NMM assumed first-order polynomial displacement field,
to define the integration fields. The common area of several thus resulted in constant stress/strain within each element. In
PCs is termed as a ME. order to improve the approximation accuracy, Cheng et al.
To make the above concepts clearer, a simple example in (2002) incorporated Wilson nonconforming elements, while
Figure 1 is discussed. A two block system is covered by two Chen et al. (1998) implemented high-order local approxima-
MCs, denoted by M1 and M2 . The M1 forms two PCs, denoted tions into the NMM code.
by P11 and P21 . The M2 also forms two PCs, denoted by P12 Extension of the NMM for crack problems. Tsay et al.
and P22 . The four PCs forms totally four MEs, denoted by ei , (1999) applied the NMM together with a local mesh refine-
i = 1–4. ment and auto-remeshing schemes to predict the crack
In the above example, we use general MCs to illustrate the growth. The crack opening displacement is adopted to eval-
concepts. In actual implementation, we usually employ finite uate the stress intensity factors. Chiou et al. (2002) com-
element mesh to generate the MCs. Since the NMM does bined the NMM with the virtual crack extension method
not require MC edges to conform to the problem geometries, to study the mixed-mode fracture propagation. Li et al.
thus a regularly-patterned finite element mesh can always be (2005) developed enriched meshless manifold method for
employed. Meshing task is greatly released. For example, we two-dimensional crack modeling. The meshless manifold
can use the structured triangular mesh in Figure 2, where the method was proved to be more efficient than the traditional
union of six triangles sharing a common node forms a MC. meshless methods. Ma et al. (2009) extended the NMM for
Linear triangular shape functions naturally form a partition of complex cracks by introducing singular physical covers and
unity (PU) for the MCs. enriching them with asymptotic crack tip functions. Zhang
On each MC MI , a PU ϕI , which satisfies et al. (2009) further extended the NMM to simulate the growth
of complex cracks, with the maximum circumferential stress
criterion adopted as the crack growth criterion.
Extension of the NMM for weak discontinuity problems.
Terada et al. (2003) introduced the finite cover method (FCM)
as an alias of the NMM, presented the formulation for the static
equilibrium state of a structure with arbitrary physical bound-
aries including material interfaces, and extended the FCM
is defined. to analyze heterogeneous solids and structures involving the
j j
On each PC PI , a local approximation function uI , which discontinuities in strains and discontinuities in displacement
reflects the local characteristics of the solution, is defined. caused by debonding or rupture of material interfaces. The
101

displacement compatibility conditions on the material inter- numerical methods are pointed out.The distinct features which
face are enforced by Lagrange multiplier method. Kurumatani differentiate the NMM from other numerical methods are
and Terada (2005) extended the FCM to elastoplasticity prob- emphasized.
lems, in which the interface elements (or mortar elements)
are introduced to impose the essential boundary conditions,
displacement compatibility conditions on material interfaces. 4.1 NMM vs FEM
Terada et al. (2007) extended the FCM for analyses of pro- The NMM can covert to the standard FEM if the following
gressive failure processes involving cohesive zone fracture, conditions are satisfied:
starting from an interface debonding and evolving toward
one of the constituents of heterogeneous solids and struc- • A finite element mesh coinciding with the external bound-
tures, in which constituents fail according to the maximum aries and the internal discontinuities is adopted to generate
principal stress, while debonding is judged by multipliers of MCs;
interface elements. Kurumatani and Terada (2009) extended • Finite element shape functions are adopted as PUs;
the FCM to crack simulations for quasi-brittle heterogeneous • Constants are adopted as local approximations of PCs;
solids by introducing a modeling method called multi-cover- The NMM combines the FEM within its framework, but
layer modeling, in order to capture both weak and strong exceeds the standard FEM in the following aspects:
discontinuities.
Development of 3D NMM . Terada and Kurumatani (2005) • The NMM is more flexible than the FEM in discontinuity
introduced an integrated procedure for three-dimensional modeling without meshing conforming to discontinuities
structural analysis using the FCM. Formulations of the FCM and without remeshing for discontinuity evolution;
with interface elements for the static equilibrium state of a • Block rotation, complete detachment and large-scale open-
structure and detailed procedure of the 3D finite cover modeling which is difficult to be treated in the FEM can well
ing, including the geometry modeling with 3D-CAD and the represented in the NMM because of its discontinuum
identification of the geometry covered by a regular patterned nature;
unit of mathematical covers for numerical integration are pro- • High-order approximations can be more easily imple-
vided. Ma and He (2009) also proposed a three-dimensional mented by incorporating higher-order local approximations
NMM based on tetrahedron elements. Generation of mathe- without edge and middle nodes in the FEM;
matical covers, formulation of discrete equations, generating • Non-polynomial approximations which is not possible in
discrete blocks from a given fracture pattern are discussed in the FEM can be conveniently implemented because of the
detail. partition of unity.
Development of Eulerian NMM . Terada et al. (2007)
proposed an Eulerian FCM to perform Eulerian numerical 4.2 NMM vs DDA
analyses of hyperelastic bodies undergoing large deformation.
In order to realize the implicit solution scheme for quasi-static The discontinuous deformation analysis (DDA) was devel-
equilibrium problems, they used both the Lagrangian and the oped originally to solve problems in which a rock mass is
Eulerian mathematical meshes simultaneously, and proposed delimited into blocks by joints. Thus, in addition to the con-
a numerical algorithm to repeatedly perform re-covering and tact detection and frictional contact modeling, the DDA also
mapping over generalized elements. Okazawa et al. (2009) presents algorithm to form blocks from joints. Since rock
developed an Eulerian FCM for solid dynamics, which con- blocks generally undergo small deformation, the DDA sim-
siders the kinematic and dynamic boundary conditions in ulates each discrete body as a simple deformable block with
the Eulerian framework by incorporating the approximation only six degrees of freedom (DOFs) (i.e. normal and shear
strategy of the FCM into the existing Eulerian explicit FEM. strains of the block, rigid body translation of a specified point
Other developments and applications of the NMM . Lin within the block, and the rotation angle of the block with the
(1997) examined the NMM from the partition of unity’s per- rotation center at the specified point within the block) in a
spective, and demonstrated that the framework of the NMM two-dimensional setting. The NMM is developed based on
is similar to meshless methods, and the most significant the DDA. It preserves all the characteristics of the discrete
differences of the NMM come from its unique way of mod- element modeling such as the kinematics constraints, contact
eling discontinuities and its adoption of analytical integration detection and modeling from the DDA. However, the NMM
scheme. Li et al. (2005b) derived the governing equations employs a number of covers to raise the DOFs to more accu-
of the NMM from the method of weighted residual (MWR), rately describe the displacement field and stress field in each
which enriched the mathematical foundation of the NMM and block. So, we can say the NMM is the DDA with each block
extended it to problems such as head conduction and potential discretized into finite covers. In other words, if one discrete
flow, where the governing equations cannot be obtained from block is one manifold element with linear displacement field,
the minimum potential energy principle or other variational then the NMM will be degraded to the DDA.
principles. A coupled discontinuous deformation analysis and
numerical manifold method (DDA-NMM) has been devel- 4.3 NMM vs UDEC
oped by Miki et al. (2009) to take both methods’ advantages
while avoiding their shortcomings. The NMM differs from the UDEC in the following aspects:
• The UDEC is an explicit method. It calculates the state of
4 COMPARSION BETWEEN THE NMM AND a system at current step from the states of previous steps.
OTHER NUMERICAL METHODS It requires small time step t to keep the error in the result
bounded. The NMM is an implicit method. It finds a solu-
In this section, the NMM is compared with various other tion by solving an equation involving both the previous step
numerical methods. The similarities of the NMM with other and the current step. Much larger time step t can be used.
102

To achieve given accuracy, the NMM takes less computa-
tional time than the UDEC, even taking into account that
the NMM needs to solve the equilibrium equation;
• The UDEC uses numerical integration techniques (e.g.
Gauss quadrature) while the NMM adopts closed-form inte-
grations (i.e. simplex integration method) to evaluate the
weak form; Figure 3. Complex crack problems using NMM.
• The joints in the UDEC are always assumed as persistent,
failure occurs along pre-existing persistent joints. How-
ever, it is highly unlikely that such a network of fully
persistent discontinuities exist in nature. The NMM allows
non-persistent joints. In the NMM, cracks propagate and
coalesce with each other to form a continuous failure path,
thus seems more realistic;
• The UDEC uses a finite difference mesh while the NMM
uses a cover system to resolve the stress / strain vibration
within each block;
• The UDEC sometimes uses artificial joints to connect all
the blocks together to represent an intact material, and then
fracturing in intact material is realized by changing artifi-
cial joints to real joints. The fracturing follows pre-defined Figure 4. Crack propagation in NMM.
artificial joints, thus the results are sensitive to the block
configuration. The NMM allows the cracks arbitrarily align
with the elements, thus mesh dependency is avoided to some can be used to simulate complex crack problems. The physical
extent. domain and cracks/discontinuities are created independently
• The UDEC is a relatively mature method. Various material in the regular mathematical covers. Propagation of cracks do
models (e.g. elastic, Mohr-Coulomb plasticity, double- not disturb the regular mathematical covers. Figure 4 shows
yield, strain-softening, etc) are available in the code. In crack propagation in a machine element simulated by using
addition, relative motion along the discontinuities can be NMM.
linear or non-linear. The UDEC has been applied to var- The NMM enabling fracturing was firstly calibrated
ious engineering applications. The NMM is a relatively through several typical crack initiation and propagation exam-
new method. Only linear-elastic material model is avail- ples. The effectiveness and efficiency of the NMM enabling
able, and only linear motion along the discontinuities can fracturing has been demonstrated. Then, the NMM enabling
be accounted for. Its application is also limited. Further fracturing and the sub-block DDA are compared in fractur-
developments of the NMM to include various rock and rock ing modeling. Both methods can be viewed as modifications
joint material models are needed. to the original DDA method. Both methods can resolve the
displacement field and stress field within each block and can
allow fracturing in each block. However, crack propagation
5 RECENT DEVELOPMENTS ON NMM trajectory predicted by the sub-block DDA is highly mesh
dependent. In contrast, the NMM does not suffer from such
Various efforts have been dedicated to develop discontinuous a limitation, thus is more suitable for the practical rock fail-
deformation methods in analysis of rock mass stability. The ure analysis. Details can be found in Ma et al. (2009), Zhang,
main focus of the author’s research team is to improve the et al. (2010) and An (2010).
numerical manifold method to suits for engineering applica-
tions. The following will give a brief introduction on the recent
developments by the author’s team. 5.2 Slope stability analysis
Discontinuity (or referred to as joint) persistence is one of
5.1 Fracture modeling
the most important parameters affecting the rock slope fail-
The traditional numerical manifold method (NMM) has been ures. In the traditional rock engineering analyses, persistent
extended for simulation of complex crack problems. A singu- discontinuities are often assumed. It is partially due to that
lar physical cover concept has been introduced to represent fully persistent discontinuities are consistent with the fail-
the physical covers which contain the crack tips inside. The ure surface in post-failure investigation, it is also due to
asymptotic crack tip functions are incorporated into the local the constraints of the analysis techniques employed. Such
functions of singular physical covers. By such a modifica- an assumption is only valid in cases where major persistent
tion, the crack tip singularity can be well represented and the faults and/or bedding planes are present. However, in massive
crack tips can stop at arbitrary positions within a manifold natural rock slopes and deep, engineered slopes (e.g. open
element. The discrete equations are formulated. The Gauss pit mines), it is highly unlikely that such a network of fully
quadrature is implemented to carry out the integrations. The persistent discontinuities exist.
domain form of the interaction integral and the maximum Figure 5 shows a slope model with non-persistent joints and
circumferential stress criterion are adopted to evaluate the the final failure with different friction angles of the joints. In
stress intensity factors and predict the crack growth, respec- the simulation, only gravitational loading is applied to the
tively. Eight examples conducted demonstrate the accuracy slope. The gravitational loading is increased by 10 times to
and robustness of the extended NMM for complex crack prob- bring the slope to failure. It is equivalent to the strength
lems and their growth. Figure 3 shows that the extended NMM reduction method usually adopted by various researchers.
103

The NMM enabling fracturing is extremely suitable for pro- footwall slopes. Parametric studies regarding to orientation of
gressive failure analysis because: (1) regular mathematical bedding plane, orientation of predominant joint set, bedding
cover system is always applicable; (2) its two sets of cov- profile (with or without bedding roll), joint persistence are
ers make it convenient to describe complex cracks and crack carried out. Two typical failure modes of footwalls are shown
propagations; (3) the fracturing algorithm makes the transition in Figure 7. Details can be found in An (2010),
from continua to discontinua possible; (4) its mature con-
tact logic makes the interactions among the finally generated 5.3 A 3-D rock mass modelling tool
discrete block system be modeled in a realistic manner. The
current approach can greatly save computational time. A frac- Due to the presence of faults and joint sets, rock engineer-
turing algorithm based on the Mohr-Coulomb criterion with a ing problems must always be simulated by three dimensional
tensile cutoff has been implemented. Both tensile failure and numerical models. An algorithm for three dimensional rock
shear failure can be captured. Numerical results indicate that mass reconstruction has been developed (Fu et al. 2010). The
the NMM enabling fracturing is able to capture the fracturing information of discontinuities is obtained from 3-D disconti-
in intact rock bridge and finally allow the kinematic release. nuity network simulation. The discontinuities, each of which
The NMM enabling fracturing is promising for such problems. is defined by dip, dip direction and coordinates of a point
NMM enabling fracturing is also applied to analyze the in the discontinuity, can be of finite or infinite extent. If
potential failure mechanisms of footwall slopes in surface coal the discontinuity is finite, the geometry in a 2-D plane and
mines. Figure 6 depicts the NMM model of a typical footwall. the angle between the strike line and the reference line are
Numerical studies indicate that the NMM enabling fractur- also needed. The block generation process involves mainly
ing can reproduce the major failure mechanisms observed in four steps, namely, forming the zone of interest, boundary
planes cutting, discontinuities cutting and block integration.
The results can be checked by several methods. The generated
blocks can be convex or concave, which will certainly com-
pose complex block systems. Both simply connected blocks
and multi-connected blocks can be treated.
The geometrical rock mass modeling is useful, not only in
investigating the stability of rock slope and tunnels and their
support designs, but also in giving input information for more
advanced numerical simulations. Since detail information of
the discontinuities and rock blocks has been obtained from the
geometrical modeling, seepage analysis of underground water
and its effect on the rock mass behavior can also be studied.
Figure 5. Slope stability analysis results with different friction
angles.
Figure 8 shows some typical block shapes cutting out by the
3-D rock mass modeling tool. A few engineering slope and
rock tunnel examples are shown in Figure 9.
5.4 Development of 3-D Numerical manifold method

The numerical manifold method has been extended to the
3-D domain based on the 2-D fundamentals (Ma & He 2009).
Firstly, the general framework of the 3-D NMM is estab-
lished, including to give the definitions of the mathematical
cover, physical cover, manifold pattern and manifold element
in the 3-D domain. A general 3-D contact algorithm is subse-
quently developed for the 3-D NMM. A checking procedure is
implemented to ensure correctness and accuracy of the contact
algorithm. Locating and tracking the contact zone in the devel-
oped contact algorithm including three operation phases, i.e.,
contact warning, contact detection, and contact transferring.
Figure 6. A typical NMM models of footwall. The background mathematical cover which is a special feature
Figure 7. Different failure modes of footwall slope.
104

of the NMM is used in the contact warning phase. The con- failure in some locations. The total calculation time takes 6
tact detection phase matches lower level primitive hierarchies, seconds in the failure process. It is worth to emphasize that
and confirms the contact type of each contact pair candidate. the physical domain and discontinuities in the 3-D NMM
Finally, four basic types of contact pair are transformed into model are created separately from the mathematical covers.
only two essential contact modes, i.e., point to plane mode They interact with the mathematical covers to generate 3-D
and crossing lines mode. Totally 10 numerical examples have manifold element which can have different shapes, sizes and
been analyzed, including verification of the 3-D NMM formu- neighborhood conditions.
lations, analysis of a discrete blocky system, verification of
the contact algorithm, and two rock engineering applications,
i.e., stability analysis of rock slopes and tunnels. The proposed
3-D NMM is proved to be very efficient and stable. It has great
potential to be applied in practical rock engineering.
Figure 10(a) illustrates the three basic concepts of the NMM
in a 3-D model. There are two MCs in total, a sphere MC1 and
a hexahedron MC2. A pyramid represents the physical domain
in this instance as shown in Figure 10(b). A discontinuity
inserts into the pyramid body. Interactions with the physical
domain, four PCs, as shown in Figure 10(c), are generated.
These PCs finally form five manifold elements as shown in
Figure 10(d) and (e).
The following example shows the capability of the 3-D
NMM in simulating rock slope stability in a jointed rock mass.
The configuration of this example is illustrated in Figure 11.
After the 3-D NMM preprocess treatment, the intersection
between the physical model and mathematical covers, 1067
tetrahedron manifold elements are finally generated for the
case, as shown in Figure 11.
Obviously distinct with a 2-D case, these discrete blocks
also have crossed configuration along horizontal direction,
besides major joints, which are parallel to slope surface. The
slope would keep stable under the friction angle of 19.673◦ .
Figure 12 presents the failure process under the friction angle
of 12◦ . For clearance, the contour lines for the isolated blocks
are shown only. In the process of collapsing, its failure mode is
similar to the case of planar failure mode, mixed with toppling
Figure 10. Illustration of the 3-D NMM.
Figure 8. Arbitrary rock block cutting.
Figure 9. Discontinuous rock mass. Figure 11. Manifold treatments of a Rock slope.
105

Figure 12. Rock slope failure process using 3-D NMM method.
6 CHALLENGES AND SUGGESTIONS • Complexity of rock discontinuity properties

Usually rock and rock joint properties are obtained from
The rock mass as a natural geological material contains various laboratory tests on samples of small size which may not
discontinuities in different sizes and forms. Rock mass stabil- reflect the real discontinuities well. On the other hand,
ity is significantly affected by the discontinuous structural in situ experiments are difficult and expensive. There are
geology. Due to limited information from geological survey, gross uncertainties for the rock mass parameters in the
various assumptions have to be made in the existing numer- constitutive models.
ical methods. Obviously the current numerical methods to • Difficulties in numerical tolerance management
use deterministic analysis are not able to represent the real In discontinuous numerical modeling, appropriate toler-
structural geology realistically. Uncertainties in the geologi- ance management is very important when modelling rock
cal and material models were often ignored or over-simplified mass structures with discontinuities of high density. For
because of insufficient computational facility and inadequate example, the generated blocky system usually contains
methodology, especially in the 3-D numerical analysis. On some small or skinny blocks, contact algorithm may not
the other hand, leaving out these uncertainties may fail to be stable.
capture different failure patterns of the rock mass. The crit- • Extremely high computational cost
ical point is to determine to which level the geological and 3-D analysis of discontinuous rock mass is highly
material uncertainties are approximated in order to retain the expensive, especially when implicit contact and integral
critical characteristics of the rock mass. In summary, the cur- algorithms are adopted.
rent numerical methods in rock engineering are facing the
Toward realistic rock mass modelling: to conquer these
following challenges.
challenges, a cost-effective rock mass modelling strategy is
• Randomness of rock mass structural geology recommended. Different realization levels of the rock mass
The orientation and forms of the discontinuities in rock geology are classified. Corresponding affordable analysis
masses are highly random due to random distribution of dip, methods are defined as simplified, refined and advanced
its direction, persistence and shape. In the current prevailing approaches. In high level of realization, a comprehensive
deterministic numerical analysis, it is infeasible to cover geological tool should be used to cover the uncertainties of
such randomness due to not only the limited geological the discontinuous rock mass structures. Figure 12 shows a
information but also the constraints in the computational probability based rock mass model, which is generated by
facilities. considering uncertainties.
106

ACKNOWLEDGEMENT
The author would like to show his deep gratitude to Dr Genhua

Shi for his various support and encouragement in the past
years.
REFERENCES
An X.M. 2010. Extended numerical manifold method for engineering

failure analysis. PhD thesis. Nanyang Technological University,
Singapore.
Figure 13. Probability based rock mass structure. BelytschkoT. & BlackT. 1999. Elastic crack growth in finite elements
with minimal remeshing. International Journal for Numerical
Methods in Engineering, 45(5): 601–620.
Chen G., Ohnishi Y. & Ito T. 1998. Development of higher-order
manifold method, International Journal for Numerical Methods
in Engineering, 43: 685–712.
ChengY. M., ZhangY. H. & Chen W. S. 2002. Wilson non-conforming
element in numerical manifold method, Communications in
Numerical Methods in Engineering, 18: 877–884.
Chiou Y. J., Lee Y. M. & Tsay R. J. 2002. Mixed mode fracture prop-
agation by manifold method, International Journal of Fracture,
114: 327–347.
Daux C., Moes N., Dolbow J., Sukumar N. & Belytschko T. 2000.
Arbitrary branched and intersecting cracks with the extended finite
element method. International Journal for Numerical Methods in
Engineering, 48: 1741–1760.
Fu G.Y., He L. & Ma G.W. 2010. 3-D rock mass geometrical model-
ing with arbitrary discontinuities, International Journal of Applied
Mechanics 2(4): 871–887.
Goodman R. E., Taylor, R. L. & Brekke, T. L. 1968. A model for
the mechanics of jointed rock, Journal of the Soil Mechanics and
Foundations Division-ASCE, 94:637–659.
Granet S., Fabrie P., Lemonnier P. & Quintard M. 2001. A two-phase
Figure 14. Toward realistic rock mass modeling. flow simulation of a fractured reservoir using a new fissure element
method. Journal of Petroleum Science and Engineering, 32(1):
35–52.
Gu J. & Zhao Z.Y. 2009. Consideration of the discontinuous deforma-
tion analysis on wave propagation problems, International Journal
High level realization/simplified analysis: in order to con- for Numerical and Analytical Methods in Geomechanics, 33(12):
1449–1465.
sider randomness of the discontinuities and creating numerous Itasca. 1999. PFC 2D (Particle Flow Code in 2 Dimensions),
3-D geological models based on probability analysis, a sim- Version 1.1. Itasca Consulting Group, Inc. Minneapolis, MN:
plified analysis e.g. using key block search algorithm is ICG.
more preferable, which is not only able to capture instable Itasca. 2004. UDEC-Universal Distinct Element Code, Version 4.0.
blocks and critical discontinuities, but also greatly reduce the Itasca Consulting Group, Inc., Minneapolis.
computational cost to an tractable extent. Itasca. 2007. 3DEC-3-dimensional Distinct Element Code, Version
Medium level realization/refined analysis: based on sim- 4.1. Itasca Consulting Group, Inc., Minneapolis 2007.
plified analysis, those critical cases of rock mass structure Jiang Q. H. & Yeung M.R. 2004. A model of point-to-
which result in rock mass failure or low factor of safety can be face contact for three-dimensional discontinuous deformation
identified. A refined analysis, e.g. the DDA method, should analysis. Rock Mechanics and Rock Engineering, 37: 95–
116.
then be applied in calculating more detailed results of rock Jing L.R., Ma Y. & Fang Z.L. 2001. Modeling of fluid flow and solid
mass deformation. The inefficiency in this methodology is that deformation for fractured rocks with discontinuous deformation
the stress distribution is still a coarse estimation. Although the analysis (DDA) method, International Journal of Rock Mechanics
refined analysis has comparatively higher computational cost, and Mining Science, 38: 343–355.
the analysis cases are significantly reduced. Kurumatani M. & Terada K. 2009. Finite cover method with multi-
Low Level realization/advanced analysis: for those most cover layers for the analysis of evolving discontinuities in hetero-
critical cases, an advanced analysis, e.g, numerical mani- geneous media, International Journal for Numerical Methods in
fold method should be applied. It will give more accurate Engineering, 79(1): 1–24.
results of rock mass deformation and stress distribution. An Kurumatani M. & Terada, K. 2005. Finite cover method with mortar
advanced analysis is highly time-consuming. However, only a elements for elastoplasticity problems, Computational Mechanics,
36: 45–61.
very limited number of cases need to be analysed. Li S. C., Li S. C. & Cheng Y. M. 2005a. Enriched meshless mani-
Figure 14 schematically shows the concept towards a fold method for two-dimensional crack modeling, Theoretical and
realistic and cost-effective rock mass modelling. Applied Fracture Mechanics, 44: 234–248.
For large-scale problems, parallel computing can also be Li S., Cheng Y. & Wu Y. F. 2005a. Numerical manifold method based
implemented to fully explore the capacity of multi-core CPU on the method of weighted residuals, Computational Mechanics,
and GPU technologies. 35: 470–480.
107

Lin J. S. 2003. A mesh-based partition of unity method for discon- using numerical modeling techniques. Engineering Geology, 83:
tinuity modeling, Computer Methods in Applied Mechanics and 217–235.
Engineering, 192: 1515–1532. Terada K. & Kurumatani M. 2005. An integrated procedure for
Ma G. W. & He L. 2009. Development of 3D numerical manifold three-dimensional structural analysis with the finite cover method,
method, International Journal of Computational Methods, 7(1): International Journal for Numerical Methods in Engineering, 63:
107–129. 2102–2123.
Ma G. W., An X. M., Zhang H. H. & Li, L. X. 2009. Model- Terada K., Asai M. & Yamagishi M. 2003. Finite cover method for
ing complex crack problems with numerical manifold method, linear and nonlinear analyses of heterogeneous solids, Interna-
International Journal of Fracture, 156(1): 21–35. tional Journal for Numerical Methods in Engineering, 58: 1321–
Miki S., Sasaki T., Koyama T., Nishiyama S. & Ohnishi, Y. 2009. 1346.
Development of coupled discontinuous deformation analysis and Terada K., Ishii T., Kyoya T. & Kishino Y. 2007. Finite cover method
numerical manifold method (NMM–DDA), International Journal for progressive failure with cohesive zone fracture in hetero-
of Computational Methods, 7(1): 131–150. geneous solids and structures, Computational Mechanics, 39:
Moes N., Dolbow J. & Belytschko T. 1999. A finite element method 191–210.
for crack growth without remeshing. International Journal for Terada K., Ishii T., Kyoya T. & Kishino, Y. 2007. Finite cover
Numerical Methods in Engineering, 46: 131–150. method for progressive failure with cohesive zone fracture in het-
Ning Y.J., An X.M., & Ma G.W. 2011. Footwall slope stability analy- erogeneous solids and structures, Computational Mechanics, 39:
sis with the numerical manifold method, International Journal of 191–210.
Rock Mechanics and Mining Sciences (in press) Tsay R. J., Chiou Y. J. & Chuang W. L. (1999) Crack growth predic-
Okazawa S.,Terasawa H., Kurumatani M.,Terada K. & Kashiyama K. tion by manifold method, Journal of Engineering Mechanics, 125:
2009. Eulerian finite cover method for solid dynamics, Interna- 884–890.
tional Journal of Computational Methods, 7(1): 33–54. Wen P.H. & Wang Y. 1991. The calculation of SIF considering the
Shi G. H. 1988. Discontinuous Deformation Analysis - A new numer- effects of arc crack surface contact and friction under uniaxial
ical model for the static and dynamics of block systems, PhD tension and pressure. Engineering Fracture Mechanics, 39: 651–
Dissertation, Department of Civil Engineering, U.C. Berkeley. 660.
Shi G. H. 1991. Manifold method of material analysis, Trans. 9th Zhang H. H., Li L. X., An X. M. & Ma, G. W. 2010. Numerical
Army Conf. Applied Mathematics and Computing, Minneapolis, analysis of 2D crack propagation problems using the numerical
Minnesota, pp. 57–76. manifold method, Engineering Analysis with Boundary Elements,
Shyu K. 1993. Nodal-based discontinuous deformation analysis, 34(1): 41–50.
PhD thesis. University of California, Berkeley. Zhang X. & Lu M. W. 1998. Block-interfaces model for non-linear
Stead D., Eberhardt E. & Coggan J.S. 2006. Development in the numerical simulations of rock structures. International Journal of
characterization of complex rock slope deformation and failure Rock Mechanics and Mining Science, 35: 983–990.
108

Synthetic rock mass applications in mass mining
M.E. Pierce & C. Fairhurst

Itasca Consulting Group, Inc., Minneapolis, Minnesota, US
ABSTRACT: A new approach to the prediction of rock mass behaviour, called Synthetic Rock Mass (SRM) modeling, has
been developed in response to the demands of the mass mining industry, where successful caving and fragmentation of the rock
mass demands a greater understanding of the joint fabric and its impact on rock mass strength brittleness and fragmentation.
The method involves embedding a Discrete Fracture Network (DFN) into a Bonded Particle Model (BPM) of rock. Much of
the work in producing a realistic SRM sample centers on accurate reproduction of the joint fabric present in the real rock mass.
This paper focuses on some of the challenges associated with this process in the context of mass mining applications. It also
touches on some of the challenges associated with representation of intact rock (and embedded joints) and concludes with a
short review of SRM validation efforts.
Keywords: rock joints, rock mass, numerical modelling
1 INTRODUCTION
The problem of characterizing the mechanical behaviour of a

rock mass was well recognized as the central challenge of rock
mechanics almost 50 years ago. In fact, Professor Leopold
Müller identified this problem as his prime raison dêtre in
founding ISRM in 1962. The complexity of the challenge was
beyond classical continuum analysis and practical engineering
design has relied heavily on development of empirical guides. Figure 1. Example of a three-dimensional Synthetic Rock Mass
(SRM) consisting of a Discrete Fracture Network embedded in a
Significant advances have been and are being made, but real-
Bonded Particle Model.
istic representation of discontinuities and their influence on
rock mass deformation and failure behavior remain the central
challenge of rock mechanics. A new approach to this problem, industry-funded research consortium. The samples are cre-
called Synthetic Rock Mass (SRM) modeling, has been devel- ated using the Particle Flow Code (PFC2D/PFC3D) and are
oped in response to the demands of the mass mining industry, intended to replicate both the intact rock and joint fabric
where successful caving and fragmentation of the rock mass observed at representative scales in the field. The method
demands a greater understanding of the joint fabric and its involves embedding a Discrete Fracture Network (DFN) into a
impact on rock mass strength brittleness and fragmentation. Bonded Particle Model of rock (Potyondy and Cundall, 2004)
The SRM method combines Discrete Fracture Network (DFN) (Fig. 1). A more detailed explanation of the SRM method
simulation with the discrete element numerical modelling pro- and examples of where it has been applied to the analy-
cedures of Cundall (1971) and Potyondy and Cundall (2004). sis of rock mass behavior may be found in Cundall (2008),
While these techniques have been around for some time (and Cundall et al. (2008), Esmaieli et al. (2008, 2009, 2010), Mas
have been used in a variety of practical applications) the ability Ivars et al. (2008, 2011), Pierce et al. (2006, 2007, 2009) and
to combine the two in three-dimensions at high joint densi- Sainsbury (2008).
ties and large scales has only recently been enabled through Some advantages of the SRM approach include the ability to
advances in discrete element modelling, coupled with recent represent joint terminations inside of blocks (e.g., see Fig. 2)
advances in computer power. and the ability to represent new fracture growth explicitly.
Much of the work in producing a realistic SRM sample cen- Also, because a single contact within the BPM may accept
ters on accurate reproduction of the joint fabric present in the only a reorientation from one DFN joint, it is not necessary
real rock mass. This paper focuses on some of the challenges to resolve thin slivers of rock that often exist between sub-
associated with this process in the context of mass mining parallel and closely spaced joints within stochastic DFNs (and
applications. It also touches on some of the challenges associ- are notoriously difficult to resolve efficiently with zones or
ated with representation of intact rock (and embedded joints) particles). Hence, small features in a very large sample (which
and concludes with a short review of SRM validation efforts. are unlikely to affect the overall behavior significantly) are
filtered out automatically, with the scale of filtering being
controlled by particle size.
2 SYNTHETIC ROCK MASS For the large-scale problems encountered in mass mining,
the SRM technique essentially is used as a “virtual laboratory”
The Synthetic Rock Mass (SRM) methodology was devel- to obtain estimates of rock mass strength, modulus, brittleness
oped within the Mass Mining Technology (MMT) project, an and fragmentation for use in larger-scale predictive caving and
109

Figure 2. Three increasingly detailed views of a two-dimensional
Synthetic Rock Mass. The colors denote intact rock blocks bounded
by joints. Figure 4. Example of a joint-pole distribution from a combina-
tion of underground mapping and drilling. Note the wide range in
sub-vertical joint orientation (maximum pole concentration of 5%).
Figure 3. Use of Synthetic Rock Mass (SRM) testing within the

methodology for caving prediction.
gravity flow models. A number of 3D SRM samples normally Figure 5. Anisotropy in rock mass modulus derived from lab tests
are produced for each geomechanical domain at representative on 12-m diameter SRM samples with the joint orientations shown in
scales (normally around 5–20 m, depending on the joint spac- Figure 4.
ing) and inputs are derived from standard core logging, lab
testing and underground mapping. The flow chart in Figure 3 model. These are well-suited to situations in which there is
outlines the general methodology developed within the MMT often insufficient data to justify the use of more complex (e.g.,
research project to obtain predictions of cave growth and mate- terminating) joint models and have (as a result) been employed
rial flow at the mine scale from a combination of empirically in most of the SRM studies conducted to date. The goal is to
derived and SRM-derived rock mass properties. This process produce joint networks that honor the measurements of open-
also includes validation of both the SRM-scale and cave-scale joint orientation, density (spacing) and size distribution (trace
models through comparison with measurements of microseis- length) that are obtained from drilling and mapping. The fol-
micity, time domain reflectometry (TDR) breakages (or open lowing sections discuss a few of the challenges associated with
hole blockages) and drawpoint fragmentation made on site. measuring each of these three properties, as well as their poten-
tial impacts on SRM and rock mass behavior. Hadjigeorgiou
3 DFN SIMULATION (2011) outlines some further challenges associated with the
data gathering and DFN generation process for mining appli-
Much of the work in producing SRM sample centers on accu- cations (e.g., the identification of open versus sealed joints).
rate reproduction of the joint fabric present in the real rock
mass. Discrete Fracture Network (DFN) simulation provides
3.1 Joint orientation
a convenient and practical means to build joint networks for
this purpose. There are many different approaches to DFN When there is a wide range in joint orientations (e.g., Fig. 4),
simulation and they vary considerably in their complexity and the process of joint-set selection is bound to exclude a large
underlying assumptions (Staub et al., 2002, Hadjigeorgiou, number of randomly orientated joints. As a result, the orien-
2011). Research is being conducted within MMT to inves- tations are generally bootstrapped, i.e., sampled directly from
tigate how sensitive the strength and deformation behavior the entire range of mapped orientations. Results of SRM test-
of SRM samples is to the underlying DFN model, and how ing suggest that marked anisotropies in rock mass strength
important it may be to consider more state-of-the-art models and modulus still emerge from weak concentrations of joint
such as the Universal Fracture Model (Davy et al., 2010). This orientations (Fig. 5).
paper focuses on the challenges associated with the devel- Joint orientations generally are the easiest of the three
opment of DFNs employing a relatively simple random disk main inputs to obtain from mapping. The main challenge is
110

2008). During the DFN simulation process, the joint density
is adjusted (generally via trial-and-error) until the target frac-
ture frequency (or frequencies) are met along one or more
synthetic boreholes or scanlines (whose orientations match
those of the real boreholes or scanlines along which the origi-
nal measurements were obtained). SRM testing of the different
DFNs can provide insight into how the rock mass strength is
likely to vary within a given domain as a result of variations
in joint density. There is significant uncertainty in how this
joint density varies spatially through the domain, therefore it
is logical to run a number of cave-scale predictive models,
each containing a unique stochastic realization of the esti-
mated rock-mass strength distribution. If this proves to be
too expensive computationally, an alternative is to construct
a number of numerical “samples” of the rock mass and test
these to quantify the decrease in variability and drop in mean
strength that is expected to occur as we enlarge our “win-
Figure 6. Variation in fracture frequency (averaged over 10-m dow” of observation from the scale of the SRM sample to
intervals) from a borehole drilled through volcanic rock. the cave-scale. Note that both of these approaches require an
understanding of the distance over which joint density is likely
to ensure that there is not a strong sampling bias, particularly to vary, as discussed by Lorig (2009).
where jointing is strongly subvertical and drilling data comes
from mostly subvertical boreholes. A strong bias also may be
present in mapping data if it is obtained primarily from under- 3.3 Joint size distribution
cut and/or extraction-level drifts, which tend to be parallel. To We know that joints range considerably in size and shape, and
help reduce this bias, undercut/extraction-level mapping often often are much larger than the exposure being mapped. This
is supplemented with mapping from other access such as rim makes the joint-size distribution a difficult thing to quantify,
drives, declines and conveyor drives. Some degree of sam- and so is generally the greatest source of uncertainty in the
pling bias always will be present in the data and so a Terzaghi DFN generation process. To help simplify the quantification
correction (Terzaghi, 1965) generally is applied. of joint size, we often assume that the size distribution is in
the form of a power law with a negative exponent (i.e., many
3.2 Joint density small joints and few large joints). While there is debate as
to the existence of a single scaling law for fractures found at
Fracture frequency measures are the simplest and most useful
the outcrop scale (m-scale) to the 1000-m-long fault scale, the
way to characterize joint density, and are required in order to
DFNs generated for the purposes of SRM testing are restricted
calibrate joint density within the DFN. A fracture frequency
typically to scales less than 20–30 m (because joint spacing
count is simply a count of the number of fractures crossing a
is typically on the order of 0.1–1 m). Larger-scale structures
scanline or borehole interval of fixed length and orientation,
(e.g., faults, lithologic contacts) generally are accounted for
and is expressed as the number of fractures per meter (ff/m).
deterministically within the cave-scale model; i.e. not within
We attempt to produce DFN that exhibit fracture frequency
the DFN.
measures equivalent to those measured through mapping and
Because joints below a certain diameter cannot be resolved
logging.
by the particle assembly (and because most mapping cam-
Jointing is strongly subvertical in many orebodies, and so
paigns ignore joints below a certain size anyway) a lower cutoff
the fracture frequency measured horizontally often is much
to this distribution also is employed. Joints smaller than this
higher than the fracture frequency measured vertically. This
lower cutoff (generally on the order of 0.25 to 0.5 m diameter),
fabric has a strong impact on rock mass response to caving
as well as sealed features (veins, healed joints, microdefects),
and so it is very important to quantify it by measuring frac-
must not be ignored, however, and attempts normally are made
ture frequency in both the horizontal and vertical directions.
to account for them implicitly within the intact rock part of
Although it is often not possible, a third measure from the roof
the SRM (the bonded particle model). This is discussed further
or end face of a drift can be taken to provide three orthogonal
below in the section on BPM modeling.
measures. At a minimum, both horizontal and vertical mea-
Most mapping takes place on two-dimensional surfaces and
sures of fracture frequency should be obtained from vertical
so joint trace lengths are normally recorded rather than joint
exposures.
diameter. By a trial and error procedure, one can adjust the
The joint density often varies significantly within a sin-
exponent in the power law of joint size distributions used in
gle domain (e.g., Fig. 6). As noted by Lorig (2009), it is
DFN generation until the simulated trace statistics (e.g., trace
non-conservative to ignore such variability when conducting
length per unit area or average trace length) match well with
stability or caving analysis at a large scale, because fail-
those obtained from mapping.
ure mechanisms often can exploit the weakest parts of the
Trace length data often is scarce because it is time consum-
rock mass (i.e. rock masses rarely fail according to the aver-
ing to collect. Consider three increasingly detailed means of
age properties). As a result, a range of DFNs usually are
trace length mapping:
constructed to cover the range of joint spacing likely to be
experienced in a given domain. The size of the DFN (and 1. Record the termination (one, both, none) of all joints
associated SRM sample) is selected to ensure that it cap- mapped.
tures a reasonably representative volume (estimated to be 2. Record the length of every trace crossed during the fracture
approximately 20 joint intersections wide, see Cundall et al., frequency measures.
111

Figure 7. Results of three-dimensional DFN simulations employ-
ing the same density but two different assumed length distributions:
a) Exponent of −4 results in a rock mass dominated by the lower
boundary of fracture length; and b) An exponent of −3 results in
a rock mass dominated by the upper boundary of fracture length.
The plots indicate how the three-dimensional joints would appear as Figure 8. Results of UCS testing on 6-m diameter SRM samples
traces on a 10-m high vertical plane. with the same joint spacing (0.5 m), but differing joint size distribu-
tions. Four of the samples have joints with a with uniform radius, r,
and one sample has a power law distribution of radii employing an
3. Record the length of every trace in the exposure being exponent of −4 and a minimum and maximum radius of 0.3 m and
mapped. 3 m respectively. Units for stress are MPa.
The first measure should be employed in any case, and must
be considered as a minimum. The third measure is not practi-
cal in most cases, due to the time and effort involved. There in intact strength. While it may be possible to achieve this scale
have been advances in high-resolution digital stereographic effect within the BPM in some manner (e.g., through the addi-
imaging, however, that may make it possible to conduct such tion of microdefects), at present the BPM is calibrated simply
detailed mapping automatically. The success of this technol- to exhibit the scaled intact strength of a rock block whose
ogy likely will depend on the ability to distinguish between dimension equals the average joint spacing. As discussed by
natural and blast-induced fractures, distinctions that are built Pierce et al. (2009), the relations of Laubscher and Jakubec
into manual mapping campaigns. The second method is a good (2001) and Yoshinaka et al. (2008) appear to provide a reason-
trade-off: it is less prohibitive time-wise, but still offers a sig- able starting point for estimation of rock block strength for
nificant amount of information to help constrain the joint-size veined/fractured rocks. Accounting for intact strength-scale
distribution and calibrate the DFN. Even this method may effects within SRM samples more directly should be a focus
prove too time consuming for some mines. In this case, one of future research.
might consider taking a second set of fracture frequency mea- As noted by Potyondy and Cundall (2004), the traditional
sures that consider only joint traces larger than a specified size Bonded Particle Model tends to exhibit a macroscopic fric-
(e.g., the average size of the exposure being mapped). These tion angle and a macroscopic unconfined compressive:tensile
additional fracture frequency measures can be very helpful in strength ratio that is low compared to many rock types. Several
calibrating the DFN, providing simultaneous information on researchers have demonstrated that these shortcomings can
joint density and size distribution. be overcome through use of non-spherical particles (e.g., by
By varying the exponent, but keeping the joint density clumping or clustering individual disks or spheres together).
constant, it is possible to generate DFNs in which jointing While several explanations have been offered for this effect
is dominated by the lower or upper boundaries of fracture (including the introduction of additional heterogeneity into the
diameter (Fig. 7). It is important to constrain this exponent system) recent research conducted by Itasca suggests that it is
with trace length data, because the two rock masses in Figure 7 the retention of resistance to moment following bond break-
would be expected to exhibit different strengths and age that is key to achieving a more realistic response. This is
responses to caving (e.g., fragmentation). The results of SRM the subject of ongoing development and a number of varia-
tests shown in Figure 8 illustrate the sensitivity of rock mass tions on the traditional parallel bond model (e.g., Potyondy,
strength, modulus and brittleness to the underlying joint-size 2011) promise to provide for a more realistic intact rock fric-
distribution. tion angle and unconfined compressive:tensile strength ratio
without the computational burden associated with clumping
or clustering of particles.
4 BONDED PARTICLE MODELING Using SRMs for the study of hydraulic fracturing also
present a challenge, because particle sizes much larger than
There are several challenges associated with the represen- the actual grain size generally are used. As a result, the BPM
tation of intact rock via the Bonded Particle Model and tends to exhibit larger fracture toughness than the real rock
the embedment of joints within a particle assembly. As dis- and breakdown pressures can be higher than the real rock.
cussed previously, small open joints (below the cutoff in size Techniques have been developed to overcome this problem,
employed for DFN generation) and sealed features (e.g., veins, and essentially involve scanning the particle assembly to iden-
healed joints, microdefects) often are accounted for implicitly tify process zones where crack tips can be identified and
within the intact rock part of the SRM (the bonded particle treated separately. Once these techniques mature, the impact
model). This is complicated by the fact that in real rocks, the of fracture toughness on the behavior of SRMs under com-
presence of such features often leads to a marked scale effect pression and tension should be studied in greater detail to
112

determine its significance in the study of rock mass strength to establish realistic DFNs given the constraints imposed in
and deformation behavior within the mining context. these situations.
As discussed previously, an individual contact within the
SRM currently is restricted to a single joint orientation. While REFERENCES
this offers a convenient means to filter out slivers of rock that
may not be relevant to the larger-scale behavior, it sometimes Adhikary D.P., Dyskin A.V., Jewell R.J. & Stewart D.P. 1997, A study
can cause individual particles to inhibit shear unrealistically of the mechanism of flexural toppling failure of rock slopes. Rock
where two sub-orthogonal joints cross at a single contact. Mech and Rock Engin, 30:75–93.
These “impassable elastic asperities” can be identified prior Cundall, P. A. 1971 “A Computer Model for Simulating Progressive
to simulation and removed, or can be monitored throughout Large Scale Movements in Blocky Rock Systems,” in Proceedings
the simulation and removed based on a force criterion. This of the Symposium of the International Society for Rock Mechanics
is most important in simulations involving a small number of (Nancy, France, 1971), Vol. 1, Paper No. II-8.
persistent joints, any one of which might strongly control the Cundall P. A. 2008. Recent advances in numerical modelling for
large-scale mining projects, ACG News, 30, 1–7.
rock mass response. Cundall, P. A., M. E. Pierce & D. Mas Ivars, 2008 “Quantifying
the Size Effect of Rock Mass Strength,” in SHIRMS 2008 (Pro-
ceedings, 1st Southern Hemisphere International Rock Mechanics
5 VALIDATION
Symposium, Perth, Western Australia, September 2008), Vol. 2,
pp. 3–15. Y. Potvin et al., Eds. Nedlands, Western Australia:
The Mass Mining and Large Open Pit research projects have Australian Centre for Geomechanics, 2008.
initiated a series of studies to validate the SRM approach Davy, P, R. Le Goc R., C. Darcel, O. Bour, J.-R. de Dreuzy &
more systematically through comparison of model results with R. Munier, 2010. “A Likely-Universal Model of Fracture
well-constrained experimental and case study results. To date, Scaling and its consequence for crustal hydro-mechanics”, J.
validation within the MMT project has focused on the com- Geophys. Res. Solid Earth, (2010), 115, B10411, doi:10.1029/2009
parison of new fracture orientations from SRM tests with JB007043.
those inferred from Enhanced Microseismic Analysis (EMA) Deisman, N., D. Mas Ivars & M. Pierce, 2008 “PFC2D Smooth Joint
(Reyes-Montes et al., 2007), as well as the back analysis of Contact Model Numerical Experiments,” in GeoEdmonton ’08:
A Heritage of Innovation (61st Canadian Geotechnical Confer-
laboratory tests on discontinuous joints (Deisman et al., 2008).
ence, Edmonton, September 2008), Paper No. 83.
Additional validation studies currently underway within Esmaieli K., J. Hadjigeorgiou, M. Grenon, & R. Harrisson 2008.
MMT involve simulation of lab tests on heavily veined intact Ore pass stability analysis at the Brunswick Mine using PFC3D,
rock samples of various scales from the Esmeralda Mine in In Proceedings of the First International FLAC/DEM Symposium
Chile (comparison of modeled versus measured strength, fail- on Numerical Modeling. R. Hart et al., Eds. Minneapolis: Itasca
ure mechanisms and acoustic emissions) and simulation of Consulting Group, Inc.
primary fragmentation at Ridgeway Deeps Mine (comparison Esmaieli K., J. Hadjigeorgiou & M. Grenon 2009. Estimation of
of fragment size distribution as a function of stress path and synthetic rock mass strength accounting for sample size. In: Proc.
lithology). Research also is being conducted in parallel with International conference on rock joints and jointed rock masses,
the MMT project to investigate the ability of SRM to represent Tucson, AZ.
Esmaieli K., J. Hadjigeorgiou, & M. Grenon 2010. Estimating geo-
pillar behavior in hard rock mines and roadway performance metrical and mechanical REV based on synthetic rock mass mod-
in coal mines. els at Brunswick Mine. International Journal of Rock Mechanics
The Large Open Pit project conducted a comprehensive set & Mining Sciences 47 (2010) 915–926.
of validation studies for SRM samples employing a simpler, Hadjigeorgiou, J. 2011 “Rock Characterization: From Empiricism to
more computationally efficient lattice-based particle model Synthetic Rock Mass,” in Bergmekanikdag 2011 (Swedish Rock
for the intact rock. These studies covered the analysis of intact Mechanics Day 2011), Stockholm: Rock Engineering Research
rock with anisotropic fabric (discontinuous joints), direct Foundation and National Group ISRM (International Society for
shear of a discontinuous joint (Lajtai, 1969) and the behavior Rock Mechanics).
of toppling slopes as observed in centrifuge tests (Adhikary Lajtai, E. Z, 1969. Strength of discontinuous rocks in direct shear.
et al., 1997) and in situ (Chuquicamata Mine). Geotechnique, 19, 218–233.
Laubscher, D.H. & Jakubec, J. 2001.The MRMR Rock Mass Classifi-
cation for Jointed Rock Masses. in Underground Mining Methods:
6 FURTHER APPLICATION Engineering Fundamentals and International Case Studies, pp
475–481, Society of Mining Metallurgy and Exploration, SMME.
Lorig, L. J. 2009 “Challenges in Current Slope Stability Analysis
Most SRM studies to date have been conducted in the context Methods,” in Slope Stability 2009 (Universidad de Los Andes,
of surface and underground mining operations. These situa- Santiago, Chile, November 2009).
tions provide three-dimensional access to the rock mass and Mas Ivars, D., M. Pierce, D. DeGagné & C. Darcel 2008. “Anisotropy
associated observation and mapping of rock discontinuities. and Scale Dependency in Jointed Rock-Mass Strength — A
Other engineering situations – such as petroleum extraction Synthetic Rock Mass Study,” in Continuum and Distinct Ele-
and development of geothermal reservoirs – encounter rock ment Numerical Modeling in Geo-Engineering (Proceedings,
masses where fractures and similar discontinuities are likely to 1st International FLAC/DEM Symposium, Minneapolis, August
play a significant role. While it is clearly necessary to define 2008), Paper No. 06-01. R. Hart et al., Eds. Minneapolis: Itasca
the nature of the fractures and their influence on engineering Consulting Group, Inc., 2008.
Mas Ivars, D., M. Pierce, C. Darcel, J. Reyes-Montes, D. O. Potyondy,
design, this is complicated significantly by the fact that access
R. P. Young, & P. Cundall 2011. “The Synthetic Rock Mass
to the subsurface rock mass in these situations is by bore- Approach for Jointed Rock Mass Modelling,” Int. J. Rock Mech.
hole only. Currently, emphasis is placed on study of evidence & Min. Sci., 48(2), 219–244 (2011).
of in situ rock mass fracture systems from rock core, com- Pierce, M., P. Cundall, D. Mas Ivars, C. Darcel, R. P. Young,
bined with borehole logging and other geophysical techniques. J. Reyes-Montes & W. Pettitt 2006. “Six Monthly Technical
Considerable research will be needed before it will be possible Report, Caving Mechanics, Sub-Project No. 4.2: Research and
113

Methodology Improvement, & Sub-Project 4.3, Case Study Appli- Modeling in Geomechanics — 2011 (Proceedings, 2nd Inter-
cation,” ICG, Report to Mass Mining Technology Project, 2004– national FLAC/DEM Symposium (Melbourne, February 2011).
2007, ICG06-2292-1-Tasks 2-3-14, March 2006. Paper 08-04, pp. 459–465, D. Sainsbury et al., Eds. Minneapolis:
Pierce, M., P. Cundall, D. Potyondy & D. Mas Ivars 2007. “A Syn- Itasca International Inc, 2011.
thetic Rock Mass Model for Jointed Rock,” in Rock Mechanics: Reyes-Montes J., Pettitt, W. & Young, R.P. 2007. Validation of a
Meeting Society’s Challenges and Demands (1st Canada-U.S. Synthetic Rock Mass Model Using Excavation Induced Micro-
Rock Mechanics Symposium, Vancouver, May 2007), Vol. 1: seismicity. Proceedings of the 1st Canada-US Rock Mechanics
Fundamentals, New Technologies & New Ideas, pp. 341–349. Symposium (Vancouver).
E. Eberhardt et al., Ed. London: Taylor & Francis Group, 2007. Sainsbury, D. P. 2008 “Analysis of River Bed Cracking Above Long-
Pierce, M., D. Mas Ivars & B. Sainsbury 2009. “Use of Synthetic wall Extraction Panels in the Southern Coalfield of New South
Rock Masses (SRM) to Investigate Jointed Rock Mass Strength Wales, Australia,” in SHIRMS 2008 (Proceedings, 1st Southern
and Deformation Behavior,” in CD Proceedings, International Hemisphere International Rock Mechanics Symposium, Perth,
Conference on Rock Joints and Jointed Rock Masses (Tucson, Western Australia, September 2008), Vol. 1, pp. 325–338.Y. Potvin
January 2009), Paper 1091. P. H. S. W. Kulatilake, Ed. Tucson: et al., Eds. Nedlands, Western Australia: Australian Centre for
Kulatilake & Associates, 2009. Geomechanics, 2008.
Pierce, M., M. Gaida & D. DeGagne. 2009 “Estimation of Rock Staub I., A. Fredriksson & Outters, N. 2002. Strategy for a rock
Block Strength,” in ROCKENG09 (Proceedings, 3rd CANUS mechanics site descriptive model. R-02-02. Swedish Nuclear Fuel
Rock Mechanics Symposium, Toronto, May 2009), Paper No. 4360. and Waste Management Company.
M. Diederichs and G. Grasselli, Eds. Terzaghi, R.D., 1965. Sources of errors in joint surveys. Geotech-
Potyondy, D. O., & P. A. Cundall 2004. “A Bonded-Particle Model for nique 15, 287–304.
Rock,” Int. J. Rock Mech. & Min. Sci., 41(8), 1329–1364 (2004). Yoshinaka, R., Osada, M., Park, H., Sasaki, T. & Sasaki, K. 2008.
Potyondy, D. O. 2011 “Parallel-Bond Refinements to Match Macro- Practical determination of mechanical design parameters of intact
properties of Hard Rock,” in Continuum and Distinct Element rock considering scale effect. Engineering Geology 96: 173–186.
114

How to select rock mass parameters for analysis and design
T. Ramamurthy
Department Angron Geotech Pvt. Ltd., Naraina Industrial Area, Phase II, New Delhi, India
A. Ghazvinian
Rock Mechanics Division, Faculty of Engineering, Tarbiat Modares University, Tehran, Iran
ABSTRACT: Strength and modulus of rock mass suggested from RMR, Q and GSI have been examined with modulus ratio,
Mrj , for their reliability. The design parameters adopted in some case studies based on these rock mass classifications are
presented. The modulus ratios of rock masses are found to be much higher than those of intact rock, even with back analyses.
Based on joint factor, Jf , compressive strength, modulus, cohesion and friction angle were estimated and applied in the analysis
of a few cases. The predictions of deformations agreed well with the field measurements. Estimation of normal (Kn ) and shear
(Ks ) stiffness of rock mass under various conditions of joint system is covered in some detail. When dilatancy component is
separated in direct shear tests, the scale effect appears to be insignificant. Further, when to adopt equivalent continuum model
for rock mass is suggested.
Subject: Rock mass characterisation
Keywords: numerical modeling, rock joints, rock mass, rock mass classification, rock properties and site characterisation.
1 INTRODUCTION
Soil and rock are geotechnical materials formed through very

complex processes. Soil is treated most often as a contin-
uum which is homogeneous and isotropic, for the purpose of
analysis and design, even though it is a particulate material.
It is not so in the case of rock masses. They are dis-
continuous, non-homogeneous, anisotropic and prestressed.
Collection of undisturbed specimen of rock mass to test in a
laboratory is considered uneconomical and mostly not practi-
cable. In some cases large field shear and plate load tests are
conducted to assess rock mass properties. Even these tests
are being considered time consuming and expensive. So tests
are conducted as a routine on intact rock specimens in the lab-
oratory to arrive at the upper bound values of rock parameters. Figure 1. Hypothetical stress-strain curves.
Attempts have been made to correlate strength and modulus of
intact rock with those of rock mass through rock mass classi- ratio, Mri = Ei /σci , where, Ei = tangent modulus at 50% of
fications. These correlations are often being adopted to assess failure strength, σci = compressive strength at failure and
compressive strength (σcj ), modulus (Ej ), cohesion (cj ), fric- i = subscript for intact rock.
tion angle (φj ) and stress strain response, etc. of rock mass
during the last three decades, subscript j for rock mass. How
reliable are these correlations is ones confidence to choose 2 MODULUS RATIO CONCEPT
these parameters to predict/estimate the performance of rock
mass/structure. Deere and Miller (1966) presented a classification of intact
Hypothetical stress-strain curves for three different rocks rocks based on modulus value (Ei ) at 50% of the failure stress
are presented in Fig. 1. Curves OA OB and OC represent and the unconfined compressive strength (σci ). Vast experi-
three stress-strain curves with failure occurring at A, B and mental data of 613 rock specimens from different locations
C respectively. Curves OA and OB have same modulus but covering 176 igneous, 193 sedimentary, 167 metamorphic
different strengths and strains at failure. Whereas the curves and 77 limestones and dolomites was presented by them to
OA and OC have same strength but different moduli and fail- classify intact rock on the basis of (σci ) and modulus ratio,
ure strains. So neither strength nor modulus alone could be Mri (=Ei /σci ). It is found that for basalts and limestones one
chosen to represent the overall quality of the rock. There- could expect Mri values upto 1600 and minimum for shale
fore, strength and modulus to gether will give a realistic could be above 60. Even weathered Keuper showed Mri more
understanding of the rock response for engineering usage. than 50. Only chalks with high porosity have shown Mri
This approach to define the quality of intact rocks was pro- greater than 1000 (Hobbs 1974) due to high modulus and
posed by Deere and Miller (1966) by considering the modulus low strength.
115

3 ROCK MASS CLASSIFICATIONS The ratio of σcj1 /σcj2 varies from 1:7 to 54: 1 depending upon
the values of Qc . The values of modulus ratio, Mrj , will be more
The most commonly adopted rock mass classifications are or less constant and are around 800; in fact, Eqs. (4 & 5) give
RMR, Q and GSI systems at present. Significant contributions this ratio as 800 for a value of γ = 2.5 g/cc, irrespective of
have come from Bieniawski (1973) and Barton et al. (1974) Qc varying from 0.001 to 1000, i.e. whether the rock is intact,
based on their vast experience in number of tunnels. GSI clas- jointed, isotropic or anisotropic, (refer Ramamurthy 2004).
sification by Hoek (1994) and Hoek and Brown (1997) is
a combination of RMR and Q approaches with some mod-
ification. The value of GSI is same as RMR (Bieniawski 6 STRENGTH AND MODULUS FROM GSI
1976) for RMR > 18 and for GSI < 18, Q values are adopted.
Often designers adopt one or all the classifications to arrive Hoek (1994) and Hoek & Brown (1997) advocated to adopt
at conservative values. Geological Strength Index, GSI. GSI adopts both RMR &
Q systems with some modifications mainly to estimate the
compressive strength of rock mass.
4 STRENGTH AND MODULUS FROM RMR
Bieniawski (1973) suggested shear strength parameters, cj For estimating the deformation modulus, Hoek (1994) rec-
(cohesion) & φj (friction angle), subscript j refers to rock mass, ommends the use of Eq. (2) as per Serafim & Pereira using
for five levels of rock mass classes. With these values of cj & RMR as per Bieniawski (1973) and not the GSI system. The
φj the uniaxial compressive strengths (σcj ) of the mass are values of GSI, sj & Ej given in Table 5 of Hoek (1994) have
calculated as per Mohr-Coulomb criterion, Eq. (1). been adopted in calculating the values of modulus ratio, Mrj .
The values of Mrj are surprisingly high, ranging from 1500 to
1875 with an average value of 1642. These values of Mrj do not
decrease, with the decrease of GSI values, (refer Ramamurthy
The values of Ej are estimated from Eq. (2) as per Serafim & 2004).
Pereira (1983),
7 PARAMETERS USED IN ANALYSIS AND DESIGN
Only a few cases are presented in the following.

The values of modulus ratios, Mrj = Ej /σcj are extremely
high for rock masses. For RMR = 80, the Mrj = 28528 and for (i) For the underground work, pump storage, Rio Grande
RMR = 20, it is 6846. No. 1, Argentina (Moretto et al. 1993), massive gneiss of
Cerro Pelado was encountered.
Mri = 273 (min) and 546 (max.); Mrj = 48000, i.e. Mrj /
5 STRENGTH AND MODULUS FROM Q Mri = 87.9 using max. Mri . The measured deformations
were twice the calculated values.
Barton (2002) suggested modification to Q values (Barton (ii) Hoek and Moy (1993) dealt with various aspects of
et al. 1974) by considering the influence of uniaxial compres- “Power house caverns in weak rock”. For siltstone
sive strength of the intact rock (σci ) in the form (RMR = 48), Mri = 89 and Mrj = 1627; Mrj /Mri = 18.28.
(iii) For Masua mine, Italy (Barla 1993) carried out 3D anal-
ysis; displacement discontinuity model used; also 2D
FEM; the displacements on the hanging wall & southern
and recommends Qc values for estimating the compressive open stope were in the same range of predicted values.
strength and modulus of rock mass as For the dolomite limestone, Mri = 888 and Mrj = 2722
(max.), Mrj /Mri = 3.07.
(iv) In their contribution on “Practical estimates of rock
mass strength”, Hoek & Brown (1997) recommendations
indicate
(a) For good quality rock (GSI = 75), Mrj = 648
where, γ = density of rock mass in g/cc or t/m3 .
(b) For Avrg. Quality rock (GSI = 50), Mrj = 692
Another recommendation of Barton (2002) is to assess cj &
(c) For poor quality rock (GSI = 30), Mrj = 824
φj of rock mass from the following expressions,
(d) For Braden braccia, El Teniete mine, chile
GSI = 75, Mrj = 1546
(e) For Nathpa Jhakri HE Project, India,
Quartz mica schist: Mrj = 1585
(f) Athens schist – decomposed, GSI = 20,
Mrj = 1060 (min) & 1474 (max.)
where, Js is for joint set number (Barton uses Jn ), SRF is the
(g) Yacambu Quibor tunnel, Veneguela, For poor quality
stress reduction number, Jr is joint roughness number, Jw is
graphittic phyllite, GSI = 24, Mrj = 870.
for seepage and its pressure and Ja is joint alteration number.
Note: Mrj values are rather high; good rock low and
With the data provided in the Tables 6 & 7 of Barton (2002),
poor rock high values of Mrj !
the values of compressive strength of rock mass are calculated
from cj & φj as per Mohr-Coulomb criterion, Eq. (1) and are (v) Large span tunnel and cavern, Singapore, Constructed in
referred to as σcj2 . The values of compressive strength from Bukit Timah qranite (Zhou et al. 2003). Mri = 400 and
Eq. (4) are say σcj1 . Mrj = 818 with Mrj /Mri = 2.04.
116

Table 1. Values of n for different joint orientation angles, β0 for types of anisotropy (Ramamurthy 1993 & 2001).
Joint orientation angle, β0 0 10 20 30 40 50 60 70 80 90

Joint inclination parameter, n U- shaped 0.82 0.46 0.11 0.05 0.09 0.30 0.46 0.64 0.82 0.95
Shoulder shaped 0.85 0.60 0.20 0.06 0.12 0.45 0.80 0.90 0.95 0.98
(vi) For Masjed – e – Soloiman, HEPP, Iran (Stabel & Samani is given by Eq. (9) and the corresponding modulus by Eq. (10).
2003), 2D – elasto-plastic hybrid FEM & BEM applied Equation (9) as per Ramamurthy (2001),
using the properties of the layer mass.
Conglomerate: Mri = 789, Mrj = 1136, Mrj /Mri = 1.44
Sandstone: Mri = 373, Mrj = 1022, Mrj /Mri = 2.74
Siltstone: Mri = 333, Mrj = 2372, Mrj /Mri = 7.12
Claystone: Mri = 348, Mrj = 3896, Mrj /Mri = 11.2.
(vii) Intake tunnel, Karuna III, HEPP, Iran: the analysis of the wherein Jf is a joint factor defined by Eq. (11)
tunnel was carried out using 2D, UDEC and 3D elastic
programmes (Tabanrad 2003). Measured deformations
agreed with the estimated values from block analy-
sis. Back analysis was carried out using direct method where, Jn = joint frequency, i.e., number of joints/meter,
UDEC software and σcj and Ej were estimated. For Marly which take care of RQD and joint sets and joint spacing;
limestone/marl: Mri = 221, Mrj = 2885, Mrj /Mri = 13.05 n = inclination parameter depends on the inclination of sliding
(viii) A stochastic analysis was carried out to estimate σcj and plane with respect to the major principal stress and anisotropy;
Ej of three grades of Ankara andesites by calculating the joint or set which is closer to (45 – φj /2)◦ with the major
the influence of correlations between relevant distribu- principal stress will be the most critical one to experience
tions on the simulated RMR values (Sari et al. 2010). sliding; r = a parameter for joint strength; it takes care of
The model was also used in Monte Carlo simulation to the influence of closed or filled up joint, thickness of gouge,
estimate possible ranges of the Hoek – Brown strength roughness, extent of weathering of joint walls and cementa-
parameters. tion along the joint. This factor could be assessed in terms
From minimum strength & modulus values of an equivalent value of friction angle along the joint as
Grade A: Mri = 520, Mrj = 7000; Mrj /Mri = 13.46 tan φj = τj /σnj obtained from shear tests, in which τj = shear
Grade B: Mri = 470, Mrj = 2113; Mrj/ Mri = 5.89. strength along the joint under a normal stress, σnj . The values
Grade C: Mri = 359, Mrj = 1311; Mrj/ Mri = 3.65. of n and r are given in Tables 1, 2 and 3.
From maximum strength & modulus Now from Eqs. (9) & (10), the modulus ratio of the jointed
Grade A: Mri = 425, Mrj = 1568; Mrj/ Mri = 3.69. mass with respect to that of the intact rock is given by Eq. (12)
Grade B: Mri = 431, Mrj = 1378; Mrj/ Mri = 3.20.
Grade C: Mri = 326, Mrj = 963; Mrj/ Mri = 2.90.
(ix) Power house Cavern, Rogun project, Kajikistan:
To predict deformations of roof and slide walls using 3D
FEM with M-C criterion (Bronshteyn et al. 2007) adopted For values of σcj & Mrj for different values of Jf varying
reduced ci , from 0 to 500 for σci = 100 MPa & Mri = 500 of intact rock,
φi and Ei as indicated here. the Mrj values rapidly decrease to 87 with the increase of Jf .
For sandstone: c = 3.0 to 1.1 MPa (F.S. = 2.7) The relationship between Ej & σcj (i.e., Mrj ) can not be taken
φ◦ = 50 to 38 (F.S. to tan φ: 1.53) as constant when the rock mass is experiencing continuous
E = 9000 to 5000 MPa (F.S. = 1.4) fracturing and undergoing change in its quality.
Mri = 546, Mrj = 1109; Mrj /Mri = 2.03 Even though the original classification due to Deere &
For aleurolites: c = 2 to 0.5 MPa (F.S. = 4.0) Miller was suggested only for intact rocks, it was modified to
φ◦ = 45 to 32.5 (F.S. to tan φ: 1.56) classify rock masses as well (Ramamurthy 2004). The main
E = 5500 to 2670 (F.S. = 2.06) advantage of such a classification is that it not only takes into
Mri = 569, Mrj = 1467; Mrj /Mri = 2.5. account two important engineering properties of the rock mass
(x) Waitaki dam block No. 10, New Zealand: A 3D FEM but also gives an assessment of the failure strain (εf ) which
analysis was carried out for the Waitaki dam (Richards & the rock is likely to exhibit in the uniaxial compression, when
Read 2007). The tiltmeter deformations under the block the stress-strain response is near linear. That is,
No. 10 was matched to obtain in situ modulus Ej . The
ratio Ej /Ei was 0.15 for GSI = 20 of class II greywake
and found to be very high even for disturbance factor,
D = 0 as per Hoek & Diederichs (2006). Further, the ratio of the failure strain of the intact rock to
that of the jointed rock is given by,
8 STRENGTH AND MODULUS FROM JOINT

FACTOR, Jf
On the basis of experimental data (Ramamurthy 2001), the
Based on the extensive experimental results in uniax- following expression was also suggested,
ial compression on jointed rocks and rock-like materials
(Ramamurthy 2001), the compressive strength of jointed mass
117

Table 2. Suggested joint strength parameter, r, for filled up joints
at residual stage (Ramamurthy 1993).
Friction angle Joint strength,

Gouge material φj , degrees r = tan φj
Gravelly sand 45 1.00

Coarse sand 40 0.84
Fine sand 35 0.70
Silty sand 32 0.62
Clayey sand 30 0.58
Clayey silt
Clay – 25% 25 0.47
Clay – 50% 15 0.27
Table 3. Suggested values of r for values of σci (Ramamurthy 1994). Figure 3. Influence of weathering on modulus ratio of rocks.
Compressive strength, Joint strength

σci , (MPa) parameter, r Remarks
elemental cubes. Specimens have undergone sliding, shearing,
2.5 0.30 Fine grained splitting and rotational modes of failure.
5.0 0.45 Unconfined compression tests were also carried out on
15.0 0.60 three weathered rocks, namely quartzite, granite and basalt,
25.0 0.70 to (Gupta & Rao 2000), which have gone through different stages
45.0 0.80 of weathering on five levels of weathering of quartz and four
65.0 0.90 levels of weathering of both granite and basalt. The values of
100.0 1.00 Coarse grained compressive strength and modulus are presented together in
Fig. 3.
Ideally when field tests are conducted, the test block is to be
isolated from the parent mass by careful cutting and dressing
operations to assess σcj and Ej in the unconstrained condi-
tion. Such a test block should have a slender ratio more than
one, preferably two. Unfortunately the data from such tests
is extremely limited. Whenever some data is available, it is
projected to indicate the effect of the specimen size rather
than the change in the quality of the rock within the test
specimen/block. As the size increases, the number of joints,
their inclination, even if the strength along some of the joints
remains same, would effect the response of the block. From
test results of Natau et al. (1995) on three sizes of specimens
80, 235 and 620 mm in unconfined state, showed 620 mm size
specimen having Mrj = 250 while the extrapolated NX size
having Mrj = 1000 (Ramamurthy 2004).
Most of the data of modulus is obtained by conducting tests
on limited area in tunnels, in drifts, in boreholes and even
Figure 2. Influence of jointing on modulus ratio. if plate loading tests are conducted on a level surface under-
ground or in open excavation, there is always some degree
of lateral confinement. The measured modulus values tend to
Figure 2 is an extended version of Deere & Miller approach be higher particularly for weaker rock masses. Such results
(1966) and will cover very low strength to very high strength need to be corrected for lateral confinement to obtain values
rocks. A modulus ratio of 500 would mean a minimum failure corresponding to the unconfined condition. When such data
strain of 0.2 %, whereas a ratio of 50 corresponds to a mini- is provided, the designer has the freedom to choose or modify
mum failure strain of 2 % as per Eq. (13). Very soft rocks and the strength and modulus depending upon the in situ stress
dense/compacted soils would show often failure strains of the expected in the field. Using Eq. (16), Ramamurthy (1993),
order of 2%. Therefore, the modulus ratio of 50 was chosen the influence of confining pressure on Ej can be estimated,
as the lower limiting value for rocks (Ramamurthy 2004).
In Fig. 2, the location of the intact specimen is shown at
‘I’on the σci,j & Ei,j plot. When the experimental data of σcj &
Ej of the jointed specimens having the same material as that
of the intact specimen is plotted, all the points fall along an where, the subscript 0 & 3 refer to σ3 = 0 and σ3 > 0, σ3 is the
inclined line originating at ‘I’, cutting across the constant effective confining stress. For σcj = 5 MPa, Ej3 for σ3 = 2 MPa
boundaries of modulus ratio. This suggests that as fractur- confining pressure will be 4 times and for σ3 = 1 MPa, it will
ing continues, the locations represented by σcj and Ej follow be 2.3 times of Ej0 . This is likely to happen in field plate load
a definite trend (Singh et al. 2002). This data is from test tests conducted underground in a limited surface area or lateral
specimens, each of which had on an average more than 260 in situ stress not being released.
118

The strength criterion for the jointed rocks is given by
where, σ1 and σ3 are major and miner principal stresses,
respectively, σcj is the uniaxial compressive strength of jointed
rock obtained from Eq. (9), αj and Bj are strength parameters
of the jointed rock. The values of αj & Bj are obtained from
Eqs. (18 & 19),
where, αi & Bi are values of strength parameters obtained from

triaxial tests on intact rock specimens for the failure criterion
(Ramamurthy 1993, 2001).
Figure 4. Calculated and experimental stress-strain plots for block
9 ELASTOPLASTIC ANALYSIS WITH JOINT jointed specimen of Gypsum plaster for different confining pressure.
FACTOR, Jf
to predict the response of jointed rocks of sandstones, gran-
The power house complex of Nathpa Jhakri hydropower ite & gypsum plaster and compared with the experimental
project in North India consists of two major openings, results. Only sample stress–strain plots for multiple-jointed
i.e., machine hall 216 × 20 × 49 m3 (length × width × height) block specimens of gypsum plaster (Brown & Trollope 1970)
with an overburden of 262.5 m and a transformer hall for different confining pressures are shown in Fig. 4 along
198 × 18 × 29 m3 which is located downstream of the with the experimental results.
machine hall. The in situ stress for the rock was determined
using hydraulic fracturing technique. The vertical stress was 10.1 Analysis of Shiobara power house cavern
found to be 5.89 MPa with an in situ stress ratio of 0.8035. The
constitutive model based on disturbed state concept (Desai The equivalent continuum model was applied for the analy-
1994) was used to characterize the behavior of the intact rock. sis of a large cavern in jointed rock mass for the Shiobara
Material parameters for the model were determined for the power station in Japan (Horii et al. 1999). The cavern mea-
rock specimens as indicated in Varadarajan et al. (2001a). sures 161 m length, 28 m width and 51 m heigh, located at a
Strength and modulus of the jointed rock mass, quartz mica, depth of 200 m below the ground. The three in situ principal
were determined with joint factor, Jf (Varadarajan et al. 2001b) stresses were recorded as 5.0, 3.9 and 2.8 MPa. The reported
to carry out the finite element analysis of the powerhouse average intact rock compressive strength and elastic modu-
cavern. The failure criterion based on Jf (Ramamurthy 2001) lus were 83.3 MPa and 42.1 GPa, respectively. The rock mass
was adopted to estimate the strength under different in situ was characterized as rhyolite consisting of platy and columnar
stresses. The analysis was carried out using computer code joints. The jointed rock mass surrounding the cavern has been
DSC-SST-2D developed by Desai (1997). Twelve stages of analyzed by the Finite element method using the proposed
excavation were used in the study. equivalent continuum approach. Equivalent material proper-
The predicted values of the displacements by FEM were ties for jointed rock were modeled using the relations with
compared with the observed values at five out of six locations Jf which was taken as 41, 12 and 111 per meter for joint
and found to be, in general, close in agreement. sets I, II&III respectively. The ‘variation of relative displace-
ment at the end of different excavation steps with progress of
caver excavation with time-displacement measurements along
10 EQUIVALENT CONTINUUM MODELLING WITH Jf the measurement line agreed well with the observed values.
These are presented here only for B110&B111 in Figs. 5 &
Using the strength, modulus and failure strain relations for 6. The variation of relative displacements from FEM and as
rock mass with Jf and the corresponding value of intact measured (Horii et al. 1999) at the completion of whole exca-
rock, a few cases have been analysed, (Sitharam et al. 2001, vation along the measurement line were compared for B117,
Sitharam & Madhavi Latha 2002). The tangent elastic modu- Fig. 7. The strength parameter r was chosen from Table 3 based
lus of intact rock was represented by confining stress depen- on intact rock strength.
dent hyperbolic relation (Duncan & Chang 1970).A numerical
model was developed from an existing finite element code for
10.2 Analysis of Kiirunavaara mine, Sweden
a non-linear soil–structure interaction program to account for
material non-linearity of both the intact and jointed rocks. The Kiirunavaara mine, which is 4000 m long, with an average
This model was incorporated in the commercial finite width of 90 m, is located 144 kilometers north of the arctic
difference code FLAC. A FISH function was written to incor- circle in the city of Kiruna in North Sweden. The magnetic
porate joint factor model with Duncan-Chang non-linear iron ore body is relatively strong surrounded by competent
hyperbolic relationships in FLAC. The model has been applied quartz porphyry on the hanging wall and syenite porphyry on
to two large power station caverns, one in Japan and the other the footwall. The rock mass is jointed with two to three joint
in Himalayas and a slope at Kiirunavaara mine in Sweden. For sets. One joint set is oriented roughly parallel to the ore body
validation purposes, the finite element analysis was applied as the other two strike obliquely to the ore body. All joints dip
119

Figure 5. Time history of displacements near the cavern wall along
measurement line B110.
Figure 8. Shear strains for FLAC model for −586 m mining level.
Table 4. Measured and predicted deformations at Nathpa-Jhakri

cavern.
Location of Deformation along the line (mm)
Stage MPRXEL Observed Predicted FLAC 2D
1 1024 13–18 10–14.0

2 1022 6–12 8.2–13.5
3 1022 −1.3–2.5 1–2.3
4 1018 1–4 1.4–3.7
5 1006 10–45 13–42.2
Figure 6. Time history of displacements near the cavern wall along 6 996 1–3 1.3–4.2
measurement line B111.
The excavation steps were simulated in the numerical analysis

FLAC2D) and the locations of the installation of exten-
someters were identified for obtaining the displacements for
comparison with the measured displacements from instrumen-
tation of the cavern. The variation of displacements with time
is also obtained from numerical analysis by solving for equilib-
rium after each excavation step. Comparison of the observed
and predicted deformations along the measurement line at
Figure 7. Displacement along measurement line B117 at the different locations for various excavation levels after the com-
completion whole excavation. pletion of excavation is presented in Table 4. The joint factor
was estimated as 22 per meter for the analysis and the value
of r was chosen from Table 3.
fairly steep, 60◦ to 90◦ . The locations where first set of cracks
were observed in 1985 were mapped by Sjoberg (1999). 10.4 Abutment stability of Chenab bridge
The value of joint factor (Jf ) was obtained as 13. The
strength parameter r was chosen from Table 3. The total size Slope stability analysis of the right abutment (359 m high) at
of the model used was 2000 × 1300 meters (Sitharam and Kauri side of Chenab river between Katra and Laole, Jammu
Madhavi Latha 2002). Sequential mining was simulated in and Kashmir, India, was carried our using FLAC of plane
the FLAC by modeling the excavated material with null model strain case for pseudo – static approach with earthquake inten-
and solving after each stage of excavation. Failure can only be sity of 6.5 (Sitharam and Maji 2007). With Jf values and intact
observed from the concentration of shear strains in the model. rock
The path of concentrated shear strains represents the failure Properties, obtained σcj , Ej , stress – strain response, cj and
surface in the model. The failure thus simulated by the numer- φj for the rock mass. By applying varying factors of safety to
ical model using practical equivalent continuum approach was cj and tan φj , failure conditions in the slope was determined.
compared with the failure observations in the field. Shear fail-
ure was observed in the footwall of the model while excavating 10.5 Synthetic rock mass
at the mining level of −586 m, agreeing with the field obser-
vations as reported by Sjoberg (1999). Typical failure surface From “synthetic rock mass” model constructed with ubiqui-
for the mining step of −586 m is presented in Fig. 8. tous joints taken from their actual orientations, Clark (2006)
showed the variation of σcj /σci with RMR covering scale
and anisotropy effect using FLAC (ITASCA 2005). This data
10.3 Analysis of Naptha-Jhakri power house cave
matches totally with the Jf relationship with σcj /σci , Eq. 9 by
This cavern was also analyzed by (Sitharam and Madhavi adopting the relationship,
Latha, 2002). The finite difference grid used for the analy-
sis was of size 210 m 450 m with 1320 rectangular zones.
120

Figure 10. Effect of joint orientation on the stress strain response
of Agra sandstone.
Figure 9. Comparison of stress-strain curves predicted by ECM and
ANN with experimental values for block jointed Gypsum plaster. where, σn = normal effective stress increment
δ = the resulting deformation.
The shear stiffness (Ks ) is obtained by conducting a shear
10.6 Stress-strain response test on the joint plane similar to a direct shear test. Kn value
is often many times larger than the Ks value (10 – 100 times).
Garaga and Latha (2007) introduced user defined FISH func- Kn is not a content value for jointed rock mass. It varies with
tions to incorporate Jf rations in to the explicit finite difference normal stress, loading and unloading cycles, joint roughness
code FLAC (version 4.0, ITASCA 1995) to simulate exact variation, disturbance to the joints and presence of opening
joint behavior. Adopting hyperbolic formulation of stress- along the joints, (Barton 1990, 2008; Bandis 1990, Hart 1993).
strain response of jointed rock, estimated elastic initial tangent With the increase of block size, Kn has been found to decrease
modulus as per Duncan and Chang (1970), they predicted due to variations in the joint characteristics. With the σn ,
stress – strain, strength envelopes and variation of strength Kn value varies nonlinearly; often larger values are chooses
with joint orientation of tested specimens. A very good (Pande et al. 1990) for the analysis.
agreement has been shown. The value of Ks is also considered to be scale dependent. If
the dilatancy component, i, (from tan−1 v/h) as indicated
10.7 Critical strain in para on scale effect is separated from the measurement peak
shear strength, the response may not show such scale depen-
Critical strain, εcr , beyond which squeezing in likely to occur dency. To apply such a result to assess the actual value of Ks in
in an underground excavation was estimated using Jf (Singh the formation at a different location, the dilatancy component
et al. 2007). of the location is to be added as determined from stereonet or
The critical strain is given as any other procedure. As for as possible Kn and Ks should be
obtained from field tests. Since the joint characteristics play
an important role in the prediction of rock mass response, they
are to be chosen with due care.
Or in terms of Jf on an average basis for different modes of
failure of rock mass,
11.1 Kn and Ks for equivalent continuum model
(i) Joint thickness, t is negligible
For the elastic deformation analysis of a transversely isotropic
rock mass loaded normal to the joints which are continuous,
10.8 Joint factor in ANN model parallel, spaced uniformly, much smaller in number than the
By constructing an ANN model, stress-strain response and rock mass and their volumetric space (thickness) occupied is
variation of Ej /Ei with Jf were predicted for jointed rocks, also negligible, Goodman et al. (1968) and Goodman (1989)
by specifying intact rock properties, σ3 , Jf and axial strain suggested that
as inputs (Garaga & Latha 2010). Out of number of cases
presented by them, only a sample stress – strain predicted by
equivalent continuum (ECM) (with Jf ) and ANN models are
presented in Fig. 9 with experimental results of block jointed
gypsum plaster (Brown & Trollope 1970). Figure 10 shows where, Ej = equivalent modulus of rock mass
the predictions as per ANN with experimental results of Agra Ei = modulus of intact rock
sandstone. Kn = normal stiffness of the joint
s = spacing of the joint,
subscripts i & j refer to intact and jointed
11 ESTIMATION OF Kn AND Ks rocks respectively.
The above expression is obtained by comparing the defor-
For the analysis of jointed rock mass to predict deformations, mation in the rock mass with summation of the deformations
normal (Kn ) and shear (Ks ) stiffness values are required. The in the intact rock and the joint. The spacing, s, is also given by
Kn values are obtained from direct loading normal to the joint 1/Jn ; Jn = joint frequency. Rearranging the above expression
plane; the resulting expression is given by (24),
121

Ej can be obtained from Joint factor Jf (Eq. 10). If gap exists in the joints in the x – direction, the term Ksx
Similarly, the expression for shear stiffness in terms of shear will not come in to play.
modulus is given by
(iv) Kn from plate load test
When the joint system is not well defined as indicated in the
foregoing paras an approximate values of Kn could be obtained
where, G = shear modulus = E/2 (1 − ν). for the rock mass from a field plate load test data. Timoshenko
and Goodier (1970) presented the out come of such a test as
In terms of E values by considering s = 1 unit length and writ-

ing s Kn as equivalent modulus (Eg ) of the joint, the Eq. (24)
will be where, δ = average deformation of rock mass
P = total load applied
Ejf and νj are elastic constants of rock mass
A = loaded area of plate
i.e. Ej = Ei Eg /(Ei + Eg ) n = factor related to the plate shape.
This Eq. (37) may be written as
(ii) Joint thickness, t > 0

If an open joint is filled with soil or a sheared zone exists
with gouge material, comparing the deformation of jointed For a circular plate n = 0.96 and for square plate n = 0.95.
mass with the sum of the deformations of intact rock and fill Assuming a circular plate of 60 cm dia. and υj = 0.2
material, one obtains under a normal stress, σ̂n as
For a plate diameter of 50 cm.
Note: Ejf values are usually high; they need to be reduced by

a factor more than 2.
Fossum (1985) derived a constitutive model for an equiva-
lent elastic continuum to represent rock mass having randomly
oriented joints with constant Kn , Ks and joint spacing. Pande
By considering sg /Eg as 1/Kn (1993) presented an elastoplastic formulation of a constitutive
model for rock mass with two sets of filled joints.
12 ON EQUIVALENT CONTINUUM MODEL – A

Ej could be estimated with joint factor, Jf or if the thickness of SUGGESTION
gouge in sufficient, one dimensional compression test gives
the value of modulus. Equivalent continuum model (ECM) is being applied to rock
masses for quite some years now. It is not well defined when
to apply ECM and/or Discrete element model (DEM). If a
jointed mass is to be treated as a continuum, it is implied that
the mass is homogeneous equivalent. Neither anisotropy, size
(iii) Ej and G j for three orthogonal joints
nor extent of joints matters any more. There must be some
Assuming again the thickness of joints as negligible for the
minimum relationship between the extent of excavation (say
case of rock mass with three orthogonal tight joints, Kulhaway
tunnel or chamber size) to the spacing of joints/block size.
(1978) presented the following well known expressions, based
Whenever a soil containing large sized particles (like gravel,
on the general analytical approach of Goodman et al. (1968).
cobble) is tested in shear, in triaxial and direct shear, the maxi-
The elastic equivalent orthotropic properties of rock mass
mum particle size allowed should not be more than 1/5th of the
in the z-direction (vertical) are
minimum dimension of the test specimen. This is done essen-
tially to obtain reliable and consistent results for the mass to
behave as a continuum. So is the criterion when a proctor com-
paction test is conducted with gravelly material. To find the
field density of a gravelly soil, a sand pouring set up is used.
For the sand to flow freely to fill the cavity, without building
arching within the flowing sand at the vent, the diameter of the
vent is 21.67 times the size of sand particle (0.6 mm) flowing
for better estimation of the cavity volume. Even in the design
122

of silos for easy flow of grains or materials with out develop- 13.1 In direct shear
ing arching, the size of the vent is kept more than 14 times the
(i) Infilled joint: studies conducted by Salas (1968), Lorde
particle size as per Eq. 41 (Durand & Condolios 1976)
(1973) and Infanti & Kanji (1978) on specimen sizes of
4 × 4 m2 , 4.4 m2 & 4 m2 respectively with infilled mate-
rial revealed practically no influence of test specimen size
where, Ae = effective area of the passage, D = diameter of on shear strength. The effect of asperities is submerged.
vent and d = average particle dia. of material allowed to flow. (ii) Continuous joint: Among others Leichnitz & Natau
For an effective area of 90%, say for the case of tunnels dia. (1979) and Bandis et al. (1981) attributed reduction of
of 5, 10 or 15 m, the ratio of effective diameter of tunnel to the shear strength with increase of size of test specimen to
average block size will be 13. 5. Since free flow of rock mass charge in the size, shape and distribution of asperities. If
does not take place and the rock mass is allowed to develop mineralogy and weathering grade change from block to
arching, this ratio would be certainly less than 13. block or location, they may also reflect as size effect.
As for as rocks are concerned, the minimum size (Ex size) (iii) Discontinuous joint: Due to the presence of rock bridges
core dia., 21.5 mm, is about 11 times the maximum size of sand along the joint, the shear strength found to be decreasing
particle (2 mm) as per IAEG classification and 4.52 times with with the increase of test specimen size (Lorde 1973).
respect to Unified Classification System (UCS) of soil, for the (iv) At residual stage: Lorde (1973) also showed that when the
sand size of 4.76 mm. In the case of NX size core as per UCS, rock bridges are broken along the joint during shear and
this ratio would be 11.3. For assessing the P and S waves in reach residual stage, the effect of test specimen size did
rock cores in the lab., the minimum dimension should be not not exist. Barton (1982) also showed constant residual
less than 10 times the wave length and the travel distance of friction angle (φr ) after separating dilation angle (i) in
the pulse in the rock shall be at least 10 times the average the tests sizes of 0.1 m to 1.0 m.
grain size. For in situ stress measurement, using door stopper,
the rock crystal is not supposed to be more than 12 mm which
is 1/4.5 times the rock core diameter. When borehole over 13.2 In uniaxial compression
coring technique is used, the over coring ratio i.e. external
Hoek & Brown (1980) and Herget (1988) showed decrease of
to the internal dia. ratio of the annular core is supposed to
compressive strength with increase of size of the specimen and
be at least 5 for effective stress release. In the field for plate
attributed changes to varying joint geometry, orientation, dis-
load or direct shear tests, the minimum size of plate or test
tribution and mineralogy resulting increase of potential failure
block respectively is 70 cm × 70 cm, most probably keeping
paths. Studies of Natau (1995) in uniaxial compression sug-
in view spacing of joints/bedding planes in most sedimentary
gest decrease of strength and modulus with the increasing size
formations.
of the specimen. The reason would be same as per the above
In underground works, it is desirable that arching should
observations.
be allowed to develop so that the rock around the excava-
tion shares rock load through arching of blocks. Based on the
above mentioned various prevailing practices one may con- 13.3 Field plate load test
sider the ration of size of excavation or loading area to the
block size/joint spacing, to be at least 10 for the rock to be A statistical analysis of the data collected from the litera-
considered as a continuum. For the zone within an extent of ture suggested no scale effect on modulus of deformation for
5 times the block size/joint spacing around the opening/under different rock masses and from different test methods in the
loaded area, DEM may be more appropriate. Between 5 to field, He (1993).
10 times the block size, consideration of a transition between
DEM and ECM may be in order. 13.4 Dilatancy component
In direct shear test for dense mass (close joints), volume
13 SCALE EFFECT ON ROCK MASS expansion and for looser mass volume contraction may occur
during shear. This volume change with shear displacement is
To evaluate the shear strength response of a contact joint or the usually not same for different test blocks in a formation. To
residual strength of gouge material, direct shear test is usually obtain Coulomb shear strength (or friction angle) correspond-
preferred for its simplicity in conducting the test. It is the ing to a plane failure i.e. no volume change, the component of
best option to estimate normal (Kn ) and shear (Ks ) stiffness dilation (expression +ve, contration −ve ) has to be accounted
values of joints. When a direct shear test is conducted, shear is either from forces consideration or from energy consideration
essentionally along a predetermined plane of sliding. A joint as indicated below.
may be continuously close, open, filled with gouge material (i) Force consideration
or has rock bridges. When testing is carried out on cylindrical When cohesion is not present (i.e., cj = 0), shear strength at
specimens in uniaxial or in triaxial compression, all joints will failure is given by
not undergo shear simultaneously.The critical joint orientation
(45 − φ/2)◦ with the major principal stress direction controls
the response. The shearing joint may not be the same in the
direct shear and triaxial tests. Field shear tests are usually where, σ́ n = normal effective stress on plane of sliding
carried on blocks of specimen less that 1 × 1 m2 . To extend φc = Coulomb’s friction angle on a plane
the results to a large area/joint, a number of studies have been i = dilation angle, tan−1 (v/h), +ve for expansion,
carried out to understand the scale effect on shear strength −ve for contraction
and modulus of rock mass. A brief review of the outcome is v = horizontal displacement, at failure
presented. h = vertical deformation, at failure.
123

(ii) Energy consideration Barla, G. (1993). Case study of rock mechanics in Masua mine, Italy.
When cohesion is present, Comprehensive Rock Engineering, 5:291–334, Pergamon Press
Ltd., UK.
Barton, N. (1982). Modelling rock joints behavior from in situ block
test: implications for nuclear waste repository design, Office of
Nuclear Waste Isolation, Columbus, OH, 96 p, ONWI -308, Sept.
Barton, N. (1990). Scale effects or sampling bias? Proc. 1st int. work-
shop on scale effects in rock masses, Loen, Norway, AA Balkema,
Rotterdam, pp. 31–55.
where, τf = failure shear stress measured Barton, N. (2002). Some new Q-value correlations to assist in site
τc = shear stress with no volume change as per Coulomb characterisation and tunnel design. Int. J. Rock Mech., Min. Sci.
σ́ n = normal stress, effective on plane of sliding Geomech. Abstr., 39: 2: 185–216.
v/h = maximum slope of vertical and horizontal Barton, N. (2008). Important aspects of petroleum reservoir and
displacements curve. crustal permeability and strength at several kilometers depth. Proc.
A plot of τf & τc with σ́ n may be established from the block ISRM Int. Symp & 5th ARMS, Tehran, 1:23–36.
shear tests. To apply the results of for a larger area than the test Barton, N. Lien, R & Lunde, J. (1974). Engineering classification of
blocks, measure of tan−1 (v/h) is obtained either from the rock masses for the design of tunnel support. J. Rock Mech., 6: 4:
spread of stereo plot of the joint or from the measure of the 189–236.
waviness and applied to τc to obtain shear strength at failure. Bieniawski, ZT. (1973). Engineering classification of jointed rock
masses. Trans. S. African Instn. Civ. Engrs., 15: 12: 335–344.
Curves of τf & τc with h could also be plotted for estimation Bieniawski, ZT. (1976). Rock mass classification in rock engineer-
of shear stiffness. ing. Proc. Symp. Exloration for Rock Engineering, (Ed. Z. T.
Bieniawski), A.A. Balkema, Rotterdam, 1: 97–106.
Bieniawski, ZT. (1989). Engineering Rock Mechanics Classification,
14 CONCLUSIONS Wiley-Interscience, New York.
Bronshteyn, VI. Zhukov, VN. Yufin, SA. Zertsalov, MG &
A critical examination of the most commonly adopted rock Ustinov, DV. (2007). Proc. 11th ISRM Congr., 2:1015–1018,
mass classifications, namely, RMR, Q & GSI, has revealed Lisbon.
that the compressive strength and modulus values suggested Brown, ET & Trollope, DH. (1970). Strength of model of jointed
need some definite modifications based on the modulus ratio rock. J. Soil MecFound Div., ASCE, 96: SM2: 685–704.
criterion, which defines the quality of rock mass. In practice Clark, IH. (2006). Simulation of rock mass strength using ubiqui-
tous joints. Proc. 4th Int. FLAC Symp. on Numerical modeling
as well, the modulus ratios of rock masses have been found to in Geomechanics. Paper: 08–07, (R Hart & P Varona, eds),
be much higher than those of intact rocks. Predicted deforma- Miniapolis.
tions did not agree with the field measured values. Application Deere, DU. (1964). Technical description of rock cores for engineer-
of joint factor, Jf , to solve some field problems and prediction ing purposes. Felsmechanik and Inge-nieurgeologie, 1: 1: 16–22.
of the response of laboratory tests seem to be encouraging. Deere, DU. & Miller, RP. (1966). Engineering classification and
The relationships based on Jf for rock masses appear to be index properties for intact rocks. Tech. Report no. AFNL – TR –
more realistic since these are based on experimental verifica- 65 – 116, Air Force Weapons Laboratory, New Mexico.
tions. Estimation of Kn & Ks for different situations of joint Desai, CS. (1994). Hierarchical Single Surface and Disturbed State
system for numerical analyses has been presented. Equivalent Constitutive Models with Emphasis on Geotechnical Applica-
continuum model may be applied when the excavation or load- tion, Chapt. 5 in Geotechnical Engineering, K.R. Saxena (Ed.),
Oxford & IBH Pub. Co., New Delhi, India.
ing area is more than 10 times the joint spacing. When this Desai, CS. (1997). Manual for DSC-SST-2D: Computer Code for
ratio is less than 5, discrete continuum model may be more Static and Dynamic Solid, Structure and Soil-Structure Analysis,
appropriate to adopt; a transitional approach may be suitable Tucson, Arizona, USA.
between these limiting boundaries. Duncan, JM. & Chang, CY. (1970). Non-linear analysis of stress and
Finally, scale effect may not be present in various conditions strain in soil. J. Soil Mech. Found. Engg. ASCE, 5:1629–1652.
of joints, if dilatancy component is suitably separated to assess Duran, R. & Condolios, E. (1956). A flow of granular material
shear strength and Ks values. through a circular orifice, Revue Travaux, March. (referred in
Silos –Theory and Practice, by Reimbert ML and Reimbert AM,
Lavoisier publishing Inc. New York, 1987).
ACKNOWLEDGEMENTS Fossum, AF. (1985). Technical note: Effective elastic properties of
a randomly jointed rock mass. Int. Jnl. Rock Mech. & Min. Sci.
22:6:467–470.
Many thanks to Prof. K G Sharma, Prof. T G Sitharam and Garaga, A & Latha, GM. (2010). Intelligent prediction of stress –
Prof. G. Madhavi Latha for sharing thoughts and discussions strain response of intact and jointed rocks. Computers and
during the preparation of this article. Geotechnics, 37:629–637.
Goodman, RE. (1989). Introduction to Rock Mechanics, 2nd ed.
John Wiley & Sons, New York.
REFERENCES Goodman, RE, Taylor, RL & Brekke, TL (1968). A model for the
mechanics of jointed rocks. Jnl. ASCE, 94: SM3: 637–659.
Arunakumari, G & Latha, GM. (2007). Effect of joint parameters Gupta, AS & Rao, KS. (2000). Weathering effects on the strength
on stress – strain response of rocks. Proc. 11th Int. Congr. ISRM, and deformational behaviour of crystalline rocks under uniaxial
1:243–246, Lisbon. compression state. Int. J. Engg. Geol. 56: 257–274.
Bandis, SC. (1990). Scale effects in the strength and deformability Hart, RG. (1993). Introduction to distinct element modeling for rock
of rocks and rock joints. Proc. 1st Int. workshop on scale effects in engineering. Comprehensive Rock Engg., 2:245–261, Pergamon
rock masses, Loen, Norway, AA Balkema, Rotterdam, pp. 59–76. Press Ltd; UK.
Bandis, SC. Lumsden, AC. & Barton, N. (1981). Experimental stud- He, J. (1993). A case review of deformation modulus of rock mass:
ies of scale effects on shear behavior of rock joints. Int. Jnl. Rock scale effect. Proc. Scale Effects in Rock Mechanics’ 93, (ed) Pinto
Mech. Min. Sci., 18:1–12. da Cunha, A A Bal Kema, Rotterdam
124

Herget, G. (1988). Stresses in Rock, A A. Balkema. Rotterdam. Salas, IAJ. (1968). Mechanical resistances. Proc. 1st. Int. Symp. on
Hoek, E. (1994). Strength of rock & rock masses. ISRM News Rock Mech., Madrid, II–115–130.
Journal, 2(2): 4–16. Sari, M. Karpuz, C & Ayday, C. (2010). Estimating rockmass prop-
Hoek, E & Brown, ET. (1980). Underground Excavations in Rocks. erties using Monte Carlo simulation: Ankara andesites. Computer
Instn. Min & Metl., London. & Geoscience, 36: 959–969.
Hoek, E & Brown, ET. (1997). Practical estimates of rock mass Serafim, JL & Pereira, JP. (1983). Consideration of the geomechanics
strength. Int. Jnl. Rock Mech. & Min. Sci., 34:8:1165–1186. classification of Bieniawski. Proc. Int. Symp. on Engg. Geology
Hoek, E & Diederichs, MS. (2006). Empirical estimation of rock and Underground Construction, Lisbon, Portugal, pt. II, pp. 33–44.
mass modulus. Int. Jnl. Rock Mech. Min. Sci. 43:2: 203–215. Singh, M. Rao, KS & Ramamurthy, T. (2002). Strength and defor-
Hoek, E & Moy, D. (1993). Design of large power house caverns in mational behaviour of a jointed mass. J. Rock. Mech. Rock Engg.,
weak rocks. Comprehensive Rock Engineering, 5:85–110. 35: 1: 45–64.
Horii, H. Yoshida, H. Uno, H. Akutagawa, S. Uchida, Y. Morikawa, Singh, M. Singh, B. Samadhya, NK & Choudhary, J. (2007). Critical
S. Yambe, T. Tada, H. Kyoya, T & Fumio, I. (1999). Comparison strain: A measure of squeezing problem in tunnels. Proc. 11th Int.
of computational models for jointed rock mass through analysis Cong. ISRM, 2:815–818, Lisbon.
of large scale cavern excavation. Proc. 9th Int. Cong. Rock Mech., Sitharam, TG & Madhavi Latha, G. (2002). Simulation of excava-
Paris, 1: 389–393. tion in jointed rock masses using practical equivalent continuum
Infanti, N & Kanji, MA. (1978). In situ shear strength, normal & model. Int. J. Rock Mech. Min. Sci., 39: 517–525.
shear stiffness determination at Agua Vermelha project. Proc. Intl. Sitharam, TG & Maji, VB. (2007). Slope stability analysis of a
Congr. of IAEG, Madrid, 2:175–183. large slope in rockmass: A case study. Proc. 11th Int. Congr. of
Kulhaway, FH. (1978). Geomechanical model for rock foundation ISRM,2:1185–1188.
settlement. Jnl. ASCE, 104: GT 2: 211–227. Sitharam, TG. Maji, VB & Varma, AK. (2005). Equivalent contin-
Londe, P. (1973). The role of rock mechanics in reconnaissance of uum analyses of jointed rock mass. 40th US Rock Mech. Symp.,
rock foundation. Qrtly. Jnl. Engg. Geol., 6:1. 25–29 June, Anchorage, Alaska, paper no. 05–776 (in CD ROM).
Leichnitz, W & Natau, O. (1979). Influence of peak strength deter- Sitharam, TG. Sridevi, J & Shimizu, N. (2001). Practical equivalent
mination of the analytical slope stability. Proc. 4th Int. Congr. of continuum characterization of jointed rock masses. Int. J. Rock
ISRM, Montreux, 2: 335–341. Mech. Min. Sci., 38:437–448.
Moretto, O. Pistone, RES. & DelRio, JC. (1993). A case history Sjoberg, S. (1999). Analysis of large scale rock slopes. Doctoral
in Argentina – Rockmech. For underground works in pump stor- thesis, Dept. Civ. Min. Engg. Lulea, Uni. Tech., Sweden.
age development of Rio Grande No. 1. Comprehensive Rock Stabel, B & Samani, FB. (2003). Mashed-e-soloiman Hydroelectric
Engineering, 5: 159–192, Pergamon Press Ltd., UK. Power Project, rock engineering investigations, analysis, design
Natau, O. Fliege, O. Mutcher, TH & Stech, HJ. (1995). True triaxial and construction. Proc. 10th Int. Congr of ISRM, 2: 1147–1154.
tests of prismatic large scale samples of jointed rock masses in Tabanrad, R. (2003). Monitoring and stability analysis of intake tun-
laboratory. Proc. 8th Int. Cong. Rock Mech, Tokyo, 1: 353–358. nels, Karun III hydroelectric power project. Proc. 10th Int. Cong.
Pande, GN. (1993). Constitutive models for intact, rock joints Of ISRM, 2: 1189–1193, Johanesburg.
and jointed rock mass. Comprehensive Rock Engg, 1:427–441. Timoshenko, SP & Goodier, JN. (1970). Theory of Elasticity. 3rd ed.,
Pergamon Press Ltd. UK. McGraw–Hill Book Co., New Delhi.
Pande GN, Beer, G & Williams, JR. (1990). Numerical Methods in Varadarajan, A. Sharma, KG. Desai, CS & Hashemi, M. (2001a).
Rock Mechanics. John Wiley & Sons. Ltd. New York. Constitutive modelling of a schistose rock in the Himalaya,
Ramamurthy, T. (1993). Strength and modulus responses of International Journal of Geomechanics, 1:1:83107.
anisotropic rocks. Chapter 13, Comprehensive Rock Engineering, Varadarajan, A., Sharma, KG. Desai, CS & Hashemi, M. (2001b).
Pergamon Press, U.K., 1: 313–329. Analysis of a powerhouse cavern in the Himalaya, International
Ramamurthy, T. (2001). Shear strength responses of some geological Journal of Geomechanics, 1:1:109–127.
materials in triaxial compression. Int. J. Rock Mech. Min. Sci. 38: Zhou, Y. Zhao, J. Cai, JG & Zhang, XH. (2003). Behaviour of large –
683–697. span rock tunnels and caverns under favourable horizontal stress
Ramamurthy, T. (2004). A geo-engineering classification for rocks conditions. Proc. 10th ISRM Congr., 2:1381–1386. Johanesburg.
and rock masses. Int. J. Rock Mech. Min. Sci. 41: 1: 89–101.
Richards, L & Read SAL. (2007). Newzealand Greywacks charac-
teristics and influences on rock mass behavior. Proc. 11th ISRM
Congr., 1:359–364.
125

Risk oriented design and construction of tunnels
W. Schubert
Graz University of Technology, Austria
ABSTRACT: Accidents, cost- and time overruns at tunnel projects are still rather common. It appears that design and con-
struction procedures still do not comply with the requirements of an up to date engineering approach. Too often dominant
behaviors during the design process are ignored or the models simplified in a way that critical mechanisms cannot be detected.
To improve the situation, design and construction processes need to be adopted, which focus on the inherent risks of tunneling.
The paper presents such an approach, which has been used successfully over the last decade.
Subject: Analysis techniques and design methods
Keywords: tunneling, risks and hazards, monitoring
1 INTRODUCTION
In terms of the insurers language tunneling is a bad risk. Too

many costly accidents not only are a source of headache for
the insurance companies, but also create a bad reputation of
the whole industry. The question now is, whether or not we
really use all means and methods available to improve the
situation. When doing design checks, one sometimes does not
have the impression that the persons acting have understood
the problem, but merely do a copy and paste job. In no other
field of engineering such an approach would be accepted. The
reason for this rather unprofessional behavior may be that we
have to deal with a lot of uncertainties. So people may be
tempted to believe that whatever they do, it will be wrong
anyway, so they don’t even attempt to produce a consistent
and traceable design. In many cases they simply follow rather
unspecific recommendations of dubious source.
Analyzing the tunneling problem, it quickly becomes clear
that only a risk oriented approach to tunnel design and
construction will lead to acceptable results. Figure 1. General procedure for risk assessment (from PIARC,
2008).
2 HAZARD AND RISK 3 HAZARDS IN TUNNEL CONSTRUCTION
Hazard is commonly understood as an unfavorable event. Risk In the course of this paper we are only addressing geotechnical
then is defined as the product of the consequences of such and environmental hazards, while hazards during operation
an event and the probability of occurrence. In engineering are not dealt with.
it is understood that a risk assessment process has to start Hazards during tunnel construction may be various modes
with the definition of the system followed by hazard iden- of failure, large displacements, excessive water inflows and
tification, evaluation of the probability of occurrence and a other effects, like intolerable emissions by blasting vibrations
consequence analysis. The next steps include the risk evalu- or dust. Like in other risk assessment processes, also in tunnel
ation and the check whether the risk is acceptable or not. In design those hazards and their consequences should be known
case of an unacceptably high risk, mitigation measures have before deciding on hazard or risk mitigation measures in the
to be implemented and the risk re-evaluated. This is done until form of construction method and supports.
an acceptable risk level is obtained (Figure 1).
It is quite clear that without hazard identification and
4 RISK ORIENTED DESIGN APPROACH
evaluation of the consequences, as well as the determi-
nation of probability of an event, all further steps are
4.1 Basic idea
meaningless.
This principle now can be relatively easily used in tunnel The generation of a cavity in the ground always triggers a reac-
design. tion from the ground. The “quantity” and the sort of reaction
127

Information on expected stresses and ground water condi-
tions supplements the ground model. The geometry of the
underground structures and their location complete the ground
model.
Characterization includes properties of the intact rock and
the joints and – depending on the project phase and rock
type – smeared ground parameters. The expected spread of
the parameters should be indicated to allow for evaluating the
range of behaviors.
One should be aware of the fact that oversimplification by
“homogenizing” the ground can cause severe loss of infor-
mation. This can lead to a misjudgment of behaviors in
subsequent analyses, as will be presented in chapter 3.3.
4.3 Hazard identification/determination of ground behavior

The next step of a risk oriented design approach is the identifi-
cation of hazards during tunnel construction. Possible hazards
(also referred to as behaviors) among others are various fail-
ure modes, excessive ground deformation, changes in the
ground water table, damage to third parties or the environ-
ment by ground deformation, change in ground water regime,
or emissions, such as noise, vibration, and dust.
In a first step the reaction of the ground to excavating the
underground opening without consideration of any additional
construction measures should be assessed. The reason for this
basic step is simple: only if one knows the not influenced
behavior of the ground it is possible to devise appropriate
mitigation measures.
Possible behaviors of the ground roughly can be grouped
into discontinuity controlled, gravity driven and stress induced
modes. Ground water, like stress influences ground behavior.
Figure 2. The design methodology as presented in the Guideline of
Indications on general types of behaviors are given in OeGG,
the Austrian Geotechnical Society (2008), and the interpretation of 2001, OeGG, 2010, Goricki, 2003, Schubert and Goricki,
its different design steps in the light of a standard risk analysis. 2004, Palmstrom and Stille, 2007.
For the evaluation of discontinuity controlled failure modes
depend on the geological and geotechnical conditions, size of a number of convenient programs are available, while for the
excavation, ground structure and influencing factors, making detection of stress controlled failure modes closed form solu-
the act of collecting and interpreting these data clearly a “sys- tions in a first step are acceptable. Only for very complex
tem definition”. The act of determining the possible failure conditions, numerical simulations are required in this step.
modes and their associated magnitudes allows us to identify What definitely is important for an unbiased identification of
hazards, and to start a transparent reasoning along the lines of behaviors is to have a mathematical model or clear criteria for
a sound risk oriented design. Assuming that the allowable risk each behavior category.
levels have been set (dictated by the issues of tunnel safety, When strictly keeping to the risk analysis procedure in this
operability and environmental impact), appropriate risk miti- step also the probability of failure and the consequences of a
gation measures (support measures and construction methods) possible failure have to be evaluated. As the models used for
can be chosen. the determination of the behavior are simple – but should not
Finally, the outcome of their interaction with the ground be simpler than admissible – a probabilistic assessment is not
can be analyzed, allowing identification of the residual risks. much effort.
Once the residual risk reaches an acceptable level, the design 4.3.1 The importance of failure mechanism identification:
can be considered satisfying the requirements. an example
Although the engineer’s first instinct is trying to express every-
thing in numbers, design and construction of underground
4.2 Definition of system/Establishment of ground model and
structures require, at least at the initial stages, a completely
geometry
different approach. Before anything is calculated or put to
An indispensible basis for tunnel design is a sound and con- numbers, one has to reflect on the likely failure mechanisms.
sistent geological model. Owners are not always aware of As already stated, the outcome of this endeavor depends on
the importance of a detailed investigation program. Prefer- the quality of the ground characterization: any kind of bias or
ably already at the beginning of a project the whole team “simplification” due to homogenization lowers the amount of
including designer should be appointed to allow for a targeted information and the quality of the results. Serving both as a
investigation and testing program. clarification of this statement and as an example accompany-
Once the geological setup is known, characterization of the ing the methodology and the approach presented in this paper,
ground has to be performed, and the properties of the ground an “application example” is analyzed and discussed. For the
assigned to the respective ground volumes in the project area. sake of simplicity, we will assume a shallow tunnel through
128

Table 1. Intact rock and discontinuity parameters used in the
Monte-Carlo analysis.
Min. Max.
Joint alteration factor Ja [−] 4.00 8.00
Joint set 1
(vertical)
Joint roughness (small scale) Js [−] 0.75 2.00
Joint roughness (large scale) Jw [−] 1.00 3.00
Friction angle ϕ [◦ ] 15 30
Normal joint spacing [cm] 20 100
Joint alteration factor Ja [−] 2.00 4.00
Joint set 2
Joint roughness (small scale) Js [−] 0.75 2.00
(horiz.)
Joint roughness (large scale) Jw [−] 1.00 3.00
Normal spacing [cm] 20 100
UCS [MPa] 15 50
Intact
rock
Hoek-Brown Constant mi [−] 10 20
Unit weight γ [MN/m3 ] 0.028
Eq. friction angle ϕ [◦ ]
Granite
40 60
Gravel Hom.
Eq. cohesion c [MPa] 0.10 0.25
Cohesion c [MPa] 0.00 0.05
In the first step, the safety factor in case of full-face excava-

tion and no support installation is calculated by a probabilistic
analysis of the limit equilibrium for blocks sliding along the
vertical discontinuity set.
In order to demonstrate the importance of proper character-
ization, the limit equilibrium is also determined for “smeared
joints” and rock mass properties determined by applying the
Figure 3. General categories of ground behavior; from OeGG 2010. relationships proposed by Hoek et al. (2002) and Cai et al.
(2004). The assumed range of the parameters has been chosen
as shown in Table 1, according to the geological description
of the problem.
The results clearly demonstrate the effect of the different
characterization methods on the safety factor (Figure 5 and
Figure 6). The discontinuous approach predicts a definitive
vertical shear failure and downward sliding of blocks along
joints if no support is installed, while the homogenous ground
model implies stable ground with a slight potential for day
lighting failure in case of low confinement stresses. Hence,
not only the estimation of the hazard occurrence probability
is wrong, but also wrong risk mitigation measures could be
the result.
When assessing the consequences, one has to distinguish
between behaviors causing only a cost increase, and such
being a hazard for the overall stability of the tunnel, or lead-
ing to unacceptable influence on the environment, like large
displacements. Structurally controlled, limited overbreak or
spalling with limited depth may be geotechnically accept-
able, while other behaviors with severe and unacceptable
consequences call for mitigation measures.
Figure 4. The assumed geological set-up and probable failure mode.
4.4 Definition of requirements and boundary conditions

jointed and weathered granite. Three perpendicular joint sets
are intersecting the rock mass, with the vertical joint sets hav- Prior to selecting a construction concept, the requirements
ing strongly weathered and open joints with a silt/clay filling. associated with the underground opening have to be identi-
The horizontal joint set is assumed to be less weathered and fied. They include requirements with respect to stability and
closed due to vertical loading. Three meters thick, neogenous durability, environmental restrictions, etc. Rules, guidelines
layer of gravel is assumed above the crystalline rock mass. and standards vary from country to country, and thus should
The sketch of the geological conditions is shown in Figure 4. be clarified prior to the continuation of the design process.
129

Figure 5. Calculated safety factors against day lighting failure, low
confinement pressure.
Figure 7. Applicable support concept (top heading with shotcrete

lining and systematic rock bolting in the sidewalls).
In jointed rock mass with the strike of one or two joint

sets parallel to the tunnel axis, where a potential failure mode
is shearing along the joints, bolts can efficiently reduce the
failure potential, while linings might shear as well. In some
cases, where large blocks can slide into the tunnel from the
sidewalls, a sequential excavation with installation of support
in the top heading prior to benching can considerably reduce
the size of the potential failure block. In case of ground water
considerably decreasing the strength of the ground, lowering
the water table may be an option. If the lowering of the ground
water table is not acceptable due to environmental reasons or
the potential of creating excessive surface settlements, exca-
vation under compressed air, grouting, freezing, or excavation
with an appropriate shield TBM are feasible options.
Figure 6. Calculated safety factors against day lighting failure, 4.5.1 Appropriate support for the identified failure mode:
higher confinement stress.
an example
The process of application of mitigation measures shall be
Boundary conditions here are defined as non-technical
demonstrated with the help of the previously shown exam-
influencing factors, such as specific regional experience, or
ple. As the probable failure mode without support would be
availability of materials, etc.
a shearing along the vertical joints, rock bolting in the side-
Project specific requirements, regulations, and boundary
walls and shoulder areas, in order to intersect and “dowel” the
conditions make it obvious that there cannot be a universally
sliding surfaces would be a logical measure (Figure 7).
applicable design concept. Each case has to be dealt separately
If the sliding portion of the rock mass is sufficiently sup-
with, and carefully engineered solutions developed to arrive at
ported by the rock bolting, the excavation of the bench is
an economical, safe and environmentally acceptable design.
rendered an uncritical endeavor, disregarding the bench round
length or top heading feet abutment capacity. When basing
4.5 Hazard mitigation and risk reduction the decision on the support on the results of the homogenized
Once hazards, their probability of occurrence, and the conse- model, one would probably think that a thin shotcrete lining
quences are identified, construction concepts can be devel- would be sufficient to prevent single blocks from falling into
oped. The aim of each construction measure is to prevent the opening. Needless to say, the “true” failure mode in combi-
or reduce the probability of failures or their consequences, nation with the wrong support would dangerously compromise
and obtain a system behavior meeting the project specific the overall stability.
requirements.
Efficiently mitigating hazards and thus reducing risk
4.6 Determination of system behavior
requires the knowledge of potential behaviors, thus empha-
sizing the importance of identifying ground behaviors and Once the range of feasible construction methods and support
potential failure modes. measures for each scenario under consideration of expected
The process of selecting appropriate construction measures ground behavior, requirements, and boundary conditions has
is rather complex, as aspects of costs and labor safety have to been determined, the interaction between the ground and the
be considered in addition to the purely geotechnical aspects. construction measures needs to be analyzed.
130

Figure 9. Probability density distributions of the displacement
development compared to the monitoring data (lines). From Radoncic
(2011).
Figure 8. The effect of support measures on the safety factor, as

obtained from probabilistic limit equilibrium analysis.
The focus of the analyses is on overall stability of the struc-

ture, deformations, and influences on other structures or the
environment. If only the stability of the overall concept has
to be examined, then simple analytical methods can be used
very successfully, whereas the issue of predicting the surface
settlements has to be treated by numerical methods.
The effect of the support measures in the example presented Figure 10. Effect of the advance rate on the shotcrete utiliza-
before has been determined by a probabilistic analysis once tion (same ground conditions and support concept as used for the
MC-simulation results shown in the previous figure).
again. The interaction between the installed rock bolts and the
ground is analyzed by including the simple analytical model
(Goricki et al, 2005, Panet & Guenot, 1982, Pilgerstorfer,
provided by Pellet & Egger (1996) into the relationships used
2009, Hoek et al, 2008). Information has to be provided also
for determining the ground behavior.
on the utilization of the supports in relation to advance rate and
It captures the rock bolt contribution to the sliding resis-
time. This is important in particular when using shotcrete in
tance both by axial elongation and shear (dowel) action. A
poor ground, as high advance rates cause also high displace-
regular shotcrete lining (thickness 10–15 cm) would be suffi-
ment rates, which might lead to overstressing of the young
cient as sealing, preventing smaller blocks from detaching and
shotcrete (Figure 10).
falling down. Figure 8 shows the results of the probabilistic
analysis of the model with rock bolt installation in sidewalls
and shoulders. 4.8 Check system behavior on compatibility with
requirements
4.7 Required information on system behavior As construction methods and measures in the previous step
have been determined in a semi empirical way, there is a certain
To be able to judge during construction, whether the observed probability that the chosen solution is off the optimum, or that
system behavior is as expected, it has to be defined dur- the system behavior does not comply with the requirements.
ing the design. Information to be provided does not only Thus it is mandatory to check the compatibility of the system
include the absolute magnitude of displacements, but also behavior with the requirements.
their development with face advance and time, as well as the In case the check shows that there is need for modifica-
characteristics. tions in the construction concept (for example construction
Figure 9 shows the result of a probabilistic analysis of sequence, timing of support, etc,) or supports in quantity
top heading excavation in weak ground, with a support con- and/or quality, those modifications have to be made and
cept featuring deformation gaps and yielding elements. The the system behavior checked again and compared to the
likely (and thus, assumed) range of ground properties has been requirements.
determined by an extensive in-situ test program (Pilgerstorfer,
2011) and used as a basis for a Monte-Carlo simulation cou-
4.9 Optimize construction concept and evaluate residual
pled with a calculation based on the convergence confinement
risk
method (Radoncic, 2011). As it can be seen, the frequency
plots of the respective longitudinal displacement profiles are While previous steps in the process were mainly focused on
in excellent agreement with the monitoring data (Figure 9). solving technical problems, the overall construction concept
Once again, this implies that relatively simple methods, if of a project needs to be considered as well to arrive at a sat-
applied to the correct problem, yield very reliable and sound isfactory result. Changes of construction methods within a
predictions of the system behavior. project only in exceptional cases will be feasible due to time
It is well known that the ground structure and the ground and cost considerations. This simply means that in some cases
utilization play an important role in the development of dis- non-optimal construction concepts have to be used for limited
placement characteristics and their longitudinal development sections of a tunnel. As an example a project may serve, where
131

Figure 11. Expected spread of costs when comparing two construc-
tion methods at an Alpine tunnel (Goricki, 2003).
an open TBM could be used on most of the length. However,

a fault zone of some tens of meters thickness would cause a Figure 12. Steps during construction to determine excavation and
serious hazard for blocking the machine. In this case addi- support and control system behavior (OeGG, 2001).
tional measures, like improvement of the ground ahead of the
TBM, or conventional excavation via a bypass would have to
be applied.
Usually again various options for the overall project layout
exist. Finding the optimal solution requires consideration of
technical aspects, budget, construction time, as well as resid-
ual risks associated with the different options. Residual risks
originate from the uncertainties in the ground model and actual
ground behavior, as well as simplifications made during the
design. For providing a sound basis for the decision, which
option eventually should be applied, quantification of the risk
in terms of costs and time is required. This process again
calls for a probabilistic approach. Such methods have been Figure 13. 3D image of tunnel face with measurement of the
successfully used on various projects for feasibility studies, discontinuity orientations (courtesy of 3GSM).
preliminary and tender designs (Goricki et al, 2003b; Pötsch
et al, 2004). 5.2 Collection of geological information
Figure 11 shows the result of a comparison of conven-
tional drill and blast and TBM excavation at an Alpine tunnel The system behavior to a great extent is influenced by the
project of some 23 km of length. In this project some major ground surrounding the tunnel. In order to be able to apply
fault zones were expected, requiring occasional conventional appropriate construction methods with minimal risk, the
excavation in addition to the TBM excavation. Shown are the ground structure and quality in a representative volume around
expected cost ranges for the two methods under consideration the tunnel has to be known. This means that the so called face
of the uncertainties in the ground model. The spread of costs mapping can be only the basis for further modeling. The geo-
is smaller with the TBM excavation due to its relatively low logical modeling is supported by appropriate evaluation of the
flexibility with respect to support, but at higher average costs. monitoring data, as will be shown later.
It has to be mentioned that this refers to a special case, and The quality of the face mapping can be significantly
the results cannot be generalized. increased, when image based recording techniques are used
(Gaich et al. 2005). In this way information is scaled, complete
and unbiased.
5 APPROACH DURING CONSTRUCTION 5.3 Decision on excavation and support

5.1 General approach After establishing the ground model around the tunnel, poten-
tial hazards for the next excavation steps have to be identified.
Even with a very extensive investigation program, a sound This includes identification of the previously defined ground
characterization, and an up to date engineering design, uncer- types and influencing factors, just as it has been done during
tainties in the ground model, the ground properties and the the design phase, but now with more accuracy, as the level of
behavior remain. information is significantly increased.
As a consequence, the design process needs to be extended
into the construction phase, with a continuous update as
5.4 Prediction of system behavior
more information becomes available. Continuous collection
of information and monitoring are essential preconditions As the majority of the decisions, like required profile size,
to optimize construction, and reduce risks. Generally this is thickness of lining, and round length cannot be changed after
referred to as the observational approach. excavation, a clear prediction of the expected system behavior
132

Figure 15. Comparison of longitudinal deformation profile of two
Alpine tunnels. Note the large influence length of the face advance,
where the strike of the foliation was parallel to the tunnel axis.
part containing the detailed regulations for different sections

Figure 14. Predicted displacements for a top heading-bench exca- with different ground or boundary conditions.
vation with temporary top heading invert (GeoFit). In the general part of the Safety Management Plan the
function, addresses, phone and e-mail numbers of all parties
involved in the project and their role and responsibilities are
has to be made prior to excavation. Basis for this prediction on listed. Also listed are the contact details of institutions and
the one hand is the design, on the other hand the experience organizations, which have to be informed in the event of a
gained from situations under similar conditions. Again here, crisis, where third parties are at risk.
as for the prediction of the geological situation, the results of Generally the requirements for a successful risk manage-
monitoring can be very helpful. It is well known that besides ment on site can be listed as follows:
ground quality, structure and other influences, like stresses and
ground water, the face advance, time, construction sequence • Definition of expected behavior of the underground struc-
and support dominantly influence the development of the dis- ture in all construction stages
placements. Panet et al (1982) and Sulem et al (1987) have • Definition of the parameters to be observed
developed a relatively simple relationship for the development • Definition of monitoring methods, locations, amount, and
of the displacements in relation to face advance and time. Fur- reading frequency
ther development by Barlow (1986) and Sellner (2000) now • Definition of warning and alarm levels
allows considering sequential excavation and different types • Establish organization required to conduct monitoring, data
of support. Figure 14 shows an example of the predicted dis- handling, evaluation and interpretation and communication
placements for a road tunnel with sequential excavation and structure between the parties involved
temporary top heading invert. • Devise procedures and measures in case the behavior
As already mentioned earlier, the discontinuity orientation deviates from the expected
in relation to the tunnel axis plays a dominant role not only • Establish procedure for the case of a crisis, including alarm
for the behavioural mode, but also for the development of dis- criteria, organization, and protection priorities
placements. In most cases, when the dominant discontinuities • Define what can be changed on site without involvement
strike parallel to the tunnel axis, the length influenced by the of the designer
excavation is much larger than in cases, where the foliation ori- • Continuously update the prediction made during the design,
entation is perpendicular to the tunnel axis. Figure 15 shows as new information is available
an example of the influence of the orientation of the foliation The designer defines the expected system behavior of the
on the development of the displacements. While at the Inntal- tunnel during the excavation and the influence of the tunneling
tunnel with the foliation orientation in an oblique angle to the works on surface and subsurface structures being influenced
tunnel axis, displacement rates decreased at a short distance for the single sections of the tunnel. This serves as a basis
behind the face, at the Strenger tunnel with the foliation ori- for the geotechnical engineer on site, who has to evaluate the
entation parallel to the tunnel axis, displacements continued monitoring results.
over a long distance from the face. Final displacements for the
top heading in both cases were around 300 mm. To be able to
interpret the monitoring results, the expected behaviour under 5.6 Monitoring
the actual conditions must be defined. This includes besides
the effects of discontinuities on the longitudinal deformation For minimizing risks during construction, monitoring of the
profile, also their influence on anisotropic displacements. system behavior is mandatory. Fortunately during the last two
decades, monitoring, evaluation, and interpretation techniques
have been advanced considerably. In particular the introduc-
5.5 Safety management plan tion of measurements of absolute displacements has created
a tremendous increase in information quality, compared to
To be prepared for deviations from the expected behavior, a the traditional convergence meter measurements. Parallel to
safety management plan has to be established, which serves the improvement of the measurement technique, also data pro-
as basis for action in case of deviating behavior. cessing, evaluation, and interpretation were further developed.
The geotechnical safety management plan consists of two A few examples of the potential of measurement data will be
parts, the first being valid for the whole project, the second shown.
133

Figure 16. Measured displacements and predicted displacements
(see Figure 14) show good agreement for a sequential excavation
Figure 18. Displacement vectors plotted in the cross section, reveal-
with temporary top heading invert.
ing strong influence of a fault zone at the right side outside the
excavation area.
Figure 19. Deviation of the displacement vector orientation before

excavation reaches a fault zone.
Figure 17. Monitoring data plotted in the traditional way of
convergence between single points. When plotting the displacement development as vectors
in the cross section, it can be clearly seen that the displace-
5.6.1 Compare observed to predicted system behavior ment due to a fault zone outside the right sidewall is strongly
The traditional method of evaluating monitoring data consists anisotropic. The situation easily could escalate and lead to a
of observing the development of displacements over time. The tunnel collapse. With the early warning by appropriate evalu-
general criterion applied is that the displacement rate should ation of the monitoring results, targeted mitigation measures
decrease with ongoing time. The setback of this approach is in the form of additional bolting at the right sidewall could be
that in most cases the face advance dominates the displace- successfully implemented in time.
ments rather than the time. Thus it is advisable to plot the
displacements versus face advance rather than against time. 5.6.3 Detection of geological structures ahead of the face
Even better is to use software (for example GeoFit), which Recording of spatial displacements and appropriate evalua-
combines effects of face advance and time, allows predicting tion of the measured data cab also efficiently help in detecting
the displacements and comparing the measured values to the changing ground conditions ahead of the face. The changing
predicted. This allows detecting deviations from the “normal” stress situation in the ground around the tunnel also leads to a
behavior very quickly, even when advance rates are unsteady change in the displacement characteristics. It has shown that
and different excavation steps influence displacements at the by evaluating the trend of the displacement vector orienta-
same monitoring section. tion, features outside the tunnel with different properties can
be detected (Schubert et al. 1995, Budil, 1996, Steindorfer,
5.6.2 Detection of geological structures outside 1998, Grossauer, 2001, Jeon et al, 2005, Baek et al, 2006,
the visible area Mösslacher, 2006). This can be illustrated with the example
Experience shows that the majority of the problems during in Figure 19, showing the trend of the displacement vector ori-
tunneling are caused by structures, like slickensides or faults, entation significantly deviating from the normal range several
which are situated outside the tunnel. Early warning about tens of meters ahead of the fault zone, which was first visible
the existence of such features generally is not possible with in the excavation area at around station 995.
traditional monitoring methods, while this is rather easy, when
recording spatial displacements. For illustration, measurement 5.6.4 Evaluating utilization of lining
data have been evaluated in the traditional way, by plotting In particular for tunnels with shallow overburden in soils
the convergence between the sidewalls and the sidewalls and and urban areas, the shotcrete lining is the main supporting
the crown (Figure 17). The development of the displacements element. Failure of the lining in many cases has severe conse-
appears rather “normal” und would not trigger any alarm. quences, reaching from large displacements to total collapse
134

Figure 22. Distribution of expected factor of safety for an advance
rate of 4 m/day (left) and 2 m/day (right).
All examples shown above are from the daily practice and
not academic. Precondition for successful monitoring and
using the results to the benefit of the project and risk reduc-
Figure 20. Utilization of the lining of an urban railway tunnel with tion is, besides quality monitoring, the awareness of owner
multiple drifts (Moritz et al, 2008). and contractor of the benefits.
6 CONCLUSIONS
The increasing demand for underground structures both in

urban areas and for infrastructure projects, like high speed rail
lines calls for state of the art design and construction practices.
Accidents, cost and time overruns at many projects indicate
that those practices are not always implemented. In some cases
the inherent uncertainties are taken as an excuse to keep the
level of engineering low and to rely on experience, how ques-
tionable and applicable for the project at hand this experience
may be. In other cases apparently sophisticated models are
used to produce colourful plots, apparently indicating accu-
Figure 21. Predicted range of displacements for an advance rate of
racy. Often those models do not come close to reality, as during
4 m/day.
the modelling process important features or properties have
been smeared into a continuum, not allowing to model realistic
behavior.
with day lighting failure. To minimize the probability of lining Attempting to reduce the risk of tunnel construction there is
failure thus is an important issue. no way around applying sound engineering practices, always
From absolute displacement measurements the strain in the keeping in mind the peculiarities of an underground project
lining can be calculated, and using an appropriate material with its inherent uncertainties. A key issue for success is the
model for the shotcrete the stresses can be evaluated and com- understanding of mechanisms and potential failure modes.
pared to the actual strength (Schubert, 1988, Hellmich, 1999, This is a precondition for all further design steps, including
Macht, 2002). selection of modelling and construction methods.
The utilization of the lining can be checked during construc- Focussing on behaviours and using practices of risk analysis
tion also for predicted displacements. Feeding the predicted and management not only leads to safer and more economi-
displacements into the utilization analysis, allows studying cal underground construction, but also satisfies engineering
future lining stresses for assumed progress rates. When the standards.
development of the displacements cannot be predicted with The means and methods to do so are available. We only need
high accuracy, the expected range should be estimated. shows to use them.
an example of such a prediction of the expected range of
displacements for an advance rate of 4 m/day.
A probabilistic approach then allows assessing the proba- REFERENCES
bility of lining failure.
Analyzing the distribution of the expected range of safety Aldrian, W. 1991. Beitrag zum Materialverhalten von früh belastetem
factors for an advance rate of 4 m/day, leads to a relatively Spritzbeton. Dissertation. Montanuniversität Leoben
high probability of failure (Figure 22, left). Baek, S.H., Kim, C.Y., Kim, K.Y., Hong, S.W., Moon, H.K. 2006.
As this is not acceptable, various options exist. One of the A Numerical Study on the Effect of Heterogenous/Anisotropic
options is to increase the support. Another option would be to Nature of Rock Masses on Displacement Behaviour of Tunnel.
reduce the advance rate to allow for the shotcrete to set. The Tunnelling and Underground Space Technology 21(2006) pp. 391–
396. Elsevier
effect of a reduction of the advance rate to 2 m/day is shown in
Barlow, J.P. 1986. Interpretation of tunnel convergence measure-
Figure 22, right. It appears that the probability of lining failure ments. MSc thesis, University of Alberta
is now in an acceptable range. Budil, A. 1996. Längsverformung im Tunnelbau. PhD thesis at the
The reduced advance rate causes smaller displacement Graz University of Technology
increments per time, while the shotcrete develops more Cai, M. Kaiser, P.K., Uno, H., Tasaka, Y. & Minami, M. 2004. Esti-
strength. mation of rock mass deformation modulus and strength of jointed
135

hard rock masses using the GSI system. Int. J. of Rock Mech. & OeGG, 2001. Richtlinie für die geomechanische Planung von
Min. Sci. 41 (2004). 3–19 Untertagebauarbeiten mit zyklischem Vortrieb. Austrian Society
Cai, M. Kaiser, P.K., Uno, H., Tasaka, Y. & Minami, M. 2004. Esti- for Geomechanics, Salzburg
mation of rock mass deformation modulus and strength of jointed OeGG, 2010. Guideline for the geotechnical design of underground
hard rock masses using the GSI system. Int. J. of Rock Mech. & structures with conventional excavation. Austrian Society for
Min. Sci. 41 (2004). 3–19 Geomechanics, Salzburg
Gaich, A., Schubert, W. & Poetsch, M. 2005. Three-dimensional Palmstrom, A., Stille, H., 2007. Ground behaviour and rock engi-
rock mass documentation in conventional tunnelling using Joint- neering tools for underground excavations. Tunnelling and Under-
MetriX3D. In Yücel, E., Solak, T. (eds), Underground Space Use: ground Space Technology 22 (2007) 363–376
Analysis of the Past and Lessons for the Future; Proc. 31st ITA- Panet, M., Guenot, A. Analysis of convergence behind the face of a
AITES World Tunnel Congress, Istanbul, 7–12 May 2005: 59–64. tunnel. Tunnelling 1982. The Institution of Mining and Metallurgy,
A.A. Balkema Publishers. 197–204.
GeoFit. http://www.3-g.at/geofit/en/home.php Panet, M., Guenot, A. Analysis of convergence behind the face of a
Goricki, 2003. Classification of Rock Mass Behaviour based on tunnel, Tunnelling 1982, The Institution of Mining and Metallurgy,
a Hierarchical Rock Mass Characterization for the Design of 197–204
Underground Structures. Doctoral thesis, Graz University of Pellet, F., Egger, P. 1996. Analytical model for the mechanical
Technology behaviour of bolted rock joints. Rock Mechanics and Rock
Goricki, A., Button, E., Schubert, W., Poetsch, M. & Leitner, R. 2005. Engineering 29 (2), pp. 73–97. Springer Verlag, 1996.
The Influence of Discontinuity Orientation on the Behaviour of PIARC, 2008. Risk Analysis for road tunnels
Tunnel. Felsbau 23 (5), 12–18, Verlag Glückauf: Essen Pilgerstorfer, T. & Radoncic, N. 2009. Prediction of spatial displace-
Hellmich, C. 1999. Shotcrete as part of the New Austrian Tunnelling ment development. Geomechanik und Tunnelbau 2 (3), 250–258,
Method: from thermochemomechanical material modelling to Ernst & Sohn: Berlin
structural analysis and safety assessment of tunnels. Dissertation, Radoncic, N. 2011. Tunnel design and prediction of system behaviour
Institute for Strength of Materials, Wien University of Technology in weak ground. PhD Thesis, Institute for Rock Mechanics and
Hoek, E., Carranza-Torres, C. & Corkum, B. 2002. Hoek -Brown Tunnelling, Graz University of Technology. 2011.
Failure Criterion – 2002 Edition. In: Proceedings of the fifth North Schubert, P. 1988. Beitrag zum rheologischen Verhalten vom
American rock mechanics symposium, Toronto, Canada, vol. 1, Spritzbeton, Felsbau 6/3, 150–153, VGE Verlag GmbH
2002. pp. 267–73. Schubert, W., Budil, A. 1995. The importance of longitudinal defor-
Hoek, E., Carranza-Torres, C., Diederichs, M.S. and Corkum, mation in tunnel excavation, In: Fujii (ed), Proc. 8th Congress
B. 2008. Integration of geotechnical and structural design in on Rock Mechanics, Vol. 3, Tokyo, Japan, 1995: 1411–1414
tunnelling. Proceedings University of Minnesota 56th Annual Rotterdam: A.A. Balkema
Geotechnical Engineering Conference. Minneapolis, 29 February Schubert, W., Goricki, A. 2004. A Consistent Engineering Approach
2008, 1–53 toTunnel Design. In Logar, J., Gaberc,A. (eds), Proc. 4th Slovenian
Jeon, J.S., Martin, C.D., Chan, D.H., Kim, J.S. 2005. Predicting Geotechnical Conference, Rogaska Slatina, 9–11 June 2004: 25–
ground conditions ahead of the tunnel face by vector orienta- 32. Ljubljana: Slovenian Geotechnical Society.
tion analysis. Tunnelling and Underground Space Technology 20, Sellner, P. 2000. Prediction of displacements in tunnelling. Doctoral
344–355. Elsevier thesis, Graz University of Technology
Macht, J. 2002. Hybrid analysis of shotcrete tunnel linings: assess- Steindorfer, A. 1998. Short Term Prediction of Rock Mass Behaviour
ment and online monitoring of the level of loading. Dissertation, in Tunnelling by Advanced Analysis of Displacement Monitoring
Institute for Strength of Materials, Wien University of Technology Data. In Schubert, Riedmüller & Semprich (eds.) Schriftenreihe
Moritz, B., Matt, R., Graf, F., Brandtner, M. 2008. Advanced Obser- der Gruppe Geotechnik, Heft 1
vation Techniques for Sophisticated Shallow Tunnel Projects – Sulem J., Panet M., Guenot, A. Closure Analysis in Deep Tunnels,
Experience Gained Using Innovative Monitoring Methods at the Int. Journal of Rock Mechanics and Mining Science, 24, 1987,
Lainzer Tunnel LT31. Geomechanics and Tunnelling 1 (5), Ernst pp. 145–154, Pergamon Press
und Sohn, Berlin
Mösslacher, A. 2006. Numerische Untersuchungen des Ein-
flusses einer Störungszone auf Verschiebungen während des
Tunnelvortriebes. Diploma Thesis, Institute for Rock Mechanics
and Tunnelling, Graz University of Technology, Austria
136

Support of excavations subjected to dynamic (rockburst) loading
T.R. Stacey
School of Mining Engineering, University of the Witwatersrand, Johannesburg, South Africa
ABSTRACT: Rockbursts are violent events that manifest in the ejection of rock from excavation walls, often at high velocity.
Although attempts have been made to link the seismic source magnitude via peak particle velocities to predicted ejection
velocities, back analyses of ejection velocities from rockburst damage indicate inexplicably large magnifications of the peak
particle velocities obtained from seismic measurements. This indicates a lack of understanding of the mechanisms of rock
mass behaviour in seismic conditions. There is an equal lack of understanding of the mechanisms of action and interaction of
support elements in a rockbursting environment. The design of rock support for these conditions is therefore problematic, since
neither the demand imposed on the support, nor the capacity of support systems under dynamic loading, can be defined. A
conventional engineering design approach is therefore not possible owing to the design indeterminacy. Since testing of support
has shown that both rockbolts and containment support systems such as wire mesh plus wire rope lacing can absorb large
amounts of energy, an alternative philosophy is to specify the support conservatively. A very important requirement is that the
containment support, and the connecting elements between this support and the retainment support, should not fail – they must
ensure that the dynamic loading is passed fully into the rockbolts. The conservative specification approach may be perceived to
be an expensive solution. However, if all costs of consequences are taken into account, containing rockburst damage is almost
certain to create improved safety and increased value for the operation.
Subject: Rock engineering in hazardous geo-environments
Keywords: Rock bursts, rock support
1 INTRODUCTION crush; shear rupture; and fault slip. Of these five mechanisms,
the first three are events in which the source and damage loca-
Research into the seismicity in mines, and consequently rock- tions are coincident. The remaining two are events in which
burst research, began in South Africa a century ago (SAIMM, the source location and resulting damage location can be sepa-
2010), and possibly even earlier in other countries. Despite this rated by significant distances, sometimes hundreds of metres.
lengthy period of research, solutions to the problem of rock- The magnitudes of these events can be very large, of earth-
bursts appear to be elusive. Mining is progressing to greater quake scale, whereas the magnitudes of the former group are
depths, in higher stress environments, and rockbursting is per- usually much smaller. When seismic source location and rock-
haps an increasing problem world-wide, rather than a problem burst damage location are not coincident, even if one knew
whose solution is understood and known. Although advances exactly the source location, the prediction of the possible dam-
have been made, rockbursts remain a hazard. In their book age location is an additional problem. The philosophy of the
dealing with safety in the mining industry, Simpson et al design of rock support for this severe rockburst problem is the
(2009) quote from Simpson (1998), “Most of the required focus of this paper.
science for most mining hazards is now well understood (with
one or two notable exceptions, for example, rock bursts in
deep hard rock mines) ….” 2 MANIFESTATIONS OF ROCKBURSTS
As has been done for earthquakes, locations of greater seis-
mic hazard can be identified, but predictions of exactly when Rockbursts in mines and civil engineering excavations are
and where the seismic event will occur, and what its magni- very violent events that commonly result in considerable dam-
tude will be, are not known. One of the recent large magnitude age to excavations, well illustrated by Ortlepp (1997). Rock
earthquakes in Chile in March 2010 was actually predicted in is usually ejected, and when this is the case, the ejected rock
terms of both location and magnitude (Ruegg et al, 2009), but, is observed to be fragmented into relatively small blocks and
although there was an expectation of such an event, the tim- slabs, as illustrated in Figure 1.
ing was unknown. The result was unfortunate loss of life due In rockburst events, rock reinforcement such as rockbolts,
mainly to an associated tsunami, as well as widespread dam- and surface support elements such as wire mesh and shotcrete,
age to structures. The rockburst problem, similarly, remains often fail. In such events, conventional rockbolts and cables
one in which the “critical” seismic event location, magni- commonly exhibit brittle failures. Another common observa-
tude and time are unknown. In mines, much of the significant tion is that when the surface support fails, the ejection of rock
seismic activity is often associated in time with blasting. Per- often leaves the reinforcement elements exposed, protruding
sonnel can therefore be removed from the mine at such times. out of the rock as shown in Figures 1 and 2.
However, “rogue” seismic events do occur at other times and Ejection in a rockburst event is not dictated by the action
these events represent a hazard to life. Ortlepp and Stacey of gravity, as is the case with rock falls, but can be in any
(1994a) suggested a classification of seismic event sources direction, and floor-heave and sidewall ejections are common,
that included the following: strain bursting; buckling; face examples being shown in Figures 3 and 4. The latter figure
137

Figure 1. Rockburst damage and fragmentation (photograph W D Figure 4. Floor heave of nearly 2 m caused by a rockburst (photo-
Ortlepp). graph T R Stacey).
• Is it a function of the seismic waves, their interactions, and

their interactions with the excavation surface?
• Why is greater damage often associated with small magni-
tude events than with large magnitude events? Is it because
of the proximity of the event? Is it because of direction-
ality of wave propagation? Is it because of magnification
effects?
3 APPROACHES TO THE DESIGN OF ROCK SUPPORT

FOR ROCKBURSTING CONDITIONS
A rock mechanics design philosophy, consisting of six design

principles and a corresponding ten-step design methodol-
Figure 2. Exposed, protruding rockbolts after a rockburst (photo- ogy or process, was developed by Bieniawski (1992). This
graph W D Ortlepp).
process has been cast into a circular “wheel of design” for-
mat, as shown in Figure 5 (Stacey, 2009), which facilitates
understanding of the interaction of the steps in the process.
In this design process, the “defining the design” stage is
the most critical. In this stage the design objectives must be
clearly stated and understood. With regard to the design of sup-
port for excavations subjected to rockburst loading, the design
objective should be to contain rockburst damage so as to pro-
mote safety and limit disruption of operations. To achieve this,
there must be sufficient input data available to carry out the
design, and the following mechanisms must be understood
and known:
• the mechanisms of loading;
• the mechanisms of rock and rock mass failure that the
support is being designed to combat;
• the mechanisms of action of the support elements and their
Figure 3. Floor heave caused by a rockburst (photograph W D interactions in the support system.
Ortlepp). If this information is known, it is usually possible to deter-
mine the demands imposed on the support elements, the
shows a case in which the concrete in the floor of the
capacities of the support elements and support systems can
excavation has been heaved upwards by nearly 2 metres.
usually be calculated, and these data can then be used to
It can be seen from these illustrations that rockbursts are
design the required support. In concept therefore, the design
very violent events, and that the damage that they cause is
of support for rockbursting conditions is a straightforward
“unpredictable”.
engineering process.
Some questions that are posed by observations of damage
Attention has been given to rock support and its design
are:
for rockbursting conditions over several decades. As early
• Why is such damage often very localized? as the 1960’s Ortlepp (1968; 1969) recognized that yielding
• Is this localisation a function of the performance of the rock support was necessary in dynamic, rockbursting conditions.
support, type of support elements, quality of installation, He developed a yielding rockbolt, and demonstrated its effec-
and strength of grout? tiveness under simulated rockbursting conditions (Ortlepp,
138

to be possibly of crucial importance in the area of design of
rockburst support for tunnels.
• Ejection velocities of the order of 10 m/s and possibly higher
can result from seismic events of moderate magnitude.
• Such velocities will cause severe damage in tunnels rein-
forced with conventional active support even if they are
heavily supported.
• Using the energy approach, yielding support systems based
on presently available components, can be designed to with-
stand these velocities and so contain the damage that would
otherwise occur as a result of even very large seismic events.
“The effectiveness of these yielding systems can realisti-
cally be demonstrated by means of controlled blasting tests.
Suggestions for testing for this purpose are given by Ortlepp
(1992b).”
In a more recent approach, Kaiser et al (1996) provide
detailed recommendations for energy-based design of rock-
burst support that take into account the magnitude of the
expected seismic event, the likely mode of failure and the
Figure 5. The engineering wheel of design (Stacey, 2009). capacity of the support system. They consider that, through
this process, a suitable support system can be chosen. They
1969; Ortlepp & Reed, 1969). It appears that this rockbolt state that “… both the type of damage mechanisms and the esti-
design was never commercialized, nor introduced as support mated level of damage severity must be identified explicitly
in rockburst areas. Further attention was given to rock sup- before it is possible to rationally design a support system …”
port and its design for rockbursting conditions by Ortlepp To determine the expected ejection velocities for design (for
et al (1975), Wagner (1982), Ortlepp (1983), Jager & Roberts the highest energy level conditions), they recommend the
(1987), Roberts & Brummer (1988), Jager et al (1990), Jager following:
(1992), Ortlepp (1992a), Kaiser (1993; 1994), Ortlepp (1994), • “Examine the seismicity records at the mine to establish
Kaiser et al (1996), Ortlepp & Stacey (1997) and Stacey & the spatial and temporal distributions of seismic events;
Ortlepp (2002b). This list of references on the subject is by • Choose a location of a design event;
no means exhaustive, but it indicates that the design of sup- • Select a design magnitude for the seismic event;
port for rockbursting conditions has been given a considerable • Select appropriate scaling law parameters for predicting the
amount of attention over many years. peak particle velocity … and hence, the ejection velocity.
A straightforward engineering design process convention-
ally makes use of the concepts of stress and strength. However, This process is summarized into convenient charts corre-
in dynamic loading situations it has been found that these lating seismic magnitude, distance from source to damage
concepts are inappropriate for rock support design, and an locations and PPV.
energy-based design method is more satisfactory. Wagner Kaiser et al (1996) also provide recommendations for the
(1982) recognized that the design of rapid yielding hydraulic estimation of the mass of rock that is likely to be ejected.
props for support of tabular stopes should be based on energy Once the mass and velocity are known, the kinetic energy (i.e.,
principles rather than stress-strength. Ortlepp (1992a) pre- the energy demand) can be calculated. The capacity of rock
sented an energy-based design rationale, with an application support has been evaluated by Kaiser et al (1996), based on
of the design principle focused on rockbolt elements. He theoretical evaluations as well as on test data from laboratory
concluded, “… it has been possible to establish some very and field testing of support. This allows the choice of suitable
important design principles. rockburst support to be made to contain the expected (pre-
The more important are: dicted) rockburst damage. A similar approach was developed
by Stacey and Ortlepp (2002b).
• It is neither practicable or economically possible to contain Therefore, as stated above, in concept, the design of support
severe rockburst damage by increasing the strength of the for rockbursting conditions is a straightforward engineering
tunnel support. process, and the recommendations of Kaiser et al (1996)
• Designed yieldability or compliance is essential to prevent are commendable. However, rockburst damage and accidents
support components being broken by rockbursts. resulting from rockbursts continue to occur, indicating that
• The ejection velocity imposed on the rock walls by the seis- designed rock support is, in those cases, not performing its
mic wave is probably the single most important determinant desired function. The reason is that, in practice, the determi-
of damage intensity in a tunnel. (Combined seismological nation of both the demand and the capacity under rockbursting
and rock engineering research is urgently needed to estab- conditions is problematic. These two aspects are dealt with in
lish how ground motion parameters in the seismic wave more detail in the sections below.
translate into ejection velocity).
• Energy considerations rather than stress and strength cal-
culations should form the basis of the design of a tunnel 3.1 Evaluation of the demand on rock support in
support system and its elements. rockbursting conditions
“Additionally, while pursuing the basic engineering It is commonly understood that rockburst damage is associ-
approach, some observations were made which are thought ated with the peak particle velocities (PPVs) generated within
139

the rock mass by a seismic event. When these interact with the that even small fragments of rock, ejected at high velocities,
surface of the excavation, the rock at the surface is subjected to can be hazardous.
“ejection” at some ejection velocity. Attempts have been made In summary, therefore, it is clear that the demand parameter
to estimate the applicable PPVs from the expected seismic cannot be defined satisfactorily for rock support design pur-
magnitudes (McGarr, 1982; McGarr et al 1981; Cichowicz, poses using a conventional design approach. The following
2001), or to correlate seismic magnitude with distance to main items of design information are not known with suffi-
source and capacities of support systems (Kaiser, 1993; Kaiser cient confidence: the magnitude and location of a potential
1994; Kaiser et al, 1996; Stacey & Ortlepp, 2002b). Based on seismic event; the ejection velocity which will result from the
the work of Yi & Kaiser (1993), Kaiser et al (1996) state that an seismic event; the direction of action of the ejection force; and
ejection velocity equal to the PPV can confidently be used for the mass of rock that will be involved in the ejection. Some of
engineering purposes, and this assumption is built into their these are items to be chosen or selected in the design method
support design process described briefly above. They indi- recommended by Kaiser et al (1996), implying that the data
cated that the ejection velocity could be double the PPV, that are not known with confidence, but are based on the designer’s
is, an amplification factor of 2, in special cases. This ampli- engineering judgement.
fication factor was used in the design approach suggested
by Stacey & Ortlepp (2002b). Contrary to these opinions,
Cichowicz (2001) indicates that amplification factors of 4
3.2 Evaluation of the capacity of rock support
to 10 can apply when the PPVs on the surfaces of excava-
tions are compared with the PPVs measured within the rock The evaluation of the capacity of rock support systems is
mass. However, these amplification factors appear to be inap- no easy task. The capacities of individual elements of sup-
propriately low, since they result in much smaller ejection port can be calculated from their mechanical properties and
velocities than are observed in practice. From observations the assumed loading conditions to which they are subjected.
of ejected material in rockburst events Ortlepp (1993) calcu- Thus, for example, it is simple to calculate the capacity of
lated possible ejection velocities. In one case the velocity he a rockbolt under tensile loading using the strength proper-
calculated exceeded 50 m/s. He concluded, “… will highlight ties of the steel and the dimensions of the bolt. It is also
paradoxes such as, importantly, how the relatively low PPV simple to check on this capacity by carrying out pull tests
generated in the rock mass does not reconcile with the high in the field (rockbolt performance will depend on installa-
velocity displacements of the wall rock that has so often been tion quality – effectiveness of anchoring, extent of grouting,
observed to break rock support tendons and produce other strength of grout, surface condition of bolt, etc). Capacities
phenomena …” Ortlepp’s (1993) findings are supported by of mesh and other components can similarly be calculated
recent observations in two mining operations. In these, more theoretically.
than 50 rockburst events that occurred over a 12 year period A rock support system is a combination of individual sup-
were evaluated. For these events, measurements of distance of port components that work together to retain and contain the
ejection, and original location of ejected rock material, were rock. In doing this, the components are subjected to loading
used to back analyse the ejection velocities in each event. by the rock and to interactive loading between one component
PPVs in the rock mass were determined from the seismic and another. Therefore, a rockbolt could be subjected to a com-
records. From these data an attempt was made to correlate bination of tensile, shear, bending and torsional loading by the
PPV with ejection velocity. This correlation resulted in calcu- rock under static and, particularly, dynamic conditions. Sim-
lated amplification factors of between about 10 and 50, with ilarly, other components of support – wire mesh, shotcrete,
an average value of about 30, which far exceed those indicated fibre-reinforced shotcrete, face plates, straps, lacing, etc –
by Cichowicz (2001), Kaiser (1993), Kaiser et al (1996) and could be subjected to combinations of loading mechanisms.
Stacey & Ortlepp (2002b). It may therefore be concluded that Connection between the rockbolts and the surface support also
the processes leading to amplification of PPVs at the surfaces implies that the surface support will impose loadings on the
of excavations, resulting in significant ejection velocities, are rockbolts, and vice versa. Owing to these complex situations,
complex and not sufficiently understood at this stage. It is a theoretical approach to the determination of the capacity
clear from the “real” results produced from back analysis of of a support system, in which the capacity of the system is
observed ejections that estimation of ejection velocity from calculated from the capacities of the individual components
seismic parameters produces data that are unsatisfactory for of the system, is very unlikely to be successful, particularly
support design. in a dynamic loading environment. The weakest link prin-
The determination of “demand” requires knowledge not ciple applies – if one component fails, it is then likely that
only of the ejection velocity, but also of the direction of the whole support system will be incapable of containing the
ejection and of the mass of rock that is ejected. This infor- damage.
mation can only be determined from examination after the There is a common tendency nowadays to turn to the
event. Although roof falls are often focused on, ejection of the computer, using numerical analyses to “solve” the problem.
floor (floor heave) and sidewalls is commonly observed. For However, such approaches can only be successful if they
example, rockbolts in the sidewall of a tunnel were observed take into account the relevant mechanisms of support action
to be bent (inclined upwards) after ejection of material in a and loading of the support. Since, as indicated above, the
rockburst. Observations indicated that rock from the sidewall mechanisms could be very complex and are not thoroughly
was ejected upwards and across the tunnel, some of that rock understood, it is unlikely that such analyses could provide the
impacting the roof of the tunnel. “correct” solution. As stated by Ortlepp (1992a), “… it is still
The “thickness” or mass of rock ejected, and the ejection impossible to use these [powerful numerical methods] realis-
velocity, will determine the kinetic energy involved in the tically in any formal design method because of the difficulties
event that must be contained if safe and stable conditions are in defining the structure, estimating the transient loads and
to be maintained. With regard to safety, it is important to note modeling the damage mechanism.”
140

An alternative approach to determining the capacity of rock the order of 12 m/s and demonstrated that “low strength” cone
support systems is to carry out physical testing of support bolts yielded satisfactorily in tension, without breaking, at
components and support systems. A very useful summary these velocities. Yield displacements of the order of 0.5 m
of dynamic testing of rock support that has been carried out occurred in these experiments in absorbing the energy of
in various countries is presented by Hadjigeorgiou & Potvin ejection produced by the blasts. In contrast, much stronger
(2007), and an interpretation of all the results obtained has rebar bolts failed in the tests and had low energy absorption
been carried out by Potvin et al (2010). These two papers capacity. All of these tests were of groups of rockbolts, not of
deal with testing using blasting and using drop weight impacts support systems involving a combination of support elements.
to simulate rockburst loading. They do not refer to the early In addition, they involved tensile loading only, and bolts were
blasting tests carried out by Ortlepp (1969), who was proba- not subjected to shear, or combinations of stresses such as
bly the first to carry out simulated rockburst testing of rock tensile and shear.
support. He carried out two tests on rockbolt and mesh sup- The use of blasting for rockburst test purposes has dis-
port systems installed in a tunnel, one with conventional advantages – it is costly, it requires special environments,
rockbolts and the other with the yielding rockbolts that he usually underground, and it is usually not repeatable. As a
had developed (Ortlepp, 1968). The rockburst loading was result, alternative laboratory testing methods, usually involv-
simulated by blasting, with blastholes 430 mm apart drilled ing some form of drop weight system, have been developed
parallel to the tunnel axis about 600 mm outside the tunnel for evaluation of rock support. Laboratory “rockburst” testing
perimeter. One may question whether blast loading is a sat- of rock support components and systems has been carried out
isfactory simulation of rockburst loading – shock wave and in several countries using somewhat different testing meth-
subsequently, and substantially, gas pressure (i.e., a blow- ods (Ortlepp & Stacey, 1994b; Yi & Kaiser, 1994; Kaiser
ing outwards) provide the loads, whereas in a rockburst a et al, 1996; Ortlepp & Stacey, 1997; Ortlepp & Stacey, 1998;
mass of rock is suddenly accelerated, with no gas pressure Stacey & Ortlepp, 1999; Stacey & Ortlepp, 2001; Stacey &
involved. In fact, the dilation of the rock mass will probably Ortlepp, 2002a; Stacey & Ortlepp, 2002b; Gaudreau et al,
result in a brief negative air pressure. However, the results of 2004; Li et al, 2004; Player et al, 2004; Player et al, 2008a;
Ortlepp’s (1969) testing showed outcomes or damage effects Player et al, 2008b; Plouffe et al, 2008; Li & Charette, 2010).
that were typical of rockburst damage, and his conclusion Further, the references and bibliography provided by Kaiser
was that such loading could be realistic, and therefore a sat- et al (1996) indicate numerous unpublished reports of testing
isfactory means of “rockburst” testing of support (Ortlepp, authored mainly by Tannant. Most of these test methods and
1992b). results have been described by Hadjigeorgiou & Potvin (2007)
More recent blasting “rockburst” tests carried out by sev- and the results summarized by Potvin et al (2010). Figure 6
eral researchers (Hagan et al (2001) provide a summary of shows one of the graphs from Potvin et al (2010) that summa-
tests, with details provided by Milev et al (2001), Reddy & rizes the capacities of various components and combinations
Spottiswoode (2001), Haile & Le Bron (2001) and of surface support.
Hildyard & Milev (2001); Espley et al (2002); Archibald It can be seen that very significant levels of energy can
et al, 2003; Tannant et al, 1993) are described by be absorbed by appropriate support, provided that yield, or
Hadjigeorgiou & Potvin (2007). The results, including the displacement, can take place. The value of wire rope lacing in
tests of Heal & Potvin (2007), are summarized by Potvin et al absorbing energy is also apparent from Figure 6, a contribution
(2010). The summary of results indicates that gas pressure that was specifically identified by Stacey and Ortlepp (2002a).
was a “problem” in some of the tests. The testing described by As can be seen from this Figure, it can enhance the capacity
Hagan et al (2001) indicated ejection velocities in the range of of mesh and shotcrete by as much as 7 times.
0.7 to 2.5 m/s, determined from high speed video recording. The above results show the capacities of the individual
Ground velocities of 3,3 m/s were recorded by an accelerome- elements of support, that is, retainment and reinforcement
ter. Rock support involved in the test consisted of fully cement support (rockbolts, cables), and containment support (mesh,
grouted rockbolts only. “Rockburst” damage occurred on the shotcrete, liners, straps, lacing), but they give no data on the
tunnel wall where the PPV exceeded 0.7 m/s. High intensity capacity of rock support systems, which are combinations of
damage occurred where the ground velocity of 3.3 m/s was retainment and containment support elements as well as the
recorded. connecting components (nuts, faceplates, loops, etc. The per-
A similar blasting geometry was used by Potvin & Heal formance of a support system will depend on the performance
(2010) to ensure that the dynamic testing of the rock support of all of these components. The link between a rockbolt and
was not influenced by gas pressure. In their first test they containment support such as mesh usually involves a steel
measured PPVs in the range of 0.3 to 2.4 m/s. Two support face plate on the bolt. Such plates often fail because of irreg-
systems were used: cone bolts (yielding rockbolts) and High ular rock bearing surfaces or non-axial loading on the bolt.
EnergyAbsorption (HEA) mesh; and cone bolts with mesh and Tests carried out, in which loading simulated “real” conditions
fibre-reinforced shotcrete. Minor damage of the support was rather than idealized flat-surface bearing conditions, demon-
observed. The same location was used for the second test, with strated that plate capacities were much less than their specified
the implication that the rock mass was possibly “damaged” values (Van Sint Jan & Palape, 2007) because the plates failed
(fractured) by the first blast. In this second test, PPVs of 0.6 in a folding mode. Nut failures were also observed in these
to 3.0 m/s were recorded, and significant damage occurred. A tests. Plate capacities are enhanced by the use of lacing and/or
mass of rock of about 100 tons was ejected, with both support straps, which spread the load rather than allowing it to be
systems sustaining damage. concentrated on the plate alone.
The ejection velocities in these tests were significantly Thus, whilst data are available on individual support
lower than the velocities back-analysed from observations of components, knowledge of the capacities of rock support
rockburst damage (Ortlepp, 1993). Ortlepp’s later blast test- systems, from theoretical calculations or in the form of data
ing method (Ortlepp, 1994) achieved an ejection velocity of from practical testing programmes, is absent.
141

4 CLASSIFICATION OF ROCKBURST CONDITIONS
AND SPECIFICATION OF SUPPORT FOR THESE
CONDITIONS – AN ALTERNATIVE PHILOSOPHY
Since the use of the conventional rock support design approach

is not possible, the suggested philosophy in determining sup-
port requirements for situations in which rockbursting occurs
or may occur is to specify the support rather than attempt to
design it. This will ensure that “appropriate” rock support will
be installed, whose level in many cases may be judged to be
rather conservative. However, since the design parameters are
unknown, and safety is paramount, conservatism is fully justi-
fied. Such conservatism could be viewed as uneconomic, and
this question will be dealt with in Section 4.2 below.
4.1 Specification of rock support commensurate with

rockburst conditions
The specification of “appropriate” rock support should follow
from a classification of the operation as “rockbursting” or
“non-rockbursting”. The decision on the classification can be
made by considering a series of questions:
a) Does the mine have a history of rockbursting? A positive
answer to this question will be required if the mine has
experienced any form of rockbursting (popping, strain-
bursting, or more severe rock bursting) in its operation.
The operation will then be classified as a “rockbursting
operation or mine”.
b) Has any seismicity has been recorded, measured, or sim-
ply noted, that could indicate that there is potential for
rockbursting? The operation will then be classified as a
“rockbursting operation or mine”.
If the answers to these two questions are negative, no rock-
burst support is necessary. It is to be noted, however, that
Figure 6. Performance of surface support systems under dynamic yielding support applicable for rockburst conditions will also
loading (Potvin et al, 2010). be of great benefit in maintaining stability under squeez-
ing conditions should these occur. Positive answers to the
two questions will require that rockburst support is specified,
which prompts a further question:
3.3 Conclusions regarding capacity and demand
c) How would the rockbursts that have occurred at the mine,
It will be clear from the above two sections that, under dynamic
or are anticipated to occur at the mine, be classified?
(rockburst) loading conditions, neither the capacity of rock
With reference to the suggested classification presented
support systems, nor whether a rockburst will occur, and if
by Ortlepp & Stacey (1994a), two categories of rockburst
it does, the demand that will be imposed by the rockburst,
severity can be defined:
are known. Therefore, with regard to a conventional design
Category 1: “local” bursts, in which the seismic source
process for rock support, neither of the two essential param-
location and rockburst damage location are coincident.
eters required to enable a robust rock support design to be
This occurs in popping, strainbursting, buckling and face
carried out, namely, the demand and the capacity, are known.
bursting situations.
This is a clear case of design indeterminacy, and it is there-
Category 2: the seismic source location and the rockburst
fore impossible to determine the required support using the
damage location are not coincident. These events, whose
classical engineering design approach.
sources are due to shear rupture and fault slip, usually
However, there is positive information that arises from the
result in much more severe rockburst damage than the first
physical testing of the capacity of support elements:
category.
• Yielding rockbolts have been proven to have the capability Support in category 2 situations, the focus of this paper, will
of yielding under tensile loading at high velocities (12 m/s); need to have much greater energy absorbing capability than
• Containment support has proven capacity of absorbing large in category 1. The general support specification for category
amounts of energy (about 30 kJ for a deformation of 0.5 m 2 should be:
as indicated in Figure 6). Rockbolts: capable of yielding in tension at a velocity of
10 m/s, spaced typically 1 m apart. There are no adequate data
This information is of use in the proposed way for- regarding yielding performance of rockbolts under shear load-
ward regarding “design” of rock support for rockbursting ing, such as on joints and fractures, or due to the orientation
conditions. of the ejection (ie, not only axial). It is logical to assume that
142

readily available yielding rockbolts will perform better than is to determine where damage will occur and where it will
conventional rockbolts in shear yielding. A suggestion is that not. A conservative approach is recommended in making such
the rockbolts should be installed at randomly variable angles, determinations.
such as ±20◦ to the normal to the excavation surface. This Support component costs are only one of many costs associ-
will ensure that bolts cross joints and fractures over a range ated with the consequences of rockbursts. Other consequential
of angles, allowing at least some of the rockbolts to act more costs include:
favourably in shear yield. It could also cater better for variation
• Accidents and associated costs, including work stoppages;
in ejection orientation.
• Clean up costs and rehabilitation costs;
Containment support: in the form of high quality steel
• The cost of loss of production in operations directly affected
mesh.This could be weldmesh or chain-link mesh. Introducing
by the damage;
corrugations in the weldmesh will improve its yieldability.
• The costs of loss of production in areas more remote from
Enhancement of containment support: wire rope lacing
the damage, owing to loss of access, such as blockages of
or tendon straps over the mesh. Such elements have been
tunnels, damage to roadways, damage to ventilation, etc;
shown to provide enormous enhancement to the capacity of
• Cost due to reassignment of crews;
the mesh, and also reduce the problems of weakness at joins
• The cost associated with loss of ore;
between adjacent mesh sections. In extreme cases, yielding
• Costs that are difficult to quantify – public perception,
capability in lacing can be introduced by using appropriate
reduction of mining company share price, reduced worker
yielding clamps. An interesting support development facili-
morale, labour unrest, etc.
tating mechanized installation is High Energy Absorption
mesh described by Potvin (2009) and Potvin & Heal (2010), Therefore, if rockburst damage can be reduced or prevented,
which incorporates wire mesh and wire rope lacing in a both direct and indirect costs that might be associated with
support “panel”. damage will be minimized. In determining the value that
Connecting elements: steel faceplates and nuts to connect would be represented by the installation of the “appropriate”
the containment support and the retainment rockbolts. The rockburst support, all direct and indirect costs should be taken
faceplates must be large enough to ensure that they do not into account. Recent research by Rwodzi (2010) has shown
pull through the mesh (wire rope lacing or straps, will assist in that indirect costs far outweigh direct costs in the overall eval-
this regard). Kaiser et al (1996) recommend that plates should uation of consequences, with loss of production usually being
cover four squares of mesh. Sharp-edged plates can cause fail- the major contributor.The direct costs of the rock support com-
ure of the mesh by “guillotining” the wire strands. Faceplates ponents and their installation are usually a small fraction of the
and nuts, which are often the weakest links, must not fail overall consequential cost. The conservative support specifi-
if the rockburst support is to perform its required function. cation philosophy indicated above must be viewed in the light
Shepherds crook rockbolts and wire rope lacing can eliminate of overall costs: damage and failures destroy value, and their
the requirement for a faceplate. prevention generates value. Therefore, “more” support may
In summary, the capacities of the containment support and be considered excessively expensive as far as direct costs are
the connecting elements must be sufficient to ensure that they concerned, but if it prevents or contains damage, the resulting
do not fail, but transfer the loads to the rockbolts, forcing savings may create great value for the operation. A further
them to yield, to absorb the rockburst energy, and therefore to point is that greater demand for yielding support will create
ensure that the support system contains the rockburst damage. greater competition from manufacturers and reduced prices
In mechanized operations, shotcrete will be required to protect because of increased volumes.
the support from mechanical damage.
As indicated in Section 2, rockburst ejection can occur from
any surface, and this raises the question of support of the floor 5 CONCLUSIONS
of excavations. Floor support is rarely installed in hard rock
excavations. However, there is no engineering justification for Rockbursts are severe events that result in the violent ejec-
the omission of such support since the influence of such ejection of rock from excavation walls. The source of the energy
tions on safety is no different from other ejection directions, involved in such events is the seismicity, but a satisfac-
and floor heave causes considerable disruption. tory, quantified, link between the seismicity and the ejection
velocity has not been established. Back analyses of ejection
velocities from rockburst damage indicate inexplicably large
4.2 Economic justification of a conservative level of support
magnifications of peak particle velocities measured in the rock
The implementation of substantial, rockburst resistant support mass. This indicates a lack of understanding of the mecha-
is likely to result in an increase in rock support costs compared nisms of behaviour of the rock mass in seismic conditions. As
with conventional rock support practice, owing to the direct a result of this lack of knowledge and understanding, it is not
costs of the support components and their installation. This possible to define the demand to which rock support will be
is not necessarily the case, however. Ortlepp & Stacey (1995) subjected.
showed that, in replacing rebar rockbolts with yielding rock- Although the capacities of individual rock support elements
bolts, installed support costs were equivalent if the spacing of can be measured or calculated, the mechanisms of action and
the yielding bolts was increased slightly from 1 m to 1.05 m. interaction of support elements in a rockbursting environ-
Most importantly, however, the energy absorption capacity of ment are unknown. It is therefore not possible to determine
the support in kJ/m2 increased more than tenfold. Kaiser et al the capacity of a support system under dynamic loading for
(1996) state that, “It is unnecessary and excessively expen- support design purposes. Since neither the demand compo-
sive to install the enhanced ground support systems needed nent nor the capacity component can be defined, the result
to prevent rockbursts everywhere in a mine, since most areas is “design indeterminacy” and a conventional rock support
will not ever be damaged.” This is correct, but the key issue design approach is not possible. An alternative approach to the
143

problem is to specify the support, using conservative assump- Hildyard, M.W. & Milev, A.M. 2001. Simulated rockburst exper-
tions, and this philosophy has been promoted in this paper. iment: development of a numerical model for seismic wave
Testing of support has shown that both rockbolts and con- propagation from the blast, forward analysis, Jl S. Afr. Inst. Min.
tainment support systems, such as wire mesh plus wire rope Metall., Vol 101, No 5, pp 234–245.
Jager, A.J. 1992. Two new support units for the control of rock-
lacing, can absorb large amounts of energy. Therefore, specifi-
burst damage, Proc. Int. Symp. on Rock Support, Sudbury,
cation of support systems incorporating elements with known pp 621–631.
large energy absorption capacities could provide a satisfactory Jager, A..J & Roberts, M.K.C. 1987. Recommended performance
solution to the rockburst support problem. A very important requirements for yielding rock tendons, Symp. Design of Rock
requirement is that the containment support, and the connect- Reinforcing: Components and Systems, S. Afr. National Group of
ing elements between this support and the retainment support, Int. Soc. Rock Mech., pp 71–74.
should not fail. They must ensure that the dynamic loading is Jager, A.J., Wojno, L.N. & Henderson, N.B. 1990. New developments
transferred fully to the rockbolts. This philosophy is likely to in the design and support of tunnels under high stress, Technical
promote greater safety in rockbursting situations. Challenges in Deep-Level Mining, Johannesburg, S.Afr. Inst. Min.
The conservative specification approach may be perceived Metall., pp 1155–1172.
Kaiser, P.K. 1993. Keynote Address: Support of tunnels in burst-
to be an expensive solution. However, if all costs of conse-
prone ground – toward a rational design methodology, in Young
quences are taken into account, prevention or containment (ed), Rockbursts and Seismicity in Mines, Balkema, Rotterdam,
of rockburst damage is almost certain to create value for the pp 13–27.
operation. Kaiser, P.K. 1994. Support in burst-prone ground, Proc. Rockburst
Workshop, ISRM Rock Mechanics Symp., Santiago, Chile, May
ACKNOWLEDGEMENTS 1994, 24p.
Kaiser, P.K., McCreath, D.R. & Tannant, D.D. 1996. Canadian Rock-
Mr E Rojas of El Teniente Mine is thanked for permission burst Support Handbook: Volume 2: Rockburst Support, Canadian
Rockburst Research Program 1990–1995, CAMIRO.
to use the illustration in Figure 4, and Prof Y Potvin of the
Li, T., Brown, E.T., Coxon, J. & Singh, U. 2004. Dynamic capa-
Australian Centre for Geomechanics is thanked for the use of ble ground support development and application, in Villaescusa
Figure 6. and Potvin (eds), Ground Support in Mining and Underground
Construction, Perth, Balkema, pp 281–288.
Li, C.C. & Charette, F. 2010. Dynamic performance of the D-Bolt,
REFERENCES in Van Sint Jan & Potvin (eds), Proc. 5th Int. Seminar on Deep
and High Stress Mining, Santiago, Chile, Australian Centre for
Archibald, J.F., Baidoe, J.P. & Katsabanis, P.T. 2003. Rockburst dam- Geomechanics, pp 321–328.
age mitigation benefits deriving from use of spray-on rock linings, McGarr, A. 1982. Some implications of seismic source mechanisms
Proc. 3rd Int Seminar on Surface Support Liners: Thin Spray-on studies to assessing underground hazard, in Gay & Wainwright
Liners, Shotcrete and Mesh, Quebec City, Section 19. (eds), Proc. Int. Symp. on Rockbursts and Seismicity in Mines,
Bieniawski, Z.T. 1992. Invited Paper: Principles of engineering Johannesburg, S. Afr. Inst. Min. Metall. Symp. Series No 6 (1984),
design for rock mechanics, Rock Mechanics, Proc. 33rd U S pp199–208.
Symp. on Rock Mech., Ed. Tillerson & Wawersik, Balkema, pp McGarr, A., Green, R.W.E. & Spottiswoode, S.M. 1981. Strong
1031–1040. ground motion of ground tremors: some implications for near-
Cichowicz, A. 2001. The meaningful use of peak particle velocity at source ground motion parameters, Bull. Seism. Soc. Am., Vol 71,
excavation surface for the optimisation of the rockburst support pp 295–319.
criteria for tunnels and stopes, Final Report, SIMRAC Project No Milev, A.M., Spottiswoode, S.M., Rorke, A.J. & Finnie, G.J. 2001.
GAP7096. Seismic monitoring of a simulated rockburst on a wall of an
Espley, S.J., Heilig, J. & Moreau, L.H. 2002. Assessment of the underground tunnel, Jl S. Afr. Inst. Min. Metall., Vol 101, No 5,
dynamic capacity of liners for application in highly-stressed pp 253–260.
mining environments at INCO Limited, in Potvin, Stacey & Ortlepp, W.D. 1968. A yielding rockbolt, Research Organisation
Hadjigeorgiou (eds), Surface Support in Mining, Australian Centre Bulletin of the Chamber of Mines of South Africa, No 14, pp 6–8.
for Geomechanics, pp 187–192. Ortlepp, W.D. 1969 An empirical determination of the effective-
Gaudreau, D., Aubertin, M. & Simon, R. 2004. Performance ness of rockbolt support under impulse loading, in Brekke &
of tendon support systems submitted to dynamic loading, in Jorstad (eds), Proc, Int. Symp. on Large Permanent Under-
Villaescusa & Potvin (eds), Ground Support in Mining and ground Openings, Oslo, September 1969, Universitats-forlaget,
Underground Construction, Balkema, pp 299–312. pp 197–205.
Hadjigeorgiou, J. & Potvin, Y. 2007. Overview of dynamic testing Ortlepp, W.D. 1983. Considerations in the design of support for
of ground support, in Potvin (ed.), Deep Mining 07, Australian deep hard-rock tunnels, Proc. 5th Int. Cong. Int. Soc. Rock Mech.,
Centre for Geomechanics, Perth, pp 349–371. Melbourne, pp D179–D187.
Hagan, T.O., Milev, A.M., Spottiswoode, S.M., Hildyard, M.W., Ortlepp, W.D. 1992a. Invited lecture: The design of support for the
Grodner, M., Rorke, A.J., Finnie, G.J., Reddy, N., Haile, containment of rockburst damage in tunnels, in Kaiser & McCreath
A.T., Le Bron, K.B. & Grave, D.M. 2001. Simulated rockburst (eds), Rock Support in Mining and Underground Construction,
experiment – an overview, Jl S. Sfr. Inst. Min. Metall., Vol 101, No Balkema, Rotterdam, pp 593–609.
5, pp 217–222. Ortlepp, W.D. 1992b. Implosive-load testing of tunnel support, in
Haile, A.T. & Le Bron, K. 2001. Simulated rockburst experiment – Kaiser & McCreath (eds), Rock Support in Mining and Under-
evaluation of rock bolt reinforcement performance, Jl S. Afr. Inst. ground Construction, Balkema, Rotterdam, pp 675–682.
Min. Metall., Vol 101, No 5, pp 247–251. Ortlepp, W.D. 1993. High ground displacement velocities associated
Heal, D., Hudyma, M. & Potvin,Y. 2004. Assessing the in-situ perfor- with rockburst damage, in Young (ed), Rockburst and Seismicity
mance of ground support systems subjected to dynamic loading, in Mines, Balkema, Rotterdam, pp 101–106.
in Villaescusa & Potvin (eds), Ground Support in Mining and Ortlepp, W.D. 1994. Grouted rock-studs as rockburst support:
Underground Construction, Perth, Balkema, pp 319–326. A simple design approach and an effective test procedure, Jl S.
Heal, D. & Potvin, Y. 2007. In-situ dynamic testing of ground sup- Afr. Inst. Min. Metall., Vol 94, pp 47–63.
port using simulated rockbursts, in Potvin (ed), Deep Mining 07, Ortlepp, W.D. 1997. Rock Fracture and Rockbursts – an illustrative
Australian Centre for Geomechanics, Perth, pp 373–394. study, S. Afr. Inst. Min. Metall., 255p.
144

Ortlepp, W.D., More O’Ferrall, R.C. & Wilson, J.W. 1975. Support Mining, Santiago, Chile, Australian Centre for Geomechanics, pp
methods in tunnels, Symp. on Strata Control & Rockburst Problems 251–272.
of the S A Goldfields, Association of Mine Managers of S. Afr. Reddy, N. & Spottiswoode, S.M. 2001. The influence of geology
1972–73, Johannesburg 1975, pp 167–195. on a simulated rockburst experiment, Jl S. Afr. Inst. Min. Metall.,
Ortlepp, W.D. & Reed, J.J. 1969. Yieldable rock bolts for shock load- Vol 101, No 5, pp 267–272.
ing and grouted rockbolts for faster stabilization, Proc. AIME Roberts, M.K.C. & Brummer, R.K. 1988. Support requirements in
Intermountain Minerals Conference, Vail, Colorado, July 1969, rockburst conditions, Jl S. Afr. Inst. Metall., Vol 88, No 3, pp 97–
21p. 104.
Ortlepp, W.D. & Stacey, T.R. 1994a. Rockburst mechanisms in tun- Ruegg, J.C., Rudloff, A., Vigny, C., Madariaga, R., De Chabalier,
nels and shafts, Tunnelling and Underground Space Technology, J.B., Campos, J., Kausel, E., Barrientos, S. & Dimitrov, D. 2009.
v 9, no 1, pp 59–65. Interseismic strain accumulation measured by GPS in the seismic
Ortlepp, W.D. & Stacey, T.R. 1994b. The need for yielding support gap between Constitución and Concepción in Chile, Physics of the
in rockburst conditions, and realistic testing of rockbolts, Proc. Earth and Planetary Interiors, Vol 175, pp 78–85.
Int. Workshop on Applied Rockburst Research, Santiago, Chile, Rwodzi, L. 2010. Rockfall Risk: Quantification of the consequences
SOCHIGE, pp 265–275. of rockfalls, MSc Eng Dissertation, University of the Witwater-
Ortlepp, W.D. & Stacey, T.R. 1995. The spacing of support – safety srand.
and cost implications, Jl S. Afr. Inst. Min. Metall., Vol 95, May– SAIMM 2010. Centenary of Rockburst Research, http://www.saimm.
June, pp 141–146. co.za / saimm-events / item / 24-centenary-of-rockburst-research#
Ortlepp, W.D. & Stacey,T.R. 1997.Testing of tunnel support: dynamic magictabs
load testing of rock support containment systems, Safety in Mines Simpson, G.C. 1998) It’s the people – stupid, Proc. Minesafe
Research Advisory Committee, SIMRAC GAP Project 221. International 1998, Perth, WA Dept. of Minerals and Energy.
Ortlepp, W.D. & Stacey, T.R. 1998. Performance of tunnel support Simpson, G., Horberry, T. & Joy, J. 2009. Understanding Human
under large deformation static and dynamic loading, Tunnelling Error in Mine Safety, Ashgate Publishing Ltd, 160p.
and Underground Space Technology, v 13, no 1, pp 15–21. Stacey, T.R. 2009. Design – a Strategic Issue, Jl S. Afr. Inst. Min.
Ortlepp, W.D., Stacey, T.R. and Kirsten, H.A.D. 1999. Containment Metall., Vol 109, No 3, pp 157–162.
support for large static and dynamic deformations in mines, Proc. Stacey, T.R. & Ortlepp, W.D. 1999. Retainment support for dynamic
Int. Symp. Rock Support and Reinforcement Practice in Mining, events in mines, Proc. Int. Symp. Rock Support and Reinforcement
Kalgoorlie, Australia, March 1999, Balkema, pp 359–364. Practice in Mining, Kalgoorlie, Australia, March 1999, Balkema,
Player, J.R., Villaescusa, E. & Thompson, A.G. 2004. Dynamic test- pp 329–333.
ing of rock reinforcement using the momentum transfer concept, Stacey, T.R. & Ortlepp, W.D. 2001. Tunnel surface support – capac-
in Villaescusa & Potvin (eds), Ground Support in Mining and ities of various types of wire mesh and shotcrete under dynamic
Underground Construction, Balkema, pp 3597–3622. loading, Jl S. Afr. Inst. Min. Metall., Vol. 101, No. 7, pp 337–342.
Player, J.R., Thompson, A.G. & Villaescusa, E. 2008a. Dynamic test- Stacey, T.R. & Ortlepp, W.D. 2002a. The contribution of wire rope
ing of reinforcement systems, in Stacey & Malan (eds), Proc. 6th lacing in surface support, Proc. 2nd Int. Seminar on Surface Sup-
Int. Symp on Ground Support in Mining and Civil Engineering port Liners: Thin Sprayed Liners, Shotcrete, Mesh, Johannesburg,
Construction, Cape Town, S. Afr. Inst. Min. Metall. Symposium S. Afr. Inst. Min. Metall., pp 1–8.
Series S51, pp 581–595. Stacey, T.R. & Ortlepp, W.D. 2002b. Yielding rock support – the
Player, J.R., Morton, E.C., Thompson, A.G. & Villaescusa, E. 2008b. capacities of different types of support, and matching of support
Static and dynamic testing of steel wire mesh for mining appli- type to seismic demand, Proc. Int. Seminar Deep and High Stress
cations of rock surface support, in Stacey & Malan (eds), Proc. Mining, Perth, Australia, November 2002, Australian Centre for
6th Int. Symp on Ground Support in Mining and Civil Engineering Geomechanics, Section 38, 10p.
Construction, Cape Town, S. Afr. Inst. Min. Metall. Symposium Tannant, D.D., McDowell, G.M., Brummer, R.K. & Kaiser, P.K.
Series S51, pp 693–706. 1993. Ejection velocities measured during a rockburst simulation
Plouffe, M., Anderson, T. & Judge, K. 2008. Rock bolts testing under experiment, Proc. 3rd Int. Symp. on Rockbursts and Seismicity in
dynamic conditions at CANMET-MMSL, in Stacey & Malan Mines, Kingston, A A Balkema, Rotterdam, pp 129–133.
(eds), Proc. 6th Int. Symp on Ground Support in Mining and Civil Van Sint Jan, M. & Palape, M. 2007. Behaviour of steel plates during
Engineering Construction, Cape Town, S. Afr. Inst. Min. Metall. rockbursts, in Potvin (ed), Deep Mining 07, Australian Centre for
Symposium Series S51, pp 581–595. Geomechanics, pp 405–412.
Potvin, Y. 2009. Surface support in extreme ground conditions – Wagner, H. 1982. Support requirements for rockburst conditions,
HEA MeshTM , in P Dight (ed), Proc. 1st Int. Seminar on Safe and in Gay and Wainwright (eds), Proc. 1st Int Symp. Rockbursts and
Rapid Development Mining, SRDM 2009, Australian Centre for Sesimicity in Mines, S. Afr. Inst. Min. Metall. Symp. Series No 6
Geomechanics, pp 111–119. (1984), pp 209–218.
Potvin, Y. & Heal, D. 2010. Dynamic testing of High Energy Absorp- Yi, X. & Kaiser, P.K. 1993. Mechanisms of rockmass failure and
tion (HEA) mesh, in Van Sint Jan & Potvin (eds), Proc 5th Int. prevention strategies in rockburst conditions, Proc. 3rd Int Symp.
Seminar on Deep and High Stress Mining, Santiago, Australian Rockbursts and Sesimicity in Mines, Kingston, Balkema.
Centre for Geomechanics, pp 283–300. Yi, X. & Kaiser, P.K. 1994. Impact testing for rockbolt design in
Potvin, Y., Wesseloo, J. & Heal, D. 2010. An interpretation of ground rockburst conditions, Int. J. Rock Mech. Min. Sci. and Geomech.
support capacity submitted to dynamic loading, in Van Sint Jan & Abstr., Vol 31, No 6, pp 67–68.
Potvin (eds), Proc. 5th Int. Seminar on Deep and High Stress
145

Advances in rock dynamics modelling, testing and engineering
J. Zhao
Ecole Polytechnique Fédérale de Lausanne (EPFL), Laboratory for Rock Mechanics (LMR), Lausanne, Switzerland
Y.X. Zhou
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore
K.W. Xia
Department of Civil Engineering, University of Toronto, Canada
1 INTRODUCTION mature, e.g., LS-DYNA, ABQUS, FLAC and RFPA are com-
mercial codes capable to model dynamic fracturing. However,
Rock dynamics deals with the responses of rock (materials the continuum based methods are not the most suitable meth-
and masses) under dynamic stress fields, where an increased ods for dealing with complete detachment and large-scale
rate of loading (or impulsive loading) induces a change in the fracture opening problems. Because of the models are con-
mechanical behaviour of the rock materials and rock masses tinuum bases, they are not seen as an ideal tool dealing
(Zhou & Zhao 2011). with complex discontinuity, typically represented in naturally
Dynamic loads include explosion, impact, and seismic, that jointed rock masses and rocks in post-failure fracturing state.
are typically in the form of time histories of particle accelera- Discontinuum based methods treat modelling material as
tion, velocity, or displacement. Distribution of dynamic stress an assembled model of blocks, particles or bars, e.g., the Dis-
field is in the form of propagating stress waves. Wave prop- tinct Element Method (DEM) (Cundall 1971), Discontinuous
agation in rock masses is governed by the wave transmission Deformation Analysis (DDA) (Shi 1988) and Distinct Lattice
and transformation across the rock discontinuities in the rock Spring Model (DLSM) (Zhao 2010). In these methods, the
masses, which forms a major topic of rock dynamics. detachment of rock mass and fracturing process of rock mate-
Response of rock materials and rock masses under dynamic rial is represented by the breakage of inter-block contacts or
stress are at different scales, including rock material fracturing inter-particle bonds. Specifically, discontinuum based meth-
and failure, sliding along the rock joints and rock block move- ods can reproduce realistic rock failure process at the post
ments. Rock material fracturing, for example, is a dynamic failure stage. Available commercial computer codes which
micro-scale process leading to macro-scale deformation and based on DEM are UDEC/3DEC and PFC (ITASC 2010) and
failure. Rock fracturing is a dynamic process often associated DDA (Shi 1988). There also exist some research codes, for
with sudden energy release forming dynamic stress waves. example, DLSM (Zhao 2010).
Rock dynamics has applications in civil, mining, energy In order to optimize the advantages of both continuum and
and environmental engineering encountering dynamic loads discontinuum methods, coupled methods are developed in
and behaviours, e.g., rock excavation and fragmentation by recent years. For example, the Numerical Manifold Method
blasting, tunnelling and slope stability and support under (NMM) (Shi 1991) was developed to integrate DDA and
earthquakes, protection of rock falls, rock burst in deep mines, FEM, the FEM/DEM method (Munjiza 2004) is to couple
fracturing of hot rock in geothermal fields, hazard and risk FEM with DEM and the Particle based Manifold Method
control due to explosion and blast. (PMM) (Zhao 2009) was proposed to combine DLSM and
This keynote addresses advancements in some of the topics NMM. The coupled method is capable of capturing both the
of rock dynamics and applications, specifically on numerical pre-failure and the post-failure behaviour of rock materials.
modelling methods, laboratory testing techniques and tunnel However, its implementation is difficult.There only exist some
stability under explosive loading. research codes, e.g., NMM (Shi 1991), Y2D (Munjiza 2004),
LDEC (Morris et al. 2006), m-DLSM (Zhao 2010), and no
2 MULTISCALE CONTINUUM-DISCONTINUUM commercial codes are readily available yet. Table 1 provides
NUMERICAL MODELLING a summary on these numerical methods and corresponding
computer codes.
Numerical methods used in rock mechanics are classified into The coupling of DEM and FEM emerged in the late 1980s
continuum based, discontinuum based and coupled contin- and various coupled models have been developed. The most
uum/discontinuum methods (Jing 2003). direct coupling methodology is to model the discontinuous
The continuum based methods are based on continuum zone with a DEM model and the continuous zone with a FEM
assumptions. They include the Finite Element Method (FEM), model, e.g. the model used by Pan & Reed (1991) and Chen &
the Finite Difference Method (FDM), and the Smoothed Zhao (1998).Yet, this simple FEM/DEM coupling cannot fully
Particle Hydrodynamics (SPH).The methods utilize the macro take advantage of the strength of each method and is limited
mechanical parameters to model the stress-strain state. Com- on handling complex rock failure problems. Recently devel-
puter codes for continuum based methods are relatively oped coupled FEM/DEM methods aims at modelling failing,
147

Table 1. Numerical methods for rock dynamic problems.
Numerical Methods Typical Software General Applicability
Continumm based: LS-DYNA, Displacement without

FEM, FDM, ABQUS, FLAC, element detachment
BEM, SPH RFPA
Discontinuum UDEC/3DEC, Element detachment,
based: PFC, DDA, rock fracturing, rock
DEM, DLSM DLSM block movement
Coupled/hybrid FEMDEM, NMM, Multiscale,
based: Y2D, LDEC, displacement, Figure 1. Particle Manifold Method (PMM) element (after Zhao
combined methods m-DLSM, PMM fracturing, and block et al., 2011).
movement combined
Lagrange multipliers are used to treat this problem. More-
over, in most of multiscale models, the microscopic domain
and macroscopic domain are always unchanged during the
fracturing and fragmenting of solids, with the most represen- computation process and translation from macroscopic model
tative one being the combined finite-discrete element method to microscopic model is not considered.
developed by Munjiza (2004). In the combined finite-discrete Very recently, a new coupling method termed Particle
element method, each body is represented by a single dis- Manifold Method (PMM) is proposed by Zhao (2009) and suc-
crete element that interacts with other discrete elements that cessfully used to develop a new multiscale model m-DLSM
are close to it. In addition, each discrete element is divided (Zhao 2010) for rock dynamic study. The PMM element sim-
into FEM elements, which can be broken into smaller blocks plifies the contact detection between the particle in the DLSM
during calculation. There also exist some numerical methods model and NMM model and serves as the cushion layer which
which attempt to combine FEM and DEM into one single can reduce the spurious wave reflection between micro and
framework. The most representative one is the Numerical macro models. The PMM element is realized by replacing
Manifold Method (NMM) developed by Shi (1991), in which the physical domain of the manifold element by the particle-
the concept of a cover system is used to unite the mathe- based DLSM model. The 3D PMM element used in m-DLSM
matic description of continuum and discontinuum problems. is illustrated in Figure 1. The eight-node FEM element is used
Due to this merit, the NMM has been widely used for frac- as the mathematic element and the DLSM model is used as the
ture progress simulation by many researchers, e.g. Chiou et al. physical domain. In m-DLSM, the interaction between PMM
(2002), Terada et al. (2007) and Ma et al. (2009). A compre- and DLSM is realized through the interaction of the DLSM
hensive review on other FEM/DEM coupled methods can be particle with the PMM particle. The interaction between PMM
found in Zhao et al. (2011). and NMM is realized by sharing common manifold nodes. The
The multiscale methods refer to coupling atomistic mechan- PMM model is used as the midst scale layer of the m-DLSM to
ics or quantum theory and classical continuum mechanics, e.g. realize coupling of the DLSM and the NMM. The work flow
the work of Mullins & Dokainish (1982), Tradmor et al. (1996) of the coupled calculation cycle in m-DLSM is shown in Fig-
and Hasnaoui & Van Swygenhoven (2003). Recently, a few ure 2. The DLSM and NMM computations are performed in
studies are conducted on the multiscale numerical techniques parallel. Interactions between them are finished by the PMM
for geomaterial. In these models, the microscopic model is the model. Information exchange only happens at the beginning
particle based discrete model and the macroscopic continuum and the end of each cycle. The implementation details of this
model is Finite Element Method (FEM) usually, examples are method are given in Zhao (2010)
presented in (Onate & Rojek 2004, Elmekati & Shamy 2010, Here an example of multiscale modelling on rock engi-
Zhao 2010, Yan et al. 2010). There are three main components neering problem using m-DLSM is presented. The multiscale
in a multiscale model: a microscopic model, a macroscopic computational model studying blast wave and tunnel stabil-
model and a handshaking/coupling model. The most widely ity is given in Figure 3. As shown in Figure 4, the left side
used technique is to link particle based discrete model with of the tunnel is damaged under blast loading. With regard to
FEM by using contact treatment between the FEM block the failure patterns obtained, the multiscale simulation gives a
model and particles around the interface, e.g. Onate & Rojek realistic description of the fracture process of the tunnel under
(2004) developed a contact algorithm in their model, Yan et al blasting loading. It should be mentioned that more than two
(2010) used ghost particle method and Lei & Zang (2010) million particles are needed to build this computational model
used a penalty function method. Another recently developed for pure DLSM model. It is an inaccessible problem for the
approach is using a bridging domain to link two different scale normal PC. Yet, only about half a million particles are needed
models (Rousseau et al. 2009), where the coupling technique for the m-DLSM which can easily be run on a normal PC.
is called as bridging domain method which was firstly devel- It shows that the m-DLSM can largely reduce computational
oped for coupling Molecular Dynamics (MD) and FEM by resources needed for purely microscopic numerical model.
Xiao & Belytschko (2004). Even though there are a number of
coupling approaches are available, many of them are not per-
fect and further developments are still needed. For example, 3 ADVANCED LABORATORY TESTING USING
directly using the interface coupling method always induces HOPKINSON BAR
apparent spurious wave reflections. Using bridging domain
method can eliminate this spurious wave, but needs to deal To test the dynamic mechanical properties of rocks, one needs
with the miss marching between the Degree Of Freedoms a reliable testing device. Different testing techniques are avail-
(DOFs) of FEM and these of particle DEM, e.g. complex able, as summarized in Table 2. For testing rocks under high
148

Figure 4. Schematics of the split Hopkinson pressure bar (SHPB)
system.
Based on the one dimensional stress theory, we can deter-

mine the histories of velocity v(t) and force P(t) on the front
surface of the sample denoted by 1 and the back surface of the
sample denoted as 2:
where C is the one dimensional longitudinal stress wave veloc-

ity of the bar, E is the Young’s Modulus of the bar, and A is
the area of the bar.
3.2 Pulse shaping techniques
Figure 2. Coupled calculation cycle in m-DLSM (after Zhao et al.,

To achieve accurate measurements in SHPB tests, one has to
2011). make sure that that the specimen is experiencing an approx-
imately equilibrium stress state (dynamic stress equilibrium
or dynamic force balance, i.e., P1 ≈ P2 ). Because it takes the
stress wave to travel in the specimen 3–4 rounds for the stress
to achieve such an equilibrium state, the dynamic loading then
should be slow and long enough (Subhash et al. 2000). The
pulse shaper technique was proposed to slow down the load-
ing rate and thus to minimize the so-called axial inertial effect
associated with the stress wave loading (Frew et al. 2002, Xia
et al. 2008). Li et al. (2000) used a tapered striker to generate
an approximate half-sine loading waveform, with which the
equilibrium stress state can also be achieved.
Figure 3. Multiscale model for the blasting wave propagation 3.3 Dynamic Brazilian disk (BD) method
interact with a tunnel. The BD specimen in the SHPB system is shown schematically
in the insert of Figure 5, where the sample disc is sandwiched
strain rates from 10 to 103 s−1 , the split Hopkinson pressure between the incident bar and the transmitted bar. This method
bar (SHPB) is an ideal choice. SHPB was invented in 1949 by was first used to measure the dynamic tensile strength of con-
Kolsky for testing metallic materials (Kolsky 1949). Shortly cretes (Ross et al. 1989, Ross et al. 1995), and was used for
after that, SHPB was attempted by researchers to test brittle studying marbles (Wang et al. 2006) and argillites (Cai et al.
materials such as concretes, ceramics and rocks. We will dis- 2007). However, it is only recently found that the dynamic
cuss in this section about some recent applications of SHPB tensile strength can be determined using the quasi-static equa-
to rocks on the quantification of rock dynamic tensile strength tion only if the dynamic force balance has been achieve in the
and fracture toughness. dynamic test (Dai & Xia 2010). This method was then used to
generate the data shown in Figure 5.
3.4 Dynamic semi-circular bend (SCB) method

3.1 The split Hopkinson pressure bar
The semi-circular bend specimen in the SHPB system is
SHPB is composed of a striker, an incident bar, and a trans- shown schematically in the insert of Figure 5. Provided a quasi-
mitted bar (Figure 4). The impact of a striker bar on the free static state has been achieved in the sample during the test,
end of the incident bar induces a longitudinal compressive using a dimensional argument, the equation for calculating
incident pulse (εi ). Upon reaching the bar-specimen interface, the tensile stress is (Dai et al. 2010b):
part of the incident wave is reflected (reflected wave, εr ) and
the remainder passes through the specimen to the transmitted
bar (transmitted wave, εr ).
149

Table 2. Rock dynamic tests (after Zhao 2011)
Strain Rate (s−1 ) Test Apparatus Testing Principle Applicability
≤102 Hydraulic machines Dynamic load applied by Uniaxial compression (e.g., Green & Perkins 1969,
movement of a piston Zhao et al. 1999a) ; dynamic triaxial compression
hydraulically driven by (e.g., Li et al. 1999)
gas or oil Direct tension (e.g., Yan & Lin 2006, Asprone et al.
2009); dynamic Brazilian indirect tension (e.g.,
Zhao & Li 2000)
Punch shear test (e.g., Zhao et al. 1998)
Shear of rock joints (e.g., Barbero et al. 1996,
Kana et al. 1996)
100 –103 Drop-weight machines Gravitational potential Flexural loading (e.g., Banthia et al. 1989)
energy Impact and fragmentations (e.g., Whittles et al. 2006)
101 –103 Hopkinson bar One dimensional stress Uniaxial compression (e.g., Li et al. 2000, Li et al.
wave propagation theory 2005, Cai et al. 2007, Zhou et al. 2010) ; Triaxial
compression (e.g., Christensen et al. 1972, Li et al.
2008, Frewet et al. 2010)
Direct tension (e.g., Cadoni 2010, Huang et al. 2010a);
Brazilian indirect tension (e.g., Wang et al. 2006,
Cai et al. 2007, Dai & Xia 2010);
Flattened Brazilian disk (FBD) tension (e.g., Wang
et al. 2009);
Semi-circular bend (SCB) test (e.g., Dai et al. 2008);
One-point impact test (e.g., Belenky & Rittel, in press);
Spalling test (e.g., Erzar & Forquin 2010)
>103 Gas gun High-pressure gas driven Equations of state (e.g., Shang et al. 2000)
projectile
where a is the crack length and P(t) is the time-varying loading

force. The dimensionless geometric function Y (a/R) depends
on the crack geometry, and can be calculated with a stan-
dard finite element software package (e.g., ANSYS). The
schematic of NSCB in SHPB is shown in Figure 6. Fracture
toughness KIC is obtained at the maximum load. The fracture
toughness is shown to be strongly rate dependent.
3.6 Dynamic cracked chevron notched Brazilian disc

(CCNBD) method
CCNBD specimen has been suggested by ISRM to determine
static rock fracture toughness (Fowell et al. 1995). Provided a
quasi-static state of the specimen has been achieved during the
SHPB test with pulse shaping, the initiation fracture toughness
Figure 5. Rock tensile strength measured using BD and SCB KIC of the specimen can be determine in a similar way (Dai
methods in SHPB system. et al. 2010a):
where P(t) is the time-varying load recorded in the test, S
is the span of the supporting pins and R is the radius of the
disc. The dimensionless stress Y (S/2R) can be calibrated using
finite element analysis.The flexural tensile strength σf is taken
as the maximum tensile stress in the history of σ(t) and the where, Pmax is the measured maximum load, B and R are the
corresponding loading rate is measured from the slope of the ∗ is the
thickness and the radius of the disc respectively, Ymin
pre-peak linear portion of the curve. The reason why strengths minimum value of Y ∗ , and Y ∗ is the dimensionless SIF and can
measured using SCB method are higher than those by BD has be determined in advance by numerical calibrations accord-
been explained (Dai et al. 2010b). ingly. We can see from Figure 6 that the fracture toughness
obtained from dynamic NSCB and dynamic CCNBD methods
3.5 Dynamic notched semi-circular bend (NSCB) method are consistent (Dai et al. 2010a).
Based on theASTM standard E399-06e2 for rectangular three-
point bending sample (ASTM 2002), a similar equation for
calculating the stress intensity factor for mode-I fracture for 4 TUNNEL STABILITY AND ENGINEERING UNDER
SCB specimen was proposed (Chen et al. 2009): EXPLOSIVE LOADING
Stability of rock tunnels and other engineering structures

under explosions loading is of great interest to engineers
150

Figure 7. Comparison of ground shock from coupled and decoupled
Figure 6. Rock fracture toughness measured using NSCB and explosions (after Zhou 2011).
CCNBD methods in SHPB system.
Prediction of ground shock due to decoupled explosions is
designing for tunnel blasting and for protection of under- more complicated because several factors may affect the trans-
ground structures, including safety of underground explosives fer of explosion energy to the ground. Basically, two types of
storage. Engineering designs involving dynamic loading from techniques can be used for predicting the peak particle veloc-
explosions can follow roughly the following steps: character- ity from decoupled explosions. The first method uses the wall
isation of the explosion source; chacterisation of the ground, or chamber pressure as the starting point and then predicts the
propagation of stress waves in geological media, analysis of ground shock based on some form of wave theory. The predic-
tunnel response and assessment of tunnel damage, and finally tion cannot be easily done in analytical form unless significant
dynamic rock support. simplifications are made. The second method applies a decou-
pling factor to the predicted ground motion based on a fully
4.1 Ground shock coupled explosion of the same quantity. The decoupling factor
is often given as a function of the loading density, expressed
Explosions in rock generate a dynamic stress wave, or ground in kg/m3 (Hulgren 1987, McMahon 1992, Joachim & Smith
shock, that propagates through the geological media. This 1988, Zhou & Ong 1996, Zhou & Jenssen 2009).
stress wave is typically represented by the time history of the This decoupling effect can be seen clearly from Figure 7
acceleration or particle velocity. For the purpose of engineer- based on tests conducted in hard rock (Chong et al. 2002,
ing design, the dynamic load generated from an underground Zhou 2011), where at the same scaled range of 1 m/kg1/3 ,
explosion in rock can be represented by the peak particle the average peak particle velocity was 15 m/s, 1 m/s, and
velocity (PPV), which has been shown by studies to be the 0.4 m/s, for fully coupled and loading densities of 10 kg/m3
most representative parameter when describing the ground and 2.5 mg/m3 , respectively. The respective decoupling factor
motion and tunnel response (Dowding 1984). for the PPV, for loading densities of 10 kg/m3 and 2.5 mg/m are
The PPV from a fully coupled explosion can be given in the 0.067 and 0.027. In other words, at a distance of about 22 m,
following general form: the peak particle velocity produced by a 10,000 kg explosion
with a loading density of 10 kg/m3 will produce only about
7% of the PPV of a fully coupled charge of the same quantity.
4.3 Tunnel damage

where H and n are constants for a certain geology and explo-
sion set up; R = actual distance (m); Q = charge weight (kg). Various criteria have been found in the literature to describe
The term R/Q1/3 is the scaled range, expressed in m/kg1/3 . rock damage, often with significant differences in definition
The exponent B, representing the energy scaling law, is and practical meaning. The damage criteria for Swedish hard
a function of the geometry of the charge and depicts the rock by Persson (1997) suggested threshold damage (incipient
energy transmission from the explosive to the surrounding damage) at a PPV of 1 m/s. Li & Huang (1994) discussed
medium. Many mining applications tend to use 1/2 for B, or the damage criteria for rock tunnels, with the following damage
square root scaling while most military and civil engineering definition:
applications tend to use 1/3 for B, or cube root scaling. • Slight damage – initial cracking
• Medium damage – partial collapse
4.2 Decoupled explosions • Serious damage – large-area tunnel collapse.
Decoupling is a very important concept in blasting engi- The PPV for slight damage seems to correspond to ini-
neering because it strongly affects the ground shock from tial swelling described by Persson (1997). Perhaps the most
explosions. Decoupling occurs when the explosive charge comprehensive study of tunnel damage is the US Army’s
does not fill up the volume completely, and at least some Underground Explosion Tests (UET), as reported in Hendron
parts of the charge are not in direct contact with the rock. (1977). The study classifies tunnel damage into four damage
When decoupling occurs, the peak particle velocity generated zones. The intermittent failure is most likely random spalling
by the same quantity of explosives at the same distance is of loose rocks. Serious spalling (or damage) does not occur
substantially reduced. until the PPV reaches 4 m/s. For practical applications, the
151

tunnel damage criteria defined in the UET tests seems to be summary has been given by Zhao (2011) and presented in
the most realistic one. In studies to examine the damage of this keynote.
rock support due to blasting ahead of the tunnel face, it has
been found that even very close to the blasting face, damage to 5.1 Wave propagation in rock joints
the rock bolt grout and shotcrete is negligible. Studies carried
out by Stjern & Myrvang (1998) and Ortlepp & Stacey (1998) The focus should be on the coupling of wave transmission ad
have shown that PPV up to 1 m/s will not cause any measur- transformation characteristics with joint geometrical proper-
able damage of the rock support. Dowding (1984) suggests ties (spacing, orientation, aperture, and surface profile). For
that the threshold PPV for lined tunnels would roughly double examples, spacing, frequency and orientation can remain as
that for unlined tunnels. For modern tunnel design, steel-fibre geometrical parameters and can be the input for either analyti-
reinforced shotcrete can be treated an effective liner for this cal solutions or numerical modelling. Aperture and roughness
purpose. can be correlated to mechanical properties such as joint normal
Based on the above literature, theoretical analysis and stiffness and shear strength. Therefore, it is possible to incor-
results of large-scale tests conducted in hard rock (Chong porate those rock joint parameters in the wave propagation
et al. 2002, Zhou 2011), it can be concluded that damage analysis, particularly in DEM numerical modelling, to esti-
of unlined tunnels in competent rock does not seem to begin mate wave attenuation in the jointed rock masses. Mechanical
until the incipient PPV reaches a value of at least 1–2 m/s. properties of joint filling (e.g., sand or clay) can be incorpo-
With the addition of tunnel support (such as rock bolts and rated into the wave propagation analytical solutions by treating
fibre reinforced shotcrete), it is expected that the tunnel can the filling as a viscous material.
sustain a much higher load, probably at least 2–4 m/s in PPV.
For such load cases, normal static rock support is probably suf- 5.2 Wave propagation in rock masses
ficient although the use of steel fibre reinforced shotcrete is Studies along this line are to develop equivalent medium wave
recommended for its high energy capacity. The use of dynamic propagation parameters for jointed rock masses, by incor-
support in hard rock is not necessary unless the dynamic porating rock mass parameters. A statistic approach may be
loading, as expressed by the incipient peak particle velocity, adopted to represent the geometrical distribution of joints and
reaches more than 2–4 m/s. of the joint properties for the rock masses.This can be achieved
by performing parametrical study using numerical modelling
4.4 Dynamic rock support to generate a representative rock mass and then to obtain
wave attenuation coefficient for that rock mass.
Support against dynamic loads has been developed based on
the concept of increasing the energy absorbing capacity of the
5.3 Interaction of wave transmission and joint damage
support system in the form of yieldable support (Ansell 1999;
Li, 2010). The earliest and most representative of yielding Joint damage associates with energy loss, can be incorporated
support is the cone bolt. The most recent dynamic rock bolts into the wave propagation equation, by considering the energy
make use of the deformation of the steel bar anchored at balance before and after wave transmission at the joint whose
the end, or anchored at multiple points in the grout hole. surface asperities are under compression and/or shearing.
Both make use of a smooth section of the rock bolt that The interaction between wave transmission and joint damage
easily detaches from the grout when subjected to dynamic can be explored by physical and/or numerical modelling. For
load, allowing the steel bar to deform and thus absorbing numerical modelling, the challenges will be the simulation of
energy, with energy capacities up to 40 KJ per bolt. On the rock joint surface damage under dynamic loads. Microme-
tunnel rock surface between these rock bolts, shotcrete rein- chanic and discrete element modelling is likely required in
forced with wire mesh is the most common method of support order to model the fracturing and surface failure of rock joints.
for containment or restraints. It is well known that steel-fire
reinforced shotcrete greatly increases the energy capacity of 5.4 Rock fracture induced seismic energy and wave
the shotcrete. However, there are currently no standards for
shotcrete design against dynamic loads. Current design stan- When a highly stressed (or strained) brittle material fails, the
dards and guides for steel-fibre reinforced shotcrete defines stored strain energy is released at the facture plane. If the
the energy capacity based on standard tests on panels of 600- energy released is sufficiently large, it can cause induced seis-
mm by 600-mm and 100-mm thickness, supported on four mic events. Physical experiment may offer direct observation
sides and loaded at the centre to a deformation of 25 mm. The on energy release pattern. Numerical micromechanical mod-
latest design chart using the Q-system has included energy elling can also be good tools to capture the dynamic behaviour
capacity specifications up to 1000 Jouls for qualities of very of statically strained rock materials at sudden failure, and the
poor or below or rock tunnels with spans greater 20 m. How- induced dynamic stress wave.
ever, there are no similar specifications for dynamic loads.
This is an area that requires further research and development. 5.5 Mechanics of rock fracturing and rate effects
While the high loading rate leads to increased rock material
5 FUTURE RESEARCH DIRECTIONS OF ROCK strengths and fracture intensity. However, it is not well under-
DYNAMICS AND ENGINEERING stood yet the cause of high fracture density. The mechanical
and physical causes of rate effects on rock strength and failure
Rock dynamic research has a very wide scope, ranging from pattern need to be explored and revealed. Rate effects on frac-
wave propagation, to response of rock material and rock mass, ture branching, on multiple fracture initiation, and on crack
to engineering applications, dealing with microscopic fractur- propagation velocity are some examples. It also includes the
ing of rock material to blocky movements of rock masses. study on the shear strength of rock joint under dynamic loads,
There are many issues requiring further investigations. A to understand the rate effects on shear strength and dilation.
152

5.6 Micromechanic modelling of rock fracturing and failure chevron-notched Brazilian disc (CCNBD) specimens. Interna-
tional Journal of Rock Mechanics and Mining Sciences &
Micromechanic modelling of rock fracturing can be per- Geomechanics Abstracts 32: 57–64.
formed by incorporating micromechanical constitutive laws Frew, D.J., Forrestal, M.J. & Chen, W. 2002. Pulse shaping techniques
and input parameters into the existing codes, such as UDEC for testing brittle materials with a split Hopkinson pressure bar.
and DDA. New microscale numerical codes with specific Experimental Mechanics 42: 93–106.
focus on modelling fracture initiation, propagation and Hasnaoui, A., Van Swygenhoven, H. & Derlet, P.M. 2003. Dimples
branching should also be developed. The other question that on nanocrystalline fracture surfaces as evidence for shear plane
micromechanics modelling should address is the effect of formation. Science, 300: 1550–1552.
element size, incorporating the Theory of Critical Distance Hendron, A.J. 1977. Engineering rock blasting in civil projects. In:
W.J. Hall (ed), Structural and Geotechnical Mechanics: A Volume
(Taylor 2007). Honouring Hathan M. Neumark: Prentice Hall.
Hultgren, S. 1987. Ground Motion – Measurements, KLOTZ II.
Report C4:87, Fort F, 1987.
5.7 Rock and earthquake engineering applications Joachim, C.E. & Smith, D.R. 1988. WES underground magazine
model tests. In Proceedings of the 23rd Department of Defence
Rock dynamics will be continuously applied to engineering Explosives Safety Seminar.
and construction. One particular area requiring research is Kolsky, H. 1949. An investigation of the mechanical properties of
damage assessment and rock support under dynamic load. materials at very high rates of loading. Proceedings of the Royal
For examples, stability of slopes and tunnels under various Society A-Mathematical Physical and Engineering Sciences B62:
dynamic conditions (earthquake and explosion), reinforce- 676–700.
ment and support of rock slope and tunnels for dynamic Lei, Z. & Zang, M. 2010. An approach to combining 3D discrete
and finite element methods based on penalty function method.
loads, use of explosives and blast damage control, seismic and
Computational Mechanics 46: 609–619.
vibration hazards control, and many other rock engineering Li. C.C. 2010. A new energy-absorbing bolt for rock support in high
applications involve dynamic loading and dynamic behaviour. stress rock masses. International Journal of Rock Mechanics and
Mining Sciences 47: 396–404.
Li, X.B., Lok, T.S., Zhao, J. & Zhao, P.J. 2000. Oscillation elim-
REFERENCES ination in the Hopkinson bar apparatus and resultant complete
dynamic stress-strain curves for rocks. International Journal of
Ansell, A. 1999. Dynamically loaded rock reinforcement. TRITA- Rock Mechanics and Mining Sciences 37: 1055–1060.
BKN. PhD Thesis. Royal Institute of Technology, Stockholm, Li, Z. & Huang, H. 1994. The calculation of stability of tunnels under
Sweden. the effects of seismic wave of explosions. In Proceedings of 26th
ASTM 2002. Standard E399-90. Standard test method for plane Department of Defence Explosives Safety Seminar.
strain fracture toughness of metallic materials. Annual book of Ma, G.W., An, X.M., Zhang, H.H. & Li, L.X. 2009. Model-
ASTM Standards. ing complex crack problems with numerical manifold method,
Cai, M., Kaiser, P.K., Suorineni, F. & Su, K. 2007. A study on the International Journal of Fracture 156: 21–35.
dynamic behavior of the Meuse/Haute-Marne argillite. Physics McMahon, G.W. 1992. Ground motions from detonations n under-
and Chemistry of the Earth 32: 907–916. ground magazines in rock. In Proceedings of 25th Department of
Chen, R., Xia, K., Dai, F., Lu, F. & Luo, S.N. 2009. Determina- Defence Explosives Safety Seminar.
tion of dynamic fracture parameters using a semi-circular bend Morris, J.P., Rubin, M.B., Block, G.I. & Bonner, M.P. 2006. Simu-
technique in split Hopkinson pressure bar testing. Engineering lations of fracture and fragmentation of geologic materials using
Fracture Mechanics 76: 1268–1276. combined FEM/DEM analysis. International Journal of Impact
Chen, S.G. & Zhao, J. 1998. A study of UDEC modelling for blast Engineering 33: 463–473.
wave propagation in jointed rock masses. International Journal of Mullins, M. & Dokainish, M.A. 1982. Simulation of the (001)
Rock Mechanics and Mining Sciences 35: 93–99. plane crack in alpha-Iron employing a new boundary scheme.
Chiou,Y.J., Lee,Y.M. & Tsay, R.J. 2002. Mixed mode fracture propa- Philosophical Magazine A 46: 771–787.
gation by manifold method. International Journal of Fracture 114: Munjiza, A. 2004. The combined finite-discrete element method.
327–347. London: John Wiley & Sons.
Chong, K., Zhou, Y., Seah, C.C. & Lim, H.S. 2002. Large-scale Munjiza, A, Owen, D.R.J. & Bicanic, N. 1995. A combined finite-
tests – airblast, ground Shock, and debris. In Proceedings of the discrete element method in transient dynamics of fracturing solids.
International Symposium on Defence Construction, Singapore, Engineering Computations 12: 145–174.
17–18 April 2002. Onate, E. & Rojek, J. 2004. Combination of discrete element
Dai, F. & Xia, K.W. 2010. Loading rate dependence of tensile strength and finite element methods for dynamic analysis of geome-
anisotropy of Barre granite. Pure and Applied Geophysics 167: chanics problems. Computer Methods in Applied Mechanics and
1419–1432. Engineering 193: 3087–3128.
Dai, F., Chen, R., Iqbal, M.J. & Xia, K. 2010a. Dynamic cracked Ortlepp, W.D. & Stacey, T.R. 1998. Performance of tunnel sup-
chevron notched Brazilian disc method for measuring rock fracture port underground large deformation static and dynamic loading.
parameters. International Journal of Rock Mechanics and Mining Tunnelling and Underground Space Technology 13: 15–21.
Sciences 47: 606–613. Pan, X.D. & Reed, M.B. 1991. Coupled distinct element-finite
Dai, F., Xia, K.W. & Tang, L.Z. 2010b. Rate dependence of the flex- element method for large deformation analysis of rock masses.
ural tensile strength of Laurentian granite. International Journal International Journal of Rock Mechanics and Mining Sciences
of Rock Mechanics and Mining Sciences 47: 469–475. 28: 93–99.
Dowding, C.H. 1984. Estimating earthquake damage from explosion Persson, P.A. 1997. The relationship between strain energy, rock
testing of full-scale tunnels. Advances in Tunnel Technology and damage, fragmentation, and throw in rock blasting. International
Subsurface Use 4: 113–117. Journal of Blasting and Fragmentation 1: 99–110.
Elmekati,A. & Shamy, U.E. 2010.A practical co-simulation approach Ross, C.A., Tedesco, J.W. & Kuennen, S.T. 1995. Effects of strain-rate
for multiscale analysis of geotechnical systems. Computers and on concrete strength. Aci Materials Journal 92: 37–47.
Geotechnics 37: 494–503. Ross, C.A., Thompson, P.Y. & Tedesco, J.W. 1989. Split-Hopkinson
Fowell, R.J., Hudson, J.A., Xu, C. & Chen, J.F. 1995. Suggested pressure-bar tests on concrete and mortar in tension and compres-
method for determining mode-I fracture toughness using cracked sion. Aci Materials Journal 86: 475–481.
153

Rousseau, J., Frangin, E., Marin, P. & Daudeville, L. 2009. Multi- Xiao, S.P. & Belytschko, T. 2004. A bridging domain method for
domain finite and discrete elements method for impact analysis of coupling continua with molecular dynamics. Computer Methods
a concrete structure. Engineering Structures 31: 2735–2743. in Applied Mechanics and Engineering 193: 1645–1669.
Shi, G.H. 1991. Manifold method of material analysis. In Transac- Yan, B., Regueiro, R.A. & Sture, S. 2010. Three-dimensional ellip-
tions of the 9th Army Conference on App Math and Comput, U.S. soidal discrete element modeling of granular materials and its
Army Research Office, Minneapolis, US. coupling with finite element facets. Engineering Computations
Stjern, G. & Myrvang, A. 1998. The influence of blasting on grouted 27: 519–550.
rock bolts. Tunnelling and Underground Space Technology 3: Zhao, G.F. 2010. Development of micro-macro continuum-
65–70. discontinuum coupled numerical method. PhD thesis. École Poly-
Subhash, G., Ravichandran, G. & Gray, G.T. 2000. Split-Hopkinson technique Fédérale de Lausanne (EPFL), Lausanne, Switzerland.
pressure bar testing of ceramics. In Kuhn, H. & Medlin, D. Zhao, G.F., Shi, G.H., Zhao, J. 2011. Manifold and advanced
(eds), ASM Handbook Vol 8, Mechanical Testing and Evaluation. numerical techniques for discontinuous dynamic computations.
Materials Park: ASM Int. In Y.X. Zhou & J. Zhao (eds), Advances in Rock Dynamics and
Tadmor, E.B., Ortiz, M. & Phillips, R. 1996. Quasicontinuum analysis Applications. Taylor & Francis, 321–344.
of defects in solids. Philosophical Magazine A 73: 1529–1563. Zhao, J. 2011. An over view of some recent progress in rock dynamic
Taylor, D. 2007. The Theory of Critical Distances: A New Perspective research. InY.X. Zhou & J. Zhao (eds),Advances in Rock Dynamics
in Fracture Mechanics. Oxford: Elsevier. and Applications. Taylor & Francis, 5–33.
Terada, K., Ishii, T., Kyoya, T. & Kishino, Y. 2007. Finite cover Zhou, Y. 2011. Explosion loading and tunnel response. In Y.X. Zhou
method for progressive failure with cohesive zone fracture in het- & J. Zhao (eds), Advances in Rock Dynamics and Applications.
erogeneous solids and structures. Computational Mechanics 39: Taylor & Francis, 457–481.
191–210. Zhou, Y. & Jenssen, A. 2009. Internal separation distances for
Wang, Q.Z., Li, W. & Song, X.L. 2006. A method for testing dynamic underground explosives storage in hard rock. Tunnelling and
tensile strength and elastic modulus of rock materials using SHPB. Underground Space Technology 24: 119–125.
Pure and Applied Geophysics 163: 1091–1100. Zhou,Y. & Ong,Y.H. 1996. Ground shock prediction methods – a crit-
Xia, K., Nasseri, M. H.B., Mohanty, B., Lu, F., Chen, R. & Luo, ical appraisal. In Proceedings of the 1st Asia-Pacific Conference
S.N. 2008. Effects of microstructures on dynamic compression on Shock and Impact Loads on Structures.
of barre granite. International Journal of Rock Mechanics and
Mining Sciences 45: 879–887.
154

Analysis and design methods

Mine-by experiment performed in the Callovo-Oxfordian claystone at the Meuse Haute

Marne Underground Research Laboratory (France)
G. Armand, A. Noiret, M. Cruchaudet & N. Conil

National Radioactive Waste Management Agency (ANDRA), Centre de Meuse Haute-Marne, Bure, France
Subject: Hydro mechanical behavior of claystone
Keywords: mine-by experiment, in situ experiment, claystone, hydro mechanical behavior, Excavated damaged zone
In 2000, the French National Radioactive Waste Management

Agency (ANDRA) started to build an Underground Research
Laboratory (URL) at Bure (East of the Paris sedimentary
Basin) to perform experiments in order to get in situ data
and to demonstrate feasibility of deep geological repository
in a claystone. The main objective of the research was to
characterize the confining properties of the clay through in
situ hydro-geological tests, chemical measurements and dif-
fusion experiments and to demonstrate that the construction
and operation of a geological repository will not introduce
pathways for waste migration (Delay et al. 2005). In 2005, a
report (ANDRA 2005) exhibited that the Callovo-Oxfordien
claystone in the area of the URL is a reliable host rock to
perform a deep geological repository.
Further researches are ongoing in the URL to increase the Figure 1. General layout of the Meuse/Haute-Marne URL.
knowledge on the rock properties, to test repository concepts
at full scale and to optimize its. In geomechanic, the long-
term time-dependent deformation characteristics of this rock section of the drift and a view of the completed drift with the
control the evolution of the Excavation Damaged Zone (EDZ) compressible concrete wedges.
around underground openings and are a key issue to design A mine-by experiment is a state-of-the-art project to char-
drift support. It is of first importance to well understand the acterize excavation-induced damage and determine relation-
impact of the excavation and support methods on the EDZ ships governing the behaviour of a rock mass around an
and the hydro-mechanical behaviour of the claystone around underground opening. Before the beginning of the excavation
the drift. The discontinuities formed within the EDZ represent work, 15 boreholes (nearly 80 measurement points) have been
one of the possible pathways for the migration of radionuclide emplaced from surrounding drifts to measurement evolution
towards the surface.The knowledge of the mechanisms of their pore pressure and deformation due to excavation the GCS drift
creation and the control of their evolution will built confidence (fig. 3). 6 convergence sections and 2 section of measurement
in the performance assessment of nuclear waste repositories on support (deformation of the compressible concrete wedges
in claystone host rocks. and total pressure on it) have been performed during the exca-
A huge program of experiment is planned to characterize the vation work. 4 extensometers (30 m long) have been emplaced
response of the rock to different drift construction methods. during a work stop to follow deformation when the excavation
Before 2008 at the main level of the laboratory, most of the resumed.
drifts have been excavated with pneumatic hammer and sup- Further boreholes will be performed to characterize extend
ported with rock bolts, sliding steel arches and fibre shotcrete. and properties (mainly permeability) of the EDZ and there
Other techniques (road header, stiff and flexible support, evolution as a function of time, by the way of seismic mea-
tunnelling machines, . . .) will be tested later to characterize surements, pore pressure and permeability measurements and
their impacts (fig. 1). geological mapping on cores.
This paper present the mine-by experiment concept and The paper shows the first data recorded during the digging
results obtained during the excavation with a road header of of the GCS drift. Five converge ropes between 6 anchored plots
a 60 m long drift called “soft concept gallery” (GCS). The are used to measure convergence in each section. Figure 4
support is flexible in order to let developing long term tun- shows the horizontal and vertical convergences measured in
nel closure (convergences .The new support techniques is GCS drift as a function of time. The horizontal convergences
rock bolts with a layer of nearly 20 cm of fiber reinforced are perceptibly higher than the vertical ones (ratio ≈ 2), even if
shotcrete within 12 yieldable concrete wedges (hiDCon® ). the state of stress is nearly isotropic. However, time evolution
Figure 2 shows excavation result of the road header, the cross of horizontal and vertical convergence is similar. More than
157

Figure 4. Convergence measurements (a) vertical, (b) horizontal.
Figure 2. GCS drift: (a) Road header, (b) section, (c) view of GCS
80% of the convergence has been reached during the first 100
drift.
days. The convergence rates decrease as a function of time.
Other results in terms of deformation, loading of the support
and in term of pore pressure are presented and discussed.
As a conclusion, a new drift, parallel to the horizontal in
situ major stress, has been excavated at the main level of
the Meuse Haute Marne URL in order to perform a mine-
by experiment. Among the 200 sensors installed previously
to the digging of the GCS drift, more than 95% working
during the shaft sinking, thus providing a important data
base to understand the coupled hydro mechanical behaviour
of the Callovo-Oxfordian claystone. Measured parameters
(mainly displacements and pore pressures) are consistent.
Further analyses of the hydro mechanical behaviour have to be
conducted, accounting the result on the ongoing EDZ charac-
terization in the GCS drift. The long term displacements and
pore pressure evolution will continue to be recorded in order
Figure 3. Concept of the GCS experiment. to study the long term behaviour.
158

Burgers rock creep around axisymmetric tunnels
S.D. Athanasopoulos, P.P. Nomikos & A.I. Sofianos

Tunnelling Laboratory, School of Mining and Metallurgical Engineering, National Technical University, Athens, Greece
Keywords: numerical modelling, weak rock, tunnelling
Squeezing in tunnelling may provoke various severe problems installation of a shotcrete lining during excavation. These
during construction, and therefore it is of high importance for are installed in stages, following the support installation
tunnel design. The squeezing phenomenon is closely asso- procedure of an actual tunnel case.
ciated with creep, which is essentially a time – dependent Figure 3 displays the normalized radial displacement pro-
behaviour of the rock mass. In this study, the time – depen- files (normalized LDPs) along the tunnel axis for various
dent rock deformation in tunneling is examined numerically, times up to 1 year time period, for an unsupported tunnel. It
using the Burgers viscoelastic rheological model. Burgers’ is observed that both the elastic and the viscoelastic normal-
body is a four – constant viscoelastic model, formed by com- ized deformation profiles are almost identical. Furthermore,
bining in series Kelvin’s and Maxwell’s creep bodies. The a good correspondence of these profiles with an empirical
four viscoelastic parameters are the Maxwell’s and Kelvin’s LDP is observed.
shear moduli G M and G K and viscous coefficients nM and nK
respectively. This idealized model is considered to be a first
approximation of the actual behaviour of rocks, as it is able to
represent both primary and secondary creep.
The problem of the Burgers rock creep around an advanc-
ing axisymmetric tunnel is investigated numerically with a
three dimensional finite differences code. The model prepared
is shown in Figure 1. Appropriate boundary conditions are
applied in order to take into consideration the axisymmetric
properties of the problem (Fig. 2).
Two model series are prepared and examined. In the first
series (A–series model), the excavation is performed incre-
mentally without considering the rock mass creep. Then the
rock mass creep is activated and the model response is exam-
ined for a time period of 1 year. In the second model series
(B–series model), the rock mass creep during excavation is Figure 2. Boundary conditions of the three dimensional axisym-
taken into account. Both supported and unsupported mod- metric models (xy–plane and zx–plane).
els are examined. Shell elements are used to simulate the
Figure 3. Longitudinal deformation profile of the normalized

Figure 1. FLAC 3D grid for the axisymmetric tunnel models with radial displacement (normalized LDP) for 1 year time period
3.840 finite differences zones. (A–series unsupported model).
159

Figure 6. Comparison of the longitudinal deformation profiles of
the normalized radial displacement of a measuring point at y = 30 m
Figure 4. Longitudinal deformation profile of the radial displace- for 1 year time period, of an unsupported and a supported tunnel
ment (elastic response) of a measuring point at y = 30 m (A–series (A–series model).
unsupported model).
Figure 7. Axial thrust developed on the shotcrete lining at sections

Figure 5. Longitudinal deformation profile of the radial displace- far from the tunnel face for the A–series model and comparison to a
ment of a measuring point at y = 30 m (B–series unsupported closed form solution.
model).
Figure 7 displays the development of the axial thrust on the

In Figures 4 and 5 the radial displacement at the tunnel shotcrete lining at sections far from the excavation face, for
boundary is shown with respect to the distance from the tun- the A–series model, in comparison to a closed form solution.
nel face, for the A–series and B–series unsupported models Comparing the analytical results with the numerical results of
respectively. As the excavation face advances, the curve of the theA–series models, although the final value of the axial thrust
displacement of the A–series model is completely smooth, of the two curves is approximately equal, an overestimation
following an elastic LDP, while the one of the B–series takes of the initially developed forces, by the closed form solution
a stepping form, due to rock mass creep during excavation. is noticed.
After the excavation has passed for about 20 m from a mea- Concluding, a comparison of the longitudinal deforma-
suring point at y = 30 m (i.e. a distance of 30 excavation steps tion profiles of the A–series models without support, to an
from the model boundary), the displacement of the A–series empirical LDP curve shows good agreement. The deforma-
model tends to stabilize. On the other hand, the displacement tion profiles of the respective supported models, are clearly
of the B–series model is continuously increasing. Until the steeper near the tunnel face.
excavation is stopped, at 80 m from the model boundary, the When rock mass creep during excavation is taken into
displacement of the B–series model is almost 2.5 times larger account (B–series models), the development of the radial
than the one of the A–series model. displacement indicates a constant rate linear progress of the
In Figure 6 the development of the radial displacement at creep phenomenon. The allowance of creep behaviour during
the same measuring point of the supported B–series model excavation leads the rock mass earlier into the stage of steady
is shown. Comparing this curve to that of the unsupported state deformation.
tunnel (Fig. 5), it becomes clear that the magnitude of the In all supported models, the shotcrete lining leads the dis-
displacement is much lower due to the support, even though placements to equilibrium very fast where the axial forces tend
it is not fully stabilized when the tunnel face is approximately asymptotically to their ultimate value.
50 m ahead. Furthermore, the activation of creep during each
excavation step does not seem to affect the curve into taking
a stepping form.
160

Multi-parameter responses of soft rocks during deformation and fracturing and their
implications in geomechanics and geoengineering
Ö. Aydan
Department of Marine Civil Engineering, Tokai University, Shizuoka, Japan
Y. Ohta
Japan Nuclear Energy Safety Organization, Tokyo, Japan
H. Tano
Department of Civil Engineering, Nihon University, Koriyama, Japan
Subject: Rock material and rock mass property testing
Keywords: rock failure, lab testing, dynamics, monitoring, rock stress, rock bursts
When rock starts to fail, the stored mechanical energy in

rock tends to transform itself into different forms of energy.
Experimental studies by Aydan and his group (Aydan et al.
2001, 2003) showed that rock indicates distinct variations
of various measurable parameters such as electric poten-
tial, magnetic field acoustic emission, resistivity etc. besides
load and displacement, which are called multi-parameters,
during deformation and fracturing processes. Furthermore,
some in-situ monitoring schemes were developed for struc-
tural safety of tunnels as well as for earthquake prediction
studies (Aydan et al., 2005). These variations may be useful
in predicting the failures of rock structures in geoengineering
well as earthquakes in geoscience.
The authors undertook a new series of experiments on soft
rocks, using the multi-parameter measurement system involv-
ing electric potential (EP) variations, electrical resistivity,
acoustic emissions (AE), rock temperature (RT), and accel-
eration in addition to the conventional displacement (strain) Figure 1. Experimental set-up.
and load (stress) measurements (Figures 1 and 2). This new
series of experiments was carried out under purely mechan-
ical manual loading without using the loading device, which
may utilize AC current as power source. Therefore, the load-
ing system is free of electric noise, which may be caused by
the loading system. This experimental study was undertaken
to make further contributions to the understanding of electric
potential variations during fracturing and sliding process of
piezoelectric and non-piezo-electric geomaterials. From the
experimental results described in this report, the following
conclusions may be drawn:
1) The experimental results clearly indicate that the defor-
mation, fracturing and sliding processes induce electric
potential in geomaterials.
2) The magnitude of induced electric potential both depends
upon the piezo-electric characteristics of minerals or grains
and the moment caused by the separation of electrons of
minerals as a result of deformation and inter-crystal or
inter-grain separation and/or sliding during dislocations as
a result of fracturing or sliding.
3) The amplitude of accelerations of the mobile part of the Figure 2. Multi-parameter response of soft rocks.
loading system is higher than that of the stationary part.
This feature has striking similarities with the strong motion
records nearby earthquake faults observed in the recent
large in-land earthquakes.
161

Fractal approach to determine rock mass strength and deformation
M.N. Bagde, A.K. Raina, P.B. Choudhury & A. Sinha*

Central Institute of Mining and Fuel Research, Regional Center, Nagpur, India
*Central Institute of Mining and Fuel Research, Dhanbad, India
Keywords: Rock mass, rock mass classification, blasting, mining
Rock tests constitute one of the bases for theoretical and The rock mass strength were determined averaging three
numerical analysis and design, and are important tools for most popular methods recommended by Sheorey (1997),
studying rock mechanics and engineering. Rock mass strength Truemann (1999) and Laubscher (1990). The following rela-
and modulus have got several applications in rock excavation tion has been obtained between the pre and post-blast Frac-
engineering. Presently it is difficult to determine accurately tal Dimensions, specific charge (q) and average rock mass
the relevant parameters used in computation, since, rock mass strength.
is characterized by its discontinuities in the form of joints, frac-
tures, and other micro-fissures. Many empirical approaches
are presently available for this purpose. The data collection
procedure for this purpose requires considerable field study
and laboratory work. Also it was felt many a time that the end where,
result is person specific. Hence there is a possibility of loos-
ing preciseness and reliability of the conclusions drawn on the
basis of the data.
The term “fractal” is perhaps best described as a general
concept which is relevant in describing the geometry of irreg-
ular objects or patterns. The fractal concept was originally Similarly, the average Deformation modulus (Ed ) was esti-
introduced by Mandelbrot (1967). A simple definition of a mated from three popular relations given by Serafim and
fractal is a shape made of parts similar to the whole in some Pereira (1983), Hoek & Brown (1997) and Mitri (1994). The
way. The fractal concept has been applied to many aspects of following relation is obtained between the pre and post-blast
nature. Fractal concept has been applied to rock mechanics Fractal Dimensions, specific charge (q) and average Ed .
since its initiation in the early 1980s. A number of researchers
have been devoted to the fractal characterization of fragmen-
tation. The fragmentation method can be well understood by
in-situ and blasted block size distribution using digital image where,
analysis technique. The Fractal Dimensions can be well related
with the distribution.
In the present study, fractal approach to estimate various
rock and rock mass properties from the Fractal Dimensions
of the in-situ rock mass and blasted rock fragments will be
presented.
The study includes the field investigations to collect the rock
mass and rock samples, blasting information and photographs
of the in-situ and blasted fragments, evaluation of the rock and The above relation is in line with that proposed by Mitri
rock mass properties, estimation of Fractal Dimensions of the (1994).
in-situ blocks (pre-blast) and blasted fragments (post-blast) In the present study all the observation were made in the
and relating those with the rock and rock mass properties. A field using the available resources and without much varia-
total of 22 sites were covered in the field investigations. The tion in the on-going blast design parameters. Studies in the
various formations covered are sandstone, limestone, granite, models may be explored with single hole blasting varying one
manganese, basalt, sillimanite, flourite, iron ore (hematite), or more design parameters for further corroboration of the
laterite and bauxite. The pre and post-blast data were collected derived conclusions.
in the referred sites.
162

A theoretical analysis of core extrusion and squeezing behaviour in tunnelling
L. Cantieni & G. Anagnostou

ETH Zurich, Switzerland
R. Hug
Ernst Basler + Partner AG, Zurich; formerly ETH Zurich, Switzerland
Keywords: Tunnelling, weak rock, numerical modelling, monitoring
1 INTRODUCTION
Squeezing intensity can vary greatly over short distances, even

where there is no change in the excavation method, tempo-
rary support, depth of cover or lithology (Kovári 1998). The
ability to predict conditions ahead of the face is therefore of
paramount importance, because it enables corrective adapta-
tions to be made to the temporary support, excavation diameter
and final lining during construction.
The analysis of deformation measurements in the ground
ahead of the face looks promising with regard to ground
response predictions, as the radial loading and axial defor-
mation of the core ahead of the face can be seen as a large
scale in-situ test.
The present paper investigates numerically the possibility
of predicting the ground response to tunnelling by assessing
the axial extrusion of the core ahead of the face.
2 OBSERVATIONS
Figure 1. Axial strain εy as a function of the distance d between the
The ground may respond more or less quickly to tunnel exca- face and the measuring point for advance rates of 3.2 and 0.8 cm/h
vation, depending on its rheological properties. Slow ground (after Myer et al. 1981).
response may reduce the extrusion of the core significantly,
thus making it difficult to predict squeezing intensity. Myer
et al. (1981) investigated the effect of the advance rate on the is modelled as linearly elastic (according to Hooke’s law)
axial strain ahead of the face through the use of physical mod- and perfectly plastic (according to the Mohr-Coulomb yield
els. Figure 1 shows the measured effect of the advance rate on criterion, with a non-associated flow rule).
extrusion. It shows clearly that the faster the advance rate, the As a reference point, the case of an unsupported tunnel
smaller will be the extrusion of one and the same material. excavated through ground with time-independent behaviour
will be discussed first. Figure 2 shows the tangential strain
εt,c as a function of the axial strain at the centre of the face
3 THEORETICAL ASPECTS εy (0) for different values of the normalized uniaxial compres-
sive strength fc /p0 and the friction angle φ. The conditions
The present section analyses the response of the core ahead of that lead to high axial strains at the face lead also to larger
the face numerically in order to gain a better understanding of convergences of the tunnel. As the relationship is unique, pre-
the observed behaviour, and to investigate whether there is a diction is theoretically possible. The dashed lines in Figure 2
correlation between extrusions and convergences. Emphasis show that most values are in the range of εt,c /εy (0) = 1 to 2.
will be placed on the effect of the time-dependency of ground Squeezing ground often exhibits a pronouncedly time-
behaviour. dependent response to tunnelling. The deformations in a cavity
The numerical analyses will focus on an axisymmetric may continue for several weeks or even months after exca-
model of a deep, cylindrical tunnel crossing a homogeneous vation. As the time scales for core extrusion (a short-term
and isotropic ground which is subject to uniform and hydro- phenomenon) are different to those for convergence (a long-
static initial stress. The mechanical behaviour of the ground term phenomenon), it is interesting to investigate the extent to
163

Figure 2. Unsupported tunnel: Normalized tangential strain Figure 3. Tangential strain at the tunnel boundary εt,c as a function
εt,c ∗ E/p0 over normalized axial strain at the centre of the face of the axial strain at the centre of the face εy (0), the normalized
εy (0) ∗ E/p0 (after Cantieni 2011). viscosity η ∗ v/(a ∗ p0 ) and the normalized uniaxial compressive
strength fc /σ0 (after Cantieni 2011).
which the rheological behaviour of the ground might influ-
ence the correlations between these two manifestations of difficult to predict ground response, because extrusions are
squeezing behaviour. governed by short-term behaviour, while the final ground
The time-dependency of the ground behaviour is handled by response is governed by long-term behaviour. On the one hand,
means of the elasto-viscoplastic creep model after Madejski if the ground behaviour is time-dependent, the fact that the
(1960), which introduces only one additional parameter to core extrusion is low does not necessarily mean that conver-
the parameters used in the preceding elasto-plastic compu- gences will be small. On the other hand, large core extrusions
tations. The micro-mechanical model consists of an elastic are always associated with poor quality ground. It can also be
spring in series with a Bingham model. The calculations have said that a large extrusion represents a sufficient, but not a
been carried out for different values of the viscosity η. necessary, condition for large convergences to occur.
Figure 3 shows that as a consequence of the viscous
behaviour (which is decisive mainly for deformations ahead
of the face) the ratios of convergence to axial displacement REFERENCES
are in general higher than in the case of time-independent
behaviour. The ratio εt,c /εy = 1−2 is valid only if the dimen- Cantieni, L. 2011. Spatial effects in tunnelling through squeezing
sionless parameter ηv/(ap0 ) is lower than about 2.5. At very ground. Dissertation ETH No. 19508, ETH Zurich, Switzerland.
Kovári, K. 1998. Tunnelling in Squeezing Rock. Tunnel 5:12–31.
high viscosities η and advance rates v, the axial strain at the
Madejski, J. 1960. Theory of non-stationary plasticity explained on
face does not depend significantly on the uniaxial compres- the example of thick-walled spherical reservoir loaded with inter-
sive strength fc of the ground, because the strains developing nal pressure. Archiwum Mechaniki Stosowanej 5/6 (12):775–787.
ahead of the face are almost entirely elastic. In such cases it is Myer, L.R., Brekke, T.L., Dare, C.T., Dill, R.B., Korbin, G.E.
impossible to predict the convergences of the opening on the 1981. An Investigation of Stand-up Time of Tunnels in Squeez-
basis of the observed extrusion. ing Ground. In: Rapid Excavation and Tunnelling Conference,
San Francisco, California, pp. 1415–1433.
4 CONCLUSIONS
It is theoretically possible to predict ground response when the

ground exhibits only a moderately time-dependent behaviour.
Pronounced time-dependent ground behaviour makes it very
164

Applications of strain based damage criterion in geotechnical engineering
Yanting Chang
Geton Teknik, Stockholm, Sweden
Subject: Analysis techniques and design methods, earthquake monitoring
Keywords: Rock failure, rock bursts, stability analysis, monitoring, filed measurements
For geotechnical engineering, displacements or ground move-

ments are often measured for stability monitoring or for design
evaluations, since it is much easier and more economical to
measure displacements or strains than stresses. To enable eval-
uations of stability issues on the basis of the acquired strains,
it is preferable to have a rock damage criterion in term of
strains. For this purpose, the following 3-dimensional crite-
rion for rock damage in term of strains has been proposed by
Chang (2006):
where εv is the total volumetric strain; ε1 is the total major

principal strain; κ and εc are material constants. Note that
compressive strain is defined as positive. Another damage cri-
terion in term of strains, similar to the Drucker-Prager stress
criterion, is also proposed by Chang (2006):
where α and β are the hardening/softening parameters

for general cases; and J2 (ε) = 1/6[(ε1 − ε2 )2 + (ε2 − ε3 )2 + Figure 1. Typical ε1 − εv curve of a triaxial test. Quartz monzodi-
(ε3 − ε1 )2 ] is the second invariant of the deviatoric strain ten- orite, confinement 2 MPa (Eloranta, 2004).
sor. For perfectly plastic models, α and β can be treated as con-
stants. Theoretically equation (2) is easier to be implemented
in numerical formulations than equation (1).
Figure 1 shows a typical ε1 − εv curve obtained from a
triaxial test (Eloranta, 2004), where the point for maximum
contraction is indicated. At this point, Lockner et al (1992)
suggested that the microscopic failure of the sample initiates
and crack interaction occurs, so that the sample starts to dis-
play macro-scale dilation.Therefore, it has been suggested that
this point corresponds to the initiation of progressive cracking
in rock samples (e.g. Martin, 1994).
Figure 2 shows the compiled test data collected from the
literatures, where the pairs of the volumetric and axial strains
corresponding to the maximum contraction are plotted. The Figure 2. Volumetric strains and axial strain corresponding to
results strongly suggest that there is a linear correlation maximum compaction. Compiled test data collected from literatures.
between the volumetric strain and the major principal strain
at the maximum contraction. The linear regression of the test well as for intact rocks. It could have significant impor-
data shown in Figure 2 yields a correlation coefficient with tance for geotechnical engineering if this hypothesis would
R2 = 0.950 and the values of the material constants defined in have been validated. More research work is therefore highly
equation (1) are κ = 0.634; εc = −0.060%. recommended in this field.
The findings of Sakurai (1999) suggest a hypothesis: a To enable practical applications of the strain based damage
strain based damage criterion has, in general, negligible scale criterion for safety assessment, an engineering methodol-
effects. In other words, a strain based damage criterion could ogy called strain path analysis (SPA) in ε1 − εv space was
be, for engineering purposes, applied for rock masses as introduced by Chang (2006) and further developments are
165

Figure 5. Interpretation of strain monitoring data for tunnel.
Figure 3. The ε1 − εv space with different domains and an example

of a strain path.
Figure 4. Different domains of failure behavior defined in the Figure 6. Illustration of stress and strain cycles associated to
ε1 − εv space. seismological events.
presented in Figure 3 and 4. Different rock behaviors, such REFERENCES

as brittle or ductile behaviors, can be detected by studying
strain paths in the ε1 − εv space. Chang, Y. 2006. Criterion for rock strain strength and its application.
For tunnel applications, it is proven theoretically that strain- Proc. 4th Asian Rock Mechanics Symposium, 8–10 November,
meters installed in the tunnel surfaces could detect surface Singapore.
failures e.g. rock burst or spalling by analyses of strain Eloranta, P. 2004. Oskarshamn site investigation – Drill hole
increments (see Figure 5), without knowing the initial strains. KSH01A, Triaxial compression test. Report no. P-04-183.
For applications in seismological monitoring (see Figure 6), Swedish Nuclear Fuel and Waste Management Co, Stockholm,
it suggests that volumetric dilatancy associated with rock Sweden.
damage/cracking could be a new viewpoint for understanding Sakurai, S. 1999. Interpretation of field measurements in tunneling
the mechanisms of precursory phenomena for earthquakes. practice. Proc. International Congress on Rock Mechnics. Paris.
Scholz, C.H. 2002. The mechanics of earthquakes and faulting.
The strain damage criterion could be used as a complemen-
2nd ed. Cambridge University Press, UK.
tary tool in the earthquake dilatancy models (Scholz, 2002),
for analyses and explanations of the various precursory phe-
nomena, e.g. uplifting of ground levels, changes of water flow
and wave velocity changes etc.
166

Solving some problems of geomechanics on the base of defining relations of post-limit

deformation of rocks
A.I. Chanyshev, I.M. Abdulin & O.E. Belousova

Institute of Mining, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Subject: modelling and numerical methods.
Keywords: numerical modeling, physical modeling, problem rocks.
1 INTRODUCTION
Rocks are media with different resistance under tension and

compression. There are many variants of the theory elasticity
for describing such medium, but these variants are such that it
is difficult to determine the criteria of plasticity, the transition
from elastic to plastic, from plastic state to failure. Besides the
existing versions of the theory of elastic and inelastic defor-
mations of rocks are such that they contain the contradictions
of a discontinuous nature, when there is no continuous tran-
sition from one loading area to another, there is no potential
(local and global).
2 STARTING POSITIONS
The underlying idea is the idea of the block character of a

structure of rocks. In this case, the simplest types of deforma-
tion characterizing the contact areas of blocks are the simple
extension and shear, as illustrated in Fig. 1.
3 DEFINING RELATIONS FOR ROCKS

(PLANE STRAIN)
Figure 1. Illustration of mechanisms of deformation of the initial
Contact pads are formed for rocks with a form that coordinates medium at flat deformation: a) two systems of planes of sliding;
with the content of the phenomenological mechanical model b) the scheme of deformation in the form of simple lengthening;
of the medium with the angle which is determined by using c) simple shift.
the experimental data. The obtained constitutive relations in
elasticity reflect the different resistance of rocks in tension and where
compression (different modules) and in the case of plasticity
and fracture they reflect an irreversible effect of dilatancy.
The equations in the plane case have such structure:
In the case of inelastic deformation and post-limit deforma-
tion, instead of (*) we have
where ϕ∗ the angle is defined in Fig. 2, which shows the

phenomenological structure of the rocks.
Putting in the elasticity where f – the experimentally determined function and the
modulus K is the same that we had in elasticity.
4 CONDITIONS ON BORDERS OF DIFFERENT

we obtain AREAS OF DEFORMATION
For the case of plane deformation necessary conditions on

borders between areas of elastic and non-elastic deformations
167

Figure 3. Dynamic loading plane of rectangular opening. The
black region shows part of the plate, located on the post-limit region
(failure).
6 CONCLUSIONS
1. New mathematical and mechanical models of deformation

and fracture of rocks, which describe the experimentally
observed effect of dilatancy of rocks, different resistance
in tension and compression are constructed.
2. Static and dynamic problems of loading of rocks with the
excavation of round and rectangular shapes are solved. The
Figure 2. Illustration of the deformation mechanism of rocks under regularities of the arrangement of dilatancy, the post-limit
plane strain: a) two systems of sliding planes; b) l – the direction of
simple shear, m – the direction of a simple extension for a system of
zone of deformation are defined.
sliding planes.
ACKNOWLEDGEMENT
are presented. It is shown that conditions of a continuity
of pressure and deformations on these borders should be The study is financially supported by the Russian Founda-
satisfied. tion for Basic Research, Grant No. 09-05-00327a, and the
Integration Projects Nos. 61, 69 and 74 of Siberian Branch,
Russian Academy of Sciences.
Static and dynamic problems of loading the plane with an

REFERENCES
aperture (circular, rectangular) are solved. In the static prob-
lem it is shown that near to the points removed from a contour
Barenblatt, G. I. 1961. The mathematical theory of the equilibrium
only deformation of simple extensions is occurred and closer cracks formed at fragile destruction, Journal ofApplied Mechanics
to a contour of excavation deformations of simple lengthening and Technical Physics, Vol. 4, pp. 3–56.
and simple shifts are occurred. Post – limit deformations occur Bulychev, N.S. 1989. The mechanics of underground structures in
from an excavation contour. An example of their distribution the examples and problems. M. “Nedra”.
in a dynamic problem is received at two-dimensional loading. Chanyshev, A.I. 1999. Block phenomenological mechanical model of
an element of the deformable mrdium. Part 3: Originally isotropic
mediums, J. Min. Sci., Vol. 4
5 EXAMPLES OF PROBLEM SOLVING Chanyshev, A.I. & Abdulin, I.M. 2006. Deformation and failure
of originally isotropic media under Mises condition of strength
In Fig. 3, the instant photo of the plate (50 × 50) with a square violation, J. Min. Sci., Vol. 42, pp. 322–334.
shape opening (8 × 8) under dynamic loading is shown. The Chanyshev, A.I. & Abdulin, I.M. 2008. Characteristics and the
black region shows a portion of the plate, located on the relations on them at the stage of post-limit deformation in rocks,
post-limit region (destruction). Calculations were performed J. Min. Sci., Vol. 44, pp. 451–463.
with the following parameters: E = 3 × 1010 Pa, ν = 0.3, Ruppeneyt, K.V. & Lieberman, J.M. 1960. Introduction to rock
mechanics, M. Gosgortekhizdat, p. 356.
τs = 40 × 106 Pa, ρ = 2600 kg/m. Loading time before failure
t ≈ 0.09 s. Fig. 3 shows that the failure begins at the internal
contour of the opening, which is free from stress.
168

The structural geology contribution to rock mechanics

modelling and rock engineering design
J.W. Cosgrove & J.A. Hudson

Department of Earth Sciences and Engineering, Imperial College of Science, Technology and Medicine,
South Kensington, London, UK
Subject: Structural geology and rock mechanics
Keywords: Structural geology, numerical modelling, fractures, scale, rock engineering design
1 INTRODUCTION
In 1997, we presented a paper to the 36th U.S. Rock Mechan-

ics Symposium NYRocks ’97 (Hudson & Cosgrove, 1997)
entitled “Integrated structural geology and engineering rock
mechanics approach to site characterisation”. Since that time,
the capabilities of computer modelling for rock engineering
design have increased by leaps and bounds, especially with
discrete element modelling and ‘intelligent’ methods such
as neural networks. Thus, it is appropriate now, as we cel-
ebrate the upcoming 50-year anniversary of the ISRM, to
reinforce the importance of the structural geology input to rock Figure 1. The regular succession of sandstones and shales that
mechanics, particularly in the context of computer modelling. characterise the folded turbidites of Northcote Mouth, Bude, S.W.
There are two main overall aspects that should be consid- England represent an important intrinsic mechanical anisotropy.
ered: what should be the content of a computer program so
that it provides an adequate simulation of rock mass behaviour;
and how can we obtain the necessary input parameters for any
particular modelling or design exercise? Within that context,
in this paper we discuss the nature of geological structures and
rock masses (Price & Cosgrove, 1990) with special emphasis
on fractures and scale in order to highlight some of the relevant
issues.
2 GEOLOGICAL STRUCTURES AND THE ROCK

MECHANICS MODELLING PROBLEM
In Figures 1 & 2, we begin by illustrating the anisotropy and

inhomogeneity that can occur in natural rock masses. The Figure 2. Folded slaty cleavage, an anisotropy induced in the rock
angular box folds that are visible in both examples reflect by metamorphism.
the high mechanical anisotropy of the two systems.
an explicit attempt in the modelling to represent the geometry
3 MODELLING and operating mechanisms on a 1:1 basis. For example, if there
is a major fracture in the vicinity of a tunnel, this can be incor-
The key to rock engineering design is the ability to predict porated explicitly in a numerical model. However, in the lower
the future, i.e. to be able to predict what will happen when an row, ‘Level 2, not-1:1 mapping’, the modelling techniques are
excavation with a given geometry is made at a certain depth not designed to include the geometry and mechanisms directly.
and at a certain orientation in a given rock mass. This means For example, in the neural network techniques, nodes in the
that some form of model is required. This modelling support network are used but, unlike the finite element approach, a
for rock engineering is illustrated in Figure 3 which sum- specific node cannot be directly associated with a location
marises the eight main methods supporting rock engineering or a physical variable. An example of the use of the Level 1,
design. Method C approach for modelling the geological evolution of
The two rows of boxes in the dashed main box represent fractures is illustrated in the companion paper to this Congress
the two main ways in which the mechanisms are mapped in by Latham et al. “Modelling the evolution of fractures using
the modelling. In the top row, ‘Level 1, 1:1 mapping’, there is a combined FEM-DEM numerical method”.
169

4 CONCLUSIONS
We have presented some aspects of the importance of struc-

tural geology understanding in supporting rock mechanics
modelling and rock engineering design. In particular, we have
emphasised
• the importance of geological information in establishing
rock mechanics properties, and
• the importance of scale, i.e. the size of the geological
structure relative to the size of the engineering project.
REFERENCES
Feng, X.T. & Hudson, J.A. 2011. Rock engineering design. London:
Figure 3. Flowchart of rock mechanics modelling and rock engi- CRC Press, Taylor & Francis.
neering design approaches (Feng & Hudson, 2011). Hudson, J.A. & Cosgrove, J.W. 1997. Integrated structural geology
and engineering rock mechanics approach to site characterisation.
Int. J. Rock Mech. Min. Sci., 34, 3–4, paper 136.
Hudson, J.A., Cosgrove, J.W. & Johansson, E. 2008. Estimating
the mechanical properties of the brittle deformation zones at
Olkiluoto. Posiva Work Report 2008-67. Downloadable from
www.posiva.fi
Jing, L. 2003. A review of techniques, advances and outstand-
ing issues in numerical modelling for rock mechanics and rock
engineering Int. J. Rock Mech. Min. Sci. 40, 3, pp. 283–353.
Latham, J.-P., Guo, L., Wang, X. & Xiang, J. 2011. Modelling the
evolution of fractures using a combined FEM-DEM numerical
model. Proc. ISRM Congress, Beijing. London:Taylor and Francis.
Price, N.J. & Cosgrove, J.W. 1990. Analysis of Geological Structures.
Cambridge University Press, UK, 502p.
Turner, S.A, Liu, J.G. & Cosgrove, J. W., 2011. Structural evolution
of the Piqiang Fault Zone, NW Tarim Basin, China, Journal of
Asian Earth Sciences 40, 1, pp. 394–402.
Figure 4. Collapse of sidewall of the transformer chamber as

affected by a fault (Jinping II hydropower project on the Yalong river,
China—from Feng and Hudson, 2011).
The geological support for the eight approaches supporting

rock engineering design in Figure 3 is most self-evident in
• Method A, Level 2,
• Method C, Level 1,
• Method D, Level 1
because the geometry and associated parameters are explicitly
required. However, if the rock excavation proceeds step-
by-step, e.g., the layer-by-layer downward excavation of a
large hydro-electric cavern, it is possible to use ‘intelligent’
modelling methods to predict the result of the next step in
excavation and hence to adjust the process as necessary,
Figure 4.
170

Rock mass instability caused by incipient block rotation
A.V. Dyskin
School of Civil and Resource Engineering, University of Western Australia, Australia
E. Pasternak
School of Mechanical and Chemical Engineering, University of Western Australia, Australia
Keywords: General, Rock failure, Stability analysis
The role of rotational degrees of freedom in the mechanisms

of instability has recently been recognised, however they are
usually thought of as being associated with rotation of spher-
ical particles and modelled as such. Real blocks (and even
rock grains for that matter) are far from being spherical and
this may have considerable effect on the rock mass stability
under compression.
Consider a blocky rock mass whose bocks are separated
by gouge and suppose the shear stress can force some blocks
to start rotational movement. Figure 1 presents schematics of Figure 1. A single block in incipient stage of rotation.
this situation and moment equilibrium of the block under com-
bined action of shear, T , and compressive, P, forces. At the
beginning of rotation the moment equilibrium about O yields
a dT = −P du, where u is the relative displacement of the
upper contact associated with the grain rotation. The dilation
related to the rotation is infinitesimal of the second order. Thus
the first moment of block rotation, due to its non-sphericity,
forms a virtual negative stiffness element (dT /du < 0). We can
treat it as a negative stiffness element embedded in an other-
wise positive stiffness matrix representing the rest of the rock
mass.
By referring forces to unit area and horizontal displacement
to the block size, a, one obtains the value of apparent tangent
shear modulus which characterises the deformation of the cor-
responding part of blocky rock mass at the beginning of block
rotation:
Figure 2. Dependence of the critical volumetric fraction of rotating

blocks upon the value of negative shear modulus.
√
where for square block sections the shape factor δ = 2.
We model the blocky rock mass with rotating blocks as an Here c is the volumetric fraction of inclusions and
isotropic elastic medium with spherical inclusions. The inclu-
sions have negative shear modulus, while the bulk modulus
is the same as the bulk modulus of the medium. We assume
that the main effect of block shape has been already taken into
account by the negative shear modulus; the particles are tak-
ing as spherical to simplify the analysis. We use McLaughlin’s
(1977) solution, which for the case of matrix and inclusions where µeff , is the effective shear modulus µm , is the shear
having the same bulk modulus is reduced to modulus of the rock mass just before block rotation is started,
where p is the magnitude of compressive stress acting in the
direction shown in Figure 1.
This equation has a critical volumetric fraction ccr , Figure 2.
At c = ccr , µ abruptly becomes negative and the rock mass
loses stability. There is a special value of m = 1 when the
171

critical volumetric fraction ccr = 0 which means that rotation before block rotations. This parameter is important in both
of any block causes instability and rock mass collapse. For ensuring excavation stability and block caving.
other values of m a number of rotating blocks is required for
the rock mass to lose stability.
Thus, on top of the stress concentration produced by the REFERENCES
excavation and rock mass strength usually considered in exca-
vation design we introduce a new controlling parameter – the McLaughlin, R. 1977. A Study of the Differential Scheme in
ratio of compression magnitude to the overall shear modulus Composite Materials, Intern. J. Engin. Science, 15, 237–244.
172

Attempt to advanced observational construction considering predictive

analysis of long-term deformation
T. Fukuda
Geoscience Research Laboratory, Co., Ltd., Tokyo, Japan
T. Takahashi
East Nippon Expressway Company, Hokkaido, Japan
H. Yamada
Konoike Construction Co., Ltd., Osaka, Japan
Y. Jiang
Nagasaki University, Nagasaki, Japan
Keywords: back analysis, tunnelling, rock properties, rock mass, rock support
1 INTRODUCTION
We report a case of analysis for the advanced observational

construction, focusing on the following two points.
• Using the back analysis method in which Kalman filter
is applied to finite element method, physical properties of
rock mass are identified from the measured information for
designing optimal support.
• If deformation lasts for a long time, the stability of the
lining has to be concerned. Predictive analysis for long-term
deformation is conducted, utilizing the rock mass physical Picture 1. Support pattern.
properties obtained from back analysis.
Table 1 shows the amount of subsidence at the measuring
points during excavation of the upper and bottom halves.
The values in parentheses are the ones measured in the field.
As a result, the predictive analysis was conducted, based on
the physical properties of rock mass identified by the back
analysis. It led to the selection of support pattern determined
to be optimal (Picture 1).
Table 1. Displacement of measurement points and identification

results.
Figure 1. Displacement of the lining.
Figure 1 indicates time-dependent change of the lining. The

predictive analysis shows the results up to about one year later.
There is approximately 8 mm of subsidence occurred at the
tunnel crown, and almost the same amount of displacement
was caused in the inner space.
As a result, the back analysis to estimate the physical prop-
erties of rock surrounding the tunnel led to the selection of the
optimal support patterns. It could be indicated how much the
lining will be deformed in the future by showing the results
of the predictive analysis of long-term deformation.
173

Vector sum method: A new method for anti-sliding stability analysis
X.R. Ge
Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan, China
State Key Laboratory of Geomechanics and Geotechnical Engineering, Wuhan, China
Institute of Geotechnical Engineering, Shanghai Jiao Tong University, Shanghai, China
Keywords: stability analysis; rock slopes and foundations; numerical modeling
1 INTRODUCTION certain direction. This projection direction should have clear

physical meaning. In VSM the projection direction is defined
The anti-sliding analysis is significantly important for the sta- as the potential direction of the potential sliding body. The
bility analysis of the dam foundation, slope and underground safety factor is defined as the ratio of the projections in VSM.
cave. By far the limit equilibrium method (LEM) has been the For the stability analysis by VSM is based on the real stress
most extensively used method in this field. The finite element state and physical parameters, it has more advantages than
method (FEM) based on strength reduction principle is also a those based on the virtual stress state. Moreover, the VSM
popular method used to analyze the stability problem of slope analysis is based on the deformed body theory rather than
and dam (Ducan 1996; Feng et al. 1990; Zheng et al. 2002). the rigid body assumption. Therefore, VSM is more physical
In the LEM and FEM-based methods, the stability is usu- sound.
ally evaluated by the so-called safety factor calculated based
on the strength reduction principle. Although the philosophy
of safety factor and strength reduction has been extensively 4 ANTI-SLIDING STABILITY ANALYSIS FOR
accepted in engineering, two basic problems are worthy of 2D PROBLEM USING VSM
reconsidering. One is the rationality of the strength reduction
principle and another is the superposition principle of force in For 2D problems, the calculating region, the potential slid-
the solution of anti-sliding safety factor. The strength reducing interface, and the sliding region are assumed to be prior
tion method involves more factitious treatment. To explore known. The current stress state has been obtained by FEM
more physically sound method for sliding stability analysis, or other methods. The search for critical slip surface is not
we proposed the vector sum method (VSM) (Ge 1987; Ge included in this paper. Take Mohr-Coulomb criterion as the
et al. 1995). Some advances on this method have been made strength criterion. When the current stress states are known,
recently (Liu 2007; Ge 2008). the anti-sliding shear stress on the slip curve is calculated as
2 DISCUSSION ON THE RATIONALITY OF

STRENGTH REDUCTION PRINCIPLE Take the potential sliding direction as the projection direc-
tion. The angle between potential sliding direction and the x
Firstly, it is not quite reasonable for the strength parameters c coordinate axis is θ, shown in Figure 1.
and tan ϕ divided by the same reduction coefficient. Secondly, According to definition of safety factor of VSM, the 2D
the parameter ϕ can’t be arbitrary reduced in Mohr-Coulomb safety factor KVS is calculated as
criterion according to (Zheng et al. 2002). Thirdly, when the
strength parameters are reduced, the calculated stress state is
not its real state, but a virtual state. The derivation from the
virtual state is not physically sound.
3 THE BASIC CONCEPT OF THE VECTOR SUM

METHOD
5 ANTI-SLIDING STABILITY ANALYSIS FOR
Since force is vector, its sum should be vector sum. The resul- 3D PROBLEM USING VSM
tant sliding force is the sum of the sliding forces of each
segment li of the potential sliding surface. Similarly the The safety factor in VSM is defined as the projection ratio
resultant anti-sliding force is the vector sum of the anti-sliding of the resultant anti-sliding and sliding force vector on the
forces of each segment li of the potential sliding surface. potential sliding direction. The resultant sliding force is a vec-
When computing the safety factor, the resultant sliding force tor sum of normal forces and shear forces acting on the Si of
vector and the anti-sliding force vector must be projected to a the potential slip surface. Similarly, the resultant anti-sliding
174

Figure 1. The coordinate systems and the stresses state acting on Figure 12. Main faults and potential slip paths in the profile of the
the li . dam foundation.
The normal stress acting on the sliding mass at the point A

by bed rock is
To facilitate the derivation, assume that the tensile stress

is positive and the compressive stress is negative. The safety
factor is expressed:
Figure 5. The stress state at point A on potential slip surface. where R is the projection of resultant anti-sliding force on
potential sliding direction d̂; T is the projection of resultant
force is a vector sum of anti-sliding normal force and anti- sliding force on potential sliding direction d.
sliding shear force acting on the Si of the potential sliding The VSM is used to analyze the anti-sliding stability of a
surface by bed rock. dam foundation, which is one of the typical dam foundations of
Shown as Figure 2, let σ s , σ τ , σ n stand for stress vector, a hydraulic power station in China. The main natural faults and
shear stress and normal stress, respectively at the point A on the potential sliding interfaces of the representative section are
the sliding interface. n is the unit normal vector of tangent shown in Figure 3. These sliding paths are ABCD , ABCD,
plane at point A (positive pointing to outside of the sliding EFD and EFGI, and the corresponding results by 2D vector
mass). d is the unit vector of the potential sliding direction. sum method are listed in figure 3. For this example 3D VSM
S is the slip surface. Then, result is also given.
6 CONCLUSIONS
The safety factor of VSM is derived from the real stress state
rather than the virtual stress state. The safety factor is defined
as the ratio of the projection of resultant anti-sliding and slid-
where σ is the stress tensor at the point A on the potential slip ing force vector on the potential sliding direction, which has
surface. clear physical meaning.
175

Design of a railway tunnel parallel to an existing tunnel
R.K. Goel & A. Swarup

CIMFR Regional Centre, Roorkee, India
R.K. Dayal & V. Bhardwaj

Rites Ltd., Gurgaon, India
Subject: Site investigation and field observation
Keywords: Site characterization; Rock mass classification; Numerical modeling; Tunnelling; Rock support
Mumbai Rail Vikas Corporation Limited has planned to con- basaltic lava flows. Basalts are predominantly the amyg-
struct 5th and 6th railway lines from Thane to Diva. On the daloidal basalts. In the area near Tunnel 1, the rocks are
route, the rail line passes through various geological features grayish to brownish fine grained amygdaloidal basalt rock.
like creeks, lineament, hills and flood creek plains. Because The empirical and the numerical approach using FLAC3D were
of the varied topography, various structures like bridges, foot used for the analysis and design of the tunnel.
over bridges and tunnels are to be constructed. As per the In the empirical approach the Q-system has been used to
Geotechnical investigation report, two tunnels are planned on classify the rock mass for the design purposes. The Q value of
the route. amygdaloidal basalt is about 25 indicating a ‘Good Rock’ for
The proposed new tunnel having length 165 m, width 12 m tunnelling. Different Q values were obtained for portal areas
and height 8.5 m is to be driven parallel to the existing tunnel. and intersection of trolley refuse. With factor of safety of 1.5,
The tunnel is to accommodate double track. There is also a the Q-value is back calculated and accordingly the supports
provision of trolley refuse in the tunnel. The existing Tunnel have been designed. Using the Q values, the shotcrete and
1 is around 110 m long having a gentle curve. The tunnel is rock bolts support is suggested in the tunnel. The slope above
aligned to N120◦ at Thane end and N290◦ at Diva end. The portal is suggested to be protected by erecting the fence or
tunnel is around 100 years old. The roof of the tunnel upto protection wall.
springing level throughout its length is brick masonry lined. At The numerical analysis using the FLAC3D code was also
the portal upto about 20 m inside the tunnel the walls are also carried out to study the effect of adjacent tunnel on the induced
brick lined. At places inside the tunnel also the walls are brick stresses. Accordingly it was found out that a minimum 16 m
lined. On the uphill side walls in the tunnel, small seepage pillar width between the two tunnel (existing and the proposed)
of water is also observed. The existing tunnel is horse-shoe is required. The results of analysis also show that the induced
shaped. stresses are within reasonable limits.
Geologically the region falls in Deccan volcanic province Details about the tunnel design and numerical analysis are
that is made up of a number of practically horizontal sub-aerial given in the full length paper.
176

Shear failure mechanism in oil wells due to reservoir compaction

“A case study in Gachsaran formation, Iran”
M. Gorjian, M. Moosavi, H. Memarian & H. Ajamzadeh

School of mining, University of Tehran, Iran
Subject: Rock engineering in hazardous geo-engineering
Keywords: reservoir compaction, rock stress, Gachsaran formation, casing collapse
A large number of casing collapses are reported in one of the For this reason, the shallow cores had been heated by oven
largest carbonate oil fields in south of Iran. Most of the casing and kept in constant temperature for at least 24 hrs. Then the
collapses have been occurred in Gachsaran overburden forma- laboratory tests were carried out promptly. From these data,
tion. Compaction of reservoir is one of the prime suspects of relations between E and UCS with temperature were generated
this phenomenon. It is very difficult to take core samples bel- both for marl and anhydrite which are summarized in Table 1.
low 200 m depth in Gachsaran formation due to high solubility Additional triaxial compressive tests were carried out on
and weakness of the rocks. Moreover there are not adequate the anhydrite and marl cores at various temperatures to obtain
logs in this formation. So obtaining the geomechanical param- Coulomb failure criterion. The coulomb failure criteria in
eters of rocks in depth associated with casing collapses (about anhydrite and marl at various temperatures were determined
3000 m) is virtually impossible. As a result, the main reason as equation 1 and 2 respectively:
of casing collapse in Gachsaran formation has been not clear
yet; however, reservoir compaction has been named as eminent
cause by some researchers.
To quantify the compaction, the lithology across the fail-
ure depth (3200 m for well M-42, as the first failed well) was
reviewed and an interface between anhydrite (as a stiff rock)
and marl (as a soft rock) was detected. One dimensional reser-
voir compaction was calculated to be as high as 0.15–0.3 m in
50 years due to 9 MPa reservoir depletion.
Anhydrite and marl cores were taken from depths of 12 m
to 195 m of Gachsaran formation at the Gotwand-E-Olya dam
Moreover, the relation between failure angle (β) and tempera-
site (Fig. 1). However the thermal condition at shallow level
ture (T◦ C) in Gachsaran anhydrite and marl were obtained as
is quite different from high depths (collapse depth), which
expression 3 and 4 respectively: The triaxial compressive test
requires testing at elevated temperatures on the samples.
results were consistent with those from uniaxial compressive
test in anhydrite and marl.
To analyze whether the anhydrite/marl interface in well 42

was prone to failure due to depletion, a model was prepared
in ABAQUS finite element package. It was assumed that the
specimen is under triaxial loading condition (Fig. 2). In the
model, the loading plates were also modeled. The specimen
is composed of anhydrite (upper part) and marl (lower part)
Table 1. The relations of UCS (MPa) and E (GPa) with temperature
(T◦ C) in anhydrite and marl.
Rock type UCS, T E,T
Anhydrite UCS = 15.93e(0.008T) E = 0.476e(0.034T)

(R2 = 0.97) (R2 = 0.95)
Marl UCS = 0.653e(0.025T) E = 0.016e(0.051T)
Figure 1. M oil field geographical location (MahabGhods Consult- (R2 = 0.98) (R2 = 0.93)
ing Engineering, 2007).
177

Figure 2. Schematic overview of FE model.
Table 2. Geomechanical parameters of anhydrite and marl used in

numerical modeling. Figure 3. Shear stresses produce across the anhydrite/marl interface
due to the reservoir depletion within 50 yrs.
Lithology Anhydrite Marl
E (GPa) 11.9 1.9

ν 0.12 0.16
C (MPa) 5.3 1.9
ϕ(◦ ) 55 30
σc (MPa) 34 6.8
which are separated by a 10◦ interface. This is the same con-

dition encountered in well 42. The confining stress (σ3 ) was
80 MPa. The axial stress (σ1 ) was 89 MPa which is equal to the
initial stress plus 9MPa depletion pressure. Steel plates in the
model were assumed rigid. Contacts between steel plates and
rock specimen were generated by joint elements characterized
by friction values (fa ) of 0.1. Rock behavior was assumed to be
homogenous and isotropic, with behavior according to Mohr-
Coulomb yielding criterion. The Mohr-Coulomb criterion was Figure 4. Failure criterion for marl and the stresses across the
assumed to characterize the shear behavior of interface too. It anhydrite/marl interface.
has been proved that shear behavior of an interface between
two rocks is dictated by the shear behavior of weaker layer, situ rocks, especially marl, will have lower strength properties
therefore the shear behavior of anhydrite/marl interface was than the tested samples in the laboratory. The numerical anal-
assumed the same as marl. ysis based on rock properties obtained from dry samples
By using temperature dependency relations (Refer to tab. 1, showed that the compaction alone cannot cause such failures.
eq. 1 & eq. 2), the elastic and Mohr Coulomb parameters However, if the real samples with higher water contents can
of anhydrite and marl were obtained (Tab. 2). Note that the be obtained may return mechanical values which may explain
temperature of collapse depth was obtained through mud logs such failures.
analysis. Figure 3 shows that the maximum shear stress (τ), It is also important to know that interface of thin layers of
which builds up across the interface, is around 39 MPa. salt in between, may also enhance the slip. Creep behavior and
If the stresses across the interface exceed the strength of long term displacement of such interfaces are also possible
marl (weaker layer), we can assume that anhydrite/marl inter- parameters that will contribute to this process. Therefore at
face is prone to failure due to reservoir compaction. As can be this stage, one can say that the biggest role is played by the in
seen from Figure 4, the shear stress at the interface is less than situ stresses together with extra pressure due to depletion. The
the shear strength. However, it shall be noted that the labora- next biggest effect might be the creep behavior of materials
tory tests were done on rock samples which had lost most of such as salt and also tectonic forces due to local faults. These
their water content. In another words, it is expected that the in issues need more studies to shed more light on this subject.
178

Fractal analysis on the fracture development of sandstone using AE measurement
Z.Y. Yang
Department of Civil Engineering, Tamkang University, Taipei
T.H. Huang
Department of Civil Engineering, National Taiwan University, Taipei
Y.X. Zhao
Department of Civil Engineering, China University of Mining & Technology, Beijing, China
M.C. Tsai
Keywords: rock failure; rock property; lab testing
1 INTRODUCTION
The AE technique has been applied to obtain the crack initi-

ation and propagation (Chang & Lee 2004). This is the basis
for seismic monitoring of active faults to predict earthquake
and highly stress zones in tunnel to predict rock burst. A dra-
matic decrease will appear before reaching the maximum AE
events. Therefore, the rate of AE events is commonly used as
a qualitative precursor of rock burst or main earthquake.
A number of studies show that the statistical properties of
AE events and seismicity are self-similar (Turcotte 1997). The
fractal parameter thus has been employed to well describe Figure 1. Arrangement of the six AE sensors and four strain gauges
the AE or seismic events. The aim of this paper is to describe in a sandstone sample.
the fracturing behavior of sandstones under axial compres-
sion from AE technology. The correlation dimension is used event is obviously measured at 0.5εf (shear dilation stage)
to describe the development of microcracks during uniaxial and violent increased after 0.75εf (unstable cracking stage).
compressive loading. The event of rock fractures observed from the AE hit event is
concentrated in the range of 0.5∼0.75εf . The event of fracture
2 EXPERIMENTAL PROGRAM somewhat decreases before peak strength failure.
The rectangular sample of sandstones from Taipei of

Taiwan was used for AE test and the sample size was 3.2 Development of fracture distribution and propagation
50 × 50 × 100 mm. The Poisson ratio of sandstone is 0.16 Figure 3 shows that at low stresses the fracturing is initialized
and Young’s modulus is 13 GPa. The average P-wave veloc- from somewhere in the rock and is close to the actual fracture.
ity is 2610 m/sec and S-wave velocity is 1580 m/sec. A The AE events localize in a shear zone before 0.5εf and form
servo-controlled hydraulic testing machine was used for the main fracture. At the stage of 0.5∼0.75εf , the numerous
this compression tests. For each AE test, the location of fractures are to extend the major fracture surface and cluster
AE hypocenter for a micro-crack was determined by the to the main fracture.
arrival time difference method, using six data measured by
piezoelectric sensor arranged shown in Figure 1.
4 FRACTALS OF FRACTURE DEVELOPMENT
3 FRACTURE BEHAVIOR OF SANDSTONE
The distribution of an object set (see Fig. 4) with small
3.1 Development of fractures by measuring AE event correlation dimension is more clustered than a higher one
(Coughlin and Kranz, 1991). Figure 5 shows the change of
A typical AE event accompanied with the mentioned axial correlation dimensions (Dcorr ) accompanied with the axial
compression stress-strain curve is shown in Figure 2. The AE stress-strain curve. This shows that the values of correlation
179

Figure 5. The tendency of correlation dimensions with the axial
applied stresses.
dimensions (Dcorr ) of sandstones during compression are less

than 1. It means the distribution of rock fracture in compres-
sion failure is clustered in a small region. The value of Dcorr
increases with the increasing stresses before 0.5εf . The value
of Dcorr is almost kept constant after 0.5εf .
However, an increasing tendency in Dcorr value is observed
near the peak strength. This special phenomenon is similar to
the decreasing AE events near peak. The parameters of AE
event and correlation dimensions (Dcorr ) both will be helpful
to forecast a major fracture such as the earthquake by fault
fracturing.
Figure 2. The variation of AE events of sandstone during
compression.
5 CONCLUSIONS
Most of the fractures before peak strength are localized to

a main fracture zone. This clustering is also explained by
the small values in correlation dimensions. The AE event
of fractures in sandstone is decreasing before peak strength.
However, the correlation dimension is rapidly increasing near
the peak strength. Both of the AE rate and correlation dimen-
sion therefore can be applied to forecast a main rupture,
however, it needs more studies.
REFERENCES
Chang, S.H. & Lee, C.I. 2004. Estimation of cracking and dam-
Figure 3. Development of AE hypocenters before peak strength age mechanisms in rock under triaxial compression by moment
during uniaxial compression. tensor analysis acoustic emission, International Journal of Rock
Mechanics and Mining Science 41: 1069–1086.
Coughlin, J. & Kranz, R. 1991. New approaches to studying
rock burst-associated seismicity in mines. In Proceedings of the
32nd U.S. Symposium on Rock Mechanics as a Multidisciplinary
Science, Oklahoma, July 10–12, 1991. Rotterdam: Balkeman.
Turcotte, D.L. 1997. Fractals and chaos in geology and geophysics,
Cambridge University Press: 106–108.
Figure 4. Two examples of grain distribution with different Dcorr

display the different clustering degree.
180

Static and dynamic response analyses of rock mass considering

joint distribution and its applicability
N. Iwata
Chuden Engineering Consultants Co. Ltd., Hiroshima, Japan
T. Sasaki
Rock Engineering Institute, Suncoh Consultants Co. Ltd., Tokyo, Japan
R. Yoshinaka
Saitama University, Saitama, Japan
Keywords: numerical modelling, rock joints, case studies, rock slopes and foundations, dynamics
1 INTRODUCTION To simulate the recorded strong motions in foundation rock

of nuclear power reactor building, dynamic response analy-
The rock mass is essentially composed of intact rock and dis- sis is performed. Fig. 2 shows the comparison of response
continuities such as joint set, joint system et al.As well-known, spectrum of horizontal acceleration at 60 m in depth. The
the mechanical properties of rock mass are greatly depend- prominent spectral peaks of MYM are roughly corresponds to
ing on the influence of discontinuities. The authors proposed that of the recorded motion, but in the result of elastic analysis
Multiple Yield Model (MYM) for analyzing discontinuous the peak at around 0.4 seconds wasn’t simulated.
rock are analyzes two practical examples of large scale verti-
cal excavations 20–30 m in depth and two earthquake response 5 CONCLUSIONS
analyses of the discontinuous rock slope and rock foundation
of large building, and is presented the applicability of MYM. As the result of these case studies, the actual deforma-
tion behavior of discontinuous rock is highly influenced by
the geometrical distribution and doesn’t correspond without
2 OUTLINE OF THE MULTIPLE YIELD MODEL considering discontinuities. The authors are confirmed that
Multiple Yield Model is an effective method for static and
MYM is a kind of finite element method constituted the dynamic response on jointed rock masses.
mechanical properties of intact rock and discontinuity sys-
tems in rock mass, and assumed that the total strain of rock
mass is expressed summation both of joint sets and intact rock
strains. MYM can be analyzed the non-linearity of deforma-
tion under loading and unloading stress paths considering the
confining pressures of rock joints. Fig. 1 shows the stress-
deformation characteristics of joint sets of the cyclic motions
on the normal and the shear directions.
3 EXAMPLE OF LARGE SCALE EXCAVATION
For analyzing, the geometrical models of rock mass were

determined from site investigations and the physical parame- Figure 1. Stress-deformation characteristics of joint; (a) Normal
ters were determined by laboratory test and also considering direction, (b) Shear direction.
scale effect. The result of horizontal and vertical deformation
mode are well corresponding to measurement and the settle-
ment of ground surface behind vertical wall coincide with
horizontal deformation.
4 EXAMPLES OF EARTHQUAKE RESPONSE
The authors are also applied MYM to large scale slope col-
lapse due to earthquake and compared the collapse mode.
The analytical results indicate that the collapse was occurred
beginning at the early time, and shear failures extended to
the middle of slope along bedding joints and tension failures
were occurred along orthogonal joints, it caused toppling at Figure 2. Comparison of response spectrum of horizontal acceler-
the surface. These results agree with eyewitness reports. ation at 60 m in depth.
181

Numerical study on zonal disintegration of rock mass

around deep underground openings
P. Jia
School of Resources and Civil Engineering, Northeastern University, Shenyang, China
C.A. Tang & Y.B. Zhang

Center for Rock Instability and Seismicity Research, Dalian University of Technology, Dalian, China
Subject: Modeling and numerical method
Keywords: rock mass; rock failure; numerical modeling
With the increasing of the axial load, cracks are first initiated
around the periphery of the opening and propagated gradually
until the first order of fracture ring is formed (Fig. 5b). Then
the second and third order are generated sequentially. These
fracture rings are not closed rings, but discontinuous fracture
rings, i.e., in the fracture zones, there includes intact zones;
while in intact zones, there includes fracture zones too.
The fracture rings are not continuous both in radial and
axial direction, which is different from our former cognition
that fracture rings are continuous rings. This discontinuity is Figure 6. Configuration of Zonal disintegration in different cross
sections.
mainly due to the heterogeneity of rock mass.
Numerical results indicate that the elastic modulus as well
as compressive strength have a very important impact on ZD
phenomenon. In layered materials, the fracture ring is mainly
initiated and propagated from weak layers. There are only few
integrated fracture rings in the model and the number of the
fracture rings is very limited compared with that in the intact
model. As for the rock mass with lower strength and higher
deformation ability, the radial strain is relatively high and the
radial tensile stress thus decreased. Due to the intensive het-
erogeneity of layered materials, the distance between fracture
rings is also longer than that in intact model. Figure 7. Longitudinal section of numerical specimen after loaded.
Figure 5. Zonal disintegration process in intact rock mass.
182

Figure 9. Zonal disintegration in layered specimen (cross section). Figure 13. ZD phenomenon around parallel openings in multi-
layered materials.
For ZD around parallel close-spaced openings, for the

fracture rings near to the opening, they enclose each open-
ing separately; however, as for those fracture rings far from
the opening boundary, they enclose the two opening like an
envelop line. In multi-layered materials, the space between the
second and the third order of fracture ring is much larger and
the number of the fracture rings is limited due to the strong
heterogeneity of the material.
183

The secondary lining design of subsea tunnel due to water inflow
Kwang-Su Joo, Young-Ho Suh & Kyung-Ho Park

Engineering & Design Department, Civil & Environment Division, Hyundai Engineering & Construction
Do-Hak Huh, Eui-Joon Hong & Sang-Jo Moon

Department of Tunnel, Yooshin Engineering Corporation
Seok-Cheon Lee
Department of Tunnel, Taeso Engineering Co.
Keywords: numerical modeling, rock stress, rock support, stability analysis, subsea tunnel, secondary lining
1 INTRODUCTION is very complicate and very poor because it is consist of the

sandstone, mudstone, shale and some coal.
The Boryung tunnel will be the first mined road subsea tunnel
in Korea. This tunnel with a length of 6.927 km, two-lane two
tubes, has been planned to pass approximately under max- 3 WATER INGRESS AND DRAIN CONDITION
imum 60 m from seabed and 80 m from sea level between OF THE SUBSEA TUNNEL
Boryung and Taean. It is excavated by blasting, supported by
shotcrete, rock-bolt, lattice girder and is drained under control The drain material of the Boryung tunnel is better discharge
of water ingress by grouting for watertight. The subsea tun- capacity than fleece because that is ensured the discharge
nel is characterized with the heavy water ingress and the high space with a convexo-concave shape. The maximum water
water pressure because of the infinite water ingress source. inflow rate of Boryung tunnel is presumed with a 0.5 m3 /min-
This project status is on the end of the detail design stage and km because it is located at the rock mass and pregrouted for
construction will be started in the end of 2010. the purpose of making the cutoff region (Seo et al., 2010).
This paper intends to introduce some design considera- Basically, the drained type tunnel is not activated with the
tions about the secondary lining design considering the water residual water pressure. But, the residual water pressure will
ingress in the mined subsea tunnel. Firstly it is introduced be activated to the secondary lining in case of the clogging of
the design concept of the combination of the rock loads and drained material and water ingress exceeding the drain capac-
residual water pressure is studied and proposed. ity. Fig. 2 is shown as a change of residual water pressure with
a permeability change of drain material in case of support
pattern SP-4 and SP-5. H is the maximum water pressure at
2 BORYUNG SUBSEA TUNNEL this region and Pw is the maximum residual water pressure
It is the geological condition of Boryung tunnel that there are

the small mountains at the end bounds and almost flat seabed
(Fig. 1). The smallest rock cover of this tunnel is located at a
third from the both end bound because the end bound is near
the seaside.
There is the gneiss penetrated by granite and metamorphic
sandstone in the front part of the tunnel and granite in the rest
part of the tunnel. The geological condition of the gneiss and
granite is very good. But, the metamorphic sedimentary rock
Figure 2. The residual water pressure at the roof with a deterioration

Figure 1. The geological condition of Boryung Tunnel. in the drain capacity.
184

activated at the roof of secondary lining. The coefficient of Chang et al., 2010) model that consider the load transfer of
permeability of rock mass is 1.2∼4.0 × 10−5 cm/sec and the primary and secondary supports and interaction of ground and
rock cover and maximum high water level from the roof of tun- supports.
nel of SP-4/SP-5 are 49.7 m/24.3 m and 84.5 m/38.3 m. If the If Concrete lining is designed by using GLI model, it
drain material is good condition, the residual water pressure assumed that the excavation and the support installation are
is not activated at the secondary lining. However, the residual performed at the same time. For the more conservative design,
water pressure is increased with deterioration of permeability. the application of the load distribution ratio for the analysis of
If the decreasing ratio variation of the permeability coefficient the support is excluded. Nevertheless, the amounts of the sup-
is 0.1 to 0.01, the residual water pressure is increased largely. port is less estimated than the case of applying the Terzaghi’s
But, the residual water pressure increasing is very slightly rock load.
with after that ratio. Because the long time monitoring data
of clogging is nonexistent, it is properly determined that the
deterioration of drain material capacity is selected with 1/100. 5 CONCLUSION
The Boryung tunnel with a length of 6.927 km is the first

4 COMBINATION OF THE ROCK LOAD AND mined road subsea tunnel in Korea. This tunnel will be a good
RESIDUAL WATER PRESSURE FOR THE data of Korea and international construction. In this study,
SECONDARY LINING DESIGN it is introduced that the method considering the residual water
pressure for secondary lining design. This is the one of the
There are various kinds of loads that affect the secondary lin- many studies for design of the Boryung tunnel. Consider-
ing. The important loads among them are the rock loads and ing the lack of experiences of subsea tunnel construction, the
the ground water pressure. The ground water inflow rate of residual water pressure is applied conservatively to secondary
rock tunnel is various because it depends on the joint condi- lining design. But, it is expected that this method will be
tion of the rock. So, it is considered that 2 cases of combination improved based on the construction experience of the Boryung
of the rock loads and the residual water pressure. tunnel.
Case 1: Considering the water inflow. Applying the submerged
unit weight (γsub ) of the rock.
Case 2: Not considering the water inflow. Applying the wet REFERENCES
unit weight (γt ) of the rock. Considering the weight of the
sea water as a dead load. Chang S.B., Lee S.Y., Seo Y.H., Yun K.H., Park Y.J. and Kim S.M.,
2010, “A Study on Field Change Case of Tunnel Concrete Lining
At the same ground condition, the stress resultants of the Designs Using GLI(Ground Lining Interaction) Model”, Tunnel
lining of “Case 1” are higher than “Case 2”. So, “Case 1” is and Underground Space, Vol. 20, No. 1, pp. 58–64
adopted as the load combination for design of the secondary Seo Y.H., Nam H.Y., Chang S.B., Huh D.H. and Lee S.C., 2010,
lining. The stress resultants of secondary lining by the rock “The Treatment Plan of Inflow Water into The First Mined Road
loads are induced by using GLI (Ground-Lining Interaction, Subsea Tunnel in Korea”, Conf. of KTA Spring, pp. 1–3
185

An assessment of rock pillar considering excavation damaged rock zone
Won-Beom Kim
Department of Energy and Resources Engineering, Chonnam National University, Gwangju, Korea
Taewook Ha
Hakyong Engineering Corporation, Seoul, Korea
Hyung-sik Yang
Department of Energy and Resources Engineering, Chonnam National University, Gwangju, Korea
Keywords: case studies, tunnelling, stability analysis
As the tunnel construction in the urban areas gets vital- In this study, considering the excavation damaged rock
ized, the planning and execution of construction reflecting zone, the behaviors of the rock pillars have been assessed
a variety of factors such as linear conditions, restrictions on by the 3-dimensional numerical analysis. In order to assess
land procurement, the protection of some obstacles, etc. are the damaging area of rock pillars based on excavation meth-
required. For example, the distance between two horizontal ods, the study adopted a seismic exploration method to the
tunnels (pillar width) increases the construction example of a blasting section as well as the mechanical excavation section
very near parallel tunnel. The excavation damaged rock zone with the breaker. After the size and degree of the damage
which is inevitably occurred when excavating the underground were re-evaluated as the decrease in the strength parame-
structures is the major cause of the stability loss during the ters of rock pillars was applied to the results of numerical
construction of the rock structures. In case that the parallel analysis, it appeared that the decrease did not influence the
tunnels of which distance between the two horizontal tunnels safety factor of mechanical excavation while it was certainly
is very near, this affects the stability of the tunnels depend- causing a decline in the safety factor a general blast. Added
ing on the size of the excavation damaged rock zone occurred to this, the damaging area of the rock resulting from an
to the rock pillars if the rock pillars themselves between the excavation method proved to be influential in improving the
tunnels act as the supporting points. range of the minimized distance between the two horizontal
tunnels.
In particular, the result showed that the safety factor reduced
to maximum 28% in case that the excavation damaged rock
zone depending on the excavation methods was considered
than the result of numerical analysis without considering the
excavation damaged rock zone.
In this study, as the size of the excavation damaged rock
zone occurred by blasting appeared to be bigger than that by
the mechanical excavation, the minimized distance between
the two horizontal tunnels to which the mechanical excavation
and blasting could be applied in constructing the very near
parallel tunnels was assessed by the numerical analysis.
Figure 1. Schematic diagram of SHT wall transmission method.
Table 1. Assessment of safety factor.
Safety factor
Pillar width Ground depth EDZ
Excavation method RMR (m) (m) (m) Without EDZ Considered EDZ
Mechanical Excavation IV 1.8 14.3 0.4 1.28 1.25

Blasting III 6.3 36.9 3 1.75 1.2
186

Response of granite under strain controlled loading and effect of support system on
behaviour of large underground cavern
Rakesh Kumar
GMR Consulting Services Pvt. Ltd., Delhi, India
K.G. Sharma & A. Varadarajan

Department of Civil Engineering, Indian Institute of Technology Delhi, India
Keywords: lab testing, rock properties, numerical modeling, stability analysis
The hard rocks like granite, basalt, quartzite etc. show the The rate of softening is less under low confining pressures and
sudden failure due to loading and unloading. The phenom- high under high confining pressures.
ena of rock bursting during underground excavation for such The paper also presents the application of softening behav-
hard rocks under high rock cover is quite common. The rocks ior of granite to the analysis of large underground caverns
in general show strain softening behavior under loading and (Power house and transformer yard) under plane strain con-
unloading. The strain softening behavior of rocks is captured ditions using FLAC. The power house cavern is 22 m wide
using strain controlled tests in the laboratory. The rate of and 44 m high and transformer hall cavern is 16 m wide and
strain softening (residual strain/peak strain) is generally more 22 m high. The analyses of large underground caverns have
in softer rocks than that of harder rocks. been carried out using conventional Mohr-Coulomb model
The paper presents the post peak response of granite and Mohr-Coulomb model with strain softening effect using
under strain controlled loading for confining pressures of FLAC. The support system has been installed in various stages
0 to 30 MPa in the laboratory. The strength and deformation in the form of shotcrete and rock bolts and effect of softening
parameters are determined from laboratory test results. The on cavern behavior under each stage is studied. The effect of
peak and residual strength parameters are determined using support system on the softening behavior is also studied in
Mohr Coulomb criteria. The variation of cohesion and friction detail.
angle with plastic strain is also determined. The rock shows It is seen from the analyses that there is significant effect
the vertical splitting under confining pressures of 0 and of the softening on caverns behavior and support systems also
10 MPa and shear failure under other high confining pressures. play a major role in softening behavior.
187

Determination of backfill requirements in multi-reef mining using numerical modelling

Anglo American Platinum, Johannesburg, South Africa
Keywords: stability analysis, rock stress, mine design, mining, numerical modelling, multi-reef mining
1 INTRODUCTION
1.1 Background
Primary Ground Control Districts have been established for
the UG2 based on the middling between reefs (UG2 and
Merensky), and rock mass conditions or quality. The mid-
dling between reefs varies between ∼14 m (UG2 underneath
Pothole Merensky) and ∼38 m (UG2 underneath Normal
Merensky). In between the Pothole Merensky and the Normal
Merensky, the transition zone is present where the middling
is between 14 m and 38 m.
The higher stress concentration on UG2 faces when min-
ing underneath Merensky pillars normally results in Poor
(RMR1976 ) ground conditions that have in the past led to large
scale falls of ground in the UG2 stopes. In many instances,
this resulted in the abandonment of the mining panels and
consequently reduced the UG2 extraction ratio between 10 Figure 1. Schematic showing stress driven failure mechanism when
Level (∼500 m below surface) and 20 Level (∼1000 m below mining UGII underneath a Merensky pillar.
surface). Experience has indicated that ground conditions
become worse in panels undermining Merensky pillars from
16 Level (∼800 m below surface). Experience gained with mining of the UG2 under Merensky
reef pillars over the last 10 years at Union Mine, indicates
that this mining to be most difficult for pillar widths varying
1.2 Postulated failure mechanism between 15 m and 65 m. The influence of different Merensky
pillar widths on the stability of the UG2 stope panels does
The large falls of ground that occurred may be referred to as not form part of this analysis. Merensky pillar widths of 35 m
back-breaks. A number of theories have been put forward by (Unpublished consulting report, 2002) and 50 m have been
numerous investigators as to the main cause that lead to the modelled and these results are presented below.
occurrence of back-breaks on the UGII mining underneath the
Merensky at low middlings (14–38 m). The authors agree that
at Union Mine, although it may be a combination of rock mass 2 REVIEW OF PREVIOUS WORK
stress and structure, the failure is primarily driven through
the changes in the stress field. Figure 1 shows a schematic 2.1 Review of numerical modelling
of the postulated failure mechanism as a result of stress.
Previous numerical modelling results have been reviewed and
are presented below to highlight the influence of backfill as
1.3 Objectives well as the stress levels on a UGII stope below a 35 m wide
Merensky pillar. A UDEC model was setup to simulate a 35 m
The objective of this analysis is to:
wide Merensky pillar under which the UGII was extracted
(i) identify the type of failure of the UG2 stope hanging wall in 20 m face advance increments (approximately 1–2 months
through numerical modelling, mining). The stope in the UDEC model was at a depth of
(ii) quantify the level of potential failure of the UG2 stope 1200 m (approximately 23 Level) below surface.
hanging wall through numerical modelling, Ultra-deep mining conditions may be experienced in the
(iii) determine whether the current in-panel support strategy UGII panel when the UGII stope face is mining close to the
would be able to cater for the predicted failure, centre of a 35 m Merensky pillar at a middling of 16 m at a
(iv) determine if and where backfill support would be suit- depth of approximately 1200 m below surface.
able for the conditions experienced through numerical The influence of leaving a 5 m wide pillar underneath the
modelling. Merensky pillar with only elongates as stope support was also
188

investigated. The model predicted failure of the elongate sup- Table 1. Summary of Analysis.
port and shearing of the hanging wall along the 5 m pillar. The
elongate combined with pack support also fails which allow Backfill Depth of Mine support
rock falls to occur. The model predicted that a combination Depth below surface options unstable zone std. FOS > 1.5
of the 5 m pillar, backfill and pencil sticks may prevent large 1020 m (∼20 Level) No B/fill ∼4 m Yes
opening within the UGII hanging wall beam and appears to 1120 m (∼22 Level) No B/fill ∼9 m No
have significantly reduced the potential for falls of ground in 1120 m (∼22 Level) B/fill 0m N/A
the UGII stope.
The UDEC modelling has indicated the importance of back-
fill from a depth of 1200 m (approximately 23 Level) below
surface, however the transition from no-backfill to back- unstable height of up to 8.9 m into the hanging wall at a Factor
fill mining i.e. from which depth is backfill a pre-requisite of Safety of 1. However, a Factor of Safety of at least 1.2 is
for undermining of Merensky pillars, has not been clearly used for support design at Union Mine, which means that a
established. From the numerical modelling performed previ- height of approximately 6 m could be supported on the current
ously it was recommended for stopes below 20 Level that no Mine Standard 10.9.9 (b).
undermining of Merensky pillars must be done beyond a pre-
determined limit (referred to as shadow-lines), which must
be indicated by the Rock Engineer. In addition, it was rec- 5 CONCLUSIONS
ommended that early warning system instrumentation (such
as the closure robot system) be installed to indicate the onset Table 1 presents a summary of the UDEC modelling results
of a large fall of ground or instability, where undermining is and the suitability of the Mine Standard 10.9.9 (b) when
planned beyond the shadow lines. undermining a 50 m wide Merensky pillar.
The analysis indicates that backfill support may be suitable
to limit tensile failure of the rock above UGII stope panels
3 NUMERICAL MODELLING for all stopes mined below a depth of ∼1100 m below surface
(22 Level).
3.1 Results
The UDEC model predicted no tensile failure of the hang- ACKNOWLEDGEMENTS
ing wall above the UGII stope when simulating a UGII stope
at a depth of 1120 m (approximately 22 Level) with backfill The authors would like to thank Anglo American Platinum for
(a-value ∼15 MPa and b-value ∼ 0.3). However, with no allowing us to publish this work.
backfill, unstable depths of ∼4 m and ∼9 m were predicted on
the UGII at depths of 1020 m and 1120 m.
REFERENCES
4 MINE SUPPORT STANDARD Anon. 2000. UDEC User Manual. ITASCA Consulting Group, Inc.
1st Edition, Minnesota, USA.
The current stope support standard (Mine Standard 10.9.9 (b)) Groundwork Consulting (Pty) Ltd. 2002. Consulting Report submit-
for UGII mining underneath Merensky pillars can support an ted to Union Mine, Anglo American Platinum.
189

Comparisons between two empirical yield criteria for rock masses
A.J. Li & R.S. Merifield

School of Engineering, Deakin University, Australia
Centre for Geotechnical and Material Modelling, The University of Newcastle, Australia
Keywords: slope stability, factor of safety, non-linear, limit analysis
1 INTRODUCTION
In general, rock masses are inhomogeneous, discontinuous

media composed of rock material and naturally occurring
discontinuities such as joints, fractures and bedding planes.
Because of these features, the strength of rock masses is
notoriously difficult to assess. Nonetheless, many criteria have
been proposed for estimating rock mass strength. Based on
the finite element upper and lower bound limit analysis meth-
ods, this study examined two empirical yield criteria for rock
masses, the Hoek-Brown failure criterion (2002) and the Dou-
glas criterion (2002). The solutions obtained in this study are
presented using Equation 1 in order to compare with those
presented by Li et al. (2008).
where N is the stability number, γ is the unit weight of the rock

mass, and H and F are the height and the safety factor of the
slope respectively. It showed that very different results may
be obtained using the same input parameters, by comparing
Figure 1 and Figure 2. Therefore, it is interesting to discuss
the source of these differences in this paper.
Figure 2. Average upper and lower bound solutions based on

Douglas failure criterion.
Figure 1. Average finite element limit analysis solutions for stabil-

ity numbers for β = 45◦ (Li et al. 2008).
190

Buckling failure of highly-stressed hard rocks surrounding deep underground

openings and confining effects of backfilling material
D.Y. Li
School of Resources and Safety Engineering, Central South University, Changsha, China
X.B. Li, Z.X. Liu & C.D. Ma

School of Resources and Safety Engineering, Central South University, Changsha, China
L.N.Y. Wong
Keywords: rock caverns; rock failure; rock stress; stability analysis
Buckling is a type of failure that has been observed around shown in Figure 2. It was evident that the smaller the slen-
underground openings in highly stressed as well as in jointed derness ratio, the larger the required stress for buckling. The
rock masses. Slabs can be formed due to the presence of effects of the slenderness ratio and eccentricity on the q/σθ
joint planes parallel to the excavation surfaces in the case of ratio are demonstrated in the parametric analysis presented in
highly jointed rock masses, or under a high compressive stress Figure 3. It could be seen that for a slenderness ratio of 10, a
that acts parallel to the excavation boundaries in intact rock confining pressure of no more than 5% of the axial stress was
masses. Buckling stability analysis is an important concept adequate to stop any deflection on the column. The effects of
for slim column under compression in material mechanics. the eccentricity of maximum tangential stress and the radial
The buckling failure of highly-stressed hard rocks surround- stress gradient were considered in the analysis. The maximum
ing deep underground openings is analyzed in this paper. The lateral displacement and the maximum moment on the rock
stress redistribution of surrounding rock masses around the slabs were calculated by using vector superposition method. It
underground opening can be shown in Figure 1. The tangen- showed that the slenderness, the eccentricity, the elastic mod-
tial stress (σ θ ) will be increasing and the radial stress (σ r ) ulus of the slabs and the loading stress played important roles
will be decreasing for the surrounding rock masses near the on the buckling failure of rock slabs. The confining pressure
excavation surfaces. Under the approximately uniaxial com- provided by filling material could prove adequate to control
pression, the rock masses may split and form a number of buckling failure under certain loading conditions. The way to
surface-parallel slabs. obtain the basic parameters in buckling analysis was also dis-
The buckling analysis was carried out by considering the cussed in the paper. These parameters can be basically divided
axial loading condition, the eccentric loadings and the stress into three categories: mechanical parameters (P, σθ and σr ),
gradient in the rock slabs. It indicated that the buckling stress material parameters (E and q) and geometric parameters
is a function of the modulus of elasticity and the slenderness (l, l0 , e, t and T ).
ratio. The effect of these parameters on the buckling stress is
Figure 1. The stress distribution of surrounding rock masses around

deep opening (a) The splitting slabs of highly-stressed rock masses; Figure 2. Buckling stress versus slenderness ratio for buckling
and (b) the simplified mechanical model of one slab. failure using Euler’s formula.
191

Figure 4. Photos of buckling failure in the surrounding rock masses
of laneway at the Level +700 m in Maluping Mine.
Figure 3. The ratio of confining stress and the tangential stress
versus slenderness ratio when considering the confining effect.
Table 1. In-situ stress measurement results at the testing locations in Maluping Mine.
σ1 σ2 σ3
Testing Stress Dip Dip Stress Dip Dip Stress Dip Dip
location /MPa direction/◦ angle/◦ /MPa direction/◦ angle/◦ /MPa direction/◦ angle/◦
M700-1# 34.49 207.16 16.95 8.53 111.30 18.50 0.61 336.81 64.47
M750-2# 8.45 143.47 17.05 7.46 28.48 54.03 −2.53 243.92 30.60
M750-3# 8.94 123.54 3.79 4.11 31.58 27.32 −1.80 220.80 62.38
M600-4# 17.02 44.08 28.46 2.83 313.67 0.64 −2.64 222.54 61.53
According to the buckling theory, a case study was pro- into the filling materials to partially replace the cement.
vided at the Maluping Mine in the Guizhou Kailin Group, Several groups of compressive strength of the phosphorus
China. Buckling and slabbing failure has occurred in the gypsum tests with different ratios of consolidated filling were
level +700 m at the Maluping Mine. Figure 4 shows the obtained. It was found that the confining pressure with 1.5–
buckling failure of the surrounding rock masses even after 2.1 MPa provided by filling material is sufficient to stop
shortcreting at the Level +700 m in the subsidiary company of buckling and slabbing failure in the surrounding rock masses.
Maluping Mine. By measuring the in-situ stresses, it was It can realize the stability of the highly-stressed underground
found that the maximum principal stress is as high as 35 MPa stopes.
at the Level +700 m, shown in Table 1. Cutting-and-filling
method is widely used nowadays in underground mines. How-
ever, the reinforcement and support mechanism of the filling ACKNOWLEDGEMENTS
material has not been fully understood by the researchers yet.
The back filling material is mainly composed by the phos- The paper was financially supported by the National
phorus gypsum, 425# ordinary Portland cement and fly ash. Basic Research Program of China (973 Program No.
To reduce the cost and to improve the flow properties of 2010CB732004) and the Natural Science Foundation of China
the paste, an appropriate combination of fly ash was added (grant No. 50934006 and 51074177).
192

Bemposta II powerhouse shaft: Geotechnical characterization, design and construction
C. Lima, E. Resende, C. Esteves & J. Neves

EDP, Portugal
Keywords: site characterization, rock support, monitoring, numerical modelling
1 BEMPOSTA REPOWERING SCHEME
The Bemposta hydropower plant, built in the 60 decade of

the past century, is the most downstream of a group of three
plants exploited by EDP (Energies of Portugal, the biggest util-
ity of the country), located at the international Douro River’s
Portuguese stretch, in the North-East of Portugal (Figure 1).
The Bemposta II project, now under construction, corre-
sponds to the repowering scheme of Bemposta plant. The
main elements of this new project are a 60 m deep power-
house shaft with 22 meters in diameter and a hydraulic circuit
of tunnel: headrace tunnel length of 400 m with a cross section
measuring 12 m in diameter, and tailrace tunnel with a vari-
able size cross section whose maximum height reaches 21 m.
Bemposta II is located at the right bank of Douro river, next
and surrounding the existing scheme.
Bemposta II is being constructed in a heterogeneous
metamorphic rock mass made up of Hercynian rocks origi- Figure 1. Schematic representation of international stretch of
nated from the partial melting and injection of pre-existing Douro river cascade. Portuguese plants identified.
Cambric rocks. Most part of the headrace tunnel is located in
an anisotropic micaxist and migmatitic rock while the power- The paper portrays the Bemposta II scheme, summarises
house shaft and the tailrace tunnel are in gneissic granite with the geotechnical field and laboratory investigation results
xenoliths of the pre-existing micaxist. and focuses the staged construction adopted. The monitoring
The support system installed during excavations of the new scheme implemented and results obtained from a numeri-
powerhouse shaft allowed erecting the concrete structure at the cal model employed to analyse the mechanical behaviour
upper half of the shaft independently and before completion of the rock mass surrounding the powerhouse shaft are also
of the excavations at the lower half of it. presented.
193

Initiation and evolution of failure zones in the vicinity of heterogeneities in rock masses:
Seismology and deep oil well stability problems
L.A. Nazarova & L.A. Nazarov

M.I. Epov & I.N. Eltsov

Institute of Petroleum and Gas Geophysics, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Keywords: rock mass, numerical modelling, rock failure
1 INTRODUCTION
The irreversible deformation zones may arise in the vicin-

ity of rock mass heterogeneities when certain geomechanical
and geotechnical factors (natural stresses magnitude, rock
strength properties, shape of an underground object etc.) occur
together. It is necessary to determine dimensions and con-
figuration of such zones for the stability assessment in mine
workings and oil wells, or for planning in situ stress measure-
ment experiments. Similar zones initiate in the Earth’s bowels
in the vicinity of focuses of forthcoming seismic events. Just
failure processes are the probable cause of seismic activity.
2 ESTIMATE OF THE FOCAL PARAMETERS OF AN

INCIPIENT SEISMIC EVENT BY USING GEODESY
DATA OBTAINED ON THE DAYLIGHT SURFACE Figure 1. Calculation domain and the equivalent source model.
Earthquakes in the Earth crust mostly take place nearby faults The surface GPS-monitoring allows quantitative assess-
so we associate the hypocenral area of an incipient seismic ment of increments in horizontal strain tensor components
event with an section S of a fault F that differs in properties. Dkm (xn , yn ) (k = 1, 2; n = 1, …, N ) on a discrete point set. Is it
Even a slight variation in natural tectonic stresses due to, for possible to find parameters of an anomalous area with such
instance, movement of lithospheric plates change geomechan- data at hand? Evidently, it is practically impossible to deter-
ical fields, and it is possible to record this disturbance on the mine shape, spatial position and properties of S based on the
Earth surface with the help of space geodesy methods. The “passive” information Dkm , as peculiarities of the additional
estimates of displacement variations in the Altai earthquake strain field smoothed over already at the distance of few diam-
epicenter (27.09.2003, M = 7.5) showed that the strain range eters of S. For this reason, we use the equivalent point source
accuracy (10−8 ) of contemporary GPS techniques is sufficient concept, which is applied to inversion of seismic data, and
to identify the preparation of the moderate and strong crustal formulate another boundary-value problem: find a source that
earthquakes. generates such a strain field in the weightless domain P that
We analyzed a typical situation (Fig. 1): the Earth’s crust is close to Dkm on the daylight surface. Introduce an objective
part occupied by a parallelepiped P (x, y, z-coordinates; z = 0 – function:
the daylight surface) contains an inclined discontinuity F.
Geomedium deformation in P\F is described by the incre-
mental elastoplastic model with the Mohr-Coulomb condition
(Nikolaevsky 1996), whereas state equations for F were set in
pursuance of (Barton 1986). where Ekm = Gi ekmi ; ekmi is the strain field generated in P by
As the anomalous area S (Fig. 1) we take a rectangle with a unit force applied at the point (px , py , pz ) and oriented in the
the center (sx , sy , sz ) so that stiffnesses in this rectangle are line of i(i = x, y, z). The required parameters of the equivalent
less than corresponding one of the remaining part of F. The source are the arguments such that is minimal: components
deformation and strength properties were typical of the rocks of the force Gi and the location pi .
that compose the modeling object, i.e. the hypocenter area of Minimization of is implemented with the original com-
the Altai earthquake. bination of methods of steepest descend and least squares.
194

analyzed the process of evolution of irreversible strain zone D
in the vicinity of a deep well bored through a pack of thin seams
with weak strength characteristics (internal friction angle ϕ
and cohesion C) – quite a common case for the oil and gas
well drilling.
A well with radius a, filled with drill mud is drilled verti-
cally downward through a horizontally seamy rock mass. In
this case, the problem is axially symmetrical, the coordinate
system is cylindrical, axis z is oriented along the well, axis r
goes along the well radius, b is the well bottom depth (Fig. 2).
The drilling process was modeled by using the initial stress
method: the well bottom advance for a distance h was fit-
ted with replacement of forces at the analyzed zone contour
{0 < r < a, b < z < b + h} by the forces generated by the drill
mud. Hereat, given the Mohr-Coulomb condition was violated
in an element, the stresses were reduced to yield surface.
Figure 2 shows a standard situation: a well crosses a pack of
three thin seams composed of different rocks (which strength
characteristics are presented in the same figure on the right).
The shapes of the zone D are shown for the well advance
h = 0.1 m. It appeared that in such conditions, the zone D
can be up to 1–1.5 well radius ahead of the well bottom, and
the weak seam 2 experiences failure earlier than the stronger
neighbor seam 1 (thick black line, b = 4000.3 m).
The originated zone of irreversible strains can alter filtration
properties in the medium. Let the radial permeability of seams,
kr , be governed by a certainly oriented system of joints with an
Figure 2. Evolution of shapes of failure zones under well bore average opening d, then: kr , ∼ d 3 . Additional vertical strains
advance. ezz alter the opening by dezz , thus, relative variation in the
permeability equals (2):
According to the numeral experiments, the most suitable
equivalent source is pair of concentrate forces applied at the
opposite surfaces of a fault (“couple of forces with moment”)
in the center of S, and the inverse problem to be solved needs For models with the discussed parameters and depths of the
N = 30–40 points to be selected in the epicenter area of an order of 4 kilometers, the maximum value of the permeability
incipient seismic event. Thus, the proposed approach allows variation is 2–5% depending of value of lateral thrust coeffi-
determination of focal parameters of an future seismic event cient which characterized ratio of horizontal and vertical stress
by using a priori data on variation of strains on the daylight sur- components in natural field.
face as well as assessment of the seismic event intensity based Financial support of the study is from the Russian Foun-
on the statistical information about earthquake efficiency. dation for Basic Research, Projects Nos. 10–05–00736 and
10–05–00835, às well as from the Siberian Branch, Russian
Academy of Sciences, Integration Project No. 60, and Russian
3 EVOLUTION OF IRREVERSIBLE STRAINS ZONES Academy of Sciences, Program No. 16.8.
AND THE ALTERATION OF ROCK PERMABILITY
OF NEARBY A WELL BORE IN THE COURSE OF
DEEP DRILLLING REFERENCES
Deep hydrocarbon exploration and extraction boring usually Nikolaevsky, V.N. 1996. Geomechanics and Fluid Dynamics,
faces such negative issues as deviation of a well bore from Moscow: Nedra (in Russian).
a pre-set trajectory, failure and permeability variations in the Barton, N.R. 1986. Deformation phenomena in jointed rock.
well bore zone. Based on the same constitutive relations, we Geotechnique, 36(2): 147–167.
195

Monitoring of cracks opening in the walls of deep pits
Victor Oparin & Vladimir Vostrikov

Subject: Information system, artificial intelligence and other advanced techniques
Keywords: monitoring; field measurements
1 INTRODUCTION
In many areas of production it is often required to know

if strains and displacements may occur in elements of dif-
ferent constructions and facilities. Fracturing of an element
can arise from the limit stresses, and for this reason contin-
uous monitoring of fracture propagation becomes essential
to ensure safety for personnel and equipment under emer-
gency conditions (Bykhovets, A.N., 2002). The solution of
this problem is an instrumentation system that: possesses sev-
eral sensing passages; can be integrated into an instrument
communication net and can transfer data to a data acquisition Figure 1. Signal generated by the large explosion.
and processing center in real time; can run in a wide range
of working temperatures, etc. Every branch of industry or an
object to be monitored naturally needs a specifically designed
instrumentation system (Kozyrev, et al., 1996).
There exist such monitoring instruments to be used to the
purposes of mining safety, but they are inapplicable in coal
temperature areas (Dimaki and Psakhie, 2009). The given
paper describes an advanced instrumentation system suitable
for operation at open pits in the severe climatic conditions of
Siberia.
Figure 2. A fragment of the recording implemented by two strain
2 DESCRIPTION OF INSTRUMENTATION SYSTEM cells installed near a large block at Level 125 m, next to the site of
extensive collapse in 2006.
The new developed multi-channel system consists of a strain
cell and antenna, logger, heat-protective housing and indepen- in Russia, the Udachny open pit mine extracted to a depth of
dent power supply; the slot strain cell is meant for measuring 560 m by now.
displacements in cracks in rocks. In Figure 1 a fragment of sensor recording after a large blast
There are several designs developed for strain cells. The is presented. The blast to sensor distance was 350–370 m. The
strain cells can simply and readily be installed in vertical and sensor caught the blast-generated elastic wave (Blast Moment
horizontal fractures in rocks, dams, etc. Equipped with aux- in Figure 1) and, in approx 3 seconds, a deformation wave that
iliary devices, these strain cells become applicable to strain caused the monitored rock block to displace by 60 µm. So,
assessment in various industrial constructions. the conclusion was drawn that major impact on the block dis-
Using a number of the designed loggers, it is possible to placement belonged to the slower (117–123 m/s) deformation
develop a multichannel instrumentation net with the data wave, in succession to the elastic wave of the explosion.
transfer to an information acquisition center. The logger The other example of the sensing system application is as
modems operate in a pass-through mode, which enables infor- follows. The instrumentation system composed of two strain
mation communication for a considerable distance by means cells was installed on one of dangerous sites (hereinafter,
of message relay points. block). One strain cell was placed above the block, the other
cell was installed 6 m beneath the block. Figure 2 presents a
fragment of the signal record received from the two strain cells.
3 APPLICATION OF THE INSTRUMENTATION The upper curve is the upper cell signal, and the lower curve
SYSTEM is the lower cell signal. The atop-placed cell showed that the
fracture first narrowed, somewhat by 0.1 mm, and then opened
This section illustrates application of the developed instru- abruptly by 0.5 mm. At the time the lower installed strain cell
mentation system in terms of the deepest diamond open pit recorded 0.4 mm constriction of the fracture. Based on this
196

shown in Figure 4. The Data Acquisition Center (DAC) is
located on the surface, approximately 2 km away from the
open pit, thus, beyond the line of sight in relation to the open
pit bottom. The data of the measurement systems are first to be
“elevated” from the bottom to the DAC’s line-of-sight point.
To this effect, the message relay point R1 is positioned at the
edge of the open pit. There is another difficulty: P1 does not
“see” the pitwall above which it is situated. Another message
relay point is positioned on the opposite pit wall for this rea-
son. It receives signals from the IS1 and IS2 and transfers them
to R2 relay point. Every instrumentation system consists of a
Figure 3. Experiment at Level −170 m. logger, three strain cells, accumulator, rf modem and antenna.
5 CONCLUSION
Based on the instrumentation system designed, in the context

of a real mining project, it has been succeeded in simplifying
the installation procedure for strain cells and the system itself,
operation of the cells, as well as the data transfer, processing
and analysis. This result is of key importance for open mining
in the severe climatic and industrial conditions of the extreme
north.
The primary advantage of the described instrumentation
Figure 4. Udachny open pitwall monitoring. system is the continuous monitoring of fracture propagation
in pitwalls via surface data acquisition and processing centers.
it was supposed that the upper portion of the block lurched. In case of an off-normal situation, when dangerous gravita-
The strain cell placed above the block showed sufficiently high tional phenomena shape up, for instance, collapses, landslides
displacement rates, 0.1 mm/day within a few consecutive days. or quick subsidences of large rock blocks in slopes, the devel-
After the measurement results had been analyzed, the decision oped instrumentation system will detect initiation of the event
was made to blast this block. long before emergency, and will ensure the just-in-time deci-
One more evaluation survey at the Udachny open pit mine sion making on precautions to prevent or mitigate social and
was executed at Level −170 m. economic damage.
The experimental results are shown in Figure 3. Initial stress
was created by means of the screw jack up to 70 T. Within this
time the crack opened by 0.45 mm (according to sensor 1). REFERENCES
Later on the crack continued gradual opening by 0.04 mm for
the next 10 minutes. Then the stress was dropped, and the Bykhovets, A.N., Tarasov, G.E., Kozyrev, A.A., Reshetnyak, S.P.,
crack opening became half as much, so, the block came back Sery, S.S., Cherevko, N.V. 2002. Geomechanical and engineering
to its position, shifted a little although. The conclusion was that grounding for optimization of open pit benches and walls. Mining
Magazine, Special Issue. (in Russian)
the rock block was in the stable condition. Existing dynamic Dimaki, A.V., Psakhie, S.G. 2009. Spaced monitoring system for
effect at the test site (large explosions, trucks moving down displacements in block media, designed based on SDVIG-4MR
the overlying bench) is much smaller in value than it was in complex. Journal of Mining Science, Vol. 45, No. 3.
the experiment, and the block exhibited weak response and Kozyrev, A.A. et al (1996). Strata Control in the Tectonically Stressed
re-set. Rocks. Apatity: Kola Scientific Center of the Russian Academy of
Sciences. (in Russian)
4 MONITORING OF THE UDACHNY OPEN PIT WALLS
The instrumentation systems IS1 and IS2 are installed at Level

+115 m and Level −170 m, respectively, as schematically
197

Mine plug design for the San Genaro Ramp
A.A. Samaniego & R.L. Cabrera

SVS Ingenieros SAC
Keywords: case studies, mine design, stability analysis
1 INTRODUCTION The rock mass strength differs from that of the intact rock
because of the discontinuities such as joints, fissures, faults,
Castrovirreyna Compania Minera SA operates the San Genaro etc. present in the former, which is why the failure crite-
mine in Central Peru, is planning the expansion of its exis- ria for the rock mass were defined using the classification
tent tailings deposit, which will require the construction of index proposed by Hoek & Brown (1988), updated by Hoek,
an underground mine plug to prevent the flow of water and Carranza-Torres and Corkum (2002).
tailings into the mine. The plug should be in operation dur- This approach takes into account the strength of the intact
ing the mine life and after mine closure. The aim of the study rock and the constants mb, s and a, which were estimated
carried out was to assess the geomechanical conditions of the Based on the structure and condition of rock mass discontinu-
area where the plug is to be located and design it to ensure its ities, being finally reflected by the geological strength index
stability in the long term. GSI. By using the generalized form of the failure criterion pro-
posed by Hoek and Brown, the value of the elastic modulus for
2 GEOMECHANICAL MODELS the rock mass (Em) at the site proposed for the construction
of the plug was estimated at Em = 5.296 MPa, a value which
San Genaro ramp was excavated in the Caudalosa volcanic corresponds to a competent andesite.
formation of the Miocene, which is the one with the largest During the field study, some areas showing water flows and
extension in the area. This formation consists of volcanic filtrations were identified, these being the result of seepage of
rocks composed of lava flows and breccias of dark gray and rain water through the fractured rock mass located above the
green andesitic flows, with inter-bedded lenticular pyroclastic ramp. It was anticipated that along major fault zones, due to
and tuffaceous sandstones. Some additional sub-units such as the increase in permeability of the rock mass, the flow of water
porphyritic andesites, andesitic tuffs, basalt and pyroclastic and transport of tailings material would preferably follow the
tuffs have also been identified in the surrounding areas. direction of the structures within the rock mass and would
The main structures found in the Castrovirreyna mine have eventually fill the fractures, which may lead to a decrease the
a general EW trend, with another system showing a NW-SE permeability of the rock mass.
and to a lesser extent a NE-SW trend. The structures in the During the field research stage, a geomechanical cell and
study area of San Genaro ramp have the following characteris- detail lines mapping was carried out from the mine entrance
tics: major fault zones and some shear zones (between 0 + 000 (0 + 000) to chainage 0 + 180.
and 0 + 088 m), generally sub-vertical and following a general The study area was divided into four structural domains
trend EW, having been affected by hydrothermal alteration, the showing similar behaviour, lithological, structural and alter-
shear zones being up to 20 m thick. The minor faults located ation characteristics.
between 0 + 088 and 0 + 180 m have thicknesses varying Table 1 summarizes the main geomechanical character-
from 50 to 100 mm, being filled with gypsum, calcite and istics of the sections that were assessed during the field
fractured rock with some quartz crystals. study.
Table 1. Main characteristics of the structural domains.

Description DE-1 DE-2 DE-3 DE-4
Spacing (cm) 14–20 15–25 20–35 20–40

Roughness Slight Slight Slight Slight
Aperture (mm) 0.1–1 0.1–1 0.1–1 <0.1
Filling Soft < 5 mm Soft < 5 mm Soft < 5 mm Soft < 5 mm
Weathering High Moderate Slighy Slight
Presence of water Dripping Wet Humid Slightly humid
RMR 30–39 40–49 50–59 60–69
Type of rock IV III III II
Chainage 0 + 000 to 0 + 027 0 + 027 to 0 + 032 0 + 032 to 0 + 051 0 + 111 to 0 + 180
0 + 051 to 0 + 088 0 + 088 to 0 + 111
198

3 DESIGN OF THE PLUG long hollow section located downstream, with a 3 m wide and
2.7 m high internal opening. The solid concrete wedge sec-
The type of plug most appropriate for the purpose of the tion was extended slightly along the upstream end in order
San Genaro ramp must be water-tight and built of plain con- to anchor the plug in the rock mass and thus ensure a better
crete, the length of which should be larger than 6.0 meters, contact and improve the shear strength between the rock wall
considering the relatively high hydraulic pressures expected. surface and the solid concrete.
The plug was designed using empirical calculation tech-
niques generally accepted in civil and mining engineering
practice. The geometry of the plug was defined by analysing 4 MODELING AND STRESS AND STRAIN ANALYSIS
the potential failure mechanisms that could occur, which are
largely independent of each other. As a complement to the empirical methodology used to design
Using the empirical methodology, the shear strength anal- the plug, numerical analysis tools were also used through the
ysis of the plug was performed assuming the maximum shear application of specialized computer software.
strength of unreinforced concrete as indicated in ACI (1972). The stresses developed around the ramp area were cal-
The sizing of the plug length was based on the concrete/rock culated using the Phases 2 geotechnical software and the
interface shear strength, using the equation developed by thickness of the bottom slab and the walls were calculated
Garrett & Campbell-Pitt (1961). The bending stress was using the SAP2000 (Structural Analysis Program) software.
analysed using the ACI 318-95 Standard. The stresses along the walls of the ramp were estimated
Calculations were then carried out to verify the stability of by means of a finite element model, consisting of a total of
the area against hydraulic fracturing, by means of the crite- 2794 shell type elements. The rock properties used for the
rion suggested by Bergh-Christensen (1988). The excessive calculations were: specific gravity = 2.70 t/m3 , Poisson ratio
hydraulic gradient due to seepage was calculated using the (µ) = 0.32, mi = 18, mb = 1.79, s = 0.023, a = 0.503 and a
equation of Benson (1987) and the calculation of dynamic modulus of elasticity Em = 5296 MPa (5.4 × 105 t/m2 ).
loads water and water hammer impacts was done using the The results show that the maximum bending moment in
equations developed by Westergaard (1931) and Seed and the 1 m thick side walls is 15 t-m and the maximum shear
Idriss (1983). For the stress-strain analysis, the Phases finite stress reaches 2.45 kg/cm2 . With these results the thickness
element software was used and for the structural analysis, the and reinforcement of the walls of the plug were designed,
SAP2000 software. resulting in a 1 m thick concrete wall along the gallery, which
The parameters used for the design, required by the above is sufficient to ensure the long-term stability of the plug.
mentioned criteria were.
According to the results of the calculations, the length 5 CONCLUSIONS
required for the plug is 5.3 m to withstand the tailings and
the water static pressure. However, a final length of 10 m was Based on the geomechanical assessment carried out, it was
adopted, to ensure a proper structural behaviour of the plug found that the most appropriate location for the plug was along
and minimise the leakage of water and tailings though it. the structural domain DE-4, between chainages 0 + 125 and
The construction of a concrete plug was considered, com- 0 + 115 m.
posed of a 5 m long upstream solid concrete section and a 5 m Due to its length, the reduction of the maximum potential
pressure load on the plug crown pillar and the longer trajectory
Table 2. Design Parameters. of the infiltration flow-lines, it was determined that a plain
concrete plug was appropriate.
Description Parameters A final length of 10 m was adopted for the plug, in order to
ensure proper load bearing and reduce the potential leakage of
Site location of the plug San Genaro Ramp: chainage water and tailings through it, considering that it only requires
0 + 115 m to 0 + 125 m, a length of 5.3 m of the plug to retain the tailings and water
4682 m.a.s.l. pressure according to the calculations and analysis carried out
Rock type of site emplacement Rock type II (RMR = 63)
under static and dynamic conditions.
Ramp section 4m×4m
Area and perimeter A = 14.3 m2 , P = 14.3
The plug geometry finally adopted was thus a wedge-
Hydraulic load 43 m shaped concrete wedge, consisting of a 5 m upstream solid
Current level of tailings 4725 m.a.s.l. concrete section and a 5 m long hollow section at the down-
Projected height of tailings 4732 m.a.s.l. stream end.
Lithostatic height 50 m
Thickness of topsoil 3m
Density of tailings 2 t/m3 REFERENCES
Density of rock 2.7 t/m3
Density of topsoil 1.8 t/m3 Auld, F.A. Design of Underground Plugs. International Journal of
Maximum seismic acceleration 0.25 g Mining Engineering, Vol. 1, pp. 189–228, 1983.
Concrete 280 kg/cm2 (30 MPa) Carter T., Castro L., Guía para el Diseño de Tapones, Golder
Maximum hydraulic gradient 10 m/m Associates Ltd., diciembre 2006.
Cement Portland type V Carter T., Castro L., Guía para el Diseño de Tapones para Cierre
Steel Tensile strenght: fy = 4200 kg/cm2 de Labores Mineras, Dirección General de Asuntos Ambientales
Mineros, Sub Sector Minería del Perú, setiembre 2007.
199

A case study of alteration effects in granitic rocks on rock cavern stability at

Jinzhou of NE China
Y.J. Shang
Key Lab of Engineering Geomechanics, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
Y.Y. Shi
Sichuan Ertan International Engineering Consulting Co. Ltd, Chengdu, China
Subject: Rock materials and rock mass property testing
Keywords: site characterization, rock caverns, lab testing, alternative rocks
1 INTRODUCTION Statistic results of over 200 nos dykes shown that the main
strike of basic veins is nearly SN, dip westward, with dip angle
In China the magmatic activity is frequently occurred espe- of 30◦ (Figure 2).
cially in the Yanshan Movement, and various igneous rocks
are often appeared in most of China (Sun & Peng, 1985).
China sits at west rim of Pacific Plate, the granite is widely 3 DIFFERENTIAL WEATHERING OF SHALLOW
spread and various magma activities were so intensive that ALTERNATIVE ROCKS
they form different kinds of lower hydrothermal metal deposit.
In a long time, the mining experts paid more attentions on In shallow depth or even in ground surface, the alternative
the ore deposit formation with alteration and alterative rocks. rocks were further weathered to gray green, red brown clays,
Besides, the alteration makes smectite to form from many which are very similar to red clays in open fissures or joints.
intermediate to basic igneous rocks. As a result, more types Figure 3 is red clays weathered from basic dykes and filled
of expansive rocks are appeared with wide distribution and in gently dipping joints. In shallow depth or the surface, the
distinctive features, which incurred many serious engineering original alternative rocks dominated by smectite are weath-
geological problems in engineering construction and mining ered to illite/smectite mix layer and kaolinite clay minerals.
(ISRM, 1994). They constitute the weak rocks to form weak intercalated lay-
ers, which control the natural slope stability. For instance, in
field survey, it was observed that the alternative rock as weak
2 DISTRIBUTION OF INTERMEDIATE-BASIC interclated layer, directly controlled the stability of rock mass
DYKES AND ALTERNATIVE MINERALS above it, or even covering as smooth slope surface. In natural
IN JINZHOU GRANITE status, the slope angle is less than 40◦ , and height is less than
30 m.
During field investigation and drilling works at the Jinzhou
granites at Tianqiao, it was observed that lots of acid and
intermediate-basic dykes and sills are intruded into granite 4 ALTERNATIVE ROCK STABILITY IN WATER AND
(Liao, 1989). From survey and thin section identification of 25 ENGINEERING MEASUREMENTS
nos dyke samples, it was found that the dykes include diabase,
aplite, granophyre, diorite porphyrite, andesite pophyrite, and The rock engineering practice shows that for swelling rocks,
lamproite, etc. The distribution of dykes and sills is not well especially those under groundwater table, the swelling is
uniform, which is several centimeters to more than 10 meters
thick. Among them, the dark green basic dykes (diabase) are
well emerged, in the form of sills in a steady thickness and an
altitudes appearing in granites (Figure 1).
Figure 1. Diabase vein (sill) intruding into completely decomposed Figure 2. Rose map and equal density chart of intermediate-basic
granites in shallow depth and keeping well. veins.
200

Figure 3. XRD chart and plate of red clay from alterated veins filling in gently dipping joints.
difficult to occur without previous drying or the water lost 4) Strengthen discharging groundwater measures to prevent
no more than boundary values. Thus drying and water loss groundwater from accumulated in caverns.
are necessary conditions for alternative rocks to expand with
large deformation. Understanding of this phenomenon is sig-
5 CONCLUSIONS
nificant to adjust shotcrete time and closing measures in due
course (Qu, 1984).
There exist various kinds of dykes in Jinzhou granite of NE
When encounter with smectitic swelling rocks, the follow-
China, especially smectitic intermediate-basic dykes, which
ing methods are useful for prevention of geohazards in rock
constitute one challenge for construction of the rock caverns
caverns.
there. The altitude of basic veins is steady and behaves cer-
1) Advance geological prediction, make clear the distribu- tain magnitude sills with a strike of SN, dipping SWW and
tion, magnitude and features of expansive alternative rocks, angle 30◦ . In ground surface these alterative rocks become
adopt measures to keep away from their affects; weak intercalated layer to control rock slope stability. When
2) In sections with alternative rocks, micro-step excava- the buried depth is over 90 m the cored smectitic alternative
tion, weak explosion, more times excavation, decreasing rocks behave moderate expansion, which threaten the rock
disturbance, setting up shotcrete and anchor-bolts in time; cavern stability. Quick closing and isolating inflow of ground-
3) Drying wash before shotcrete, if not, clean out of alterna- water in time are main engineering measures for prevention
tive rocks with obvious swelling; of associated geohazards.
201

Design of tunnels in jointed rock on the basis of mechanical models and

classification systems
R. Sommer & W. Wittke

WBI Professor Dr.-Ing. W. Wittke Consulting Engineers for Foundation Engineering and Construction in Rock,
Aachen/Stuttgart/Guadalajara
Keywords: case studies, numerical modelling, rock mass classification, rock support, stability analysis, tunnelling
1 INTRODUCTION
Two frequently applied design methods for tunnels in jointed

rock are the design based on rock mechanical models and
corresponding analyses and the design based on classification
systems.
The first method is mainly based on the results of geotechni-
cal investigations, stability analyses as well as on monitoring
during construction.This design method was recommended by
the study group “tunneling” of the German Geotechnical Soci-
ety [DGGT, 1995] as well as the European Regional Technical
Committee no. 9 for Tunneling and Underground Construc-
tion (ERTC 9) of the International Society for Soil Mechanics
and Foundation Engineering (ISSMFE) [ERTC 9, 1997] and
has been successfully applied by WBI and other consulting
engineers for several decades (see e.g. [Wittke, 1990; Wittke,
2000; Wittke et al., 2002]).
A characteristic feature of the so-called classification sys-
tems is that rock mass properties and other factors influencing
the stability of a tunnel such as in-situ stresses and ground-
water conditions are condensed to a single numerical value
referred to as “rock mass rating index”.
According to Bieniawski [Bieniawski, 1989], the developer
of the RMR classification system, “a classification system
is not intended to replace analytical modeling, site investiga-
tions and monitoring but should be used in conjunction with
these tools of rock engineering design”. Different from this
statement, classification systems in the recent past have been Figure 1. Design based on rock mechanical models.
used more and more as design methods without any kind of
analyses.
In this paper the two design methods for tunnels in jointed of the rock mass for all units the planned tunnel is passing
rock are critically reviewed and compared also by means of through. Also the in-situ stress state must be considered. The
examples. orientation of the discontinuity system with respect to the
tunnel considerably influences the type and amount of sup-
port measures and thus is an important aspect of every model
2 DESIGN METHOD BASED ON ROCK (fig. 1).
MECHANICAL MODELS On the basis of a rock mechanical model stability analyses
and serviceability proofs are carried out considering parame-
The basic procedure of this design method is outlined in fig. 1. ter variations. The rock – structure interaction and anisotropic
Based on the results of geotechnical investigations as well rock mass behavior are accounted for using a suitable analysis
as engineering and geological judgment a rock mechanical method such as the finite element method (FEM). Supple-
model is established. This includes models describing the mentary analyses with regard to the stability of rock wedges
structure and texture of the intact rock and the discontinuities. adjacent to excavation surfaces may become necessary as well.
In addition, the rock mechanical model includes the parame- On the basis of the analysis results the final design is carried
ters which describe the stress-strain behavior and permeability out and followed by construction of the tunnel (fig. 1).
202

During construction a documentation of the different con- In addition, the rock mass rating indices do not adequately
struction phases and a monitoring program are carried out cover all influencing parameters such as orientations of the
(fig. 1). In addition to monitoring of displacements and even- discontinuity sets, deformability and potential anisotropies.
tually also of anchor loads and stresses in the shotcrete and in As a consequence classification systems can lead to an unsafe
the lining of reinforced concrete geotechnical mapping of the and noneconomic design [Sommer, 2009; Sommer, 2010].
temporary face is carried out. By these means, the assump-
tions the design is based on can be reviewed by comparing the
results of monitoring and analysis. Also back analyses may REFERENCES
be required if the result of the monitoring is not in agreement
with the prediciton. Thus, if necessary, the support measures Barton, N.: Some new Q-value correlations to assist in site charac-
can be adapted or modified during construction. terisation and tunnel design. Int. J. Rock Mech. Min. Sci. 39(2),
2002.
Barton, N.; Grimstad, E.: The Q-System following thirty years of
development and application in tunnelling projects. In: Proc. ISRM
3 DESIGN METHOD BASED ON CLASSIFICATION Symp. Eurock 2004, 2004.
SYSTEMS Barton, N.; Lien, R.; Lunde, J.: Engineering classification of rock
masses for the design of tunnel support. Norwegian Geotech. Inst.,
In principle all classification systems are based on coeffi- Publ. Nr. 106, 1974.
cients by means of which it is attempted to account for the Bieniawski, Z. T.: Engineering rock mass classifica-tions. Wiley:
rock mechanical parameters, in-situ stresses as well as for the New York, 1989.
groundwater conditions. By means of empirical functions of Deere, D. U.; Hendron, A. J.; Patton, F. D.; Cording, E. J.: Design
the coefficients the so-called “rock mass rating index” is cal- of surface and near surface construction in rock. Soc. Min. Eng.
culated. Based on such a single index design recommendations AIME, New York, 1967.
DGGT (German Geotechnical Society): Empfehlungen des Arbeit-
for support measures denoted as “support classes” are given. skreises “Tunnelbau” (ETB). Ernst & Sohn: Berlin, 1995.
The basis of each classification system is empirical. This ERTC 9: Recommendations of the ISSMFE European Regional
means that the selection of the influencing variables, their Technical Committee no. 9 for Tunneling and Underground Con-
rating and the recommended support classes are based on struction – Geotechnical aspects of the design of shallow bored
the experience of the developer gained from practical cases. tunnels in soils and soft rock. Ernst & Sohn: Berlin, 1997.
As a consequence, classification systems generally have a Grimstad, E.; Barton N.: Updating the Q-system for NMT. In: Proc.
subjective aspect. Int. Symp. on Sprayed Concrete, Oslo, 1993.
Examples for classification systems are the “rock quality Palmström, A.: RMi – a rock mass characterization system for rock
designation index” [Deere et al., 1967], which serves as an engineering purposes. Doctoral Thesis, University of Oslo, 1995.
input parameter for different classification systems as well as Palmström, A.; Broch, E.: Use and misuse of rock mass classification
systems with particular reference to the Q-System. Tunnelling and
the Q system developed by Barton [Barton, 2002; Barton et al., Underground Space Technology 21(6), 2006.
2004; Barton et al., 1974]. Another widely used classification Sommer, R.: Kritische Wertung der Methoden der Gebirgsklassi-
system is the RMR system developed 1974 by Bieniawski and fizierung im Tunnelbau. Geotechnik in Forschung und Praxis,
updated in 1989 [Bieniawski, 1989]. The RMi system was WBI-PRINT 16, VGE: Essen, 2009.
developed more recently by Palmström [Palmström, 1995]. Sommer, R.: Entwurf von Tunnelbauwerken in klüftigem Fels auf
Grundlage eines felsmechanischen Modells oder auf Grundlage
von Gebirgskennziffern und Klassifizierungssystemen. Geotech-
4 CONCLUSIONS nik 33(2), 2010.
Wittke, W.: Rock mechanics – Theory and application with case
The design method based on rock mechanical models and histories. Springer: Berlin, Heidelberg, New York, Tokyo, 1990.
adequate analysis procedures has proven to lead to a safe and Wittke, W.: Stability analysis for tunnels – Fundamentals. Geotech-
nical Engineering in Research and Practice, WBI-PRINT 4,
economic design. When applying this procedure no rock mass Glückauf: Essen, 2000.
classification system is needed. Wittke, W.; Pierau, B.; Erichsen, C.: Statik und Kon-struktion der
The review of classification systems has shown that the Spritzbetonbauweise. Geotechnik in For-schung und Praxis. WBI-
correlation between rock mass rating index and required sup- PRINT 5, Glückauf: Essen, 2002 (free download of the English
port is not unambiguous, i.e. the same rock mass conditions version “New Austrian Tunneling Method (NATM), stability
may lead to distinctly different rating indices and vice versa. analysis and design” available at www.wbionline.de).
203

A study on mechanical behaviors of concrete lining and rock caused by shaft sinking
at the Horonobe underground research laboratory
K. Tsusaka & D. Inagaki

Japan Atomic Energy Agency, Hokkaido, Japan
M. Koike, Y. Ijiri & Y. Hatsuyama

Taisei Corporation, Tokyo, Japan
Keywords: tunnelling, rock support, field measurements, numerical modeling
Japan Atomic Energy Agency (JAEA) has been conducting an

underground research laboratory (URL) project at Horonobe
in order to establish reliable technologies related to geological
disposal of radioactive waste in sedimentary rock formations.
The Horonobe URL consists of a 4.5 m diameter ventilation
Shaft, two 6.5 m diameter access Shafts and experiment gal-
leries. At the end of 2010, the ventilation shaft and one of
the access shafts have reached to 250 m in depth. The authors
investigated circumferential stress distribution in a concrete
lining and displacement of rock around the shaft wall induced
by the shaft excavation. Field measurements were conducted
between 218 m and 220 m in depths. The lining stress distri-
bution was also analyzed using a three-dimensional numerical
model. The stress meters were installed at upper (S-1 & S-2),
middle (S-3) and lower positions (S-4 & S-5) in a 2 m high
Figure 1. Circumferential stresses in a concrete lining between
and 400 mm thick concrete lining in the eastern and southern 218 m and 220 m in depths (+: compressive, −: tensile). (a) Measured
sections, which correspond with the direction of horizon- result. (b) Analyzed result.
tal minimum and maximum initial stresses, respectively. The
measured and analyzed results in the southern section are
represented in Figure 1a and b. It was found that the maxi-
mum stress occurred at inner and lower position (S-5) in the
southern section. It was also clarified that the difference in
the stress between the maximum and minimum values was
beyond 10 MPa. The trends of stress change measured with
an excavation face advancing were analyzed by the three-
dimensional simulation. Displacements in rock around the
shaft wall after installation of a concrete lining were monitored
with multi-interval borehole extensometer arrays at 219.3 m
in depth. In the section, the mechanical anchors were set
0.5 m, 1 m, 1.5 m, 2.5 m, 4 m, and 6 m from the shaft wall.
It was found that compressive axial strain occurred around
the shaft wall after a concrete lining was casted since dis-
placement of rock in the vicinity of shaft wall was controlled
by the stiff lining. Compared with tunneling in rock with thin
layer of shotcrete, the rock behavior was a specific aspect of
the shaft excavation whose procedure is to cyclically install
a concrete lining every 2 m. The results of lining stress dis-
tribution in this study suggest that the authors should install
one of stress meters of a concrete lining at an inner and lower
position in the direction of the horizontal minimum initial
stress in order to estimate the stability of support system Figure 2. Observed result of axial strains in the rock around the
of a shaft. shaft at 219.3 m in depth.
204

Use of rock mass classifications for design: Recommendations and suggestions
C. Vibert
Coyne et Bellier-TE., Gennevilliers, France
P. Vaskou
Géostock, Rueil-Malmaison, France
Subject: Analyses techniques and design methods
Keywords: numerical modelling, rock mass classification, rock failure, rock support, stability analysis, tunnelling
1 EXTENDED ABSTRACT of the rock mass. The GSI method is now widely used, but
the complexity of the phenomena occurring in the deforma-
The rapid spread of the use of the geomechanical rock mass tion of a rock mass is such as, obviously, a single rate can
classifications at an international scale is a worldwide and not pretend represent the whole range of possible behaviours
irreversible trend driven by its applications in the profession. depending on the mineralogy, geology, tectonic features, etc.
In particular, rock mechanical engineers value these tools in Moreover, the behaviour is also variable with the type of the
assessing characteristics and failure criteria of rock masses. Works.
The implications on design are therefore obvious. Having witnessed tendencies to a blind application of the
Using case studies, the authors present an update on the ben- recommended methods, especially related to GSI, the authors
efits of the rock mass classifications actually used at an inter- have pointed out some of the most frequent misuses, which
national scale, namely the Q-system, RMR and GSI, although most of the time actually discarded the limitations or posterior
other classifications, such as the Chinese and Japanese one, notices from the originators of the rock mass classifications
exist. With regard to the limitations of these classifications, themselves. They recall that the method is to be used with
much has already been published; therefore attention of the judgement, and that, despite the comfortable assurance the
Reader is focused on the specific aspects where it is felt that strict application of formulas provides, it might have to be
strong recommendations are to be done for a correct, conve- adapted to fit the requirements of the Project, and that critical
nient and efficient utilisation, from on-site estimation of rock acumen shall apply in any case.
mass ratings to the use made of results for design purposes. Different examples from the experience of the two authors
First, a review is made of the positive aspects of the rock are listed to illustrate the limitations in the application of the
mass classifications, and principally: rock mass classifications.
• to impose a minimum description of the characteristics of
The consequences of the simplification of rock mass char-
acteristics to a single rate is a first one, since similar ratings
the rock masses, which have to be recorded,
• to allow, in similar contexts, direct comparison between
may be obtained for fractured hard rock on one side, and quasi-
homogeneous soft rock on the other side, while deformation
different rock masses, at various scales, between different
pattern at tunnel excavation is likely very different.
sites, or within a same project, and/or to monitor the varia-
The scale effect, when assessing ratings from a small pilot
tion in the geotechnical characteristics of the rock with the
gallery, and comparing it to ratings finally obtained once the
progress of the Works.
cavern is excavated, which demonstrates that the Q-index is
A review of the most frequent difficulties encountered when actually not an intrinsic characteristics of the massif, but more
assessing the rock mass classes is made, with insistence on the or less depends upon the relative size of the Work with regard
absolute need of a minimum of geological and geotechnical to the rock mass features.
skills for the one performing the assessment. Some important Improper use of serial correlations between Q-index, RMR
aspects may be neglected if the responsible staff in charge of and GSI is pointed out, too.
the Project does not have an experience in rock mechanics, The presence and frequency of specific zones – karsts,
such as occurrence of alterable or swelling material. Ability faults – requiring special treatment like grouting ahead of
to 3D visualisation is also required for a proper assessment. the face, can not be assessed from the only rock mass
Assessment of the different parameters used for ratings, classification.
such as RQD, Jn and SRF, also require caution and engineering The effect of saturation on the strength of the rock matrix
judgement. is to be checked in any case, and especially in hydropower
But the most sensitive aspect – the most interesting, too – projects, where the rock foundation is to be saturated. Some
is the possibility given to the Engineer to extrapolate the rocks exhibit a very significant drop in strength in presence
geotechnical characteristics of the rock matrix and disconti- of water, which cannot be assessed from the tests performed
nuities until obtaining a range of geotechnical characteristics on dry samples.
205

The selection of the rock matrix design parameters, espe- The second group of recommendations applies to the deter-
cially when using it for mi estimation in the GSI method for mination of design parameters for the rock mass, to be used as
determination of failure criterion of the rock mass shall: input data in software calculations, in particular when assess-
ing of the rock mass failure criterion by way of GSI. The
• be really representative of the rock mass to be assessed,
first to be done is to check if the method is applicable to the
since ignoring zones of weak strength, only because no
considered rock mass.
adequate samples could be taken from them, leads to
It is then to check if the design values retained for calcula-
overestimation of the rock mass strength,
tion of the required parameters are really representative of the
• include the necessary safety factor with regard to the
rock massif, such as if strength of the saturated rock matrix
possible mechanism of failure.
has been duly considered if the rock mass is to be submerged
Although obvious, it is always worth to recall that the rock by the Project. A necessary factor of safety is to be introduced
mass classifications poorly apply to anisotropic rock massif. at the stage of selection of a design value.
They are mostly designed for rock masses made of stiff blocks The range of ratings (and not only one rating) shall be
separated by discontinuities randomly oriented in space. One assessed, for the purpose of input data in calculations, as dry
example is provided, showing how use can be made of the and without in-situ stress effect.
GSI method to get an assessment of the shear strength of an In any way, since rock mass assessment does not apply
anisotropic massif when the failure mode is known. to every aspect of a rock massif, it is recommended to
Finally, the paper concludes with listing practical rec- describe each independent “geotechnical unit” according to
ommendations for the use of the here analysed rock mass checklist of rock mass properties, e.g. the one established by
classifications. the French Association of Tunnels and Underground Space
The first recommendations apply to general use of the clas- (AFTES).
sification, such as assessing the ratings on site, or designing The experience and engineering judgment shall finally
a tunnel support. prevail to validate the data.
Among others, is to perform both Q and RMR assessment As a conclusion, the opinion of the authors is that the
from the early stage of a project, to obtain a range of values actual rock mass classifications can be a valuable tool, if
for both parameters, that may eventually linked by a site- applied with discernment, critical acumen, good sense and
specific formula if required. The assessment is to be made honesty.
as much as possible by the same and experienced staff, for Nevertheless, even if those methods have shown their lim-
which a minimum of five years experience in this task should its, it remains that their immediate success proves that they
be required. answered an accurate need for Designers, allowing some kind
The geological mapping is to be made as the work progress, of “logical” jump from the measurable characteristics of rock
since they will provide valuable information that a simple matrix and discontinuities, to the characteristics of the rock
rating cannot produce, such as 3D vision of the massif, mass.
support of potentially unstable wedges, etc. The authors thereby suggest, with the help of numerical
The ratings shall be provided as a range of values; in any modelling, an improvement of the presently existing rock
case, avoid to calculate one rating, and deduce that the ade- mass classifications, which would better fit the expectancies
quate bolt length is the one mentioned on the charts; these of Designers. Charts of equivalent geotechnical characteris-
lengths shall remain indicative values. tics for cut slopes in an anisotropic rock mass, depending on
Wherever possible, the assessment shall be rectified the matrix properties, shear strength and spacing of the dom-
according to new information, and notably to feedback from inant family of joint, taking into account the orientation of
monitoring. slope face, are suggested as an example.
206

Analysis to the design factors of the diversion tunnel plug based on

additional rock elastic resistance
K. Wang, M.J. Zhao & X. Sun

River & Sea Department, Chongqing Jiao Tong University, Chongqing, China
Keywords: stability analysis, numerical modeling, rock stress
1 INTRODUCTION used more and more widely in pug design now, but the mechan-
ics effect of plug is still not studied systematically. Based on
The hydro diversion tunnel should block with a permanent the consideration mentioned above and combined with the
concrete plug before impoundment of the reservoir while diversion tunnel project, the mechanical characters of the plug
finished diversion. Although the amount of the project of are analyzed, and the optimization method of plug design has
hydraulic tunnel plug generally small, but as one of the main been studied in the theoretical meaning and practical level.
control conditions of hydropower engineering, its construc- (Wu, et al. 2006, Xu, et al. 2007).
tion are urgent. It has an extremely important significance for
the latter part of the project whether the blocking job to finish
promptly. At present, the Hydraulic Tunnel Design Specifica- 2 ANALYSIS OF THE MECHANICAL EFFECTS OF
tion (SL279–2002) gives only general principles for design HYDRAULIC TUNNEL PLUG
of the plug, the actual project has been formulated in their
design principles and selecting parameters, also does not have 2.1 Project profile
uniform standard. It is very important for the safe operation
of hub to block diversion tunnel, but design length of many Jin-Jia-Ba hydropower project diversion tunnel cross section
plugs tend to over-indulge as result of the irrational design and the concrete plug section are shown in Figure 1 and
principles and any other reasons (Zhao, 2004, Chen, 2007, Figure 2.
Su, 2008).
At present, the major method of sealing body design for
construction diversion tunnel are as follows: 2.2 Surrounding rock and plug three-dimensional finite
element numerical simulation
1) The length of the plug is determined according to the
multiples of the sealing diameter. In order to reduce the effect of boundary element analysis
2) The length of the plug is determined according to the model for calculating results, take cross-sectional area as:
experiential formula, the major formulas are as follows: 109 m × 111.5 m, which consider 50 m around the diversion
L = (3∼5)H /100 or L = mHD. tunnel cross-sectional area. The length of the model was
3) The length of the plug is determined by all kinds of for- 120 m, which take 30 m upstream and downstream of the plug.
mulas, such as cylindrical punching shearing formula and Finite element model shown as Figure 3.
against shearing formula. But still many factors of the The simulation of the support effect of bolts is considered
plug’s stability are not considered sufficiently. according to equivalent principle. The cohesive strength of
4) The interaction between plug and surrounding rock can be anchorage body after applying bolts is as follows (Li, S.C.
simulated with 3-D finite element method. et al. 2002):
Although, the development of finite element theory and
computing software is great, the 3-D finite element method is
Figure 1. Dimension of diversion tunnel cross-section (m). Figure 2. The dimension and shape of the plug section (m).
207

Figure 3. The FEM model of plug and surrounding rock mass.
Figure 4. The assumptions of mechanical state of the plug.
The simulation of the effect of support for steel is considered
with equivalent principle, make the steel arch frame convert of the concrete plug, the surrounding rock of seal section is
into concrete, the method is as follows (Li, J. et al. 2002): extruded laterally, so the stresses of surrounding rock are
reduced. The water pressure is borne by the friction of inter-
face and the cohesion between surrounding rock and plug,
while the friction is reduced from upstream to downstream.
2) Increasing the elastic modulus of surrounding rock is bene-
ficial to the sliding stability of the plug, so the consideration
In order to facilitating the analysis, a combination of the should be taken not only to the construction of the plug,
slice analysis and key point analysis method is adopted. but also to the avoidance of the tunnel with complex
cross-section and geological conditions.
3 ANALYSIS OF DIVERSION TUNNEL PLUG 3) Increase the strength of concrete can not increase the anti-
DESIGN FACTORS BASED ON ADDITIONAL sliding stability of plug, as the strength of concrete itself is
ROCK ELASTIC RESISTANCE strong enough.
4) The additional elastic resistance of surrounding rock is
The conventional design method of plug is based on the for- assured when the plug is designed as wedge shape with
mula of gravity dam stability analysis. But as traditional design large cross-section on upstream and small cross-section on
ideas, the beneficial effects of concrete plug caused by lateral downstream. But the angle of wedge should be appropriate,
constraint of surrounding rock, cannot be considered. In prac- because the angle is not the bigger, the better.
tice, the shape of pug generally was designed as “wedge body”.
The wedge plug press the surrounding rock under the action
of water in the upstream, but because of the concrete plug is ACKNOWLEDGMENTS
also non-rigid body, with the plug under axial compression,
and it will induce the radial expansion effect and press the Authors would like faithfully to acknowledge the Open-
surrounding rock. Therefore, the concrete plug will endure end fund of the Hydraulic Engineering Laboratory
additional elastic resistance caused by rock deformation (SLK2009B02), Chongqing Jiao tong University, and all
pressure. colleagues in the laboratory.
1) For the convenience of considering additional elastic resis-
tance of surrounding rock, to simplify the calculation REFERENCES
model, taking circular section tunnel, and this figure same
as the tunnel. Chen C.M., Liu B.X. & Zhang Y.J. 2007. Design and optimization
2) Assume upstream water pressure P is the only external load, of Shuibuya diversion tunnel plug. Yangzi River, Water Resources
and it distributed on the whole section of the plug. and Hydropower Engineering, 38(7): 44–45.
Hydraulic tunnel design specification, (SL279 – 2002).
The radius of upstream section is R0 , the radius of upstream Su, K. & Wu, H.G. 2008. Application of strength reduction FEM to
section is r0 , and the wedge angle is θ. stability analysis of diversion tunnel plug. Water Resources and
Figure 4 shows the mechanical state of the plug. Hydropower Engineering, 2008, 39(1): 43–46.
Wu H.G., Su K. & Chun G.K. 2006. Analysis of stability of plug
for diversion tunnels. Engineering Journal of Wuhan University,
4 CONCLUSIONS 39(3): 6–10.
Xu, G.G. & Wu, H.G. 2007. Optimization design of diversion tunnel
The major conclusions are as follows: plug based on FEM. China Rural Water and Hydropower, (3):
100–102.
1) After the operation of water-retaining with plug, the Zhao Q., He J.D. & Liang Z.J. 2004. Modification to conventional
surrounding rock of seal section is deformed reversely. design method of plug for diversion tunnel. Chinese Journal of
Because of the water pressure is borne by the middle part Rock Mechanics and Engineering, 23(8): 1336–1338.
208

The influence of bedding angle on failure mechanisms in sandstone
P.L.P. Wasantha, P.G. Ranjith, D.R. Viete, A. Haque, J. Kodikara & A. Bouazza
Department of Civil Engineering, Monash University, Melbourne, Victoria, Australia
Keywords: Lab testing, Mine design, Rock joints, Stability analysis, Rock failure
Civil and mining engineering projects that deal with rock loading direction) has a significant influence on failure mech-
are confronted with difficulties that arise due to the inherent anisms, with failure occurring by brittle ‘splitting’ for bedding
heterogeneous behavior of jointed rock masses. Prediction of angles between 0◦ and 30◦ , ‘shearing along bedding planes’
the mechanical behavior of jointed rock masses can be dif- for bedding angles between 45◦ –90◦ and by a combination
ficult due to uncertainties in the independent contribution of of these mechanisms for bedding angles between 30◦ –45◦
intact rock properties and discontinuity properties to the rock (Table 1). Strain concentrations on the surface of the samples
mass behavior. Currently, models such as the Hoek-Brown cri- (as shown by the ARAMIS camera results of Table 3) display
terion and the Mohr-Coulomb criterion are popularly used to distinct differences in the location and orientation of zones of
determine the strength of jointed rocks, but these treatments incipient failure for the various failure modes associated with
do not provide any indication of the expected failure mecha- different bedding orientations. Distinctions between the three
nisms for a jointed rock mass. A thorough understanding of failure modes can also be observed in the AE results of Table 1,
the influence of discontinuity properties on failure mecha- which show a sudden increase in AE counts for ‘splitting’
nisms remains elusive. This study addresses this shortcoming failure, a two-stage increase in AE counts for ‘mixed’ fail-
by focusing on the role that discontinuity geometry plays in ure (representing initial splitting and later failure by ‘shearing
governing failure mechanisms in natural rock. Uniaxial Com- along bedding planes’or vice versa) and intermittent increases
pressive Strength (UCS) testing was carried out on natural in AE counts (representing numerous discrete episodes of fail-
sandstone samples with various bedding orientations (Fig. 1). ure that contributed to the overall failure) for the ‘shearing
To better identify the operation of the various mechanisms along bedding planes’ failure mechanism.
available for failure, an optical deformation and strain measur-
ing (ARAMIS camera) and Acoustic Emission (AE) system
were employed for all the UCS tests (Fig. 2).
The results of the UCS tests revealed that bedding orien-
tation (as measured from the direction perpendicular to the
Figure 1. Typical samples with different bedding orientations (bed-

ding angles of 80◦ , 55◦ and 20◦ are shown from left to right, Figure 2. Typical testing setup with ARAMIS camera system, AE
respectively). system and strain gauges.
209

Table 1. Examples of typical failure patterns, and ARAMIS camera and AE results for the observed failure mechanisms.
Splitting Mixed Shearing along a defect
Failed specimens
ARAMIS camera
AE system
210

Estimating the porosity-depth relation of sedimentary rocks from

an effective stress/stress history-dependent porosity model
W.-J. Wu & J.-J. Dong

Graduate Institute of Applied Geology, National Central University, Taoyuan
J.-H. Wu
Graduate Institute of Geophysics, National Central University, Taoyuan
S.-S. Lin
Department of Harbor & River Engineering, National Taiwan Ocean University, Keelung
Keywords: lab testing, physical modelling, rock properties
1 INTRODUCTION 1111 m and 1707 m, respectively. In addition to continuously

sampling of rock cores, well logs in 0.15 m interval are also
The porosity-depth relation is useful in geological application available (Wu et al., 2008).
such as geothermal energy development, oil and gas extrac-
tion, nuclear waste disposal, geological sequestration of car-
bon dioxide and deep well injection of liquid or solid wastes. 3 THE POROSITY MEASUREMENT AND
In sediments compacted under self-weight, the porosity-depth STRESS-HISTORY DEPENDENT POROSITY MODEL
curve often show an exponential decrease called Athy’s law
(Athy, 1930): 3.1 Porosity measurement
A permeability/porosity measurement system, YOKO2, was
utilized for measuring the fluid flow properties of rock sam-
ples from TCDP Holes. The tests were performed using an
where φ is the porosity varied with depth, φo is the porosity intra-vessel oil pressure apparatus at room temperature. The
at atmospheric pressure, c is a constant and z is the burial samples, test procedure and experimental results are described
depth. This porosity- depth relation was widely used. How- in detail in previous research (Dong et al., 2010).
ever, the phenomenological porosity-depth relations failed to
explain the reason of their high variability, such as lithology 3.2 The stress history dependent porosity model
or different tectonic setting (e.g. Fowler et al., 1985).
In this study, we proposed a stress history dependent model The stress history dependent porosity model (Wu and Dong,
of porosity. Based on the model, the porosity-depth relation 2011) is utilized to evaluate the porosity-depth relation. When
of the sedimentary rocks can be evaluated quantitatively. To the in-situ effective stress is less than the maximum effective
validate the proposed method, the derived porosity-depth rela- stress, the porosity of sediments are expressed as the follows:
tion from porosity measurement of the rock cores of Taiwan
Chelungpu-fault Drilling Project (TCDP) was compared with
the density-derived porosity logging of the TCDP borehole
A. Based on the proposed model, the influence of the stress
history on the porosity-depth relation is elucidated.
where φpc and φo are the porosity under the maximum effective
2 TAIWAN CHELUNGPU-FAULT DRILLING PROJECT stress (σpc ) and the atmospheric pressure (σo ), respectively.
q and q are material constants. When the effective stress (σe ) is
The major scientific objective for the TCDP is to explore the equal to the maximum effective stress indicating the sediments
physical mechanisms involved in the large amount of fault are in virgin-compaction stage, the porosity can be expressed
slip during the 1999 Chi-Chi earthquake in Taiwan (Ma et al. as follows:
2006). Two vertical holes were drilled 40 m apart, hole A is
at a depth of 2 km and hole B is at a depth of 1.3 km. The
holes were drilled through the Chelungpu fault (CLPF) which
was ruptured during the Chi-Chi earthquake. Hole-A of the The parameters of the proposed model for the tested samples
TCDP penetrates the Chelungpu and Sanyi faults (SYF) at are listed in Table 1.
211

Table 1. Parameters of the tested rocks.
Run φo q q
Ss. R261-1 21.41 0.0457 0.0270

R261-2 26.26 0.0790 0.0436
R307 22.68 0.0533 0.0322
R160 25.24 0.0635 0.0403
R168 26.22 0.0644 0.0371
R773 26.83 0.0518 0.0244
R781 25.51 0.0713 0.0281
Ms. R255 14.86 0.0741 0.0165
R287 14.72 0.0606 0.0150
R316 14.54 0.0929 0.0222
R351 12.46 0.0617 0.0215
R390 15.93 0.0706 0.0241
R437 16.61 0.0333 0.0058
R693 17.74 0.0623 0.0233
Ss. = sandstones; Ms. = mudstone (shales and siltstones)
4 DETERMINING THE PARAMETERS OF THE

PROPOSED POROSITY MODEL
For evaluating the porosity-depth relation, the parameters of

fourteen rock samples (Table 1) are divided into two sets: (1)
sandstone and (2) mudstone. When the API value is more than
75, the averaged φo , q and q of mudstones samples were used
to calculate the porosity. While the averaged φo , q and q of Figure 1. (a) Density log-derived porosity-depth relation and the
sandstones were used to calculate the porosity. porosity-depth relation calculated from the proposed model. (b)
The maximum effective stress (σpc ) is estimated by in-situ Location of samples and lithology.
effective stress (σe ) plus eroded effective stress. The eroded
effective stress was estimated by thickness of eroded forma-
tion. The geological profile was used to evaluate the thickness
influence of the lithology and the in-situ effective stress on the
of eroded formation. The in-situ effective stress was calculated
porosity of sediments is well depicted. The estimated porosity-
from the density log.
depth relation is reasonably matched to the well log. Besides,
the effect of uplift on the porosity is identified.
5 RESULTS AND DISCUSSION
The modeling result is shown in Fig. 1. In the first column, REFERENCES

black curve is porosity derived from density log, blue curve is
Athy, L.F. 1930. Density, porosity and compaction of sedimentary
porosity derived from sonic log and red curve is the porosity
rocks. AAPG Bulletin 14(1): 1–24
evaluated by the proposed effective stress/stress history poros- Dong, J.-J., Hsu, J.-Y., Wu, W.-J., Shimamoto, T., Hung, J.-H., Yeh,
ity model. The porosity curves are plotted by data running in E.-C., Wu, Y.-H. & Sone, H. 2010. Stress-dependence of the per-
average per 3 m. meability and porosity of sandstone and shale from TCDP Hole-A.
Evidently, the lithology affects the porosity-depth relation. International Journal of Rock Mechanics & Mining Sciences 47:
In addition, the porosity decreased with increasing depth due 1141–1157.
to the effective stress induced compaction. Most interesting, Fowler, S.R., White, R.S. & Louden, K.E. 1985. Sediment dewatering
a larger porosity of the sediments below the Sanyi fault than in the Makran accretionary prism. Earth and Planetary Science
the one above it can be observed from the well log derived Letters 75(4): 427–438.
porosity-depth relation. The modeled porosity above the Sanyi Ma, K.-F., Tanaka, H., Song, S.-R., Wang, C.-Y., Hung, J.-H.,
Tsai, Y.-B., Mori, J., Song, Y.-F., Yeh, E.-C., Soh, W., Sone, H.,
reverse fault (10.0–17.1%), which experienced a deeper burial
Kuo, L.-W. & Wu, H.-Y. 2006. Slip zone and energetics of a large
depth than present, is about 2% smaller than those below the earthquake from the Taiwan Chelungpu-fault Drilling Project.
fault (12.6–18.6%). That is, the effect of the stress history on Nature 444: 473–476.
the porosity-depth relation is successfully modeled. Wu, W.-J. & Dong, J.-J. 2011. Determining the maximum overburden
in a sedimentary basin using a porosity versus effective confining
pressure curve. Tectonophysics.
6 CONCLUSION Wu, Y.-H., Yeh, E.-C., Dong, J.-J., Kuo, L.-W., Hsu, J.-Y. &
Hung, J.-H. 2008. Core-log integration in hole-A of Taiwan
We proposed a methodology to evaluate the porosity-depth Chelungpu-fault Drilling Project. Geophysical Journal Interna-
relation using a stress history dependent porosity model. The tional 174: 949–965.
212

Experimental study on constitutive models of rocks under loading

with active confining pressure
J.Y. Xu
Engineering College, Air Force Engineering University, Xi’an, China;
School of Mechanics and Civil Architecture, Northwest Polytechnic University, Xi’an, China
J.Z. Liu, XC.C. Lv, Z.D. Wang & E.L. Bai

Engineering College, Air Force Engineering University, Xi’an, China
Keywords: Rock properties; lab testing; physical modeling; rock stress
ABSTRACT: The research of constitutive model of rock- Through research of this paper, four aspects of conclusions
like materials is one of the most important aspects in the are demonstrated as follows:
rock mechanics study at all times. These models are essen-
1. By using the theory of new rock micro-unit strength and
tial for studying blasting mechanism, stress wave propagation,
Drucker-Prager failure criterion, a statistical damage con-
destruction effect analysis of underground structure, etc. At
stitutive model which satisfies weibull distribution was
present, constitutive model of rock-like materials in static
developed. The model can reflect confining pressure effect
uniaxial stress state has been investigated widely, and can
and variation law of rock strength with confining pres-
be proved by experiment accurately. However, as for compli-
sure. It can also simulate the deformation after pear point
cated stress state, especially static-dynamic coupling loading,
and residual strength of rock perfectly. Moreover, it reflects
the investigation of constitutive model is much more com-
that the damage of rock failure proceeding not only changes
plicated, and still at exploring stage. The development of
with rock micro-unit strength, but also depends on the stress
constitutive model and impact mechanical properties tests
state of rock.
of rock are closely linked. At present, Split Hopkinson
2. Through discussing the physical meaning of parameters
Pressure Bar and light air gun instrument were mainly used
in the constitutive damage model, it can be believed that
to study the uniaxial dynamic characteristic of rock. But there
F0 can judge the strength and m can judge the brittleness
were rare reports on the dynamic mechanical properties of
of rock, and damage correction coefficient q can reflect
rock-like materials under confining pressure. In fact, rock
the residual strength of rock. The relationship between
is under the condition of high static stress or ground stress
parameter F0 , q and strain-rate is approximately linear,
before bearing dynamic stress in engineering practice and
and parameter m obeys the quadratic multinomial function
nature. So it is meaningful to investigate constitutive model
change law with strain rate.
and dynamic mechanical properties of rock under loading with
3. Experimental results show that stress peak and integral
active confining pressure.
level of stress-strain curves of amphibolites under con-
Firstly, on the basis of damage mechanics and statistical
fining pressures were higher than uniaxial dynamic load.
theory, through combining visco-elastic model and statistic
Under the same confining pressure, the dynamic compres-
damage model, a new statistical damage constitutive model
sive strength of amphibolites are strain-rate dependent.
of rock based on weibull distribution under dynamic loading
The strengthening effect becomes more prominent with the
is developed. Then, in order to simulate real stress state of
increase of confining pressure at the same range of strain
rock in underground project, the traditional SHPB apparatus
rate.
was improved. The impact test of amphibolites with differ-
4. Stress-strain curves calculated from proposed model has a
ent confining pressure and strain rate was carried out by using
good agreement with that from experimental data directly,
improved 100 mm diameter SHPB. Thus parameters of consti-
which indicates the constitutive model is reliable. What’s
tutive model were obtained, and the influences of parameters to
more, the new model with less parameter is simple and easy
constitutive model were analyzed. In order to improve the pre-
for application, and it can offer reference for both further
cision, the statistical damage constitutive model of rock was
research of rock mechanical properties and engineering
modified reasonably, so as to set up dynamic damage constitu-
application of underground engineering.
tive model of rock which corresponds to reality better. Finally,
compared with SHPB experimental curve, the rationality of
constitutive model has been tested and verified.
213

Global stability analytical method of cavern group in salt rock based on energy criterion
Q. Yang, J.Q. Deng, Q.C. Lu & Y.R. Liu

State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Beijing, China
Keywords: numerical modeling; oil reservoir; rock caverns; rock failure; stability analysis
1 INTRODUCTION unbalanced force exists is unsteady potential failure zone. The

unbalanced force is the driving force of the time-dependent
Salt rock possesses good leak tightness, low permeability, deformation of storage cavern group.
property of damage self-healing, so deep underground salt This paper applies the deformation reinforcement the-
mine becomes the main media for energy reserves in countries ory to the global stability and failure evolution analysis of
of the world. Different from the thick salt dome abroad formed underground storage cavern group. The stability and failure
by marine deposit, the salt rock in China is the stratified struc- evolution law of single storage cavern, double storage cavern
ture formed by lacustrine deposit. It has three characteristics: and multi- storage cavern.
small thickness of salt layer, much insoluble intercalation,
shallow burial depth. When intensive underground storage 2 DEFORMATION REINFORCEMENT THEORY AND
cavern group is constructed in such salt rock, the global sta- THE KEY CRITERION OF QUANTIFYING THE
bility, the spatio-temporal development law of failure and the GLOBAL STABILITY OF CAVERN GROUP
corresponding cataclysm prevention theory of cavern group is
a focused problem. This is of great significance to the secure A typical incremental loading process is illustrated in Figure 1.
implementation of energy strategic reserve in our country. The initial stress state σ 0 is stable, i.e., f (σ0 ) ≤ 0.
When the FEM strength reduction method is used to ana- By imposing a strain increment ε on initial state, the
lyze stability, one of the key problems is how to determine elastic loading stress state is σ1 = σ0 + σ e where σ e = D:
that the rock mass reaches the ultimate failure state according ε. The strain increment ε results in plastic loading if
to the calculated results of FEM. In the stability analysis of f(σ 1 ) > 0. In order to satisfy yield condition, the stress has
slope, usually the plastic zone connection, the divergence of to be adjusted according to orthogonal flow rule. The final
FEM numerical calculation, mutation in strain and displace- stress state adjusted is σ = σ1 − σ p and f (σ) = 0. where
ment are used to judge whether the slope reach the ultimate σ p is plastic transfer stress. Yang et al. derived the analytic
failure state. But the study specifically aiming at the global solution of plastic transfer stress of elastic-perfectly plastic
stability and failure evolution of underground cavern group of material under Drucker-Prager yield criterion and computa-
energy reserve is few and there are quite a few problems to be tion in this paper uses this model. The plastic complementary
solved. First, quantitatively determining the global stability of energy norm of material is defined as
intensive cavern group is essential to the optimization of cav-
ern group arrangement and mode of operation. However, the
current conventional elastic-plasticity and rheological analy-
sis are still too difficult to give quantitative stability criteria.
Second, there is a wide range of plastic zone in cavern group on
the verge of failure, so it needs to seek more effective indicator
in plastic zone to better reflect the complex surge mechanism
of chain destruction of cavern group.
To solve the above problems, Yang et al. first applied
the deformation reinforcement theory to the global stabil-
ity and chain destruction analysis of underground cavern
group of energy reserve. On the basis of the concept of
constrained equilibrium state in non-equilibrium thermody-
namics, this paper develops and improves the elastic-plasticity
theory taking account of unsteady state, that is, deforma-
tion reinforcement theory, and redefines it as deformation
reinforcement theory.
Deformation reinforcement theory uses the relation curve
between plastic complementary energy and global strength
reduction factor, K∼E, as the global stability criterion of
cavern group. The plastic zone of cavern group in which Figure 1. Elasto-plastic stress adjustment.
214

Figure 2. Plastic complementary energy norm versus the strength Figure 3. (a) plastic zone, (b) unbalanced force.
reduction factor.
energy is analyzed to determine the position of failure and

Yang et al. have proved that the process of adjusting stress destabilization mode of cavern group.
according to orthogonal flow rule is the process of seek- Figure 3 indicates that unbalanced forces mainly exist in
ing plastic transfer stress to minimize plastic complementary the connected position of two caverns when plastic zone con-
energy norm i.e. minimum plastic complementary energy nects on a large scale. Unbalanced force is more effective than
principle. The equivalent nodal force corresponding to plas- plastic zone to illustrate the failure location of cavern group
tic transfer stress field σ p is the unbalanced force in FEM and the trend of deformation convergence and development.
analysis. Plastic zone connection does not illustrate that cavern group
produce overall failure and it is the necessary condition but
not sufficient condition of failure.
Wherein, subscript e indicates summation of all elements and 3 CONCLUSION

B is strain matrix. Obviously, plastic complementary energy
norm is also the norm of unbalanced force. Deformation stability theory expands the traditional methods
The stress field adjusted everywhere satisfies yield condi- of classical elastic-plastic theory that utilize stress, displace-
tion. But in some nodes where unbalanced forces exist the ment and plastic zone etc. as stability criterion, proposes a
equilibrium condition is unsatisfied and the stress field is more effective criterion of evaluating stability and failure and
unstable. To let the structure remain stable, reinforcement improves the accuracy and pertinence of analysis.
force which is equal in size and opposite in direction to unbal- The K∼E relation curve provides the uniform quantifi-
anced force has to be exerted. If reinforcement force not cation scale of a variety of factors causing failure. It can
exerted, instable elastic-plastic structures continue to deform quantitatively evaluate the influence to the global stability
under the drive of unbalanced force and tend to the state that of cavern group caused by factors such as cavern diameter,
self-bearing capacity is maximized and reinforced force is cavern spacing, weak intercalation, internal pressure, pressure
minimized. If structures are unable to eliminate unbalanced loss and cavern layout.
force by self-adjustment, it is indicated that structures lose To make single cavern remain stable, its diameter is not
stability. suitable to exceed 90 m and its height is not suitable to exceed
The relation curve between plastic complementary energy 150 m. The weak intercalation has adverse effect to the cav-
and strength reduction factor, K∼E, is used as the stabil- ern stability. The influence degree depends upon the number,
ity criterion of cavern group, and the strength parameters the strength parameter and the location of intercalation. The
of salt rock are reduced to simulate the progressive failure rational pillar width between caverns can be controlled from
process of cavern group, as shown in figure 2. From the per- 1.5 to 3.0 times the cavern diameter. The minimal stable inter-
spective of entirety, the variation of plastic complementary nal pressure of cavern group is 4 MPa. It is better to utilize
energy in the process of strength reduction is analyzed to rhombus layout when choosing the layout mode of cavern
judge if cavern group generate instability and failure, and then group. The pressure loss of storage cavern group of Dong-
from the perspective of locality, the variation of distribution gang well zone in Jintan salt mine is not probable to cause
of unbalanced force corresponding to plastic complementary chain destruction.
215

Analytical study on underground continuous impervious curtain for shaft lining

treatment at eastern Chinese coal mines
Q. Yu, H. Shimada, T. Sasaoka & K. Matsui

Department of Earth Resources Engineering, Kyushu University, Fukuoka, Japan
Keywords: Coal mines, numerical modelling, stability analysis
Serious shaft lining ruptures have often occurred in the east- The UCIC is constructed in the bottom aquifer, can interrupt
ern part of China, such as Datun, Xuzhou, Huaibei, Yanzhou, the connection with the seepage path, and eliminate the effect
Yingxia, Hebi, Dongrong, etc, since 1987. The geological of bottom aquifer drainage to shaft lining. Furthermore, the
conditions of all these spots are the almost same. All these UCIC has the strength to reinforce the aquifer around the shaft
shaft linings pass through deep topsoil of Quaternary strata lining, so that the effect of subsidence can be decreased and
for which the composition of the bottom aquifer is complex; the additional force can be exterminated from this point.
and before shaft lining rupture, the water head of the bot- Finite element software was used to simulate the rein-
tom aquifer has fallen dramatically and large scale subsidence forcement effect of the UCIC method. Phase2 7.0 is a 2D
occurred. During the water head falling, the soil layer applies elasto-plastic finite element analysis program for underground
additional force to the outer shaft lining. As the additional or surface excavations in rock or soil. An axisymmetric model
force is increasing with the effect of other joint forces, the was built for the simulation and the UCIC is set near the
condition of shaft lining goes into a plastic state gradually shaft lining. There are four stages in this model for simu-
and then can be ruptured. As the maximum additional stress lation. Stage 1: Initial Model; Stage 2: Formation of a shaft
occurs in the bottom aquifer and near the bedrock surface, (diameter 8 m); Stage 3: Formation of UCIC (thickness 1 m);
most of the ruptures have happened in this zone. Stage 4: Increasing pressure around the shaft. The location of
Based on an analysis of other methods and the mechanism of the UCIC is mostly built in the aquifer and in this analysis the
the shaft lining rupture, a new treatment method was proposed range and width of the UCIC are expanded for simulation.
for shaft lining reinforcement with underground continuous The widths of the UCIC were in three widths as 1 m, 2 m
impervious curtain. From the results of a series of numeri- and 3 m. From the results of various conditions, the most
cal analyses, the UCIC built in bottom aquifer can avoid the effective width of the UCIC is 1 m. When the UCIC was only
stress concentration in shaft lining. The current researches are built in aquifer or built in aquifer and 10 m above were more
mostly focused on the causes, mechanism and solutions for effective than built in other ranges.
these kinds of geotechnical issues in the engineering projects The changes of the safety factors show that the effect of
and they still have been the hot topics in recent 20 years in the UCIC depends on the UCIC materials.
the shaft construction. In this research, the various range and In this study, the failure criterion is Mohr-Coulomb fail-
width of UCIC built in aquifer were analyzed, and the effect ure criterion. The safety factor of the shaft lining was the
of the divers range was discussed, then the optimum condition evaluation criterion for the effect of the UCIC.
was determined. When the UCIC were built around the shaft lining, the safety
In this analysis, we choose Baodian coal mine as a typical factor in the boundary between upper layer and aquifer didn’t
example for the mines which located in the eastern China. change with the various conditions. The lowest safety factor
This mine belongs to Yanzhou Mining Group in Shandong was happened in the boundary between the aquifer and the bed
province, and is a concealed coalfield which the shafts pass rock. When the range and width were both larger, the safety
through the Quaternary stratum and Jurassic stratum. factor of the shaft lining had obvious reducing.
Underground continuous impervious curtain (UCIC) is an The most effective width of the UCIC was 1 m near the shaft
underground wall that has the anti-seepage capacity. This lining. When the UCIC was only built in aquifer or built in
method for shaft lining rupture treatment is based on the aquifer and 10 m above were more effective than built in other
mechanism of shaft lining rupture and borrows ideas from ranges. Material’s property played an important role in shaft
the existing treatment method such as strata grouting. The lining treatment using the UCIC.
UCIC is constructed outside of the shaft lining in a certain
area, adopting the vertical cutting mixing technique, mixing
cement grout with soil or concrete to be a wall.
216

Back-analysis of schistose rock mass strength parameters based on

displacement monitoring data
X.P. Zhang
Key Laboratory of Engineering Geomechanics, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
L.N.Y. Wong
S.J. Wang
Key Laboratory of Engineering Geomechanics, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
Keywords: back analysis; field measurements; monitoring; rock mass
Anisotropic and foliated rocks such as slates, schists and to provide data for predicting the behavior of rock masses
phyllites, the behavior of which is dominated by closely spaced after the excavation of the underground powerhouse and main
planes of weakness, cleavages or planes of schistosity, present transformer hall of the DanBa hydroelectrical power project,
particular difficulties in the determination of rock strength an exploration adit, which was 298 m long and 3 m diameter,
parameters. In choosing the strength values for anisotropic was first excavated for survey. Convergences after the exca-
rocks, one has to decide whether the highest, the average or the vation were measured, based on which a back-analysis was
lowest values obtained from testing should be used. Accord- performed to determine the strength parameters.
ing to our best knowledge, a widely adopted guidance on the To ensure the uniqueness of the back-analysis solution
choice of anisotropic rock mass parameters is unavailable. and to minimize the number of iterations of back-analysis
This paper presents a displacement based back-analysis required, the parameters for back-analysis were chosen based
method (flow chart as shown in figure 1) for the determina- on the following principles: (1) Select parameters that are of
tion of the elastic modulus (E), cohesive strength (C), friction greater influence on the deformation of underground exca-
angle (ϕ) and in-situ stress of a schistose rock mass. In order vations. (2) Select parameters that are very difficult to be
obtained accurately by other methods. (3) The number of
unknown parameters should be as small as possible. There
are a number of parameters affecting the rock mass defor-
mation which include elastic modulus (E), cohesive strength
(C), friction angle (ϕ), Poisson’s ratio (ν), tensile strength
(σt ), dilation angle (ψ) and in-situ stress. According to the
above-stated principles, E, C, ϕ and in-situ stress were cho-
sen for back-analysis in the present study. A preliminary
assessment revealed that the rock mass deformation was rel-
atively less sensitive to the variation of the Poisson’s ratio,
tensile strength and dilation angle. Constant single values of
ν = 0.25, σt = 2.7 Mpa, ψ = 10◦ were thus assigned based on
preliminary laboratory test results and recommendations in
the literature. In a 2D calculation model, there are three in situ
stress components σxx , σyy and σxy . In the present study, the
rock mass is assumed to be isotropic. Including E, C and ϕ,
a total of six parameters thus took part in the back-analysis.
The second step is to define a range for each of the six cho-
sen parameters to be used for back-analysis. The range of
each parameter was decided with reference to the preliminary
laboratory test results, which was then slightly broadened to
different extents: E from 1 GPa to 20.5 GPa, C from 1 Mpa
to 8 Mpa, ϕ from 24◦ to 49◦ , σxx from 8 MPa to 12 MPa,
σyy from 8 MPa to 12 MPa, and σxy from 0 MPa to 2.4 MPa.
Figure 1. Flow chart of back-analysis calculation. The designed range of each parameter is then respectively
217

subdivided into five equal intervals. As a result, a total of in-situ monitored deformation when relatively low C values
56 = 15625 times of calculations are needed to cater for all were adopted. Based on this finding, new sets of parame-
possible combinations of the six parameters. Such a large ters, which have narrower ranges of values were defined:
number of calculations would be heavily resource-demanding, C value from 0.5 MPa to 2.0 MPa, ϕ from 25◦ to 40◦ , σxx
which was not practical to implement. Orthogonal experi- from 9 MPa to 12 MPa, σyy from 9 MPa to 12 MPa, σxy
mental design (OED) method was hence used to design 25 from 0.6 MPa to 2.4 MPa, while keeping E = 10 GPa. Each
representative groups for back-analysis calculation. of these modified ranges of parameter was respectively sub-
Since convergence measurements were conducted after divided into four equal intervals. As a result, another set
excavation, the elastic deformation can be assumed to have of 16 representative groups were designed for back-analysis
completed before the monitoring equipment was in place. calculation. Out of the 16 calculated groups, the calculated
The monitored deformations were thus due primarily to the plastic deformations of one group fit very well with the
plastic deformation. In the back-analysis calculation, both in-situ monitoring deformation. The parameters of this par-
elastic model and elastic-plastic model were used. The plas- ticular group hence obtained were E = 10 GPa, C = 1 MPa,
tic deformation component can be obtained by subtracting ϕ = 40◦ , σxx = 11 Mpa, σyy = 9 MPa and σxy = 2.4 MPa. In
the elastic component from the elastic-plastic deformation, addition to the initially assumed parameters of ν = 0.25,
i.e. the net deformation is attributed to the plastic component. σt = 2.7 MPa, ψ = 10◦ , this set of parameters was used in
The back-analysis calculation of the 25 groups revealed that subsequent underground excavation and support design.
the calculated plastic deformation can become close to the
218

Different analytical models and methods to study wave propagation across rock joints
J.B. Zhu, W. Wu & J. Zhao

Laboratory for Rock Mechanics (LMR), Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Subject: Others – Rock dynamics
Keywords: Dynamics, blasting, rock joints, rock mass
Rock Masses are different from other engineering materials reflections are taken into account. Based on one-dimensional
by the existence of a great number of joints. They results in wave equation, relations between particle velocity and stress
the discontinuous characteristics of rock masses, and greatly along right- and left-running characteristics can be built.
affect the properties of rock masses. When a wave propa- Response at certain point and time can be derived from those
gates in jointed rock masses, it is both slowed and attenuated, on previous times.
mainly due to the existence of joints. The effects of rock With the RM, relations among different layers with respect
joints on wave propagation are of fundamental interest in geo- to potential amplitude or stresses and displacements can be
physics exploration, rock dynamics, earthquake seismology, established with a standard algorithm. The standard algorithm
and hydrocarbon production. is a process to take into account the effects of all the joints on
This paper conducted an overview of different models to seismic response at certain point in the form of a product of a
describe the physical properties of joints in wave propagation lot of functions or matrices.
and methods to take into account multiple wave reflections With the VWSM, virtual wave source exists at the joint
among joints. The joint models include the layered medium position and represents the mechanical properties of the joint.
model (LMM), the displacement discontinuity model (DDM), It produced reflected transmitted waves each time an incident
the wave scattering model (WSM) and the equivalent medium wave arrives at the joint. The overall reflected wave and trans-
model (EMM). Bases on these models, the effects of joints on mitted wave are the results of all the reflected waves and all
wave propagation can be studied. However, when multiple par- the transmitted waves arriving at different times, which are
allel joints exist, wave propagation is more complicated due created by VWSs.
to multiple wave reflections among joints. There are mainly The DDM can be regarded as an approximation of the
three methods to take into account multiple reflections among LMM or WSM. Parameters in DDM can be obtained from
joints, i.e., the method of characteristics (MC), the recursive those in the LMM or WSM. Similarly, parameters in the LMM
method (RM), and the virtual wave source method (VWSM). can be deduced from those in the WSM. Changes of equivalent
Comparison of these models and methods is also performed moduli of the EMM due to the presence of joint are function
to detect the relationship and difference among them. of parameters used in boundary equations of the LMM, DDM
With the LMM, both the stresses and displacements across or WSM.
the interface are continuous. There are two kinds of treat- The LMM can be used to study wave propagation across
ment of joints within the LMM. One is that the joint is treated filled joints or perfectly bonded joints. The DDM is applica-
as a perfectly bonded interface, and both the displacements ble to non-perfectly bonded joint with thickness much smaller
and stresses across the interface are continuous. The other is than wavelength. With the WSM, the effects of cracks in joints
that the joint is treated as a layer of the filled weak medium on wave propagation can be taken into account. The EMM
sandwiched between two fully-bonded interfaces. has advantages over other models in engineering estimate
The basic assumption of the DDM is that, as a wave prop- the overall influence of joints on wave transmission and
agates through a joint, the stresses are continuous, while the disadvantages of lose of joint discreteness.
displacements are not. The MC works only in time domain, while the RM and
The WSM treats the joint as a plane boundary with a dis- VWSM function in both time and frequency domains. The
tribution of small cracks and voids. The wave reflection and RM and VWSM can be adopted to study two dimensional
transmission across a joint is the results of wave scattering wave propagation across joints, while it is difficult for the MC.
through all cracks. Due to the discretization of time, the MC is semi-analytical.
The EMM treats problems from the viewpoint of entirety. The VWSM is also semi-analytical because of the prescript
From the EMM, the rock material and the contained joints of times of wave propagation among joints. The RM is purely
together are approximated by an equivalent continuous, analytical method without any approximations.
homogeneous and isotropic medium. The effect of joints is All models and methods have their own advantages, dis-
lumped into effective moduli of the equivalent medium. advantages and application areas. Depending on the problem
The MC is a mathematical tool to study wave propagation to be solved, specific model or method can be chosen and
across different layers or discontinuities, where multiple wave adopted.
219

Numerical modelling

Coupled flow-deformation simulation for mine scale analysis of

cave initiation and propagation
D.A. Beck
Beck Engineering Pty Ltd, Sydney, Australia
G. Putzar
Beck Engineering Pty Ltd, Berlin, Germany
Keywords: numerical modelling, monitoring, mine design, back analysis, stability analysis, rock mass classification
In most stress-deformation simulations of block or sub computes a new solution for stress, deformation, damage and
level cave induced rock mass damage and subsidence, the fault movement. As a consequence of the draw, an unstable
cave growth is approximated using the outputs of empiri- zone in the cave back – measured using nodal velocity – may
cal or stochastic tools that assume propagation without due thus develop at each coupling step and slough into the cave,
regard for the stress, strain, strength and structural conditions at which point these elements become available to be drawn.
required for cave propagation. The cave growth is usually The process is repeated following the draw schedule, and is
assumed to occur following simple rules related to draw, and able to simulate most cave propagation phenomena including
will propagate in the stress-deformation model whether cave stalling and chimneying.
back instability is indicated to occur or not. Example results are shown from an example mine where a
To overcome the problem and better estimate the likely comparison of the model results and field observations over
performance of caves, a tool that properly accounted for the 12 months showed a good correspondence between forecast
physical coupling of the cave material to the un-caved rock and actual induced seismicity, extraction level damage and
mass and the draw schedule, driven by the known physics of cave growth. A similar result has been achieved at several
both parts of the problem was needed. other mines.
After analysing a number of alternatives, a coupled Dis- It is concluded that the coupled DFE-LGCA simulation
continuum Finite Element (DFE) – Lattice Grain Cellular procedure enables rapid simulation of cave propagation, flow
Automata (LGCA) scheme was developed. In this scheme, and induced deformation driven by the cave draw schedule.
the LGCA part computes the velocities within the cave and The method can be calibrated directly using observations of
changes in airgap geometry due to an increment of draw. cave back location, grade and recovery, seismicity, tunnel
Particles are removed through drawpoints, element by element damage, tomography or ground movement. The close match
following the draw schedule. A relation between average par- between results of coupled flow deformation analysis and field
ticle flow velocity and modulus for cells within the cave is measurements suggests the technique is useful for forecasting
then used to imposed cave state changes node by node in the of cave induced deformation, and is especially useful for sim-
DFE model. These state (modulus) and geometry changes in ulating cave propagation and assessing risk related to the draw
the cave mass in the DFE model disturb the quasi-equilibrium schedule.
between the cave and the rock mass and the DFE model then
223

A numerical analysis of a four-way coal mine intersection with

primary and secondary supports
Y.P. Chugh & B. Abbasi

Mining and Mineral Resources Engineering
Southern Illinois University Carbondale, Illinois, USA
Keywords: mine design, numerical analysis, bolts and anchors, rock mass, stability analysis
Partial extraction room-and-pillar mining systems provide the value of the ratio of the horizontal stress to vertical stress
about 60% of the underground coal mined in the USA. This (M), width of opening to height of opening (Wo/Ho), and
mining system develops 3-way and 4-way intersections. About whether the immediate roof layers are bonded or un-bonded,
70% of these falls occur at intersections. Similarly, over 80% 2) The ratio of the Wo/Ho is most critical for intersection
of these falls occur at intersections in Illinois. Thus, there is a stability, 3) The peak value of the compressive stress due to
significant need to improve stability of intersections. The goal curvature is located about 0.2 m from the edge of the open-
here was to analyze stress distribution and instability around ing toward the center of the intersection, 4) The effect of
a 4-way intersection in a typical geologic environment in the de-bonding the layers is to increase the SCFs values and the
Interior Coal Basin, USA. An additional goal was to evaluate effect of increasing the M1 (M-value in the E-W direction)
primary and secondary support layouts for improving their is also to increase the HSCF and SSCF values, 5) The effect
stability. The results of numerical analyses of stress distribu- of de-bonding the layers is much more significant than the
tions and associated instabilities with and without primary and M1 value and explains why more failures are associated with
secondary supports installed are presented. thinly laminated strata in the immediate roof, 6) Increasing
Three-dimensional finite element analyses were performed the M1 values increases the peak VSCF only slightly but the
with estimated Hoek-Brown failure criterion parameters for HSCF and SSCF values are significantly increased, 7) Fail-
different lithologies associated with the coal seam. Vertical ures are likely to initiate around the edges of intersection
stress of 2.5 MPa (100 m mining depth) was applied at the top pillars, 8) Stability of an intersection would be significantly
of the model. Different lateral stress ratios were applied in improved if additional primary and secondary supports were
the E-W (M1 ) and N-S directions (M2 ), where M1 and M2 provided around the corners of an intersection to minimize
represent the ratios of lateral stress to vertical stress in the two failure initiation, 9) Increasing mining depth results in initi-
orientations. The immediate roof strata above the 2.0 m thick ation of shear and tension failure at intersection corners, 10)
coal seam are Black Shale (2.0 m), Gray Shale (0.6 m), weak Weak floor strata results in slight increase in the yielded zones
limestone (0.6 m), weak shale (0.9 m), competent limestone in the roof, and pillar ribs and significant yielding around
(1.2 m) and 6.7 m thick shale. The strata below the coal seam the floor corner, 11) Primary supports are not adequate to
consist of Gray Shale (1.0 m), weak limestone (1.0 m) and ensure stability of intersections, and 12) Fully-grouted bolts
10 m thick shale below the weak limestone. Models 1 to 6 with pre-tension, located around the corners of an intersec-
used above lithologies. In model 7, a 0.6 m weak claystone tion demonstrate the best potential to improve stability of an
layer below the coal seam was included. For unbonded layers intersection. A 5-bolt pattern around intersection corners, with
analysis, the cohesion between different layers was assumed 2.4 m long vertical, #7 rebar, pre-tensioned bolts appears best
to be zero. as secondary support. A coal company has agreed to demon-
The results indicate that 1) The peak value for each SCF is strate the effectiveness of proposed secondary supports and
located around the edge of the opening and its precise location planning is underway at this time.
shifts toward the opening or toward the pillar depending upon
224

3-D modeling and parameters research of irregular shape particles for DEM
based on computerized tomography
X. Du, Y.W. Zeng, R. Gao & J. Yan

School of Civil Engineering, Wuhan University, Wuhan, Hubei, China
Keywords: numerical modeling; physical modeling; rock properties
1 INTRODUCTION 2 THE COMPUTERIZED TOMOGRAPHY AND 3-D

COORDINATES INFORMATION OBTAINING
Granular materials have been widely used in various fields of
civil engineering, such as railway ballast; gravel embankment 2.1 The computerized tomography of real particles
of roads, et al. A notable characteristic of these materials is
In this paper, computerized tomography (CT) equipment was
irregular-shape, which creates a good deal of build-ins and
used to obtain railway ballast’s CT images, and the interval of
interlocks among particles and makes the feature of sliding
CT scanning is 0.6 mm. 157 CT images were obtained from
and friction more complex than the regular-shape particles.
sample ballast. The CT images were saved as BMP format and
In the granular materials research with Discrete Element
DICOM format.
Method (DEM), the common problems are shape and inertia’s
representation of real particles. So, modelling methods for
irregular-shape particles have been proposed constantly. 2.2 Processing of particle’s CT images
One method is combination algorithm and it represents In order to collect accurate 3-D coordinates information of a
real particles with a number of basic geometries. Currently, real particle, firstly, the image noise is eliminated by the binary
two categories of the combination have been developed and image processing with C language; then, based on the binary
adopted extensively in research because of the simplicity of image, the canny operator is applied to detect the edge points
ball’s contact simulating and the high efficiency of calculation. of ballast-shape boundaries, and the 3-D coordinates of bound-
One category of the combination methods is non- aries are collected; finally, by reading boundary coordinates in
overlapping modelling. For example, Wang et al. proposed to ANSYS and implementing the uniform discretization pro-
an advanced burn algorithm. However, the contacts between cess, spatial 3-D coordinates of this ballast are collected. In
balls need to be detected and calculated in this method, which this paper, the ballast is discreted into 59,014 surface points
causes an expensive cost of computer resources. and 73,330 inner points.
Another category is overlap-able algorithm. For instance,
Ferellec proposed an algorithm using inscribed balls following
a normal direction of surface to build irregular-shape particles. 3 THE ALGORITHM OF 3-D DEM PARTICLE
As the DEM analysis program ignores the internal contacts, MODELLING
which can save a lot of computation resources.
However, the overlap-able algorithm has two shortcomings. In our study, the algorithm adopts the combination manner of
One is the generated model relying on the initial conditions; overlap-able. The authors hope the algorithm could be simple,
another is that the model is not unique by executing the nonrandom and unrelated to initial conditions. The detail of
programme repeatedly. the algorithm is described below.
In order to optimize the model generation process, in this
paper, firstly, the 3-D coordinates of a real ballast particle are
3.1 The ideas and steps of modeling algorithm
collected by dealing with computerized tomography imag-
ing. Then, a computing programme is compiled basing on For every inner point, the distance between every surface point
the 3-D coordinates for the generation of particle model. and itself is calculated, and the smallest distance of the inner
The particle model is built by combining the basic-spheres point is recorded. Obviously, a group of B-S can be generated
(B-S) automatically under a preset three governing parameters by setting every inner point to be spherical center and the
algorithm. The program isn’t related to the initial conditions. corresponding recorded smallest distance to be radii. In this
And the model generation process can be repeated. Finally, way, the generated basic-spheres can characterize the shape
the parameters of governing are discussed for the purpose of of real particle.
modelling optimization, some suggestions and conclusions To reduce the B-S used in the model, a governing param-
are given. eter Rlim is introduced. The parameter Rlim controls the
225

minimum radii of B-S. If the B-S’s radius is less than Rlim , Above all, the DEM model of real particles can be accepted
the corresponding ball will be deleted. with the B-S number and accuracy of characterization by
Subsequently, the efficient space factor K is introduced. adjusting three governing parameters.
The mechanism of efficient space factor K is given by the
expression:
4 CONCLUSIONS AND DISCUSSIONS
4.1 Conclusions
where: djm is the distance between B-S j and m; Rj , Rm are In this paper, the 3-D coordinates of irregular shape parti-
the radii of B-S; min and max indicate the smaller one and the cle had been collected by dealing with CT images. A model
bigger one in this two B-S. generate program had been compiled based on the spatial
Finally, the split-radius Rd is introduced. The function of coordinates. The program built particle model by searching
Rd is dividing the B-S into two groups. The B-S which radii the minimum distance between the inner points and sur-
larger than Rd are given a relatively high value of K, and the face points. And a quantitatively evaluate method of model
B-S which radii smaller than Rd are given a relatively low value accuracy has be proposed.
of K. By this approach, the number of B-S will be reduced By the error analysis, some meaningful rules are found:
extra and the accuracy of characterization will be decreased (1) the function of Rlim error-reduction goes to strengthen
insignificantly. progressively, while the function of K goes to weaken (2) Rlim -
decrease should be the first choice for improving accuracy
and if the accuracy is not advanced obviously, the K-decrease
3.2 Model generation and parameters research
method should be executed. (3) The parameter Rd can miti-
Base on the software platform of Visual C Studio, the mod- gate the conflict of error and B-S number partly. Finally, the
elling programme is developed. The general programme is balance of B-S number and model errors had been searching
divided into four modules corresponding to the four steps by adjusting three governing parameters.
of modelling ideas. Moreover, particle models simulated in
different accuracy grade can be obtained by adjusting the
three governing parameters. 4.2 Discussions
The purpose of introducing parameters is reducing the num- Comparing with the method of inscribed sphere, the boundary
ber of B-S in models and the meantime guaranteeing the error in our study is circumscribed sphere error, and the two
accuracy of simulation.As the B-S number is positively related methods are similar. But in the sight of calculation speed, the
with the accuracy of model, reducing the number of B-S radius searching is more effective and faster than the inscribed
will cause the accuracy of model shape loss, adding the B-S sphere method.
number will cause computer resource waste. Therefore, three Three governing parameters are interrelated, and it makes
governing parameters are introduced for the aim of finding a the optimization between B-S number and accuracy becomes
suitable balance between B-S number and accuracy of model. more complex, which needs more deep study.
If the accuracy has been determined, the parameters should As the method introduced in this paper is overlap-able
maximize the reduction of B-S number. Then, the number of algorithm, the model’s density distribution is probably dif-
B-S and accuracy of model can be accepted simultaneously. ferent from the real particles. The enhancement of similar-
Comparing the B-S number and the errors of the models ity between numerical model and archetype needs further
which change with the values of parameters, it is found that research.
with the increasing of K and Rlim , the B-S number is reduced,
and this reducing has a high pertinence and sensibility with
K. The B-S number reduced with the increasing of K, how- AUTHORS
ever, the errors of model increased correspondingly, and vice
versa. The third parameter Rd is trying to mitigate the intense Du, X. School of Civil Engineering, Wuhan University, Wuhan,
effect caused by parameter K, moreover, it can mitigate the Hubei, 430072, China. duxin0218@126.com.
increasing of model errors which are caused by the reducing Zeng, Y. School of Civil Engineering, Wuhan University, Wuhan,
of B-S number. Hubei, 430072, China. zengyw@whu.edu.cn.
226

Three-dimensional thermomechanical modelling of backfilling

of old mining rooms in the Morsleben repository
S. Fahland, R. Eickemeier & St. Heusermann

Federal Institute for Geosciences and Natural Resources, Hannover, Germany
Keywords: nuclear repository, numerical modelling, stability analysis
1 EXTENDED ABSTRACT complex geological structure (Behlau & Mingerzahn 2001),

which is simplified with respect to homogenous rock layers
The Morsleben nuclear waste repository was used until 1998 with uniform material behaviour. These layers were especially
for the disposal of low- and medium-level radioactive wastes. classified with respect to the steady-state creep behaviour.
The repository was established in the old Bartensleben mine, The main units of the Zechstein strata (salt layers Z2SF,
a former salt and potash mine consisting of several mining Z3LSBKBD, Z3AMSS, and anhydrite layers Z3HA) were
parts, e.g. the southern, western, and eastern part located considered. The Hauptsalz Z2HS was separated into two
at the periphery of the mine and used for waste disposal several parts (Z2HSW and Z2HSO), due to different creep
(Heusermann et al. 2007). With respect to the stability of the behaviour. The structure of the overburden was idealised taken
mine and the integrity of the salt barrier, the central part of into account the main layers caprock CR, Jurassic-Cretaceous
the Bartensleben mine with the most considerable degree of and Quaternary J-Q (Fig. 2).
excavation and a complex geological structure is the most A three-dimensional finite-element model was discretised
important area from a geomechanical point of view (Fig. 1). taking the idealised structure of the geological layers and the
Due to the repository’s closure concept several large old geometry of the old mining rooms into account. Figure 3
mining rooms were selected to be backfilled in advance with shows a plot of the entire 3-D model, 835 m in height, 790 m
saltcrete in order to stabilise precautionarily the central part wide and 177.5 m in length, comprising 320,000 nodal points
of the mine and to improve the long-term integrity of the salt
barrier. During and after backfilling additional stresses which
are thermally induced by emitted heat of hydration occur in
the salt rock around the mining rooms.
To investigate the thermomechanical loading of rock caused
by generation of heat during hydration of saltcrete finite-
element calculations have been carried out. These thermome-
chanical calculations have been made to analyse the stability
of old mining rooms, e.g. pillars and roofs, and the integrity
of the salt barrier in the central part of the Bartensleben
salt mine. Basis of the modelling and the analysis is the
Figure 1. Mine situation in the Morsleben repository. Mapping and

scanning data by DBE company, model generated with ERAM-SIS
(Heller et al. 2004). Figure 2. Idealised geological structure with homogenous layers.
227

(Hunsche & Schulze 2003). Calculations were made using
the JIFE code and the INCA/PATRAN tool for pre- and post-
processing. It was assumed that the rooms were excavated in
the vears 1938 to 1943. After a time elapse of about 65 years
the successive backfilling of the rooms was simulated in the
model calculations. The behaviour of the saltcrete was taken
into account using a special hydration model. Thus, the ther-
mal impact of the saltcrete to the surrounding rock could be
investigated.
The calculated thermal und thermomechanical results of
calculations, e.g. temperature, stress, strain, displacement, and
dilatancy of the rock mass, are presented and discussed. These
results are basic values for the evaluation of the stability of the
repository mine and the long-term integrity of the geological
barrier.
REFERENCES
Behlau, J. & Mingerzahn, G. (2001): Geological and tectonic inves-

tigations in the former Morsleben salt mine (Germany) as a basis
for the safety assessment of a radioactive waste repository. (In:
LANGER, M. & TALBOT, C. J. (Eds.): Geosciences and Nuclear
Waste Disposal).– Engineering Geol., Spec. Issue, Vol. 61, Nos. 2–
3: 83–97, 4 Fig., 2 Tab.; Amsterdam (Elsevier) – ISSN 0013-7952,
2001.
Heller, M., Mauke, R., Mohlfeld, M. & Skrzyppek, J. (2004): ERAM
SIS: A spatial information system for visualization and manage-
ment of geotechnical data. Proc. Int. Conf. on Radioactive Waste
Disposal (DisTec2004), p.130–136, Berlin.
Heusermann, S., Nipp, H.-K., Eickemeier, R., Fahland, S. & Preuss, J.
(2007): Geomechanical integrity of waste disposal areas in the
Morsleben repository. REPOSAFE 2007 – Int. Conf. on Radioac-
tive Waste Disposal in Geological Formations, Nov. 6–9, 2007,
Braunschweig, Germany.
Hunsche, U. & Schulze, O. (2003): The dilatancy concept – a basis for
the modelling of coupled TMH processes in rock salt. European
Figure 3. Three-dimensional finite-element model of the central Commission CLUSTER Conference on the Impact of EDZ on
part. the Performance of Radioactive Waste Geological Repositories,
Nov. 3–5, Luxembourg, 2003.
and 300,800 isoparametric 8-node elements. The deforma-
tion behaviour of the ductile rock salt layers was described
by a constitutive equation including both elastic and steady-
state creep deformations. In addition, the dilatant behaviour
of rock salt was considered using a new dilatancy concept
228

Development of a non-contact survey method of tunnel excavation

face for DEM modelling
A.M. Ferrero, G.F. Forlani, R. Migliazza & R. Roncella

Department of Civil engineering, DICATeA, University of Parma, Italy
Subject: Site investigations and field observations
Keywords: field measurements, numerical modelling, monitoring
1 INTRODUCTION Apennines. The tunnel passes through low mountain sections,

passing reliefs with rather steep scope that are characterised
The quality of the design of tunnels is closely linked to the by the presence of Solignano and Cassio flyschoids.
degree of knowledge of the rock mass, which is based above all The mechanical characteristics of the materials were inves-
on the investigation of samples that are gradually enriched dur- tigated directly, through a series of dilatometric and pressio-
ing the excavation phases which allow access to the uncovered metric tests that were carried out in the survey boreholes
rock faces. The survey of discontinuities on the excavation and in laboratory tests, considering the knowledge of the
faces is therefore current practice in rock engineering and same lithotypes in analgous geotechnical-geomechanical
it allows the validity of the design, and more in general, the contexts.
design safety conditions to be assessed during the work. More- Survey of the discontinuities was conducted using the
over, the excavation face is very often a zone of high access Rockscan programme, which was developed by the Univer-
risk because of the disturbance caused by the excavation itself sity of Parma, and which allows the automatic identification
and because of the inevitable stress release that can be caused of the orientation and of the position of opportunely chosen
by more or less localised instability phenomena. The necessity discontinuity surfaces, on the basis of data obtained from a
of obtaining quantitative and reliable geostructural informa- photogrammetrical survey of the walls (Ferrero et al., 2009).
tion in safe conditions has led to the development of the The Rockscan programme makes it possible to choose one
research described in this work in which a survey procedure, of the digital photographs of the excavation face, taken during
based on photogrammetrical techniques, has been applied the photogrammetrical survey phases, and opportunely over-
to conduct a survey and has been utilised to set up a DEM lap it onto the point cloud that represents its geometry, in such
(Cundall, 1971, 1976) numerical model to analyse the stability a way that it is possible to visually recognise the discontinuities
conditions. that are present. In this way, it is possible to choose portions
of the rock mass of various extensions directly on the photo-
graph, inside which discontinuity surfaces that are emerging
2 DESCRIPTION OF THE WORK along the tunnel walls are highlighted.
It is therefore possible to identify a vast number of plains
The necessity of obtaining quantitative and reliable geostruc- in a very short period of time, which allows the structural
tural information in safe conditions has led to the development layout of the entire excavation wall to be studied (and not only
of the research described in this work in which a survey those parts that can be reached with traditional surveys), thus
procedure, based on photogrammetrical techniques, has been obtaining a statistically representative data sample.
applied to conduct a survey and has been utilised to set up a The geometrical characteristics (orientation and spacing) of
DEM numerical model to analyse the stability conditions. 135 plains emerging on the excavation face were determined
The Marta-Giulia tunnel, which is being constructed as part using the Rockscan programme. The plains were determined
of the Pontremolese Infrastructural Expansion in order to dou- by choosing and defining each single portion of the surfaces
ble the Solignano-Osteriazza (PR) section, was studied in this emerging on the excavation face and by choosing the frame
research. The tunnel has an overall length of about 4200 metres with the most suitable setting for each surface while taking
and a maximum overburden of about 275 m. The tunnel project care not to choose the same zone more than once, in order
was developed using the ADECO-RS method, on the basis of to not condition the statistical analysis of the distribution of
analyses and checks of the deformations of the rock mass the poles and therefore the identification of the discontinuity
involved in the excavation and of which detailed knowledge families.
is therefore already available. The excavation face has been
surveyed using the proposed method and the obtained data
have been compared and validated with data obtained utilis- 3 UDEC MODELLING
ing traditional methods. The validated surveys were then used
to set up the DEM method. On the basis of the described in situ investigations and labora-
The tunnel extends in an East-West direction along the mid tory tests a DEM model has been set up with the code UDEC
and lower Taro Valley on the Emiliano slope of the northern (Itasca, 2000) code.
229

Figure 2. Excavation face, manual selection of the areas coinciding
with the discontinuità plains – photograph taken centrally.
Overall, the model represents dimensions of 120 m × 120 m

and, in the centre, reproduces the real geometry of the exca-
vations, that is, a 10 m high and 12 m wide horseshoe shaped
tunnel.
An isotropic model with linear elastic behaviour charac-
terised by aYoung modulus equal to 5 GPa, a shear modulus G
equal to 2.9 GPa and a density r equal to 2500 kg/m3 was cho-
sen for the intact rock The discontinuities were characterised
by a perfectly plastic elastic behaviour (Mohr Coulomb); the
first was defined with normal and shear stiffness values equal
to 480 kN/mm and 240 kN/mm, respectively, while the second
was reached with a friction angle equal to 30◦ .
Figure 2 shows details of the model at the excavation Figure 5. Details of the model at the excavation boundary and the
boundary and the stiffness curve of the lining-convergence stiffness curve of the lining-convergence displacement obtained by
displacement obtained by means of the modelling. means of the modelling.
The displacement values obtained from the numerical
modelling were compared with those monitored during the Pinto; Project Engineer Ing. Saverio Lo Presti for his con-
excavation advancement phases. stant support in all working phases. Part of the work hereby
Comparing the values obtained by means of the UDEC presented was financed by the CRUI ICE Project N. 23 – Disc.
modelling with the horizontal and vertical ones of the tar- N. 13 and by the PRIN 2008 project, financed by the Italian
gets, it is possible to observe how the orders of magnitude Minister of the Instruction, University and Research and titled:
are the same. The maximum displacement revealed with the “Analysis of the stability of tunnel under seismic conditions.”
numerical simulation is 1.7 cm per node, while the mean value
falls perfectly within the values measured in situ.
REFERENCES
4 CONCLUDING REMARKS
Cundall, P.A. 1971. A computer model for simulating progressive
large scale movements in blocky rock systems. Proc. Symp. Int.
The Rockscan software is a valid method for the analysis of Soc. Rock Mechanics, Nancy, France, Vol. 1, no.II-8.
the structural layout of a rock mass and the advantages are Cundall, P.A. 1976. Explicit finite-difference methods in geomechan-
evident. When Rockscan is used, the photographic taking pro- ics. Numerical Methods in Geomechanics, C.S. Desai (Ed.).
cess (which is useful for the subsequent analysis) becomes a Ferrero A.M., Forlani G., Roncella R., Voyat H.I. 2008. Advanced geo
systematic part of the excavation process, while, on the other structural survey methods applied to rock mass characterization.
hand, the photographic taking should be designed so that it Rock Mec Rock Eng 4(2):631–665
interferes as little as possible with the excavation cycle. Itasca. 2000. User manual for UDEC, Version 3.1. Itasc Consulting
Group Inc., Minnesota, USA.
ACKNOWLEDGMENTS
Authors would like to thanks Italferr spa for the constant sup-
port and data and in particular: the Project Manager Ing. Giulio
Marcheselli; the Project Manager Assistant Ing. Giuseppe
230

Numerical simulation of crack propagation in rock by clumped particle model
T. Funatsu
Institute for Geo-Resources and Environment, National Institute of Advanced Industrial Science and Technology, Japan
N. Shimizu
Department of Civil and Environmental Engineering, Yamaguchi University, Japan
Keywords: numerical modelling, rock failure
In order to evaluate the application of particle flow code with 2. The fracture toughness obtained by the numerical modeling
clump model to crack problem on rock, we conduct the sim- is lower than that obtained by the experiments. However,
ulation of fracture toughness testing using SCB specimen, the difference is within 10%.
and conventional uniaxial compressive testing and Brazilian 3. The deformation behavior during fracture toughness testing
testing. The simulation result of uniaxial compressive test- is different from that of numerical modeling. This differ-
ing is shown in Figure 1. In addition, the Brazilian testing ence may be caused by the difference of particle of element
shows that the mechanical properties obtained by the simula- and mineral particle of natural rock. The further study is
tions are close to those obtained by the experiments. And then needed to solve the difference.
SCB specimen for evaluating fracture toughness is modeled
by using the input parameters collected above. The SCB model
is shown in Figure 2. And load versus displacement curve is
recorded shown in Figure 3. The clump model can reproduce
the value of fracture toughness as well as fracture process zone,
followed by the brittle failure. In conclusion, the particle flow
code with clump model is able to model the rock with strength,
elastic property, and crack propagation property. The main
results obtained from this work are summarized as follows:
1. The distinct element method can model Kimachi sandstone
by giving the appropriate micro-parameters. Especially,
matching the properties of compressive strength and ten-
sile strength at the same time is successfully conducted by
adopting clumped particle model.
Figure 2. Model of SCB specimen.
Figure 1. Comparison of stress-strain curves obtained by simulation Figure 3. Load versus displacement curve obtained by numerical
to experiment. simulation.
231

Stability modelling of isolated rock blocks at the surface of underground

excavations taking into account in-situ stresses
R. Ghazal, F. Hadj-Hassen & M. Tijani

Centre de Géosciences, Mines Paris-Tech, Fontainebleau, France
Keywords: numerical modelling, rock joints, stability analysis, rock stress, rock failure
1 EXTENDED ABSTRACT where σ is the stress variation vector on a considered

point of the joint and F0 and M0 are respectively the
When an underground cavity is excavated in a discontinuous forces and moments acting on the free face before it is
rock mass, blocks formed at the free surface may become released.
unstable and fall into the cavity. Predicting such occurrences Normal and shear stresses along the joints vary linearly with
is essential for the design of the excavation and the support. normal and shear displacements respectively:
Because of the complexity of the problem, simplified iso-
lated rock block methods have been adopted to model block
instabilities.
These methods consist of studying only the blocks located
at the surface of the excavation. Each block is treated inde- The block moves as a rigid body translating and rotating.
pendently of the others, considering the rock as infinitely rigid Any vector of displacement (u) can be related to a vector of
and the discontinuities as perfectly plane. The displacement displacement (U) of a reference point and a vector of rotation
is supposed to occur only along the joint faces. Despite these (W) of the block by:
simplifications available isolated rock block methods present
a lot of drawbacks. In fact, the use of equilibrium equations
to determine the forces acting on the block’s faces is not
sufficient. Successive assumptions should are usually made
to solve this indetermination. Finally, combining previous equations leads to solving a
This article presents a review on these methods. The linear system where the only unknowns are (U) and (W).
‘Keyblock’ method considers only simplified displacements:
falling, sliding on one face or sliding on two faces. Rotational
movements are generally not included since this requires a
very complex procedure. The mechanical analysis depends
strictly on the kinematical analysis, thus it is mainly the geom-
etry of the block that controls its stability. The joint normal
and shear behaviour are not included and in-situ stresses are Once the displacement at any point is assessed, its state of
ignored. Approaches that intended to include in-situ stresses stress is deduced and stability evaluated according to a Mohr-
used a continuum model hence ignoring the effect of joints Coulomb criterion. The method can also calculate a minimum
on the redistribution of stresses. ‘Relaxation’ methods added support pressure force required to ensure stability.
joint behaviour to the analysis but the cases studied were lim- Integrals are calculated by partitioning the blocks poly-
ited to simple 2D roof blocks. Finally, in all of these methods hedral faces into simple elements and using Gauss points.
the consideration of one force per face limits the possibilities Thus any block geometry formed of n polyhedral faces can be
of a rigorous modelling. studied. This new method is named ‘Isobloc’.
A new isolated rock block method is presented in this arti- Comparisons between different approaches show the advan-
cle. It analyses the equilibrium of the block during the process tages of the newly developed model. Blocks determined to
of excavation. Final stresses on the joint faces are determined be unstable using ‘Keyblock’ can be stabilized by in-situ
by considering joint behavior laws and a displacement of the stresses according to ‘Isobloc’. Besides, calculating stresses
block as a rigid body. Subtracting equilibrium equations of using a continuum model and integrating them in ‘Key-
forces and moments between the initial state before excava- block’ analysis can overestimate stability. Furthermore, it is
tion and the final state after excavation leads to the following shown that the block movement is more rigorously modelled
equations: using the new approach. Finally, the increase of the ratio of
normal to shear stiffness has a destabilizing effect on the
block.
232

Back analysis of tunnel response using the Differential Evolution Genetic Algorithm
(DEGA)
Marte Gutierrez
Division of Engineering, Colorado School of Mines, Golden, CO, USA
Sotirios Vardakos
Parsons Brinckerhoff Geotechnical and Tunneling, New York, NY, USA
Caichu Xia
Department of Geotechnical Engineering, Tongji University, Shanghai, PRC
Keywords: back-analysis, field measurements, monitoring, numerical methods, tunneling
1 INTRODUCTION an optimization problem. The main elements of a genetic algo-

rithm can be summarized as follows: 1) The procedure starts
Field monitoring can provide the means to verify predicted from a population of trial vectors, instead of a single point. For
response, and to check the validity of values of material prop- an n-variable problem, the usual population size is 2n to 4n.
erties and loads used in design of tunnels.The determination of For a highly non-linear objective function, higher sizes may
parameters required in computational models using monitored be required. 2) There is no gradient or pattern search direc-
data is referred to as back-analysis.. Back-analysis is partic- tion exploitation. 3) Each variable resembles a chromosome
ularly suited to underground constructions such as tunneling. in genetics. 4) The objective function value is the equivalent of
Underground constructions provide unique situations where fitness in genetics. In minimization problems, a vector corre-
more information about ground conditions and response sponding to a very low function value is a strong candidate and
become available as the construction progresses. A crucial may survive in the future. 5) The process is based on repeated
component of systematic back-analysis procedures is an algo- generations of new population vectors. The trial vectors of
rithm to find a set of input parameters that will minimize the these populations are the result of randomized parental selec-
difference between predicted and measured performance (e.g., tion, crossover and mutation processes. They are essentially
in terms of deformations, stresses or tunnel support loads). the offspring of previous trial vectors.
Methods of back-analysis can be broadly classified as direct DEGA combines the concepts of Differential Evolution and
and gradient-based optimization techniques. Genetic Algorithms to handle optimization of non-binary val-
An alternative to direct and gradient-based methods is the ued nonlinear functions. DEGA uses two arrays to store a
use of heuristic algorithms, which are problem-solving tech- population of NP, D-dimensional vectors of parameters that
niques in which the most appropriate solution of several found are being back-calculated (D = number of parameters = n).
by alternative methods is selected at successive iterations for The parameter vector may consist of heterogeneous data from
use in the next step of the iteration. The present paper focuses different types of monitoring (e.g., displacements, strain, loads
on the application of Differential Evolution GeneticAlgorithm on structures, pore pressures, etc.) The first array contains the
(DEGA) in modeling of tunneling using a commercially avail- primary values, which contains the present vector population,
able computer code. DEGA is a heuristics-based global search and the secondary array stores sequentially the products for
algorithm belonging to a wide class of GeneticAlgorithms that the next generation. The algorithm starts by filling the primary
have been used in a wide range of optimization problems in a array with NP vectors with randomly generated parameter
variety of applications. The use of DEGA in back-analysis of values. The initial random generation should satisfy the con-
tunnel response is analyzed in terms of the stability and effi- straints on the parameters. The primary array is also called as
ciency under highly non-linear circumstances, the sensitivity the “trial vector” since it contains NP vectors that will later
of the solution to the initial trial assumption and the sensitivity be tried for fitness.
of the solution to the monitoring data. The application of the Each of these individual randomly generated vectors Xi is
proposed back-analysis procedure is demonstrated using the considered sequentially for genetic operations. For each of the
Heshang Highway Tunnel project in China. chosen vectors, three other vectors XA , XB , XC are randomly
chosen from the remaining vectors of the primary array. To
2 USE OF DEGA IN BACK-ANALYSIS generate a new trial vector, a mutation is performed by using
OF TUNNEL RESPONSE the following relation:
DEGA belongs to a family of global optimization procedures

that aims to find globally optimum solution (if there is one) to
233

Figure 1. Tunnel cross-sections and construction stages at Station
K6+300 of Heshang Tunnel.
where Xm 1 is the new mutant vector, and F is a scaling factor

in the range 0 < F ≤ 1.2. According to Price and Storn (1997),
the optimum value of F is in the range 0.4 to 1. A small modi-
fication of DEGA is applied in this study whereby the mutant
vector X1m is checked for constraint violation. Even though
the initially generated XA , XB , XC vectors are feasible, their
linear combination may be violating the constraints. If any
parameter constraint is violated, then the sampled vectors XA ,
XB , XC are discarded and a new sampling is performed until
a feasible vector X1m is found. This intervention is seamlessly
integrated in the algorithm and also keeps the true heuristic
process of the problem intact.
3 APPLICATION OF DEGA TO THE BACK-ANALYSIS

OF HESHANG TUNNEL
To illustrate its applicability to real tunneling problems, the

DEGA-based back-analysis procedure was applied to the
analysis of the response during construction of the Heshang
Highway Tunnel. This tunnel is located in South East China
at the Fujian province, about 5 km southeast of the Fuzhou
City, and is a part of the transportation system between the Figure 2. Predicted extensometer displacements above the right
tunnel from forward modeling.
local airport and Fuzhou city. The tunnel construction was
completed on August 30, 2006. It is a twin tunnel project and
approximately 450 m long. A typical tunnel cross-section and elastic properties E and ν, cohesion c and friction angle φ ,
construction stages are shown in Fig. 1. Each of the two tun- and the average unit weight of the rock γ. The back-analysis
nels is 11.5 m high and 15 m wide. The tunnel passes through is performed using monitoring data obtained from Stages 2
highly weathered volcanic material. and 4 in the excavation of the left tunnel. These include exten-
Selected sections of the tunnel were fully instrumented someter displacements measured at KO1 and KO2. It is noted
using extensometer, internal convergence and surface dis- that Stages 3 (left tunnel top core) and 4 (right tunnel top
placements. The goal of the back-analysis is to determine heading) had already been completed before the second set of
model parameters using monitoring data from the earlier exca- measurements was obtained.
vation stages and use the parameters for forward modeling Figure 2 shows a comparison for the forward modeling part
of subsequent excavations. Thus, material parameters deter- between monitored extensometer data and predicted response
mined from the back-analysis of the left tunnel, which was at the end of Stage 4 of the right tunnel. Due to excavation
constructed first, were then used in the prediction (or forward of the right tunnel, single top heading large displacements
modeling) of the response of the right tunnel. The combined tend to develop around the right tunnel and all extensometers
back- and forward modeling was performed for all the stages are now registering detectable displacements. There is rela-
of tunnel construction at Station K6+300. It is noted that based tively good agreement for extensometer KO5, but generally
on the available information, no lateral differences in rock the back-analysis resulted in a slight overestimation of the
types were encountered in the left and right tunnels at Station displacements developed at KO3 and KO4, possibly due to
K6+300. Thus, material parameters back-calculated from the local variations in the quality of the rock mass, which were not
left tunnel should be directly applicable for the right tunnel. explicitly modeled here. The differences between predicted
The ground is modeled by an elasto-plastic Mohr-Coulomb and measured response are in the range of 1 cm in most cases.
model. The pre-reinforcement was simulated using finite The relatively good agreement between predicted and moni-
difference zones, instead of using structural elements. The tored response indicates the validity of the model parameters
unknown properties that were back-analyzed are the rock mass obtained from the back-analysis of the tunnel response at the
earlier stages of construction.
234

The assessment of the hydro-jacking risk of the pressurized water tunnel by theoretical and
numerical method
T.W. Ha & Y.W. Shin

Hakyong Engineering Co. LTD, Seoul, Korea
H.G. Lee
Doosan Engineering & Construction, Seoul, Korea
H.S. Yang
Dept. of Energy & Resources Engineering, Jeonnam Nat’l Univ. Gwangju, Korea
Keywords: case studies, fluid flow, hydraulic fractures, risks and hazards, stability analysis, pressurized water tunnel
Different from the common tunnel such as road/railroad tun- occurrence of hydro-jacking in the condition of before and
nel, a pressurized water tunnel is known to be exposed to water after constructing emergency spillway tunnel, as well. Safety
pressure applying on an inside of tunnel. The internal water factor by Norwegian theory for the diversion tunnel at present
pressure acts on the concrete liner of tunnel, and it causes state was evaluated as 2.4 which were evaluated over allowable
damage in the liner as well as leakage into the surrounding safety factor of 1.3. The minimum safety factor evaluated by
rock mass. Apart from damages in the liner, the surround- numerical analysis is 1.47 which calculated from proportion
ing rock mass also can be damaged from crack initiation and between pore pressure with minimum principal stress at the
propagation by water leakage. Therefore, a careful analysis on current state. An also, the safety factors calculated by theo-
the hydraulic damage should be carried out when the stabil- retical and numerical method at the state after constructing
ity of internal water tunnel is needed to be concerned against
“hydro-jacking” which status that pre-existing joint is opened
and grown by leakage is predicted when stability analysis of
the water tunnel is carried.
In this study, hydro-jacking related analysis has been per-
formed on the stability of old drainage tunnel at dam site with
theoretical and numerical methods. It was predicted from the
stability analysis that hydro-jacking can be occurred at the
existing pressurized tunnel when new drainage tunnel is con-
structed nearby. On close examination it was found that the
safety factor of the old drainage tunnel was higher than a
permissible range.
The Norwegian criterion among theoretical method was
adopted for evaluating hydro-jacking hazard of existing
drainage tunnel. The numerical method was carried to predict
Figure 1. Schematic diagram of Norwegian Criterion. Figure 2. Present and planned state of construction site.
235

emergency spillway tunnel were 1.52 and 18.83 which were
over the allowable safety factor, repectively.
As a result of that, safety factor of the existing drainage
tunnel against hydro-jacking was calculated higher than allow-
able value by both of theoretical and numerical method.
Consequently, old drainage tunnel was estimated stable for
the hydro-jacking not only at the current state but after new
drainage tunnel construction.
Figure 3. Safety factor distributions after constructing emergency

spillway tunnel.
236

Dynamic analysis of rock tunnel failure by using 3-D Numerical Manifold Method
L. He & X. Huang
Department of Structures and Mechanics, Nanyang Technological University, Singapore
G.W. Ma
Department of Civil and Resource Engineering, The University of Western Australia, Australia
Keywords: Rock dynamics; Numerical modeling; Rock tunnel; Rock failure; Key block; Stability analysis; NMM
A number of methods are available to analyze tunnels, from NMM simplifies the modeling process by avoiding the tedious
the earliest closed-form analytical solutions to the most meshing operation. The numerical description on the irregu-
recent numerical modeling methods. Normally, we have two lar joint network in rock engineering is easily achieved. In
branches of approaches to rigorously analyze tunneling prob- this paper, weak form of governing equation, NMM interpo-
lem (Barla & Barla 2000): Equivalent continuum approach lations, and discrete equations are introduced for the dynamic
and Discontinuum approach. The continuum assumption has analysis. Rock falling from the side and roof of a horseshoe-
obvious disadvantages since the rock mass failure is largely shaped tunnel is studied using the proposed 3-D NMM, setting
dominated by the joint structure. Strong tectonic discontinu- as in Figure 1, and it is an important task to find and describe
ities cut the surrounding rock blocks of a tunnel surface into the most critical key block around an excavation in such
discrete state.An equivalent continuum cannot reflect the main engineering problem.
aspects of such blocky system. It might be improper to ana- The calculation consists of two stages. At first, the fric-
lyze rock tunnel with only emphasizing the deformation or tion angle is set to 59 degree. The designed system reaches
strain based on the continuum assumption. Block deformation a steady state with proper systematic damping. Secondly,
could not substitute the displacement in the block system com- through reducing the friction gradually, it could reach the case
pletely. The enforcement scheme design, based on continuous until one of key blocks starts to detach out from surrounding.
assumption, might not be safe especially in 3-D situation. Such Gradually reduce the value of friction angle, there are two
averaging design of bolt and lining might not be sufficient for
locally loose blocks.
In general sense, the NMM can be thought as a bridge
connecting two branches of numerical method: continuum
method (e.g. FEM) and discontinuum method (e.g. DDA). The
achievements on “The description of irregular joint network”
in rock mass and “merging FEM and DDA” as a combined
method, are through the concept of “Dual-Cover system” in
this method. Different with traditional finite element method,
it divides the solution domain into two domains: mathematical
and physical domain. More specifically, it builds interpo-
lation function in the mathematical domain,meanwhile the
functional integration is done in the physical domain. The sep-
aration also frees the shape of cover-based element. Therefore, Figure 2. Blocks with diverse shapes in tunnel.
Figure 1. Scenario setting for a rock tunnel.
237

Figure 3. Close-up view of the key-blocks and failure of key block located at the side wall.
Figure 4. Dynamic failure processes of key-blocks in a tunnel.
Figure 5. Displacement history of key blocks.
potential falling blocks, indexed with Blk #11 (Critical fric- moving downward. From the curvature change along the
tion angle for the roof key block is 11.1070◦ ) and Blk #12 curve, we can also identify that block #11 will be completely
(Critical frictionangle for the sidewall key block is 17.6450◦ ) detached from surrounding at time 0.776 s. It later falls to
(see Fig. 3). the bottom of the tunnel at 1.172 s. Similarly, side block #12
Fig. 4 gives rock tunnel failure process with a frictional is detached at 0.709 s, and falls to the bottom at 1.605 s.
angle of 8◦ , which means that both two key blocks in this This paper strongly provides evidence for 3-D NMM anal-
tunnel model will be detached from the surrounding blocks ysis. The proposed program can simulate the whole process of
and fall down to the bottom. failure, and it can also get more detailed information, despite
The maximum collapsing volume is 0.148 m3 under the the critical friction angle (including collapse time, collapse
setting condition, which will not collapse as a whole. Enforce- mode, volume velocity, etc.). These provide the fundament
ment scheme design should focus on how to prevent small for forward and backward analysis to engineering scenario.
block from falling. Generally speaking, lining and rock bolt It also demonstrates the potential ability to analyze the interac-
should be enough for stabilization. Fig. 5 plots the displace- tive contact interaction among complex blocks in the platform
ment history and indicated that the position of them is mainly of 3-D NMM.
238

Thermomechanical modelling of the Gorleben exploration site to analyse

the integrity of the salt barrier
St. Heusermann, P. Vogel, R. Eickemeier & H.-K. Nipp

Federal Institute for Geosciences and Natural Resources, Hannover, Germany
1 EXTENDED ABSTRACT
Geoscientific exploration work at the Gorleben salt dome has

been underway since 1979 to investigate the site for its suit-
ability as a final repository for high-level radioactive wastes
(BMWI, 2008). Underground exploration at the site started
in 1983, mine excavation work began with the sinking of
two shafts from 1986 to 1994 and the excavation of drifts
of the first exploration area EB1 within the following years.
The exploration work continued until 2000 when a ten-year
moratorium was imposed. Actually, the restart of the explo-
ration is planned for October 2010. The Federal Office for
Radiation Protection (BfS) is responsible for the planning,
exploration, construction, and operation of the Gorleben site.
The Federal Institute for Geosciences and Natural Resources
(BGR) deals with the primary geoscientific questions, e.g. Figure 1. Geological cross section through the Gorleben salt dome.
geology, geophysics, geotechnics, and safety analyses.
During the past decades, BGR has been carrying out exten-
sive geoscientific research and practical project work on
the Gorleben site. A highly important item is the numerical
analysis of the Gorleben salt dome taking into account the
geological structure of the rock as well as appropriate consti-
tutive material models to describe the time-, temperature- and
stress-dependent deformation and strength behaviour of the
rock salt. Model calculations have been performed on the one
hand to analyse the mechanical behaviour of mine components
like shafts and drifts. On the other hand, thermally induced
stresses and strains have been predicted on a large scale. These
stresses and strains will be caused over large time periods by
the disposal of heat-generating high-level wastes and are basic
values to evaluate the thermomechanical loading of the rock
as well as the mechanical integrity of the salt barrier. To this Figure 2. Far-field modelling of the Gorleben salt dome (cut-out of
end, BGR has been developing several finite-element codes the model without bedrock and overburden).
and tools adapted to the special requirements of rock salt:
The well-proven ANSALT code and the new JIFE code for The paper describes two different numerical models of the
thermomechanical modelling, as well as the INCA/PATRAN Gorleben salt dome. The first model deals with the regional
tool for pre- and postprocessing, data handling and storage. far-field situation covering an investigation area of 30 × 30 km
Figure 1 shows a vertical cross section through the Gorleben (Fig. 2). This three-dimensional model incorporates the com-
site describing the geology of the salt dome and the overburden plete Gorleben-Rambow salt dome which is discretised in a
(Bornemann et al., 2008). The central section of the salt dome more general way with respect to the geological structure.
consists of Zechstein strata Hauptsalz z2HS. Other main units Referring to the geological situation depicted in Figure 1, the
of the salt dome are the salt layers of the Leine-Folge (z3) and second model describes a characteristic vertical cross section
the Aller-Folge (z4). of the salt dome and considers the main geological layers of the
To establish a geomechanical model for subsequent numer- salt structure and overburden in a more detailed way (Fig. 3).
ical calculations, the model of the geological structure had to With respect to the German borehole disposal concept it was
be idealised by classifying geological layers with respect to assumed that the waste canisters will be emplaced in vertical
the steady-state creep behaviour. boreholes at a depth of 870 to 1170 m.
239

The model calculations were performed for a time elapse
of 10,000 years. The obtained results, e.g. temperatures at the
top of the salt dome, stresses and strains in the salt rock and in
the overburden, dilatancy of the rock salt, displacement of the
rock, and uplift of the ground surface, are presented and dis-
cussed. Calculated stresses are the basic values to evaluate the
long-term integrity of the salt barrier from a rock-mechanical
point of view. To this aim, two safety criteria, the dilatancy
criterion and the frac criterion, are considered (Heusermann
et al., 2009).
As an example, Figure 4 shows the calculated temperature
distribution in the salt dome for a time elapse of 127.5 years
after emplacement of wastes. Maximum temperature values
of up to 175◦ C arise in the disposal area. At the top of the
salt dome maximum values of 40◦ C are predicted for a time
Figure 3. Modelling of a characteristic cross section of the Gorleben period of 5,500 years. Figure 5 depicts the predicted deviatoric
salt dome with detailed geological structure. stresses in the rock salt and the anhydrite layers after a time
period of 10,000 years. The calculated stresses are basis to
evaluate the long-term mechanical integrity of the salt barrier.
REFERENCES
BMWI (2008): Final disposal of high-level radioactive waste in

Germany – The Gorleben repository project. Federal Ministry
of Economics and Technology (BMWI), Public Relations/IA8,
Berlin, Germany.
Bornemann, O., Behlau, J., Fischbeck, R., Hammer, J., Jaritz, W.,
Keller, S., Mingerzahn, G. & Schramm, M. (2010): Description of
the Gorleben site – Part 3: Results of the geological exploration of
the salt structure. Geol. Almanac, Series C, 73, 212 p., Hannover,
Germany.
Heusermann, St., Fahland, S. & Eickemeier, R. (2009): Geomechan-
ical stability and integrity of radioactive waste repositories in salt
rock. SINOROCK2009 – Proc. International Symposium on Rock
Mechanics: Rock Characterisation, Modelling and Engineering
Figure 4. Predicted temperatures in the salt rock after a time elapse Design Methods, May 19–22, 2009, Hongkong, China.
of 127.5 years.
Figure 5. Predicted deviatoric stresses in the rock salt and anhydrite

layers after a time period of 10,000 years.
240

Response of nuclear waste repository host rock to thermo-mechanical loads
H. Hökmark & M. Lönnqvist

Clay Technology AB, IDEON Research Centre, Lund, Sweden
Keywords: Numerical modeling, nuclear repository, rock stress, site characterization, stability analysis
1 INTRODUCTION
Within the one million year timeframe established by the reg-

ulators for the safety assessment (SKB 2011), the repository
host rock will be subject to a number of mechanical, thermo-
mechanical and hydro-mechanical loads. Here, examples of
the calculations, conducted within the SR-Site assessment
to estimate changes in host rock transport properties during
construction and the heated period following deposition and
closure, are given.
2 THERMAL EVOLUTION
The thermal evolution of the repository host rock depends on

the heat output of the individual canisters, on the rock heat
transport properties and on the deposition geometry, i.e. on
the layout. The spacing between canisters at the Forsmark site
is established by SKB (2009) using the strategy devised by
Hökmark et al. (2009), and the thermal property model given
by Back et al. (2007). The complete site description is given
by SKB (2008).
2.1 Forsmark host rock thermal properties Figure 1. Contours of temperature increase in the plane of the
repository at four instances of time. Positions of vertical scan-lines
The thermal property model accounts for the distribution as are indicated in the upper right.
well as for the spatial variability of the heat transport prop-
erties. The effective equivalent global thermal conductivity
corresponding to the conditions in the hottest positions at the distribution of thermal properties are not important on
the time of the peak is called the dimensioning conductivity the scale of the large scale model mesh resolution, mean-
(Hökmark et al. 2010). Some distance away from the canisters ing that mean values of the thermal properties are valid
the mean conductivity is an adequate approximation. approximations. In high resolution near-field models, rock
wall temperatures are monitored at canister mid-height. For
the dimensioning case, the peak buffer temperature reaches
2.2 Calculations the 100◦ C design limit, less a 5◦ C margin specified in the
All numerical analyses of the thermo-mechanical evolution dimensioning rules to account for uncertainties, some 10 years
related to the safety assessment are performed using the dis- after deposition. The corresponding maximum rock wall peak
tinct element code 3DEC (Itasca 2007). The 3DEC thermal temperature amounts to about 66◦ C.
logic is based on analytical solutions and can be used to
analyze temperature fields generated by thousands of time-
dependent heat sources on all scales in time and space. 3 MECHANICAL EVOLUTION
Variations in heat transport properties cannot, however, be
taken into account. Here, the mean thermal conductivity is The thermally-induced stresses are calculated on three model
used for large-scale analyses, whereas small-scale models are scales: The large, the medium and the small scale. All analyses
analyzed for two different values: the mean conductivity and are performed using the 3DEC code.
the dimensioning conductivity. The large scale models include the entire repository. The
Figure 1 shows the large-scale thermal evolution at 460 m horizontal cross section area is about 60 km2 and the model
depth calculated using the 3DEC thermal logic. The details of height about 3 km.
241

4 TRANSMISSIVITY IMPACT
The thermally-induced stress changes will impact on the frac-

ture transmissivity on all scales. On the large scale, increases in
fracture transmissivity are found on near-surface vertical frac-
tures striking along the major horizontal stress above heated
regions and on horizontal fractures between heated regions at
repository depth.
5 DISCUSSION
The following conclusions are judged to be particularly

relevant:
– Peak temperatures will amount to about 40◦ C as a deposi-
tion area average and to about 65o C at maximum.
– Thermally induced stresses have a modest impact on frac-
ture transmissivities: a factor of about 2.5 at maximum,
even assuming a worst-case sensitivity model. At most
locations, transmissivities are decreased.
– The majority of the deposition holes will be stable after
Figure 2. Scanline stresses from large scale model. excavation. After some time of heating, however, the nom-
inal spalling strength will be exceeded in significant wall
sections of most deposition holes.
The medium scale models include five deposition tun-
nels and have dimensions corresponding to 165 canister
positions. Time-dependant mechanical boundary conditions, REFERENCES
corresponding to the thermally-induced expansion and fol-
lowing contraction of the box-shaped model volume, are Back, P.-E., Wrafter, J., Sundberg, J. & Rosén, L. 2007. Thermal
determined from displacement histories recorded at relevant properties. Site descriptive modelling Forsmark stage 2.2. SKB
positions of the large scale model. R-07-47, Stockholm: Svensk Kärnbränslehantering AB.
The small-scale models include one deposition tunnel with Hökmark, H., Lönnqvist, M. & Fälth, B. 2010. THM-issues in
seven canister positions. The main purpose of analyzing repository host rock. Thermal, thermo-mechanical and hydro-
mechanical evolution of the rock at the Forsmark and Laxemar
small-scale models is to assess the potential for stress- sites. SKB TR-10-23, Stockholm: Svensk Kärnbränslehantering
induced brittle failure (spalling) in the walls of the deposition AB.
holes. Again, time-dependent mechanical boundary condi- Hökmark, H., Lönnqvist M., Kristensson O., Sundberg J. &
tions are obtained from the large scale model results.A number Hellström, G. 2009. Strategy for thermal dimensioning of the
of models are run with in situ stress data covering the stress final repository for spent nuclear fuel. SKB R-09-04, Stockholm:
uncertainty ranges (magnitudes and directions) given in the Svensk Kärnbränslehantering AB.
site report. Itasca. 2007. 3DEC – 3-dimensional Distinct Element Code. Ver.
Figure 2 shows the total stresses on scanelines I and II. Note 4.10 users guide. Minneapolis: Itasca Consulting Group Inc.
that the coarsely meshed large scale model cannot capture SKB. 2008. Site description of Forsmark at completion of the
stress maxima and local stress gradients. site investigation phase. SDM-Site Forsmark. SKB TR-08-05,
Stockholm: Svensk Kärnbränslehantering AB.
According to the stress results of the small model analyses, SKB. 2009. Site engineering report Forsmark. Guidelines for under-
spalling is unlikely during construction for the most typical ground design Step D2. SKB R-08-83, Stockholm: Svensk
in situ stress case, but likely to be initiated in the upper part Kärnbränslehantering AB.
of the deposition hole within the first year after deposition. SKB. 2011. Long-term safety for the final repository for spent
During the following years the spalled zone will grow slowly nuclear fuel at Forsmark. Main Report of the SR-Site project.
downwards and eventually reach close to the bottom of the SKB TR-11-01, Stockholm: Svensk Kärnbränslehantering AB.
deposition hole.
242

Mixed complementary formulation of discontinuous deformation analysis and solution
W. Jiang
Key Laboratory of Geological Hazards on Three Gorges Reservoir Area, China Three Gorges University,
Ministry of Education, Yi Chang, Hubei, China
H. Zheng
Chinese Academy of Sciences, Wuhan, Hubei, China
Keywords: numerical modeling, rock failure, stability analysis, dynamics
1 INTRODUCTION (stiffness parameters). Too small factors lead to undue devia-

tion from the contact conditions; while too large factors result
The object studied in the DDA is usually a system of blocks, in an ill-conditioned stiffness matrix which incurs numerical
and the treatment of contact between blocks is a major task in problems. The Lagrange multiplier method enforces the con-
the DDA. A contact problem can be reduced to an optimization tact conditions, but might fail because of rank deficiency of
problem (or a variational problem) with inequality constraints. the stiffness matrix if some inactive constraints are imposed
These constraints are classified into active ones expressed by during an open-close iteration. The Augmented Lagrange
equalities and inactive ones by strict inequalities. Only those Multiplier method can alleviate difficulties of the DDA in
active constraints can affect the solution of the problem. In selecting penalty factors and overcome rank deficiency of the
general, however, in advance we by no means know which stiffness matrix in the LDDA. But still the ALDDA adopts
constraints are active. A process called the open-close iteration the open-close iteration whose drawbacks have been stated in
is thus utilized to pick out those active constraints. the above.
In the open-close iteration, firstly the contact modes for the To avoid the penalty factors and the open-close iteration, we
contact-pairs are assumed. Here, a contact-pair is composed reformulate the DDA as a mixed complementarity problem,
of an angle and an edge belonging to two blocks respectively and design a solution procedure for the problem.
that might be in contact at the end of a time step. Then, the
potential function of the block system is minimized subject
to a set of equality constraints corresponding to the assumed 2 SYSTEM OF EQUATIONS OF CONSERVATION
contact modes. Last, the contact modes are corrected for those OF MOMENTUM
contact-pairs that violate the contact criteria. The three steps
are repeated until no contact-pair violates the contact criteria. Using the weak form of the law of conservation of momen-
Obviously, the open-close iteration belongs to the state itera- tum and the Euler backward integration over time, we have a
tion and falls into the category of the test-and-error schemes. system of linear equations, denoted by system-M
Simple and intuitive as it is, the scheme cannot assure the
iteration is always convergent. During the iteration the change
in the contact mode of any contact-pair will cause a discon- whose basic unknowns include contact force vector p as well
tinuous change in the stiffness matrix, and lead accordingly as block generalized displacement vector d, with K = stiffness
to abrupt changes in degrees of freedom in two consecu- matrix and C = contact matrix.
tive iterations. Furthermore, in considering block nonlinearity The reason that the weak form of the law of conservation of
such as block nonlinear constitution or crack propagation in momentum is selected as the start instead of the principle of
blocks, an iteration process for the block nonlinearity must be minimum potential functional is to facilitate the consideration
involved. In this case, the block nonlinearity iteration and the of block nonlinearity in future. If block nonlinearity is taken
open-close iteration are interwoven together and a “nested” into account, the principle of minimum potential function will
structure of programs based on the two kinds of iterations not hold any more.
must be produced. The DDA has a procedure for detecting potential contact-
If the active constraints are supposed, the penalty function pairs at the beginning of a time step. These contact-pairs might
method and the Lagrange multiplier method or its vari- vary in the contact iteration. If no new contact-pair enters in
ants are generally utilized to solve contact problems. In the the time step, a frequently-encountered situation in the analy-
penalty function method, adopted in the original DDA and sis, equation (1) will keep invariant as well. In the conventional
its improved version, small penetration between blocks is DDA, however, the case is quite different: the stiffness matrix
allowed, and even necessary. Although the penetration is inap- is always changing because of the modification of the con-
preciable, it is not easy to select reasonable penalty factors tact states of contact-pairs, causing the abrupt change in the
243

generalized displacements in two consecutive iterations. This Considering this fact, we use τr and ġ τ pτ , instead of |ġ τ |,
explains why it is not easy to assure the open-close iteration to form the complementarity
is always convergent.
3 SYSTEM OF EQUATIONS OF CONTACT

which is the tangential contact equation.
Compared to the conventional DDA, more unknowns, i.e., the Gathering together the normal contact equation (4) and
contact forces, are included in the new formulation. Hence the tangential one of all contact-pairs, we have the system
equation (1) is not enough to solve for p and d. We will have of contact equations denoted by system-C,
to resort to other equations from the contact conditions.
To avoid being entangled in the open-close iteration, we
introduce a C-function, the Fisher-Burmeister function, as
follows, Here, ψ is a vector-valued function with ψ in equations (4)
and (5) being its component functions.
4 SOLUTION PROCEDURE
with the set of its zeros being the two nonnegative semi axes of
the planar system. That means that ψ(a, b) = 0 if and only if the To this point, we can combine system-M with system-C and
following complementarity condition on two real numbers, a obtain a system of nonlinear equations, called system-MC,
and b, is satisfied
ψ(a, b) given in equation (2) is continuous and smooth every

where function H is continuous and piecewise smooth.
where except at the origin (0, 0).
Without directly solving system-MC, we minimize the merit
For any contact-pair, the normal displacement g n and the
function associated with function H
normal contact force pn has a complementary relationship.
So, we have the normal contact equation
through modifying the Fisher-Burmeister Line Search Algo-

In the tangential direction, the situation is a bit more com- rithm (FBLSA) designed for the nonlinear complementary
plicated because there are two potential sliding directions problems. Here,
opposite to each other.
It is true that the sliding speed ratio |ġ τ |, and the resid-
ual shear τr ≡ C + µpn − |pτ | form a complementarity rela-
tion, with C = cohesive force, µ = frictional factor, and
pτ = frictional force. However, such a complementarity alone The improved FBLSA has very excellent numerical properties
is not sufficient to reflect the fact that ġ τ has the same sign because the worst situation is taken into account during the
as pτ . contact iteration.
244

X-ray CT technology-based finite element simulation of coal samples

under uniaxial compression
Y.D. Jiang
China University of Mining & Technology, Beijing, China
State Key Laboratory of Coal Resources and Safe Mining (CUMTB), Beijing, China
W.J. Wang
Y.X. Zhao
State Key Laboratory of Coal Resources and Safe Mining (CUMTB), Beijing, China
L. Zhang
Bluocean Technology Inspection Co., Ltd, Tianjin, China
J.L. Han
Keywords: lab testing, numerical modeling, rock failure
ABSTRACT: There are many researches on studying char- corresponding to different substances. Hounsfield units were
acteristics of coal and rock by using X-ray Computed Tomo- converted into by using experience formula and density values
graphy (CT) technology and finite element method (FEM). were assigned to the corresponding volume mesh. Then elastic
However, there are only a few papers to combine the two modulus and poisson coefficient were defined for every sub-
methods together to study this kind of issues. Combining stance. Finally volume mesh which was assigned the material
X-ray CT and FEM together is a new analysis method. We would be output to ABAQUS files, as inp file.
can get the physics model of coal and rock based on real The commercial FEA suite ABAQUS was used to simulate
material with three-dimensional reconstruction of CT images. the process of uniaxial compression of coal sample. Input the
Finite numerical modeling of materials can directly predict its inp file which was converted from MIMICS to create one
behavior under different loading conditions based on simula- model with meshing situation of 270460 tetrahedron units.
tion rather than indirectly prediction by measured parameters. The loading path was 0-5MP-10MP-15MP-20MP.
Therefore, in this paper we present an example to combine From the simulation results which were shown in figure 1
these two methods to study the X-ray CT technology-based and 2, we can observe the development of maximum princi-
finite element simulation of coal samples under uniaxial pal stress in the vertical section of model during the loading
compression. process and get the following points:
In order to study the effect of compressive loads on coal
1) Stress was firstly produced at some points of inner model;
in finite analyze software based on real materials, we used
2) Secondly, with increasing loading force, high stress points
X-ray microCT scanner to scan the original sample. Then, CT
gradually formed an area;
images of coal sample was import and threshold.The coal sam-
3) Thirdly, high stress area would continue to expand to both
ple was divided into two types of material based on different
ends of the model, which mainly expanded outward along
gray value range, namely coal matrix and mineral impurities.
the interface between two kinds of materials;
The 3D model was established according to the scan data and
4) Finally, stress increased more quickly in high stress area,
threshold classification. After 3d reconstruction of coal sam-
distributed along with 45◦ direction.
ple, in order to facilitate the follow-up finite element analysis,
the meshing on the surface and its optimization and remeshing The variation of maximum principal stress in the middle
was conducted in MIMICS FEA module, and then generated cross-section of model was similar as it in vertical section
the volume mesh. After loading volume mesh data, gray value in the early step, namely stress was firstly produced at some
of hounsfield units was calculated for each mesh unit based points of inner model, but with increasing loading force, high
on scan images data in FEA module. Then hounsfield units stress points gradually formed an area from the outside of
in volume mesh were divided into equal areas, and each area model to the interior, as shown in Figure 2.
245

Figure 1. Distribution of the maximum principal stress in the vertical section of model during different analytical steps (unit: Pa).
Figure 2. Distribution of the maximum principal stress in the middle cross-section of model during different analytical steps (unit: Pa).
The micro FEA simulation agrees well with the experiments of the effect of coal under uniaxial compression load. This
and theory, which present the point that the interface between method can be more specific responses of physical properties
two materials will more likely get stress concentration result of small specimen. It has certain significance to study the
in finally fracture. The methods described in this paper can be basic mechanical properties of rock and failure mechanism,
applied to a variety of coal samples in an attempt to predict but it is limited to apply this method for large area (such as
their destruction areas, and to gain a deeper understanding coal seam) to study its stress and strain.
246

Pumping simulations using 3D analysis on multi-pumping wells
T. Kitaoka
Graduate School of Science and Engineering, Kansai University, Osaka, Japan
H. Kusumi & M. Nakamura

Faculty of Environmental and Urban Engineering, Kansai University, Osaka, Japan
Keywords: field measurements, fluid flow, numerical modelling
1 GENERAL INSTRUCTIONS the fluctuation of water level in the shallow monitoring wells.
The amount of the rainfall greatly influences the fluctuation
In the management and a usage of a groundwater resource, of water level in the shallow monitoring wells, and the water
we need to work on an accurate grasp of the current situation level tends to rise in summer when the amount of the rainfall
and to make a future vision in the region where the ground- is large, and to decrease in winter when the amount of the
water exploitation is active. In the past research, a common rainfall is a little. We used monitoring data when we do the
aquifer was examined in one municipality. This simulation analysis. Fig. 3 shows the analysis result. As a result, it can be
was undertaken where the scope was expanded, the subject of seen from this that the fluctuation of water level group wells
which was a pumping well for water service – water levels cause accuracy by using our model made in the present study.
for one pump were large. To use the groundwater resource in If the pumping well is newly established, using this model
this region effectively, we examine the influence on periph- based on the result of measurement of the water level for 20
eral groundwater properties especially where a large amount years or more, we want to forecast its future influence on the
of pumping from the wells for waterworks is being practiced. water level. We also want to use it as a model that can propose
an appropriate groundwater control for the long-lasting use of
ground water in this region, and for the prevention of troubles
2 CONCLUSIONS concerning the wells, subsidence, and uplifts.
Fig. 1 shows the 3D mesh model. In the examination, we con-
struct the 3D mesh model in detail, analyze the seepage flow
by the finite element method, and compare the change of the
groundwater level that the group of pumping wells caused with
the actual measurement values. Fig. 2 shows the relation of
Figure 2. The amount of the rainfall and groundwater level of

shallow monitoring.
Figure 1. The 3D mesh model. Figure 3. The analysis result.
247

Abutment stability assessment of the Hongrin arch dam using 3D distinct element method
A. Koliji
Stucky Ltd, Switzerland
Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland
T. Bussard
Norbert SA, engineering geology and hydrogeology, Switzerland
A. Wohnlich
Stucky Ltd, Switzerland
J. Zhao
Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland
Keywords: rock joints, stability analysis, rock support, numerical modeling, monitoring
The Hongrin north dam is a double curvature concrete arch role of rock discontinuity in the failure mechanisms. The rock
dam located in western Swiss Prealps, which attains 125 m mass is defined as deformable distinct blocks which inter-
high. The right bank abutment of the dam mainly consists of act along frictional discrete discontinuities representing the
intensively jointed Neocomian limestone and exhibits zones of rock joint sets. The water pressure is introduced as fluid pres-
potential instability. At the time of construction (1965–1969), sure boundary condition along the discontinuities, and the
this slope was reinforced with rock anchors. Subsequent rock reinforcement is modeled as structural elements working
hydrogeological study and groundwater monitoring revealed across the discontinuities. The dam reaction forces, derived
the presence of water pressure due to a slight seepage flow from a separate finite element analysis, are evaluated for their
through the rock joints in the dam foundation. This lat- possible effects on the stability. The model examines the sen-
ter evidence raised an additional concern about the stability sitivity of the abutment stability to the presence of joint water
of the abutment. In a dedicated study, the stability of the pressure and evaluates the improving effects of rock reinforce-
right abutment the Hongrin north dam abutment has been ment. The results of the analysis allow achieving an enhanced
assessed using continuum-discontinuum numerical analysis. understanding of potential failure mechanisms and helps in
3DEC (3-Dimensional Distinct Element Code) has been used proposing further suitable measures to improve the stability
to model the complicated slope geometry and to explore the of the abutment.
248

Development of numerical model for grout injection and its application to

the real in-situ injection tests
T. Koyama & Y. Ohnishi

Kyoto University, Kyoto, Japan
K. Takahashi
Suimon-Gijyutu Consultant, Co. Ltd., Chiba, Japan
A. Kobayashi & T. Katayama

KANSO Technos Co. Ltd., Osaka, Japan
Keywords: numerical modelling, fluid flow, site characterisation
For the construction of underground facilities and dams, the and advection-dispersion equations using 2D finite element
control and sealing groundwater is one of the key issues. method (FEM) assuming the grout is Newtonian fluid. In the
Grouting (grout injection) is one of the commonly used meth- numerical model, the time dependent physical properties of
ods to seal groundwater during the constructions. Grout is the grout (increase of viscosity) and the reduction of porosity in
mixture of water and cement with a certain mixing rate and is the porous media were considered to express the reduction
injected into the rock fractures from the injection boreholes of hydraulic conductivity during grout injection (improving
drilled into rock masses in order to improve hydraulic con- hydraulic conductivity field). In the numerical simulations,
ductivity fields of target rock masses. Recently, underground two different grout injection methods mentioned above were
facilities have been built under various geological conditions simulated using the in-situ grout injection data including the
with limited budget, and for grouting system, more effec- change of the grout injection pressure and grout density. The
tive and economical injection method has been required. For simulation results were compared with the ones obtained from
grout injection, so far the injection method which changes the in-situ grout injection tests. The simulation results clearly
the water/cement mixing ratio (W/C) stepwise from high to showed that the change of the grout viscosity as well as the
low has been commonly used in order to inject cement into decrease of porosity in time played important roles for the
the fractured rock masses effectively. Unlike this conventional grout injection mechanism. The simulation results also agreed
method, in the newly developed method, the W/C is changed well with the ones obtained from the in-situ grout injection
from high to low continuously. This method is more effective tests. Figure 1 shows the comparison of hydraulic conductivity
injection method and starts being applied to the construction of field after 18, 30 and 180 minutes between the two injection
various underground facilities. In this study, to investigate the methods. The newly developed injection method was more
grout injection mechanism and processes, the numerical sim- effective and was able to inject more cement in shorter time
ulations were carried out by solving unsteady seepage flow comparing with the conventional injection method.
Figure 1. The distribution of hydraulic conductivity after 18, 30 and 180 minutes, a) conventional injection method and b) newly developed
injection method.
249

Modelling the evolution of fractures using a combined FEM-DEM numerical method
J.-P. Latham, L. Guo, X. Wang & J. Xiang

Department of Earth Science and Engineering, Imperial College London
Keywords: numerical modelling, rock joints, rock stress
1 INTRODUCTION composite specimen subjected to a rock mechanics test in a

plane strain rig (zero strain in the z-direction) with constant
The growth of fractures in layered units has been described velocity control ±Vy for the upper and lower boundaries and
in terms of “Sequential Infilling” where the term “Fracture constant confining pressure P for the left and right end bound-
Saturation” is given to the strata when no more fractures aries. Multi-layered rocks undergoing increasing layer-normal
can infill—even with increasing strain. Bai & Pollard (2000) loading during slow burial in a geological basin setting may
simulated an elastic three-layer material model where ver- develop a steady inelastic deformation in the shale layers as
tical fractures were inserted in the central layer before the they yield and stretch pervasively, or they may deform with
model was loaded by progressive application of strain and more localised shearing deformation styles. In the virtual lab-
provided an explanation for saturation. Tang et al. (2008) mod- oratory experiment, continuum plasticity is not implemented.
elled the entire evolution of fractures with a strain-dependent Only brittle shear and tensile failure occurs, so a variable and
FEM-based degradation model and showed fracture saturation somewhat localised inelastic as well as elastic stretching in the
occurs. Both studies used 2-D models to consider the boundary shale is generated which contributes to tensile stresses being
condition in which tensile stress was directed parallel to the imparted to the limestone layer.
associated layers. Discussion of the fracture evolution caused Four model conditions spanning a range of confining pres-
by layer normal compressive stress in the layered materials sures were chosen to illustrate fracture development with
(e.g. induced by overburden or burial) was outside the scope results shown in Figure 1. In all models, the friction coef-
of their papers. ficient µ was set at 0.5 for shearing on fractures within either
The objective of this paper is to use FEMDEM (Munjiza rock type, as well as for the horizontal interface between them
2004) to mechanically model the growth of realistic two- which behaves like a pre-existing closed crack with no cohe-
dimensional fracture networks in stratified models. A layer- sion. Sliding on the interface will therefore contribute to local
normal maximum compressive stress regime is considered. delamination when σyx ≥ 0.5σyy . Mean fracture spacing to
We take the classical three-layer sandwich as the problem layer thickness ratio, S/tf at saturation is 1.8 (P = 0 MPa) and
domain to study the controls on the fracture spacing of the 1.7 (P = 1 MPa) for the lower confinement cases, i.e., similar
central competent layer. The fracture model gives an explicit to that for the homogeneous case obtained by Tang et al. (2008,
geometric realisation of fracturing and fragmentation. The Fig. 13H). Critical spacing values at saturation are larger than
failure criteria modelled are for opening mode tensile fail- Bai & Pollard’s (2000) bonded models with pre-existing frac-
ure at the tensile strength and shear failure using a Rankine tures would suggest and delamination appears to be the cause
model with failure at a critical shear stress independent of of this. At higher confinement, the mean stress is higher and
mean stress, set here at 3.5 times the tensile strength. Crack normal stresses across the frictional layer interface inhibit
extension is constrained by the fracture energy release rate, a sliding, so that for P = 5 MPa, S/tf = 1.0 and at P = 10 MPa,
material property. The fractures develop along the pre-defined S/tf = 0.8 these fractures are more clustered into swarms or
mesh boundaries according to a smeared crack cohesive zone zones.A possible mechanism that can lead to further fracturing
fracture model; its implementation and mesh sensitivity is and values of S/tf lower than 0.8 is also revealed.
described in Munjiza et al. (2002).
3 CONCLUSIONS
2 NUMERICAL MODELLING OF FRACTURES
Spacing to thickness ratios for fracture saturation from 1.8
The problem investigated here has identical dimensions to to 0.8 were obtained for several different confining pressures.
a case investigated by Tang et al. (2008) where the cen- These were compatible with both high and low ratios observed
tral layer of 0.030 m thickness and 0.120 m long is sand- in geological sequences. Possible mechanical explanations for
wiched between two layers of 0.045 m thickness and 0.120 m many of the spacing ratios and fracture associations observed
length. Material stiffness and yielding properties that are in nature can be offered for different materials and loading
considered reasonable for a limestone–shale multi-layer are conditions based on simulations with FEM-DEM modelling
selected. The simulation performed is equivalent to a large tools.
250

Figure 1. FEM-DEM numerical simulation stages of fracture patterns forming in a shale-limestone composite subject to steady shortening
normal to the layering at 0.05 m · s−1 . a–d. shows four models with different degrees of constant layer-parallel confining pressure. Note:
shear fractures develop first in the shale and soon after, the limestone develops extension fractures which infill in a sequence, see numerals.
e details the horizontal stress during tensile fracture propagation.
251

A constitutive contact law for discrete element modeling of sandstone
L. Li
SINTEF Petroleum Research, Trondheim, Norway
Keywords: numerical modeling, rock properties, stability analysis
1 INTRODUCTION
The discrete element method (DEM) has been widely used

to simulate complex deformation and failure phenomena in
rocks. The material behavior in the models was often simply
treated as linearly elastic before the micro-cracking occurs due
to relatively high stress. This does not exactly represent the
mechanical behavior of some rocks such as sandstone which
usually has stress-sensitive mechanical parameters. This paper
presents a study to make the DEM model capture the consti-
tutive behavior of sandstone, including the stress-dependent
stiffness, the failure behavior, the plastic and the elastic defor-
mation. The objective was to improve the accuracy of the
DEM modeling. As an application example, a 3-dimensional
(3D) model was used to analyze the stress alteration and the Figure 1. Triaxial compression test results: peak axial stress vs.
mechanical damage induced by hole drilling in sandstone. confining stress.
2 CONSTITUTIVE CONTACT LAW FOR SANDSTONE
The constitutive contact law in the DEM determines how to

calculate of the interactions between a pair of elements in
response to their relative displacements. The formulation of
the contact law for the sandstone considers its microstructure
and the micromechanics.
According to the presented contact law, the contact between
two elements will be in one out of three possible states:
unbonded state, bonded state and broken state. The calcula-
tions of the interactions between the element pairs consider
the non-linear elasticity and the plasticity which may occur in
the sandstone even if it is under low stress. In the case that
a contact is in the bonded state, the normal contact force is
given by:
Figure 2. Triaxial compression test results: E50 and ν50 vs. confin-
ing stress.
where kc and kb are two coefficients, un is the total relative

normal displacement while un,p is the plastic part of un . results with laboratory experiment data of Castlegate sand-
The bonded contact may break in the tensile or the shear stone. In Figure 1, the peak axial stresses from the triaxial
mode. Then it turns into the broken state. compression tests are plotted versus the confining stress. Both
the numerical results and the laboratory data show that the
peak axial stress increases with the increasing confining stress.
3 SIMULATION OF TRIAXIAL COMPRESSION In Figure 2, the E50 and ν50 are plotted versus the confining
TESTING stress. The E50 is the tangent static modulus calculated from
the slope of the stress-strain curve at 50% of the peak dif-
The constitutive contact law has been validated through simu- ferential stress. ν50 is calculated at the same stress state, but
lating triaxial compression tests and comparing the modeling from the curve of the lateral strain and the vertical strain. E50
252

Figure 3. The model before hole drilling.
Figure 5. The stress alterations (the differences) before and after

drilling the hole.
if only the stress alterations are examined as shown in Figure 5,

the curves are smooth. This proves that this type of models
can be used to study stress alteration with reasonably good
accuracy.
5 CONCLUSION AND DISCUSSIONS
This study demonstrated the possibility to make a DEM model

simulate the constitutive behavior of the sandstone properly
through the development of a contact law. This contact law
considers the microstructure and the micromechanics of the
Figure 4. Bond breakages (shown in discs in 3D) in the model due deformation and the failure of sandstone. It has included
to drilling the hole. non-linear elastic deformation and plastic deformation of the
sandstone. The DEM model using such a contact law cap-
and ν50 are important parameters for the characterization of tures the important aspects of the mechanical behavior of the
reservoir sandstone. The E50 from the modeling quantitatively sandstone, including the strength properties and deformation
agree with the laboratory data. Both show the trend that the E50 properties.
increases with the increasing confining stress. The ν50 from The DEM model has also been applied to the analysis of the
the modeling show the same trend as the laboratory data: the stress alteration and the mechanical damage in the rock when
ν50 decreases with the increasing confining stress. a hole is drilled. The calculated stresses in the DEM model
fluctuate due to the heterogeneity of the model. However, the
calculation of the stress alteration (“repeated stress calcula-
4 DRILLING-INDUCED STRESS ALTERATION AND tion”) was seen reasonably accurate. The observed mechanical
DAMAGE IN SANDSTONE damage and near hole failure mainly locate in the azimuth of
the minor horizontal principal stress direction. It agrees with
The model shown in Figure 3 was used to simulate well drilling field and laboratory observations.
in sandstone. The input parameters were the same as in the
model which was used for the simulations of triaxial com-
pression tests. The model was first loaded to an initial stress ACKOWLEDGEMENTS
state: σ x = 40 MPa, σ y = 50 MPa, σ z = 20 MPa. Then keeping
the boundary stresses constant, a vertical hole with diameter The Research Council of Norway supported this work. The
0.2 m was drilled along the y-axis. Bond breakages occurred experiment data are from the Formation Physics Laboratory
near the hole, in the azimuth of the minor horizontal principal in SINTEF Petroleum Research.
stress direction (Figure 4). The damage zone may develop into
a breakout.
Due to the heterogeneity of the DEM model, fluctuations REFERENCES
are commonly observed in the average stress calculations in
a small domain with a small number of elements. However, See full text.
253

Numerical modeling of rock cutting
J.-S. Lin
U.S. Department of Energy, National Energy Technology Laboratory, Pittsburgh PA, USA
Department of Civil and Environmental Engineering, University of Pittsburgh, Pittsburgh PA, USA
J.A. Mendoza, M.C. Jaime & Y. Zhou

Department of Civil and Environmental Engineering, University of Pittsburgh, Pittsburgh PA, USA
J. Brown, I.K. Gamwo & W. Zhang

U.S. Department of Energy, National Energy Technology Laboratory, Pittsburgh PA, USA
Keywords: lab testing, oil reservoir, rock properties
Rock cutting is at the core of all construction on rocks or into successfully. In the discrete element analysis, assemblies of
rocks. A cutting action, by its nature, is defined by causing circular disk particles, in 2-D, and spherical particles, in 3-D,
failure in a rock so that part of the rock can be removed. For a were created following the grain size distribution of a par-
practical application the numerical solution is often the only ticular rock. The assemblies were then packed under a preset
credible approach. From modeling perspectives, rock cutting level of compressive stress and bonded together afterwards at
constitutes a sequence of difficult problems: A contact prob- the points of contact between particles. The micro-parameters
lem first arises as a cutter advances and interacts with a target were calibrated with laboratory mechanical tests.
rock. This is followed by the problem of determining when We have found both the discrete element and the finite
and if the rock would fail. If the rock does fail, a modeler element methods were capable of providing cutting force
is subsequently faced with the problem of how to initiate the estimates and capturing the important characteristics of the
fragmentation process. Moreover, this sequence of problems fragmentation process. One of the major remaining issues is:
repeats itself each step of the way. how to select a velocity in the modeling that is a proper refec-
In this study, we employed both the finite element method tion of the actual cutter velocity? This question can only be
and the discrete element method for modeling rock cutting. resolved through calibration provided sufficient test data are
Our immediate goals were to develop a framework within the available. Our study also showed that particle crushing could
two methods so that a laboratory rock scratch test could be have important impacts both on the cutting forces and fail-
well modeled. The reasons for the focus were twofold: first, a ure modes. Again, the answers require the aid of a carefully
scratch test possessed all essential characteristics of a general planned experiment program. For drilling applications, the
rock cutting problem; second, there were test data available logical next steps of rock cutting research include incorpo-
for validation. We used LS-DYNA for the finite element anal- rating high pressure effects and addressing rock-fluid-cutter
ysis, PFC2D and PFC3D for the discrete element analysis. interaction problems. These types of research will have appli-
In the finite element analysis, the Continuous Surface Cap cations over a wide range of problems beyond drilling. A
Model, a visco-elastic-plastic damage model, was the mate- research on the modeling of complete drill bit mechanics is
rial model we used. After substantial tuning, this material also desirable as the experiences should be readily applicable
model together with element erosion that removed elements to other areas such as modeling the mechanics of a complete
with excessive damages, modeled rock cutting fragmentation tunnel boring machine.
254

Strength symmetry with respect to the stress state imposed by

Mogi’s theory against long-prevailing observations
X.D. Ma
Shandong University, Shanghai, China
R.J. Rodriguez
ETSI Caminos, C. y P. Technical University of Madrid, Spain
Keywords: lab testing; numerical modeling; rock failure
1 INTRODUCTION 2000, Haimson & Rudnicki 2010), thereby confirming the

important effect of σ 2 on rock strength.
The most commonly used rock failure criteria, Mohr-Coulomb Based on this evidence, various researchers proposed 3D
(M-C) and Hoek-Brown (H-B), are both in essence two- failure criteria to better fit existing experimental data. Among
dimensional (2D). However, even though it is well acknowl- them, Mogi (1971) proposed that rock failure criteria should
edged that traditional 2D failure criteria are only a limited have the form of:
simplification of reality (since they do not account for the
effect of the intermediate principal stress, σ 2 ), they have been
widely employed in practice because they captured (in a con-
servative way) one essential aspect of a failure criterion; i.e. Mogi’s assumption provides failure criteria that produce
the “limiting effect” of the minimum principal stress σ 3 . good fit to existing empirical data, and has therefore been
From this perspective, a traditional 2D failure criterion can widely employed for 3D rock failure criteria.
be seen as a function σ 1 = f (σ3 , K) of σ 3 (which we will here- We will take this form proposed by Mogi (1971). Following
inafter call “limiting factor”) and the material constants K. Al-Ajmi & Zimmerman (2005), we focus on Mogi’s funda-
(For simplicity, material constants will not be further consid- mental contribution (Eq.(1)), while the specific algebraic form
ered in this paper.) When a 2D failure criterion is upgraded of the function becomes secondary concern.
into a 3D failure criterion, the intermediate principal stress σ 2 In this paper we show that Mogi’s fundamental assumption
needs to be considered, so that we have an expression in of the (Eq.(1)) imposes symmetry of the failure criterion with respect
form of σ 1 = f (σ2 , σ 3 ). However, it is often more convenient to the stress state factor value of β = 1/2 (this is referred to as
to work with a stress state factor β (β = (σ2 − σ3 )/(σ1 − σ3 )) pure shear state, where σ2 = (σ1 + σ3 )/2). This observation,
to consider the effect of σ 2 . An additional benefit of using this which occurs independently of the actual form of failure cri-
stress state factor is that it allows us to unify the treatment of teria considered (linear, exponential or quadratic), does not
strength results corresponding to different testing modes or, seem to have attracted much attention in the literature before,
equivalently, stress states. That results in a failure criterion in even though it goes against empirical evidence suggesting that,
the form of σ 1 = f (β, σ3 ) which depends on the stress state for constant σ 3 , rock strength is higher in triaxial extension
factor and the limiting factor. (σ 2 = σ 1 ) than in triaxial compression (σ 2 = σ 3 ). Note, how-
Over the last century, there has been intense experimental ever, that we do not aim to directly negate Mogi’s work (Mogi’s
research to consider the effect of σ 2 on rock strength (defined criterion does reproduce experimental data quite well for a
as peak strength or fracturing strength in this paper). Murrell wide range of β values (from 0 to 0.7) comprising the range
(1963) was stimulated by previous testing results by Von of data available from his testing at the time); rather, we see it
Kármán (1911) and Böker (1915) to investigate the effect of σ 2 as an opportunity to discuss about possible reasons for such
on rock strength and he conducted triaxial compression (TC) a discrepancy between theory and empiricism and to present
and triaxial extension (TE) tests in which he found empiri- some ideas about what a general 3D failure criterion should be.
cal evidence of the strengthening effect of σ 2 for constant σ 3 . Finally, it has to be noted that there is still probably a scarcity
Conversely, almost at the same time, Brace (1964) conducted a of data related to 3D failure behavior of rock. At an early stage,
set of triaxial tests (one at uniaxial stress state, while another at the testing done by Mogi (1967, 1971) only obtained rock
quasi-triaxial extension state) and he postulated that σ 2 has lit- strength when β is no more than approximately 0.7. Those
tle influence on failure based on his (limited) observation that data generally confirm the qualitative adequacy of the theory
almost identical strength occurred at different values of σ 2 for of Wiebols & Cook (1968). Similarly, Fjaer & Ruistuen (2002)
the same σ 3 .The results of Murrell (1963) were later replicated presented a numerical and an alternative probabilistic way
by other researchers working with improved apparatus (Mogi to estimate rock peak strength by considering the influence
1967, 1971, Takahashi & Koide 1989, Haimson & Chang of σ 2 and using the basic Coulomb-type of microscopically
255

shear-in-plane failure mechanism. Their DEM-based numeri- simplifications. However, if we can confirm that real results
cal simulation yielded strength monotonically increasing from are very close to symmetry, the effort of estimating and
β = 0 to β = 0.5; this trend was found to (symmetrically) employing failure criteria in practice can be greatly reduced, as
reverse as β rised from. (Similar results have been reported it would greatly reduce the number of data needed to calibrate
in the soil mechanics field by Ng (2003).) However, Fjaer & the failure criteria.
Ruistuen’s probability model gave a small difference in the
expected strength in vicinity of σ1 = σ2 . More recently, how-
ever, Haimson & Rudnicki (2010) used a testing apparatus SELECTED REFERENCES
to replicate that of Mogi’s and conducted tests for β values
around 1, obtaining some results which indicated the strength Brace, W. F. 1964. Brittle Fracture of Rocks. In W. R. Judd (ed),
when σ1 = σ2 are equal or even (perhaps surprisingly) lower State of stress in the earth’s crust: 111–180. New York: Elsevier.
Cai, M. 2008. Influence of intermediate principal stress on rock
than at σ2 = σ3 . Such conflict between the imposed symmetry fracturing and strength near excavation boundaries – Insight from
and the available empirical data triggers the question of the numerical modeling. International Journal of Rock Mechanics and
range of β values in which a rock failure strength criterion Mining Sciences 45: 763–772.
should be employed, and consequently, a concern about the Fjaer, E. and Ruistuen, H. 2002. Impact of the intermediate prin-
application of failure criteria based on Mogi’s law over the cipal stress on the strength of heterogeneous rock. Journal of
whole range of β values (β in [0,1]). Geophysical Research 107(B2): 2032.
It might be argued that for high β values, the corresponding Haimson, B. and Chang, C. 2000. A new true triaxial cell for
stress states are rarely realistic in field situations. However, testing mechanical properties of rock, and its use to determine
these high β values might occur in engineering practice such as rock strength and deformability of westerly granite. International
stress redistribution after excavation (it is suspected that stress Journal of Rock Mechanics and Mining Sciences 37: 285–296.
Haimson, B. and Rudnicki, J.W. 2010. The effect of the intermediate
state within a certain range is a prerequisite for rockburst), or principal stress on fault formation and fault angle in siltstone.
in laboratory simulation (e.g., a dynamic unloading usually Journal of Structural Geology 32: 1701–1711.
captures transitions from low β values to high β values), or in Mogi, K. 1967. Effect of the intermediate principal stress on rock
field research (the formation of strain-localized bands). failure. Journal of Geophysical Research: 5117–5131.
In any case, it seems clear that further research is needed in Mogi, K. 1971. Fracture and flow of rocks under high triaxial
this topic. Experiments are never perfect because of specimen compression. Journal of Geophysical Research 76: 1255–1269.
anisotropy, inhomogeniety, systematic errors and imperfec- Rudnicki, J. W. & Rice, J. R. 1975. Conditions for the Localization
tions. Similarly, numerical and analytical simulations of rock of Deformation in Pressure-Sensitive Dilatant Materials. Journal
strength are constrained by their intrinsic assumptions and of the Mechanics and Physics of Solids 23: 371–394.
256

Scale effects in a synthetic rock mass
C.D. Martin, Yun Lu & Hengxing Lan

Department of Civil & Environmental Engineering, University of Alberta, Edmonton, Canada
Keywords: lab testing, rock mass, rock properties, numerical modelling
1 INTRODUCTION
There are two contributors to rock mass strength: (1) the

strength of the intact rock and (2) the strength of the frac-
tures. The intact and fracture strengths can be measured in
routine laboratory tests at small (typically centimetre) scales.
Testing at larger scales is seldom practical and laboratory tests
are replaced with empirical approaches such as the Geological
Strength Index. More recently, with the significant advances
in computer modelling, there is a growing effort to replace
the empirical approaches with a synthetic rock mass (SRM).
The SRM utilizes a numerical modelling technique that com-
bines discrete element modelling of intact rock with discrete
planar fractures (Pierce et al. 2007). The discrete element
modelling can utilize the bonded particle approach or the Figure 1. Summary of UCS data, normalized to the minimum
voronoi tessellation scheme. The planar fractures are captured representative diameter, (Hoek & Brown = 50 mm, Jackson &
using a Discrete Fracture Network, a convenient method for Lau = 63 mm and Blanks & McNamara = 152 mm).
geometrically describing the rock mass fractures.
In this paper the effect of scale on intact low porosity rock
strength is reviewed and guidelines are proposed for selecting Figure 1 shows a marked scale effect for normalized diameter
the intact rock strength for SRM simulations. less than approximately 3 but that there is no scale effect when
the sample diameter is at least 3 times the representative diam-
eter. The Hoek-Brown equation for intact scale effects given
2 SCALE EFFECTS: INTACT ROCK AND CONCRETE in Hoek & Brown 1980) was also normalized and, as shown
in Figure 1, provides a reasonable fit to the data showing the
The first step in creating a Synthetic Rock Mass is to cali- scale effect. However the Hoek-Brown equation significantly
brate the numerical intact rock properties to the laboratory under-predicts the strength for normalized diameter greater
properties. It is generally agreed that the volume tested does than 3, where the data shows no scale effect. A better fit to the
not significantly affect the deformation properties (Young’s normalized data can be obtained using:
modulus and Poisson’s ratio) of intact rock. This is not sur-
prising for low porosity rocks as long as the samples being
tested are homogeneous and the sample size does not influ-
ence the test method. To minimize the effect of the sample size
on the test results the ISRM states that the “test specimen shall
be right cylinders having a height to diameter ratio of 2.5–3.0
and a diameter preferably of not less than 54 mm.” The method where A = 0.81, B = 0.4, C = 1.21 and D is the normalized
then states that the diameter of the specimen should be greater diameter. Equation 1 illustrates that the reduction in UCS due
than 10 times the largest grain in the rock. to scale effects is limited to 0.8 of the representative labo-
The results from Jackson and Lau (1990) and those from ratory UCS determined for the intact material, which is also
Blanks and McNamara (1935) reinforce the recommendation consistent with the original data from Hoek and Brown (1980).
by the ISRM that for consistent results the grain size of the The test results demonstrate that the scale effects in intact
intact material must be considered in determining the min- rock are relatively minor and caused by the sample test size,
imum test diameter for establishing consistent compressive and not a material property.A reasonable lower bound estimate
strengths. The data from all the tests results discussed thus far for the intact strength using laboratory test results based on
are summarized in Figure 1.The uniaxial compressive strength traditional sample diameters of 50 to 63 mm, is 80% of the
and the diameters in Figure 1 have been normalized to the UCS UCS. This assumes that the intact rock is homogeneous. If the
and diameter that are considered representative of the material sample contains micro-defects, such as filled veins, the rock
tested in order to compile the data on a single plot. The data in strength may decrease to less than 80% of the UCS.
257

Figure 2. The rock mass model with fracture traces in PFC2D, data
from Park et al. (2004).
3 EFFECT OF STRUCTURE ON ROCK STRENGTH
The structure in a rock mass can occur at several scales: Figure 3. Effect of fracture intensity and fracture orientation on
(1) centimetre scale where the micro-defects are healed with confined rock mass strength.
infilling that is usually weaker than the intact rock and (2)
metre-scale where the fractures generally lack infilling and are 4 CONCLUSIONS
classed as open. In the synthetic rock mass, the rock fractures
are described using the geometrical descriptions associated Laboratory results were used in conjunction with physical
with the Discrete Fracture Network (DFN). The fracture inten- and numerical model results to examine (1) the strength of
sity description in a DFN is given by: P10 = No of fractures the intact rock and (2) the strength of a fractured rock mass.
per unit length, #/m, P21 = Trace length of fractures per unit The findings suggest that scale effects in intact rock are lim-
area, m/m2 , P32 = Area of fractures per unit volume, m2 /m3 . ited to 80% of the standard laboratory uniaxial compressive
The rock mass at the Äspö Hard Rock Laboratory (HRL), strength. The physical and numerical modelling results sug-
Sweden, is typical of the granitic rock masses found in the gest that the rock mass strength decreases as the fracture
Scandinavian Shield. It contains several joint sets that vary in intensity, expressed as fracture area per unit volume, increases.
both trace length and fracture frequency. A discrete fracture However, the lower limit to this decrease is controlled by the
network model was developed for the Äspö HRL and the three fracture orientation and persistence, regardless of the fracture
fracture sets with their fracture intensity values (expressed as intensity. Hence fracture intensity alone may not be an ade-
P32 ) are provided in Figure 2. quate parameter when estimating rock mass strength using
Park et al. (2004) used the Particle Flow Code to examine discrete fracture networks.
the effect of fractures on rock mass strength. In their mod-
elling the fracture sets were incrementally added to assess the
effect of each fracture set on rock mass strength. The results REFERENCES
from their modelling are provided in Figure 3 as a function
of fracture intensity, and show that as the fracture intensity Blanks, R. and McNamara, C. 1935. Mass concrete tests in large
increases the rock mass strength decreases. If the rock mass cylinders. ACI Journal Proceedings, 31(1):280–303.
was intact the expected strength using the results from Figure Brown, E. T. and Trollope, D. H. 1970. Strength of a model of jointed
rock. ASCE Journal Soil Mechanics and Foundations Division,
1 would be 80% of the peak strength while the a lower bound
96(2):685–704.
strength would be that provided by an optimally oriented sin- Einstein, H. H., Nelson, R. A., Bruhn, R. W., and Hirschfeld, R. C.
gle through-going fracture as shown in Figure 3. These limits 1969. Model studies of jointed-rock behaviour. In Proc. 11th U.S.
appear to adequately bound the rock mass strength determined Symposium on Rock Mechanics, Berkeley, pp. 83–103.
by Park et al. (2004). Hoek, E. and Brown, E. T. 1980. Underground Excavations in Rock.
Brown and Trollope (1970) and Einstein et al. (1969) con- The Institution of Mining and Metallurgy, London.
ducted a number of physical model tests to establish the effect Jackson, R. and Lau, J. S. O. 1990. The effect of specimen size on
of fractures on rock mass strength. Their tests were carried the mechanical properties of Lac du Bonnet grey granite. In Proc.
out using a plaster model with continuous fracture planes. The 1st. Int. Workshop on Scale Effects in Rock Masses, Loen, Norway,
results from their model studies are also shown in Figure 3. pp. 165–174.
Park, E-S., Martin, C. D., and Christiansson, R. 2004. Simula-
The fracture intensity in Figure 3 is expressed as fracture area
tion of the mechanical behavior of discontinuous rock masses
per unit volume. For the model tests the volume of each test is using a bonded-particle model. In CD-ROM Proc. 6th NARMS04,
considered the unit volume. The results in Figure 3 show that Houston, pp. 1–8. NARMS 04–480.
regardless of the fracture intensity, the persistence of a single Pierce, M., Mas Ivars, D., Cundall, P. A., and Potyondy, D. O. 2007. A
fracture can reduce the strength of the rock mass to its lower synthetic rock mass model for jointed rock. Proc. 1st Canada-US
bound strength. Rock Mechanics Symposium, Vancouver, vol. 1, pp. 341–349.
258

2D-/3D-stress analysis of deep-seated gold mines in South Africa
F. Menschik, K. Thuro, H. Käsling & M. Bayerl

Lehrstuhl für Ingenieurgeologie, Technische Universität München
Keywords: case studies, lab testing, mining, numerical modeling, rock properties, rock stress
Today many mines are built in higher depths to prospect for the tensile strength and the abrasivity (e.g. Cerchar abrasivity
needed resources and obtaining them. Not only mines but also index, LCPC abrasivity coefficient, petrographic composition
many construction and tunneling projects like the Gotthard with thin section analysis) were determined. In addition to the
base tunnel or the Brenner base tunnel advances into these laboratory testing the performance and the wear data are col-
deep laying rock zones where typical high stress phenomena lected directly in the mines and will be generated in additional
like rock bursts can occur. Higher stress can also lead to prob- drilling tests. For the evaluation and the interpretation all gath-
lems like high costs due to high wear or less performance of the ered data will be combined in a MySQL database and will be
rock cutting tools. Due to the complexity of the measurement analyzed statistically using Matlab (The MathWorks).
methods the stress rates were seldom directly quantified. In Three 2D cross sections of a regular stope geometry were
most cases the stress is only determined by numerical models. modeled and calculated with Phase2 (Rocscience) to define
This paper and the associated research project deals with the the occurring stress and its distribution. For a detailed analysis
influence of these high stress rates on wear and performance further complex 3-D models (Flac 3D, Itasca) are calculated.
of rotary hammer drills in deep-seated mines in South Africa The previous collected field data will be compared to the cal-
up to a depth of 3900 m. The encountered rocks are high abra- culated stress data from the numerical models to benchmark
sive quartzites with high uniaxial compressive strength. These the influence of stress on wear and performance of the used
rocks were mined to reach the decimeter thick gold bearing tools. The aim of this research work is to take the advantage
layers – the so called reefs. of these effects to increase the efficiency and to optimize the
The most common geomechanic properties including the mining progress.
rock strength like uniaxial compressive strength or brazilian
259

Numerical modeling for the evaluation of grout penetration in fractured rock masses
Y. Mito
Department of Urban Management, Graduate School of Engineering, Kyoto University
M. Ashraf
School of Civil Engineering, Universiti Sains Malaysia
T. Yamamoto
Kajima Corporation

Keywords: numerical modelling, rock joints, fluid flow, site characterization
Grouting aims to reduce the permeability of rock mass below model. The goal achievement probability under the specific
the required value and has been widely used in rock engineer- target permeability is computed with the help of geostatisti-
ing field for a long time. However, the injection procedure cal interpolation method. The applicability of the method is
needs several empirical decision makings and it is rather verified through the several field applications.
difficult to reliably confirm the achievement of improve-
ment, because the grout penetration process is not sufficiently
clarified.
This study proposes a stochastic prediction method of spa-
tial distribution of rock hydraulic properties after grouting in
fractured rock mass by using the grout penetration process
Figure 2. The computed goal achievement probability and the

measured Lugeon value under the target Lugeon values.
Figure 3. The computed goal achievement probability and the

Figure 1. The calculation method of goal achievement probability. measured achievement ratio under the target Lugeon values.
260

Numerical method for determining contact areas of a rock joint under shear load
Jung-Wook Park, Yong-Ki Lee and Jae-Joon Song

Department of Energy Resources Engineering, Seoul National University, Korea
Keywords: rock joints, roughness, contact area, back analysis, lab testing
Joint roughness is one of the most important factors in under-

standing and estimating the shear behavior of rock joints.
Several methods to quantify the joint roughness by using some
indices such as JRC, statistical parameters and fractal dimen-
sion have been proposed. In these methods, however, a whole
area of a joint surface is characterized with a single value that
the joint roughness is kept constant regardless of the shearing
area or shearing process, while only certain fractions of the
joint are in contact during the practical shearing process and
every part of the joint surface has different roughness.
In this study, a new method to determine the contact
areas of a rock joint under normal and shear loads by using
back-analysis of direct shear tests is proposed. The algo-
rithm requires only three dimensional surface coordinates
at the initial stage before shearing, while in other conven-
tional methods disparate materials are inserted between joint
surfaces or particular equipments are adopted for the test,
such as pressure-sensitive paper (Ducan & Hancock, 1966),
deformable film (Iwai, 1976; Bandis et al., 1983) and CT scans
(Re & Scavia, 1999). In the proposed method, a joint surface
is expressed as a group of triangular planes and each plane
is examined on contact condition by calculating the relative
displacements from initial locations of both surfaces. Assum-
ing that two surfaces are completely matched at initial stage
and the asperities are not significantly damaged, the contact
condition of a point (xi , yj ) can be written as follows:
Figure 1. Schematic view of the surface information on XY grid.
between upper and lower surfaces were observed. The loca-

where, Zlow (x, y) is the initial elevation function of the lower tions of the actual contact points showed good agreement with
surface, and dx and dz are displacements of the upper block to those estimated by the proposed algorithm.
the fixed lower block in shear and normal directions, respec- In order to characterize the steepness of contact areas which
tively. Examination on the contact condition based on the point greatly affect the shear behavior, the micro-slope angle θ i in
coordinates can show us the location of contact points, but it is range of −90◦ and 90◦ was defined as the apparent dip angle
still difficult to get information about the distribution of size of a triangular element i of a joint on the shearing direction
and roughness of the contact areas. To overcome this defi- plane (XZ-plane in Figure 1). The elements dipping toward the
ciency, each plane comprised of three neighboring points is shear direction may come into contact condition and actively
examined on the contact condition, e.g. ABC and DEF in contribute to the shear behavior of the joint at the initial stage,
Figure 1. The contact condition for a plane is more compli- and the elements inclined opposite may be detached, causing
cated than that for a point because a plane can be in contact voids. Therefore, we classified the elements into two groups:
although all the three points don’t satisfy the contact condition. the former is ‘active’ elements and the later ‘inactive’ ele-
Therefore, the coordinates of the centers of planes on both sur- ments. The magnitude and sign of θ i indicate steepness and
faces were chosen for the examination as an alternative. This orientation of the surface element with respect to the shearing
method was verified by using PFC simulation of direct shear direction: The plus and minus signs denote active and inactive
test, in which the physical contact points transmitting the force elements, respectively.
261

indicates the shear direction of the block. The white color
denotes non-contact elements, and the active and inactive
elements are displayed in warm and cool colors, respectively.
In pre-peak stage, all of the active elements were in contact,
while the inactive ones with relatively low slope angle sur-
rounding the active zones were partially brought into contact.
The contact area gradually decreased with increasing shear
displacement until the peak stage, however, which was mainly
ascribed to the separation at the inactive zones. At the peak
stage, the normal dilation was initiated causing a sharp drop
in contact area, and about 53% of the surface area came into
contact and supported the loads. The contact areas were almost
comprised of the active elements, and the locations of contact
areas were remarkably consistent with those of active zones.
Figure 2. Variation in contact area ratio with shear displacement. After peak stage, only the active elements with high slope
angle >10◦ remained in contact, and all of the inactive ele-
ments were detached. The contact area ratio decreased rapidly
with increasing shear displacement, and few inactive elements
got into contact until the residual stage. As the residual stage
was reached, only small fractions <3% were involved in con-
tact. It is remarkable that the inactive elements with high slope
angle started to be detached, and the active ones with high
slope remained in contact until residual stage in reverse.
The effects of normal stress and normal stiffness on the con-
tact area were examined by analyzing the test results under 6
different normal load conditions. Under CNL condition, more
elements were kept in contact throughout the test at a higher
normal stress, but contact areas decreased dramatically after
peak stage regardless of normal stress. At peak stage, the
contribution of inactive elements to contact areas increased
remarkably with increasing normal stress, while the contact
area ratio showed little variations. Under CNS condition, the
contact area ratio varied gradually and a dramatic decrease
Figure 3. Estimated contact areas with their micro-slope angles at after peak stage was not observed. The contact area at peak
peak stage. increased with normal stiffness due to the suppressed dila-
tion. The inactive elements in contact were distributed widely
around the active zones, which means that peak strength was
For the application of the proposed method, replicas of developed with both surfaces completely matched, contrary
joints created by splitting a granite block were prepared and to CNL tests.
direct shear tests under CNL and CNS conditions were carried In summary, a numerical method to determine the contact
out. The test results under constant normal stress of 0.5 MPa areas of a rock joint under shear and normal load was pro-
was analyzed for the distribution of contact areas at each of posed with verification, and the 3D characterization of the
four shearing stages: pre-peak, peak, post-peak and residual surface was developed in consideration of shearing direction.
stage. Figure 2 shows the change in contact area ratio together The change in contact areas could be continuously estimated at
with test results, where the contributions of active and inactive each shearing stage under various normal load conditions. The
elements were expressed in different colors. Figure 3 shows distribution of asperities which dominantly affected the shear-
the predicted contact areas with their micro-slope angles at ing behavior was discussed with visualization of micro-slope
peak stage (dx = 0.615 mm), where the arrow in the bottom angles.
262

Approximate methods in numerical rock mechanics modelling
J.J. Peura
VTT Technical Research Centre of Finland, Finland
J.P. Pöllä & I. Konstantas

Fundatec Ltd., Finland
Keywords: numerical modelling, weak rock
1 INTRODUCTION material softening. At the time support is installed at the exca-

vated section, it would have experienced deformations that can
The Ring Road I is the busiest highway in Finland and up to be modelled in two dimensions by reducing the deformation
70 000 vehicles pass Ring Road I daily in Leppävaara. The modulus of the excavated material (Curran et al., 2003). In the
traffic volume is predicted to grow by 20% in the following third method fore poles are projected to the plane and the bolt
decades and an extensive improvement program will be car- properties and spacing are the same as in the 3-dimesional
ried out during 2008–2012. As a part of this improvement case.
traffic will be shifted into the 500 m long Mestaritunnel, of The methods mentioned above were tested by parame-
which 300 m is double rock tunnel. The excavation span of ters derived from bedrock investigations for Mestaritunnel –
the tunnel varies between 19–22 m and the width of the pillar project. According to geological information (Kuula et al.,
between tunnels is about 10 m. According to bedrock investi- 2006) the GSI values were 24 (very poor rock quality) and
gations poor quality rock mass will be met at the halfway in 40 (poor rock quality). A uniaxial compressive strength was
the tunnel. Preliminary rock mechanical calculations showed 220 MPa and Poisson’s ratio was 0.25. The Young’s modulus
that there will be difficulties in stabilities due to the displace- for very poor rock quality was 4.7 MPa and for poor rock
ments during the excavation process. To maintain the stability quality 13.2 MPa. Other rock parameters were estimated by
and safety when excavating the very poor rock zone required using Roclab-software.
sequential excavation the use of fore poling. Because the The tunnel geometry corresponded to the real rock-building
problem was a 3D-problem, it would need a lot of calculations project only its dimensions because in this study tunnel geom-
and time to analyse the effect of different support system to etry wanted to keep simple as possible to make sure that
the stability. calculated displacements are generated only due to the rock
It was decided to carry out a study to find out if it is possible mass properties. The span of the tunnel was 19.2 meters and
the use 2-dimensional approach in solving the 3-dimensional the height was 8.8 meters. The length of the modelling tunnel
support problem with reasonable accuracy. Main problem was was 120 meters. The thickness of the rock roof was 8 meters.
in modelling the properties and behaviour of fore poling in The major vertical stress was based on the gravity and the
2-dimension. The programs to be used were Phase2D and major horizontal stress was about 30% of the vertical stress
FLAC3D and the rock properties are derived from the bedrock (Peura, 2008).
investigations. The tunnel was excavated in three stages – first in the middle
and then on the sides. The width of the opening was 6 meters
and the length of the round was 4 meters. The tunnel was
2 NUMERICAL MODELLING supported in both cases as follows:
– Forepoling with 12 m self-drilling rock anchors (Belbor
In this study the three variable methods were studied: 1) so R44/24), c/c 1.5 m, fan interval 4 m
called fore pole umbrella, 2) use of soft material in order – CT-22 M rock bolts, length 5 m, c/c 1.5 m
to describe the initial deformations before the actual sup- – Steel fibre shotcrete layer 60 mm as a temporary support
port is installed and 3) projecting fore poles to 2-dimensional
plane. In the first case the effect of fore poles are modelled Rock mechanical analysis was carried out using Phase2D
roughly as a zone of ‘reinforced rock’. The strength of ‘rein- and FLAC3D – software.
forced rock’ is calculated by multiplying the strength of each
component (rock, steel and grout) by the area of the com-
ponent and summing them up. This new rock mass strength 3 CONCLUSION
describes the features of reinforced rock (Hoek, unpubl.).
The second method is based on the softening behaviour of The conclusions derived from results of the 2- and
material inside an excavation. The 3-dimensional stress dis- 3-dimensional calculations were: In fore pole umbrella
tributions at the face of a tunnel can be interpreted in terms of method supported area is modelled as reinforced rock. In
263

Table 1. Total displacements calculated by Phase2D and FLAC3D Table 2. Vertical displacements calculated by Phase2D and
(Peura, 2008). FLAC3D (Peura, 2008).
Total Total Vertical Vertical

displacement displacement displacement displacement
Model (GSI 24) (GSI 40) Model (GSI 24) (GSI 40)
FLAC3D 2.25–2.5 mm >0.7 mm FLAC3D 2.0–2.5 mm >0.6 mm

CT bolts, Phase2D 3.00 mm 0.99 mm CT bolts, Phase2D 3.00 mm 0.98 mm
Fore poles, Phase2D 2.55 mm 0.99 mm Fore poles, Phase2D 2.60 mm 0.98 mm
the phase of “a new rock mass properties” calculation it was During the excavation work the rock quality in pilot tunnel
noticed that this method can not be used these rock param- was checked continuously on daily basis by design engineer
eters because the new calculated rock mass properties were and careful observations showed that the rock quality was
only few hundredth parts smaller than the original properties. better than predicted and no fore poling was required. Exten-
Fore pole umbrella-method can be used only if the strength someter measurements showed displacements from 0.2 mm
and the deformation modulus of the rock are very diminutive up to 4 mm depending on the location of the extensometer.
in relation to quantity of the support.
Usage of the soft material in the excavation: The proposed
method is based on softening behaviour of the material within REFERENCES
the excavation. In this method the soft material describes
deformations before the support system is installed using Curran J. H., Hammah R. E & Yaboub T. E. 2003. A two-dimensional
minor Young’s modulus in the excavation area than outside Approach for Designing Tunnel Support in Weak Rock. Proceed-
excavation. In this case calculated results were not reliable ings of the 56th Canadian Geotechnical Conference. October
because in the all other calculations displacements were 2003. Winnipeg, Manitoba, Canada. 6 p.
much smaller. The failure was caused by unsupported tunnel Hoek E. 2003. Numerical Modelling for Shallow Tunnels in Weak
modelling; the multi-element model collapsed and further cal- Rock. www.rocience.com (2007–2008). 12 p.
culations were done by using only 2 phase’s method to achieve Peura J. J. 2008. Likimääräismenetelmät kalliomekaanisessa ele-
the balance. Despite the failure the method itself seems to be menttilaskennassa (Approximate Methods in Numerical Rock
Mechanics Modelling). Master thesis. Helsinki University of
promising. Technology. Espoo, Finland. 114 p.
When the fore poles are formulated in plane in Kuula, H., Remes, H., Somervuori P, Syrjänen P., Hakala M.
2-dimensional space, the results of the displacement are the 2006. Mestarintunnelin rakennettavuusselvitys (Rock mechanical
same than 3-dimensional model. The results are presented in feasibility study). WSP Gridpoint Finland.
Table 1 and 2. Horizontal displacements were insignificant.
264

Hydromechanical modeling of an underground excavation with an elastoviscoplastic

behavior law and regularization
R. Plassart, R. Fernandes & F. Laigle

EDF CIH/R&D, Le Bourget-du-Lac/Clamart, France
A. Giraud
Laboratoire LaEGO, ENSG INPL, Vandoeuvre-lès-Nancy, France
D. Hoxha
Institut PRISME, Université d’Orléans, Orléans, France
Subject: Underground storage
Keywords: fluid flow, nuclear repository, numerical modeling
The long term behavior of underground excavations is a social Creep phenomenon has already been studied for COx-
and economic stake, in particular in the context of nuclear argillite (Gasc-Barbier et al. 2004 for instance), but the
waste storage in deep geological formation. Several experi- association of both creep and hydro-mechanical effects are
mental galleries have been dug in the underground research quite rare in the literacy (Jia et al. 2008). The specific elas-
laboratory (URL) of Bure in the Meuse/Haute-Marne region toviscoplastic model used in this study is the L&K model
in France, where studies are leaded in order to understand the (Kleine 2007): it offers a coupling between instantaneous and
global behavior of the constitutive rock which is the Callovo- delayed behavior, and an influent variation of dilation which
Oxfordian argillite (Cox argillite) (see Jia et al. 2008; Martin governs the volume strains of the material during a solicita-
et al. 2010 for instance). tion. Rock strain is strongly coupled with fluid flow thanks
The purpose of this paper is to establish a modeling with to Biot’s equations (Coussy 1995). Even if taking unsaturated
Code_Aster of one of the Bure laboratory galleries taking into conditions would be possible afterwards, modeling is limited
account non local approach, creep effect and hydromechanical for this study to saturated ones.
coupling in the framework of the mechanics of porous media, Furthermore, another main novelty of this work concerns
and then to compare numerical results with available experi- the coupling between such complex rheological behavior and
mental data. An example of result can be seen in Figure 1: we non local approaches. Among available methods of regular-
compare water pressure changes in a point of the rock mass ization, the second gradient of dilation (Fernandes et al. 2008)
with and without activating the viscosity in the model. is well fitted to dilatant geomaterials. Its aim is to avoid mesh
dependency and numerical localized solutions, and allow the
simulations to be achieved.
REFERENCES
Coussy O. 1995. Mechanics of porous media, Wiley (ed.).

Fernandes R., Chavant C., Chambon R. 2008. A simplified sec-
ond gradient model for dilatant materials: Theory and numerical
implementation. Int. J. of Solids and Strucutres, 45: 5289–5307.
Gasc-Barbier M., Chanchole S., Bérest P. 2004. Creep behavior of
Bure clayey rock. Applied Clay Science. 26: 449–458.
JiaY., Bian H.B., Duveau G., Su K., Shao J.F. 2008. Hydromechanical
modelling of shaft excavation in Meuse/Haute-Marne laboratory.
Physics and Chemistry of the Earth, 33: 422–435.
Kleine A. 2007. Modélisation numérique du comportement des
ouvrages souterrains par une approche viscoplastique. PhD Thesis,
Inst. Nat. Polytech. Lorraine.
Martin F., Laviguerie R., Armand G. 2010. Geotechnical feed-
back of the new galleries excavation at the ANDRA underground
Figure 1. Experimental and simulated water pressures in rock mass research laboratory – Bure (France). Rock Mech. in Civil and Envir.
during the excavation. Viscosity effects. Engineering: 779–782.
265

Implementation of ANN-based rock failure criteria in numerical simulations
H. Rafiai & A. Jafari

School of Mining Engineering, University of Tehran, Tehran, Iran
Keywords: neural network; numerical modelling
Strength of rock in different loading conditions has been one

of the most important issues in the field of rock mechan-
ics, and many different failure criteria have been proposed
by researchers. However, it has been recognized that these
criteria are usually suitable for particular rock types or cer-
tain modes of failure. These limitations are because of the
presumed forms of the failure criteria which are not flexible
enough to be applied for various conditions.
The idea of using Artificial Neural Networks (ANNs) as a
basis for new generation of rock failure criteria was proposed
by the authors.They usedANNs to predict the strength of seven
different rock types on a wide range of loading conditions. It
was observed that using flexible formulation of ANNs instead
of rigid formulation of conventional failure criteria can lead
to more accurate predictions of strength.
ANNs may be defined as structures comprised of densely
interconnected adaptive simple processing elements called
neurons that are capable of performing massively parallel
computations for data processing and knowledge represen-
tation. Feed-Forward Neural Networks (FFNNs) are especial
type of ANNs in which, inputs are received and simply for-
warded through all the next layers to obtain the outputs.
FFNNs are well suited for function approximation, and are
successfully applied in different fields of rock mechanics.
Hence, these type of ANNs were adopted a basis for rock
failure criteria.
The strength data used in this study for illustration of the
Figure 1. The results of brittle and ductile failure of Indiana
application ofANN-based criteria are those obtained for brittle limestone in triaxial compression tests and the fitted solutions.
and ductile failure of Indiana limestone in triaxial compression
tests. These data show the typical trend observed in transition
from brittle to ductile modes of failure and are frequently used
in literature. Figure 1 shows the measured data points as well of Mohr-Coulomb and Hoek-Brown criteria are almost coin-
as Mogi’s line (σ 1 = 3.4σ 3 ) separating the brittle and ductile cident and cannot satisfactory fit data points. However, the
failure data. ANN-based criterion has reproduced the trend quite well.
The ANN used to fit these data has two neurons in the The values of root mean squared error of the predictive mod-
hidden layer with hyperbolic tangent activations function and els showed that the ANN-based criterion is considerably more
one neuron in the output layer with linear activation function. accurate than conventional criteria.
The portion of data used for training and testing of the ANN The trained ANN uses an explicit formulation to calculate
were 80 and 20%, respectively. The utilized training algorithm the value of output for each value of input:
was Levenberg–Marquardt back-propagation combined with
Bayesian regularization in order to negate any risk of over-
fitting problems. MATLAB software was implemented for
random selection of training sets, initializing the values of
weights and biases, and training the ANN. where σ 1 = the value of major principal stress at failure in
In order to evaluate the relative accuracy ofANN-based fail- MPa; σ 3 = the value of minor principal stress at failure in MPa,
ure criterion Mohr-Coulomb and Hoek-Brown were also used tanh(.) is the hyperbolic tangent function, and the other param-
to fit the data. The fitted curves of different models are also eters are the values of weight and bias of the trained network
shown in Figure 1. It can be observed that the failure envelopes given in Table 1.
266

Table 1. The values of weight and bias of the trained ANN.
1 1 2 2
Weight/Bias w1,1 w2,1 w1,1 w1,2 b11 b12 b21
Value −0.02 −0.06 −59 −57 0.97 −0.05 83
In order to implement the ANN-based failure criterion in

numerical analyses, it is useful to express it as an equivalent
Mohr-Coulomb criterion with variable parameters as:
Figure 2. Relation between the measured values of strength in

Indiana limestone in triaxial compression tests and obtained values
from numerical simulations using Mohr-Coulomb and ANN-based
failure criteria.
where ceq is the instantaneous value of cohesion in MPa; ϕeq

is the instantaneous value of friction angle in radians.
In order to examine the feasibility of incorporating ANN-
based failure criteria in numerical analyses, it was tried to
solve two examples involving Indiana limestone with the finite
difference code, FLAC. The basic idea for introduction of the
new failure criterion to the software was using an equivalent
Mohr-Coulomb criterion and updating the values of cohesion
and friction angle in each iteration of computations according
to Equations 4, 5. The results of the simulations were com- Figure 3. The obtained profiles of plastic zone around circular
pared with those obtained by using the fitted Mohr-Coulomb tunnels excavated in Indiana limestone from numerical simula-
criterion. tions. Solid lines and dashed lines represent the results of using
Mohr-Coulomb and ANN-based failure criteria, respectively.
The first analyzed example was the axisymmetric problem
of triaxial compression test on cylindrical specimens of rock
with height to diameter ratio of 2.5. The numerical tests were and ANN-based failure criteria for σ v values equal to 30 and
performed at σ3 levels which were used in real laboratory tests 60 MPa is shown in Figure 3. The difference between the pre-
on Indiana limestone (Fig. 1). The relation between the mea- dictions of the two criteria is negligible for low values of
sured and predicted values of strength using the fitted Mohr- vertical stress. However, as the level of stresses increases, the
Coulomb and ANN-based criteria is shown in Figure 2. It can difference becomes more apparent.
be observed that the predictions using ANN-based failure cri- Although there is no reference solution for this problem to
terion are in excellent agreement with measured values, which evaluate the relative accuracy of the results, it is expected that
indicates the successful implementation of this criterion in the predictions of the ANN-based criterion were more accu-
numerical simulations. rate than those of Mohr-Coulomb criterion because it showed
The second example which involves more compli- higher accuracy for the first example.
cated stress distribution and more practical condition was a The results of these simulations showed that ANN-based
plane strain problem of a circular tunnel in an anisotropic failure criteria can be successfully incorporated in numerical
in-situ stress field with the ratio of horizontal stress, σ h , analyses and lead to more accurate results. The promising
to vertical stress, σ v , of 1/3. The problem was analyzed for findings of this study can lead to the beginning of the practical
the values of vertical stresses equal to 10 to 60 MPa. The use of ANN-based failure criteria in numerical simulation and
obtained profiles of plastic zone predicted by Mohr-Coulomb design of real-world rock structures.
267

Stochastic model generation for discontinuous rock mass media and numerical analysis
using the NSCD method (case study: Saint-Béat gallery in the southwest of France)
Ali Rafiee
Zanjan University, Iran
Muriel Gasc-Barbier
CETE du Sud Ouest, LRPC de Toulouse, France
Marc Vinches
Alès School of Mines, France
Keywords: numerical modeling, rock joints, stability analysis
1 EXTENDED ABSTRACT
The processes of the stochastic generation of a discontinuous

rock mass model are presented in this paper, using the orien-
tation data measured directly in the abandoned underground
gallery of Saint-Béat castle, in the southwest of France. Since
this gallery was excavated in a marble mass next to a major
regional fault, all its walls and the roof are perfectly stable
even without reinforcement.
The fracture measurements are carried out in order to inves-
tigate the stability and design of a 1 km long tunnel, which will
be a part of the new ring road of the town of Saint-Béat, The
town is built on a narrow pass along the Garonne River at the
bottom a steep slope.
The measured fractures in the gallery are grouped in four
main families and then the parameters such as orientation and
mean spacing are calculated for each family. These parameters
are used as input data within a numerical development of the
AutoCAD software in order to generate 3D stochastic discrete
models of jointed rock mass. In the present study, seven 3D
models are generated using different methods of integration
(hierarchical or random methods), different values for fracture Figure 1. Instable blocks in the tunnel for deteriorated contact
persistence and also different tolerances for fracture orienta- conditions.
tion. These discrete models are used to analyse the mechanical
behavior of the tunnel.
The stability analysis of the models is carried out using the collections of rigid blocks under unilateral constraints and
LMGC90 numerical code that is based on the non-smooth con- frictional Coulomb’s law.
tact dynamics (NSCD), a method developed in the Mechanical The first results obtained from these seven models about the
and Civil Engineering Laboratory, of Montpellier University stability in the abandoned gallery are a basis for the studies of
(France). The NSCD method simulates rock masses as large the projected tunnel.
268

Rock fracture propagation from mechanical, thermal and hydraulic loadings
M. Rinne
Aalto University School of Engineering, Espoo, Finland
B. Shen
CSIRO Earth Science and Resource Engineering, Brisbane, Australia
O. Stephansson
Helmholtz Center Potsdam, GFZ German Research Centre for Geosciences, Potsdam, Germany
Keywords: numerical modelling, hydraulic fractures, fluid flow, rock failure, geothermal
1 INTRODUCTION F-criterion, in an arbitrary direction (θ) at a fracture tip there

exists a F-value, which is calculated by
Rock failure is often controlled by fracture initiation, propa-
gation and coalescence, especially in hard rock where explicit
fracturing rather than plasticity is the most dominant factor
during failure process. It is therefore desirable to predict the
explicit fracturing process when investigating the rock mass
stability for engineering purposes. The field of rock mechan-
ics is being advanced and widened to address the complex where GIc and GIIc are the critical strain energy release rates
behaviour of mechanical, thermal and hydraulic responses for mode I and mode II fracture propagation; GI (θ) and GII (θ)
of rocks for industries such as deep geological disposal of are strain energy release rates due to the potential mode I and
nuclear waste, geothermal and underground LNG storage. mode II fracture growth of a unit length.
Consequently, it is required that the fracture modelling tool The direction of fracture propagation will be the direc-
can predict not only the mechanical processes but also cou- tion where F reaches the maximum value. If the maximum
pled F (fracture) – T (thermal) – H (hydraulic) processes in F reaches 1.0, fracture propagation will occur.
rock masses.
This paper introduces a code, FRACOD, for modelling frac-
ture initiation and propagation in rocks. It also describe the
recent developments in FRACOD for modelling the coupled 3 THERMO-MECHANICAL COUPLING
Fracture (F) – Thermal (T) and Fracture (F) – Hydraulic flow
(H) processes. Several applications cases have been included An indirect method is considered for simulation of the
in the paper that demonstrate the effectiveness of the code temperature distribution and thermal stresses due to inter-
in handling the complex rock fracturing processes due to nal and boundary heat sources. The indirect approaches
mechanical, thermal and hydraulic loadings. have been found efficient in modelling poroelasticity and
thermal-poroelasticity using boundary element methods. The
two-dimensional fundamental solutions for temperature and
stresses induced by a continuous point heat source in thermo-
2 THEORETICAL BACKGROUND OF FRACOD elasticity are used.
The basic principle of the indirect boundary element
FRACOD is a two-dimensional code which is based on approach for thermoelastic analysis is the assumption that a
the Displacement Discontinuity Method (DDM) principles. fictitious line heat source exists at each element. The strengths
It predicts the explicit fracturing process including fracture of the line sources are unknown and should be determined
sliding/opening, fracture initiation and fracture propagation based on the boundary conditions. For example, if the tem-
in rocks. perature at all boundary elements is zero, the combined effect
Both tensile and shear failure are common in rock masses. of all the line heat sources on the boundary elements should
Therefore, to effectively predict rock fracture propagation, result in a zero temperature. Once the strength of each ficti-
FRACOD adopted a fracture propagation criterion developed tious heat source is determined, the temperature, thermal flux,
by Shen and Stephansson (1993) for both mode I and mode II and thermal-induced stresses and displacements at any given
fracture propagation, namely the F-criterion. According to the location in the rock mass can be calculated.
269

4 FRACTURING-HYDRAULIC FLOW COUPLING
In fractured hard rock such as granite, fluid flow occurs pre-

dominantly through explicit fractures rather than intact rock
due to the low permeability of the intact rock. Fluid pres-
sure in rock fractures may cause rock fracture movement,
increase fracture aperture or even cause fracture propagation.
On the other hand, fracture movement and propagation will
change the fracture hydraulic conductivity and create new flow
paths. The dynamic interaction between fracture mechanical
response and fluid flow is critically important in studying the
coupled fracture – hydraulic flow (F-H) processes.
The explicit approach is used in F-H coupling, where both
fluid flow and mechanical response are simulated using a
time marching iteration process. Comparing with the implicit
approach, the explicit approach is mathematically simple and
easy to adopt complicated boundary conditions and chang-
ing model conditions. However it often requires significantly
longer computational time as small time steps are required to Figure 1. Cooling of a borehole. Predicted process of thermal
achieve convergence for flow solution. cracks growth after 106 seconds of cooling.
The mechanical calculation (including rock deformation
and fracture propagation) is done using the Displacement
Discontinuity (DD) method with an iteration scheme for mod-
elling fracture propagation processes. The fracture fluid flow
calculation is conducted through the time-marching iteration
based on the cubic law.
5 APPLICATION EXAMPLES
5.1 Cooling fractures in borehole wall

We consider a borehole with radius r = 0.1 m in a geothermal
reservoir with an in situ rock temperature T0 = 200◦ C. The
borehole wall is cooled by drilling fluid and maintained at
temperature Tw = 80◦ C.
Figure 2. Simulated rock fracturing process due to high pressure
The numerical model for this problem has 60 constant ele-
fluid injection.
ments on the borehole boundary. Six different cooling times
were considered: 10, 102 , 103 , 104 , 105 and 106 seconds.
During the last step of the modelling, fracture initiation and
drives the short fracture at the borehole wall to propagate
propagation were considered in the model. Figure 1 shows the
toward the preexisting fractures and eventually coalesce with
predicted pattern of thermal cracks in the immediate vicinity
these fractures. The pre-existing fractures then propagate
of the borehole wall due to the cooling effect. Note that the
under the high fluid pressure.
initial major fracture initiations occur at the borehole wall.
These fractures then propagate in radial direction and coalesce
with other newly initiated short fractures, eventually forming ACKNOWLEDGEMENTS
several long radial fractures.
The coupled functions were developed with support from
5.2 Hydraulic fracturing in rock mass with preexisting CSIRO, SKEC, KIGAM, LIAG, Geomecon GmbH, TEKES
fractures and FRACOM. Major contributions from H.M. Kim, E.S.
Park, M. W. Wuttke, T. Backers, T. Y. Ko, S. C. Lee, J. Y.
A typical case of hydraulic fracturing is simulated to demon-
Kim, T. K. Kim and J. M. Lee are gratefully acknowledged.
strate the effectiveness of the coupled F-H function in
FRACOD. A borehole is drilled in a rock mass with several
isolated pre-existing fractures (Figure 2a). A high fluid pres- REFERENCES
sure is then applied in the borehole to propagate the existing
fractures. Shen, B. & Stephansson, O. 1993. Modification of the G-criterion
The simulated fracture propagation pattern after 0.15 sec- of crack propagation in compression. Int. J. Engineering Fracture
onds is shown in Figure 2. In this model, the fluid pressure Mechanics. 47(2), 177–189.
270

Meshless simulation of rock mechanics problems by Element Free Galerkin Method
H. Salari-Rad & M. Rahimi Dizadji

Rock Mechanics Department, Faculty of Mining & Metallurgical Engineering,
Amirkabir University of Technology (Tehran polytechnic), Tehran, Iran
Keywords: Meshless Methods, Element Free Galerkin Method (EFGM), Rock Fracture Mechanics, Three point bending
(3PB) load test
1 INTRODUCTION
The computational methods in rock engineering and geome-

chanics branch are applied to solve many problems such as
stress-strain and fracture mechanics analysis, which grow
ever more interceding. Although, the mesh-base methods
such as FEM, BEM and DEM are robust and well estab-
lished techniques, they have some deficiencies due to their
mesh-base structure. In order to solve the problem of the dis-
cretization interfering with arbitrary crack extension we have
to remove meshing from the simulation process. Using the
meshless methods, the crack can propagate through a sys-
tem of nodes to precisely capture the stress singularity. In
this research, some problems in rock mechanics are modeled
using EFG code developed in MATLAB programming envi-
ronment. The obtained results from the developed Element Figure 1. (a) Problem statement and (b) the meshless nodal distri-
Free Galerkin (EFG) code are compared with other numerical bution for EFG stress analysis.
and experimental researches. In order to carry out a meaning-
ful comparison, the input parameters of EFG code selected
based on available experimental and numerical data in the
references and also the values are in units.
2 NUMERICAL ANALYSIS OF ROCK MECHANICS

PROBLEMS
2.1 A circular tunnel in homogeneous medium

In rock mechanics in the first step we have generalized the Figure 2. Computed values of (a) vertical stresses and (b) horizontal
Kirsch problem to stress analysis of a tunnel or a shaft exca- stresses by EFG developed code.
vated in rock. As a reliable validation, the EFG developed code
is validated by solving Kirsch problem as a basic problem
in rock mechanics in stress analysis of underground excava- 2.2 Rock fracturing in uniaxial tensional test
tions with circular opening. The model geometry and meshless
discretization are shown in Figure 1. Fracture mechanics can be applied to many of rock engineer-
The EFG computed results of tangential stress variations ing issues where hydraulic fracturing, rock slopes, explosion,
by the distance from the tunnel center, are in good agree- rock cutting and hot dry rock geothermal wells are closely
ment with the closed-form solution. Stress analysis contours related to the strength, stability and fracture of rock materials
around the tunnel are depicted in Figure 2. As can be seen in and structures. In order to show capability of EFG developed
this figure, the vertical stress in tunnel roof and tunnel walls code as a tool in solving computational geomechanics, it is
is predicted −1 and 3 units, respectively. These results are used to predict the fracture propagation in some rock mechan-
coinciding with the closed-form methods as the following: ics test under linear elastic fracture mechanics (LEFM). The
algorithm of fracture propagation is depicted in Figure 3.
Erdogan and Sih performed uniaxial tension tests on
isotropic Plexiglas sheets containing a central crack. Also,
Chen et al. reproduced these tests with Boundary Element
271

Figure 3. The algorithm of crack propagation process.
Figure 5. Predicted fracture path using (a) experimental test (b)

EFG developed code in MATLAB and (c) adopting the results of
crack trajectory by EFG-developed code and experimental data.
Liu et al. researches with the same domain geometry and

boundary conditions and the angle of fracture initiation is
calculated. The results obtained from EFG developed code
are compared with that of experimental results under the
same boundary and loading conditions in Figure 5. The two
fracture trajectories are coinciding surprisingly.
3 CONCLUSIONS
Figure 4. Variation of crack initiation angle θ0 with the crack angle β
under uniaxial tension (Experimental results, BEM and EFGM). Meshless EFG method has shown a great promise in rock
mechanics and rock engineering applications due to their flex-
ibility in solving those problems with discontinuities within
Method (BEM) with the same domain geometry and bound- fracture mechanics and stress analysis. The evaluated frac-
ary conditions. The angle of fracture initiation in different ture growth trajectory using EFG developed code in uniaxial
crack orientation angle in uniaxial tensional test is calculated tensional test and also the three point bending (3BP) test
and is depicted in Figure 4. Regarding to this figure, the cal- demonstrated the ability of the code for solving fracture prop-
culated values of initiation angles are closer to experimental agation problems in geomechanics. As a future research, this
results than the numerical boundary element method results. code can be extended to the anisotropic materials which con-
For example, in the case of β = 30◦ , BEM has about 5% error tain numerous discontinuities like faults and joints within a
and EFG code has about 1% error relative to the experimental rock mass to make it more applicable in rock engineering
data. practical problems.
2.3 Rock fracturing in 3-Point Bending (3PB) test

In order to predict the crack propagation procedure in 3PB
test were simulated using the developed EFG code based on
272

Applications of two dimensional discontinuous deformation analysis
G.H. Shi
DDA Company, Belmont, California, USA
Keywords: numerical modeling; stability analysis
In the field of practical rock engineering, there are two

independent computations: continuous computation and limit
equilibrium computation. Limit equilibrium is still the
fundamental method for global stability analysis. For any
numerical method, reaching limit equilibrium requires large
displacements, discontinuous contacts, precise friction law,
multi-step computation and stabilized time-step dynamic
computation. Therefore three convergences are unavoidable:
convergence of equilibrium equations, convergence of open-
close iterations for contacts and convergence of the contact
forces of dynamic computations. This paper focuses mainly on
applications of two-dimensional discontinuous deformation
Figure 1. Dam foundation failure with increased water pressure and
analysis (DDA). The applications show DDA has the ability to reduced friction angle.
reach limit equilibrium of block systems. This paper presents
dam foundation damage computation and slope reinforcement
computation, where the block sliding is a main issue. This
paper also presents underground chamber rock stability and
bolting computations.
Figure 1 shows the mode of failure of a gravity dam founda-
tion by increasing total water pressure and reducing the friction
angle. Based on the mode of failure, the sliding blocks are
chosen. Based on the assumed sliding blocks, following
different factors of safety are computed:
Case 1 is limit equilibrium method. Here normal loads are
applied, the factor of safety of the chosen sliding blocks are
Figure 2. Result of dam foundation of case 1 to case 5.
computed. The factor of safety is 1.94.
Case 2 is the fictitious force method. Keeping the stability,
increase the water pressure as much as possible. The factor of The computation results of all case 1 to case 5 are stable.
safety is the ratio of applied total water pressure and the total All case 1 to case 5 didn’t reach limit equilibrium. Except case
normal water pressure. The factor of safety is 2.80. 1, all case 2 to case 5 are close to limit equilibrium. Figure 2
Case 3 is the strength reduction method. Keeping the stabil- shows the results of all case 1 to case 5. The computation
ity, reducing the friction angle as much as possible. The factor requirement of case 2 to case 5 is highly difficult because
of safety is the ratio of the tangent of real friction angle and these cases are very close to limit equilibrium or very close
the tangent of reduced friction angle. The factor of safety is to failure. The computation of case 2 to case 5 also requires
2.91. stability and three convergences.
Case 4 is the limit equilibrium method under the fictitious The rock falling of an underground chamber is simulated
force. All parameters are the same as case 2. The computation using two-dimensional DDA, as shown in Figure 3.
is also identical with case 2. After increasing the water load, The rock bolting computation of the underground chamber
the computed factor of safety is still greater than 1.0. The is carried out using two-dimensional DDA. Figure 4 shows the
resulting factor of safety is the factor of safety of case 2 plus location of each rock bolts. The length of the rock bolts are
the marginal factor of safety 1.14 − 1.0 = 0.14. 10 meter and 12 meters alternatively. The bolt spacing is one
Case 5 is the limit equilibrium method under the reduced meter. The diameter of the rock bolts is 32 mm. Figure 12
friction angle. All parameters are the same as case 3. The shows the time depending resulting bolting forces (unit ton)
computation is also identical with case 3. After reducing the of bolt number 1 to 30. Figure 13 shows the time depending
friction angle, the computed factor of safety is still greater resulting bolting forces (unit ton) of bolt 31 to 60. The bolting
than 1.0. The resulting factor of safety is the factor of safety forces were not changed in 6 digits after the dismal point under
of case 3 plus the marginal factor of safety 1.39 − 1.0 = 0.39. large enough time steps.
273

Figure 5. Time depending resulting bolting forces of bolt number
Figure 3. Rock fall computation of an underground chamber under
1 to 30 counting from top centre down and right to left in each level
horizontal initial stress 1800 tons per square meter and vertical initial
of the underground power house section.
stress 1500 tons per square meter by 2D DDA.
Figure 6. Time depending resulting bolting forces of bolt number

31 to 60 counting from top centre down and right to left in each level
Figure 4. Rock bolting computation of an underground chamber of the underground power house section.
under horizontal initial stress 1800 tons per square meter and vertical
initial stress 1500 tons per square meter by 2D DDA.
274

Stochastic boundary element method for fracture reliability analysis
K.J. Shou & H.Y. Yu

Department of Civil Engineering, National Chung-Hsing University, Taichung
Keywords: numerical modelling, rock failure
The analysis of random stress-strain fields and the rock struc- of crack simulation. The integral equations and solutions
ture reliability are important in the rock engineering design. are established by the partial derivation of the original inte-
Since the joint systems of rock mass are always with uncer- gral equations and solutions. And the fundamental solutions
tainties, it is very difficult to solve the random stress-strain were derived and implemented to a computer program. For
field of a rock structure in a jointed rock mass. Stochas- calibration, an example problem was analyzed by the devel-
tic finite element methods have been applied to solve the oped model, and the results were compared with those from
problems of random stress-strain fields. However, boundary Monte Carlo analysis. The results show that developed model
element methods provide alternatives with the advantages can accurately analyze fracture reliability problems. And the
including efficiency and simplicity. In this study, a stochastic developed model is much more efficient than the Monte Carlo
boundary element was developed for the analysis of fracture analysis approach. The developed stochastic boundary ele-
reliability. Displacement discontinuity method, an indirect ment method can be applied to the reliability analysis of rock
boundary element method, was adopted due to its suitability structures.
275

Numerical anisotropic fracture mechanics modelling in crystalline rock
T. Siren, & K. Kemppainen

Posiva Oy, Eurajoki, Finland
B. Shen
CSIRO Earth Science and Resource Engineering, Queensland, Australia
M. Rinne
Fracom Ltd, Kirkkonummi, Finland
Keywords: numerical modelling, rock stress, nuclear repository, case studies
1 INTRODUCTION under thorough research for many years, but there are still
some uncertainties, related to the in situ stress and to the rock
1.1 Fracture propagation code (FRACOD) spalling strength. (Siren et al. 2011a).
To answer these questions, an in situ experiment called
Most numerical methods for rock engineering are based on
Posiva’s Olkiluoto Spalling Experiment (POSE) was started
methods that cannot genuinely describe the fracture initiation
(Aalto et al. 2009). The objective of POSE is to establish the
and propagation. Fracture propagation code (FRACOD) is a
in situ spalling strength of the rock in Olkiluoto and also to
two-dimensional Displacement Discontinuity Method (DDM)
establish the state of in situ stress at the −345 metre depth
computer code that was designed to simulate fracture initiation
level.
and propagation.
The code development was started in early nineties by
Shen & Stephansson (1993). The FRACOD code is based on
the principles of the Boundary Element Method (BEM). Since 2 DEVEPLOMENT OF ANISOTROPY OF
the BEM doesn’t cope with problems with a fracture – that ROCK MEDIUM
is, two surfaces coinciding with each other – Crouch (1976)
developed a method called the Displacement Discontinuity In FRACOD, the fracture initiation occurs when the combi-
Method (DDM). nation of two principal stresses reaches a critical value. More
FRACOD uses a modified G-criterion, which is also known closely, the tensile and shear stresses and strengths are used to
as the F-criterion, introduced by Shen & Stephansson (1993). determine the initiation of a new fracture. The fracture prop-
The problem with the original G-criterion is that, in some agation is, however, determined by using fracture toughness
cases, it predicts shear failure when the failure, in fact, is parameters.
tensile. However, the F-criterion can predict the propagation To take anisotropy of the rock into account, the parame-
of a crack system in both mode I and mode II, which is crucial ters (σt , c, φ, KIC , KIIC ) have an elliptical variation from θ to
for such material as rock in which shear and tensile failures θ + 90◦ . For each point, calculations are done in all anisotropy
are common. directions, and the fracture will initiate in the direction of
The latest development, described in later chapters, intro- maximum tension or shearing, and will propagate in the
duced in the code allows the possibility to simulate anisotropy direction where the maximum F value is reached.
of rock medium using strength anisotropy instead of just
explicit joints and bedding planes. Anisotropy related to new
fracture initiation is described by direction dependent Mohr- 3 CASE EXAMPLE OF POSE
Coulomb and direct tensile strength criteria.The fracture prop-
agation function is converted to anisotropic by formulating The input parameters for the case example are determined
F-criterion to be direction dependent. by using existing test results for pegmatitic rock (PGR),
which is assumed to be isotropic, and for migmatitic gneiss
(MIGN.GN), which is assumed to be anisotropic. Both
1.2 Posiva’s Olkiluoto Spalling Experiment (POSE) types are assumed to be homogeneous and linearly elastic.
For migmatitic gneiss, individual rock strength and fracture
Currently in Olkiluoto, the construction of the underground strength parameters are determined for two perpendicular
rock characterisation facility for the final disposal of spent foliation directions. The input parameters, as listed in Table 1,
nuclear fuel named ONKALO is on the way. The site has been are used.
276

Table 1. Values of the mechanical parameters of the intact rock and Table 2. Results of the models.
fractures used as input in FRACOD.
1st hole 2nd hole
Rock type PGR M.GN. M.GN. Model
Anisotropy Isotrop. Paral. Perpendic. direction – Width Depth Width Depth
depth mm mm Type mm mm Type
Young’s modulus E (GPa) 55 55 55
Poisson’s ratio ν 0.20 0.20 0.20 Peg – 1 m 44 409 3Nm,Cs 148 593 Ns,2Cm,Cx
Anisotropy dip – 52◦ 52◦ Peg – 3 m 40 295 3Nm 241 421 Nm + Cx
Anisotropy direction – 175◦ 175◦ 84◦ – 1 m 71 295 Cl,2Cm 48 296 3Cm
Cohesion c (MPa) 12.9 12.4 13.8 84◦ – 3 m 55 294 3Cm 48 295 3Cm
Friction angle φ (MPa) 47◦ 45◦ 47◦ 99◦ – 1 m 98 468 2Nm,2Cl 85 468 Nm + Cl,2Cl
Tensile strength σT,I 12 10 14 99◦ – 3 m 94 352 2Ns,Cl 109 352 Nm + Cl,Cl
√ 114◦ – 1 m 82 409 Nm,2Cl,Cm 78 410 Nm + Cl,2Cl
Fracture toughness I (MPa √m) 1.96 1.87 3.05
Fracture toughness II (MPa m) 3.30 3.00 3.86 114◦ – 3 m 97 410 Ns,3Cl 75 410 Nm,2Cl
Fracture cohesion c (MPa) 10 10 10
F. normal stiffness kn (GPa/m) 20000 20000 20000 Abbreviations: N = Notch, C = Crevice
F. shear stiffness ks (GPa/m) 2000 2000 2000 Scale: s = Small (depth < 50 mm), m = Medium (50 . . . 100 mm),
Major principal stress σ1 25.1 166◦ (horizontal) l = Large (>100 mm), x = Extra (400 . . . 500 mm).
Intermediate principal stress σ 217.1 256◦ (horizontal)
Minor principal stress σ3 12.3 (vertical)
close-ups of the pillar between the holes. From the figures
it can be observed that small notches, which are about the size
of the grid density, form at the right side at 2nd cycle and the
large wedge on left forms last.
3.2.2 Spalling
Figure 1. The fracture propagation in stages (numbered) from left Fracture growth happens in all of the models with the POSE
to right with the mean anisotropy direction (99◦ ). Fractures marked holes. In the models with the anisotropy of 84 degrees, the
with grey are slipping or open. cracks form only crevices, but in the rest of the models the
fractures form notches—or in other words, spalling. The
3.1 In situ stress results are assembled in Table 2. In the table the number of
crevices and spalling is stated as observed from the results.
Several stress measurements has been carried out in the
ONKALO in the past few years and it is possible to determine
different stress domains to describe the stress distribution at 4 DISCUSSION
different depths. However new LVDT-cell measurements have
been applied in the POSE niche with promising results and The case example shows results that seem to be realistic. The
therefore new result have been used in this study. local geological features that were neglected in this study,
however can affect to the failure plane so that the failure will
3.2 Results proceed through existing weaker surfaces that penetrate the
holes.
3.2.1 Results of the assumed anisotropy direction
Migmatitic gneissic rock with interpreted anisotropy in the
direction of 99◦ results in some fracture propagation at both 5 CONCLUSIONS
sides of pillar at the depth of both 1 metre and 3 metres. The
results indicate only minor differences between the depths The results show that the fracture propagation is very sensi-
and that spalling will occur in both simulated depths. Notches tive to changes in the anisotropy direction, friction angle and
that form are deeper in anisotropic models than in isotropic cohesion. However, the fracture toughness is observed not to
models. be a very sensitive parameter. Modelling results suggest minor
The fracture propagation in stages with the mean anisotropy spalling on the pillar surface while observations from the field
direction (99◦ ) is presented in Figure 1. The figures are shows slight fracture slipping of an existing fracture.
277

Contact description in numerical simulation for rock mechanics
L. Sun & J. Zhao

Laboratory of Rock Mechanics, École polytechnique fédérale de Lausanne (EPFL), Lausanne, Switzerland
G.F. Zhao
School of Civil and Environmental Engineering, The University of New South Wales, Sydney, Australia
Keywords: numerical modeling, rock joints, dynamics, rock properties
In this paper, methods and technology of contact description Contact detection is an important issue which is the prereq-
are shortly reviewed. A new spherical contact model using uisite to perform the contact response. Actually detection of
numerical manifold method is presented to describe the col- contact between arbitrary shapes is very difficult and it is still
lision between two balls. This algorithm is based on penalty the puzzle in modern applied mathematics. In modern con-
theory and open-close iteration. The test results show that it is tact detection algorithm, contacts are detected in two ways: a
capable to simulate both elastic and inelastic collision. With post-detection (after the collision occurs) or a pre-detection
the given maximum interpenetration rate and coefficient of (before the collision occurs).
restitution, the penalty number can be chosen wisely. The benefits of the pre-detection algorithm are increased
Analytical contact theory dates back to 1881 by Heinrich fidelity and stability. But it needs to integrate the physical
Hertz who tried to solve the problem involving contact simulation algorithm with the detection algorithm and this is
between two elastic bodies with curved surfaces. In the very difficult. The main benefits of the post-detection method
next one hundred years, the framework of contact mechan- are as follows. In this case, the contact detection algorithm
ics was established step by step, for example, the well-known need not be aware of the myriad physical variables; a sim-
DMT model and JKR model which are used to represent ple list of physical bodies is fed to the algorithm, and the
adhesive elastic contact. For numerical implementation, uni- program returns a list of intersecting bodies. This means
lateral constraints are added to restrict the motion of bodies the contact detection algorithm does not need to understand
but this process is challenging especially including stress friction, elastic contact and even motion and deformations. In
analysis. addition, the post algorithm is in effect one dimension simpler
The second model is fully based on physical laws (e.g. than the priori algorithms. The pre algorithm must deal with
Newton’s and Coulomb’s) and is mainly used in computa- the time variable, which is absent from the post algorithm. On
tional graphics. To achieve high efficiency, this model usually the other hand, the post algorithm causes problems in the fixed
deals with rigid bodies and gives less accuracy. There are time step, where intersections, which may not be physically
mainly impulse-based method and constrain-based method. correct, need to be corrected.
The former models all forces between bodies through a series Open-close iteration algorithm was proposed by Shi (1991).
of impulses and it cannot completely handle a prolonged static One of its advantages is to overcome the disadvantage of the
contact. The latter works by applying a series of constrains on post-detection algorithm. This algorithm adopts an iteration to
rigid bodies but its simulation is still plausible. check the fixing positions of bodies. And this guarantees that
The so-called parameterized model is widely used in energy all the contact responses really exist during the whole time
principle based numerical methods (e.g. FEM and Manifold step.
Method). There are mainly two ways to parameterize the Using the penalty method and open-close iteration algo-
contact: Lagrange multiplier method and penalty method. rithm, a spherical contact model is presented. This model
The former introduces Lagrange multiplier λ into the energy adopts the numerical manifold method (NMM) to establish
system to describe the reaction force. The new variables the system equation.
are treated as unknowns which are solved together with For a 3D problem, 12 unknown variables are used to
displacements. The latter minimizes the potential energy with describe the motion and deformation of a body. They are trans-
an additional penalty term. Although penalty method does lation (u0 , v0 , w0 ), rotation (θx , θy , θz ), normal strain (εx , εy , εz )
not require additional independent variables, the parameter, and tangential strain (γxy , γyz , γzx ). So the displacement of an
penalty number α, needs to be defined by users and it influ- arbitrary point in a body is:
ences the simulation a lot. These contact models above have
different applicability, accuracy and efficiency.
The choice of contact models is a compositive considera-
tion depending on computational scheme, quantity of poten- The dynamic system equation without damping is:
tial contacts, simulation time, material properties, required
accuracy, etc.
278

Figure 1. Spherical contact model.
The contact condition is given as:
Figure 2. Effect of the penalty number on the maximum interpen-

Using the penalty method, the object system equation in every
etration rate and COR.
time step is:
In the simulation, the penalty number is the only artifi-

cial parameter. So, different α are tested to test the effect of
this parameter. As a reference, α is given as a multiple of the
This is the system equation when the contact is closed. In other elastic module of the collision subjects (Eq. 6).
words, at the end of the time step the contact still exists. If the
contact is open at the end of the time step, the system should
adopt the format without the penalty term. This process is
controlled by the open-close iteration algorithm.
When relevant time step is as small as 10−3 (when V = 1m/s,
With the contact model above, a frontal elastic collision
t = 10−3 s), the simulation is stable. This time scale is quite
between two metal balls is simulated (Fig. 1). In physi-
large enough for dynamic simulation. As shown in Figure. 2,
cal theory, this collision follows the law of conservation of
the simulation follows the law of conservation of momentum.
momentum. The total kinetic energy of the system depends
The total kinetic energy varies with different penalty numbers.
on many factors. The coefficient of restitution (COR) Cr is
With certain values, a perfect elastic collision is achieved.
used to express the change of the kinetic energy (Eq. 5). When
During the interval [1, 100], the COR is about 0.8. The max-
Cr = 1, it is the perfect elastic collision. When Cr = 0, it is the
imum penetration rate is small (<0.1%) when the penalty
perfect non-elastic collision. Usually, Cr is between 0 and 1.
number is large enough. So the interval [1, 100] is a good
approximator for the penalty number.
279

Numerical investigation of the potential applicability of surface deformation measurement

by InSAR technology for monitoring CO2 behavior in deep underground
Hajime Yamamoto
Taisei Corporation, Yokohama, Japan
Takumi Onuma
JGI, Inc., Tokyo, Japan
Tsukasa Kumagai
JGC Corporation, Yokohama, Japan
Jonny Rutqvist
Lawrence Berkeley National Laboratory, Berkeley, USA
Keywords: numerical modelling, fluid flow, case studies, monitoring, field measurements
CCS (carbon dioxide capture and storage) is a promising TOUGH-FLAC (Rutqvist et al, 2002), that links the mul-
approach for reducing the greenhouse gas content in the tiphase fluid flow simulator TOUGH2 with a commercial
atmosphere, through capturing carbon dioxide (CO2 ) from geomechanical code FLAC3D . As numerical grid of the three
large emission sources and injecting it into reservoirs (such dimensional region of 10 km × 10 km wide and 4 km deep
as deep saline aquifers). Large-scale storage projects will around the horizontal injection well KB-501 was used. The
likely involve very long-term storage of huge amounts of geological model and properties basically follow Rutqvist
CO2 , potentially exceeding hundreds of millions of tonnes et al. (2010). A sensitivity study was performed for investigat-
(Mt). Effective site monitoring is crucial for demonstrating ing the effect of model parameters on surface deformation pat-
the effectiveness and safety of a site, calibrating predictive terns, and for investigating how coupled fluid-geomechanical
models, as well as providing warning and identify measures models can be calibrated in the light of surface-deformation
of any potential surface leakages (Chadwick, 2010). temporal and spatial changes detected in detail by DInSAR.
Differential interferometry SAR (DInSAR) is a promising As the results, the coupled fluid flow and geomechanical
remote sensing technique for detecting surface deformation on model successfully simulated the temporal and spatial surface
the order of millimeters. The technique has been applied to the deformation, and illustrated and demonstrated that:
CO2 injection site at the In Salah CO2 storage project, Algeria,
1) As expected, information on the CO2 flow is indirectly
and successfully demonstrated its usefulness and effectiveness
transferred to the surface displacement through pressure
for measuring, in remarkable detail, spatial surface defor-
propagations. Even if the CO2 flow were highly localized
mations due to pressurization of the reservoir during CO2
along with reservoir heterogeneities such as faults, surface
injection (Onuma et al., 2010). The detected surface deforma-
deformation would not be sharp but smoothened through
tion patterns are anisotropic, likely due to fluid-geomechanical
the overburden.
effects of fractures and faults within the site.
2) Temporal and spatial changes in surface deformation data
In the In Salah project, since August 2004, CO2 removed
acquired by DInSAR potentially provide a unique opportu-
from natural gas production has been injected from three injec-
nity for inferring reservoir spatial permeabilities, including
tion wells into the storage reservoir formation, at a depth of
anomaly features such as hydraulic compartments (low-
about 1.8 km and at a rate of nearly 1 million tonnes per/year.
permeability barriers), which are important economic
The surface deformation detected by a DInSAR (Figure 1)
performance factors for reservoirs storing CO2 .
shows that the uplifts occur near the injection well with a
surface uplift rate up to 5 mm/year.
For simulating surface deformation due to CO2 injec-
tion, we use a coupled fluid-geomechanical simulator,
Figure 1. Contour maps showing the spatial distribution of vertical

displacement (a) on the ground surface near a injection well (KB501) Figure 2. Surface deformation history near an injection well with
by DInSAR (Onuma et al., 2010), and (b) that by the simulation model. the simulation results from a sensitivity study.
280

An introduction to Distinct Lattice Spring Model (DLSM)
G.F. Zhao & N. Khalili

School of Civil and Environmental Engineering, The University of New South Wales (UNSW), Sydney, Australia
J. Zhao
Laboratory of Rock Mechanics, Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland
Keywords: numerical modelling; dynamics; rock failure
1 INTRODUCTION randomly distributed bonds. The bond here is a virtual concept

rather than a real existence. In this work, a micromechani-
There are a large number of numerical methods which have cal model which takes into account the advantages of VMIB
been applied to rock mechanics and rock engineering. Yet, and Lattice Spring model is proposed and named as the real
none can tackle all the problems encountered in rock mechan- multi-dimensional internal bond model (RMIB).
ics and rock engineering. This is particularly the case in
dealing with phenomena emanating from the micro struc-
ture such as sensitivity to size and distribution of flaws and 3 DISTINCT LATTICE SPRING MODEL (DLSM)
cracks and their evolution with imposed deformation and
loading. Indeed, a successful numerical method must be able to In DLSM, material is discretized into mass particles with dif-
fully consider the microstructure of a material when studying ferent sizes. Whenever the gap between the two particles is
its macroscopic mechanical properties and failure behaviors. smaller than a given threshold value, the two particles are
These requirements have resulted in renewed and increased linked together through a bond between their center points
interest in discrete or framework/lattice numerical model- , which consists of normal and shear springs (see Fig 2(a)).
ing techniques in recent years. In this paper, we introduce The particles and bonds form a network system represent-
a new micromechanical numerical model: Distinct Lattice ing the material. The system equation of motion is solved
Spring Model (DLSM) (Zhao et al, 2010 and Zhao, 2010), by using the explicit central finite difference scheme (see
which is proposed to overcome some of the limitations of the Fig 2(b)). Different from the conventional LSMs, the shear
existing discrete micromechanical models and may become spring is introduced making possible handling of problems
an alternative tool for rock mechanics study. The context of with a variable Poisson’s ratio. For the shear spring, the rela-
this paper covers the theoretical basis, formulation, verifi- tive shear displacement between the two particles is obtained
cation and application of the model using several numerical through a local strain based method which ensures that the
examples. model preserves the rotational invariance of LSM consist-
ing of normal springs only, details can be found in (Zhao
et al, 2010; Zhao 2010). Another distinct feature of DLSM is
2 REAL MULTI-DIMENSIONAL INTERNAL that during calculation, the micromechanical parameters are
BOND MODEL automatically calculated based on formulation provided in the
RMIB model.
In recent years, a number of attempts have been made to
develop the so-called micromechanical models by explic-
itly accounting of the micro-discontinuous structures in the
medium. The first micromechanical model in this context
may be attributed to the pioneering work of Gao and Klein
(1998), who proposed the virtual internal bond (VIB) model
to simulate crack growth in an isotropic solid. At the con-
tinuous state, VIB corresponds to a linear elastic solid with
a fixed Poisson ratio, namely, 0.25 for three-dimensional
cases and 0.33 for two-dimensional cases. To represent the
diversity of the Poisson’s ratio, Zhang and Ge (2005) devel-
oped the virtual multi-dimensional internal bond (VMIB)
model, in which a shear constraint was added into the interac-
tion between two coupled particles. In both VIB and VMIB,
materials are discretized into mass particles linked through Figure 1. The physical model and the calculation cycle of DLSM.
281

(see Fig 2(e)) and DLSM and FEM coupled model (see
Fig 2(f)), are also presented. Other applications of DLSM
include: Zhao and Zhao (2010) to study the cutting process
of coal under single blade and You et al (2010) to analyze the
anisotropic failure of rock material.
5 CONCLUISION
DLSM is different from the conventional lattice spring mod-

els in that a shear spring is introduced to capture multi-body
forces that develop due to spring deformation. Compared with
traditional numerical methods, DLSM has the advantages:
it directly uses macroscopic parameters without a calibra-
tion process, unlike particle based discrete element methods
(DEM), it has half as many degrees of freedom as DEM and it
is stable, does not require a numerical integration technique,
and is well suited to the analysis of contact and heterogeneity
problems. These advantages may lead the DLSM become an
alternative choice in rock mechanics study.
REFERENCES
Gao H.J., Klein P. 1998. Numerical simulation of crack growth in an

isotropic solid with randomized internal cohesive bond. J Mech
Phys Solid, 46: 187–218.
Kazerani T., Zhao G.F., Zhao J. 2010. Dynamic fracturing simulation
of brittle material using the Distinct Lattice Spring Model (DLSM)
with a full rate-dependent cohesive law, Rock Mech Rock Eng;
Figure 2. Numerical Examples of DLSM. DOI: 10.1007/s00603-010-0099-0.
You S, Zhao G.F., Ji H.G. 2011. Model for Transversely Isotropic
Materials Based on Distinct Lattice Spring Model (DLSM),
Journal of computers (accepted).
4 VALIDATION AND APPLICATION Zhao G.F. 2010. Development of micro-macro continuum-
discontinuum coupled numerical method. Phd thesis. EPFL,
Examples are presented to highlight the abilities of the DLSM Switzerland.
for static modeling, wave propagation through joints and Zhao G.F., Fang J., Zhao J. 2010. A 3D distinct lattice spring
dynamic fracturing (see Fig 2). For the beam bending problem, model for elasticity and dynamic failure. Int J Numer Anal Meth
results of the DLSM model with different lattice structures Geomech, DOI: 10.1002/nag.930.
are analysed and compared with the FEM solution (see Fig Zhao S.F., Zhao G.F. 2010. Modelling the cutting process of coal
2(a)). Fig 2(b) shows the application of DLSM on model- under single blade through Distinct Lattice Spring Model (DLSM).
Proc of ICACTE 2010; 241–245.
ing wave propagation problem. The results are compared well Zhang Z.N., Ge X.R. 2005. A new quasi-continuum constitutive
with UDEC and analytical solutions (Zhu et al, 2011). The model for crack growth in an isotropic solid. Eur J Mech A Solids,
DLSM is also used to predict the dynamic cracking propaga- 24: 243–252.
tion velocity of PMM plate by Kazerani et al (2010) (see Fig Zhu J.B., Zhao G.F., Zhao X.B., Zhao J. 2011. Validation study of
2(c)) and dynamic fracture toughness of rock material (see Fig Distinct Lattice Spring Model (DLSM) on P-wave propagation
2(d)). New developments of the DLSM, e.g. Implicit DLSM across multiple parallel joints, Comput & geotech, 38: 298–304.
282

A discussion about boundary conditions in numerical simulation and its application in

mining-induced rock mass movement
H.J. Zhao, F.S. Ma, Y.M. Zhang, J. Guo & A.H. Wei
Key Laboratory of Engineering Geomechanics, Institute of Geology and Geophysics, China Academy of Sciences, Beijing, P.R. China
Keywords: numerical modeling; rock stress; mining; rock mass
ABSTRACT: The geo-stress field of rock mass is a complex of which boundary conditions are established, the state of the
system affected mutually by various factors (such as topo- back-calculated initial stress field should correspond to that
graphy, lithology, and geologic structures). Currently, there of the practical stress field. In fact, no matter which kind of
are many kinds of in situ stress measurement methods, but the boundary conditions is used, the state of initial stress should
precise description of the geo-stress field remains difficult. be the same as that of the actual stress state, horizontal or
The initial geo-stress field is the basic factor and important nearly horizontal. However, previous studies have shown that
basis for the study of rock mechanics problems. To inverse the simulation results always vary considerably under different
initial stress field of rock mass reasonably is an important issue boundary conditions. If fixed displacement boundary condi-
in rock mass mechanics.The degree of approximation between tion was applied to both sides, the simulated horizontal stress
the inversed initial geo-stress field and the actual stress field contours are horizontal and smooth. This stress field is consis-
is one of basic conditions for assessing the validity of the tent with the expected distribution of the stress field. However,
numerical simulation. Especially excavating in high tectonic if stress boundary condition was applied to both sides, the dis-
stress areas, the correctness of the initial geo-stress obtained tribution contours of horizontal stresses are irregular. Aside
by numerical simulation directly influences the final results. from the horizontal distribution trend near the lateral borders,
The boundary conditions of the continuous mechanic large fluctuations can still be observed in the inner model.
models are always divided into 3 types: (1) stress boundary These simulated stress fields highly differ from the theoret-
condition, which means surface force or concentrated force ical stress field. Therefore, boundary conditions are optimal
of the body surface is given, and the force can be variable or for initial geo-stress field inversion.
not; (2) displacement boundary condition, which means dis- To study the special effects of high tectonic stress on exca-
placement component of the boundary points is known, it can vating works, a typical mining-induced ground movement
be a constant, equivalent to zero or equivalent to zero in some caused by underground mining in Jinchuan No. 2 mine is
directions. When the displacement in some direction is zero, carried out. Based on the results of years GPS monitoring
it is what is called constraint boundary condition; (3) mixed data on the ground surface, inverse analysis of mining-induced
boundary condition, which means the displacement is known rock mass movement characteristics under high tectonic stress
in some directions, and the boundary force is given in the other was studied, and a comparative analysis of the surface move-
directions. ment, deformation characteristics were given out. The results
During numerical simulation, the reasonable initial geo- indicated that the back calculated method can achieve reason-
stress field is usually acquired by regression fitting method able and reliable initial geo-stress field. Because of different
using limited on terrain, geomorphology and observed geo- mechanisms of rock mass movement in high tectonic stress
static stress as base. For the underground excavation in field, the mining-induced ground movement has distinctive
geologic body with high horizontal tectonic stress, to simulate characteristics of ground movement and deformation in the
the actual geo-stress state rationally is the key of numerical following aspects.
simulation.
To discuss the inversion of initial geo-stress field, suppose a) The volume of subsidence trough is much smaller than
there is a homogeneous geologic body with high-level tectonic the volume of the underground goaf. The slow subsidence
stress, in which the direction of the maximum principal stress ratio will not result in a sudden collapse.
is horizontal. Then, the distribution of the initial tectonic stress b) The scope of ground movement, deformation and damage
contours should be horizontal under the influence of tectonic expand distinctively outward of the subsidence center. The
stress field. ground damage types vary with different places. Internal
In numerical simulation, if the stress boundary condition region of the basin, it shows shear dislocation scarp dam-
is set, additional stresses are applied to lateral boundaries to age, which corresponds to the mining-induced subsidence
simulate the corresponding tectonic stresses. During numer- and deformation caused by its self-weight stress. However,
ical simulation, additional stresses are constant, whereas the in the outside region of the basin, it is mainly tensile fail-
boundaries are moveable during excavation. If the displace- ure, which corresponds to the subsidence and tension effect
ment boundary condition is set, tectonic stresses can be under the tectonic stress field.
assigned as internal stresses in the calculated area. Regardless
283

c) On the edge of the subsidence trough, the ratios of the rock deviated gradually to the upper surface of ore body, and
mass horizontal displacements to vertical displacements the ground movement, deformation and damage character-
are doubled or increased by orders of magnitude. istics expressed more prominent.
d) The existence of tectonic structural planes changed the
Based on the contrast study of rock mass stress field and
normal distribution patterns of the tensile deformation and
displacement field during the pre-and post-mining process, we
damage. Discontinuous deformation is frequently occurred
conclude that the fundamental reason for the special features
due to the existence of small faults and weak structural
lies in the effect of the geo-stress field on rock mass perfor-
planes, and resulting in a large number of scarps on the
mance. In the tectonic stress field, high normal stresses occur
surface. In the compression zone of subsidence trough,
on the preferred structural planes. It cause large sliding fric-
scarps are caused by shear and dislocation movement of
tion for structural planes activation or slide, and results in low
rock mass along the fault planes or weak structural planes.
rate of rock mass deformation and damage. Moreover, when
Whereas, in the tensile zone of the subsidence basin, addi-
the mining-out space is formed in high-level of tectonic stress
tional tensile stress occurred on the fault planes or weak
field, horizontal displacements are the overriding displace-
structural planes, which result in distinguished opening
ment components in the hanging wall and footwall of the mine
displacements.
goaf. It directly resulted in expansion of the horizontal move-
e) With increase of mining depth, not only the total sub-
ment and deformation scope, and caused more prominent
sidence amount is increased, but also the subsidence
macro ground movement, deformation and damage.
scope is expanded. The location of subsidence center is
284

Theoretical and experimental modeling of material hardening and

softening by compression tests
V.M. Zhigalkin, V.N. Semyonov, O.M. Usoltseva, P.A. Tsoi, A.I. Chanyshev & I.L. Abdulin
Siberian Branch of Russian Academy of Sciences, Institute of Mining, Novosibirsk, Russia
Keywords: numerical modeling, physical modeling, problems of rocks
1 INTRODUCTION Table 1. Empirical relationships between mechanical properties

and ratio h/d.
The main task of ensuring safe mining in water-soluble ores
Amount
(rock salt, sylvinite) is to prevent the risk of breakthrough of σ1cs , εcs
1 , De ,
of oversaline waters in underground openings. Usually this is Rock type samples MPa % GPa
achieved using a breast system development with the support
of the overlying strata on the pillars. Straining salt rocks must Rock salt 65 15.61 + 11.17(h/d)−1 3.42 (h/d)−1 1.7 h/d
maintain solidity for a lifetime of mine, playing the role of Sylvinite 90 15.69 + 9.6(h/d)−1 5.1 (h/d)−1 1.38 h/d
−1
impervious pillar. In order to develop ways to prevent acci- Carnallite 42 4.14 + 7.9(h/d) 0.9 (h/d)−1 1.65 h/d
dents, researches on the physical and mechanical properties Rock type Mcs , WV ,
of rocks and modeling of the stress-strain state of structural GPa MJ/m3
elements of underground openings should be carried out.
Rock salt −h/d(0.49 + 0.03h/d) 1.31 (h/d)−1.67
Sylvinite −h/d(0.16 + 0.09h/d) 1.95 (h/d)−1.97
2 EXPERIMENTAL SET-UP
Carnallite −h/d(0.04 + 0.43h/d) 0.09 (h/d)−1.82
The cylindrical samples of motley sylvinite and rock salt were
prepared, with a diameter d of 38 mm and height h of 76,
57, 38 and 19 mm, respectively. The bases of the samples (on
which the axial displacement was measured) were 50, 25 and between the sample and the testing machine. The empirical
12.5 mm. relationships between the mechanical properties of salt rocks
The choice of sample size and accuracy of processing are and the ratio h/d are shown in Table 1.
consistent with the requirements of State Standard (1985, In Table 1: σ1cs – compressive strength, εcs1 – ultimate strain,
1991). The uniaxial and triaxial compression tests were car- De – tangent modulus of elasticity, Mcs – secant softening
ried out with the testing machine model 8802 manufactured modulus, WV – energy required for failure.
by Instron. It is an automated electro-hydraulic system. It is ascertained that the relative ultimate strain εcs
1 depends
The gauge length extensometer manufactured by Instron linearly on the ratio of diameter to height of the sample. This
with a 25 mm base was used for measuring the axial deforma- suggests that the absolute ultimate strain does not depend on
tion of the samples. Lateral deformation was measured using the height of the sample, and varies only according to its diam-
the glued-on strain gauges manufactured by HBM, a German eter value. This conclusion is extremely important and can be
company. used to evaluate the critical
strain of pillars of salt rocks.
A special cell was used in the triaxial compression tests. The energy (WV = σ1 dε1 , where σ1 is the axial stress,
The sample was placed in the cell, a certain level of pressure ε1 – axial strain ) required for failure increases with decreasing
and axial load was then applied, and the sample was loaded to sample height and its variation can be well described by a
failure. power function.
The sylvinite samples required the most energy for failure
(due to plasticity), and carnallite samples needed the least
3 EXPERIMENTAL RESULTS energy for failure as carnallite was the most brittle rock.
Study on the deformation process of quasi-plastic rocks
In order to study the influence of sample shape on the mechan- under triaxial compression (Zhigalkin et al., 2008) was carried
ical properties, tests were performed on core materials (rock out on cylindrical samples of the motley sylvinite.
salt, red and motley sylvinites) (Zhigalkin, et al., 2005). Ratio Diameter (d) of the samples was 38 mm and the ratio of
of the height to the diameter of samples on the average ranged height (h) to diameter was 2.0. Tests were performed in a
from 0.5 to 2.5. The influence of sample shape on the strength cell of triaxial compression, which allowed maintaining the
and deformation properties of carnallite was studied. Tests internal pressure to 40 MPa.
were performed in uniaxial compression mode, subjected to The experiments were performed under axial load at a
the “dry” friction in the contacts (without lubricant or spacer) constant lateral pressure.
285

Loading was applied at the following levels of lateral pres- used for describing the certain material data within reasonable
sure: 0 MPa, 1 MPa, 2.5 MPa, 5.0 MPa, 7.5 MPa, 10.0 MPa limits.
and 20.0 MPa.
It is revealed that strength and deformation parameters
increase with increasing of lateral pressure, the tangent 5 CONCLUSION
softening modulus goes down.
The compression strength decreases with increasing height
of the sample. Stabilization of the strength parameters is
4 THEORETICAL MODEL AND RESULTS observed for h/d≥1.5. The lowest compression strength and
proper strain is observed in carnallite samples. Strength
Theoretical studies of Chanyshev (2004) regarding the con- properties of sylvinites are similar to rock salt ones.
struction of constitutive relations in elasticity and plasticity An increase in the compression strength and deformation
are based on the Christianovich-Shemyakin’s idea about block parameters with increasing lateral pressure and decreasing
structure of a deformable body. It is assumed that any model tangent modulus are observed.
of body consists of a deformable rigid blocks, of which an Using the constructed uniform curves of Chanyshev’s
arbitrary deformation, including elastic, is the result of mov- model, the compression strength σ1cs and the corresponding
ing one block relative to the other. Based on this mechanical strains εcs cs
1 , ε2 of salt rocks under various levels of lateral
representation of a rigid body, the following is proposed pressure can be estimated. The received dependences reflect-
to construct constitutive relations in elasticity and plasticity ing the behaviors of environmental block model is the theme
(Chanyshev and Kurenkeeva, 2004). The following functional of our other reports.
dependences are considered. The dependence of S1 on 1 is
the relationship between the average stress and the average
strain, and the dependence of S2 on 2 is the relationship ACKNOWLEDGEMENT
between the intensity of shear stress and shear strain.
The certain eigen unit tensor basis is own basis for metals, This work was supported by the Integration Project No 191
i.e. dependence Sl = Sl (l ) is linear function and universal “Deformation and fracture of rocks and rock mass under
for metals and consequently suitable for elasticity, plastic- various loading conditions”.
ity, failure. Dependence S2 = S2 (2 ) is “uniform” curve,
i.e. it is not dependent on a kind of loading. For rocks the
REFERENCES
foregoing ones are not universal, i.e. own. It is shown that
by turn of basis on some corner ϕ∗ universal functional
Chanyshev A.I., Kurenkeeva B.O. 2004. To a question of construction
dependences(Sm = Sm (m ) and Sl = Sl (l )) can be received. of rocks passport dependences. Proceedings of the international
It is the main conclusion of the theoretical part. conference “Problems and prospects of development of Mining
The angle ϕ∗ = 42◦ and dependencies Sm = Sm (m ), Science”, Novosibirsk.
Sl = Sl (l ) for samples of motley sylvinite, obtained State Standard 21153.2-84 1985 Rocks. Methods of definition of
by the method of least squares, have the expressions: uniaxial compression strength. Moscow.
Sm = 389m (−0.15 ≤ m ≤ 0) and Sl = −3552l + 145l State Standard 28985-91 1991 Rocks. Methods of strain definition at
(0 ≤ m ≤ 0.17) respectively. uniaxial compression. Moscow.
The angle ϕ∗ = 39◦ and dependencies Sm = Sm (m ), Zhigalkin V.M., Usoltseva O.M., Semyonov V.N., Tsoi P.A. 2005
Sl = Sl (l ) for samples of rock salt, obtained by the Deformation of quasi-plastic salt rocks under different conditions
method of least squares, have the expressions: Sm = 556m of loading. Report I: deformation of salt rocks under uniaxial
compression Journal of Mining Science Vol. 41, No 6, pp 507–515.
(−0.15 ≤ m ≤ 0) and Sl = −11992l + 286l (0 ≤ m ≤ Zhigalkin V.M., Usoltseva O.M., Semyonov V.N., Tsoi P.A.,
0.15) respectively. Asanov V.A., Baryakh A.A., Pan’kov I.L., Toksarov V.N., Evseev
Based on the theoretical uniform curves, inverse calculation A.V. 2008 Deformation of quasi-plastic salt rocks under different
of εcs cs cs
1 , ε2 , σ1 was performed for the motley sylvinite and conditions of loading. Report II: Regularities of salt rock defor-
rock salt samples. A fair error of all the calculated values did mation under triaxial compression. Journal of Mining Science
not exceed 30%. Consequently, the theoretical model can be Vol. 44, No 1, pp 15–23.
286

Multiple scale numerical modeling of growth process of multiple cracks in rock mass
X.P. Zhou, H.Q. Yang & Y.X. Zhang

School of Civil Engineering, Chongqing University, Chongqing, China
Keywords: numerical modeling; rock failure; rock mass
Rock mass is neither continuous material nor totally discrete a process of crack propagation for quasi brittle material under
one, it’s a kind of defect material which contains many cracks. tensile stress by Wells. A process of crack propagation under
The nonlinear deformation behavior of rock mass is induced tensile stress is obtained by Du, and the feasibility of this
by the unstable propagation of cracks which existed in the rock method is verified too. The relationship among extended
mass under external loads. Therefore, it is key important for finite element method, meshless method and traditional finite
designing rock engineering to study the propagation process element method are summarized by Jin.
of cracks existed in rock mass under compressive stress. It is well known that there are many joints, cracks and
They are crucial problems for complex system that the phe- faults in rock mass. The dimensions of these discontinuous
nomenon of cross-scale and the coupling of multiple scale. are varied from meters to millimeters. In the engineering prac-
Moreover, It is pointed that they mirror the intrinsic behavior tices, the failure of surrounding rock mass mostly related to
of material. These crucial problems have rich contents. For the interaction of discontinuous interface with varied dimen-
rock mass, it contains many discontinuous interfaces (such as sions. However, it is lack of effective numerical simulation
joints, cracks and faults), and these discontinuous interface method for crack-weakened rock mass which can take the
varied in dimension. In addition, the deformation behavior multiple scale mechanical behavior or rock mass into account.
of rock mass is proved time-dependent and temperature- In order to solve this troublesome, some breakthroughs should
dependent too. So, the behavior of rock mass is multiple be made in the fundamental theory field of numerical simu-
scale problem. In this paper, only the dimension aspect of lating method for crack-weakened rock mass. On the basis
cross-scale problem is discussed. For example, the interaction of extended finite element method, a multiple scale model for
problem between microcracks and marco faults are usually simulating the damage evolution in crack weakened rock mass
observed in rock engineering. However, no effective modeling is proposed. The coordination between the internal bound-
method for solving this problem has been proposed till now. ary (such as cracks and holes) and mesh is not needed in
It is nearly impossible for traditional numerical method which this multiple scale model. By combining the displacement
needs mesh for calculation to modeling multiscale cracks projecting method and displacement loading method, the con-
existed in rock mass. nection between different sizes of mesh behave smoothly. This
There are some numerical methods for modeling crack multiple scale model indeed surmounts difficulties in mesh-
growth which remeshing is not needed, such as mani- ing and it save the storage space. The contact constraint on
fold method, meshless method and extended finite ele- crack surface is embedded within the total stiffness matrix by
ment method. Among these, extended finite element method the penalty method, in which the load vectors are modified
(XFEM), which can be used to solve discontinuous mechan- too. The path of crack growth and stress field are determined
ical problems, is proposed by Belytschko et al. in the year of through iterative computations. Because additional discontin-
1999. It is proved that the extended finite element method is uous functions and enriched tip elements are added in the
effective in modeling discontinuous problem, such as growth displacement field, the geometry of cracks is independent of
of interface, propagation of crack and complex fluid mate- the finite element mesh. As a result, remeshing is not neces-
rial. There are three main advantages for the extended element sary to model crack propagation by using this multiple scale
method. Firstly, the details of substructure are neglected in this model. Because the frictional contact of crack surface is taken
method, and the mesh for calculating are generated according into account, the present method is fit for modeling the growth
to the geometry shape of structure. Secondly, in order to track of multiple cracks in geomaterials under compressive loads.
the growth of cracks, other methods such as level set method Finally, the method is employed to simulate multiple crack-
are adopted. Then, the location of crack is accurately deter- growth when frictional contact exists on the crack surfaces,
mined. Finally, in order to model the propagation of cracks, and the results show good agreement with the experimental
only the shape function of element is modified. Therefore, the ones.
matrix of total stiffness, which is sparse and zonal, has the The paper is organized as follows. The model problem for
same advantages as traditional finite element method. an elastic body with microcracks with varied dimensions is
At present, the method of extended finite element is adopted formulated in Sect. 2. In Sect. 3 the total stiffness matrix is
in solving the propagation problem of single crack under obtained. The asymptotic method for calculating stress inten-
axial tensile stress. For example, a object-oriented program sity factor in the crack tip is introduced in Sect. 4, and the
for extended finite element method is compiled by Stephane fracture criterion is also adopted in this section. Numerical
in C++ language. By using extended finite element method, results are presented in Sect. 5.
287

Laboratory testing and rock properties

Mechanical behavior of Callovo-Oxfordian claystone-steel interfaces at

high levels of normal stress in “dry” and “wet” conditions
M. Boulon & M. Keshavarz

Lab 3SR, University of Grenoble, Grenoble, France
G. Armand & N. Conil

National Radioactive Waste Management Agency, Centre de Meuse Haute-Marne, Bure, France
F.L. Pellet
INSA – University of Lyon, Villeurbanne, France
Keywords: lab testing, nuclear repository
The French national radioactive waste management Agency possibility that shearing might develop at the interface due to
(ANDRA) studies the feasibility of geological repository the difference in the thermal expansion coefficients of steel
for radioactive wastes. Since 2000 it has been engaged and rock. It should also be noted that during placement of the
in the development of the underground laboratory of steel lining in the micro tunnel, there was no perfect contact
Meuse Haute Marne to study the properties of a claystone, between the lining and the claystone. Therefore normal stress
called “Callovian-Oxfordian argillite”. At the Underground on the steel lining will vary over time. However, despite the low
Research Laboratory, the claystone layer lies between 420 permeability of the rock, water will appear at the wall of the
and 550 m below ground and is the target layer for the repos- micro tunnel, which means that the interface could become
itory of high-activity and long-life (HA) radioactive wastes. more or less saturated. Therefore, it is necessary to under-
Along the main gallery, smaller horizontal blind micro-tunnels stand the mechanical behavior of the steel–claystone interface.
were excavated. These subsidiary micro tunnels are approxi- This research was an initial attempt to study the properties of
mately 40 m long and 700 mm in diameter lined with an about friction and dilatancy in this type of interface at room temper-
25 mm thick steel lining along the entire length, and designed ature. Several tests on claystone-steel interfaces in both dry
to accommodate waste canister. (meaning at the water content of core sample which may be
In this study, direct shear tests were carried out using a slightly desaturated) and wet (meaning after water injection at
prototype machine that was developed in the 3S-R labora- the interface) were performed for this study. Based on the test
tory, Grenoble, and named BCR3D. This machine operates results, the following conclusions can be drawn:
with five electro-mechanical servo controlled jack and was
1. The maximum normal stress for studying the behavior of
designed to measure the hydro mechanical properties of nat-
the claystone is 12 MPa. Above this stress level, the prop-
ural or artificial joints/fractures as well as the interfaces
erties of claystone degrade due to the presence of crushed
between different materials with 3-D stresses (two orthogo-
material and removed particles.
nal shear directions and normal compression). This machine
2. The presence of water alters some of the claystone proper-
can also be used to create a hydraulic conductivity interface
ties and this inevitably affects the mechanical behavior of
by injecting a fluid at the center of the specimen, creating a
the interface. We have highlighted the fact that the shear
radial flow at the interface. Symmetrical movement of the two
resistance of the “wet” interface is significantly less than
parts of the shear box during testing ensures the quality of
for the “dry” interface. We obtained the same results for
tests. The relative normal and shear displacements (tangential
claystone-claystone interfaces, although these test results
[w] and normal [u]), between the two parts of the shear box
are not presented here.
as well as the stresses (shear τ and normal σn ) are measured
3. For the wet and dry interface tests, the values of shear
by various gauges. It is also possible to measure the flow (q)
stresses in the loading direction are more reliable than the
and injection pressure (p) of fluid or gas during the test.
parameters obtained from the reverse shearing direction.
The concept for high level, long life radioactive wastes is
4. The general behavior of the claystone – steel interface
to place the waste canisters in horizontal steel casing of micro
remains virtually iso-volumetric during shearing, except
tunnels excavated in Callovo Oxfordian claystone. It is impor-
during reverse shearing that causes damage at the edges of
tant to be aware of the Thermo-Hydro-Mechanical behavior
the specimens (about 20% of the surface for “dry” and over
of the steel-claystone interface in order to understand the
30% for “wet” specimens).
long-term modifications that can be expected around these
cells. In particular, wastes are exothermic, and this raises the
291

Microcrack growth in granite during fatigue process
Y. Chen, A. Yamazaki, H. Kusuda, E. Kusaka & M. Mabuchi

Graduate School of Energy Science, Kyoto University, Kyoto, Japan
Keywords: lab testing, rock failure, rock properties, rock stress
The structures constructed in underground spaces are usually

used over a prolonged period. For the stability evaluation of
many rock structures, it is important to reveal the degradation
characteristics of rocks over the long term. It is well-known
that materials degrade due to repeated stress changes. This is
generally known as ‘fatigue’. Fatigue occurs in many kinds
of materials, and rock is no exception. Many studies on the
fatigue characteristics of rocks have been carried out since
the 1960’s. However, the detail damaging process in gran-
ite fatigue has not been clarified. Therefore, to examine the
fatigue process in granite, cylindrical specimens were sub-
jected to the cyclic loading test, and the crack development
during the test were observed microscopically and investigated
by image analysis technique.
A fine-grained two mica granite from Rhode Island, USA,
Westerly granite, was used for the examination. Cylindri-
cal specimens 10 mm in diameter and 20 mm in length were
subjected to the cyclic loading tests under dry uniaxial con-
ditions at room temperature, with a specified maximum Figure 1. Relationship between crack number and crack length as
stress of 160 MPa and the minimum one of 80 MPa. The the fatigue process progressed.
loading-unloading cycle was 0.1 Hz. The method proposed
by Nishiyama & Kusuda (1994) was applied to identify
microcracks within the specimens. dimensions of the reference area was calculated.The analytical
The fatigue process was generally divided into three stages: result indicated that at the first phase of fatigue process, the
a primary stage in which the strain increases at a decelerating increase in crack length was dominant compared with that
rate, a second stage with linearly slight increasing rate of strain in number. Meanwhile, it was estimated that the growth in
following the primary stage, and a third stage in which the number was prior to that in crack length at the later phase. It
strain increases at an accelerating rate and culminates in speci- showed that the elongation of preexisting cracks was dominant
men failure. Consequently, crack development patterns at each during the first phase, whereas many microcracks formed at
stage were observed. In the primary stage, intragranular cracks the later phase.
in quartz grains increased compared with the intact speci- To sum up the major characteristics of granite fatigue pro-
men. During the second stage, the most characteristic feature cess revealed in this study, at the primary stage, marked crack
was that many microcracks parallel to the loading direction growth was identified in quartz grains, but no preferential
were observed in feldspar grains at the posterior half of this orientation was found in these cracks. It showed that the
stage. Whereas, in quartz grains, no marked orientation in degradation in this stage was caused by the growth of pre-
microcracks was observed, but the increase in the number and existing cracks mainly in quartz grains. During the second
length of cracks was identified. At the final stage, long cracks stage, the crack initiation was more frequently than previous
which could not be identified in the former-stage specimens stage. Crack development shifted into feldspar grains from
were identified in the failure specimen. quartz grains. It was estimated that the initiation and growth
To analyze crack growth patterns, the length and the number of these cracks resulted in the slight increasing rate of strain
of extracted cracks in each section were measured and the in this stage. At the final stage, many intergranular cracks
parameter ‘crack population’ which was determined by the parallel to the loading direction were formed by linking cracks
sum of the crack length or the number divided by the full generated during the former stages.
292

An experimental study on deformation and strength anisotropy of

transversely isotropic rocks in Korea
Jung-Woo Cho, Hanna Kim, Seokwon Jeon & Ki-Bok Min

Department of energy systems engineering, Seoul National University, Republic of Korea
Subject: Rock material and rock mass property testing (laboratory)
Keywords: transversely isotropic rock, elastic constant, apparent elastic moduli, uniaxial strength, anisotropy ratio
This paper reports the experimental investigation of deforma- planes (E/E ) were determined to be 1.3, 2.1 and 3.4 for
tion and strength anisotropy of Asan gneiss, Boryeong shale Asan gneiss, Boryeong shale and Yeoncheon schist, respec-
and Yeoncheon schist in Korea which show clear evidence tively. The anisotropy ratios of maximum to minimum elastic
of transverse isotropy. For each rock type, three sets of seven modulus (Emax /Emin ) were 1.7, 2.7 and 4.0 for Asan gneiss,
cylindrical specimen cored with angles of 0, 15, 30, 45, 60, 75, Boryeong shale and Yeoncheon schist, respectively, which
and 90 degrees with respect to transversely isotropic planes were higher than E/E due to the fact that the minimum elas-
were obtained by using laboratory directional coring system tic modulus exist when the transverse isotropic plane were
established for this study. Five independent elastic constants inclined with respect to loading direction. The applicability of
of transversely isotropic rocks were determined using various transversely isotropic model to the chosen rocks was quantita-
combinations of strain measurements obtained from speci- tively investigated by comparing the apparent elastic moduli
mens having different coring directions. The anisotropy ratios with the theoretical apparent elastic moduli predicted by tenso-
of elastic modulus parallel to and perpendicular to isotropic rial transformation of compliance matrix. The mean prediction
error (MPE) defined as the average relative difference between
measured and predicted elastic moduli were lower than 17 %,
which indicates that transversely isotropic model is a reason-
able assumption for the rock types chosen in this study. The
mean prediction error becomes lower as more strain measure-
ments were employed and this demonstrates that more strain
measurement can increase the accuracy of the determined
elastic constants even if five independent strain measurement
is theoretically sufficient. The anisotropy ratios of maximum
to minimum uniaxial compressive strength were 2.6, 2.6 and
18.6 for Asan gneiss, Boryeong shale and Yeoncheon schist,
respectively. The anisotropy ratios of maximum to minimum
tensile strength determined by Brazilian Tensile test were 3.2,
2.2 and 7.1 for Asan gneiss, Boryeong shale and Yeoncheon
schist, respectively. The strength anisotropy were greater than
deformation anisotropy for Asan gneiss and Yeoncheon schist
and the opposite were observed for Boryeong shale. The extent
of anisotropy observed in this study is rather significant and
this study reconfirms the importance of anisotropy in rock
Figure 1. Photograph of the tilt coring system. engineering.
Table 1. The five elastic constants of Asan gneiss, Boryeong shale, and Yeoncheon schist in Korea.
Rock type Set E (GPa) E (GPa) v v G (GPa)
Asan gneiss 1 69.9 52.6 0.23 0.26 16.8

2 62.3 55.1 0.27 0.21 17.0
3 72.6 55.4 0.26 0.19 17.5
Mean 68.3 54.4 0.30 0.20 17.1
Boryeong shale 1 45.8 16.5 0.13 0.23 12.0
2 38.1 20.1 0.23 0.14 6.2
3 34.0 20.5 0.17 0.22 8.1
Mean 39.3 19.0 0.18 0.20 8.7
Yeoncheon schist 1 68.4 19.2 0.19 0.13 12.1
2 74.2 20.6 0.18 0.19 17.5
3 73.7 23.9 0.37 0.17 11.5
Mean 72.1 21.2 0.25 0.16 13.7
293

Figure 2. Variation of apparent Young’s modulus (Eθ ) and theoretical results from the transverse isotropic solution.
Figure 3. Variation of uniaxial compressive strength with anisotropy angle.
294

Mechanical rock properties prediction: Deriving rock strength and

compressibility from petrophysical properties
B. Crawford, B. Alramahi, P. Gaillot & P. Sanz

ExxonMobil Upstream Research Company, Houston, TX, USA
N. DeDontney
Harvard University, Cambridge, MA, USA
Keywords: lab testing, oil reservoir, rock properties
All geomechanical models from simple analytical calcula- from production-induced fluid pressure decline and strongly
tions to complex fully-coupled finite-element simulations impacts reservoir performance.
require some knowledge of mechanical properties in order to In this study we introduce a new predictive algorithm
describe the deformation behavior of the stressed rock units. for evaluating the mechanical strength of siliciclastic rocks.
For petroleum-related applications two ubiquitous experi- The relationship is sufficiently generic that it can be applied
mental techniques include triaxial compression testing for throughout sandstone-to-shale sequences and requires only
rock strength measurement and compaction testing under basic lithology and porosity wireline logs as data input. It
uniaxial strain boundary conditions for rock compressibil- is simple and ideally suited for rock strength prediction in
ity measurement. Knowledge of compressive rock strength data-sparse regions where, for example, no acoustic logs or
is essential for modeling near-wellbore deformation while associated core measurements are available for calibration
rock compressibility controls pore volume strains resulting purposes (Fig. 1).
Single element ABAQUS modeling has been used to simu-
late rock compressibility evolution with uniaxial compaction
and the onset of pore collapse, benchmarked against an exten-
sive database of core measurements spanning unconsolidated
sands to tight gas reservoir rocks (Fig. 2). Empirical corre-
lations relating Cam-Clay elastic-plastic material parameters
and other petrophysical properties potentially derivable from
geophysical wireline logs are discussed.
Figure 1. Mohr-Coulomb cohesion and friction angle derived from Figure 2. Modeling impact of yield surface eccentricity on forma-
wireline logs. tion compressibility magnitude following pore collapse.
295

Behaviour of carbonated rocks under true triaxial compression
F. Descamps & J.-P. Tshibangu

Mining Engineering Department, Faculty of Engineering, UMONS-University of Mons, Belgium
M. Ramos da Silva, C. Schroeder & J.-C. Verbrugge

Soil Mechanics Laboratory, Université Libre de Bruxelles, Belgium
Keywords: lab testing; rock failure; rock properties
A true triaxial testing device has been developed in the nineties circular section. The range of mean pressures corresponding
in the mining department of the Faculty of Engineering, to a particular shape varies according to the competence of
Mons-Belgium. Since then data have been gathered from sev- the rock. For Sorcy limestone, we also went further by trying
eral rocks, among which various carbonated rocks exhibiting to build the octahedral envelopes on the cap surface, indicat-
different mechanical behaviours. In this work, we compare ing a triangular shape with apexes in the direction of triaxial
the behaviour of three limestones: a compact and strong extension states.
one (Soignies limestone, UCS = 100–120 MPa), a compact These observations can be associated with the evolution of
one with medium strength (Moca limestone, UCS = 60 MPa) the rocks behaviour. In triaxial conditions, perfectly plastic
and a soft and porous one (Sorcy limestone, UCS = 40 MPa). flow was observed for P between 10 and 20 MPa for Sorcy
Figure 1 compares the failure envelopes for those three limestone, between 80 and 90 MPa for Moca limestone and
rocks in the (P, Q) plane. In addition to the meridian sections, is expected beyond 240 MPa for Soignies limestone. For the
some octahedral envelopes are also sketched on this figure. In first two cases at least, strain hardening therefore occurs before
each case, we observe an evolution of octahedral sections from the deviatoric stress attains its maximum value on the meridian
an almost triangular shape to a hexagonal one, and then a quasi section.
Figure 1. Failure envelopes of three limestones.
296

Mechanical rehearsals in special slate schists to potentiate constructive solutions of

structures with great slenderness
L.M. Ferreira Gomes, C.R.V. Batalha, P. Gabriel de Almeida, L.J. Andrade Pais & E. Mendes
Department of Civil Engineering and Architecture, University of Beira Interior, Covilhã, Portugal
Keywords: lab testing, rock mass properties
1 INTRODUCTION 2 PRINCIPAL RESULTS AND CONCLUSIONS
The investigation zone of Foz-Côa schists, known as the quar- The rocks in study corresponds to slate schist unweathered
ries of Poio is located in Portugal, near the city of Foz-Côa, (W1), coloured from dark grey to black, sometimes striped,
in Guarda’s district (Europe). The rocks exploited in quarries as shown in Figure 2.
integrate the geological unit named Formação de Desejosa, When broken by cleavage S1 frequently presents a black
belonging to the Douro Group of Super-Grupo Dúrico Beirão, color, being very soft to the touch. When broken by cleavage
corresponding to ante-Ordovician rocks. S2 is dark grey and a little rough to the touch. When changed,
The quarry area is located in a syncline, diving ESE, with and in particular on the faces corresponding to fractures, it
layering (So) varying from sub-horizontal, South, to N54◦ E presents chestnuts colorations.
60◦ SE, in the northernmost area. The results of tests of physical indices are presented in
The main schistosity, S1 , is parallel to S0 . The secondary table 1, being noted the low values of porosity and water
schistosity, S2 , is generically directed to N43◦W sub-vertical. content when samples are saturated.
The joints, sometimes parallel, sometimes perpendicular to The slate schist of Foz Côa, in the 20th century were known
S2 , makes the start-up and the transformation of rock easier. with much regional importance, being exploited for the pro-
In Figure 1 are presented typical images of the quarries, where duction of props for alignment and verticality of vineyards.
it can be observed: the foot of the quarry semi-horizontal and
parallel to S0 (and to S1 ); the schistosity of the second phase,
S2 , semi-vertical; and yet the joints perpendicular to S1 and S2 .
Figure 2. Macroscopic aspect of the slate schists under study when

sawed (in core) in W1 situation (unweathered). Specimens prepared
to be subject to strength tests.
Table 1. Physical parameters of Poio’s slate schist in W1 situation

(unweathered).
γt γd WS WN na
Parameter (kN/m3 ) (kN/m3 ) (%) (%) (%)
Minimum 26.30 26.30 0.01 0.00 0.02

Medium 27.47 27.47 0.02 0.02 0.06
Maximum 28.40 28.40 0.03 0.08 0.09
St. Dev. Pad. 0.720 0.720 0.008 0.040 0.024
n. of specimens 70 70 70 70 70
γt – Total unit weight, with saturated samples condition; γd – Dry

unit weight of the material with dry samples condition; WN – Natural
Figure 1. Images on typical situations of the Quarry of Poio – moisture content; WS – Moisture content with saturated samples
Foz-Côa, Portugal. condition; na – Porosity.
297

It should be noted that traditional application took advan-
tage of the singularity of that rock, allowing, very often, the
extraction of slender elements, frequently with a section of
4 cm × 8 cm and length between 200 and 350 cm, replacing
real wooden props.
Already at the end of the 20th century, this rock started to
be used in the building industry, particularly in the cladding of
buildings, exterior and interior, in curb stones for walks, in the
garden walls, sills and windowsills, in risers among others.
In the meanwhile, it is understood that due to their charac-
teristics, it must be proved, with mechanical tests and others,
that this rock can be applied in very bold solutions in the build-
ing industry, taking particular advantage of the high capacity
this rock presents to resist especially to flexion, as well as
withstand high loads in relative terms, taking into account
the slender elements used. The authors have done some tests
and applications, as shown in Figure 3 that potentiate these
applications.
Starting from Uniaxial compression tests on perfect spec-
imens (Fig. 2), the values of uniaxial compressive strength
(σc ), on average were, 107.12 MPa for ⊥S1 and 73.93 MPa to
//S1−Tr .
When it is intended to obtain σc from the Point-load test,
using a relationship of type σc = k(IS-50 ), with k = 22 (ISRM,
1985), leads to σc values higher than those recorded in the
Uniaxial compressive test in the slate schist of Foz-Côa. Con- Figure 3. Images on slate schist applications of Poio (Foz-Côa) in
cases that take advantage of the flexural strength in various situations
sidering the values of IS-50 obtained in regular cubic shape and other cases that take advantage of the uniaxial strength in slender
specimens, for this same rock, it is obtained k = 18. elements loaded parallel to the schistosity, of top (//S1-To ).
From the in situ tests, Windsor probe and Schmidt hammer,
σc values are greater than those obtained directly on the uncon-
fined compressive strength test, with about 21% in the case On the flexural strength (σF ), generally it can only be justi-
of Windsor probe and about 36% in Schmidt hammer. In any fied the application of slate schist of Poio on the situation of
event both in situ tests have found that σc in //S1-To situation loadings according to ⊥S1 , whose values of σF on average for
is roughly equal to σc in ⊥S1 , singularity that is important to test type vertical beam, type horizontal beam and type slab,
note, as guides for these materials to have great performance were 61.29 MPa, 38.52 MPa and 54.22 MPa, respectively.
in compression along the axis of lineation. Finally, studies with essays by flexion and uniaxial com-
By the Brazilian test, the tensile resistance values, σt , are pression of the buckling type (using specimens type vertical
about 16 MPa regardless of the fact that the specimens were beams, simulating pillars with different heights) should con-
loaded parallel or cross-sectional to schistosity S1 , being this tinue, so as to optimize the relations between the respective
figure, approximately 15% of the value of σc in state ⊥S1 . strength with the slenderness of the pieces.
298

An appraisal and modification of Ramamurthy criterion on anisotropic rocks
A. Ghazvinian, M.R. Hadei, H.R. Nejati & V. Sarfarazi

Rock Mechanics Division, Faculty of Engineering, Tarbiat Modares University, Tehran, Iran
Keywords: empirical criteria, mechanical anisotropy, anisotropy effect, jointed specimen
There have been extensive efforts to develop empirical criteria so when the anisotropy is of a mechanical type (induced
by a number of rock mechanics researchers as the theoretical anisotropy).
ones are not adequately capable to demonstrate the response Studies on Ramamurthy criterion reviled that it is possible
of intact or jointed rock masses at a wide range of confining to modify the expression with the help of the term defined as
pressures. Amongst such expressions proposed Ramamurthy’s Anisotropic Effect (Ae ) given as:
criteria is easy to use and yields sufficiently accurate response
for intact and jointed rocks. However this research indi-
cates that this criterion could be further refined to estimate
even more precise compressive strength of anisotropic rocks.
An intensive laboratory investigation was undertaken to test where, σia is the compressive strength of the intact samples
jointed specimen at various discontinuity angle. Fabricated under ‘a’ confining pressure, σja is the compressive strength
artificial samples which include two perpendicular joints were of the anisotropic samples.
prepared by Plaster of Paris. These specimens were used for The anisotropic effect parameter Ae could only vary
simulation of mechanical anisotropy subjected to uniaxial and between 0 to tending 1. Hence a parameter called strength
triaxial tests (Fig. 1). reduction factor was derived as:
It was noticed that Ramamurthy’s criterion can be well
applicable for the anisotropic rocks only when the uniaxial
compressive strength is noticeable. On the other hand deter-
mination of the uniaxial compressive strength of the jointed where, Cj : strength reduction factor, σ1i : compressive strength
rocks in the laboratory or in-situ requires expenses and time. of isotropic samples under the confining pressure of σ3 , σ1j :
Furthermore, preparation of the samples that could describe compressive strength of anisotropic samples under confining
the various anisotropic conditions is a difficult task, more pressure of σ3 .
Figure 1. Artificial anisotropic specimens.
299

Figure 2. Comparison of proposed modification with experimental data in the different anisotropic specimens.
Now, considering factor Cj and the uniaxial compressive It is evident from Figure 2 that the proposed modification
strength of the intact rock, a new relationship could be derived generates a good conformity with the results obtained from
as below: the tests. Even the conformity is more pronounced when the
discontinuity angle is 30◦ or 45◦ . It can be concluded that for
the specimens which were tested for this research work the
proposed modification provides adequate precision.
where σ1 : compressive strength of anisotropic samples under

the confining pressure σ3 , σci : uniaxial compressive strength
of intact sample and α, β: strength parameters.
300

Effects of fracture size and normal stress on the shear behavior of a fracture
as estimated by mortar replica
A.A. Giwelli
Technology & Research Center, JOGMEC, Chiba, Japan
Graduate School of Environmental Studies, Tohoku University, Sendai, Japan
K. Matsuki, K. Sakaguchi & A. Kizaki

H. Sekino & K. Okatsu

Technology & Research Center, JOGMEC, Chiba, Japan
Keywords: lab testing, rock joints, rock mass, rock properties
In this study, the effects of fracture size and normal stress aperture, which can be a measure for the matedness of the frac-
on the shear behavior of a fracture were investigated in the ture surfaces, increases with shear displacement and fracture
direct shear tests by using mortar fracture replicas of three size. Thus, the matedness of the fracture surfaces decreases
different sizes of 200 × 200 × 80 mm (M200), 150 × 150 × with shear displacement and fracture size. Furthermore, the
80 mm (M150) and 100 × 100 × 80 mm (M100), which were SD of the initial aperture also increases with normal stress.
produced from a tensile fracture created in granite. Monotonic This suggests that an increase in normal stress causes more
shear loading was conducted under constant normal stresses damage on the asperities, which decreases the matedness of
of about 0.3 and 1 MPa with the maximum shear displace- the fracture.
ment of about 20 mm. To observe the evolution of surface Figure 2 shows the effect of normal stress on the non-
damage in the shearing process, the fracture surface heights linearity of the closure curve for M100, M150 and M200 at
were measured before and after the experiment, and the mean, shear displacement of 20 mm. Since the matedness decreases
the standard deviation (SD) and the power spectral densi- with normal stress, the non-linearity in the closure curve
ties (PSDs) of both the surface height and the initial aperture increases with normal stress. Figure 3 shows a log-log plot of
were determined at designated shear displacements of 2 and the PSD of the upper surface height (US) at normal stresses of
20 mm. In addition to the shear behavior, the closure curve 0.3 and 1 MPa and shear displacements (δs ) of 0 (initial) and
was measured by applying only normal stress at the desig- 20 mm for M150. This figure indicates that smaller asperities
nated shear displacements to clarify the effect of shearing on with shorter wavelengths were more damaged by shearing and
the non-linearity in the closure curve. that larger asperities with longer wavelengths are damaged as
The SDs of the initial aperture were plotted against shear the normal stress increases. Thus, the non-linearity in the clo-
displacement for different fracture sizes and normal stresses, sure curve during shear increases not only with fracture size,
as shown in Figure 1. It can be seen that the SD of the initial
Figure 1. Relation between shear displacement and the SD of the Figure 2. Effects of normal stress on the closure curve at shear
initial aperture as a function of normal stress and fracture size. displacement of 20 mm.
301

Figure 3. Power spectral density of the upper surface height at
normal stresses of 0.3 and 1 MPa for shear displacements of 0 and
20 mm for M150.
as already shown in previous researches, but also with nor-

mal stress, since large asperities with longer wavelengths are
more damaged by a greater normal stress and the matedness
of a fracture decreases with normal stress.
302

Study of the short and long-term brittle behavior of a porous rock with continuous
monitoring of (P- and S-) wave velocities under different hydrous conditions
D. Grgic
LAEGO, Nancy University (INPL-ENSG), Vandoeuvre-lès-Nancy, France
D. Hoxha
Laboratoire PRISME (EA4229), Polytech’ Orléans, Orléans, France
J. Eslami
LAEGO, Nancy University (INPL-ENSG), Vandoeuvre-lès-Nancy, France
Keywords: lab testing, rock failure, rock properties
1 EXTENDED ABSTRACT
The work presented here is part of a series of studies aimed

at increasing the understanding and at quantifying the mech-
anisms behind the short- and long-term behaviors of rocks
(iron ore, limestone, gypsum rock…) in relation to the prob-
lem of stability of underground mines and quarries in Lorraine
(North-Eastern region of France). In these shallow envi-
ronments, rocks constituting mines’ pillars are exposed to
variations in hydrous conditions which may impact on both
short- and long-term behaviors. In the brittle regime relevant
to mining conditions (low and constant temperature and low
confining pressure), the mechanisms of time-dependent dam-
age are of great importance in rocks. The main focus of this
work is to study, under different hydrous conditions, the short
and long-term micro-cracking damage of a porous oolithic Figure 1. Localization of the 5 (3-layer) piezoceramic transduc-
limestone during hydro-mechanical experiments through the ers on the surface of a cylindrical sample and directions of wave
continuous evolution of ultrasonic (US) waves velocities. polarization and propagation.
An experimental device, developed and used for the pur-
pose of this study, allowed us to measure simultaneously and independent elastic constants, the study of anisotropy evo-
continuously strains and five elastic wave velocities in two lution during loading requires the measurement of five inde-
different directions (3 P-waves and 2 S-waves) in hermeti- pendent US wave velocities (Fig. 1). Measurements of the
cally sealed uniaxial compression cells (Fig. 1). Short-term velocities of three compression waves (P) and two shear waves
compression stress-cycling tests and (long-term) multi-steps (S), in three different directions, were given by the five piezo-
creep tests were performed. Samples were subjected to satu- ceramic transducers. These velocities were referenced with
rated and partially saturated moisture conditions (the relative respect to the horizontal plane (Fig. 1), i.e., Vp(0◦ ) for the
humidity was controlled by using the saturated-salt-solutions horizontal-parallel (i.e., θ = 0◦ ) P-wave velocity, Vp(45◦ ) for
method). The piezoelectric transducers used in our study the 45◦ -to-horizontal (i.e., θ = 45◦ ) P-wave velocity, Vp(90◦ )
have a bandwidth frequency range between 100 and 230 kHz for the horizontal-perpendicular (i.e., θ = 90◦ ) P-wave veloc-
(resonant frequency = 150 kHz) and have three layers (Fig. 1); ity, Vsh(0◦ ) for the horizontal-parallel velocity of horizontally
the first layer is sensitive to compression waves (P-waves) polarized S-wave, andVsv(0◦ ) for the horizontal-parallel veloc-
while the second and third layers are sensitive to shear waves ity of vertically polarized S-wave. For an isotropic trans-
(S-waves) with two perpendicular directions of polarization. verse material, Vsv(0◦ ) is equal to the horizontal-perpendicular
Hence, with only a couple of transducers, three wave velocities S-wave velocity Vs(90◦ ) . Once the velocities were known, the
can be measured simultaneously at the same point. coefficients of the elastic stiffness tensor were calculated.
Since both geometry and mechanical loading have a cylin- The evolutions of wave velocities and dynamic elastic mod-
drical symmetry and the studied rock is initially isotropic, uli during short-term tests reproduced very remarkably the
one would expect that the anisotropy due to the micro-cracks shape of the stress-strains curves (Figs. 2). Since the wave
induced during uniaxial loading to be isotropic transverse. velocities were measured continuously, the continuous evolu-
For an isotropic transverse material, characterized by five tion of the dynamic Young’s modulus E11 during the uniaxial
303

Figure 2. Evolution of the axial, lateral, and volumetric total strains
and evolution of P- and S-waves velocities – uniaxial compression
test on sample ST2 (hr = 85%).
Figure 3. Evolution of the axial, lateral, and volumetric creep strains

loading is deduced providing evidence of the progressive dam- and evolution of P- and S-waves velocities – uniaxial multi-steps
age of the material. The anisotropy of the wave velocities creep test on sample C3 (hr = 85%).
developed progressively during the loading but this anisotropy
was reversible after full unloading, and this even for high
deviatoric stress almost up to the failure. However, after a opening of micro-cracks parallel to the direction of the applied
complete unloading, all waves velocities became smaller than uniaxial stress.
the initial values, reflecting an irreversible and isotropic dam- During the multi-steps creep tests, the P-wave velocities
age of the material. These experimental observations could remained definitively anisotropic after full unloading (Fig. 3).
be explained firstly by the nucleation of new cracks in scat- This is compatible with a subcritical crack propagation
tered and random orientation and preferentially in zones of mechanism which seems to be the dominant mechanism of
high stress concentration (e.g., grain contacts) for moderate cracking evolution during creep. The differences on crack
stress levels. Then, anisotropic coalescence and propagation evolution mechanisms between instantaneous and creep tests
of cracks occur at high stress levels (i.e., close to the failure). might signify that such irreversible mechanisms are rate
The damage of the limestone is thus mainly caused by oolithic dependent.
grains de-cohesion. In situ monitoring of seismic wave velocities could be
The anisotropy of the wave velocities during loading and used to infer macroscopic crack damage of pillars and there-
the reversibility of the anisotropy during unloading may be fore estimate the long-term stability of underground mines
explained by the preferential closure of pre-existing micro- and quarries. Particularly, since Vp(90◦ ) starts to decrease just
cracks and sharp voids perpendicular (or almost perpendicu- before the macroscopic failure, this velocity could be used to
lar) to the direction of the uniaxial stress and the preferential detect very precisely the pillars failure.
304

The effect of certain chemical solutions on some material

properties of clay bearing rocks
P. Hagan, S. Saydam, E. Elias, L. Deramore Denver & B. Hebblewhite

School of Mining Engineering, The University of New South Wales, Sydney, Australia
Keywords: lab testing, rock properties, rock support
The addition of water to rock is known to change its mate- A similar ameliorating effect trend was found with the use of
rial properties. These changes can be both beneficial as well a CuSO4 solution but it somewhat less falling midway between
as detrimental in terms of rock strength, stability of rock the results observed for water and for the KCl solution.
excavations and the process of mechanical rock excavation. It is postulated that the underlying mechanism is associated
This paper presents the results of a study to exam what effect with an exchange of cations in solution with those in the clay
certain chemical solutions have in ameliorating the changes minerals which was confirmed by analysis of the concentra-
in material properties brought about after clay bearing rock is tions of elements in solution before and after treatment of the
exposed to water. rock samples. Furthermore, the preferential cation exchange
Two chemical solutions, potassium chloride (KCl) and cop- with potassium and copper ions interferes with absorption of
per sulphate (CuSO4 ), were chosen on the basis that they water molecules and hence reduces swelling of the clay matrix.
would result in a chemical exchange with elements in the clay As well as altering the strength of rock, immersion in a
minerals. chemical solution also changed the mechanical rock cutting
The study confirmed earlier work whereby clay bearing parameters as shown in Figure 2. The cutting and normal
rock when exposed to water weakened the fabric of the mate- forces and specific energy reduced by 59%, 48% and 68%
rial. After immersing rock in distilled water for as little as respectively after immersion in water. The effect of KCl
30 minutes, rock strength as indicated by the average UCS though was found to differ depending on the period of immer-
reduced by nearly 50%. A similar weakening of the rock fab- sion with the longer immersion time of 72 hours resulting
ric was observed using the Slake Durability test and Schmidt in the cutting parameters retreating back towards the levels
Hammer. observed with untreated rock.
Immersion of rock samples in a solution of KCl was found Overall the results of this study indicate there is significant
to counteract the degradation effects of distilled water. As potential to alter the degradation in properties of clay bearing
shown in Figure 1, the UCS was of a similar level being only rocks brought about by exposure to water and hence improve
8% less than the untreated rock. the stability of excavations in rock and rock cutting conditions
with the use of certain chemical solutions.
Figure 1. Comparison of the average UCS strength for untreated

rock with the strength of rock that were immersed in water, KCl and Figure 2. Effect of water and KCl solution at different immersion
CuSO4 solutions. times on rock cutting parameters.
305

Creep test of tuff for over thirteen years and its know-how
K. Hashiba, S. Okubo & K. Fukui

Department of Systems Innovation, The University of Tokyo, Tokyo, Japan
Keywords: lab testing, rock properties, weak rock
Many researchers have studied creep of rock to understand its Occasionally the air pressure was adjusted and the pneumatic
time dependent behavior. However, important unsolved ques- lines were cleaned. During the second test, the air compressor
tions of long-term creep have still remained. In this study, a and the air conditioner malfunctioned and then were replaced
pneumatic creep testing machine was developed for long-term with new ones. The test had no other troubles and has been
continuous use. In the autumn of 1994, a uniaxial compression conducted for over thirteen years.
creep test was started, and up to now two sets of creep data Figure 1 shows the creep strain and stress from May 23,
have been obtained in two creep tests for over a year. One of 1997 to December 20, 2010 in the second test. From 1997 to
the tests is now under conducted and its period is over thirteen 1999, the final times when the displacement took some prede-
years. In this paper, the know-how and details of long-term termined values were automatically recorded in the computer
creep tests for rock were described. memory. After that the displacement and the air pressure were
In the testing machine, air was supplied from an air com- manually recorded on Monday and Friday every week. Except
pressor to a pressure-reducing valve and then delivered to an for small and large fluctuation probably caused by changes of
air cylinder with constant pressure. Displacement of a speci- temperature and creep stress, the creep strain increased under
men was measured with two cantilever-type sensors. Sample such a low stress along the thick solid line which has the slope
rock was Tage tuff yielded in Tochigi prefecture, Japan. A of 7 × 10−13 /s. From 2008 to 2010 the creep strain decreased
creep test was conducted under water-saturated condition, and though the creep stress was almost constant. The air cylin-
creep stress was 2.8 MPa, which is approximately 30% of uni- der probably had some kind of troubles and was dealt with in
axial compressive strength. The first creep test was started on October 2010.
November 22, 1994. Unnatural movement in the measured dis- Comparison of creep strain and strain-rate indicated that the
placement values was observed in August 1996 and therefore trends in the long-term creep tests with two specimens were
the test was halted. With this experience, countermeasures consistent with those for primary creep in short-term creep
against some troubles in measurement apparatus were pre- tests conducted in previous studies. These findings suggested
pared not to halt the test. The second creep test was started that deformation mechanism of long-term creep was similar to
on May 23 in 1997. During the test, the specimen, its dis- that of short-term one, and that common theory and constitu-
placement and the air pressure have been checked everyday. tive equation can be applied to both long- and short-term creep.
Figure 1. Creep strain and stress in the creep test for over thirteen years with Tage tuff under water saturated condition. The slope of the
thick solid line is 7 × 10−13 /s. Creep stress was adjusted to the predetermined value several times.
306

Macro-fracture characteristics of granite in different boundary conditions
M.C. He
School of Mechanics and Civil Engineering, China University of Mining and Technology (Beijing), Beijing, China
State Key Laboratory for GeoMechanics and Deep Underground Engineering, Beijing, China
W. Nie
Z.Y. Zhao
C. Cheng
Keywords: rock bursts; lab testing; rock failure; rock stress
1 INTRODUCTION K-feldspar (beige) and 5.0% clay minerals. The most abun-
dant clay minerals include 56% illite and 44% chlorite. The
Rock displays splitting fractures under uniaxial compression grains are visible to naked eyes, displaying a crystalline
and shear fractures under moderate confining pressure in lab. interlocking texture. The average density, uniaxial compres-
Rock exhibits plasticity under high confining pressures. Once sive strength and elastic modulus of granite are 2.57 g/cm3 ,
unloading occurs, the rock suffers a change from plasticity 131MPa and 21GPa, respectively. Both the UC test and the
to brittleness. Rockburst is one kind of sudden and violent CC test had a loading rate of 0.002 mm/s. All the tests on
changing progress from high confinement to no confinement granite were conducted on prismatic specimens of dimensions
involving large energy release in rock mass. Generally, the 150 mm (height) × 60 mm (breadth)×30 mm (thickness) by a
rock mass is under a static stress distribution before excava- true triaxial compression facility (He et al., 2007).
tion, and excavation results in the tractions on the excavation The one-dimensional variogram function, 2γ(x, h), can be
surface being reduced to zero within a short time period. expressed in the discretized form as:
Either the combined dynamic and static stresses exceeding
the strength of the rock mass, or the loss of rock structural
stability will lead to the occurrence of rockburst.
The application of laser profilometers in the rock surface
measurement offers a great tool to analyze the fracture prop- where x is the horizontal distance along a roughness profile,
erties. In this paper, the variogram method is used with a z(x) is the height of the roughness profile from the datum, M
suggested refining procedure to analyze the fracture surface of is the total number of pairs of toughness heights of the profile
granite specimens after rock burst, and compare the specimens that are spaced at a lag distance h. A program was developed to
failed under uniaxial compression with those under polyaxial apply the variogram fractal technique to analyze the profiles.
tests. The fractal dimension Drld and another parameter Kv The data, measured by laser profilometers, were available in a
are calculated and fitted by ellipses for quantifying the stress grid pattern in the form of z = f (x, y), where x-y is a referred
field. The distribution of these fractal parameters are used to coordinate system on the horizontal plane and the z coordinate
compare the rock deformable behavior under different stress at each point is the height of the fragment surface from the
boundary conditions. x-y plane in the normal direction.
2 MATERIALS AND METHODS 3 RESULTS AND DISCUSSIONS
Granite specimens, taken from Laizhou in Shandong province, The variogram method was applied to the two-dimensional
China, were used for analysis. This granite specimen consists roughness profiles that were calculated for the fragments sur-
of 27.0% quartz (gray), 31.0% plagioclase (white), 37.0% faces as a measurement in all directions (Kulatilake et al.,
307

The parameters of fractal dimensions were found with a
preferential orientation perpendicular to the direction apply-
ing the maximum force within the rupture of UC and CC tests.
The reasons are that, in UC test, the rock failed along the longi-
tudinal axis with stress-controlled tilt, whilst in the polyaxial
compression, the shear tractions are predominantly oriented
in the plane of the major and intermediate principal stresses
which caused tilting on the fracture. But in the rockburst,
the fracture twisted as the shear traction in the direction of
intermediate stress which emerged as well-distributed fractal
dimensions.
4 CONCLUSIONS
In this paper, granite specimens were studied under the uni-

axial compressive (UC) test, the polyaxial compression (CC)
test and the rockbursts simulation (RB) test, by a true-triaxial
testing machine. Laser profilometers were used to get the
macro-fracture of the granite specimen suffered different
boundary conditions. One of the self-affine fractals methods,
named variogram method, is proposed to present the surface
roughness of scanning profiles. The fractal dimensions (Drld ),
the scale-dependent fractal parameters (Kv ) and the product
of the above two parameters in all directions were shown as
rose diagrams. Ellipses envelops were drawn to describe the
non-homogeneous behavior of rock under different boundary
conditions. The maximum Drld of rock surface of UC tests was
on the direction of 10◦ from the axial force direction and 45◦
from the axial force direction. The Kv distribution changed a
lot in various orientations. Drld distribution on the surface of
CC specimen was more homogeneous than the other two tests
as the fitting ellipse envelop was more circular. On the surface
Figure 1. Representative rose diagrams to show the Drld distribu- of RB test, the maximum Drld was on the direction parallel to
tions (left) and Kv distributions (right) of (a) #5, (b) #8 and (c) #25 the major principal stress direction and the minimum fractal
specimen. dimension was nearly vertical to the major principal stress,
quite different from the other two tests. In general, rock under
RB tests has a much rougher surface than the UC test and the
2006). Mean values of fractal dimension (Drld ) and scale CC test. As this is just a preliminary investigation to analyze
dependent fractal parameter (Kv ) were estimated. the stress distribution by the variogram method, more studies
Figure 1 presents the Drld and Kv distribution in all should be carried out in the future for qualitative and quantita-
directions. The respective area of specimen #25 and #5 tive analysis on the fractal dimension about shear and tensile
is 25.6 mm × 25.6 mm, whilst the area of specimen #8 is fracture.
51.2 mm × 51.2 mm. Both on the top and the bottom of speci-
men #25, higher Drld value with little offset from the maximum
force direction, also the split direction, was presented. Their ACKNOWLEDGEMENTS
elliptical envelopes have higher eccentricity, more than 0.9.
However, in the middle of the specimen, Drld varies little and The authors wish to thank the Major State Basic Research
the eccentricity of its elliptical envelope is 0.686. Lower val- Development Program Project of China (2006CB202200),
ues of Kv between ±45◦ deviating from the maximum force GDUE Open Funding (SKLGDUEK0914) and Creative
direction are presented in all three areas even though the Team Development Project of the Chinese Education Ministry
maximum Kv of middle part is not the same as the top and (IRT0656).
the bottom parts. Directional variation of Drld of specimen #8
is much less than that of specimen #25. The eccentricities of
their elliptical envelops vary from 0.89 at the top, to 0.65 at the REFERENCES
bottom. Kv has disadvantage values near the maximum force
direction just as the same range as those of specimen #25. He M.C., Miao J.L., Li D.J. & Wang C.G. 2007. Experimental Study
on Rock bursts Processes of Granite Specimen at Great Depth[J].
Specimen #5 has a less variation for Drld distribution, whose
Chinese Journal of Rock Mechanics and Engineering, Vol.26:
eccentricities are 0.49 at the top, 0.70 at the middle and 0.76 865–867 (in Chinese).
at the bottom of the fracture surface. Unlike the directional Kulatilake PHSW., Balasingam P., Park Jinyong & Morgan R.
variation of Kv in other two specimens, the disadvantage Kv 2006. Natural rock joint roughness quantification through fractal
value is perpendicular to the maximum force direction and techniques. Geotechnical and Geological Engineering, Vol.24:
displayed by contrary. 1181–1202.
308

Experimental analysis of deformation mechanisms induced by desiccation

in Tournemire argillites using DIC method
S. Hedan & P. Cosenza

ENSIP, HydrASA CNRS UMR6269, Université de Poitiers, Poitiers, France
V. Valle
Institut PPRIME CNRS UPR3346, Université de Poitiers, Futuroscope Chasseneuil, France
A.L. Fauchille & P. Dudoignon

ENSIP, HydrASA CNRS UMR6269, Université de Poitiers, Poitiers, France
J. Cabrera
DEI-SARG Department, Institute for Radiological Protection and Nuclear Safety, Fontenay-aux-Roses, France
Keywords: lab testing, rock properties.
1 EXTENDED ABSTRACT In order to apply the desaturation path, our samples were
placed in an oven (with glass doors). The desiccation pro-
Fractures associated with the process of desaturation of cess was achieved by heating the samples to temperatures (T)
argillaceous medium were observed on faces and fronts of between 40◦ C and 200◦ C. The loss of water (w) was calculated
the galleries of the experimental platform of Tournemire. at the end of each thermal stage.
This hydric fracturing process was evidenced in situ by sub- The duration of each thermal stage (t) varied between from
horizontal cracks spaced out by about twenty centimeters, on one to a few hours. Prior to this desaturation path, samples
all vertical walls in contact with ambient air. were carefully (air) cut in two cores of about seven centimeters.
The aim of this experimental study (supported by GNR The first core was used to calculate (w) whereas strain fields
TRASSE, Transfert des Radionucléides dans les Sols, les (ε1 , ε2 ) were measured by DIC method through the glass doors
Sous-sols et vers les Ecosystèmes and IRSN, French Institute on the second core.
for Radioprotection and Nuclear Safety) was to investigate The results of this experimental investigation using optical
in laboratory this fracturing process by using a non invasive method (DIC) are the following:
and non destructive optical method (Digital Image Correla-
tion, DIC). This method consisted in (a) recording with IDS i) The cracks (aperture ≥80 µm) induced by heating began
µEye SE (CMOS 1/2 sensor) camera (1280 × 1024 pixels2 ), to be observable at 80◦ C. As observed on the gallery walls,
digital images of a sample undergoing a mechanical transfor- these cracks were parallel to the bedding planes and spaced
mation (heating) and (b) using an image correlation algorithm out by twenty-twenty five millimeters.
in order to obtain full-field displacements on the core surface. ii) These cracks appeared also at the interface between
The image correlation algorithm usually uses a random the clay matrix and the macroscopic non-clayey hetero-
speckle located at the sample surface. In our case, small geneities (e.g., calcareous nodules).
particles (<100 µm) deposed on a small thickness of glue iii) Before the observation of such cracks at a macroscale,
and covering the surface sample were chosen as the random strain localizations were measured close to the locations
speckle. The measuring resolution depends on the marking of the future cracks.
(speckle) and varies generally from 0.05 to 0.1 pixel. In this iv) When water was fully removed from samples near 100–
study, the measuring resolution was about 0.08 pixel. 150◦ C according to the duration of the thermal stage, the
Two sets of two samples, in an undisturbed state were used cracks were no more observable and seemed to be closed.
for this experimental study. These two sets were taken from This original sealing process is likely to be associated with
the boreholes GSM135 and GSM180, which were drilled in the thermal expansion of the solid phase.
2009 perpendicular to the wall of the North gallery of the v) The strain localizations were organized following two net-
Tournemire URL. The borehole GSM180 was subvertical and works: the main network was associated with the first
perpendicular to the bedding and GSM135 was drilled with cracks (parallel to the bedding planes) and a secondary net-
an angle of 135◦ from vertical. At the beginning of this study, work connected to the main network occurred in a second
the samples were in a near-saturated state. The samples were step, with a complex spatial organization.
200 mm high and had a 78 mm diameter.
309

Radiology of Canarian volcanic rocks
L.E. Hernández
Regional Ministry of Works, Government of the Canary Islands, Spain
A. Eff-Darwich, R. Viñas & J.A. Rodríguez-Losada

Department of Soil Science and Geology, University of La Laguna, Spain
Keywords: rock properties
1 INTRODUCTION analyze the geochemical, radiological and geotechnical char-

acteristics of the main lythological units of the Canary Islands.
The quantification of naturally occurring radionuclides in the The project also studies the potential of radon concentration
rock matrix, soil and building materials (including cements) measurements as a proxy of changes in the geological activ-
has been carried out in many countries in order to assess the ity as the result of seismic/volcanic processes. At the present
radiation dose affecting the public. The presence of natural stage of the project, 150 samples have been already analyzed,
radioactivity is partly due to primordial radionuclides con- although we expect to complete the project with 250 samples.
tained within the Earth’s crust, such as 40 K, 238 U, 232Th and We will present and discuss the data corresponding to the
the products of their decay series. In this sense, the levels concentrations of the main radionuclides, namely 40 K, 238 U,
of gamma radiation (external radiation) in naturally occurring 232Th, and their relations with the main chemical features of
radioactive materials (NORMs) depend upon their contents of the samples.
thorium, uranium and potassium, whereas the internal expo-
sure occurs through the inhalation of radon gas, a decay
product of 226 Ra (UNSCEAR, 2000).The specific activities of 2 RESULTS
40 K, 238 U, 232Th depend on geological and geographical con-
ditions as well as geochemical characteristics of the materials. In order to limit the external gamma radiation dose from
The worldwide average specific activities of 238 U, 232Th and building materials to 1.5 mSv/y per practice, it has been pro-
40 K in the earth’s crust are estimated at 32, 45 and 412 Bq/kg, posed a safety criterion, namely the external hazard index (He )
respectively (UNSCEAR, 2000). (Mujahid et al., 2008):
The Canary Islands, Spain, are a group of volcanic islands
where specific geological characteristics and many different
volcanic lythotypes are present; hence, these islands offer
a potential wide variety of naturally occurring radioactive where CRa , CTh and CK are the specific activities of radium,
materials. Few works have been carried out to characterize thorium and potassium expressed in Bq/kg. In addition to the
naturally occurring radionuclides in the Canaries. Some of external hazard, radon and its short-lived products are also
them (Robayna, 1998; Pinza, 1998) deal with the concentra- hazardous to the respiratory organs. To account for this threat
tion of radon in dwellings, whereas Fernández de Aldecoa the maximum permissible concentration for 226 Ra must be
et al. (2000) obtained maps of external gamma radiation half of the normal limit (185 Bq/kg). The internal exposure to
and concentration of radionuclides of the western Canaries. carcinogenic radon and its short-lived progeny is quantified
Doreste et al. (1998), Rubiano et al. (2006) and Arnedo et al. by the internal hazard index (Hi ):
(2006) obtained the natural radiation maps of the eastern
Canaries. These works were pioneering in establishing the
levels of natural radiation in the Canaries; however, a step
forward is necessary to account for some drawbacks of the The standard safety criterion requires that in both cases He < 1
abovementioned works. In the case of the western Canaries, and Hi < 1.
the existing radiation maps have a small spatial resolution, We found that the hazard indexes of our sample of rocks
namely 5 × 5 km2 or 10 × 10 km2 . In the case of the eastern closely follow their corresponding rates of magmatic differ-
Canaries, there is lack of information about the concentration entiation, as illustrated in Fig. 1. Both He and Hi are quite low
of radionuclides. In all cases, there are not data concerning at the mafic end of the distribution and trend strongly upward
the concentration of the natural radionuclides in relation to at the felsic end, exceeding in some cases the safety criterion.
the lythological and chemical features of the rock samples. The scatter of the data is also larger at the felsic end.
In this work, we present the first results of a project (funded If the data are plotted in a TAS diagram (Le Maitre, 1989),
by the Spanish Nuclear Security Agency, CSN) devoted to it is found that the samples exceeding the safety criterion
310

Figure 1. Geo-chemical variation diagram as a function of the exter- Figure 3. Geo-chemical variation diagram as a function of the
nal (solid circles) and internal (open circles) hazard indexes. Dashed uranium content.
lines indicate the upper safety limit of 1 for both activity indexes.
Figure 4. Geo-chemical variation diagram as a function of the

thorium content.
Figure 2. TAS diagram for the set of rocks used in this work. The
filled circles indicate those samples that exceed the safety criterions the geo-chemical characteristics of the materials, as illus-
expressed by the external and internal hazard indexes. trated in Figures 3 and 4. The specific activities of 238 U and
232Th of the sample of rocks exceed in most cases the aver-
(Fig. 2) are located within the fields that correspond to tra- age activities in the earth’s crust, namely 32 and 45 Bq/kg
chytes, phonolites, one case to phono-tephrite and one case to for 238 U and 232Th, respectively (UNSCEAR, 2000). There
silica oversaturated rhyolite. The last case is significant since is an increase in the content of radionuclides towards rocks
this rhyolite was extracted from a quarry located at the Tindaya associated to more differentiated magmas, although in the
Mountain, in Fuerteventura. There are plans to partially empty case of thorium and uranium, the scatter of data at the felsic
this mountain through the excavation of a cube of 50 × 50 × 50 end of the geo-chemical distribution is significantly larger.
meters within the mountain as part of a monumental project. Felsic rocks (mostly phonolites and trachytes in the Canaries)
Since the rhyolite exceedes the standard external and inter- contain significant amounts of tectosilicates (e.g. quartz and
nal hazard safety criterion, it is strongly recommended to feldespats). The crystalline network of tectosilicates are large
carry out extensive radiation dosimetry studies as part of the enough to accommodate elements such as thorium and ura-
geotechnical characterization of the site. nium that have relative large ionic sizes. In this sense, felsic
This work is allowing us for the first time to study the rocks contain more radionuclides than mafic rocks because
relationship between naturally occurring radionuclides and felsic rocks contain more tectosilicates than mafic rocks.
311

A new laboratory test for shear fracture formation and its

permeability measurement
T. Ito, T. Shono & K. Sekine

Institute of Fluid Science, Tohoku University, Japan
K. Yamamoto
Technology and Research Center, Japan Oil, Gas and Metals National Corporation (JOGMEC), Japan
Keywords: fluid flow, lab testing, rock joints, rock properties
Fractures in subsurface rocks should be closed and their sur-

faces contact each other due to rock stress in compression.
However, the fractures can maintain their permeability, since
they retain residual apertures caused by misfit of rough sur-
faces even at high normal stress. It is generally considered
that if shear sliding of fractures occur, the shear displace-
ment will enhance the misfit of fracture surfaces and it will
lead to increase in fracture permeability. Actually it has been
observed through laboratory experiments that permeability
of fractures with shear displacements were increased by sev-
eral orders compared with that of fractures with no shear
displacement (e.g., Chen et al., 2000). In the experiments, a
cylindrical specimen with a fracture was prepared by coring a
rock block containing a natural or artificially-created fracture.
The specimen with a fracture in shear condition was prepared
as follows; temporarily separating the fracture surfaces, they
were fitted again with a shear displacement of few millime-
ters in the axial direction. Then the specimen was supplied
for fluid flow experiments on the fracture with a shear dis-
placement to measure its permeability. However, under natural
conditions in subsurface rocks, it should be hard for fracture
surfaces to be separated completely as far as pore pressure Figure 1. Concept of the new testing method for shear fracture
formation and permeability measurement.
in the fracture is not raised abnormally to be equal to nor-
mal stress on the fracture. Thus the shear sliding of fracture
should occur actually while keeping contact of the fracture the proposed method for a rock sample of Kimachi sandsone.
surfaces, and such a manner is different to that of laboratory We prepared three specimens for those tests, and the distance
experiment. for pushing down the punch was changed for each specimen
Based upon these considerations, we have proposed a new from 0.7 to 3 mm. Results show that shear fractures were
testing method which allows us to create a single shear frac- formed successfully and the permeability of a rock increased
ture and measure its permeability sequentially under confining with existence of a shear fracture, however the increment was
pressure, once a specimen is set to an apparatus. Then the shear not so significant compared with the original permeability
sliding of fracture occurs while keeping contact of the frac- and there was no clear correlation between the permeability
ture surfaces, as expected to occur under natural conditions increment and shear displacements of the fracture surfaces.
in subsurface rocks. The method is illustrated in Figure 1. A Such behaviors are different from those observed in conven-
simple cylindrical specimen is initially subjected to confining tional laboratory experiments by Chen et al. (2000), where
pressure, an axial load is applied at the center of the specimen the fractures induced in a tensile mode were used in their
through a cylindrical punch, the axial load leads to shear fail- experiments. Our results are rather similar to the experimental
ure and accordingly an internal cylindrical portion is punched results obtained by Watanabe et al. (2008) using the fractures
down. In the same setup, fluid is injected passing through the induced in a shear mode similarly to the present study. Those
specimen, and finally, permeability of the specimen can be facts indicate that the dependency of fracture permeability
calculated from the flow rate and the pressure drop necessary on shear displacement is highly depend on fracture surface
for the fluid passing through the specimen. We demonstrated morphology.
312

Direct shear and tensile test on cemented healed joints from El Teniente mine, Chile
L. Jacobsson & M. Flansbjer

Building Technology and Mechanics, SP Technical Research Institute of Sweden, Borås, Sweden
A. Brzovic & C. de los Santos

División El Teniente, Codelco, Ranacagua, Chile
ABSTRACT: Direct shear and direct tensile tests have been conducted on specimens taken from drill cores of porphyry ore
from the El Teniente mine containing infilled veins and bands of e.g. copper sulphide. The direct shear tests were conducted
at different normal stress levels in order to determine the failure shear strengths. The direct tensile tests were conducted using
stiff load frame in order to be able to obtain the complete stress-deformation. The tensile failure stress and fracture energy was
computed in the cases when the failure was localized to the vein.
Keywords: rock joints, lab testing, rock failure
1 INTRODUCTION AE-measurements were carried out to see if it is possible to

get a further insight of the joint degradation process.
The geometric characteristics and strength of the geological
discontinuities are the key parameters to characterize rock 2 SPECIMEN AND TEST DESCRIPTION
masses for geotechnical and fragmentation considerations in
any panel caving underground mine. This is particularly rele- 2.1 Direct shear tests
vant at the El Teniente mine, because it has been observed that
the geological veins (cemented healed joints) that compose the The eighteen shear test specimens had veins with different
El Teniente porphyry ore deposit play a relevant role control- inclination against the core axis. The specimens were grouted
ling fragmentation and seismicity during caving propagation into the lower and upper holders with a space opening of
(Brzovic and Villaescusa, 2007, Brzovic, 2009). The weaker 10 mm at the shear plane using a high strength concrete. The
paths where the competent and massive primary copper ore setting time before testing was minimum 7 days. The tests were
breaks during the caving process are mainly thicker veins hav- carried out in a servo-hydraulic shear testing machine. A nor-
ing less than 1/3 of hard minerals as infill. The strength of the mal load was applied with loading rate of 5 MPa/min during
infill mineral assemblages has been assessed based on the loading and unloading stages according to the test schedule in
mineral abundances and their relative Mohs scale of hardness Table 1. The normal stress was kept constant during the shear
(Brzovic and Villaescusa, 2007). Tests undertaken at SP will stage and the specimens were sheared with rate of 0.5 mm/min
provide a further characterization of the strength. on the shear actuator. The forces were measured by means of
Direct shear tests and tensile tests are logical test methods electrical load cells and the relative displacement between the
to characterize joint strength properties. Direct shear tests on upper and lower shear box by LVDTs.
sealed joints were reported by e g Jacobsson and Flansbjer,
2005. The direct tensile strength test is a straight forward way 2.2 Direct tensile test
to achieve the tensile strength. Luong, 1990 pointed out that The joint planes in the six direct tensile test specimens were
imperfections during the test may cause a premature failure for approximately perpendicular to the longitudinal axis. Three
intact rock, e.g. eccentrically applied load. Direct tensile tests of the specimens were approximately 100 mm and three were
according to RILEM TC 162-TDF, 2001, aimed for concrete, shorter and extended by attaching metal distances. The spec-
have been conducted in this paper with a set-up similar to imens was glued directly to the loading platens according
Okubo and Fukui, 1996. Cylindrical specimens with parallel to RILEM TC 162-TDF, 2001. The tests were carried out
top and bottom faces are glued directly on the loading platens in a servo-hydraulic machine with a stiff load frame. The
and thereby achieving a perfect match between the load- displacement was measured locally over the vein with three
ing platens. The set-up also allows to achieve the post-peak displacement transducers with a gauge length of 30 mm cen-
response. The aim in this paper has been to explore possibility tered over the vein. The mean value of the displacement values
to obtain the mechanical properties of joints by direct shear was used for the displacement control. AE was recorded dur-
tests and direct tensile tests including the post-peak behaving four of the tests with the sensors attached to the upper and
ior and fracture energies from the load-displacement results. lower loading platens.
313

Table 1. Summary of direct shear tests.
Test sequence* Joint Shear

Area strength
Specimen 1 2 3 4 5 6 7 8 (cm2 ) (MPa)
DAC-2a 5 S 33.8 12.6

DAC-2b 5 S 35.6 12.7
DAC-3a 5 S 46.3 8.0
DAC-3b 10 S 44.8 16.1
DAC-4a 1 S 31.2 1.2
DAC-4b 5 S 38.7 2.9
DAC-4c 10 S 34.3 12.4
DAC-7 5 S 43.3 5.2
DAC-14a 15 5 S 32.4 17.5
DAC-14b 15 5 10 S 34.5 19.0
DAC-18 5 S 32.7 11.4
DAC-19a 15 5 S 30.9 19.1
DAC-19b 15 5 10 S 39.2 18.4
DAC-20a 15 5 S 28.8 12.1
DAC-20b 15 5 10 S 29.6 19.4
DAC-21b 5 S 39.5 15.7 Figure 9. Stress-displacement graph for DAC-24: (1) identifies the
DAC-22 20 0 20 0 20 0 5 S 35.5 15.3 first AE event with amplitude greater than 75 dB (red symbol), (2)
DAC-23 20 0 20 0 20 0 5 S 33.2 9.7 and (3) shows a clear intensity increase of AE-events due to increase
of the loading rate.
*The numbers 1, 5, 10, 15 and 20 refer to the normal compressive
stress level in MPa at the end of the normal load sequence and S be seen. Uneven deformations were recorded during two of
refers to the shear load sequence.
the tensile tests. This indicates that the veins may have been
undertaken an uneven opening resulting to a premature failure.
Table 2. Summary of direct tensile tests.
4 DISCUSSION AND CONCLUSION
Loading Tensile Fracture
Diameter rate strength energy
Specimen (mm) AE (µm/min) (MPa) (N/m) The mechanical behavior of veins in porphyry ore has been
examined. The strength of the veins is dependent on the
DAC-1 63.2 no 5.0 1.0 36.2 infill minerals. The tensile tests show that the full stress-
DAC-11 63.3 yes 1.5 2.7 –* displacement curve can be obtained the fracture energy for
DAC-13 63.0 yes 1.5 3.5 –* breaking the vein in tension can be calculated. Acoustic
emission recordings from two specimens provided results on
the failure process in the veins. The signals were localized to
the veins and the signals (>75 dB) yielded information of the
3 RESULTS failure progress in the veins.
3.1 Direct shear tests
REFERENCES
The fractures occurred in shear planes which fully or partly
contained the joint during the shear tests. A correlation of the
Brzovic, A. 2009. Rock mass strength and seismicity during caving
shear strength vales and the infill minerals and normal stress propagation at the El Teniente Mine, Chile. In C.A Tan (ed), 7th
can be seen. A summary of the results are shown below in Int. Symp. Rockbursts and Seimicity in Mines (RaSiM7). Dalian,
Table 1. China, Rinton Press, New York. Vol. 2, 838–52.
Brzovic, A. & Villaescusa, E. 2007. Rock mass characterization
3.2 Direct tensile tests and assessment of block-forming geological discontinuities dur-
ing caving of primary copper ore at the El Teniente mine, Chile.
The failures occurred in the joint planes in four of six cases Int. J. Rock Mech. Min. Sci. and Geomech, Abstr., 44, 565–83.
during the tensile tests which provided information of the local Jacobsson, L. & Flansbjer, M. 2005. Oskarshamn site investigation.
displacement at the vein. The fracture energy was computed Borehole KLX07A. Shear tests on sealed joints. SKB P-05-209,
as the area under the stress-deformation curve starting from Svensk Kärnbränslehantering AB.
the peak stress until failure. The stress-displacement graph for Luong, M.P. 1990. Tensile and shear strengths of concrete and rock.
DAC-24 is shown in Figure 8. A summary of the tensile tests is Eng. Fract. Mech., 35, 127–135.
presented in Table 2. AE measurements with line localization Okubo, S. & Fukui, F. 1996. Complete stress-strain curves for various
rock types in uniaxial tension. Int. J. Rock Mech. Min. Sci. and
of the AE events were conducted during four of the tensile
Geomech, Abstr., 33, 549–556.
tests. Events with higher amplitudes were registered close to RILEM TC 162-TDF 2001, Test and design methods for steel
the vein where the main fracture process occurred. An increase fibre reinforced concrete – Uni-axial tension test for steel fibre
of AE events just before the maximum stress was reached can reinforced concrete. Materials and Structures, 34, pp. 3–6.
314

Rheological behavior of sandstone under unloading
Y.Z. Jiang
Key Laboratory of Geotechnical Mechanics and Engineering of the Ministry of Water Resources,
Yangtze River Scientific Research Institute, Wuhan, China
R.H. Wang
Key Laboratory of Geological Hazards on Three Gorges Reservoir Area, China Three Gorges University,
Ministry of Education, Yichang, China
Keywords: rock mechanics; unloading; sandstone; rheology; experimental study
For investigating the rheological properties of rock mass in samples became more and more obvious under the condition
the loaded state, there’re a lot of researches both abroad and at of unloading confining pressure sequentially and maintaining
home, which makes great progress, however, there are rarely stress level and confining pressure constant, and sandstone
experiments and theoretical studies on rheological mechani- rock samples, when in the final level of confining pressure
cal properties under the condition of unloading. Engineering levels, showed the three typical rheological stages over time,
practice shows that the instability and destruction of rock which are primary attenuating rheological stage, the steady
excavation does not occur immediately after the completion uniform rheological stage and accelerating rheological stage,
of the excavation, and the deformation due to excavation and finally it came into rheological destruction of non-linear
unloading has the time effect, and unloading is a time-related accelerating.
concept, including the instantaneous unloading, as well as Based on the results of triaxial compression rheological
long-term unloading. It is quite necessary to carry out rheolog- experiments, the parameters of Burgers rheological model
ical test of rock mass under unloading for the study of the rock were identified with the optimized algorithms, the non-
mass long-term stability, which has an important engineering stationary law between rheological parameters and confining
significance. pressure, stress level were studied. From the regression anal-
The approach of step loading on the same rock was adopted ysis results, it can be seen that the precision of the results
to carry out indoor rheological test. Load hydrostatic pressure of the exponential function is slightly lower than the return
to 30 MPa at the rate of 0.05 MPa/s firstly; when deformation fitting power function fitting precision. However, the correla-
was stable, maintain confining pressure at 30 MPa, continue tion coefficients of two functions have reached more than 0.95.
applying axial force to the set value at the rate of 0.5 KN/s, and The trend function could explain the regularity of feature vari-
maintain it constant for hours; when rock rheological deforma- able parameters to some extent. But for specific rheological
tion under the level of former class became stable, maintain parameters, the follow factors should be taken into consid-
axial force σ1 constant, and unload confining pressure step eration: the physical meaning of the rheological parameters;
by step at the rate of 5 MPa each level. Pay attention to that the corresponding combination of physical and mechanical
when each level of confining pressure unloaded, the emer- mechanisms. The rheological parameter η1 of Burgers rheo-
gence of rock steady-state creep are required to be observed, logical model reflects the size of the flow rate when sandstone
and then unloaded next level of confining pressure, until the is in a steady flow stage. When the confining pressure less or
destruction of rock flow occurred. The rate for unloading equal to zero, the rheological parameter η1 will corresponding
confining pressure were all 0.05 MPa/s. Control temperature decrease, but will not reduce or close to zero, because when
and humidity condition maintained constant during the exper- the rheological parameters η1 tends to zero, the corresponding
iment. By using of rock servo-controlling rheology testing uniform steady state flow stage of the flow rate will tend to
machine, triaxial compression rheological experiments were infinity, so ,as the confining pressure tends to zero, the rhe-
carried out, the variation laws of rock deformation with time ological parameters η1 tends to a constant greater than zero.
went was analyzed under different confining pressure, the The exponential function can be used to describe the relation-
deformation properties of different rocks in the rock triaxial ship between the rheological parameters η1 of sandstone and
rheological process were studied, and the development trends confining pressure.
and characteristics of rock deformation in stage of acceler- The phenomenon that for rock nonlinear rheological will
ated rheology were discussed. At the same time, the rupture happen in long term has gradually be recognised and
forms and mechanisms were analyzed. It can be seen from concerned; the development of rock rheology damage and
tests that, the rheological phenomena of sandstone samples unloading capacity are related to action time, represented
took place significantly over time in the long-term external in non-linear rule. When the exterior loads a rock suffered
loads, the rheological deformation characteristics of sandstone is greater than the long-term strength, the rock will fall
315

into the rheology accelerates stage and will be destroyed creep deformation and the early stage of uniform steady-
eventually. The rock under a constant load typically expe- state creep strain, but can not simulate the acceleration phase
riences three rheology stages. In the initial stages of decay, of accelerated deformation rheological characteristics and
the rheology rate decrease gradually, representing a clear trends. In order to improve classical linear rheological model
hardening phenomenon; when the rheology enter into the of rock that does not reflect the nonlinear rheological prop-
uniform steady-state phase, the rheology rate remains at a erties of rock enough, the damage factor was added to the
constant value, and approximately varying linearly at a certain burgers rheological model. The corresponding damage rheol-
slope. When the rheology deformation come into the rheology ogy model was proposed. In order to verify the adaptation and
accelerating stage, the deformation slope increases gradually, rationality of the rock nonlinear rheological model, used accel-
showing the trend of non-linear acceleration, which indicates erating rheological test curve of sandstone samples under the
that rocks produces a new damage. As time arises, the dam- condition of 15 MPa confining pressure and 135 MPa stress
age of the rock increase continuously and the deformation level to verify nonlinear rheological damage model of rock,
grows lager and lager at the same time until the rock break. It and the results showed that the fitting results of rheological
represents a dramatic nonlinear characteristic. test data and the improved rheological model had good con-
Assume that at the rheology accelerating stage the evolve- sistency, and the rheological model can reflect the three stages
ment of rheology damage represent a exponential function of of rock nonlinear rheology, which can accurately reflect the
the unloading degree and time, the damage evolving function rheological characteristics under the condition of unloading
was deduced. Classic Burgers rheological model can reflect sandstone.
the instantaneous elastic deformation of rock, attenuation
316

Strength evaluation of discontinuous rock mass by multi-scale limit load analysis
J. Jung, J. Kato, K. Terada & T. Kyoya

Department of Civil Engineering, Tohoku University, Sendai, Japan
Keywords: rock mass, numerical modeling
1 INTRODUCTION
The rock mass in/on which engineering structures are

constructed intrinsically involves certain characteristic
microstructures comprised of intact rocks and spatially dis-
tributed cracks. Since the mechanical properties of the rock
mass that are nothing less than the result of the mechanical
response of such complicated microstructure, reasonable and
reliable characterization of such properties of rock masses is
an extremely important problem in rock mechanics and rock
engineering.
Multi-scale analysis method, which is based on rigor-
ous mathematical homogenization theory, seems suitable for
analysis of mechanical behavior of bodies containing such
complicated microstructures. In the multi-scale analysis the
mechanical behavior of a body containing a microstructure
is treated in terms of two sets of governing equations on
two different scales. The macroscopic equations describe a
boundary value problem for the whole body, and the micro-
scopic equations describe the behavior of a unit cell of the
microstructure. The solution of the microscopic problem plays
a role of nonlinear constitutive law for a material point of Figure 1. Macroscopic failure criterion for a unit cell with a single
the equivalent continuum of the macroscopic whole body. So, crack (θ = 0 degree) and that of the intact rock.
the multi-scale analyses are applicable to various bodies with
various microstructures, if an appropriate unit cell model of saddle point theorem and is performed by applying ordinary
such microstructure can be prepared. two-dimensional finite element method. Figure 1 presents the
In this study we propose a method of multi-scale analysis determined failure criterion used in this study together with
with a frictional contact analysis of cracks being incorpo- that for an intact unit cell model without any crack.
rated into the part of microscopic analysis of the aforemen-
tioned multi-scale framework. At the end several numeri-
cal examples are shown to validate the proposed method
3 CONCLUSION
and to demonstrate its potential usefulness in the field of
rock mechanics and rock engineering.
In this paper, we have proposed a method of strength evalua-
tion of discontinuous rock mass by as multi-scale analysis. The
2 EVALUATION OF STRENGTH CHARACTERISTICS numerical example of the block shearing test on discontinu-
ous rock mass has shown that the applicability of our method
We adopt a unit cell model with a single crack in order to to evaluation of strength characteristic of discontinuous rock
analyze the behavior of discontinuous rock mass. This study mass.
determines a failure criterion of a macro-body that contains
a microstructure by utilizing the localization process as a
numerical strength test (Kyoya et al. 1999). REFERENCES
We assume that the constituent material is linearly elas-
tic, specifically considering only the nonlinearity that is Kyoya, T., Terada, K. & Kawamoto, T. 1999, Multi-scale limit load
attributable to the frictional contact behavior of the crack. analysis for discontinuous rock mass based on the homogenization
The frictional contact analysis in the unit cell is formulated method, Int. J. Numer. Anal. Meth. Geomech., vol. 23, pp. 995–
based on the Lagrange multiplier method and the relevant 1019.
317

Estimation of fracture toughness of anisotropic rocks by SCB test and

visualization of fracture by means of X-ray CT
M. Kataoka & Y. Obara

Department of Civil and Environmental Engineering, Graduate School of Science and Technology,
Kumamoto University, Kumamoto, Japan
M. Kuruppu
Western Australian School of Mines, Faculty of Science and Engineering, Curtin University,
Kalgoorlie, Western Australia, Australia
Keywords: lab testing, rock failure
In order to investigate the influence of anisotropic charac- On the other hand, in 3-2 type, the crack propagates along grain
teristics of rock and water vapor pressure of surrounding boundaries smoothly.
environment on fracture toughness, a series of Semi-Circular The results obtained in this paper are summarized as
Bend (SCB) tests under various water vapor pressures were follows:
performed.
1) The granite has anisotropic property. The fracture tough-
SCB specimen is a semi-circular with an artificial notch and
ness shows the directional dependence;
subjected to three-point bend loading. The setup of specimen
2) The fracture toughness depends on water vapor pressure
for SCB test is shown in Figure 1. The load is applied through
and decreases with increasing water vapor pressure; and
one upper and two lower rollers. The setup of specimen is
3) Based on the X-ray CT images of the fractured specimen,
placed inside a vacuum chamber used to control surrounding
the fracture toughness is influenced by the type of crack
environment of specimen. The load is recorded using a load
propagation.
cell equipped on a loading plate.
The rocks used in the test were African granite and Korean
granite. The measurement of elastic wave velocity was per-
formed to make clear orientation of inherent micro cracks.
Considering the direction of the artificial notch, two types of
specimen were prepared, 1-2 type and 3-2 type as shown in
Figure 2.
The relation between fracture toughness and water vapor
pressure is shown in Figure 3. The values of fracture tough-
ness of 1-2 type are larger than that of 3-2 type for both
granites. The fracture toughness varies almost linearly against
water vapor pressure on the logarithmic graph and increases
with decreasing water vapor pressure. The equation of the
approximated lines is:
Figure 3. Relation between fracture toughness and water vapor
pressure.
where β is a constant and m is a slope of the line.
Furthermore, crack propagation was discussed based on the
images of fracture visualized within rock specimen after SCB
test by means of X-ray CT scanner. The gray-level CT images
near the artificial notch tip of the fractured specimen are shown
in Figure 4. It is found that the crack propagates near the notch
tip along boundaries around grains or across grains in 1-2 type.
Figure 1. Setup of specimen. Figure 2. Two types of specimens. Figure 4. CT images near notch tip of fractured specimen.
318

Laboratory measurement of poroelastic and hydraulic constants of sedimentary rocks

focused on pore water discharge from specimen with its deformation
M. Kato
Faculty of Engineering, Hokkaido University, Sapporo, Hokkaido, Japan

Advanced Industrial Science and Technology, Tsukuba, Ibaraki, Japan
K. Kaneko
Keywords: lab testing, rock properties, fluid flow, nuclear repository
The relation between rock mass deformation and groundwater kept constant during the test, second, the condition of triaxial
behavior is of importance for several fields in earth science compression, which the axial stress is suddenly subjected to
and engineering including geophysics, structural geology, both ends while confining pressure and pore pressure at both
rock physics, engineering geology, and geomechanics. The ends are kept constant, and third, the pore pressure is suddenly
knowledge of this relation is needed particularly for repos- changed at both ends while confining pressure is kept constant.
itories design for radioactive waste disposal. In terms of a During these tests, axial and circumferential strains are mea-
geological barrier of the repository system, sedimentary for- sured by the extensometers and discharge of pore water from
mations are predominant to guarantee that the radioactivity a specimen by the displacement of piston of syringe pump
cannot escape for several centuries. (Figs. 1, 2).
In this study, poroelastic and hydraulic parameters of sed- Hydraulic constants obtained from triaxial tests agreed
imentary rocks were estimated simultaneously from triaxial well with those from transient pulse permeability tests. The
compression tests with instantaneous change either of axial experimental results for discharges from the specimen dur-
stress, confining pressure, or pore pressure. The triaxial tests ing the tests permit theoretical interpretations. Therefore,
were conducted under the drained condition but under the besides the behavior of pore water, the transient response
undrained condition because of the difficulty of strict real- of rock deformation corresponding to pore water move-
ization. Stress conditions adopted in the tests are grouped ment can be theoretically treated. Accordingly hydraulic
into three types: first, the condition of hydrostatic compres- constants can be estimated from the strain data of triax-
sion, which all principle stresses are changed equally and ial tests with instantaneous change either of stress or pore
suddenly while pore pressure at both ends of a specimen is pressure.
Figure 1. Variations of strains and specific discharge with time from Figure 2. Variations of strains and specific discharge with time from
triaxial test with sudden change of axial stress. triaxial test with sharp change of pore pressure.
319

Relationship between point load strength and uniaxial compressive strength of

hydrothermally altered soft rocks
M. Kohno
Cold Regions, Environmental and Energy Engineering Course, Graduate School of Engineering,
National University Corporation Kitami Institute of Technology, Kitami, Japan
H. Maeda
Department of Civil and Environmental Engineering, Faculty of Engineering,
National University Corporation Kitami Institute of Technology, Kitami, Japan
Keywords: weak rock, rock properties, rock failure, rock mass, lab testing, field measurements
The strength of fresh rocks and altered rocks, including

hydrothermally altered or weathered rocks, is generally evalu-
ated based on uniaxial compressive strength (UCS). However,
rock core pieces for UCS tests cannot always be obtained
from outcrops of faulted, jointed or cracked rock masses.
In these cases, the point load strength (PLS) test is a con-
venient and effective alternative to the UCS test, because it
can be done promptly using on-site testing equipment with
various shaped small rock specimens taken from outcrops or
floats. Provided that we can calculate a UCS estimate from
a PLS value, the PLS test can lead more convenient and
cheaper. Many researchers have already studied the relation-
ship between the PLS and UCS of hard rocks. However, the
relationships between the PLS and UCS of soft rocks and irreg-
ular lump specimens have not been estimated. Furthermore,
Figure 1. Relationships between point load strength and uniaxial
the loading direction, specimen size and shape, and methods compressive strength in hydrothermally altered soft rocks.
used to calculate the PLS have not been standardized. For these
reasons, the relationships given above are far from practical,
for soft rocks. The purpose of this study was to clarify the
relationships between the PLS and UCS in hydrothermally In soft rocks, the relationships between the UCS and PLS
altered rocks, which are typical of the soft rocks found in in the forced-dry and forced-wet states were similar. There-
northeastern Hokkaido, Japan, using axial and irregular lump fore, the forced-dry and forced-wet states can be combined to
PLS test specimens. give an overall relationship between the UCS and PLS values.
The rock samples, which were collected primarily from the The relationships between axial PLS (Is(a) ) and UCS (qu ), and
earth’s surface in ancient hydrothermal fields, were hydro- irregular lump PLS (Is(i) ) and UCS (qu ) were qu = 11.1Is(a)
thermally altered volcaniclastic rocks. and qu = 4.7Is(i) , respectively, in soft rocks with a UCS
The rock specimens used for the axial PLS and UCS tests below 25 MPa (Figure 1). The correlation coefficients R were
in this study had a diameter of 50 mm, and length of 20 mm 0.85 and 0.78, respectively, indicating a strong correlation
and 100 mm, respectively. These shapes were easy to procure, (Figure 1).
having the advantage of using boring cores. On the other hand, The number of tested specimens satisfied accuracy require-
the rock specimens used for the irregular lump PLS test in this ments, based on the coefficient of variation. The axial and
study were sized to contain a ball with a diameter equal to the irregular lump PLSs were strongly correlated with the UCS.
distance between the two loading platens. Therefore, the relationships between PLS and UCS established
The axial PLS, irregular lump PLS, and UCS tests were in this study were highly precise. We could calculate the UCS
performed using a laboratory testing machine with specimens from PLS tests on soft rocks only when their UCSs were below
in a forced-dry state, dried in an electric oven at a temperature 25 MPa.
below 60 ± 3◦ C for 4 days or more to achieve a constant mass, Based on the results of this study, the PLS of hydrothermally
and specimens in a forced-wet state, saturated with water for altered rocks from hydrothermal alteration zone landslide
15 days or more to achieve a constant mass. The numbers of areas can be useful when evaluating modern landslide haz-
specimens tested were 1,755 rock specimens for the axial PLS ards and constructing landslide hazard maps of ancient and
test, 2,123 rock specimens for the irregular lump PLS test, and active hydrothermal fields.
329 rock specimens for the UCS test, making a total of 4,207
hydrothermally altered rocks.
320

Study on the dilatancy angle of rocks in the pre-failure domain
M. Kwaśiewski & P. Rodríguez-Oitabén

Faculty of Mining and Geology, Silesian University of Technology, Gliwice, Poland
Keywords: lab testing, numerical modeling, rock properties
It has been well established experimentally that dilatancy The Brenna sandstone was selected intentionally since it is
may appear at relatively early stages of the deformation pro- a rock that exhibits a considerably high compressive strength
cess under compressive loading conditions. In particular, in and, at the same time, behaves in a highly dilatant manner.
medium- and coarse-grained clastic sedimentary rocks, the The threshold of absolute dilatancy of the Brenna sandstone
onset of dilatancy may not exceed several percent of the peak amounted to, on average, 50% of the ultimate strength and the
differential stress under low confining pressure conditions, mean value of the dilatant volumetric strain at strength failure
and an absolute increase in volume often begins at stress lev- reached almost 1.4%.
els as low as 40 percent of the ultimate strength. Unfortunately, For modeling purposes, the stress-strain curves as obtained
these effects are often neglected in the mathematical model- from laboratory compression tests, were simplified to a piece-
ing of rocks and rock masses. Even those researchers who wise linear form (Fig. 1). The proposed idealized model has
acknowledge the pre-failure dilatancy in rocks tend to dimin- the following features: (i) it is a bilinear model in the pre-
ish its importance and when building a model, they allow failure domain which allows for the compaction and dilation
for the increase in volume only in the post-peak domain,
that is they stick to the very unrealistic assumption that it is
solely compactant behavior that rocks manifest in the pre-peak
domain (see Brown et al. 1983,Vermeer & de Borst 1984,Yuan
& Harrison 2004, Alejano & Alonso 2005, Zhao & Cai 2010).
In order to lay the foundation for a future challenge of this
common practice of excessive simplification of the volumetric
deformation mode of rocks in the pre-peak domain, a study
has been undertaken by the present authors with the aim of
creating a simple yet realistic model capable of capturing both
of the phenomena that are usually observed in the pre-failure
domain: compaction at the early stage of deformation and the
dilation that follows compaction and precedes strength failure.
In order to reveal the main features of the volumetric defor-
mation mode of a rock in the pre-peak domain and, first of all
to determine the values of the dilatancy angle in this domain,
samples of a medium-grained sandstone from Brenna in the
˛ Mountains in southern Poland were tested under
Beskid Ślaski Figure 1. Simplified, piecewise linear characteristics representing
uniaxial compression conditions. the behavior of rock in the pre-failure domain.
Figure 2. Slopes of the volumetric strain vs axial strain characteristic (a) and the lateral strain vs axial strain characteristic (b) in the
contractant domain (0-TD) and in the pre-peak dilatant domain (TD-F).
321

Table 1. Values of the parameters of the piecewise idealized model, When using the total volumetric strain vs total axial strain
as determined for Brenna sandstone. or the total lateral strain vs total axial strain characteristics
Parameter ν ψ p
ψ
simplified to a piecewise linear form, an average value of the
dilatancy angle equal to about 33◦ was obtained. In contrast,
Minimum value 0.18 26.5◦ 56.2◦ a dilatancy angle of 65◦ was obtained when only the plastic
Maximum value 0.22 38.8◦ 69.9◦ components of volumetric and axial strains were taken into
Mean value 0.20 32.9◦ 65.0◦ account. It is interesting to note that the dilatancy angle values
determined based on plastic strains (p ψ) are, on average, two
times higher than those (ψ) determined based on the total
strains (Table 1).
of the rock material, (ii) this model uses two material parame- It should be noted that the results listed in Table 1 differ from
ters, namely Poisson’s ratio and the dilatancy angle, (iii) these the data that can be found in the literature where the opinion
parameters can be estimated based on the relationship between prevails that the dilatancy angle is significantly smaller than
either the volumetric strain and the axial strain or the lateral the angle of internal friction not only for soils but also for
strain and the axial strain. Accordingly, Poisson’s ratio can concrete and rocks. It is also noteworthy that the values of the
be determined either from the slope of the volumetric strain dilatancy angle characterizing the dilatant behavior of Brenna
vs axial strain characteristic in the pre-dilatant domain (0-TD sandstone in the pre-failure domain are considerably higher
range in Figure 2a): than those determined by other researchers who focused on
the dilatant behavior of rocks in the post-peak domain only.
REFERENCES
or from the slope of the lateral strain vs axial strain character-
istic in the same domain (Fig. 2b): Alejano, L.R. & Alonso, E. 2005. Considerations of the dilatancy
angle in rocks and rock masses. International Journal of Rock
Mechanics & Mining Sciences 42:481–507.
Brown, E.T., Bray, J.W., Ladanyi, B. & Hoek, E. 1983. Ground
Similarly, the dilatancy angle can be determined either from response curves for rock tunnels. Journal of Geotechnical Engi-
the slope of the volumetric strain vs axial strain characteristic neering 109(1):15–39.
Vermeer, P.A. & de Borst, R. 1984. Non-associated plasticity for
in the pre-peak dilatant domain (TD-F range in Figure 2a): soils, concrete and rock. Heron 29(3):3–64.
Yuan, S.-C. & Harrison, J.P. 2004. An empirical dilatancy index for
the dilatant deformation of rock. International Journal of Rock
Mechanics & Mining Sciences 41:679–686.
Zhao, X.G. & Cai, M. 2010. A mobilized dilation angle model
or from the slope of the lateral strain vs axial strain character- for rocks. International Journal of Rock Mechanics & Mining
istic in the same domain (Fig. 2b): Sciences 47:368–384.
322

The thermal and mechanical properties around an underground

research tunnel in granite body
S. Kwon
Department of Energy Resources, Inha University
J.S. Kim, C.S. Lee & W.J. Cho

Korea Atomic Energy Research Institute, Daejeon, Republic of Korea
Keywords: rock properties, lab testing, nuclear repository, underground research tunnel
1 INSTRUCTIONS
In Korea, 20 nuclear power plants are now operating and pro-

duce about 40% of total electricity. From the operation of the
power plants, a significant amount of spent fuels as well as
low and intermediate wastes have been generated and the sug-
gestion of a reliable and feasible management technique for
the radioactive waste is an urgent issue for the sustainable uti-
lization of nuclear energy. Korea Atomic Energy Research
Institute (KAERI) developed a reference disposal system,
KRS, for the disposal of high-level radioactive waste, HLW,
in a deep geological formation in 2006. According to the
reference disposal system, the spent fuels of 36,000 tons gen-
erated from the nuclear power plants during their life time
will be disposed of at several hundred meters deep under- Figure 1. Variation of RMR with depth.
ground repository constructed in a crystalline rock mass. For
assuring the safety and performance of the repository for a soil to good conditioned hard rock, it is expected that the rock
long period of time, a multi-barriers system consisted of cask, properties vary along the tunnel. Fig. 1 shows the variation of
buffer, backfill, and rock was adopted in KRS. RMR with depth. Rock strengths and elastic modulus showed
In order to validate the safety, reliability, and technical fea- similar pattern as Fig. 1. Using Hoek & Brown equation The
sibility of the disposal system, an underground research tunnel deformation modulus could be estimated to vary from 3.3 GPa
was constructed in Nov. 2006. The research tunnel, KURT, is to 37 GPa along the tunnel and the average was 16 GPa.
located inside of KAERI territory and the major rock type
around the site is granite. It is consisted of an 180 m long
3 THERMAL PROPERTIES
and −10% declined access tunnel and two research modules
with 45 m and 30 m length. During the site characterization
The understanding of thermal properties of rock and other
and the excavation of the 6 m × 6 m tunnel, several disconti-
parameters, which can influence on the thermal behavior of
nuities were encountered. During and after the construction of
rock, is important for the validation of the disposal system
KURT, different in situ and laboratory tests were carried out
and the adequate design of disposal tunnel and deposition
to characterize the geological condition and rock properties.
hole. Since high-level radioactive waste generates significant
One of the in situ tests is borehole heater test for investigat-
amount of decay heat for a long time, the investigation of
ing the thermal-mechanical coupling behavior under heating
thermal behavior as well as the coupling effect with other
and cooling phases. For the design of such an in situ test and
mechanical and hydraulic behaviors is critical.
the interpretation of results, the clear understanding of the
thermal, hydraulic, and mechanical properties of rock around
3.1 Measurement of thermal properties using rock cores
KURT is essential. In this study, the initial thermal properties
and mechanical properties of rock were analyzed extensively. Thermal conductivities of KURT rocks extracted from differ-
ent locations were measured under different saturation con-
ditions. The average thermal conductivities in dry condition,
2 MECHANICAL PROPERTIES 2.89 W/(m · K), is about 90% of that in saturated condition,
3.2 W/(m · K). It was possible to determine the relationship
Understanding of the in situ stress condition and mechani- between the thermal conductivity and porosity, saturation, and
cal properties of the rock around the tunnel is important for depth(Kwon et al, 2010). In order to consider the effect of
tunnel design and stability analysis. Since the access tunnel saturation, three-phase system consisted of minerals, water,
of KURT is located in the area characterized from weathered and air, should be considered.
323

Chaudhary & Bhandari (1968) reported that the geomet-
ric mean, which was good for two-phase system, did not
agree with measurements in three-phase system except near
the saturation region. They suggested an equation, in which
a weighting factor was introduced to take into account the
fraction of parallel and serial combinations in a sample:
where, k1 and k2 are arithmetic and harmonic means, respec-

tively, and n is the fractional factor for parallel combination.
For KURT granite, the best fitting fractional factor was found
to be 0.87. Using the equation, the thermal conductivity with
different mineral composition, porosity, and saturation can be
estimated. When the equation was applied to KURT rock, it
was possible to derive that the thermal conductivity increases Figure 2. Model mesh used for the simulation of KURT.
rapidly in shallow depth and converges to 3.2 W/(m · K) from
the depth of 40∼50 m. 5 CONCLUSIONS
3.2 Thermal expansion coefficient For suggesting a reliable and safe disposal system for HLW,
Thermal expansion coefficient of KURT rock was measured the validation of the system in underground research facility
in the testing range of −150∼1000◦ C. There was a clear trend is essential. In Korea, KURT was constructed and will be used
of increasing expansion coefficient with increasing tempera- for various in situ tests related to HLW disposal program. In
ture. In a normal temperature condition, the thermal expansion order to choose an adequate testing location, to design the
coefficient varies from 2.03 to 10.36 microstrain/◦ C and tests, and to analyze the test results, the understanding of the
average is 7.48 microstrain/◦ C. rock properties and geological conditions is important. From
the laboratory and in situ tests and theoretical prediction, it
3.3 Specific heat was possible to determine the variation of rock properties with
depth and other parameters such as saturation. With the vari-
Specific heat of KURT granite was measured at different sat- ation of geological condition, topography around KURT, and
uration using rock specimens with different porosities ranges rock properties variation, the initial thermal and mechanical
from 0.5% to 2.6% (Cho and Kwon, 2010). When the rock is conditions around the tunnel could be described more clearly.
fully saturated, the average specific heat capacity was 820 J/kg
k, which was about 6% higher than that of dry rock.
REFERENCES
4 COMPUTER SIMULATION Chaudhary, D.R. & Bhandari, R.C. 1968. Heat transfer through a
three-phase porous medium. Brit. J. Appl. Phys 1:815–817.
Using the rock properties at KURT, the thermal and mechan- Cho, W.J. & Kwon, S. 2010. Estimatino of the thermal properties for
ical conditions before and after the construction of the tun- partially saturated granite, Engineering Geology 115:132–138.
nel could be predicted. In the computer simulation using Kwon, S., Choi, W.J. & Choi, J.W. 2010. Initial Thermal Conditions
FLAC3D, the topography above the tunnel, rock property around an Underground Research Tunnel at Shallow Depth, Int. J.
variation, geothermal gradient, seasonal temperature varia- of Rock Mechanics and Mining Sciences 48:86–94.
tion, and sequential tunnel excavation were considered. Fig. 2
shows the model mesh.
324

Shear behavior model of rock joints applicable to CNL and CNS conditions
Yong-Ki Lee, Jung-Wook Park & Jae-Joon Song

Department of Energy Resources Engineering, Seoul National University, Korea
Keywords: lab testing, rock failure, rock joints, rock properties, rock stress
Mechanical properties of rock mass and its stability are highly The peak friction coefficient (µp ) was formulated asequation
affected by the mechanical and geometrical characteristics of (1), and the peak shear displacement (up ) was estimated as
discontinuities. One of the most important characteristics of equation (2). The post-peak behavior of the friction coefficient
discontinuities is shear strength and shear behavior observed was modeled using the peak friction coefficient and peak shear
in direct shear test. The direct shear test is divided into constant displacement. The overall modeling process of the behavior
normal load (CNL) test and constant normal stiffness (CNS) of friction coefficient is as follows. The peak friction coef-
test according to loading condition. The CNL condition is ficient is calculated by equation (1), and then the peak shear
suitable for rock slope while the CNS condition is suitable displacement is obtained by equation (2). The post-peak curve
for underground rock mass where the shear behavior of dis- intersecting the peak friction coefficient and the peak shear
continuities depends on the stiffness of rock mass. Reliable displacement at the initial normal stress is increased contin-
prediction of shear behavior of rock joints for both conditions uously in equation (1) and (2). The normal displacement in
has been an important issue in rock engineering field. the pre-peak range was assumed to be zero based on the test
In this study, an empirical model of shear behavior of rock results. The post-peak behavior was modeled as equation (3).
joints was proposed from direct shear tests using a servo-
controlled shear testing machine. Cement mortar replicas of
rock joints were tested with various normal stiffnesses (Kn ),
initial normal stresses (σ0 ), joint roughness coefficients (JRC)
and joint wall compressive strengths (JCS). Joint roughness
was measured using a 3-D laser roughness profiling machine.
The shear test results of 120 specimens were analyzed to inves-
tigate the effects of loading condition and material properties The suggested models can be applied to predicting the
on friction coefficient and normal displacement of the joints. shear behavior of rock joints including pre-peak and post-
Empirical models of the friction coefficient and normal dis- peak shear stress levels regardless of the loading condition.
placement were suggested based on the test results. In the To verify the suggested model, additional shear tests of rock
empirical models, dimensionless terms were adopted to avoid joints were carried out and the performance of the models was
the scale effect and therefore to enhance the applicability of compared with those of other model. Based on the additional
the suggested models. In a pre-peak range, the friction coef- test results, the suggested model of peak friction coefficient
ficient showed a linear relationship with shear displacement. showed lower error of prediction than Barton’s empirical shear
Figure 1. The behavior of friction coefficient under various initial normal stresses.
325

Table 1. Comparison of error rate for peak friction coefficient models.
Proposed model Barton’s empirical Modified Barton’s model

Sample in this study model by jang et al.
R-12.72 4.31 4.17 21.47

R-13.71 11.03 11.04 29.08
R-13.86 10.39 10.13 21.22
Hwangdeung granite 2.77 8.07 45.19
Onyang gneiss 9.77 59.65 35.97
Average 7.65 18.61 30.59
Figure 2. Verification of suggested shear model at applied additional test results.
Figure 3. Verification of suggested shear model at applied test result of previous studies.
strength model (Barton, 1973) and modified Barton’s model size were selected from the previous studies for the compar-
suggested by Jang (Jang et al., 2010). Table 1 showed the ison to avoid the scale effect. The previous studies were only
comparison result. It was also verified by comparison with for CNL condition. In overall stress levels, the prediction by
previous studies of Bandis (Bandis, 1980), Homand (Homand suggested models showed good agreement with the shear test
et al., 2001), and Son (Son, 2005). Figure 2 shows comparison results. The suggested models can be effectively applied to
with additional tests while Figure 3 shows comparison with predicting peak friction coefficient and overall stress levels
previous studies. Only the test results of specimens similar in for both CNL and CNS conditions.
326

Evaluation of shear velocity dependency of rock fractures by using repeated shear tests
B. Li & Y. Jiang
Department of Civil Engineering, Nagasaki University, Nagasaki, Japan
G. Wang
College of Civil Engineering and Architecture, Shandong University of Science & Technology, Qingdao, China
Keywords: rock properties, lab testing, rock joints
Understanding the deformational behaviors of rock masses friction of a fracture surface has a positive relation with the
subjected to seismic loads is an important issue to ensure shear velocity. The dilation of fracture decreases with sheared
the stability and safety of some important structures such as loops since the asperities on fracture surface are crushed more
dam and nuclear power plant, which requires adequate evalu- and more with the increase of shear loops.
ation of the mechanical properties of discontinuities existing No obvious relation between peak shear stress and shear
in rock masses. The mechanical properties of rock fractures velocity can be observed. Since rock is one kind of brittle
are usually estimated through laboratory test on rock samples material, the failure of major asperities at peak shear stress
containing single rock fracture. Direct shear test in labora- may depend more on the material properties, rather than the
tory is considered as an effective method to estimate the shear velocity that the failure is induced. Therefore, the behavior of
behavior of rock fracture, which is commonly conducted at fracture at residual stage will be more representative to the
relatively low shear velocity (e.g. 0.5 mm/min). When earth- characteristics of fracture subjected to seismic loads. At here,
quake happens, the shear velocity on rock fractures can vary in we take the shear stress and normal displacement at shear
a large range depending on the seismic strength. In this study, displacement of 10 mm as residual shear stress and residual
repeated shear tests on artificial rock fracture samples were normal displacement, respectively, to estimate the mechanical
conducted with shear velocities ranging from 0.5 mm/min properties at residual stage.
to 50 mm/min to estimate the effects of surface roughness, The residual shear stresses at different shear velocities
shear velocity and shear history on the shear behavior of rock exhibit clear linear relations with the normal stress. The larger
fractures. the shear velocity, the steeper the straight lines connecting
A servo-controlled direct shear apparatus for rock fractures residual shear stresses become, demonstrating that in the resid-
using virtual instrument (VI) software was used in this study.A ual stage, the fracture has larger friction at higher shear veloc-
rock fracture, labeled as J10, was created by splitting a granite ity. At the first shear loop with shear velocity of 0.5 mm/min,
block, and was used as prototype to produce artificial replicas the major asperities have not been totally crushed, thus cohe-
of rock fractures. Its surface is very rough with a few major sion can be observed by extending the straight line to the
asperities on various locations, and the JRC range is 18∼20. y-axis. In the following loops, the cohesion become very small
Totally 15 tests were conducted under constant normal load approaching 0 due to the damage of the asperities. The resid-
boundary conditions with normal stresses of 1, 2 & 4 MPa. ual normal displacements decrease proportionally with the
Repeated shearing was conducted in these tests with increas- increase of normal stress and shear velocity. For the cases with
ing shear rates for each loop of 0.5, 5, 10, 20 & 50 mm/min. the initial shear velocities (velocity at the first loop) of 0.5 &
The samples were sheared with displacement up to 20 mm, 5 mm/min, the friction angles increase significantly with the
then the shear direction was reversed and the samples were shear velocity before 20 mm/min, after that, the friction angle
sheared back to 0 mm to complete a loop. trends to be constant. For the cases with initial shear veloc-
In the repeated shear test results, the peak shear stress can ities larger than 5 mm/min, the friction angles increase with
only be observed at the first shear loop. After the peak, the the increase of shear velocity. These results indicate that the
shear stress undergoes a slow decrease until around 5 mm friction angle of a fracture without shear history depends on
of shear displacement, then, the shear stress keeps almost the behavior of major asperities, which don’t have clear rela-
constant value in the residual stage. The shear stress in the tion with shear velocity. The friction angle at more shear loops
backward shearing is much smaller than that of the forward always has larger value than that with fewer loops, in the con-
shearing for all the loops, since the backward shearing is dition of the same shear velocity. During a shear, the asperities
merely a restoration process on the shearing tracks of the for- are being crushed, which generate plenty of gouge materials.
ward shearing. The shear stresses at residual stages (e.g. at The gouge will fill the void spaces between the two walls of
10 mm of shear displacement) increase with the shear veloc- the fracture, which increases the contact areas between the two
ity for both forward and backward shearing, indicating that the walls, thus gives rise to the friction angle.
327

An elastic-plastic-brittle constitutive model of rocks and its numerical validation
Q. Li
Department of Civil Engineering, The University of Hong Kong, Hongkong
The Research Center for Material Failure Modeling, Dalian University, Dalian
D.G. Wang
The Research Center for Material Failure Modeling, Dalian University, Dalian
J.H. Xiong & L. Song

Department of Geotechnical Engineering, Tongji University, Shanghai
Keywords: rock stress; numerical modeling; rock mass; rock failure
1 INTRODUCTION by the method of phenomenology. In addition, in the deep

engineering case, the rock mass is always in a high stress state
The issue of the deep rock mass mechanics is one of the most which force the rock mass to be more plastic. Considering the
significant subjects which matters a lot to the safety of State features, under the frame of the classic elastoplastic mechan-
property and people’s lives, it is also one of the core problems ics and damage mechanics, a particular constitutive model
in the field of both rock mechanics and underground structure including a plasticity strengthening flow stage for the deep
engineering. rock mass is proposed in the paper.
With the increment of the depth of engineering, the sur- In this model, there are two essential criteria, the yield cri-
rounding rock mass will yield and then fracture under the high teria based on the stress space is adopted for the judgment
ground stresses, which would form a special area called plastic of plasticity, and the failure criteria based on strain space is
zone and damage zone. These zones will cause serious threats adopted for the judgments of failure of elements. The reason
to the further excavation of caverns and the construction of is that when the material is softening, the relationship between
structure. stress and strain is not one to one corresponding, which cause a
Many researchers applied the statistical damage mechanism problem in distinguishing the stress state is whether softening
into the study of rocks. Cao et al. (1998) have introduced the or unloading based on stress space.
parameter which can describe the distribution of rock strength Also, the failure criteria of the paper differentiate the tensile
into the constitutive model, and established the equation of failure and shear failure of elements. It is obvious that the
damage evolution in order to discover the damage mechanism tension failure is a kind of brittle failure which has no residual
of rocks. Xu et al. (2002) have discussed the failure, damage strength, while the shear failure is a kind of ductile failure
and elastoplastic deformation of rocks based on the statis- which has been through the plasticity flow stage, so it has a
tic theory and damage mechanics, and also established the residual strength that can be treated as a decrease of fictional
elastoplastic damage constitutive model using the effective angle defined in the paper.
stress. Based on the existing program of elastoplastic analysis, the
The meso statistical damage method is able to simulate the efficiency of calculation and the process of code are modified,
relative complex marco characters using the simple constitu- and the subroutine of elastoplastic damage is added, also the
tive model of each element, Tang etc. (1997, 2003, 2009) have modulizaiton of preprocessing and postprocessing is carried
done much researches on this area. on. The following calculation in this paper is under this new
In this paper, based on the self-developed software and version of program.
the new constitutive model which takes both meso statis-
tic damage and plastic damage into account, two numerical
experiments are carried out.
3 COMPARISONS AND VALIDATION
Two numerical tests have been carried out to validate the new
2 CONSTITUTIVE MODEL developed constitutive model in the paper. One is the extension
of cracks in the experiment of pre-existing crack board under
In general, rocks are treated as a quasi-brittle material which uniaxial tension. It can be shown that the development path
means there is little plastic effect especially at a macro scale. of existing crack is first along the initial direction, and then
But as the different definition of plasticity in metals, the vertical to the horizontal tension stress (Shown in Figs. 1–2).
irreversible deformation is equal to the plastic deformation The other one is the uniaxial compression rock samples. It
328

Figure 1. Tension failure of prenotched plates.
Figure 2. Stress distribution of prenotched plates.
Figure 3. Failures of rock samples.
Figure 4. Stress distribution of rock samples.
is not difficult to find both the parallel and conjugate shear proposed, in which the multiple yielding criterion based on
band in the figures of failure (Shown in Fig. 3). With the stress space and ductile failure criterion based on strain space
increase of compression stress, the relatively weaker elements are adopted. What is more, the program for numerical model
first get yielded, and then fail which lead to release of their and calculations is self-developed which is open and can be
own stresses, resulting in a stresses redistribution of local area extended, therefore makes it easy to develop the corresponding
(Shown in Fig. 4). constitutive model with different material in further research.
Through the comparison with the relevant physical exper- According to the simulation of two tests including the ten-
iment and numerical method, it is shown that the present sion test of pre-existing crack board and the compression test
model is rational and accurate. of rock samples, it is shown that the present model is ratio-
nal and accurate, which can simulate the extension mode of
cracks and the development of localized shear band.
4 CONCLUSIONS
Taking the features of deep engineering into considera-

tion, a particular elastic-plastic-brittle constitutive model is
329

Drilling geological strength index in altered gneiss
D.M. Lin
Key lab of engineering geomechanics, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
Graduate University, The Chinese Academy of Sciences, Beijing, China
Y.J. Shang
W. Zhang
Graduate University, The Chinese Academy of Sciences, Beijing, China
ABSTRACT: GSI is the methodology that developed by E. Hoek and other geologists throughout the world. It is particularly
suitable for weathered and heterogeneous rock. However, there are not quantitative parameters for GSI, which leads to sub-
jectivity when getting value. Based on the study of rock mass in the region where is abundant of gneiss during geotechnical
investigation, the author propose a new parameter RCL to help the quantitative GSI; combine with geological effect of alteration
which is very normal in metamorphic rock, the author try to utilize the photos of the intact core in the field to substitute the
sketch of rock mass structure in the chart of GSI’s value, Ultimately, set up a chart of GSI’s value on the basis of gneiss. In
terms of the contrast of modulus of elasticity between calculation by the author and the value given by the unit of investigation,
the author finds that the result fit well, which means it is scientific of this chart.
Keywords: rock mass; rock properties; lab testing
1 RCL AND JC AS PARAMETERS TO HELP GSI STUDY 2 CONCLUSION
Drilling core is very important in the study of parameters In Fig. 1 the GSI chart given by Hoek is not highly applicable
for GSI; though RQD is one of most popular description in drilling project, especially the sketch of rock mass struc-
of drilling core, for drilling core can expose faults, joints ture, with the core box the author substitute the sketch with
in underground engineering, and Rock Core Length can typical photographs, in that way, it can look much more intu-
reflect the joint spacing, joint direction, joint sets and so itive. In the description of the drilling core, we combine with
on; from Fig. 2, you can find the obvious difference of the RQD, drilling core rate to have a good understanding of GSI
two underground geological condition in weathering, RQD, in drilling core.
joint roughness and filling, so the author refer the scholars Such as Fig. 3, you can see the differences obviously
around the world on rock mass structure, and find the rela- compare with Fig. 1.
tionship between core drilling and rock mass structure, at last The intact rock block length RCL that suggested by the
try to introduce RCL as one important parameter to help the author is an important parameter in rock mass structure’s study
quantitative value of GSI. at present which can reflect abundantly underground informa-
Table 1 is one the description and process of getting joint tion. Quantificative research of GSI combine with RCL can
condition value, we use this parameters to divide surface con- avoid effect of subjective factors to some extent. Many schol-
dition into several grand to help the quantitative value of GSI. ars research the GSI in underground engineering care about
At last we set up a chart of GSI’s value on the basis of gneiss’ the volume of rock, but is the same that consider the rock core
category such Fig. 3. length in drilling engineering. So this parameters is useful in
the research of GSI.
The structure surface is the most likely weathered surface,
so compile with joint roughness, filling can be practical in
Figure 2. The core core of different structure for gneiss (a: block structure b: sheared and fractured structure).
330

the quantitative research of GSI, the author introduce those
and develop a new chart of GSI for drilling engineering, when
compare the result between author and exploration institute,
we are glad to find the consequence fit the practical very
well, so we believe this chart can be widely used in Drilling
engineering when you need GSI.
Alteration is an inevitable process to gneiss. It sometimes
needs to consider influence of alteration, especial in hydrother-
mal and tectonic area, if the alteration is serious, the grades
of rock mass structure need to degrade1 grade (as Fig. 3),
Figure 3. New drilling GSI chart of gneiss which considers alteration.
331
Investigation on mechanical parameters of water-rock interaction
Yingfa Lu & Tao Lu

Key Laboratory of Geological Hazards on Three gorges Reservoir Area, Ministry of Education,
China Three Gorges University, Yichang, Hubei, China
Subject: Rock properties
Keywords: rock mechanics, water-rock interaction, true/pseudo tri-axial tests, limit yield stress, anisotropic Biot coefficient,
action mechanism
ABSTRACT: The processes are very complex for the cou-

pling of water-rock interaction, and the sensitivity of water
for porous media material cracks should be focused for
the research of fundamental mechanical properties of rocks.
In the present research, the theories and measure tech-
nologies of conventional true and pseudo tri-axial tests are
also analyzed; the mechanical parameters are obtained by
true and pseudo tri-axial tests with/without water pressures.
The idea that the tri-axial test is only suitable for the
isotropic porous media materials is proposed, the sample of
50 mm × 50 mm × 100 mm is proposed to conduct the tri-
axial test to understand the anisotropic mechanical behaviors
Figure 1. The scheme of Biot’s coefficient for different effective
of porous media materials. Based on the researches of theory confining pressure.
and test, some results are obtained for the different sample
shape and testing method. The five mechanical parameters and its theory and calculation are suggested in the following
(E11 , E22 , µ12 , µ23 , µ31 ) are obtained for the true tri-axial test form.
under both saturated and dry condition, the nine mechani-
cal parameters (E11 , E22 , µ12 , µ23 , µ31 , β11e , βe , βe , K ) can
22 33 s
be gotten for the true tri-axial test with water pressure,
the two mechanical parameters (E11 , µ12 ) are obtain for the
pseudo tri-axial test under saturated (without water pressure)
or dry porous media condition with a cylindrical sample.
The five mechanical parameters (E11 , µ12 , β11 e , βe , K ) can
22 s K = C1111 + 2C1122 + 2C1133 + 2C2233 + C2222 + C3333 βii1 is
be obtained for the pseudo tri-axial test with water pressure, Biot’s coefficient.
and the three mechanical parameters (E11 , µ12 , µ31 ) can be A testing method of limit yield stress determination is sug-
obtained for the pseudo tri-axial test with hexahedral sample gested: the limit yield stress is equal to the deviatoric stress
without water pressure, and the seven mechanical parame- corresponding to the maximum water pressure between devia-
e , βe , βe , K ) can be obtained for the
ters (E11 , µ12 , µ31 , β11 toric stress and water pressure curve for the non-drainage test.
22 33 s
pseudo tri-axial test using hexahedral sample with water pres- When the deviatoric stress is applied, the porous media mate-
sure. Finally, the pseudo tri-axial test with cylindrical sample rial is compressive firstly, the water pressure increases. When
is only fit for the isotropic porous media, and the test with the deviatoric stress is continue to add, the water pressure is
hexahedral sample is fit for the anisotropic porous media added and up to a maximum value, above the maximum water
mechanical behaviors. pressure, the fissures in the porous media materials are open,
The anisotropic Biot’s coefficients and the damaged propagate, collection and nucleation. When the porous media
mechanical parameters under different stress states are material is expensive, the water pressure decreases. When the
obtained by considering the influences of unloading water failure of sample occurs, the water pressure is nearly zero.
pressure on porous media materials. The testing results show The some mechanism are testified by the four conventional
that variable trends of anisotropic Biot’s coefficient in the tri-axial tests of non-drainage, the three proportional load-
different directions are not the same; The different behaviors ing tests of non-drainage, the three water pressure unloading
of axial and radial Biot’s coefficients are analyzed for the water tests of non-drainage, the two deviatoric stress cyclic tests of
pressure injection and the non-drainage test, the axial Biot’s non-drainage (see fig. 2), the results are focused on the Fig. 3.
coefficient change from initial value to 1, and the radial Biot’s The testing results also show the Skempton coefficient dis-
from initial value to initial porosity (see fig. 1). Based on parity is significant; and the mechanism of water pressure
the Biot’s coefficient theory and testing research, the testing action is different under various injections and drainages (see
determination of anisotropic Biot’s coefficient is proposed, fig. 4).
332

tant Project Funds (No. 50879044), the National Support Plan
(No. 2008BAB29B03), the China National Defenses Bureau
(No. 838) and the Mechanical Laboratory of Lille, France.
REFERENCES
Hongkui GE, Dehua HAN, Yong CHEN. Experimental investiga-

tion on poroelasticity of sandstones[J]. Chinese Journal of Rock
Mechanics and Engineering, 2001, 20(3): 332–337.
Y.H. HATZOR, A. ZUR, & Y. MIMRAN. Microstructure effects on
microcracking and brittle failure of dolomites[J]. Tectonophysics,
1997, 281(3/4): 141–161.
Sijing HUANG, Yuming SHAN, Weiguo LIU, et al. Relationship
Figure 2. The relationship of cyclic loading—unloading testing for between mechanical parameters of rock and formation conditions
10 MPa confining pressure. of reservoir sandstone[J]. Chinese Journal of Rock Mechanics and
Engineering, 1999, 18(4): 454–459.
M.E. KASSNER, S. NEMAT-NASSER, & Z.G. SUO, et al. New
directions in mechanics[J]. Mechanics of Materials, 2005, 37(2/3):
231–259.
Tianbin LI, Lansheng WANG. An experimental study on deforma-
tion and failure features of a basalt under unloading condition[J].
Chinese Journal of Rock Mechanics and Engineering, 1993, 12(4):
321–327.
Shuncai LI, Xiexing MIAO, Zhanqing CHEN, et al. Experimental
study on seepage properties of non-Darcy flow in confined broken
rocks[J]. Engineering Mechanics, 2008, 25(4): 85–91.
Y.F. LU. Modelisation de l’endommagement anisotrope des roches
saturees[Ph. D. Thesis][D]. Lille, France: University of Sciences
and Technologies of Lille, 2002.
Figure 3. The scheme of effective yield stress and effective con- Y.F. LU, J.F. SHAO. Modelling of anisotropic damage in brittle
fining pressure (NDT: Non-drainage test, NDWCT: Non-drainage rocks under compression dominated stress[J]. International Jour-
cyclic test, NDPT: Non-drainage Proportional Loading Tests). nal for Numerical andAnalytical Methods in Geomechanics, 2002,
26(10): 945–961.
Yingfa LU, Defu LIU, Yanchun WU, et al. Numerical simulation
of orthotropic anisotropic damage considering rock-water inter-
action[J]. Chinese Journal of Rock Mechanics and Engineering,
2007, 26(2): 23–330.
Yingfa LU, Junjie ZHENG, Huayou CHAI. Characteristics of
Skempton coefficient of representative rock and soil[J]. Chinese
Journal of Rock Mechanics and Engineering, 2005, 24(11): 1
847–1 851.
Robina, H.C., K.T. Wong, & C.P. Wang. Microcracking and grain
size effect in Yuenlong marbles[J]. International Journal of Rock
Mechanics and Mining Sciences and Geomechanics Abstracts,
1996, 33(5): 479–485.
J.F. Shao,Y.F. Lu & L. Dariusz, Damage modelling of saturated rocks
in drained and undrained condition[J]. Journal of Engineering
Figure 4. The scheme of Skempton coefficient under different Mechanics, 2004, 130(6): 733–740.
effective confining pressure. X.Y. Wu, P. Baud, T.F. Wong. Micromechanics of compressive failure
and spatial evolution of anisotropic damage in Darley Dale sand-
It is crucial for damage mechanics and civil engineering to stone[J]. International Journal of Rock Mechanics and Mining
study water-rock interaction. Sciences, 2000, 37(1): 143–160 .
Mingqing YOU, Anzeng HUA. Test of triaxial confining unloading
of rock[J]. Chinese Journal of Rock Mechanics and Engineering,
ACKNOWLEDGEMENTS 1998, 17(1): 24–29.
Jianping ZUO, Hongwei ZHOU, Heping XIE. Fracture characteris-
We acknowledge the supports from the China National tics of sandstone under different thermal effects[J]. Engineering
Nature Science Funds (No.50779030), the National Impor- Mechanics, 2008, 25(5): 124–129.
333

Surface roughness measurements using three-dimensional laser imaging
J. Mah & C. Samson

Department of Earth Sciences, Carleton University, Ottawa, Ontario, Canada
S.D. McKinnon
The Robert M. Buchan Department of Mining, Queen’s University, Kingston, Ontario, Canada
Keywords: rock mass classification, rock joints, rock properties, site characterization, field measurements
1 INTRODUCTION recorded. The collection of 3D points is termed point cloud

data.
One of the main parameters in the Rock Mass Rating (RMR) The laser camera was setup 3 m from the rock face and 7
system is the condition of discontinuities. Surface roughness images were acquired at various angles to ensure complete
is a sub-parameter and commonly quantified by the Joint coverage of the test site. The images cover an area 5.3 m wide
Roughness Coefficient (JRC). JRC is estimated by manually by 2.0 m high. The upper region of the rock face was supported
measuring, with a ruler, the maximum asperity amplitude nor- by wire mesh, which was also captured in the images. Figure 1
mal to a straight edge reference length, or with a mechanical displays a composite view of the 7 images and the sites where
profilometer. the surface roughness were measured manually.
Recently, three-dimensional (3D) laser imaging has been
applied to estimate joint orientation. Measurements are made
quickly without the need to be in direct contact with the rock
face. In this paper, the applicability of 3D laser imaging to 2.3 Image processing
estimate surface roughness of joints is explored.A new method A MatLAB program was developed to estimate the surface
to estimate surface roughness from 3D image data is presented roughness from point cloud data. The user selects a point
and tested in an underground case study. where the surface roughness is to be measured and the pro-
gram selects all the surrounding points within a specified
2 UNDERGROUND FIELD TRIAL radius. A principal component analysis is conducted on this
subset of points and the best fitting plane through them is
An underground field trial was conducted at Vale’s T1 nickel determined. Using the normal vector of the plane and the ori-
mine located in Thompson, Manitoba, Canada. The site was entation of the laser camera, the joint orientation (strike and
1524 m (5000 ft) underground and the rock mass contained dip) is determined.
several curved foliating features as well as irregular surfaces Surface roughness is measured by first selecting a rectan-
from blast fractures. gular subset of points. The 2D surface profile is extracted
by binning the points and calculating their average normal
distance to the plane. The average bin width in this study was
2.1 Manual measurements 0.635 cm (0.25 ). From this, the maximum asperity amplitude
To perform a controlled validation, 18 strike and dip measure- can be determined to estimate the JRC.
ments were made using a compass. At these same locations, 32
surface roughness measurements were made in the directions
of the strike and dip. The surface roughness measurements 3 RESULTS AND DISCUSSION
were made at various base lengths 0.08–0.61 m (3–24 inches)
and the maximum asperity amplitude was recorded. Making Validation of the laser measurements is required to assess the
compass measurements was problematic due to the close performance of the proposed methodology. For most mea-
proximity of the highly magnetic ore body. surement sites, only minor variations are observed between
the laser acquired and manual maximum asperity amplitudes.
The maximum asperity amplitude measurements derived from
2.2 Image acquisition
laser data were overall an average of 0.3 mm greater than those
In an underground mine, the available scanning distance is based on manual measurements. Inaccuracies can be intro-
limited and the asperity amplitude for a rock face can be quite duced in the manual measurements if the operator does not
subtle. Therefore a short-range triangulation-based laser cam- identify correctly the location of the maximum amplitude.
era capable of millimetre accuracy was used to acquire data. This can occur due to the limited lighting and dirty rock faces
With this camera, for each point illuminated by the laser beam, in underground conditions. For several manual measurements
the spatial coordinates (X, Y, Z) and reflection intensity are with baselengths greater than 0.30 m (12 ), the maximum
334

Figure 1. The locations of the sites where surface roughness was measured manually are displayed on the 3D laser data.
Figure 2. The 2D surface profile averaged from binned point cloud data. The maximum asperity amplitude is calculated based on
perpendicular distance to a straight line through the two highest local maxima. The 2D surface profile is from site #7, along the strike.
In a preliminary analysis, 2D surface profiles at site #7 were

generated at 5 degree increments. The 2D surface profiles
revealed a change in surface roughness where the amplitudes
are relatively low for azimuth angles 0 to 85 degrees and higher
from 90 to 175 degrees. The corresonding JRC values ranged
from 2.5 to 12. The JRC values indicate that, at site #7, the
rock face was locally smoother and the shear strength lower
in the strike direction.
4 CONCLUSIONS
Despite the challenging underground environment, the 3D

laser camera was capable of acquiring accurate data at the
rock face in the millimeter range at a distance of 3 m. In
Figure 3. The JRC is measured from the 2D surface profiles at 5
degree increments for site #7 and reveals the roughness anisotropy a controlled validation study, the laser maximum asperity
at this location. amplitudes and the corresponding JRC values were compared
directly to manual measurements, and the difference in ampli-
amplitude was over 2 mm larger than the manual measure- tude was on average only 0.3 mm. JRC values were measured
ment. It is possible that, for these longer baselengths, that in different azimuth directions and significant anisotropy
the ruler flexed towards the rock face while the measurement was revealed. The results demonstrate the potential of 3D
was taken. This effect would tend to decrease the measured laser imaging of contributing additional valuable information
maximum asperity amplitude of long-baselength manual mea- towards rock mass characterization that would be both diffi-
surements. The application of 3D imaging opens the door to cult and time consuming to obtain using conventional, manual
new possibilities. approaches.
335

Understanding failure mode in uniaxial and triaxial compression for a hard brittle rock
V.B. Maji
Department of Civil Engineering, IIT Madras, Chennai, India
1 INTRODUCTION
Numerous numbers of studies are done to understand the

failure mechanism and mode of failure under various stress
condition considering failure theories combined with vari-
ous laboratory experiments. Various different types of failure
modes are possible in rocks. Moreover the failure modes
substantially vary with the effect of confining pressure. Rock
under natural conditions experiences different stresses and
fractures occur in a rock at a certain point when it crosses the
threshold stress value. The rock fails with fractures developed
from the coalescence of several micro cracks giving rise to
different failure modes under various stress conditions. They
Figure 1. Test under progress with a 200 tons capacity compression
provide useful information for safe and economic design of testing machine and a 70 MPa capacity Hoek triaxial cell connected
various geotechnical structures involving rock. The prediction to a hydraulic pump.
of failure through better design will significantly reduce the
costs involved in construction and increase in the safety. The
present study is an attempt to review and understand the failure
mechanism of hard brittle rocks like granite. This work aimed
to understand the different failure modes of brittle rock under
different stress conditions by conducting laboratory studies on
rock samples (granulite) collected from Karimnagar district,
Indian state of Andhra Pradesh. The various failure modes of
the brittle rock samples under different uniaxial and triaxial
compression are studied.
Figure 2. Different failure modes of granulite observed during after
uniaxial compression testing.
2 BACK GROUND
occurs along a clearly defined plane of fracture. At very high
Many researchers have done extensive studies to understand confining pressure rock behaves like a ductile material.
the failure behavior of rock (Mogi, 1971, 1974, Peng, and
Johnson, 1970, Klien, Baud, Reuschle and Wong, 2000,
Prudencio Vansint, 2009, Basu, Celestino and Bortolucci, 3 FAILURE MODES IN COMPRESSION
2009). The stress-strain curve too gives a good amount of
information about the failure process. The failure modes are Rock specimen of diameter 54 mm and length 108 mm are
mostly found to be related to the strength, anisotropy, brit- tested under uniaxial and triaxial compression. A 200 tons
tleness and grain size of the rock. However, axial fracturing capacity compression testing machine is used for conducting
generally precedes failure surface and initiates at 50–95 per- the tests (figure 1). For triaxial testing, a 70 MPa capacity
cent of the compressive strength (Wawersik and Fairhurst, Hoek triaxial cell is used with a hand held hydraulic pumping
1969). The values of the UCS obtained in extension are the arrangements (figure 1).
highest, whilst the values obtained in shear are the lowest. Figure 2 shows the different failure modes of granulite
The most common failure types were failure by axial split- observed after uniaxial compression testing. Observations
ting and along single shear plane. The failure mode of a brittle were made on the crack initiation and propagation based on
rock changes on application of confining pressure because the analysis of failure modes and fracture patterns of the rock
usually under unconfined compression a rock tends to deform samples. The effect of confining pressure on the strength and
elastically until failure. With moderate amount of confin- failure mode of a rock samples are studied using triaxial com-
ing pressure, longitudinal fracturing is suppressed and failure pression. Failure strength of the rock samples found varying
336

(2004) proposed a criterion assuming critical distortion energy
increases with the effective mean normal stress, considering
a monotonically increasing function. Present study also dis-
cusses the applicability of various criteria for brittle rock
failure and understanding the fracture mechanism.
5 SUMMARY
An attempt has been made to review and understand the vari-

Figure 3. Different failure modes of granulite observed during
ous failure modes of brittle rock samples under different stress
after tri-axial compression testing.
conditions. Uniaxial and triaxial compression tests are con-
based on the failure modes and fracture patterns of the rock ducted to study the failure modes of rock and the relationship
samples. In case of uniaxial compression, it was observed that between the strength and the failure mode is studied. It is
cracks sometimes start at the circular surface either from the understood that the failure mode is very significant to decide
top or bottom or both, and sometimes forming V-shaped cracks upon true strength of rocks. Observations were made on the
at the ends and propagate along the vertical axis and along the crack initiation and propagation based on the analysis of fail-
length of the cylindrical sample (figure 2). Different modes ure modes and fracture patterns of the rock samples. Failure
are observed from the failure of the different samples tested strength of the rock samples vary depending upon the failure
under different confining pressures (figure 3). modes and fracture patterns of the rock samples.
4 FAILURE CRITERIA AND FAILURE MODE REFERENCES
There is no single failure criterion which can properly jus- Basu, A., Celestino, T.B., and Bortolucci, A.A. (2009), Evaluation
tify the brittle rock fracture process and failure. Often Mohr’s of rock mechanical behaviours under uniaxial compression for
different weathering grades, Rock. Mech. Rock. Engg, Vol. 42,
theory is used in predicting the failure of brittle rock and can No. 1, 73–89.
be opted for 2D stress but is unable to explain the tensile Klien, E., Baud, P., Reuschle, T. and Wong, T.F. (2001), Mechanical
brittle fracture in rock. Griffith theory better suits for brittle behavior and failure mode of Bensteim Sandstone under triaxial
rock but unable to follow the in-homogeneity in tock material. compression, Phys. Chem Earth (A), Vol. 26, 21–25.
It explains the two dimensional relationship between shear Mogi, K. (1971) Fracture and flow of rocks under high triaxial
and normal stress at point of failure and the mechanism of compression. J. Geophys. Res. 76, 1255–1269.
failure is based on the formation, propagation and joining of Mogi, K. (1974) On the pressure dependence of strength of rocks
microscopic cracks leading to macroscopic failure. Griffith and the Coulomb fracture criterion. Tectonophysics, 21, 273–285.
cracks edges concentrate huge stresses and eventually failure Peng, S. and Johnson, A.M. (1970) Crack growth and faulting in
occurs. The von-Mises criterion may be modified and used cylindrical specimens of chelmsfold granite, Int. J. Rock Mech.
Min. Sci. Vol. 9, 37–86.
as satisfactory criterion for fracture of brittle material like Szwedzicki, T. (2006), A Hypothesis on Modes of Failure of Rock
rock. Because of the variable distortion strain energy (τoct ) Samples Tested in Uniaxial Compression. Rock. Mech. Rock.
during high pressure dependent brittle fracture, applicability Engg. 40 (1), 97–104.
of the von-Mises criterion for brittle failure may be ques- Wawersik, R.R. and Fairhurst, C. (1969) A study of brittle rock frac-
tionable as the criterion assumes, yielding occurs when the ture in laboratory compression experiments, Int. J. Rock Mech.
distortional strain energy (τoct ) reaches a constant value. Mogi Min. Sci. Abstr. vol. 7, 227–229.
337

The mechanical properties of CO2 -saturated coal specimens
M. Masoudian-Saadabad, D.W. Airey, A. Gainey, T. Morris & J. Berger

School of Civil Engineering, University of Sydney, Australia
Keywords: coal, gas flow and gas bursts, lab testing, rock mass properties
1 INTRODUCTION
Geosequestration is the process of pumping CO2 into deep

geological formations such as coalbeds to reduce greenhouse
gas emissions into the atmosphere. Coalbeds are of interest
because they have naturally stored methane which can be dis-
placed by injecting CO2 and can help produce a clean and
valuable hydrocarbon that can partly offset the sequestration
expenses. Coal is an organic sedimentary rock that can be
considered as a dual-porosity medium, where the coal matrix
can be modeled as micro-porous blocks which are separated
by orthogonal systems of fractures or cleats. It is generally
considered that during CO2 injection gas flows through the
cleats and diffuses into the coal matrix where the adsorption
of CO2 leads to volumetric swelling. The process of swelling
is believed to be a result of a reduction in the glass transition
temperature of the coal polymer, an explanation that implies Figure 1. The peak strength of coal specimens: water saturated
a change in the molecular structure of coal. Hence changes in (•), dried (◦), CO2 saturated in 24 hrs (), and CO2 saturated in
the mechanical properties of solid coal may be expected during 72 hrs ().
CO2 saturation. Many studies have investigated the effect of
swelling on the hydraulic properties of coal and have provided
a range of analytical and numerical models to describe the per- specimens that contained visible cleats after coring were dis-
meability of coalbeds during gas production and/or injection. carded, but it was not possible to guarantee that all specimens
However, relatively few studies have been reported investi- were defect free.
gating the effect of CO2 on the mechanical properties of coal. Figure 1 illustrates that water-saturated samples show the
For example, it has been found that the uniaxial compres- highest strength, and this is similar to the oven-dried speci-
sive strength and elastic modulus of CO2 -saturated Victorian mens. In deep underground conditions, coalbeds are expected
brown coal samples are less than those of air-saturated spec- to be water saturated (fully or partially) and therefore one
imens by 13% and 26%, respectively, but this effect reduces can consider this as the strength of intact coal under in-situ
with increasing confining stress. The aim of this paper is to conditions.
investigate whether similar trends are observed for black coal It may be seen that the specimens saturated with CO2 for 24
from the Sydney Basin, Australia. hours show the lowest strength in Figure.At 1.0 MPa confining
A series of 45 triaxial tests have been performed in which stress, the reduction in the maximum strength is approxi-
coal specimens of 25 mm or 38 mm diameter have been tested mately 10.5 MPa, which is nearly 20% of the water-saturated
dry, or subject to permeation with either CO2 or water. The strength. However, when the coal was saturated for 72 hours
coal was bituminous with average porosity of 0.15 and dry with CO2 , the strength was higher than after 24 hours satura-
density of 1.29 t/m3 . tion, especially under low confining stress, and in some cases
the strength exceeded that of the water saturated specimens.
Similar trends with confining stress and saturating fluid were
observed for the coal stiffness.
2 RESULTS The results were used to develop the failure envelopes
for both CO2 -saturated and water saturated specimens using
The peak strengths from the 45 triaxial tests are shown in Hoek & Brown and Mohr-Coulomb failure criteria. It was
Figure 1. As expected, the peak strength of the coal samples found that the strength parameters of both criteria were
is higher under a larger confining stress. A few samples have reduced due to CO2 adsorption. The lower values of mi and ϕ
shown lower peak strengths under higher confining stress, but suggest that CO2 adsorption has eased the crack propagation
these can be explained by the natural variability of the coal in coal or resulted in a change in the surface properties of the
properties, and the possible presence of minor cleats. Any cracks.
338

Analysis of uniaxial tensile fracture of monomineral polycrystalline

rock based on intergranular cracking
K. Matsuki, Y. Karino, K. Sakaguchi & A. Kizaki

Keywords: lab testing, numerical modelling, physical modelling, rock failure, rock properties
In this study, creating realistic specimen models of monomin- In the initial stage (stage I), tensile failure occurs at random
eral polycrystalline rock on a computer, we performed a 3D at grain boundaries with a small angle of the normal direc-
finite element analysis of uniaxial tensile fracture under the tion relative to the loading axis and the connectivity between
assumption that fracturing occurs only at the grain bound- them is poor. Unloading always occurs before the peak at
aries. The grain boundaries were modeled by quadratic grain boundaries that do not compose the final failure plane.
joint elements and the constitutive law was derived by During the completion of the final failure plane after the peak
applying the associated flow rule in the plastic theory and (stage III), the conversion of failure mode from tension to shear
a tension-shear-softening curve to an extended Coulomb occurs at grain boundaries with a large angle of the normal
criterion. direction relative to the loading axis to form a single failure
Figure 1 shows an example of the change in the number of plane.
failure mode with the normalized axial strain, including both The normal stresses on the grain boundaries that have a
elastic and unloading modes (Model 2). Open circles with small angle of the normal direction relative to the loading
Arabic numerals in the normalized σz − εz curve indicate (1) axis are always tensile until the final stage (7 & 27 in Figure
onset of tensile fracture, (2) onset of unloading, (3) the peak 3). On the contrary, the failure mode of the grain boundaries
strength, (4) onset of shear failure, (5) the point where the that have a large angle of the normal direction relative to the
number of tensile mode is maximum, and (6) the point where loading axis (21 & 39) was changed from tension to shear.
a failure plane is completely formed, after which no fracturing For such grain boundaries, the normal stress was in compres-
newly occurs. Figure 2 shows the failed grain boundaries at sion in the elastic stage of loading, and changed to tension
the stages indicated by open squares with Roman numerals to reach the failure condition in tension. However, the shear
in the normalized σz − εz curve in Figure 1. The dotted area, stress did not decrease significantly after the failure, since
the cross-hatched area, the broken line and the thick solid line the axial displacement given at the top end of the specimen
indicate partial failure in tension, shear failure, unloading and
the grain boundaries that penetrate to the side surfaces of the
specimen, respectively.
Figure 1. The change in the number of failure mode with the nor-
malized axial strain (εz Eeff /T0 ) ( Model 2, T0 = tensile strength of Figure 2. Intergranular cracks at stages I, II, III and IV in Figure 1
grain boundaries, Eeff = effective Young’s modulus). (Model 2).
339

produces a large shear displacement, rather than an opening
displacement, to these grain boundaries and accordingly, the
shear stress does not significantly decrease even after the grain
boundary fails in tension, resulting in the conversion of the
failure mode from tension to shear with shear dilation. Thus,
grain boundaries with a large angle of the normal direction
relative to the loading axis provide a source for shear fail-
ure in uniaxial tension, while they provide a source of tensile
fracture in compression.
Figure 3. Stress paths in the fracture process at some nodes in the

intergranular cracks shown in Figure 2 (Model 2).
340

Determination of a correlation between intact rock unconfined compressive

strength and index parameters
H.J. Mendieta
Geotechnical Laboratory, Universidad Mayor de San Simón, Cochabamba, Bolivia
Keywords: lab testing, unconfined compressive strength, point load test, index parameters
1 UNCONFINED COMPRESSIVE STRENGTH TEST The failure is caused by cracks parallel to the loading axis.
Point load resistance is defined by the Point Load Index, Is ,
The concept of Uniaxial Compressive Strength refers to the given by Equation 1, where P is the load al failure and D the
maximum stress in the direction of a single axis that rock can distance between the tips.
withstand before breaking. The standard direct method for
determining compressive strength of intact rock is specified
by ISRM (1979) “Suggested Methods for Determining the
Uniaxial Compressive Strength and Deformability of Rock
Materials”. The expression that relates the Point Load Index, Is(50) , with
In this test, a right circular cylinder specimen is com- Unconfined Compressive Strength, σ ci , obtained according to
pressed between two plates of specific material and shape study carried by Bieniawski (1975) and Broch and Franklin
characteristics until failure is reached. (1972) is as follows (Equation 4) where K is the correlation
The Unconfined Compressive Strength, σci , is obtained coefficient.
dividing the force that caused failure, P, by the cross sec-
tion area of the specimen tested, A, (Equation 1). At least 5
specimens have to be tested in order to obtain representative
results.
3 INDEX PARAMETERS OF INTACT ROCK
An index parameter is a property that can be easily deter-

mined through simple tests application on rock samples. These
In conventional apparatus it is possible to control the mag- parameters provide a value that may not have a mechanical
nitude and rate of the pressure applied. For this case the rate meaning but may relate to characteristics of intact rock.
of pressure applied was based on the prescriptions of ISRM According to the definition, physical properties like unit
(1979) such that failure occurred within 5 to10 min of loading. weight, porosity, specific gravity and other properties like
wave velocity propagation and rebound sclerometer readings
are index parameters that can be related with the strength of
2 POINT LOAD TEST rock material.
Of the above mentioned index properties, physical prop-
The point load test was first proposed by Broch and Franklin erties and sclerometer readings have been selected to study
(1972). This test measures the force required to break a piece their relationship with Unconfined Compressive Strength.
of rock compressed between two spherically truncated still These parameters have been selected because the results
conical tips (Figure 2). obtained with these tests are affected by characteristics that
The standard direct method for determining point load also affect the unconfined compressive strength (i.e. Grade of
strength of intact rock is specified by ISRM (1985) “Suggested cementation, specific gravity of solids).
Method for Determining Point Load Strength”.
4 METODOLOGY
There has been a complete characterization of 9 different sam-

ples of sedimentary clastic rocks. The studies include visual
descriptive tests like the determination of the color (Munsell
scale) and visual classification of intact rock.
The physical properties determined were hardness, density,
porosity, specific gravity, slake durability, abrasion durability
and rebound index through the Schmidt hammer and Shore
Figure 2. Recommended 60◦ load tip (ISRM,1985). sclerometer.
341

Table 3. Results from the strength tests. Table 5. Results from the relevant index tests.
Is(50) Sample Shore Unit weight

σci
Sample MPa Parallel Perpendicular Ia DMR 01 61.74 24.04
DMR 02 50.16 22.61
DMR 01 62.02 2.3 5.7 2.48 DMR 03 35.49 23.06
DMR 02 50.84 2.0 4.7 2.35 DMR 04 44.91 23.31
DMR 03 14.9 0.8 1.2 1.50 DMR 05 57.64 24.39
DMR 04 38.25 2.4 4.2 1.75 DMR 06 37.29 23.18
DMR 05 74.41 3.0 5.2 1.73 DMR 07 36.45 22.18
DMR 06 45.27 1.0 2.7 2.70 DMR 08 45.00 26.22
DMR 07 18.77 0.6 1.2 2.00 DMR 09 40.00 22.51
DMR 08 50.39 1.4 2.7 1.93
DMR 09 38.85 1.5 3.1 2.07
4.3 Results
Table 4. Results of the correlation between σ ci and Is(50) .

The following tables present the results obtained from the tests
performed.
Sample Kaverage
DMR 01 15.51
5 RESULTS ANALYSIS
DMR 02 15.18
DMR 03 14.90 Statistical analysis was performed in two stages. In the first
DMR 04 11.59 stage bivariate analysis was performed in which facing each of
DMR 05 18.15 the independent variables with dependent (point load index)
DMR 06 24.47 separately. To For this purpose the results were analyzed in
DMR 07 20.86 pair series.
DMR 08 24.58 In most research problems where it is applied regression
DMR 09 21.91 analysis requires more than one independent variable in the
regression model. In the second stage a multivariate analy-
sis that investigated the influence two or more independent
variables was performed
The characterization included strength tests like Uncon-
Both the bivariate and multivariate analysis used least
fined Compressive Test, UCS, and Point Load Test, PLT. The
squares method for the determination of the trend lines.
Point Load Test included the determination of the anisotropy
From the multivariate analysis we obtained the following
index, Ia , relating the maximum Point load Index and the min-
equation where the unconfined compressive strength depends
imum Point load index obtained perpendicular and parallel to
upon the shore hardness value, density of intact rock and
stratification respectively.
anisotropy index as obtained from the point load test.
4.1 Test samples
The samples were obtained from the study area of a project
located in the capital of Bolivia, Sucre. Complementary, two In the same way, the relation between the correlation coeffi-
more samples were characterized, one from El Abra Tunnel cient K and index parameters was investigated. The result of
and the other from a highway, both located in Cochabamba, this study is the following.
Bolivia. The samples are conformed by clastic sedimentary
rocks (sandstones).
4.2 Test specimens

The specimens were obtained by core drilling. Core dimen-
sions were in the ranges suggested for the applied methods.
342

Experimental research on the electrical parameters of coal mass under load
L. Meng
State key lab of Coal Resources and Safe Mining, Beijing, China
M.J. Liu
Henan Polytechnic University, Jiaozuo, China
Y.D. Jiang & Y.X. Zhao

State key lab of Coal Resources and Safe Mining, Beijing, China
Y.G. Wang
Henan Polytechnic University, Jiaozuo, China
Subject: Rock material and rock mass property testing (Laboratory)
Keywords: dynamics, lab testing, rock failure, rock mass, rock properties
Coal or rock dynamic catastrophe is one kind of natural dis- Based on the previous research achievements and the test
aster induced by mining activities and phenomenon that coal principle of LCR meter, we design and make an experimental
or rock exerted by external stress results in dynamic effects system of electrical parameters of coal/rock mass under load
and disaster consequences in a very short time. And coal or as shown in the Figure 1, which can be able to conduct low
rock dynamic catastrophe has become worse than ever with frequency measuring for electrical parameters of coal mass in
deep coal seams being mined at higher production rate in a the process of coal mass fracturing. Electrical parameters of
more complex underground deep mining environment, so it is coal mainly include resistivity, dielectric constant and mag-
very important for disaster prevention and reducing the losses netic conductivity. However in view of less magnetic medium
caused by dynamic disaster to the minimum to accurately contained in coal bearing formation, magnetic conductivity
forecast coal or rock dynamic catastrophe. of which produces no changes and basically maintains at 1,
Geophysical methods forecasting coal or rock dynamic so electrical resistivity and dielectric constant are selected
catastrophe can overcome some traditional static prediction as main electrical parameters. In order to make experimental
methods’ shortcomings, such as not to exert great influence results more representative, these experimental samples that
over normal production, to authentically achieve real-time this paper uses include two structural types of coal samples,
dynamic monitoring and so on. Given the above mentioned such as the tectonic coal and the un-deformed coal respec-
analysis, it is necessary for cognizing coal or rock dynamic tively taken from three different mining communities, each
catastrophe to make an intensive study of geophysical meth- type of which includes 3 kinds of metamorphic grade coals
ods. For the reason that LFEW, Acoustic-Emission, EMR and and tectonic coal of which are made by using special mold
geophysical methods’ theoretical principles are all based on because this kind of coal is soft broken and can not be sam-
the electrical parameters change laws of coal or rock mass pled as block specimen. In addition, all the experimental coal
in the process of coal or rock fracturing, so in one sense, samples are made into standard specimen (5 × 10 cm) by
study on the electrical parameters change laws of coal or rock using special mold to relieve the additional friction caused by
mass under load is the same as study on the internal rela- ends of coal samples.
tion between the stress condition or energy accumulation and The experimental results show that in the process of instabil-
release in the coal or rock mass and the anomalous change of ity of coal, there is a clear anti-symmetry between the change
electrical parameters of coal or rock mass in the process of law of stress-strain curve and that of resistivity with its min-
dynamic disaster breeding, which means that research on the imum value corresponding to the cracking stress of coals,
electrical parameters of coal mass under load has great the- and the resistivity of the low metamorphic grade coal and
oretical significance and application prospect for monitoring high metamorphic grade gently decrease without appearance
and forecasting coal or rock dynamic catastrophe. Study on of abrupt change no matter what the structural type of coal
the electrical parameters measurement in the lab has attracted samples belong to, however the resistivity of the medium meta-
the attention of scholars, partly because of the development morphic grade coals abruptly decrease when external load
need of the discipline and partly because it can provide micro- reaches approximately 75% and 85% of compressive strength
cosmic mechanism explanation for change laws of electrical value for the tectonic coal and the un-deformed coal, respec-
parameters measured on spots to some extent. tively, the part of which is shown in Figure 2. In addition, it
343

Figure 1. Experimental system of electrical parameters of coal samples under load and experimental specimens.
Figure 2. Resistivity (a) and dielectric constant (b) change curve of coal samples with outburst prone taken from Hebi mining community
in the process of uniaxial compression.
is also observed that there is a maximum value correspond- the fracture revolution laws of coal under load and providing
ing to the cracking stress of coals for dielectric constant in the information of forecasting for the instability and fail-
all dielectric constant strain curves and the dielectric con- ure of coal and the coal/rock dynamic catastrophe. However,
stant strain curves of the tectonic coal are better fitting to the further investigation is necessary in order to verify the validity
stress strain curves, which is independent on the metamorphic of the above analysis to the electrical parameters of coal/rock
grade of coal. Therefore, study on the change laws of electri- contained gas under load.
cal parameters of coal under load is promising in revealing
344

Creep property of artificial methane-hydrate-bearing rock
K. Miyazaki, Y. Sakamoto, K. Aoki & N. Tenma

National Institute of Advanced Industrial Science and Technology, Tsukuba, Ibaraki, Japan
T. Yamaguchi
Toho University, Funabashi, Chiba, Japan
S. Okubo
The University of Tokyo, Bunkyo, Tokyo, Japan
Keywords: rock properties, rock failure, lab testing, oil reservoir
Methane hydrate (MH) is anticipated to be a promising energy Assuming that a specimen has been fractured at the time when
resource in the near future. Time-dependent characteristics the parameter D expressed by the following equation reaches
of MH-bearing rock are essential in constructing the con- a certain value Df :
stitutive equation to predict the long-term behaviors of a
reservoir. In this study, we conducted creep tests for artificial
MH-bearing rock under triaxial compression and discussed
the time-dependent properties. Figure 1 shows the strain rate
(dεcr /dt) plotted against the elapsed time t for MH-bearing where σ* = normalized deviator stress by σCε . Based on this
∗ relationship from the depen-
hypothesis, we obtain the tcrf –σcr
rock specimens. Larger (dεcr /dt) is observed under higher σ cr .
The slope of each log(dεcr /dt)-log(t) relationship in primary dence of σ Cσ on Cσ as shown in Figure 2. The tcrf in the
creep region varies with σ cr in the range of −1 to −0.4. The creep tests was found to be approximately estimated from the
time-dependent behavior of a MH-bearing rock specimen was Cσ -dependence of σ Cσ in constant-stress-rate tests.
found to be more apparent than that of a sand specimen. Some
specimens experience the increasing (dεcr /dt) and finally
ACKNOWLEDGEMENTS
fracture. Figure 2 shows the creep life tcrf plotted against
the creep stress level σcr∗ normalized by the strength σ
Cε in This work was financially supported by the Research Consor-
a constant-strain-rate test. The tcrf tends to decrease with
∗ . We separately conducted constant-stress-rate tium for Methane Hydrate Resources in Japan in the Japan’s
increasing σcr
Methane Hydrate R&D Program by the Ministry of Economy,
tests for MH-bearing rock specimens. The strength σ Cσ in
Trade and Industry.
constant-stress-rate tests increased with the stress rate Cσ .
Figure 1. Strain rate dεcr /dt versus elapsed time t for MH-bearing
∗
rock specimens. Figure 2. Creep life tcr versus creep stress level σcr .
345

An investigation of brittle rocks failure mechanism based on the damage mechanics
H. Molladavoodi & A. Mortazavi

Department of Mining and Metallurgical Engineering, Amirkabir University of Technology, Tehran, Iran
Keywords: numerical modeling, constitutive model, damage mechanics, damage yield function, strain softening
The rock materials surrounding the underground excavations

typically demonstrate nonlinear mechanical response and
irreversible behavior. The main cause of irreversible changes
in quasi-brittle materials such as rock is the damage process
occurring within the material.
In the definition of damage yield function for rocks, many
authors such as Salari et al. (2004), Shao et al. (2006), and
Chiarellia et al. (2003) considered only the tensile micro-
cracking (mode I). Since quasi brittle materials such as rock
degrade due to shear microcracking (mode ), the definition
of damage yield function under compressive stress field are
vital in the analysis of complex structures .
For a physically meaningful description of the thermo-
dynamic force conjugate to damage, a pseudo logarithmic
damage tensor was introduced. Hence, the thermodynamic Figure 1. Axial stress (σa ) vs. axial strain (εa ) curves recorded in a
force conjugate to pseudo logarithmic damage which is equal uniaxial compression test on an oolittic limestone.
to undamaged elastic strain energy
A damage yield function shows the state boundary of the reported in (Hudson & Harrison, 1997). The numerical com-
damage development. In this work, the tensile loading function pressive fracture energy for unit volume (gfcnum ) was calculated
introduced by Carol et al. (2001) was extended to double ten- as 110 KPa based on element characteristic length, l e , as 5 mm.
sile and compressive loading functions to represent the tensile The tensile fracture energy (Gt ), dividing the total dissi-
and compressive damage of rock materials. In this work, the pated energy by the specimen cross-section, was assumed
damage yield function under compressive loading is proposed 0.035 N/mm based on (Cicekli and Voyiadjis, 2007). Hence,
in the following form: the tensile fracture energy, gft , (the energy dissipated per unit
volume) of the tested sample is 7 KPa based on the element
characteristic length, l e , as 5 mm (Hudson & Harrison, 1997).
A summary of input data used in the numerical analysis is
given in Table 1.
The compressive damage yield function is based on the Figure 2. illustrates a view of the discretized model with
negative eigenvalues of the thermodynamic force conjugate 4096 elements and selected measuring points. The sample
−
to pseudo logarithmic damage (−y(i) ). These negative eigen- uniaxial loading was simulated imposing a velocity field of
values of the thermodynamic force is associated to pseudo −0.005 m/sec at the top of the model in the y-direction.
logarithmic damage. Fig. 3 demonstrates the typical axial stress-strain curves
In above equation, r0c is the elastic energy associated with calculated within the some measuring points.
peak uniaxial compressive strength (σc ) and kc is the ratio With regard to the results presented in Fig. 3, at initial
of elastic energy to total absorbed energy associated with the stages of loading, all elements of the sample deform homoge-
uniaxial compression test. gfc is the area under the compressive nously within the elastic region leading to a bifurcation state.
uniaxial stress-strain diagram (compressive fracture energy At bifurcation point, non-uniform deformations occur within
for unit volume). the sample as a result of strain localization and softening at
In order to verify the proposed damage model, a complete some zones and unloading at other zones. The localization of
stress-strain curve obtained from uniaxial testing of a lime- strain resulting of progressive damage can be seen from Fig. 4.
stone sample (Brady & Brown, 2005) was selected as a basis. In Fig. 5-a, the direction of displacement vectors are down-
This stress-strain curve is shown in Fig. 1. ward, but upon the progression of damage, the displacement
The area under the stress-strain curve envelope represents vectors change orientation demonstrating the formation of a
the experimental compressive fracture energy for unit vol- major diagonal crack.
ume (gfcexp ) of the tested sample. This value was calculated In the presented work, a pseudo logarithmic damage vari-
to be 110 KPa. The compressive characteristic length wc was able was used and formulated. To realistic evaluation of the
assumed as 5 mm based on the experimental observations damage process in rock, two distinct damage yield functions
346

Table 1. Rock property data used in the numerical Analysis.
Compressive Tensile
fracture fracture
Bulk Shear Compressive Tensile energy for energy for
modulus modulus Density strength strength unit volume unit volume
(GPa) (GPa) (kg/m3 ) (MPa) (MPa) (KPa) (KPa)
28.33 13.07 2600 68 6.8 110 7
Figure 4. The distribution of principal strains in: a) prepeak, b)

post-peak state for model discretized with 8192elements c) post-peak
Figure 2. A view of discretization and measuring points. state for model with 4096 elements.
Figure 3. Stress-strain curves obtained from simulation and experi-

mental data. Figure 5. The change in displacement vectors as a function of
loading and damage: a) pre peak, b) post-peak region.
under both compressive and tensile loading were proposed.

The proposed damage model was programmed in C++ envi-
ronment and implemented into the UDEC code. Important
governing mechanisms such as occurrence of damage under
combined states of compression, shear, and tension were
described by the proposed model. Strain localization states
predicted by the model are in good harmony with practical
observations and experimental findings.
347

Influence of humidity and rock fabric on subcritical crack growth in rock
Y. Nara, K. Kaneko, T. Yoneda & N. Hiroyoshi

K. Morimoto
1 INTRODUCTION contains smectite. If the water content in smectite increases,

the basal spacing increases, and the strength of sandstone
Understanding subcritical crack growth (SCG) is important containing smectite decreases. It is thus considered that the
for the long-term integrity of structures in a rock mass. In this resistance to crack growth decreases in smectite when the
study, we have investigated the influences of relative humid- relative humidity is higher and the water content increases.
ity on SCG in Kumamoto andesite (KA), Berea sandstone Consequently, this leads to a remarkable increase in the crack
(BS) and Shirahama sandstone (SS) using the load-relaxation velocity in SS. By contrast, it is considered that the changes
method of the double-torsion technique. of the crack velocity in KA and BS were smaller because the
content of clays was much less than that for SS.
In addition, it is important to investigate the decreases of the
2 RESULTS AND DISCUSSION stress intensity factor when the relative humidity increases.
In Figure 2, the relation between the decrease of the stress
Figure 1 illustrates the relations between the crack velocity intensity factor when the crack velocity is 10−5 m/s, KI (10−5 ),
and the stress intensity factor for KA (Fig. 1a), BS (Fig. 1b) and increase of relative humidity is shown. In this figure, the
and SS (Fig. 1c) under different relative humidities at the same stress intensity factor and relative humidity are normalized
test temperature. The results clearly demonstrate that the crack by the values of them under the low humidity condition. The
velocity increases dramatically when the relative humidity is decrease of KI (10−5 ) for SS is much more significant than
increased. Specifically, the increase of the crack velocity in those for KA and SS which contain few clays. It is considered
SS is more remarkable. that the crack growth resistance in rock which contains more
It is necessary to consider the reason why the increase in clays such as smectite is affected remarkably by the change of
crack velocity for SS was much greater. It is known that SS the relative humidity.
3 CONCLUDING REMARKS
This study reveals that SCG in rock is affected remarkably by

relative humidity and the amount of clays such as smectite.
Therefore, it is effective to control the humidity of surround-
ing environment for the long-term integrity and retarding
weakening of structures in a rock mass.
Figure 1. Relations between stress intensity factor and crack Figure 2. Change of stress intensity factor when crack velocity is
velocity. a: KA, b: BS, c: SS. 10−5 m/s with increasing relative humidity.
348

Estimating the basic friction angle of a very strong

meta-sandstone for rockfill strength prediction
D.J.M. Ngan-Tillard & A. Mulder

Department of Geotechnology, Faculty of Civil Engineering and Geosciences,
Delft University of Technology, Delft, The Netherlands
Keywords: rock joints, rock properties, lab testing
Rock fills are used for multi-purposes in coastal engineer- fills is expected to be related to the dilatancy measured during
ing. They form the foundation base and the core of seawater shearing. It is proposed to calibrate the R and S parameters
breakers. They serve as run ways for cranes that lift-up with dilatancy measurements made during direct shear box
large armourstones to erect seawater breakers. They are testing. It is also recommended to compare the BK model
also dumped on beaches to function as dynamic sea water inspired from the mechanics of rock discontinuities to soil
defence and mitigate coastal erosion. As testing rock fills models. It would be interesting to examine whether parameters
at the field scale requires metric to decametric set-ups and recognized to be of importance for sands, such as the relative
is therefore challenging, engineers favour simple models density, the state paramter and the gradation are indirectly
that capture reasonably well the physics of the behaviour accounted for in the BK model.
of rock fill and give a first estimation of the strength of The basic friction angle is a fundamental parameter of intact
a rock fill. Barton & Kjaernsli (1981) and Barton (2008) rock that contributes largely to the strength of rock fills. It is
took advantage of the analogy between the behaviour of dis- a key parameters of the Barton and Kjaernsli model for rock
continuities and rock fills, and proposed an elegant model fills. Experimental difficulties encountered when measuring
inspired from the Barton and Choubey model for joints the basic friction of a very strong meta-sandstone with the
(Barton & Choubey, 1977), that allows to predict the strength help of a Golder shear box are described. Values measured on
of rock fills for a wide range of stresses, included the very low diamond saw cut surfaces are very low and reflect the quality
stresses. of cutting rather than the natural rock texture. Values mea-
The Barton & Kjaernsli model, named here the BK model, sured on sand blasted surfaces are low as sand blasting caused
requires the determination of few parameters, the basic friction micro-damage to the rock surface. Values measured on rough
angle of the rock material (φb ) and the equivalent particle tensile cracks after correction for dilatancy are high and higher
roughness (R) and crushing strength (S). R and S play for than 40◦ which seems to be a upper bound for the basic fric-
rock fills the role of JRC and, respectively JCS for joints. They tion angle of silicate rocks, based on results published in the
contribute to the rock fill structural strength that is caused by literature. Such high values are explained by the very tight
particle interlocking and decreases when crushing of asperities interlocking and very high strength of the small scale surface
takes place. As for rock joints, the structural strength of rock asperities.
349

Application of pattern recognition analysis of rock acoustic

emission for determination of Kaiser Effect
M. Nikkhah, M. Ahmadi & A. Ghazvinian

Rock Mechanics Department, Faculty of Engineering, Tarbiat Modarres University, Tehran, Iran
Keywords: lab testing, monitoring
Nowadays, acoustic emission testing based on the Kaiser AE signal parameter. It is also possible to use averaging, but
Effect is increasingly used for estimating in-situ stress in lab- we need to have a sufficient number of samples. Since the
oratories. The Kaiser effect can be observed in rocks and other number of initially estimated values in the proposed system
material during their cyclic uniaxial loading by AE moni- is limited, using the averaging method cannot be justified.
toring and corresponds to a stress memory. As soon as the Weighted averaging is another method that can be used. In
load achieves its previously reached peak value, a noticeable this method, each estimator output is assigned a weight in a
increase in acoustic emission activity takes place. Different range of (0, 1] which indicates its importance. The important
methods have been proposed for determining the KE point point in using this method is determining the weight of each
by different researchers. There are cases where KE cannot be output. This study uses expected pre-stress range to assign
observed clearly by plotting data and using the different meth- weights to each output. A trapezoid function has been used
ods. Consequently, the KE results can be doubtful because for weighting outputs. The background knowledge about how
determining the KE value mostly is difficult. In this paper, sure we are that the real Kaiser point is around PG appears
AE testing and monitoring of AE signals have been accom- in the trapezoid figure. Sigari et al. (2008) have proposed a
plished under uniaxial compressive testing on the cylindrical method for combining estimator outputs which works based
samples of sandstone. After each preloading stage, AE test- on clustering data. In their method which was proposed for
ing was carried out without delay by applying a loading stress more accurate estimation of the signal period time, first out-
more than the previous loading stress level. Pattern recogni- puts from different estimators are clustered, then the cluster
tion is the study of how machines can observe the environment, with the highest frequency is chosen as the winner as in the
learn to distinguish patterns of interest from their background, majority voting method. After choosing the winning cluster,
make sound and reasonable decisions about the categories of in order to determine the final output accurately, weighted
the patterns. It is the science of the methods of data analy- averaging is used with the data in this cluster. This method
sis for the purpose of separating, clustering and classifying makes it possible to use the majority voting method for com-
them based on a number of features. One of the most impor- bining estimators whose outputs are in the continuous space
tant issues in the science of pattern recognition which has because by clustering the data the continuous space changes
recently attracted considerable attention in different sciences into a discrete one. Therefore, the majority voting method can
is combining classifiers. be used to choose the winning cluster. The proposed method
One of the simplest methods for combining results obtained uses a method similar to that of Sigari et al. In our method,
from several classifiers is Majority Voting which can be used in order to cluster estimator outputs, first data histogram is
in three different ways: a) Unanimity, b) Simple Majority Vote, computed in 20 ranges with equal distances from the range
and c) Plurality Vote. For combining estimators, averaging [0,2 PG]. Thus, by computing the histogram the data are clus-
can be used. The averaging method can also be used in two tered in a simple manner. The method proposed by Sigari
ways: a) Simple Averaging and b) Weighted Averaging. We et al. uses the Isodata method for clustering; using the Iso-
can assign weights to each estimator output and thus change data or other clustering methods in the current application
their effect on the final output. This method is called weighted can lead to producing clusters with different sizes in the range
averaging. Simple averaging is a type of weighted averaging in [0,2 PG]. Producing clusters with unequal sizes can lead to
which all the estimators have the same weights. There are two inaccurate estimation of the KE point. Therefore, we need to
ways to assign weights to estimators. The first is the manual use the histogram method and cluster the estimator outputs
weighting method in which each estimator is weighted based into the clusters. Now, each cluster is assigned a weight and
on an expert individual’s background knowledge. The second the weighting system is the same as in weighting data which
method is the automatic method in which a criterion is set was mentioned earlier. In other words, using a trapezoid func-
for estimators’ level of effectiveness. Then, each estimator is tion the clusters are weighted in such a way that those closer to
weighted based on this criterion. The most important stage in the center are assigned a greater weight. After weighting the
the proposed method is the combination of estimator outputs clusters, the cluster whose frequency multiplied by it weight
from different parameters. gives the highest value is chosen as the main cluster. If the
In the proposed method, outlier data could occur as a result center of this cluster is chosen as the final estimation, the error
of an estimator’s error in estimating the KE point from a certain of estimation could be great. In order to determine the final
350

estimation more accurately, averaging is used with the data in with an acceptable level of accuracy. The proposed method
this cluster so that we will have a smaller error. Thus, the final uses the main parameters of acoustic emission including
output would be the average of the data chosen from the main energy, count, rise time and duration but it should be noted that
cluster. the mechanism of wave emission source is excluded. There
The results of different methods of estimating the Kaiser was a good level of correspondence between the results of
effect and the combinatory methods for more accurate the proposed method of combining clusters and the weighted
automatic determination of KE used in 32 acoustic emis- averaging method. The procedures of the proposed method
sion tests on sandstone specimens has been discussed. The can also be used in similar research projects focusing on other
results show that pattern recognition methods and the pro- rocks.
posed method can be used to determine the Kaiser effect point
351

Acoustic emission of Dionysos marble specimens in uniaxial compression
P.P. Nomikos, K.M. Sakkas & A.I. Sofianos

Acoustic emission has been extensively utilized over the past capabilities. For all AE signals received the AE features were
decades to quantify microstuctural damage of rock both in determined automatically by the recording system.
laboratory and field studies. In this study the acoustic emission In Figure 2 typical stress-strain diagrams obtained from the
signature of Dionyssos marble under uniaxial compression is uniaxial compression tests of the prismatic samples is shown.
examined, and the stages of damage are identified combining For this test, a preload of 20 kN, corresponding to a stress of
acoustic emission and deformation measurements. Further, 8 MPa, was applied. In these diagrams the axial stress differ-
the evolution of some AE parameters with the applied load is ence σa -σa,preload is normalized with the maximum axial stress
examined whereas special attention is given to their usefulness difference σ max -σ a,preload obtained during the test. The axial
as precursors of the forthcoming rock failure. and lateral strain differences (ε-εpreload ) are also normalized
The Dionyssos marble is a calcitic medium crystalline with respect to their values at the maximum stress (εmax -
marble, with mechanical properties varying with the angle εpreload ). Also, the normalized volumetric strain is shown in
of the loading direction to the main plane of anisotropy. the diagram.
The marble specimens prepared for the uniaxial compres- In Figure 3 the normalized cumulative AE (excluding
sion tests are either prismatic or cubic with a cross section of any the events recorded during the preload stage) is shown
50 × 50 mm2 and 100 × 100 mm2 respectively, and 100 mm with respect to the normalized axial stress σa /σmax . In the
height. The specimens were loaded under constant loading same diagram the normalized axial strain-axial stress curve is
rate until macroscopic failure. The applied load, the axial and shown.
lateral strains were continuously monitored simultaneously From the diagrams of Figures 2 and 3 the elastic and crack
with the AE monitoring. development stages may be identified. The crack initiation
To capture the AE signals piezoelectric transducers were stress is estimated by the onset of the acoustic emission activ-
used. The transducers were attached on the surface of the ity at a normalized stress level of about 0.3. The crack damage
specimens using silicon vacuum grease as a coupling agent threshold is estimated at a normalized axial stress difference
(Fig. 1). Before each test, coupling of the sensors was exam- of about 97%. This stress level corresponds to the volumet-
ined using the PLB (pencil lead break) technique. Wideband ric strain curve reversal and to a drastic increase of AE rate
AE sensors were used. The AE signals were preamplified as shown in Figure 3. The main increase of AE activity is
by external amplifiers, with 40 dB preamplification. The AE observed near the macroscopic failure point.
triggering threshold was set to 50 dB. In each case full wave- Energy released by damage during loading may be deter-
form data were recorded for each AE signal with a multichan- mined either by rigorous or simplified approaches. A sim-
nel data acquisition system with real time signal processing plified measure of the released energy is the ASL (Average
Figure 1. Photograph of a marble prism tested in uniaxial compres- Figure 2. Typical normalized stress-strain diagrams obtained from
sion and the AE transducers mounting. the uniaxial compression tests of the prismatic marble specimens.
352

Figure 3. Typical stress-strain diagrams obtained from the uniaxial
compression tests of the prismatic marble specimens.
Figure 5. Normalized ASL values with respect to the normalized
load obtained from the uniaxial compression tests of the cubic marble
specimen #1.
Figure 4. Normalized ASL with respect to the normalized load.

Specimen Prism #1.
Signal Level) which is a measure of the continuously varying Figure 6. Normalized b-values with respect to the normalized load
and “averaged” amplitude of the AE signal. In Figure 4 the calculated for the prismatic marble specimens.
normalized ASL values (divided by the maximum ASL value)
of the AE signals for the prismatic marble specimen #1 with
respect to the normalized axial stress are shown. An increas- after 95% of the maximum load. This may be an indicator of
ing (almost linear) trend of ASL values with the normalized the damage state of the specimen. Also, this point corresponds
load is observed up to a normalized load of approximately to accelerated AE activity observed in the experiments and
0.95. An abrupt increase of the slope of a hypothetic bi-linear to the abrupt increase of the ASL values.
envelope is observed at 97% of the normalized axial stress. In conclusion, for the Dionyssos marble specimens tested
Similar response may be observed for the other prismatic spec- under uniaxial compression the onset of the AE activity is
imens tested. For the cubic specimens tested a precursory trend observed at a stress level lower than 30% of the maximum
increase is only identified very close to macroscopic failure. load. The crack damage threshold may be correlated with the
In Figure 6 the b-values calculated for the prismatic marble ASL values of the recorded AE signals. An abrupt increase
specimens #1 to #3 are shown with respect to percent loading. of the ASL-loading curves is observed at this stress level.
An increasing trend of the b-values up to approximately 90% A significant percentage of the AE activity and an appar-
of the maximum load is shown. Thereafter, b-values decrease. ent precursory drop in b-value, obtained from an amplitude
In all cases, an apparent precursory drop in b-value is observed distribution analysis, are also observed at this load.
353

An elaborative evaluation of various methods for simultaneous estimation of uniaxial

compressive strength and elasticity modulus of intact rocks
R. Noorani & M. Hesami

Metra Consulting Engineers Co, Tehran, Iran
Sabir Company – General Contractor of infrastructural Project, Tehran, Iran
Keywords: lab testing, neural network, problem rocks and rock properties
1 INTRODUCTION Table 1. Descriptive statistics of the rock sample data for 126 sets
(including training and test sets).
Mechanical properties of rocks such as uniaxial compressive
strength (UCS) and modulus of elasticity (E) of intact rocks Parameter Mean Min Max Std. dev
are of vital importance in mining and civil activities. Rock Vp (m/s) 5.34 2.2 6.82 1.03
engineers extensively use the UCS and E in designing tun- SHR 36.12 18.22 58.9 7.69
nels, foundations, slopes and other rock engineering projects. T (MPa) 8.7 3.2 20.15 3.28
In addition, analytical and numerical solutions require both IS(50) (MPa) 7.57 1.96 17.61 3.46
parameters. Measuring these parameters has been standard- Dry (gr/cm3 ) 2.51 2.05 2.77 0.16
ized by both the ISRM (1981) and ASTM (1984) which based Sat (gr/cm3 ) 2.56 2.22 2.79 0.12
on these standards, UCS test requires specimens prepared n (%) 5.03 0.15 27.3 6.21
accurately but it is often extremely difficult and time con- UCS (MPa) 61.72 3.36 133.54 25.62
suming to obtain such samples from weak, highly fractured E (GPa) 6.45 1.99 13.24 2.42
and thinly bedded rocks. Therefore, some predictive models
utilizing simple index test such as schmidt hammer, point load
index, sound velocity, physical properties, and other tests were
developed by many researchers (Hassani et al. 1980,Alvarez & desired output of models and other parameters (engineering
Babuska 1999, Meulenkamp & Alvarez 1999, Kahraman index tests) are input parameters.
2001, Lashkaripour 2002, Sonmez et al. 2004, Gokceoglu &
Zorlu 2004, Fener et al. 2005, Karakus & Tutmez 2006,
Tercan & Ozcelik 2006) owing to the fact that these index 3 NETWORK ARCHITECTURE
tests require less or no sample preparation when compared
with the uniaxial compressive strength test. In addition, they The prediction capability of the optimizedANN used for deter-
can be used easily in the field. mining the mechanical properties of rocks is assessed using
This paper presents the application of a neural network for different data pairs than those that were not used for training
the simultaneous prediction of the UCS and modulus of elas- the ANN.
ticity from index tests. To investigate the suitability of this After many trials, the optimal ANN is determined to be a
approach, the results of the network were compared with pre- model with a 7-25-15-2 ANN architecture, that there are two
dictions obtained by conventional statistical relations. For this hidden layers with 25 and 15 nodes in each layer for predicting
purpose different rock types have been tested based on ISRM the mechanical properties of rocks. The criteria used to deter-
suggested methods. These tests consist of point load, sound mine the optimal ANN model is the lowest root mean square
velocity, schmidt hammer, porosity, tensile strength, uniax- errors (RMSE) of the predicted values.
ial compressive tests, modulus of elasticity, dry and saturated Figures 1, 2 show the estimated values versus measured val-
density. ues for UCS and E from optimized ANN model. They indicate
that designed ANN model has sufficient ability to predict the
desired variables.
2 EXPERIMENTAL INVESTIGATIONS
4 COMPARISON WITH CONVENTIONAL METHODS
Rock blocks have been collected and 126 sample sets for rock
mechanics tests were obtained. Each data set consists of UCS, Several regression models were constructed in order to assess
E, point load index (IS(50) ), porosity (n), sound velocity (Vp), the ability of ANN models. Linear, power, logarithmic, and
schmidt hammer (SHR), tensile strength (T), dry and satu- exponential functions were employed so as to build a uni-
rated density. The procedures suggested by ISRM [2] for rock variate regression models. Strongest simple correlations were
mechanics tests were considered. Table 1 lists the descriptive obtained between UCS and point load index (R2 = 0.54) and
statistics of various parameters. In this paper UCS and E are between E and dry density (R2 = 0.46). Afterwards several
354

Figure 1. Predicted versus measured UCS from ANN. Figure 3. Predicted and measured value of UCS for MR.
Figure 2. Predicted versus measured E from ANN. Figure 4. Predicted and measured value of E for MR.
multiple regression (MR) models were performed in the sec- values to predict the UCS from ANN models on test data sets
ond stage of the regression analyses (Figs 3, 4). Another were 9.64 and 7.96 respectively. These indices were 1.26 and
method namely canonical correlation has been developed in 0.95 for modulus of elasticity. As seen from Figures 1, 2, the
the next step. Whereas multiple regression predicts a sin- ANN model shows a good relation with high correlation of
gle dependent variable from a set of multiple independent coefficient (r = 0.89 for UCS and r = 0.87 for E). Consider-
variables, canonical correlation simultaneously predicts mul- ing performance indices and mentioned figures the prediction
tiple dependent variables from multiple independent variables. capacities of ANN to estimate UCS and E of intact rocks are
Some researchers utilized this method to study the relation- higher.
ship between mechanical properties of rock and engineering In addition, considering all indices, improvement can bee
index tests (Tercan & Ozcelik 2006, Noorani & Ghazvinian seen in ANN model rather than univariate models and also
2008). multiple regression. According to the results, artificial neural
Several indices were calculated to control the capacity of network is a robust means to estimate mechanical properties
the developed equations in this study. The RMSE and MAE of intact rocks simultaneously.
355

Drying-induced deformation and consequent change in mechanical

properties of sedimentary rocks
M. Osada, N.M. Adikaram & T.N. Illankoon

Geosphere Research Institute, Saitama University, Saitama, Japan
Keywords: lab testing, rock properties, nuclear repository, site characterisation
1 INTRODUCTION
In this paper, we would like to introduce drying-induced

deformation behaviors of sedimentary rocks which have been
obtained in our laboratory, and the maximum volumetric
shrinkage in each rock type is compared with peak radius
of pore size distribution from mercury intrusion porosimetry
(MIP).
The four rock types are used to be compared, that is Opalinus
clay from Switzerland, Shirahama sandstone, Tage tuff, and
a pumiceous tuff.
2 METHODOLOGY
Figure 1. Relation between peak radius and maximum volumetric
shrinkage.
2.1 Drying experiment
Laboratory experiments of drying were carried out with the
cylindrical core samples. Four strain rosettes are attached on 3 RELATION BETWEEN MAXIMUM VOLUMETRIC
side surface of a cylindrical specimen in a symmetrical manner SHRINKAGE AND PEAK RADIUS OF PORE SIZE
against the cylindrical axis in order to measure successive DISTRIBUTIONS
deformation with the change in direction of principal strains.
Strain variations and weight loss of a sample were con- If the mechanism of drying-induced deformation is related
tinuously measured with logging of temperature and relative to the pore structure of a rock, pore size distribution must
humidity of environment up to that no weight loss is recorded. have adequate information. Then, as the first approximation,
The maximum shrinkage values in volumetric strain for four the relation between peak radius and maximum volumetric
rock types are different for each rock type. In case that vol- shrinkage is plotted on log-log scale as shown in Figure 1.
umetric strain does not reach to a steady state even after 15 All data are plotted on the right-upper half space of this
days passing like Opalinus clay and Shirahama sandstone, the graph and the smaller peak values seem to be lined up.
final point is taken as the maximum shrinkage value. The maximum shrinkage can be roughly estimated using this
relation.
2.2 Mercury intrusion porosimetry
Mercury intrusion porosimetry was used to evaluate the pore 4 SUMMARY
structures of four rock types. The absolute pressure applied
to inject the mercury into the dried sample is used to evaluate In this paper, we introduce drying-induced deformation expe-
the pore size from Young’s equation. riments performed in our laboratory. In addition, the first
Opalinus clay and pumiceous tuff have unimodal pore size attempt has been made to estimate the maximum shrinkage
distributions with different peak values, while Shirahama values by using MIP results.
sandstone and Tage tuff have bimodal ones. The ranges of Change in mechanical properties during drying is also
each peak radius for four rock types are summarized with the reported in this paper. The desiccation-driven hardening of
median values. Tage tuff is clearly shown.
356

Assessment of clay minerals around fault using slake durability test

and reflectance spectroscopy
Chang-Shin Park, Chang-Uk Hyun & Hyeong-Dong Park

Department of Energy Systems Engineering, Seoul National University, Seoul, South Korea
Keywords: lab testing, problematic rocks, rock properties, weak rock, clay mineral
Fault is a geological crack in the earth’s crust resulting from the located from fault planes, contain more clay minerals. Slak-
displacement of one side with respect to the other by being ing durability test and reflectance spectroscopy wasapplied to
forced, and crushed rocks around a fault plane show weak confirm assumption (Fig. 1).
physical properties. Among materials located between fault We collected rock samples from faults located inside of ◦◦
planes, especially, clay minerals, derived from weathering of mine, Samcheok-Si, Kangwon-Do, South Korea. Three kinds
bedrock or existence inside rocks, the clay has been become of samples were collected from fault gauge located between
one of the biggest problems during underground construction fault planes, rocks of fault planes, and fresh rocks located
by combining with water. For this reason, investigation of clay 1 m∼2 m apart from the fault planes (Fig. 2). Among many
minerals is essential to improve stability of underground con- faults inside the mine, 3 faults were selected and sampled.
struction because clay minerals exist inside of rocks lead to WG1, 2, and 3 were given to each fault, and then subnumber
not only physical expansion but also shotcrete malfunction 1, 2, and 3 were named to classify the samples according to
and injection efficiency reduction by bonding with water. distance from fault planes.
Slake durability test (SDT) can estimate rock’s durability SDT suggested by international society for rock mechanics
against slaking phenomenon whichis observed easily for clay- (ISRM) was used, and total 4 cycles were repeated to measure
ish rocks due to repeat drying and wetting. In this study, we slake durability index (SDI). Collected rock samples were cut
used X-ray diffraction analysis (XRD) to classify clay miner- to 10 spheroidal lumps of 40∼60 g for each specimen, total
als in rock samples collected and assumed that the closer rocks weight of 450∼550 g. Ten lumps were placed in meshed drum
Figure 1. Flow chart for this study.
Figure 2. Left, location of the OO mine in South Korea. Right, rock sampling area around fault in the mine.
357

Table 1. Result of slake durability test. by detecting diagnostic absorption feature in the vicinity
of 2,200 nm wavelength. In this study, ASD Fieldspec® 3
Slake durability index
spectrometer was used to measure and to detect the absorp-
Samples 1st cycle 2nd cycle 3rd cycle 4th cycle tion feature of various clay minerals in visible and infrared
wavelength range from 350 to 2,500 nm.
WG1-1 99.6 99.4 99.3 99.0 As a result of XRD analysis, clay minerals, e.g. chlorite
WG1-2 100.0 99.9 99.9 99.8 which is expected to come from feldspar, were detected. For
WG1-3 100.0 100.0 99.9 99.9 SDT, according to distance from fault planes, SDI of fault
WG2-1 99.8 99.5 99.2 98.9 gouge is shown the lowest value among all, and SDI of fresh
WG2-2 99.9 99.8 99.7 99.6 rocks is shown the highest value (Table 1).
WG2-3 99.9 99.8 99.7 99.6 The WG1-1, WG2-1 and WG3-1 samples from fault gouge
WG3-1 97.6 96.4 95.5 94.6
showed deep absorption depth in the vicinity of 2,200 nm
WG3-2 99.9 99.9 99.8 99.8
WG3-3 100.0 100.0 100.0 99.9 wavelength and this reflects the existence of clay minerals
but other samples showed shallow absorption feature. The
absorption feature of 2,200 nm was known to be originated
from Al-OH of clay minerals and this feature can be easily
and rotated at a speed of 20 rpm for 10 min. After finishing detected in the reflectance curves. This spectral analysis using
rotation, drum with lumps were dried for 2 hours in a dry- reflectance spectroscopy can be directly applied to the field
ing machine at 100◦ C. SDI was calculated using precision without complicate and time-consuming preprocessing steps
weighing machine with equation below (1). such as sampling or making rock powder, so this method can
improve efficiency of field investigation.
ACKNOWLEDGEMENT
Reflectance spectroscopy is a non-destructive and in-situ
analysis method which can be applied to geomaterials, and This work was supported by the National Research Foundation
has been successfully applied to detect the clay minerals of Korea (NRF) Grant (NRF-2010-616-C00044).
358

Relation between mechanical properties of rock and sedimentary structure

Keywords: lab testing, rock failure, rock joints, rock properties, rock stress
Laboratory experiments were carried out to investigate the

relationship between sedimentary structure and anisotropy of
mechanical properties. Especially, the elastic wave velocity
and uniaxial compression strength were carefully measured
for cylindrical specimens that were prepared in different direc-
tion (every 30 degree from North; size: ϕ50 mm, h 100 mm).
First of all, a plate shaped block (500 mm × 500 mm ×
100 mm) which is parallel to the bedding plane was pre-
pared. Totally 12 core specimens, then, were taken from the
block (Fig. 1a). P wave velocity test and uniaxial compres-
sion test were performed for the naturally dried specimens,
respectively. After the tests, another block which is vertical
to the bedding plane and also parallel to 11-5 direction of
the bedding parallel block was prepared (Fig. 1b). The direc-
tion (11-5; NNW-SSE) was a maximum value direction of the
bedding parallel block specimens. Therefore, 3-dimentional
anisotropy of mechanical properties could be investigated.
Figure 1. Specimen preparation (a) parallel to the bedding plane,
From the P wave velocity test result, it was confirmed that (b) vertical to the bedding plane (it is also parallel to 11-5 direction
the anisotropy of P wave velocity was not significant. For of (a)).
the bedding parallel block specimens, the velocity differ-
ence between maximum and minimum value was 0.09 km/sec
(2.85–2.77). It is about 3.06% of mean velocity (2.79 km/sec).
For the bedding vertical block specimens, the max/min veloc-
ity difference was 0.27 km/sec and it is about 9.59% of
mean velocity (2.81 km/sec). However there were significant
anisotropies of uniaxial compression strength, especially, for
the bedding parallel block specimens, the uniaxial compres-
sion strength differences between maximum and minimum
value was 14.26 MPa (47.99–33.73). It is about 35.47% of
mean value (40.19 MPa) (Fig. 2a). The specimen number 11
and 5 were placed almost same direction with the paleo-
flow direction of this area (SSE → NNW). For the vertical Figure 2. Uniaxial compression strength for each direction (unit:
block specimens, the difference between maximum and min- MPa) (a) parallel to the bedding plane, (b) vertical to the bedding
imum strength was 7.73 MPa and it is about 18.56% of mean plane.
strength (41.65 MPa) (Fig. 2b). The maximum value speci-
men direction was about 30 degree inclined from the bedding cross-lamination were not influential to the anisotropy of
plane. Moreover, the change of Young’s modulus anisotropy mechanical properties. This implies that it is necessary to
was investigated for different stress levels (10%, 30%, 50%, be more careful when the sedimentary rock is used for the
70% and 90% of uniaxial compression strength). The exper- facilities such as a radioactive waste disposal or fuel storage.
imental results indicate that the anisotropy of mechanical *This study is regulatory support research funded by the
properties in Kimachi sandstone is strongly related with the Nuclear and Industrial Safety Agency, Ministry of Economy,
sedimentation environment. However, the bedding plane and Trade and Industry, Japan.
359

Estimation of uniaxial compressive strength of weak rocks using needle penetrometer
Young-Ho Park
Department of Energy and Resources Engineering, Chosun University, Gwangju, Korea
Yuzo Obara
Graduate School of Science and Technology, Kumamoto University, Kumamoto, Japan
Seong-Seung Kang
Department of Energy and Resources Engineering, Chosun University, Gwangju, Korea
Keywords: uniaxial compressive strength, weak rock, needle penetration resistance, curing period
The various mechanical properties of rock materials, which indication ring, the UCS-NPR correlation chart, the remov-
are commonly used in rock engineering, are required for ana- able cap, and the penetration needle. Needle penetrometer can
lyzing and evaluating the stability of geo-structures such as be adapted for indirectly estimating the uniaxial compressive
tunnels, dams, and slopes. In particular, the uniaxial compres- strength of weak rocks from the needle penetration resistance
sive strength (UCS) has been used as one of the important (NPR) without sampling. The needle penetration resistance
mechanical properties in rock materials. The uniaxial com- can be calculated by an equation (1) as the ratio of load (F)
pressive strength of rock materials can be generally obtained and depth of penetration (D). An equation (2) was suggested
by the convensional uniaxial compression test. This conven- by a manufacturer from the correlation between the uniaxial
tional testing method can be obtained the value of exact compressive strength and needle penetration resistance, based
uniaxial compressive strength, whereas a lot of time and effort on data from artificial and natural rock specimens.
to prepare specimen for its tests are needed. In addition, a test The purpose of this study is to estimate the uniaxal compres-
specimen of the appropriate size demanded for the uniaxial sive strength for the weak rocks from the needle penetration
compression test is sometimes difficult to manufacture. Due resistance measured by needle penetrometer. For this purpose,
to these problems, some testing methods such as the point load we firstly made the artificial cemented-based specimens and
test, Brazilian test, Schmidt hammer test, and block punch curing period of each specimen is six steps with the ratio of
index test have been developed and suggested for estimating maxed materials of three cases. Then, the needle penetrometer
the uniaxial compressive strength. However, these methods are test and uniaxial compression test were performed to obtain
also difficult for preparing the appropriate testing specimens, the needle penetration resistance and the uniaxial compres-
especially in the case of weak rocks. That is, during coring sive strength. Secondly, the relationship between the measured
process for weak rocks, core specimens for the experiment needle penetration resistance and the estimated and measured
can be splitted into smaller sizes. uniaxial compressive strengths was compared. Finally, the
For solving above difficulties, a new testing device has been uniaxial compressive strength was estimated. In relationship
developed by a manufacturer of Maruto Corporation in Japan between NPR and UCS, the UCS gradually increased until
as non-destructive and portable testing methods for laboratory the NPR was less than 10 N/mm, then drastically increased
and field. This device is called needle penetrometer applied from more than 10 N/mm. The NPR was widely distributed
for weak rocks, and consists of eight parts, namely presser for at the curing period of 3-day to 14-day, but was narrowly dis-
measurement of the penetration and protection of the penetra- tributed at the curing period of 21-day to 28-day. In addition,
tion needle, the chuck used for fixing the penetration needle, the NPR of the curing period between 21-day and 28-day was
the penetration scale of 0∼10 mm, the load scale, the load nearly constant, but the UCS during those periods showed a
Figure 1. Needle penetrometer and its configuration.
360

Table 1. The results of the estimated unaxial compressive strength UCSe and measured unaxial compressive strength UCSm and the
calculated needle penetration resistance NPR
Specimen-1 Specimen-2 Specimen-3

(C : B : W = 1.0 : 1.0 : 1.3) (C : B : W = 1.3 : 0.7 : 2.3) (C : B : W = 1.5 : 0.7 : 2.0)
Curing NPR UCSe UCSm NPR UCSe UCSm NPR UCSe UCSm
Period (N/mm) (kN/m2 ) (kN/m2 ) (N/mm) (kN/m2 ) (kN/m2 ) (N/mm) (kN/m2 ) (kN/m2 )
3-day 1.86 0.77 0.49 2.87 1.18 1.02 4.94 2.02 1.59
7-day 4.97 2.03 1.66 6.03 2.46 2.67 11.03 4.47 3.50
14-day 7.63 3.10 3.48 8.70 3.54 3.87 14.00 5.66 5.22
21-day 9.25 3.76 4.06 11.30 4.58 4.68 16.71 6.74 6.91
28-day – – – 11.57 4.69 6.86 17.26 6.96 8.92
Figure 3. Relationship between the estimated uniaxial compressive

Figure 2. The results of needle penetrometer test on Specimen-1, strength and the measured uniaxial compressive strength.
-2, and -3 according to curing period.
big difference. In relationship between the estimated unaxial

compressive strength and the measured uniaxial compressive
strength, the gradient of UCS obtained from this study was sit-
uated in the upper limit of 1:1.7 and the lower limit of 1:0.6.
Most of the values are closely located at the accurate estima-
tion gradient of 1:1. As a result, it is considered that the UCS
estimated from the NPR is reasonable enough in the case of
weak rocks.
361

Testing of fiber reinforced sprayed concrete – influences on the energy

absorption capacity
Gerhard Pittino
Chair of Subsurface Engineering, Department Mineral Resources and Petroleum Engineering, University of Leoben, Austria
Yuewei Bian
Department of Geotechnical Engineering Tongji University, Shanghai, P.R.C
Subject: Others (shotcrete rock support)
Keywords: lab testing, numerical modeling, rock support
1 INTRODUCTION investigations at the Chair of Subsurface Engineering (SE),

in which the effect of the test specimen imperfections and
Sprayed concrete is an important support element used in thus also the support conditions (Fig. 2) are being tested and
geotechnical engineering. Generally, concrete and sprayed estimated numerically using ABAQUS and FLAC3D .
concrete can be reinforced with bars or steel mesh. However,
more and more fiber reinforced shotcrete is used in tunneling
as it offers some advantages. For instance, the use of fiber 2 TESTING OF FIBER SHOTCRETE
shotcrete saves the fixing of reinforcing mesh so no miner has
to be in the unsecured area of excavation, resulting in a signi- 2.1 Test procedure according to ÖNORM EN 14488-5
ficant safety improvement. A wide range of steel and synthetic Energy absorption is determined using the slab test accord-
fibers is used for reinforcing concrete and sprayed concrete. ing to ÖNORM EN 14488-5, “Testing sprayed concrete – Part
The Austrian Association for Concrete and Construction 5: Determination of energy absorption capacity of fiber rein-
Technology (ÖVBB) guideline “Sprayed Concrete” provides forced slab specimens”, in which a square shotcrete slab with
the classification for the evaluation of the post-crack behavior 60 cm sides, 10 cm thick and 7 days old is supported around
of fiber reinforced sprayed concrete alternatively in so-called the perimeter and deflected from above by a central testing
sprayed concrete classes EV500, EV700 and EV1000 in which piston.
e.g. EV500 corresponds to an energy absorption capacity of The slab is tested with a constant deflection rate of 1 mm per
at least 500 joules. The procedure to the determination of minute until a deflection of 30 mm is reached. The evaluation
the load-deflection behavior is defined in the ÖNORM EN is based on the work required to produce a deflection of 25 mm.
14488-5, 2006. The standard exactly regulates the test equip- SE uses a computer-controlled, servo-hydraulic testing
ment, the specimen dimensions and the test procedure. machine with a system stiffness of 9,000 kN/mm. The
Deviations from the standard specification have to be indi- hydraulic cylinder is mounted underneath so the piston works
cated in the test report. upwards. For this test rig the support frame was designed and
The various force-deflection curves (e.g. Fig. 1) from the
in-house tests and tests from third parties are part of continuing
Figure 1. Force-deflection curves with peak force at the a) first,

b) second and c) after several failures/cracks. Figure 2. Possible influences on the slab test result.
362

Figure 3. Test set-up for slab test with build-in girder and specimen.
calculated by SE and made by Östu-Stettin Stahlbau (Fig. 3).

An advantage of this test set-up is the possibility to observe the
development of the cracks on the bottom side of the specimen
during the test. Every sudden force decrease can be assigned Figure 4. Slab test with square and round test specimen and
to a crack. different support conditions.
2.2 Executed slab tests The comparison of the plate stiffnesses as expected leads to
very different magnitudes, expressed in force per vertical dis-
First of all, a reference test was carried out accordingly to the placement. The analytical solution is for a simply supported
standard ÖNORM EN 14488-5. Therefore a series of three plate and thus exaggerates the stiffness. The analysis yields
slabs was concreted with the addition of synthetic fibers as approx. 600 kN/mm. If the bedding material is modeled very
reinforcement. Because of the influence of the bedding mate- stiffly in the numerical analysis, we get obtain stiffness of
rial on the test result, which also should be investigated, the approx. 450 kN/mm. The stiffness found from the slab test
slabs were milled at the top side and underside and tested is for different bedding materials between 50 kN/mm and
without bedding material. 100 kN/mm.
In addition plain concrete slabs were tested for compari- An annular support and a square or circular slab was pro-
son purposes. Some slab tests are executed very fast with a vided in the draft of the ÖNORM EN 14488-5, issue 2003.
displacement rate of 10 mm per minute. The Swiss Standard SIA 162/6 (SN 562162-6, February 1999)
Numerous numerical analyses with FLAC3D show the influ- still specifies the slab test with round specimens and support
ence of the support condition on the failure pattern. The in addition to the test with square specimens and support.
analysis results are validated by laboratory tests using the ASTM C 1550-03a suggests a round slab supported on three
same support conditions as shown in Fig. 4. It appears that symmetrically arranged pivots. Therefore the geometric mod-
the supporting conditions have a significant influence on the els shown in Fig. 4 are examined.
results. Some results of the slab tests and numerical analyses are
described in the extended version of this contribution.
3 ANALYTICAL AND NUMERICAL ANALYSES
Parallel to the laboratory tests analytical and numerical

analyses were performed.
363

A comparative study of mi , the Hoek-Brown constant for intact rock material
S.A.L. Read
GNS Science, Lower Hutt, New Zealand
L. Richards
Rock Engineering Consultant, Canterbury, New Zealand
1 INTRODUCTION the stress space would be σ1 < σ3 < σ1 /3.4 on the basis of
Mogi’s original definition of the brittle/ductile transition.
The original Hoek-Brown failure criterion (Hoek & Brown Subsequently Hoek & Brown (1997) recommended that the
1980) was defined in terms of major and minor total stresses range should be 0 < σ 3 < 0.5σ ci .
as follows: Sheorey (1997) advocated the inclusion of tensile strength
data along with unconfined and triaxial compression data, as
better regression is obtained with data point(s) closer to the
σ 3 axis of the principal stress plots.
The calculation method to analyse the testing data has
where the relationship between the principal stresses (σ 1 and an effect on the mi values, with the RocData and RocLab
σ 3 ) at failure is defined by two constants (m and s) and the programs from RocScience (2010) using the non-linear
uniaxial compressive strength of intact rock (σ ci ). Levenberg-Marquardt method of least-squares curve fitting as
In the case of intact rock material where s = 1.0, Hoek & the default, with options for the simplex reflection and linear
Brown (1980) noted that m ≈ σ ci / |σ t | where σ t is the uniaxial regression methods.
tensile strength of the intact rock.
The generalized failure criterion (Hoek & Brown, 1997) in 2.1 Guideline values for mi
effective stress terms included three constants (mb , s and a) in
As part of the original failure criterion, Hoek & Brown (1980)
the equation as:
suggested that m for intact rock (mi ) reflected rock type,
with values increasing ≈7 for carbonate rocks to ≈25 for
coarse-grained rocks. Some 15 years later, Hoek put forward
guideline mi values, based on a more detailed lithological
classification, with subsequent revisions including ranges of
In the case of intact rock material (i.e. s = 1) the generalized values.
equation becomes:
2.2 Values for mi calculated from test result databases
Valuable databases of laboratory test data have been provided
by Sheorey (1997) and Douglas (2002). In general, the cal-
culated database values show little correlation with guideline
The full paper first summarizes procedures used to calculate values, with calculated values having a much greater range
intact material constant mi and then compares values obtained than those in the guidelines.
from a large database of testing data with guideline tables. Correlations of mi with other rock parameters, (such as
After considering further testing results, suggestions are made unconfined compressive strength) are mostly very poor.
for a simple method of estimating mi by means of a combina-
tion of unconfined compression and indirect Brazilian tensile 2.3 Estimates of mi values using tensile testing
tests.
The only rock material parameter with which mi has any rea-
sonable correlation is the ratio of the unconfined compressive
2 CALCULATION OF mi VALUES strength σ c to the tensile strength σ t . Sheorey (1997) defined
this ratio as R = σ c /|σ t .|.
The Hoek-Brown failure criterion is intended for use in the The laboratory tensile data in the Douglas database com-
brittle range of rock behaviour, and this controls the range prised results from both pure uniaxial and indirect (Brazilian)
of minor principal stress (σ3 ) values over which triaxial test- testing. Using these data (59 of the 521 sets in the database),
ing to determine mi is carried out. Initially Hoek considered the intact rock parameter mi , shows a fairly good relationship
364

The results are given with respect to measured values of σ c ,
(as measured by the unconfined compressive (qu ) test) and σ tb
(as measured by Brazilian tensile test), and R(ratio of σ ci to
σ tb ), together with calculated values of σ ci , σ t and mi , with or
without the inclusion of tensile results.
The calculated mi values are generally outside the ranges for
guideline values (e.g., schists vary from 5 to 8, which is below
the lower end of the guideline range of 12 ± 3). While the
ignimbrites at the two locations cover an overlapping range of
unconfined compressive strengths (5 to 25 and 14 to 40 MPa),
the mi values at each site are noticeably different (Figure 2).
The inclusion or exclusion of tensile results has a variable
influence on results. In the main, the mi values are reason-
ably consistent whereas others have greater differences, more
noticeably associated with higher mi values.
In most cases, there is reasonable correspondence between
the values of R and mi , reflecting good fits between testing and
calculated data from the full suite of triaxial tests, in particular
Figure 1. Plot of laboratory compressive to tensile strength ratio where the test and calculated tensile values are similar. The R
with Hoek-Brown constant mi using data from Douglas (2002). value is considered to give a good indication of the value of
the Hoek-Brown parameter mi .
4 CONCLUSIONS
Conclusions from our investigation of intact rock mate-

rial databases, combined with further laboratory testing, to
determine the Hoek-Brown constant mi are:
1 The simple relationship between mi and rock type as envis-
aged by Hoek & Brown (1980) has not proven to be
appropriate, with the benefit of a much greater data set
than was available at that time.
2 The most accurate method of assessing mi values remains
as statistical analysis of data from a full set of laboratory
test results. These should include triaxial and unconfined
compression plus tensile tests within the recommended
stress range (i.e. σ t < σ 3 < 0.5 σ ci ).
3 In absence of a full suite of laboratory tests, R – the ratio
of unconfined compressive strength to tensile strength, is
a useful indicator of mi values, particularly as the tests for
both properties are relatively straightforward to perform.
REFERENCES
Douglas, K. J. 2002. The shear strength of rock masses. PhD the-

sis, School of Civil and Environmental Engineering, University of
New South Wales, Sydney, Australia.
Hoek, E. 2010. Rock mass properties. Ch 11 in Practical Rock
Figure 2. Principal stress plots for similar strength ignimbrites from Engineering. http://download.rocscience.com/hoek/PracticalRock
Arapuni and Maraetai. Engineering.asp.
Hoek, E, & Brown, E. T. 1980. Empirical strength criterion for rock
masses. J. Geotech. Eng. Div., Proc. Am. Soc. Civil Eng. 106(GT9):
1013–1035.
between R and mi , evenly distributed about the 1:1 correlation Hoek, E. & Brown, E.T. 1997. Practical estimates of rock mass
line for both test types (either Brazilian or direct) – Figure 1. strength. Int. J. Rock Mech. & Min. Sci. 34(8): 1165–1186.
Rocscience 2010. Product overview – RocLab, RocProp and Roc-
Data. http://www.rocscience.com/products/overview.
3 ASSESSMENT OF mi AND R WITH ADDITIONAL Sheorey, P.R. 1997. Empirical rock failure criteria. Rotterdam:
DATA SETS Balkema.
Four sets of tests have been performed on different rock

types (schist, conglomerate, two ignimbrites), including at
least two uniaxial and two Brazilian tensile tests, triaxial tests
with maximum σ 3 < 0.4σ c to 0.6σ c and calculated using the
Levenberg-Marquardt technique.
365

Inconsistencies in the determination of the compressive strength of intact rock
M. Romana
Technical University of Valencia, Spain
Keywords: lab testing, rock properties
1 INTRODUCTION 3.2 Aspect (height to diameter ratio) effect

UCS is bigger in specimens with lower aspect. The sug-
The value of the uniaxial compressive strength of intact
gested method proposes an aspect of 2.5–3.0 Mogi (2007)
rock (UCS) is the most commonly used parameter in rock
“carried out careful measurements of the apparent compres-
characterization for any purpose: strength criteria, assess-
sive strength … of Dunham dolomite, Westerly granite, and
ment of boreability and mechanical excavation capacities
Mizuho trachyte (porosity 8.5%)”. Specimens with aspect
(TBM, roadheaders, dredges). Nevertheless there are many
<2.5 increased their UCS, due to “stress concentrations near
inconsistencies in the practice when using UCS.
the end of specimen, (a) clamping effect” Relative increase was
There is a scale effect for size an aspect (relationship height-
10–15%. With bigger than 2.5 aspects the strength diminishes
diameter) The normalized laboratory test specimens must be
in very small quantities.
cylindrical, with a diameter of ∼50 mm, and an aspect value
of 2.5–3.0. Less slender cylindrical specimens increase UCS.
Cubic specimens reduce it.
UCS diminishes with saturation. Strength loss depends on: 3.3 Form effect
degree of saturation of connected pores, porosity, cementa-
tion, and water pH and salts contents. Some thumb rules can In many cases the rock samples are small and irregular (for
be gathered from existing databanks of UCS of saturated and instance when sampling at a tunnel face) and for simplic-
dry specimens for the different rock classes (Romana and ity cubic specimens are arranged for testing. This produce a
Vasarhely, 2007). double bias: the bigger samples from excavation will be the
In rock mechanics textbooks, and in several geomechanics stronger ones; besides, the cubic specimen has bigger UCS
classifications, it is stated that there is a simple relationship than a cylindrical one with aspect = 1. Therefore it should not
(Co ) between point-load strength index (ISCI ) and UCS, valid be tolerated the commercial ads on equipment capacities in
for all rocks. The alleged relationship is Co = 22–25. That is “very high strength rock”, that don’t declare size and aspect.
contrary to laboratory evidences. Co can be bigger than 25 for That increase is well known in concrete testing: (British
hard rocks and much lower for medium and soft rocks. Standards Institution, 1992). For stresses less than 50 MPa it
is very frequent that concrete cylindrical UCS be multiplied
by 1.25, to get the equivalent cubical UCS.
2 TESTING METHODS
4 EFFECTS OF WATER CONTENT
ISRM suggested method is the basis for all national labora-
tory testing specifications. Ulusay and Hudson (2007) have The suggested method ask that the samples “be stored … in
compiled and reedited all the ISRM suggested methods. The such a way as to preserve the natural water content … and
ISRM Commission on Testing Methods published in 1979 the tested in that condition” In practice the specimens are tested
“Suggested method for determination of uniaxial compressive at air-dry condition. The increasing water content has influ-
strength of rock materials”, which to date has not been revised. ence on the test results. As a rule UCS diminishes when water
content increase, with minimum at saturation.
So some experienced engineers advise to test the rock in the
same humidity conditions in which the rock mass is going to
3 EFFECTS OF SPECIMEN SIZE AND ASPECT
stay, which is especially important in dam foundations and/or
in rockfills. Romana & Vásárhelyi (2007) gathered theoretical
3.1 Size effect
and testing data on that effect.
It is well known that mechanical tests results in rock mechan- As has been shown by Ballivy and Colin (1999) the sat-
ics clearly show a scale effect. Increasing the size of specimen uration degree has a direct influence in the crack openings,
diminishes UCS, reducing the size increase UCS. The effect a fact due to the decrease in surface energy of the crack
of diameter changes (until 200 mm) can be assessed from borders when the pore is full of water. A similar explana-
Hoek & Brown formula, but there are no agreed formulations tion is offered by Vásárhely and Ledniczky (1999): “moisture
for bigger sizes diminishes the spread of free surface energy, i.e. it facilitates
366

higher for compact rocks (n < 1%). This question has been
studied by Sabatakakis et al (2008) for all the mechanical
properties of sedimentary rocks.
Basically the point load test is a tension strength test, so
the quality of bonding between grains should be a governing
factor. In sedimentary (and in many metamorphic) rocks UCS
depends mainly on the cementation.
Romana (1999) made a revision of technical literature on
that question and stated “The actual value (Co = 22–24) could
be only accepted for high-strength igneous rocks, and some
well cemented (other rocks), but not for most of metamorphic
and sedimentary rocks. It is necessary to find experimentally
Figure 1. Bieniawski correlation between UCS and Isci . the appropriate Co value. The following tentative values of C o
could be used as a first guide”.
micro-cracks propagation by decreasing the elastic limit and • Igneous rocks (high strength) 20–25
the peak strength also”. • Volcanic rocks (Canary islands)* 10–18*
The conclusion of Romana & Vásárhelyi (2007) was: “a • Quartzite 18–24
tentative rule of thumb can be proposed for the preliminary • Gneiss 16–24
estimations” of the ratio UCSsat /UCSdry between uniaxial • Metamorphic foliated rocks
compression strength of saturated and air-dry samples: (High strength) (15)–22
Well indurated strong rocks: UCSsat /UCSdry = 0.8–0.9 (Medium/low strength 10–(17)
Cemented medium strength rocks: UCSsat /UCSdry = 0.6–0.7 • Sandstone 12–24
Soft argillaceous rocks: UCSsat /UCSdry = 0.30 • Siltstone, mudstone, greywacke 10–15
• Limestone
(Well cemented) (18)–24
5 CORRELATION WITH POINT LOAD TEST (PLT) (Poorly cemented) 12–(18)
• High porosity sedimentary rocks 5–10
The ISRM Commission on Testing Methods published in 1972 (Chalk, arreciphal limestone, marine conglom.)
the “Suggested method for determining point load strength”
and revised it in 1985. (Ulusay & Hudson 2007).
PLT is frequently used for determining UCS from correla- SOME ABRIDGED REFERENCES
tion with the PLT strength index IPLT UCS = Co IPLT .
Most of texts maintain former values proposed for Co (24, Ballivy G. & Colin J. C. (1999) “Stockage souterrain: Influence … du
Bieniawski, 1974; 22 Brook, 1985). C0 = 25 is customarily fluide sur les propriétés mécaniques .9th ICRM*. Paris. Vol. 2.
used in all the versions of the Bieniawski RMR system. Fig- British Standards Institution. (1992) “Concrete”. DD ENV 2006.
London.
ure 1 shows the original Bieniawski data. (Broch & Franklin, Broch E. (1979) “Changes in rock strength caused by water” 4th
1972; D’Andrea et al, 1965 and Brook, 1985). ICRM* Montreux.
Broch & Franklin (1972) proposed C0 = 23.7, but stated Broch E and Franklin J.A. (1972). “The point load test”. Int. J. Rock
that “it had to be prudently used, due to lack of enough Mech. Min. Sci. 9 Pp 669–697.
experimental data”. Former correlations were established with Mogi K. (2007) “Experimental rock mechanics” Taylor/ Francis.
most high strength rocks data. For instance Brook only uses 361 pp.
five values of UCS below 50 MPa. In the Bieniawski figure Romana, M., 1999. “Correlation between uniaxial compressive and
there are a big number of samples with high UCS values: point-load strengths for different rocks”. 9th ICRM*, Paris. Vol. 1.
UCS = 100–300 MPa. Co = 22–24 could only be accepted for Romana M. & Vásárhelyi B. (2007) “A discussion on the decrease of
high-strength rocks. Many authors have proposed different UCS between saturated and dry rock”. Proc. 11th ICRM*. Lisbon.
Ulusay R. & Hudson J. (2007) “Complete ISRM suggested meth-
Co values for the different rock classes (see an excellent and ods for rock characterization, testing and monitoring”. ISRM
complete compilation in Zhang, 2005). Turk. N. G.
The main factor for the scatter of Co seems to be the rock Vásárhely B. & Ledniczky K. (1999). “Influence of water-
porosity. The results of Patchik & Hatzor (2004) are porosity saturation …on mechanical properties of Sivac marble” 9th
(n) dependent: “an increase in porosity from 18% to 40% leads ICRM*. Paris. Vol. 2.
to a decrease (in the ratio value) from 18 to 8”. A fact also Zhang L. (2005) “Engineering properties of rocks”. Elsevier. 290 pp.
confirmed by Kahraman et al (2005). His Co value is much
367

Effect of heat on the mechanical properties of selected rock types – a laboratory study
C. Saiang
Luleå University of Technology, Sweden
K. Miskovsky
Luleå University of Technology, Sweden
Envix Nord AB, Sweden
Subject: Rock material and rock mass property testing (laboratory and in-situ testing)
Keywords: effect of heat, increased temperatures, rock mechanical properties, mineralogical changes, micro-cracks, micro-
cracks distribution
A laboratory study was conducted to study the effect of of the rocks at the different heat treatment temperature levels,
heat on the mechanical properties of diabase, granite and determine the micro-crack intensity within the major minerals
quartzitic schist at temperatures of 400◦ C, 750◦ C and 1100◦ C. and in the rock specimen as a whole which assisted in explain-
Unheated samples were also studied. The reasoning behind ing the reasons for the mechanical behaviour observed from
this study was to understand the effect of elevated temper- the mechanical test results. Scanning Electron Microscopy
atures on the rock mass, such as in the event of a fire in combined with Energy Dispersive Spectrometry (SEM/EDS
a rock tunnel. Other authors who have undertaken similar method) was employed to identify mineral phases with grain
studies include for example, Smith & Pells, (2007), who stud- size below the resolution limit of the polarizing microscope.
ied how heat exposed sandstone is likely to respond in a The uniaxial compression test showed that the strength of the
tunnel fire by means of laboratory and field tests, Hajpal rock specimens increased by 6% for granite to 29% for diabase
(2002), who presented results of the changes in the physi- at 400◦ C when compared to the UCS values of the unheated
cal and mechanical properties of sandstone when heat treated specimens. From 750◦ C to 1100◦ C the decay in the strength
at temperatures up to 900◦ C and Zhang et al. (2010) who was very rapid. From the microscopic analyses it was con-
conducted experimental study of thermal spalling of rock cluded that the increase in the strength of the rock specimens
blocks exposed to fire. Samples from the aforementioned rock at 400◦ C is attributed to the initial reaction of the rock forming
types were heat treated at controlled low heating rates, 0.8◦ C minerals, hence the rock specimens were less brittle but more
to 1.5◦ C per minute in an electric furnace at temperatures plastic. The rapid drop in the strength from 750◦ C to 1100◦ C
shown above. After heat treatment at the desired tempera- is attributed to the mineralogical changes, micro-cracking and
ture levels the rock samples were allowed to cool slowly to dehydration due to the loss of crystal bound water. At 1100◦ C
room temperature and then subjected to uniaxial compression the rocks were highly friable and crumbled very easily when
and Brazilian tests. The uniaxial compression was performed tested mechanically. The effect of mineralogical changes was
using a servo controlled hydraulic testing machine at a dis- obvious in diabase where the physical appearance of the sam-
placement rate of 5 µm/s. For the Brazilian tests the disk ples mimicked that of natural iron, which is believed to be
shaped specimens were loaded using a 0.5 MN capacity servo due to the alteration of pyroxene. The result was an increase
controlled Instron machine at the same rate as for the com- in strength by 29% at 400◦ C compared to the unheated spec-
pression tests. Thin sections were extracted from the heat imens. Even at 750◦ C the strength was slightly higher than
treated samples for microscopic analyses. The microscopic unheated specimens. In summary; the mechanical behaviour
analysis conducted employed polarizing microscopy com- of the rock specimens depended on the temperature level and
bined with petrographic digital image analysis technique to the mineralogical and physical changes that occurred at that
identify micro-cracks and determine the mineral compositions temperature.
368

Weathering mechanisms and mechanical property changes of Oya tuff

T. Seiki
Department of Civil Engineering, Utsunomiya University, Utsunomiya, Tochigi, Japan
A. Satoh
Kashima port and Airport office, Kanto Regional Development Bureau,Ministry of Land, Infrastructure,
Transport and Tourism, Kashima, Ibaragi, Japan
K. Kikuchi
Land Development Department, Iwate Prefecture, Morioka, Iwate, Japan
Keywords: weak rock, rock properties, lab testing, field measurements
1 INTRODUCTION test, it turned greenish into reddish on a* value at some con-

ditions which was likely to contact moisture through rainfall
Oya tuff, which is a welded rhyolitic tuff (Nakamura et al. and so on. At the period, daily maximum difference between
1981) and one of the most popular commercial and architec- maximum moisture content and minimum one was 5.8 g/m3 .
tural stone materials, is quarried at Oya town in Utsunomiya And the period of oscillating moisture content between 1 g/m3
City, Tochigi Prefecture, Japan. Fresh one presents light green and 4 g/m3 also caused colour tone change.
colour having brown clay mineral spots and enough soft to At dry and humid cyclic test, little colour tone change
manufacture stone bricks and statures. On the other hand envi- from greenish to reddish on a* value occurred. Image analy-
ronment easily influences to change strength of Oya tuff and sis shows matrix of Oya tuff samples swelling and shrinking
to exfoliate the surface. This study focuses on mechanisms of according to condition drying and wetting. And EDX which
weathering of Oya tuff and we tried to clear relation between analyzed samples among dry and humid condition and natu-
an environmental factor, moisture content and those changes. rally drying ones shows that sodium (Na) and calcium (Ca) of
This study tried to reproduce weathering phenomenon of Oya atom number density is about two times as much as that of sil-
tuff empirically and considered the mechanism. ica (Si) among them. Mineral including calcium and sodium
components are condensed and crystallized near the sample
2 WEATHERING OF OYA TUFF surface. Those components may change matrix structure with
accumulation of cycles. Thus, inferred uniaxial compression
Weathering feature of Oya tuff is categorized into three types. strength changed at the stage. And then weight of transpired
This study focuses on colour tone change and strength fluc- water tends to be larger than that of absorbed water. While Oya
tuation at surface which includes preliminary step of two tuff samples soaking into water, the water may fill unsaturated
exfoliation types. In this study, we carried out several kinds voids and become priming water. It may help to vaporize more
of measurements and observations to understand weathering amount of water that of absorbed water during drying stage.
and its mechanism through outdoor exposure test, weathering Following the drying stage, the suction acting among voids
test at constant moisture and dry and humid cyclic test with and matrices may be stronger than the previous cycle.
needle penetration test, powder X-ray diffusion (XRD) test, XRD detected goethite (α-FeO(OH)) or a kind of iron oxide
colour tone analysis on L*a*b* system and energy dispersive hydroxide which should present brownish colour on the sur-
X-ray spectroscopy (EDX) with scanning electron microscope face of Oya tuff sample exposed for about two years. And
(SEM). EDX detected iron (Fe) and surfer (S). And XRD detected
small amount of pyrite (FeS2 ) from another sample. We esti-
3 EXPERIMENTS AND DISCUSSIONS mated that goethite which was chemically reacted from pyrite
or iron and surfer components might cause one of colour tone
Starting outdoor exposure test inferred uniaxial compression changes.
strength which was estimated by needle penetration test clearly
fluctuated. Moisture content that relative humidity (RH) and
temperature at that time combine to form exceeded 8 g/m3 for 4 SUMMARY
several times after two weeks. The humidity is a threshold of
the change. Weathering test of constant humidity was carried This study focuses on strength fluctuation and colour tone
out in three rooms in temperature of 20 degrees centigrade in change by weathering of Oya tuff and those mechanisms. We
which humidity was 45% RH, 60% RH and 80% RH, respec- conducted several measurements and observations on outdoor
tively. Only sample weight definitely increased at the room in exposure test, weathering test at constant moisture and dry and
80% RH. In comparison with inferred relative humidity (IH) humid cyclic test. It clarified that moisture content which was
which was converted from moisture content with saturation an environmental factor showed threshold of weathering and
vapor content at the temperature of period, moisture content related with the occurrence mechanisms.
shifted to threshold of strength fluctuation. And the humidity
was converted into around 76% IH to 78% IH.Those moistures REFERENCES
should affect to the strength fluctuation.
XRD detected small amount of thenardite (Na2 SO4 ) and Nakamura, Y., Matsui, S. & Suzuki, A., 1981, Geology of the Oya
gypsum (CaSO4 · 2H2 O) on the samples. In same view point Area, Utsunomiya City, Bull.Fac. of Edu, Utsunomiya Univ. Sec.
of the results, colour tone of sample surface change during 2, 31, Fac. of Edu. Utsunomiya Univ.:105–116 (in Japanese).
369

Relationship between the rupture modulus and the flexural strength of

building and decorative stones of Iran
K. Shahriar, M.S. Dehkordi & H.A. Lazemi

Faculty of Mining and Metallurgical Engineering, Amirkabir University, Tehran, Iran
Candidate of PhD of Rock Mechanics, Sciences and Research University, Tehran, Iran
Islamic Azad University, Bafgh Branch, Bafgh, Iran
Keywords: modulus of rupture, flexural strength, Building stones, Granite, Calcite, Travertine
Estimation of tensile strength of stones is very important in between correlation of linear and nonlinear equations of
industry, because its measurement using direct tensile strength Calcite stone is approximately 5%. All stones of Sannandaj –
test is very difficult. 3 point flexural loading and four point Sirjan zone are classified in three groups consisting of gran-
flexural loading are two methods for measurement of flexural ite, calcite and travertine. The relationship between flexural
strength. The flexural strength is the maximum tensile stress strength and modulus of rupture in each group is considered.
on the bottom of the rock corresponding to peak load. It is cal- Relationship between flexural strength and modulus of rup-
culated from simple beam theory assuming elastic conditions ture are direct but difference between correlation of linear
throughout. ASTM C880-06 covers the procedure for deter- and nonlinear equations on travertine and Granite groups are
mining the flexural strength of stone by using a simple beam smaller than Calcite groups. All samples of Sahand – Bezman
using quarter-point loading. The flexural strength is found to zone, are classified in three groups consisting of granite, cal-
be greater than direct tensile strength. ASTM C99-87 is used cite and travertine, then relationship between flexural strength
for measurement modulus of rupture of building and deco- and modulus of rupture upon granite stones was considered
rative stones. To prevent various names of stones are used but it was not considered upon calcite and travertine stones
in different classifications, the Dunhan System for notation because samples were not enough. Relationship between both
of sedimentary stones the volcanic stones, it is considered the of the strength properties are direct and difference between
main and subsidiary materials, and also the tissue of the stones. correlation of linear and nonlinear equations of Granite is very
Based on obtained results from these tests all decorative little. Therefore, the linear equation which is simpler can be
stones in this project can be divided such as that mentioned. preferred. Taken samples from the quarries have been cut in
Based on geological map of Iran, it is divided in different cutting workshop and sent to the laboratories for tests. ASTM
nine zones such as Alborz, Central Iran, Zagros, Sannandaj – standards are considered during sampling. Based on standard
Sirjan, Mokran, Sahand – Bezman, Loot block, Eastern Iran, codes, physical and mechanical tests should be done in each
Kopedagh.The Building and decorative stones are one of those sample. All tests are carried on bases ASTM standards in lab-
important natural resources of Iran. According to recent statis- oratory of construction research center of Iran. The purpose
tics, Iran has been listed among the five important producers of this paper is considering relationship between the modulus
of building and decorative stones in the world. Approximately of rupture with the flexural strength on three types of building
all of decorative and building stones located in the following stone as Granite, Calcite and Travertine groups in Iran. Based
five zones such as Alborz, Central Iran, Zagros, Sannandaj – on results relationship between both of properties is direct
Sirjan, Sahand – Bezman. All available data of Alborz zone and difference between correlation of linear and nonlinear
is only four, so it is not enough to set equation upon this. All equations of stones is very little it is possible that is assumed
stones of Central Iran zone are classified, in two groups, gran- relationship between modulus of rupture and flexural strength
ite and calcite. Relationship between both strength on Granite of Granite, travertine and Calcite stones is linear and direct.
and Calcite groups is linear but difference between correlation Ultimately, relationship between modulus of rupture and flex-
of linear and nonlinear equations of Granite group is smaller ural strength for all samples are considered without assuming
than Calcite group. All of stones in Zagros zone are classified type of stones that it is direct and linear. The reason is usage
in Calcite stones group. In this case relationship between flex- of linear equation, because applicability of linear equations is
ural strength and modulus of rupture is direct and difference simpler than nonlinear equations.
370

Shear behavior of rock under different normal stiffness
A.K. Shrivastava
Department of Civil Engineering, Delhi Technological University, Delhi, India
K.S. Rao & Ganesh W. Rathod

Department of Civil Engineering, Indian Institute of Technology, Delhi, India
Keywords: lab testing, physical modeling, rock joints, rock mass
The shear behavior of rock joints greatly influence by the pres- affecting the shear behaviour of rock joint. The influence
ence of asperity and the normal stiffness of the surrounding of these parameters on shear strength of jointed rock has
rock mass. Correct evaluation of the shear strength of rock been studied by different researchers such as Patton (1966),
joints plays an important role in the design of deep under- Ladanyi & Archambault (1977), Barton (1973 and 1976),
ground openings, stability analysis of rock slopes, socketed Barton & Choubey (1977), Indrartna & Haque (1998, 1999),
piles in rock and anchored rock slopes. To study influence of Yang & Chiang (2000) and Weildeniya (2005). To study the
stiffness of the surrounding rock mass on the shear behav- influence of normal stiffness of the surrounding rock mass on
ior of non planar rock joints, test were performed on an the shear behavior of the rock joint, tests have been performed
automated servo controlled large scale direct shear testing on non planar joints of asperity angle of 15◦ -15◦ of 5 mm
machine on rock, which has been developed and fabricated thickness under CNS conditions at different normal stiffness.
at IIT Delhi. This equipment is capable of performing the A detailed account of shear deformation behaviour of non pla-
test to study the effect of joint roughness, scale, stiffness nar rock joint under different normal stiffness is discussed in
of the surrounding rock, shear rate, condition of the joint this paper.
i.e. unfilled joint/infilled joint, infill type, and infill thickess,
371

On the reliability of the uniaxial compressive strength obtained

from the point load index
L.O. Suarez-Burgoa
Department of Civil and Environmental Engineering, University of Brasilia, Brasilia, Brazil
Faculty of Mines, National University of Colombia, Medellin, Colombia
Keywords: general, lab testing, rock properties
In literature one can find plenty of formulas that relate the presented in a graphic which shows up to 100% of estimation
point load index (Is(50) ) with the uniaxial compressive strength error. It is finally recommended avoid the use of the point
of rock material (σci ). In this article it is shown that the uniax- load index as a unique tool to assess the uniaxial compressive
ial compressive strength of rock material obtained with these strength, even though the equipment and the method to obtain
kinds of empirical formulas may be in most of the cases unre- it give to the engineer an economic and in-hand information
liable. From 37 formulas proposed by different authors, values of σci . In contrast, it is incentivized to use directly the uniaxial
of Is(50) ranging between 1 to 20 were used as input values in compressive tests of rock material, because it can give many
order to obtain σci . The result of the response values of σci was other important and reliable values apart of the strength.
Figure 1. Variation of σci with Is(50) (Suárez-Burgoa, 2010).
372

Preliminary laboratory thermo-hydro-mechanical characterization of Opalinus clay
A.M. Tang & Y.J. Cui

Laboratoire Navier/CERMES, Ecole des Ponts ParisTech, Marne-la-Vallée, France
Keywords: lab testing; nuclear repository; weak rock
1 EXTENDED ABSTRACT
In the concept of geological radioactive waste disposal,

argillite is being considered as a possible host rock in various
countries such as France and Switzerland. This material is cho-
sen thanks to its extremely low permeability that is required
to limit the transfer of radioactive elements from the waste
package towards the environment. During the construction and
exploitation phase of the radioactive waste disposal, the host
rock will be subjected to various thermo-hydro-mechanical
loadings induced by the excavation of the tunnel or the installa-
tion of engineered barriers, by the drying process on the inner
wall of the tunnel during the excavation and the re-saturation
process by water coming from the aquifer after the closure
of the tunnel, and finally, by heat emitting from the radioac-
tive waste packages. On one hand, the behaviour of the host
Figure 1. Soil volume change during the heating-cooling test.
rock under each solicitation is important for the design of such
storage, on the other hand, the behaviour under the coupled
thermo-hydro-mechanical solicitations is also needed to be
well understood. waste disposal) on the mechanical behaviour was also tested.
In the present work, some preliminary results on the labo- A soil specimen, having a suction of 39 MPa was subjected
ratory thermo-hydro-mechanical characterization of Opalinus to a heating/cooling cycle (temperature increased from 25
clay are presented. The material is one of the argillites being to 60◦ C and then decreased to 25◦ C) under a low confin-
studied in several research projects in Europe in the context ing pressure (0.1 MPa). The results are shown in Figure 1.
of geological radioactive waste disposal. On a whole, heating induced an immediate dilation followed
A triaxial compression test was performed under saturated by a contraction when temperature is maintained constant.
conditions. It was observed that preparing a specimen for tri- The relationship between temperature and volumetric strain is
axial test (38 mm in diameter and 76 mm in height) may induce almost linear when heating from 37◦ C to 61◦ C and when cool-
certain damage on the material. In addition, when saturating ing. The volumetric thermal dilation coefficient calculated
the specimen under its in situ stress (2 MPa), a volumetric from these slopes is equal 5 × 10−4 ◦ C−1 . This coefficient
swell of 1.2% occurred The swelling process continued even is similar to that obtained previously in other studies. When
after 200 h of saturation. the temperature was maintained at 60◦ C, a decrease of the
The results on triaxial test suggest that the variability of soil volume was observed. The subsequent cooling induced
mechanical properties of Opalinus clay obtained from existing a decrease of the soil volume with the same thermal dilation
studies is related to the damage induced during the preparation coefficient obtained during the heating.The thermal behaviour
of specimens on the one hand and the saturation on the other of Opalinus clay (thermal dilation and then contraction dur-
hand. ing heating) is then similar to that usually observed on clays.
Secondly, a possible effect of heating (which would be The thermal contraction is to be linked to the release of pore
generated by the radioactive waste canister in the conditions of water.
373

Experimental evidences of dramatic embrittlement of hard rocks due to

rising confining pressure at triaxial compression
B.G. Tarasov
The University of Western Australia, Perth, Western Australia
Keywords: lab testing, rock failure, rock properties, stability analysis
The paper discusses the variation of rock brittleness with con-

fining pressure σ3 for rocks of different hardness, failed under
triaxial compression with σ1 > σ2 = σ3 . Experimental results
obtained for four intact rocks (sandstone, quartzite, Westerly
granite, and dolerite) tested within the range of confining
pressure 0 ≤ σ3 ≤ 150 MPa were analyzed.
A brittleness index k = dWr /dWe = (E – M)/M were used
for rock brittleness estimation at failure in the post-peak
region. Here dWr is the post-peak rupture energy; dWe is
the elastic energy released from the material body at the rup-
ture development; E and M are elastic and post-peak modulus.
The brittleness index k characterizes the capability of the rock
to self-sustaining failure due to elastic energy accumulated
in the material body during loading, and available for rup-
ture development in the post-peak region. In accordance with
this brittleness index the absolute brittleness corresponds to Figure 1. Variation of brittleness index k versus confining pressure
situation when E = M, dWr = 0 and k = 0. σ3 for four rocks of different hardness. Brittleness increases from
Figure shows the variation of brittleness index k with left to right. k = 0 corresponds to the absolute brittleness where the
confining pressure σ3 for the four rocks, with brittleness post-peak rupture energy dWr = 0.
increasing from left to right. The graph for sandstone shows
that increase in σ3 makes the rock less brittle. This behavior is
typical for soft and insufficiently hard rocks. For the quartzite, of intact hard rocks of low porosity and fine-grained structure
increase in σ3 within the range of 0–100 MPa makes it more have not been explored and are of a very different nature.
brittle. At greater σ3 the brittleness decreases. The graphs for A special shear rupture mechanism is proposed to explain
granite and dolerite show very severe rock embrittlement. At the observed phenomenon. In accordance with this mech-
σ3 = 75 MPa the dolerite became 250 times more brittle com- anism the embrittlement results from reduction of friction
pared with uniaxial compression (ko = −1.5; k75 = −0.006). within the rupture zone with rising confining pressure. It is
At σ3 = 100 and 150 MPa the brittleness increased signifi- caused by specific structure of shear rupture involving rotating
cantly further approaching the absolute brittleness.The dashed blocks (slabs) which can operate as hinges at shear displace-
lines indicate the expected brittleness variation for granite and ment of the rupture faces. The efficiency of this mechanism
dolerite at greater σ3 : the brittleness continues to increase until is a function of rock hardness and confining pressure. In
it reaches a maximum at some level of σ3 and than decreases most hard rocks failed within a certain range of confining
since all rocks become ductile at very high σ3 . For less hard pressure this mechanism can create transient negative shear
rocks (such as the quartzite) the mode of brittleness varia- resistance – referred to as ‘negative friction’ – which makes
tion is similar but the maximum brittleness is lower and the rocks superbrittle and failure abnormally violent.
range of confining pressure where embrittlement takes place It is supposed that elevated rock brittleness and superbrit-
is smaller. tle behavior at high confining pressures can represent a very
In general increasing brittleness of rock with increasing σ3 important factor determining the position of the sesmogenic
is a well known effect. Mechanisms of embrittlement asso- zone in the earth’s crust. The specific variation of rock brittle-
ciated with rock material compaction or pore collapse at ness with rising confining pressure (increasing in brittleness –
high confining pressure have been studied by a number of maximum – decreasing in brittleness) may be responsible for
researches. Another form of embrittlement was observed in similar variation in earthquake activity with depth: the more
studies of stick-slip phenomenon on existing faults, associated brittle the rock behavior the greater is the earthquake activity.
with the transition from stable sliding to stick-slip behavior. The paper advances our knowledge about hard rock properties
However, mechanisms responsible for the brittleness variation at highly compressed conditions and at great depths.
374

Method of determination of orthotropy ratio and principal material direction for

orthotropic rock material with diametrical compression test
T. Tsutsumi
Department of Urban, Environmental Design & Engineering, Kagoshima National College of Technology,
Kirishima, Japan
Subject: Rock material and rock mass property (laboratory and in situ)
Keywords: rock properties, numerical modeling, rock stress, problem rocks
The diametrical compression test is a simple and relatively two principal elastic axes) and the direction of the principal
inexpensive test for measuring the tensile strength of brittle elastic axis. Therefore, the orthotropy ratio and the direction
material. This test is performed by placing a disk between two of the principal elastic axis are needed to estimate the tensile
(rigid) plates and applying a diametrical compressive load. strength.
The test induces a biaxial stress state in which the stress at the In this study, the relation between the angle between the
center of the disk is compressive in the x-direction (σ x ), and maximum principal strain and the loading axis, and the angle
tensile in the y-direction (σ y ). Theoretically, for an isotropic between the principal elastic axis and the loading axis, is calcu-
material, the tensile stress reaches a maximum at a constant lated using the theoretical solution, and the results are shown
magnitude of P/(πR), where P represents the applied load as a graphical representation for the diametrical compression
and R represents the radius of the disk (Sokolnikoff, 1956) test as shown in Figure 1. The relation between the absolute
(Timoshenko, 1970). Because tensile strength is smaller than value of the ratio of the normal strain in the loading direction
compressive strength for many brittle materials, the mate- to that in the orthogonal direction at the center of the disk,
rial undergoes tensile failure first. However, for orthotropic and the angle of the principal elastic axis on the loading axis
materials, Lemmon et al. (Lemmon, 1996) and the authors is also shown in graphical form as shown in Figure 2. Fur-
(Tsutsumi, 2000) showed that the distribution of tensile stress thermore, by using the strain values obtained using Rosette
on loading radius is not same as that for isotropic materials. gages in the diametrical compression test and these graphi-
According to these researchers, the maximum tensile stress cal representations, a method of determining the orthotropy
which occurs on the loading radius depends on the orthotropy ratio and the direction of the principal elastic axis was
ratio (the ratio of eachYoung’s modulus in the directions of the shown.
Figure 1. Relation between strain ratios and direction of maximum Figure 2. Relation between the direction of principal elastic axis E1
principal strain. and the direction of maximum principal strain.
375

Influence of confinement dependent failure processes on rock mass strength at depth
Benoît Valley, Bo-Hyun Kim, Fidelis T. Suorineni, Navid Bahrani & Rob P. Bewick
MIRARCO/Geomechanics Research Center, Sudbury, Canada
Peter K. Kaiser
CEMI – Center for Excellence in Mining Innovation, Sudbury, Canada
Subject: Site investigation and field observations
Keywords: site characterization, rock properties
1 INTRODUCTION
An increasing number of mining and civil tunneling projects

are being developed at great depth (>2 km) where in situ
stresses exceed the confinement levels that are: (a) typi-
cally applied in standard laboratory tests; and (b) for which
empirical approaches for rock mass strength evaluation are
calibrated. Thus, such empirical relations can be misleading
when designing underground infrastructure at great depth.
Designing at depth, outside the calibration range of empir-
ical methods, may result in over conservative or inadequate
designs by not properly anticipating the behaviour of the rock
mass. Figure 1. Laboratory test results of ultimate strength of Carrara
Results which illustrate the effect of confinement dependent marble.
changes in failure mechanism for intact rock are presented
first. The second part of the paper develops this approach for tensile stress conditions created by sample heterogeneities
the strength of rock masses. decrease with increasing confinement (Valley et al., 2009).
The Apparent Compressive Strength (ACS) is obtained by pro-
jecting back to the ordinate a linear best fit line through data at
2 FAILURE OF INTACT ROCK high confinement (σ 3 > 0.1UCS). For the data set presented
in Fig. 1, and for most rock types evaluated to date, ACS is
The observation of the failure mode of laboratory samples between 1 and 3 times the UCS (1.6UCS on average).
under uniaxial loading (σ 3 = 0) clearly highlights the impor- In terms of the quality of fit, the sigmoid criteria is not
tance of tensile mechanisms. When confinement is increased, statistically better than the Hoek-Brown failure envelope (see
tensile crack propagation is progressively inhibited and at peak R2 results in Fig. 1), but gives an alternate insight into rock
strength, the failure is usually controlled by the coalescence of strength measurement results with significant consequences
micro-cracks forming a macroscopic shear band. However, the in terms of interpretation of laboratory results. It also leads
interpretation of test results in terms of failure criteria is not to the differentiated approach for design that considers two
consistent with this transition. The two most common crite- types of cases, inner shell and outer shell problems, presented
ria used in rock mechanics, Mohr-Coulomb and Hoek-Brown, below.
are monotonic linear or steadily curved, implying that shear
strength solely contributes to the rock’s ultimate strength.
Based on the assumption that the change of failure mech- 3 STRENGTH OF ROCK MASSES
anism with confinement should be reflected in the failure
envelope, Kaiser and Kim (2008) fitted laboratory testing The strength of rock masses is typically estimated by cou-
data with a sigmoid function. An example of such a fit is pling characteristic rock mass descriptors with empirical
presented in Fig. 1. This failure envelope is parameterized relationships, e.g., the Geological Strength Index GSI, and
with the common Uniaxial Compressive Strength (UCS) and the generalized Hoek-Brown empirical failure criterion, that
the Apparent Compressive Strength (ACS). Further parame- degrade the intact rock strength to some lower value to take
ters describe the overall slope of the failure envelope and the into account the effects of block size and fractures. The
shape of the transition. degradation factors have been evaluated empirically based on
The transition typically occurs at confinement levels less back-analyses of tunnels. Hence, it must be assumed that they
than about 1/10 of UCS, justified by the fact that tensile crack are most representative for the low confinement range (inner
propagation is inhibited at larger confinement levels and that shell behaviour). At high confinement, it is to be expected that
376

Figure 3. Proposed approach to differentiate failure envelope for
inner shell and outer shell problems.
Figure 2. Sigmoid failure envelope for rock masses with hypo- using smaller W /H ratios and still insuring the carrying capac-
thetical reduced degradation under high confinement overlain on ity of the pillar, and 2) pillars that would be assumed to be in
laboratory strength data of quartzite with variable intensity of
the yielding state based on the Hoek-Brown/GSI failure enve-
micro-defects and confinement.
lope could have a core that actually is in pre-peak conditions,
rough joints will be strongly interlocked and failure through i.e. still accumulating stress and potentially entering into a
intact rock (shearing of asperities and coalescence through burst prone state.
rock bridge failure) will be controlling the strength of the
rock mass. Based on this, we hypothesize that the degradation 4 CONCLUSIONS
factors back-analyzed from inner shell behavior underestimate
the rock mass strength in the outer shell. This is also supported The change of failure mode of rock and rock masses with
by the analysis of the laboratory test data presented in Fig. 2 increasing confinement has been recognized for a long time,
of a quartzite with variable intensity of micro-defects, that but assessing its impact on the strength of intact rock and
is considered as an analog for rock mass. These results sug- particularly of rock masses is not embedded in commonly
gest that there is significant strength increase under confined used failure criteria. In order to remedy this situation, we pro-
conditions compared to the Hoek-Brown/GSI relationships. pose to differentiate the approach for strength determination
For these reasons, we recommended to use, for design, a dif- when solving problems affected by low confinement, what
ferentiated approach to obtain the failure envelope and to we call inner shell problems, and problems controlled by high
assess potential failure modes in the inner or the outer shell. confinement or outer shell problems (Fig. 3).
Such differentiated approaches are discussed in the following
sections.
ACKNOWLEDGMENTS
3.1 Inner shell problems: tensile failure leading to spalling
This research is supported by CEMI’s industrial sponsors, the
In the inner shell, under low confinement, the failure mode Government of Ontario through its Ministry of Research and
will be dominated by tensile cracking leading to spalling and Innovation and by NSERC (Natural Sciences and Engineering
related geometric dilation processes. A number of field studies Research Council of Canada).
have shown that for brittle rocks, spalling initiates at tangential
stress levels significantly below the UCS (down to 0.4UCS).
The reason for such a difference in strength between labora- REFERENCES
tory and in situ measurements still remains uncertain as in
both cases the same mechanisms, unstable tensile crack prop- Kaiser, P. K., Kim, B. H., 2008. Rock mechanics advances for under-
agation, leads to final failure. Various explanations have been ground construction in civil engineering and mining. In: Korean
proposed, but none of them have been fully verified. Symposium of Rock Mechanics. Seoul.
Nonetheless, practical approaches to assess the degree to Kaiser, P. K., Kim, B.-H., Bewick, R. P., Valley, B., 2010. Rock mass
which a rock mass is affected by brittle processes (Suorineni strength at depth and implications for pillar design. In: Deep Min-
et al., 2009) as well as the practical implications for modeling ing 2010 – 5th international seminar on deep and high stress
(Martin et al. 1999) have been proposed. mining, Van Sint Jan, M and Potvin, Y. (eds.), Santiago, Chile.
Australian Center for Geomechanics: 463–476.
Martin, C. D., Kaiser, P. K., McCreath, D., 1999. Hoek-Brown
3.2 Outer shell problems: crack damage leading to shear parameters for predicting the depth of brittle failure around
failure tunnels. Can. Geotech. J. 36: 136–151.
Suorineni, F. T., Chinnasane, D., Kaiser, P. K., December 2009.
The degree to which strength degradation should be modi- A Procedure for Determining Rock-Type Specific Hoek-Brown
fied for outer shell problems is the focus of current research. Brittle Parameter s. Rock Mech. Rock Eng. 42 (6): 849-881.
For now, we tentatively propose to use a degradation constant Valley, B., Suorineni, F. T., Kaiser, P. K., 2010. Numerical analyses of
Cs = 50 for the confined conditions (see Fig. 3). The conse- the effect of heterogeneities on rock failure process. In: 44th U.S.
quences are most relevant for pillar design (Kaiser et al., 2010) Rock Mechanics Symposium and 5th U.S.-Canada Symposium.
and are two-fold: 1) pillar design can possibly be optimized, Salt Lake City. ARMA 10–648.
377

Nondestructive evaluation of granite building stones and relation with

key mechanical parameters
G. Vasconcelos
ISISE, Department of Civil Engineering, University of Minho, Portugal
C.A.S. Alves
CIGOVR, Department of Earth Sciences, University of Minho, Portugal
T. Miranda
C-TAC, Department of Civil Engineering, University of Minho, Portugal
1 INTRODUCTION
Simple, flexible and economical tests such as the ultrasonic

pulse velocity and Schmidt hammer tests are essentially suit-
able in the characterization of elastic and strength properties Figure 1. Loading directions considered in the mechanical
of structural materials. characterization.
Thus, the main objective of the work is to give an overview
of the results of an experimental campaign based on destruc- carried out in eight stones of the granites GM and GD* existing
tive and nondestructive evaluation of distinct types of granites in four masonry walls of an ancient masonry building. In spite
used in old masonry constructions and define as a basis for of planar anisotropy was not visible, it was decided to consider
attainment of better insight on the application of simple non- the orientation of the rock splitting directions marked with the
destructive techniques (ultrasonic pulse velocity (UPV) and help of a quarryman.
Schmidt hammer testing for in-situ conditions and to relate the The destructive evaluation of the most representative
NDT indexes with key engineering properties. Therefore it is mechanical properties of granites used in masonry buildings
intended to: (a) assessment of the sensitivity of each method (compressive and flexural strength) was carried out only on
to detect anisotropy of granites; (b) evaluation of the surface two types of granites, namely granite GN and MD in order to
roughness in the NDT indexes; (c) evaluation of the sensitivity gain experience on the compressive and flexural behavior and
of NDT to detect the weathering of granites; (d) assessment of obtain any relation with UPV. The compressive and flexural
the size effect on the UPV; (e) relate UPV with compressive tests were carried out in the three loading directions afore-
and flexural strength of granites. mentioned, see Figure 1, according to ISRM (1981) and EN
12372 (1999).
2 MATERIALS AND EXPERIMENTAL PROGRAM

3 NON-DESTRUCTIVE EVALUATION
To accomplish the research objectives, different types of
granites, representative of ancient masonry buildings, were The analysis of results obtained in the blocks with rough sur-
selected. Two types of fresh (GN and GD) and weathered faces shows that the mean rebound values varies from 67.5
granites (MD and VR) from the north region of Portugal were (GN) and 64.4 (GD) in fresh granites to 57.6 (MD) and 61.0
considered. Additionally, two types of granite stones of four (VR) in weathered granites. The difference in the rebound val-
masonry walls belonging to an ancient construction were also ues between fresh and weathered granites is reduced if sawn
considered for the nondestructive evaluation based on Schmidt surfaces are considered (from 10.2% to 6.4%), even if the
hammer technique (GM and GD*). rebound values remain higher in case of the fresh granite.
It was decided to carry out the nondestructive evaluation in The increase of the rebound values in fresh granite (GN) for
granite blocks with linear dimensions ranging from 1.2 to 2.0 sawn surface is about 15.4% and in weathered granite (MD) is
coming directly from the quarries. Distinct surface conditions 22.9%. It is also seen that in average the scatter is considerable
for the blocks were considered, namely natural (not graded) higher in case of rough surfaces and weathered granites. It may
and sawn surfaces (only GN and MD) so that the influence be concluded that the surface conditions play an important role
of the roughness of surface on the non-destructive indexes in the hardness of granites, particularly in case of more weath-
could be evaluated. Additionally, Schmidt hammer testes were ered granites. It is also observed that there is a considerable
378

Table 1. Average values of the flexural and compressive strength.
Coefficient of variation is inside brackets (in %).
Gran. Dir. fl (N/mm2 ) fc (N/mm2 )
GN 1–4 9.2 (31.6) 154.5 (20.3)

2–5 15.9 (7.2) 142.0 (14.6)
3–6 14.2 (16.3) 160.1 (16.4)
MD 1–4 8.4(13.3) 73.1(19.9)
2–5 6.9 (18.4) 72.9 (20.5)
3–6 9.6 (13.9) 77.1 (15.7)
Figure 2. Direct ultrasonic pulse velocity before after sawing the

surfaces: (a) granite GN; (b) granite MD.
Figure 3. Relation between ultrasonic pulse velocity and compres-

variation on the Schmidt rebound value from the first to the sive and flexural strength of granite.
second impact. This behavior is particularly relevant in case
of rough surfaces and weathered granites. This is related in
part to the rough contact surface at the first impact, which is granite and the granite MD is a medium to coarse gran-
smoothed in the second impact. The increase in the Schmidt ite, this result can be mainly justified by the weathering
rebound values in successive impacts can be also related to degree, which is in large extent related to the porosity of both
the degree of weathering of rocks. granites. Indeed, granite GN present low value of porosity
The comparison of the UPV for different surface conditions, (η = 0.35% − CV = 9.6%), whereas granite MD present high
namely rough (BR) and sawn (BS) and can be made based on value of porosity (η = 2.28% − CV = 13.0%).
the results shown in Figure 2. It is interesting to note that the
values of ultrasonic pulse velocity increase when the blocks
are sawn (BS), both for granite GN and granite MD. The values 5 RELATION BETWEEN ULTRASONIC PULSE
in the sawn blocks are very close to the ones obtained in the VELOCITY AND MECHANICAL PROPERTIES
plate specimens tested under flexure before to be dried at 105◦
during 24 hours (SW). The values of ultrasonic pulse velocity The relation between ultrasonic pulse velocity measured in
decrease after the specimens are dried (SD). In case of fresh laboratory specimens with the mechanical properties of com-
granite GN, the values obtained in the specimens are close to pressive and flexural strength of granites GD and GN, is shown
the ones obtained in the blocks with natural surfaces. in Figure 3. It is seen that a considerable scatter is associ-
In case of more weathered granite MD, the values in the ated to the mechanical properties, as already discussed, but
specimens sawn and dried are higher than the ones obtained is clear that the ultrasonic pulse velocity increases for the
in blocks with natural surfaces, especially in case of direction higher strength. Obviously that this relation is only indica-
perpendicular to the rift plane. tive and is part of a planned enlarged research program for
which further nondestructive and nondestructive evaluation
for other types of granites is needed. The idea for further
4 ANALYSIS OF THE MECHANICAL BEHAVIOR research is to relate the ultrasonic pulse velocity and addi-
tionally the Schmidt rebound values obtained for laboratory
The average values of the flexural and compressive strength and field conditions and evaluate the possibility of estimating
for granite GA and GD are summarized in Table 1. It is in-situ mechanical properties based on the measurements
seen that fresh granite GN presents considerable higher resis- of the in-situ non-destructive indexes for application in the
tance than more weathered granite GD. Even if the type of diagnosis of historic masonry structures and geotechnical
granite is distinct, as the granite GN is a fine to medium constructions.
379

Relation between tensile and compressive engineering properties of granites
G. Vasconcelos & P.B. Lourenço

ISISE, Department of Civil Engineering, University of Minho, Portugal
C.A.S. Alves & J. Pamplona

CIGOVR, Department of Earth Sciences, University of Minho, Portugal
T. Miranda
C-TAC, Department of Civil Engineering, University of Minho, Portugal
1 INTRODUCTION Table 1. Brief description of the selected granites
Masonry is the oldest building material that survived until Granite Petrologic description Loading directions
today, being used all over the world and being present in the
BA Fine to medium-grained Parallel to the rift plane
most impressive historical structures as an evidence of spirit of porphyritic biotite granite
enterprise of ancient cultures. Portugal is rich of monumen- GA, Fine to medium-grained, Parallel to the rift plane
tal buildings and there is a considerable amount of existing GA* with porphyritic trend, two
residential masonry buildings in rural areas. Conservation, mica granite
rehabilitation and strengthening of the built heritage and pro- RM Medium-grained biotite Parallel to the rift plane
tection of human lives are clear demands of modern societies. granite
This requires identification of deficiencies of existing struc- MC Coarse-grained porphyritic Parallel to the rift plane
tures and appropriate intervention techniques. Being masonry biotite granite
a composite material, composed by units and mortar, its behav- AF Fine to medium-grained two Parallel and perpendicular
ior necessarily depends on the mechanical properties of the mica granite to the foliation plane
MDB, Fine to medium-grained two Parallel and perpendicular
components and on the masonry bond (arrangement of the MDB* mica granite to the foliation plane
stones). PTA, Fine to medium-grained two Parallel and perpendicular
Thus, the present article aims at giving an overview of the PTA* mica granite to the foliation/rift plane
tensile and compressive behavior of different types of gran- PLA, Medium to coarse-grained Parallel and perpendicular
ites that are characteristic of ancient masonry buildings in PLA* porphyritic biotite granite to the rift plane
the northern region of Portugal. Besides the presentation of
the details of experimental testing, discussion of the complete
by its random texture, are likely to have isotropic properties,
tensile and compressive behavior based on the complete stress-
a more careful analysis is needed when visual evidence of
displacement diagrams and the main engineering properties
heterogeneities and preferential orientations of the minerals
are presented. The discussion of the variation of the elastic
are detected. The latter assumptions, which were identified
and fracture properties among the different types of granites
on some granitic lithotypes under study, led to the necessity
as a function of the weathering and internal texture (planar
of conducting mechanical tests according to different load-
anisotropy) and physical properties was made based on the
ing directions. Three orthogonal planes can be identified with
experimental results. Finally, the relation between tensile and
rock splitting planes (quarry planes) defined as planes of pre-
compressive behavior as well as between compressive and
ferred rupture. The rift plane is the plane corresponding to the
tensile main mechanical properties is presented based on a set
easiest splitting in the quarry being easily recognized by the
of statistical correlations.
quarryman since it is further associated to the plane of easiest
finishing. A brief description of the adopted different litho-
2 DESCRIPTION OF MATERIALS types is indicated in Table 1. The more weathered types of the
same granite facies are distinguished with an asterisk (*).
The granites adopted in the present work were mostly col-
lected from the Northern region of Portugal. The selection
of the granitic types was based on the mineralogical com- 3 RELATION BETWEEN TENSILE AND
position and grain size, aiming at providing a comprehensive COMPRESSIVE PROPERTIES
sample of the Portuguese granites. In addition to these criteria,
the presence of preferential orientation planes and weathering From the experimental research, it was possible to derive
condition were also taken into account. f some granitic rocks, the compressive and tensile behavior of the distinct types of
380

Figure 2. Relation between tensile strength and modulus of
elasticity.
Figure 1. Relation between tensile and compressive strength.
proposed between the tensile strength and the modulus of elas-
granites. Properties like compressive strength, fc , modulus of ticity, see Figure 2. In general, higher values of elastic stiffness
elasticity, E, tensile strength and normalized energy, GN , were are associated to higher values of the tensile strength, but it is
derived. also disturbed by the tensile strength anisotropy. If one con-
The dominant mechanism of brittle fracture in compression sider the data without granites PLA and PLA* in the direction
is basically mode I cracking, the same as in tension. Previous perpendicular to the rift plane, a significant linear correlation
studies have suggested that pre-peak microcracks in compres- would be proposed: ft = 0.000858E + 0.884, (r 2 = 0.83). The
sion are tensile cracks that nucleate from pre-existing flaws ductility index corresponding to the mode I tensile fracture can
and propagate predominantly as mode I fracture. This is the also be estimated based on the compressive strength and on
basis of the pre-peak dilatant behavior exhibited by granites. the modulus of elasticity. A reasonable nonlinear negative cor-
Therefore, it is expected that both tensile and compressive relations were found to exist between compressive strength
strength are correlated. In the present work, this possibil- and modulus of elasticity with the mode I normalized frac-
ity is investigated by fitting a statistical correlation to the ture energy. The normalized fracture energy decreases as the
experimental data obtained in direct tensile and uniaxial com- modulus of elasticity and the compressive strength increase
pressive tests. The average values of the compressive and behaving the granites in a more brittle manner. This means
tensile strength were considered. that compressive strength and modulus of elasticity are also
The correlation between tensile and compressive strength reasonable measures of the ductility of the material.
is shown in Figure 1, even if it is not very significant, which Finally, it should be stressed that according to findings of
can be attributed to the anisotropy and inhomogeneity exhib- Vasconcelos (2005), it was possible to confirm that the pre-
ited by granites under tensile loading. If the linear correlation peak compressive fracture process is, in fact, predominantly
is obtained without the values of highly anisotropic granites characterized by mode I facture, corresponding to the opening
PLA and PLA* in the direction perpendicular to the rift plane, of microcracks in the direction parallel to the applied load.
the relation between tensile strength and compressive strength In addition, the proportionality between the energy consumed
would be given by ft = 0.0416fc (r 2 = 0.73). This means that in the opening of the longitudinal cracks during compressive
compressive strength is roughly 24 times higher than tensile fracture process and the mode I fracture energy appears to be
strength. Since the values of the elastic stiffness calculated confirmed by this linear correlation.
in the tensile tests were solely indicative, a new correlation is
381

Modeling the deformation behavior of sandstone using a bonded-particle model
Meng-Chia Weng
Department of Civil and Environmental Engineering, National University of Kaohsiung, Kaohsiung
Hung-Hui Lee
Chung Cheng Institute of Technology, National Defense University, Taoyuan
Shang-Shu Zhang
Graduated Institution of Engineering Technology-Doctorial, National Taipei University of Technology, Taipei
Fu-Shu Jeng
Keywords: numerical modelling, weak rock, rock properties
1 INTRODUCTION reasonable macroscopic uniaxial compression behaviour. This

test also serves the purpose of determining BPM’s microscopic
Sedimentary rocks such as sandstone are composed of rock parameters. After the test, a 2-dimensional numerical model
fragments and mineral grains. The macroscopic mechanical was found to be capable of yielding a reasonable macroscopic
behaviour of sandstone, the uniaxial compressive strength mechanical response and was thus adopted as the tool for
(UCS) for instance, is inherently affected by the microscopic further analyses.
properties, e.g. the mineral types, porosity, bonding strength, Since the natural sandstone is composed of grains, matrix
etc. of the rock forming materials. Therefore, it has been and pore. Particles representing grains and matrix were
of interest to understand how the macroscopic mechanical accordingly introduced. One of the major difficulties simulat-
behaviour of rock is influenced by the microscopic proper- ing nature sandstone is to model sandstone with high porosity.
ties.Based on previous researches (Jeng et al., 2004; Weng Natural sandstone typically shows porosity of 10%∼25%, but
et al., 2005; Lin et al., 2005), the results, two parameters, the when using BPM to model sandstone with large porosity may
porosity n and the grain area ratio GAR were found to be the result in ”floating particles”, i.e. some particles are not sup-
key parameters. ported owing to the limited number of particles that can be
Though it has been shown that the petrographic factors do used. Therefore, in previous work (Hsieh et al., 2008), an
affect the mechanical behaviour of sandstone, the mechanism additional type of particle representing the “porous matrix”
of relative mechanical properties between grain and matrix was introduced to enable modelling high-porosity sandstones
is difficult to be studied either by laboratory experiments or without leading to unsupported, floating particles. As a result,
by petrographic analysis. Given this difficulty, a numerical the proposed model contains three types of particles rep-
model, provided that it is capable of accounting for the discrete resenting grain (GP), matrix (MP) and porous matrix (PP).
packing nature of grains and the bonding strength, can serve Moreover, there exists four kinds of contact types. This modi-
as convenient means to study the influence of microscopic fication enables us to vary the GAR and porosity n of the model
properties on the macroscopic behaviour of rock. Among the while observing the macroscopic behavior of sandstone.
existing discrete numerical models, bonded-particle model To simulate uniaxial compressive tests, the size of the
(BPM ) possesses the above-mentioned requirements, and numerical specimen is 55 mm in width, 130 mm in height.
this model is adopted in this research. Therefore, this study The typical distributions of PP, MP and GP particles are ran-
employs a numerical method based on BPM to analyze the domly distributed inside the specimen. The upper and lower
effects of above-mentioned petrographic factors on strength boundaries are selected to be frictionless rigid plates, and lat-
and deformation behaviours of sandstones. eral boundaries are unconstrained. The unconfined uniaxial
compression test is simulated by vertically moving both the
upper and the lower plates toward the center of the specimen
2 METHODOLOGY at a constant velocity.
Moreover, the deformation of sandstone is explored by
2.1 Description of distinct element model means of conventional triaxial compression tests (CTC), and
In this study, the uniaxial compression was first modelled simulated tests were carried out under confining pressure
using BPM, and it was revised iteratively until it yields ranging from 5 to 40 MPa to study the effect of confining
382

which is similar to actual behaviour of sandstone. Addition-
ally, Figure 2 illustrates the plastic strain trajectory at various
stages of shearing. It indicates that plastic strain occurs even
for a minor degree of shearing (stress ratio = 0∼0.4) and sig-
nificantly increases, by about two orders while approaching
the failure state (stress ratio = 0.8∼1). The simulated plastic
strain trajectory is also similar to that of sandstone.
4 CONCLUSION
This study explored how the microscopic factors, e.g. parti-

cles types and the pore and stiffness, affect the macroscopic
Figure 1. Variations of plastic flow angle β during shear loading.
Dilation occurs when β is greater than 90◦ . strength and deformation of sandstone. A model based on
bonded-particle method is adopted to simulate the macro-
scopic mechanical behaviour. The proposed numerical model
contains three types of particles: grain particles (GP), matrix
particles (MP), and porous matrix particles (PP). The model
has been tested for modeling sandstones and shows good
capability in simulating the mechanical behavior of sandstone.
The elasto-plastic deformation of sandstone is further
explored by means of triaxial tests. Simulation results demon-
strate the mechanical behavior of sandstone has the following
characteristics:
1) The greater GAR and porosity lead to the less cohesion,
and the higher porosity leads to the lower friction angle;
2) The elastic shear modulus increases with decreasing por-
Figure 2. Variations of plastic strain increment with different shear osity and increasing GAR;
stress levels. 3) The variations of plastic angle and plastic strain trajectory
stress. Meanwhile, two cycles of unloading and reloading are under shearing are similar to actual behaviors of sandstone,
conducted in every test for obtaining the elastic deformation, and the higher porosity leads to the more plastic strain.
which will be used to decompose the total deformation into
elastic and plastic components. REFERENCES
3 RESULTS AND DISCUSSION Erosy, A., Waller, M.D. (1995): Textural characterization of rock.
Engineering Geology 39: 123–136.
Hsieh, Y.M., Li, H.H., Huang, T.H. Jeng, F.S. (2008): Interpreta-
3.1 Strength
tions on how the macroscopic mechanical behavior of sandstone
The results indicate that the cohesion is significantly influ- affected by microscopic properties—revealed by bonded-particle
enced by both GAR and porosity. The greater GAR and model”, Engineering Geology, 99: 1–10.
porosity values will lead to the less cohesion. On the other Itasca Consulting Group Inc. (2002): PFC2D (Particle Flow Code
hand, the lower porosity induces the higher friction angle, and in 2 Dimensions), Version 3.0. Minneapolis, MN: ICG.
Jeng, F. S., Weng, M. C., Lin, M. L. and Huang,T. H. (2004): Influence
the influence of GAR on the friction angle is insignificant. of petrographic parameters on geotechnical properties of Tertiary
sandstones from Taiwan. Engineering Geology, 73: 71–91.
3.2 Characteristics of elastic deformation Lin, M. L., Jeng, F.S., Tsai, L.S., Huang, T.H. (2005): Wetting
weakening of Tertiary sandstones – microscopic mechanism.
For the elastic deformation, it can be found that the modulus Environmental Geology, 48: 265–275.
increases with decreasing porosity and increasing GAR. The Potyondy, D. O. and Cundall, P. A. (2004): A bonded-particle model
numerical results are similar with the tendency of Young’s for rock. Int. J. Rock Mech. Min. Sci., 41: 1329–64.
modulus with different GAR and porosity. Weng, M.C., Jeng, F.S., Huang, T.H., Lin, M. L. (2005): Character-
izing the deformation behavior of Tertiary sandstones. Int. J. Rock
3.3 Characteristics of plastic deformation Mech. Min. Sci., 42: 388–401.
For the plastic deformation, Figure 1 shows the variations of

plastic angle β under various shear stress ratios (ηf = τ/τf ),
383

Initiation and propagation of tensile wing cracks and anti-wing cracks

from a pre-existing open flaw under compression
L.N.Y. Wong & H.Q. Li

Keywords: numerical modelling
Various crack types have been reported to initiate from a propagation processes for rock specimens containing a single
pre-existing flaw in rocks under compression. In addition to pre-existing open flaw. The variations of pressure, Mises stress
the most commonly observed tensile wing cracks, anti-wing and yield stress with time in response to external loading in
cracks propagating in an opposite direction to that of the ten- various pre-selected elements (labeled in figure 3) located on
sile wing cracks are also observed in the field (figure 1) and in the crack paths were monitored. The numerical study revealed
the laboratory testing (Wong and Einstein, 2006, 2009) (figure that for the tensile wing crack and the anti-wing crack, the
2). Although a substantial portion of the anti-wing crack path initiated crack segment adjacent to the flaw tip was found to
was observed to be associated with tensile cracking, signs of be “tensile” and “shear” in nature respectively. The propa-
shearing such as the production of shearing dust were iden- gation of both crack types, which later curved towards the
tified adjacent to the flaw tip. The objective of this paper is vertical loading direction, was tensile in nature as shown in
to numerically study the cracking processes involved in the the numerical study.
anti-wing crack formation and compare it with those in the A detailed stress analysis for the anti-wing crack initia-
tensile wing crack formation. tion and propagation in a rock specimen containing an open
In the present study, numerical analysis was performed by straight flaw under uniaxial (vertical) compression was illus-
a dynamic analysis software by incorporating the static damp- trated in Figures 3 and 4. Three elements (#7, #33 and #50)
ing technique to achieve a quasi-static loading condition. The located along an anti-wing crack propagating path and two ele-
numerical tool was then used to model the crack initiation and ments (#13 and #30) along a tensile wing crack were selected
for mechanics analysis. In elements #7 and #33, yield stress,
Mises stress and pressure all increased with time in response
to the external loading before the occurrence of cracking. The
rate of increase of the Mises stress was higher than those of the
yield stress and pressure (Figure 4a). When the Mises stress
reached the yield stress, yielding commenced. It was then fol-
lowed by an increase of plastic strain until failure (cracking)
occurred in the element. Due to the prevalence of the posi-
tive pressure in the element, the cracking can be classified as
“shear cracking”. When the anti-wing crack propagated, the
pressure in element #50 gradually increased, while staying
Figure 1. Development of tailcracks (tensile wing cracks) in the
extensional quadrants and anticracks in the compressional quadrants.
The right-lateral sense of shearing is indicated by the central arrow
pair (Kattenhorn & Marshall, 2006).
Figure 3. Rock specimens containing a single pre-existing open

Figure 2. Snapshots of the development of an anti-wing crack from flaw under uniaxial (vertical) loading. Pressure, Mises stress and
the top of the left flaw tip as external vertical loading was gradually yield stress in pre-selected elements for tensile wing crack and
applied. anti-wing crack were monitored.
384

Figure 4. Relations between stress or pressure and time in (a) elements #7 and #33, (b) element #50. The variation of pressure, Mises stress
(MIS) and yield stress (YLD) in the respective elements are plotted.
positive. When the propagating crack tip was very close to the Wong, L.N.Y. & Einstein, HH. 2006. Fracturing behavior of pris-
element, the pressure dropped and subsequently became neg- matic specimens containing single flaws. Golden Rocks 2006, The
ative (tensile pressure) as shown in Figure 4b. The cracking 41st U.S. Symposium on Rock Mechanics (USRMS): “50 Years of
event in element #50, which was far away from the flaw, was Rock Mechanics – Landmarks and Future Challenges”, Golden,
Colorado, June 17–21, 2006.
thus “tensile” in origin. Similar detailed mechanics analysis
Wong, L.N.Y. & Einstein, HH. 2009. Systematic evaluation of
was performed along the tensile wing crack for elements #13 cracking behavior in specimens containing single flaws under uni-
and #30. The modeling results revealed that the cracking pro- axial compression. International Journal of Rock Mechanics and
cesses in these two elements, respectively adjacent to and far Mining Sciences, 46(2): 239–249.
away from the flaw tip, was “tensile” in nature.
REFERENCES
Kattenhorn, S.A. & Marshall, S.T. 2006. Fault-induced perturbed

stress fields and associated tensile and compressive deformation
at fault tips in the ice shell of Europa: implications for fault
mechanics. Journal of Structural Geology 28:2204–21.
385

Physico-mechanical and thermo-mechanical characterisation of sandstone
G. Wu
Department of Civil Engineering, Shanghai Jiao Tong University, Shanghai, P.R. China
Y. Wang & G. Swift

Civil Engineering Research Group, University of Salford, Salford, UK
J. Chen
Department of Civil Engineering, Shanghai Jiao Tong University, Shanghai, P.R. China
1 INTRODUCTION at room temperature, elastic wave propagation measurements

were again undertaken, followed by strain-controlled, uniaxial
The thermo-mechanical performance of rocks is a topic compression tests to determine specific mechanical proper-
of growing importance in geotechnical engineering. The ties. More than 70 samples were tested at temperatures ranging
physical-mechanical properties of rocks under high tem- from 20–1200◦ C.
perature are of significant relevance in many engineering The physical properties considered within the investiga-
applications, such as: deep geological repositories for heat- tion include the shape, volume, mass and density changes
generating radioactive wastes, oil/gas recovery enhancement, and the velocity of longitudinal and transverse waves through
underground gasification of coals, exploitation of geother- the samples, before and after exposure to high temperature.
mal resources and the protection of buildings against fire or The mechanical properties considered include the stress-strain
building restoration after exposure to fires. response, the uniaxial compressive strength, the modulus of
elasticity and the Poisson’s ratio.
2 TESTING PROCEDURE 3 TEST RESULTS

The current investigations concern the mechanical response Figure 1a,b shows the change in volume and mass within the
of the Jiaozou Sandstone, a common rock type found in the samples tested with respect to temperature. In addition, Figure
overburden above coal seams in the Henan Province, located 2 shows the change in velocity of longitudinal and shear waves
in northern China. Samples of the sandstone were collected with temperature.
from cores drilled perpendicular to the bedding, and a range of The ratio of longitudinal wave velocity to transverse wave
tests was carried out on carefully prepared samples of 50 mm velocity is important in understanding the relation between
diameter and 100 mm length. A large number of samples were compressional deformation and shear deformation, as well as
prepared. These samples were selected based on macroscopic the Poisson’s ratio and other dynamic properties such as the
studies of sample quality. In addition, the internal integrity dynamic modulus and the dynamic Poisson’s ratio. As part of
of these samples was investigated by measuring the propa- this investigation, the data collected were used to calculate
gation of longitudinal waves using an NM-4B ultrasonic test these properties by using established relations.
device used for non-destructive testing of non-metallic materi-
als. Those samples with similar longitudinal wave speeds were
selected for further testing. The mass and density of these sam- 4 DISCUSSION OF RESULTS
ples was then determined, and samples grouped according to
these parameters. From the results presented in the paper, the authors observe
Once samples had been assigned to a particular group, the that an increase in temperature has a marked effect on the
samples were again examined for elastic wave propagation mechanical properties of this rock. In particular, both the com-
velocity, in both the longitudinal and transverse directions. pressive strength and the static Young’s modulus reduced with
Once this had been done, the samples were placed in an the increase in temperature to which the samples had been
electronically controlled oven, within which the internal tem- exposed. The authors suggest that this could be attributed to
perature was increased at a rate of 5◦ C per minute until the thermal stresses induced by the variation in thermal expan-
target temperature was attained. The temperature was then sion rates of the minerals present within the sandstone. For
maintained for a mimimum of 2 hrs, after which the samples example, quartz has a relatively high coefficient of expan-
were allowed to cool naturally to room temperature. Once sion, compared with, say, calcite and feldspar, all of which
386

are present in this rock. Also, at temperatures of 573◦ C quartz
undergoes phase transformations (β to α) which can also lead
to volume changes. Clearly, with such a high quartz content
in the sandstone (72%), these phenomena could result in the
formation of micro-cracks that will readily propagate under
applied loading, leading to reduced strength and stiffness, as
reflected in the test results.
In addition, during heating the mineral composition of the
sandstone samples would vary; the initial mineralogical com-
position is as shown in Table 2 in the paper, but as temperatures
increased within the samples tested, some of these minerals
would undoubtedly be removed. For example, samples sub-
jected to temperatures in excess of 800◦ C no longer contained
calcite or charcoal; and samples subjected to temperatures in
excess of 1200◦ C no longer contained iron or feldsapr.
It is likely that the minerals contained within the sandstone
Figure 1a. Change in volume with respect to temperature. may be subjected to partial melting and recrystallisation as
a results of the high temperatures. It is equally plausible,
that this would have an effect on the mechanical properties
of the rock after heating. Since the velocity of elastic waves
is directly related to the porosity as well as the density of
the media through which they pass, with the decrease in den-
sity observed due to heating, and the associated increase in
pososity, it would follow that there would be a decrease in
the velocity of both longitudinal and transverse waves with
increasing temperature.
The data presented in this paper, certainly supports this
assumption. The unconfined compressive strength of the
samples varies with temperature, probably due in part to the
evaporation of moisture and hence reduction in the overall
moisture content of the rock during the heating process. It
has long been understood that an increase in moisture content
can have a negative impact on the strength of rocks, including
sandstone. It therefore follows that a reduction in moisture
Figure 1b. Change in mass with respect to temperature. content may have the opposite result. Certainly, this may con-
tribute to the trends observed in the data presented in this
paper, where the general trend indicates an increase in strength
with temperature up to 400◦ C. Beyond this temperature, other
factors, such as those mentioned in the preceding paragraphs
may dominate the overall behaviour of the samples.
The authors conclude that these are initial experimental
results that have only allowed preliminary discussions to
be formulated. As the test programme continues, more evi-
dence will allow more clearly formed conclusions to be drawn
relating to the effects of thermal loading on the mechanical
behaviour of sandstone.
Figure 2. Density changes due to temperature.
387

Development of experimental technique on seismic response of rock joints
W. Wu, J.B. Zhu, Q.B. Zhang, J.F. Mathier & J.P. Dudt
Laboratoire de mécanique des roches (LMR), École Polytechnique Fédérale de Lausanne (EPFL),
Lausanne, Switzerland
Keywords: dynamics; lab testing; physical modelling; rock joints
A split Hopkinson pressure bar (SHPB) is a dynamic experi- top of each ball contacts the polished bottom of the specimen
mental technique that applied to determine mechanical prop- to minimize the friction. The plate is cut into two identical
erties of materials at high strain rates (102 −104 s−1 ). The quadrilaterals, and the interface is used to simulate a rock
application of the SHPB technique has been an active research joint. An artificial rock joint is determined in 20 degree with
topic to estimate dynamic properties of rock material, and respect to the front end of the specimen. A granite plate, with
gradually becoming a standard dynamic testing technique in the same cross-sectional area as the specimen, is designed
rock mechanics. However, the technique has been subjected as the striker plate and immediately launched by a couple
of less attention on rock joints. One possibility is the configu- of identical springs, storing equal elastic potential energy, to
ration of long bars especially suitable for the one-dimensional impact the front end of the specimen and generate a seismic
longitudinal wave propagation. In fact, joint patterns often wave. The specimen is pre-compressed by an axial compres-
display higher variability rather than discontinuities formed sion system to make the plate subjected to coupled static and
from purely tensile stress. For a general case, non-welded rock dynamic loads. Two groups of LVDTs are arranged perpendic-
joints are commonly subjected to an obliquely incident wave, ular and parallel to the joint to measure its normal deformation
resulting in not only the reflected and transmitted longitudinal and slip displacement, respectively. The incident, reflected
waves but also the frictional slip and the shear wave generation and transmitted waves are recorded by two groups of strain
and propagation. gauge rosettes attached along the two sides of the joint, and
The study discusses modified Hopkinson bars that conceive the normal and shear stress distributions along the joint can be
the design concept, and introduces a two-dimensional rock obtained.
physical model to simulate a seismic wave across an inclined The experimental technique has some other applications,
rock joint and allow the frictional slip and P- and S-waves e.g. the mechanism of stick-slip friction under seismic trig-
propagation simultaneously. In the two-dimensional config- gering, a seismic wave across an infill rock joint, etc. For the
uration, a Dark Impala granite plate is employed and laid new experimental technique, there are a few critical issues to
over an aluminum table to avoid the self-weight effect due be studied, for instance, the pulse shaping technique that suit-
to the large-scale and easily apply the axial pre-stress load- able for the two dimensional case, the precise control of the
ing. The table surface consists of half-buried balls, and the striker plate to generate a plane wave, etc.
388

Mechanical properties of freeze-thaw and damage propagation law for rock
G.S. Yang
College of Architecture and Civil Engineering, Xi an University of Science and Technology, Xi an, Shaanxi, China
H.M. Zhang
Department of Mechanics, Xi an University of Science and Technology, Xi an, Shaanxi, China
Keywords: lab testing; rock failure; rock properties; stability analysis; weak rock
1 INTRODUCTION (ϕ100 mm ×100 mm). The freeze-thaw experiment and uni-

axial compression experiment were conducted from the two
The stability assessment and freeze-thaw disaster prevention types of rock on different freeze-thaw cycle (0, 5, 10, 20, 40,
and control of frozen rock are prominent problems of engi- 60, 100 times) using the experimental research method of ‘the
neering geology and geotechnical engineering in cold regions. temperature gradient under the open system’recommended by
Cold rock endures the freeze-thaw weathering caused by the International Association of Ground Freezing, the influ-
seasons change and diurnal cycle, rock micro-damage, frost ence of the lithology and freeze-thaw cycle on rock mechanical
fragmentation and unstable failure caused by the freeze and characteristics were studied systematically. It is shown that
melts of the pore and fracture water under the alternate temper- three freeze-thaw damage deterioration modes of the red sand-
ature change, leading to irreversible degradation of material stone and shale are spalling mode, fracture mode and crack
properties and decreasing of bearing capacity, so the freeze- mode. Freeze-thaw cycle leads to irreversible deterioration on
thaw disaster of Geotechnical Engineering occurs in cold physical and mechanical properties for rock. Through the non-
regions. Therefore, study on rock degradation mechanisms linear evolution of damage, the initial microscopic differences
of freeze-thaw damage and mechanical characteristics under within the material tend to be the macroscopic differences
action of the damage deterioration are of important theoretical in the final state, and the resistance to freeze-thaw of red
and practical significance for engineering stability assessment sandstone is poorer than that of shale. For red sandstone, the
in cold regions. compressive strength and elastic modulus decrease consider-
Currently, studies concerning the freeze-thaw properties of ably with the increase of the number of freeze-thaw cycle,
rock at home and abroad mostly concentrate on the experimen- and finally collapse completely after the 79th freeze-thaw
tal levels, paying more attention to aspect of freeze physics and cycle. For shale, the compressive strength and elastic mod-
less in mechanical properties of rock, while the study involv- ulus decrease to some extent within the initial 20 freeze-thaw
ing the damage model of freeze-thaw rock has never been cycles, but the freeze-thaw damage tends to be stable through
reported yet. Furthermore, studies on the damage propaga- 20–100 freeze-thaw cycle. The freeze-thaw damage of red
tion mechanical characteristics of rock at home and abroad sandstone is more serious, its anti-compression characteristic
focus mostly on damage mechanical characteristics under the is more sensitive to freeze-thaw cycle than that of shale.
external load effect, there is few studies covering the damage Regarding the loaded damage of rock engineering struc-
propagation mechanical characteristics of rock under the con- tures under freeze-thaw conditions in cold regions was equiv-
ditions of temperature effect and freeze-thaw cycle, and most alent to the damage under the two loading, a damage variable
are conducted from the microscopic perspective of materials. was proposed with more new meaning to freeze-thaw and
In fact, the damage propagation characteristics of rock is a loaded rock, three new conceptions were put forward, that
very complex system under freeze-thaw conditions, first it is is, freeze-thaw damage, loaded damage and total damage.
a mechanical system, so the rock mechanical properties under Considering the heterogeneity of rock at mesoscopic level,
different state of freeze-thaw cycle must be studied thoroughly, the damage evolution equation, using the concepts and meth-
which is the basis of research. At the same time, rock weath- ods of macro phenomenological damage mechanics and non-
ering is a combined result of the physical, mechanical and equilibrium statistics, was given in the form of freeze-thaw
environmental factors, the rock of the actual construction is cycle and strain. The unified damage propagation constitutive
generally subjected to combined effect of 2 or more than 2 relation of freeze-thaw and loaded rock under different freeze-
factors to study the stability of rock structure, therefore based thaw cycle was established by using the generalized theory of
on a single damage factor is often not sufficient to reflect the strain equality.
actual project situation. The evolutionary route of micro-structural damage and
On the background of rock engineering in cold induced mechanical properties for the rock material were elab-
regions, two typical rocks which are red sandstone and orated, the coupling properties of the freeze-thaw and stress
shale were processed to standard compression specimen on damage propagation was reflected, and the macro-rock
389

Figure 3. Evolution curve of freeze-thaw and loaded damage for
Figure 2. Evolution curve of freeze-thaw and loaded damage for shale.
red sandstone.
characteristics of freeze-thaw and loaded rock by the math-
coupling failure mechanism of freeze-thaw and load was ematical model is in accordance with the actual freeze-thaw
revealed from meso-damage evolution. breakage.
It is shown that the mechanical properties of freeze-thaw In this paper, Figure 2 and 3 are the theoretical curve of the
and loaded rock of engineering structures in cold regions are freeze-thaw and loaded damage evolution for red sandstone
determined by freeze-thaw factor, load factor and the coupling and shale using the measured parameters from the freeze-
effects, and the total damage of rock is exacerbated by the thaw mechanical experiment provided by authors and derived
combined effect of freeze-thaw and load with obvious nonlin- from the calculation of the damage model for freeze-thaw
ear properties, but weakened by the coupling effect. and loaded rock.
The weights of influence factors are determined by the
lithology and initial damage state of the rock, so the rock
performances different damage propagation characteristics. ACKNOWLEDGMENTS
The freeze-thaw and loaded mathematical model can char-
acterize the complex relationship among freeze-thaw, load Supported by the National Natural Science Foundation of
and damage, which consider not only the effect of low China (40872177, 50974102), Natural Science Foundation of
temperature environment, but also the mechanical character- Shaanxi (SJ08A21) and The Special Foundation of Education
istics of the rock material, meanwhile, the predicted damage Department of Shaanxi (08JK359).
390

Experimental study on crack coalescence mechanism of granite specimens

containing two surface cracks
P. Yin, R.H.C. Wong & K.T. Chau

Department of Civil and Structural Engineering, The Hong Kong Polytechnic University, China
Subject: Others (experiments on crack growth and coalescence mechanism)
Keywords: surface crack, crack coalescence, crack type, coalescence pattern
Fractures (faults, joints, cracks) which extend from the ground from 0◦ to 135◦ (45◦ increment) and the bridge length 2b
surface to a certain depth of rock mass are the common geo- (Fig. 1b) was changed from 16.5 mm (half of pre-existing sur-
logical structures and this kind of fractures is defined as face crack length) to 26 mm (about one pre-existing surface
surface fractures. Under the compression stress, crack will ini- crack length).The captured images from the CCD camera were
tiate, propagate and coalesce with the neighboring fractures. used to analyze the displacement field of the pre-existing sur-
However, unlike the 2-D fracture the sliding of the surface face cracks area using Digital Speckle Correlation Method
fracture is restricted by the rock under the surface fracture. (DSCM). The DSCM technique was used for the first time
Thus, the crack growth process and coalescence mechanism to classify the crack types and coalescence patterns. Accord-
of the surface fracture are different with the 2-D case. Up to ing to experimental results, six different crack types (Fig. 1)
now, the coalescence mechanisms are still not fully investi- were identified basing on their growth trajectories and initi-
gated. In order to understand it, granite specimens with two ation mechanisms. Three types are tensile cracks, other three
pre-existing parallel 3-D surface cracks (here and the fol- types are mixed mode (tensile and shear) cracks. Meanwhile,
lowing of this paper, “fracture” will be replaced by “crack”) five crack coalescence patterns (Fig. 2) were identified in
were tested under uniaxial compression. The pre-existing sur- accordance with the initiation mechanisms.
face crack length and angle were fixed at 33 mm and 45◦
respectively. But the bridge angle β (Fig. 1b) was varied
Figure 1. Six types of cracks observed in this study. T = tensile Figure 2. Crack coalescence patterns observed in this study.
crack, S = shear crack. T = tensile crack, S = shear crack.
391

Simulation of laboratory tested bolted rock beams
P. Yiouta-Mitra, N. Halakatevakis & A.I. Sofianos

Tunnelling Laboratory, National Technical University, Athens, Greece
Keywords: numerical modeling, rock joints, back analysis, bolts and anchors
Roof bolting has been commonly in practice for more than

half a century. However, the mechanisms for the roof bolts
supporting the roof strata of underground openings in layered
structures have not been quantitatively understood much fur-
ther than the insight provided by the centrifuge experiments
of Panek in 1956. The fast development in computer hardware
and software in recent years makes it possible to investigate
roof bolting mechanisms of layered strata more realistically
using numerical analysis techniques. It is the intension of
this research to simulate the experiments performed by Panek
so that the calibrated model may be used to further develop
quantification of the reinforcement rock bolting mechanism
of stratified rock roof.
The full size structure considered by the experiments is the
horizontally bedded roof of a long underground opening in
rock. Models were made of Indiana Limestone or Hydrostone. Figure 1. Numerical model setup.
Each miniature model was considered to approximate a verti-
cal slice through a mine roof and included one set of bolts. The
bolts did not penetrate the top thick bed, which only served Two well-known criteria are selected and their results com-
to maintain a constant total model thickness and therefore pared to the experimental data; Mohr-Coulomb and Barton-
a constant restraint on the ends of the lowest member. No Bandis. The former is linear while the second non-linear with
load was transmitted by this overlying material. Strain gauges respect to normal and shear loading.
were used to measure the outer fiber bending strain at loca- A total of 10 analyses have been performed, six of which
tions L/16, L/2 and 15L/16 on the upper and lower surfaces used the M-C failure criterion while the other four the Barton –
of the lowest model beam but only the average of these was Bandis failure criterion implemented by calculating the instan-
reported. Moreover, εL/16 was used to designate the average taneous equivalent cohesion and friction angle of the joints.
of four points due to symmetry. The test data results were In the M-C analyses, the angle of friction on the contact
reported in load-strain graphs of KwL/E vs εL/16 and εL/2 and plane was varied over the expected range for Indiana lime-
not tabulated. Therefore it is not possible to re-use these data stone, i.e. between 35◦ and 30◦ . A very low value of 20◦ was
without a certain loss of precision. also tested in order to confirm the range selection. It was seen
However, with a view to gain an understanding of the rock that, although the strains measured at L/16 for values of ϕ = 35
bolts function, Panek also conducted a series of preliminary and ϕ = 30 degrees completely bracket the experimental data,
experiments, the exact measurements of which he reported in nevertheless, the value of ϕ = 33 is not the best approximation
order to prove certain theoretical assumptions. One such set of as might be expected. This led the research towards implemen-
reported measurements has been used by the current research tation of a variable modulus of elasticity, while maintaining
to evaluate the response of the simulation. the friction angle at 33 degrees. There was however no ame-
A commercial distinct element code was used to provide lioration of the proximity of the results to the experimental
the numerical simulation of the centrifuge experiments. In a measurements.
case where the behaviour of the rockmass is determined by the In the B-B analyses, JRC was given values 4, 3 and 2 and
behaviour of its joints, it is important to include the presence the residual friction angle ϕr was given values of 28, 27 and
of joints in a simulation. The numerical model is a plane strain 25 degrees. In all cases, the JCS adopted was 62 MPa (9000
model of elastic rockmass, discontinuous at the contact planes, psi), while the normal stress varied according to bolt preten-
so that they are the only possible sources of failure. This is in sion values. The scale effect parameter was not taken into
consistence with the experimental data, since the specimens consideration since the experiments were laboratory-scaled.
were not loaded beyond a certain strain measurement that was As can be seen from the results plotted in Figure 3, very
within the elastic range of the limestone material. good approximation was obtained for JRC = 2 and ϕr = 25◦ .
392

Figure 2. M-C joint simulation.
Figure 3. Barton-Bandis joint simulation.
These values are close to the lower expected range because of the behaviour of rock joints under small normal stress and
the way that the specimens were prepared. at different levels of compression due to bolting action. It is
The Barton-Bandis joint model has been found to yield therefore selected to be used for any further simulations of
very satisfactory results. It provided good approximation of laboratory-tested bolted rock beams.
393

Empirical failure criterion for biaxially loaded granite
X. Yun
Cameco Corporation, Saskatoon, Saskatchewan, Canada
H.S. Mitri
McGill University, Montreal, Quebec, Canada
X. Yang
Luoyang Institute of Science and Technology, Luoyang, Henan Province, China
Subject: Rock material and rockmass property testing
Keywords: Lab testing, Rock properties, Rock stress, Rock failure, Rock bursts
1 INTRODUCTION
Although the stress state in the vicinity of a mine opening is

generally caused by tri-axial loading, this is not quite the sce-
nario at the boundary of the underground opening whereby
mining-induced planar stresses are caused by biaxial loading
and the stress in the direction perpendicular to the boundary
σ 3 = 0. Such kind of planar stresses can attribute to initiate and
progressively or immediately trigger surface failure phenom-
ena such as spalling, slabbing, face crush (or burst), buckling,
and out-of-plane shear failure. With mining activities exceed
the depth of 2,000 m below ground surface, the stresses sur-
rounding excavation and openings may become significant
enough to cause immediate failure, even rockburst. Thus, the
study of the strength and behaviour of rockmass under biaxial
loading is fundamental for better understanding of the failure
process and for the safe and economical design of deep mines.
This paper presents the results of a series of biaxial compres-
sive tests of granite cubes with size of 75, 100 and 125 mm. All
testing was implemented with a newly developed and patented Figure 1. Loading path.
biaxial loading apparatus. An empirical biaxial failure crite-
rion is derived from a regression analysis of test results, and employed: solid platens, brush platens (5.8 mm square steel
a new biaxial failure criterion for the jointed rock mass is brush with 0.2 mm gap between brushes) and solid platens
proposed in an analogous way to the Hoek-Brown criterion. pasted with MoS2 powder (with a very thin layer of grease on
the loading faces of all four solid platens). Also, conventional
uniaxial compressive strength tests were conducted.
2 TESTING PROGRAM
3 OBSERVATION AND FAILURE MECHANISM
Designed to be spatially symmetric and of equal stiffness in
both loading directions, the new biaxial loading apparatus con- When tested uniaxially, the cubic granite sample demonstrated
sists of two independent and perpendicular uniaxial loading spalling from all free faces prior to failure, and eventually
frames, having a capacity of 500 metric tonnes, and equipped failed suddenly and violently. The remaining central portion
with servo-controlled load and displacement systems. Load showed a conical shape having thicker slabs than those occur-
servo-controlled function was employed for the implemented ring on the outer skin. As for the samples tested biaxially,
investigations. The applied loading rate in both directions is the observed failure mechanism is characterized by a varying
1 MPa/s, except for one set of 100 mm cubes being loaded at degree of rock spalling, which is initiated at the free faces in the
a rate of 10 MPa/s. Four loading paths were employed during σ 1 -σ 2 plane when the stress level reaches approximately σ c /2
the investigation, as shown in Figure 1. in both directions. The typical failure mechanism is observed
Additionally, uniaxial compression tests were conducted in biaxial tests of samples that were subjected to loading path 1.
for each sample size. Three types of loading platens were For samples subjected to loading path 2, it is observed that the
394

Figure 3. New empirical failure criterion for granite rockmass.
Figure 2. Normalized empirical failure criterion of intact granite.
where R is the slope reduction factor cited from previous
work by Mitri et al.
higher confining pressure progressively causes more spalling
which occupies a large portion of the sample volume at fail-
ure. It is observed that the spall plates decrease in thickness c) Constant Cj :
with confinement. Cj is the curvature of the failure envelope for rockmass.
4 BIAXIAL STRENGTH FAILURE CRITERIA OF

Figure 3 demonstrates the biaxial failure criterion for
INTACT ROCK AND ROCKMASS
granite rockmass with GSI = 71 and s = 0.1 for GSI.
It should be noted that the proposed failure criterion for
Based on the test result of intact cubic granite sample, a regres-
jointed rockmass, equation offers a simple and reasonable way
sion analysis was conducted in order to determine the best fit
for reducing the UCS of the intact rock in order to estimate
polynomial representing the biaxial failure envelope for the
the rockmass strength, based on the Geological Strength Index
range of sample sizes investigated, it may be considered as
(GSI ). Any artificial factors, such as blasting damage, are
an empirical failure criterion for intact granite under biaxially
apparently not included in the criterion.
loaded condition, and written in the general form as
5 CONCLUSIONS
where Ai = 0.998, Bi = 1.873 and Ci = −1.533. Laboratory test results indicate that the UCS of cubic granite
As can be seen in Figure 2, the biaxial strength of intact samples is larger than that of cylindrical samples by as much
granite can be as much as 1.57 times the uniaxial strength as 22% for 75 mm cubes. The reduced strength of cylindrical
σ c when the confining pressure is 61% of σ c . It should be samples is mainly caused by their larger aspect ratio of 2.5,
noted however that the number of biaxial tests conducted in which in turn lessens the influence of sample end friction with
this study is too limited to warrant the use of Equation (1) as the loading platen. Biaxial tests on cubic granite samples of
an empirical failure criterion for intact granite. More tests are 75, 100 and 125 mm show that size effect is insignificant for
required to validate the coefficients Ai , Bi and Ci . the range of sample sizes investigated. This may be attributed
Combined with the Geological Strength Index (GSI ), a new to the reduced pore and flaw volume under the high confin-
biaxial failure criterion for the corresponding rockmass can ing pressure. The test results indicate that using MoS2 powder
be generally proposed as, is more effective in reducing end frictional effect than brush
platens. The results suggest that while the loading path has less
influence on biaxial strength, confinement increases biaxial
strength to its maximum at σ 2 ≈ 100 MPa. The observed fail-
ure mechanism of biaxially loaded samples is characterized
where Aj , Bj and Cj are constants for rockmass. by a varying degree of spalling initiated at the free faces in the
a) Constant Aj : σ 1 -σ2 plane when the stress level reaches approximately σ c /2
The constant is unified UCS, Aj = Ai . in both directions. The intensity and volume of spalling plates
b) Constant Bj : is found to increase with confining pressure. A formula for an
Bj is the slope of the failure envelope for rockmass. empirical failure criterion for biaxially loaded intact granite
and rockmass is derived based on test result, and combining
with GSI.
395

Experimental and numerical research on crack propagation

in rock under uniaxial compression
K. Zhang
Institute of Geotechnical Engineering, School of Civil Engineering, Dalian University of Technology, Dalian, China
Department of Civil and Environmental Engineering, The University of Tokushima, Tokushima, Japan
Q. Yang
The State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian, China
J.C. Jiang
S.B. Tian
Keywords: rock failure; lab testing; numerical modeling
1 INTRODUCTION This paper intends to contribute to the understanding of the

initiation and propagation of cracks in rock or rocklike materi-
It is an important subject of rock fracture mechanics to study als. A series of uniaxial expression experiments are carried out
the law of the crack initiation, propagation and failure in using rock-like specimens with a pre-existing flaw to study the
rock masses under compression. The experimental investiga- propagation process of cracks. The effects of angles, widths
tion, theoretical analysis and numerical modeling have been and other natural characteristics of the pre-existing cracks on
made on crack propagation in different materials in uniaxial the initiation and propagation of branch cracks are experimen-
compressive loading by many researchers (Bobet & Einstein, tally examined. A number of criterions are incorporated into
1998). the XFEM technique in order to simulate and analyze the ini-
In the previous papers, the pattern of crack propagation was tiation and propagation of branch cracks, and the results are
usually studied in rock and/or rock-like material specimens compared with those obtained from experiments.
with pre-existing cracks in conditions of uniaxial compres-
sion. Two types of branch cracks are regularly observed: wing 2 EXPERIMENTAL SYUDIES
cracks (sometimes called primary cracks), and secondary
cracks. The wing cracks that are generally considered as ten- 2.1 Specimen geometry and preparation
sile cracks appear first, initiate at the two tips of pre-existing
cracks and propagate in a curvilinear path finally towards the The specimens used in the experiments are prismatic blocks
direction of the maximum compressive stress. The secondary with dimension 240 mm high, 120 mm wide, and 20 mm thick
cracks appear later, initiate also from the tips of the same (see Fig. 1(a)). A pre-existing flaw with different lengths and
cracks, and propagate in a stable manner. This type of cracks inclinations is set up in the center of each specimen. The flaw
has been recognized by many researchers as shear cracks. in experiments keeps open during the loading procedure.
At present, there are two main approaches in describing Before loading, the side surfaces of specimen are polished
crack behavior of rock or rock-like material: one using stress with abrasive cloth in order to ensure that compression can
intensity factor (SIF) as the main fracture parameter, and the be applied perpendicularly to the upper and lower faces of
other being energy based. Sometimes these two methods are the specimens (Fig. 1(b)).
considered to be consistent with each other. The above meth-
2.2 Loading equipment and loading process
ods just focus on crack initiation, and can’t be used to analyze
the process of branch crack propagation after its initiation. Electronic all-powerful testing machine with type of CSS-
How to precisely predict the crack initiation angle, crack initi- 2205 is chosen as loading equipment in the experiments. Its
ation load, and propagation path of cracks is still a challenging maximum loading range can reach 50 kN. The experiments
topic in rock fracture mechanics. Some new criterions with are carried out under placement-control loading, with the rate
more parameters are proposed to simulate the propagation of 0.5 mm/min. During the experiments, digital camera and
process of cracks in rock and/or rock-like material (Khan & DV are used to record the whole process of crack initiation,
Khraisheh, 2000). propagation, and even failure.
396

are obtained. The main findings from the experiments are as
follows.
i) Wing cracks and secondary cracks are observed in all the
specimens.
ii) Wing cracks appear first with a very short length at the
tips of the initial crack, and then become wider and longer
with the load increasing, finally towards the direction of
the compressive stress.
iii) Initiation angle of wing cracks approximately remains
above 75◦ in many specimens for different angle α of the
pre-existing cracks.
3 NUMERICAL STUDIES
Figure 1. (a) The geometry of a specimen (left); (b) The model of
a flaw under compression (right). The Extend Finite Element Method (XFEM) originally pro-
posed in 1999, which is different from the Conventional Finite
Element Method (CFEM), is very powerful for discontinu-
ous problems in fracture mechanics. The major improvement
between XFEM and CFEM is that the mesh in XFEM is
independent of the internal geometry and physical interfaces,
so that the meshing and re-meshing difficulties after crack
propagation in discontinuous problems can be overcome.
The XFEM technique is used in this paper to simulate
the propagation of cracks in the specimens, and the obtained
results are compared with the laboratory test ones. The numer-
ical results seem to be close to the testing ones, which show
that above criterions and methods can provide a satisfactory
simulation for crack initiation and propagation.
Figure 2. Image of the propagation of wing cracks (α = 30◦ ).

4 CONCLUSION
The mechanism of crack initiation and propagation was stud-

ied using rock-like material with an oblique central crack (a
pre-existing flaw) under uniaxial expression. Two types of
branch cracks, i.e. wing cracks and secondary cracks were
found in the specimens and their propagation process was
observed and recorded. A XFEM technique was used to simu-
late the initiation and propagation of cracks in the specimens,
and the obtained results were compared with the laboratory
test ones.
ACKNOWLEDGEMENTS
This research is supported by the NSFC (Grant No. 50979013).

Figure 3. Curves of displacement vs. load for b = 2 mm. The authors are grateful for these supports.
2.3 Testing results

REFERENCES
A series of experiments were carried out on the specimens
with different parameters of b (width) and α (inclination). Bobet A. & Einstein H.H. 1998. Fracture coalescence in rock-type
Fig. 2 shows one of crack patterns observed in a pre-cracked materials under uniaxial and biaxial compression. International
specimen. A typical load-displacement curve is illustrated in Journal of Rock Mechanics, Min. Sci. Vol.35(7): 863–889.
Fig. 3. Khan, SMA. & Khraisheh, MK. 2000. Analysis of mixed mode
By the image processing, the initiation angles, initiation crack initiation angles under various loading conditions. Eng.
loads and propagation paths for different pre-existing flaws Frac. Mech. Vol.67(5): 397–419.
397

Investigation on the failure features of coal under uniaxial loading condition by

Micro X-ray Computed Tomography
Y.X. Zhao, Y.D. Jiang, W.F. Zhao & T. Wang

State Key Lab of Coal Resources and Safe Mining, China University of Mining and Technology, Beijing, China
Keywords: lab testing, rock failure; rock mass; rock properties
1 INTRODUCTION
The distribution of internal micro cleats or flaws can cause

anisotropy of material properties. Meanwhile, it is easy to
induce deformation anisotropy and strain localization under
various loading conditions. The initiation and propagation of
micro cracks in the specimen can also lead to the damage
accumulation and strength deterioration.
Recent advances in X-ray Computed Tomography (CT) Figure 1. Internal and surface images obtained by CT and SEM
make it possible to observe internal failure process in the techniques. (a) Digital Radiography (DR) of sample TS-4. (b) Typical
coal or rock specimen under various loading condition. There- CT slice. (c) Typical SEM image.
fore, great attempts have been made to understand the micro
mechanism of specimen failure under uniaxial compression
and methodologies have been proposed to analyze the fail-
ure features based on mainly medical CT datum (Karacan &
Okandan, 2001; Feng et al., 2004). While the resolution of
medical CT scanners is generally of the order of 100 µm, that
of a micro scanner is of the order of 1 µm. It is hardly to
observe micro cracking features in detail by using medical
CT scanners. So the application of advanced Micro X-ray
CT to investigate the initiation and propagation of micro
cracks in the specimen has been proved to be effective in Figure 2. X-Ray CT scanner and loading device.
failure features investigation (Takahashi, 2008). Meanwhile,
the three-dimensional quantitative analysis of degradation fea- Province. The 2.5 cm diameter and 5.0 cm length cores were
tures in the specimen can also provide us insights on the micro obtained parallel to the bedding plane. The selected coal
mechanism of coal or rock failure under uniaxial compression. sample was coking coal, with vitrinite reflectance (Ro) of 0.9–
However, quantitative analysis of degradation features in 1.13%. Cleating analysis identified that there were no visible
the specimen need to extract two- or three-dimensional infor- macro cleats. The mineral morphology in the coal was dis-
mation about the crack geometry in the compression process. tributed as streaks (see Figure 1). Figure 2 shows the X-ray
As a common practice in structural pattern description in CT scanner and associated loading device used in this study.
the geosciences, fractal analysis of crack size and frequency
distributions has been used to predict highly variable and char-
acteristic parameters such as permeability, conductivity and 3 EXPERIMENTAL RESULTS AND ANALYSIS
etc. These parameters are extracted from the log–log plots
of cracks size and scale (or size and frequency). Moreover, The initial damage for the tested sample varies and manifests
the fractal analysis of multiscale and multitemporal images the non-homogeneity of the internal micro-crack distribution.
allows attainment of statistically precise quantitative infor- As the stress reaches 20.2 MPa, the propagation, interaction
mation about retracted object structure in terms of gray grade and coalescence of multi-cracks occur. This state suggests the
statistics, from which the main features of structural patterns unstable crack growth and initiation of macroscopic failure
are obtained. The aim of this paper is to describe the cracking of the sample. The main cracks are connected and the cracks
behavior of coal under uniaxial compression based on micro width increase as the stress approaches 24.25 MPa. The fractal
X-ray CT technique and fractal analysis method. dimensions of slices vary with the spatial positions of slices
and increases with the cracking expansion and bifurcation and
decreases with coalescence of cracks. Figure 4 shows a com-
2 SPECIMEN, EXPERIMENT SYSTEM AND METHOD parison image of slice fractal dimension at two stress stages.
Table 1 shows the results calculated based on the 3D box-
A coal core was prepared from a coal block collected from counting method. Comparison analysis was carried out to
the No. 9 coal seam in Tangshan mine, Kailuan Basin, Hebei the estimated fractal dimension based on the 2D and 3D
398

Figure 3. X-ray computed tomography reconstruction of a coking coal sample TS-4 under various uniaxial pressures. (a)–(d) are the
reconstructed volumetric images at 0 MPa, 19.77 MPa, 20.38 MPa and 24.25 MPa, respectively. (e)–(f) are the orthographic cross-sections
in grey-scale of the sample at 0 MPa, 19.77 MPa, 20.38 MPa and 24.25 MPa.
4 CONCLUSIONS
The fractal dimension increases with the propagation of micro

cracks at pre-failure stage. The fractal dimension obtained by
the 3D algorithm does not approach the fractal dimension of
2D data set obtained by slice images analysis if the value is
simply added to one, but a little bit lower than the 2D frac-
tal dimension added to one. However, the difference between
the two types of fractal dimension decreases with increasing
stress.
ACKNOWLEDGEMENTS
Figure 4. Calculation results of fractal dimension. The research is financially supported by the Major State
Basic Research Development Program Fund (Grant Nos.
Table 1. Estimation results of fractal dimension based on two- and 2010CB226804), State Key Laboratory for Geomechanics
three-dimensional box-counting method.
and Deep Underground Engineering Open Research Fund
Average D of (No. SKLGDUEK1021) and Fundamental Research Funds
Stress (MPa) Slices (ADS) ADS + 1 Volumetric D for the Central Universities.
0 1.7630 2.7630 2.7158

19.77 1.7745 2.7745 2.7311 REFERENCES
20.38 1.7835 2.7835 2.7616
24.25 1.8257 2.8257 2.7976
Feng, X.T., Chen, S.L. & Zhou, H. 2004. Real-time computerized
tomography (CT) experiments on sandstone damage evolution dur-
ing triaxial compression with chemical corrosion. International
Journal of Rock Mechanics and Mining Sciences 41:181–192.
box-counting method. It is identified that the fractal dimen- Karacan, C.O. & Okandan, E. 2001. Adsoption and gas transport in
sion obtained by the 3D algorithm do not approach the fractal coal microstructure: investigation and evaluation by quantitative
dimension of 2D data set obtained by slice image analysis if X-ray Ct imaging. Fuel 80: 509–520.
the value is simply added to one. The result shows that the vol- Takahashi, M., Takemur, T., Lin, W.R., et al. 2008. Microscopic
umetric fractal dimension of cracks is a little bit lower than the visualization of rocks by micro X-ray CT under confining and
2D fractal dimension obtained by slice image analysis added pore water pressures, Chinese Journal of Rock Mechanics and
to one. Engineering 27: 2455–2462.
399

Field measurements and site investigations

A new method for visualizing stresses in rock support measures

by using a light emitting sensor
S. Akutagawa & Y. Minami

Department of Civil Engineering, University of Kobe, Hyogo, Japan
H. Yamachi
Sumitomo Mitsui Construction, Co. Ltd., Japan
T. Haba
KFC Co. Ltd., Japan
T. Hiroshima & T. Nozawa
SE Corporation, Japan
Keywords: bolts and anchors, field measurements, monitoring, rock slopes and foundations, rock support
Light Emitting Deformation Sensor is applied for ground

anchors and rock bolts. By adding additional elements to orig-
inal ground anchors and rock bolts, the new types are capable
of showing current axial force in them by color of LED that
changes according to magnitude of the force. Real time mea-
surement of these forces and simultaneous presentation on
site, leads to safer working environment since any symptom
for larger deformation can be grasped with no time delay. In
addition, simple axial force presentation by changing color of
LED might help upgrade the management efficiency of those Figure 1. Wire running through an anchor plate.
rock support measures and grasp axial force distribution in
underground structures as a whole.
As a method for the reinforcement of rock masses, quite a
number of ground anchors and rock bolts have been installed in
underground caverns or tunnels. Also, plenty more of ground
anchors and rock bolts are expected to be installed in future
construction projects. For evaluating the soundness of those
underground structures, it is an important point to grasp the
state of axial forces of ground anchors or rock bolts.
Generally, in order to measure the axial force, lift-off test
has been conducted to ground anchor and strain gauge method
has been carried out to rock bolt. However, due to the high cost Figure 2. Body of Light Emitting Ground Anchor.
of these measurement methods, it is difficult to cover all of
ground anchors or rock bolts. From the view of evaluating
the soundness of such a large underground structure, as many
ground anchors or rock bolts should be monitored as possible,
then the distribution of axial force increment/decrease should
be monitored. In addition, as these methods require some
procedures for notification the measured consequents, it takes
a while to inform those concerned about the state of axial
forces. When an unexpected deformation occurs, it might not
be fast enough to give emergency warning signal or message
to nearby people. Figure 3. Protective cover at night time.
To overcome these problems, a concept of Light Emitting
Deformation Sensor has been applied to the axial force mea- with measurement. Applying the concept of the Light Emit-
surement of ground anchor and rock bolt. This is one example ting Deformation Sensor to the axial force measurement, it
of those methods employed in a new monitoring scheme called enables to measure the axial force at a low cost and show the
“On Site Visualization” to enable monitoring and visual data result by the color of LED instantaneously. The new devise
presentation at the same time. The Light Emitting Deforma- with that concept for axial force measurement of ground
tion Sensor is able to be produced at a low cost, measure a anchor and rock bolt might be able to upgrade the construc-
relative deformation of two points by a simple mechanics, tion safety and the management efficiency of underground
and inform the result by the color of LED simultaneously structures.
403

Microseismic monitoring strategy as a key component of post-mining risk management:

Review and feedback experience over the past decade
P. Bigarré, M. Bennani, I. Contrucci & E. Klein

National Institute of Environment and Industrial Risks, France
H. Baroudi, R. Hadadou & I. Vuidart

Geoderis, France
J.P. Josien
Formely Geoderis, France, presently Rovconsult, Luxemburg
Keywords: mining, risks and hazards, monitoring, field measurements
EXTENDED ABSTRACT. In France, abandoned anthro- management strategy focusing on public safety and based on
pogenic underground excavations underlie tens of thousands permanent monitoring was introduced.
of squared kilometres, disseminated through most regions Risks related to large-scale subsidence and ground collapse,
in various geological settings, at depths ranging from a few involving usually large surface areas, conducted to the choice
meters down to hundreds of meters. A significant part of of the passive microseismic monitoring method, enabling
those ancient mines have been exploited by room and pil- down-hole monitoring of extended mining voids based on a
lar techniques, with extended, multi-level, varied geometries limited number of stations. Its proven capabilities to detect
and interlaced voids, creating uncountable specific situations. distant early warnings related to brittle failures of mine pillars
Most of the mines are ancient and were closed with no or very and their overlying rock strata is documented through many
poor considerations about long term stability, at a time when other underground industrial activities worldwide. This tech-
knowledge in rock mechanics was not as advanced as nowa- nique was extensively used in France in the deep active mines
days. But time adversity is ineluctable with ageing man-made till 2002. Specific field experiments were undertaken to assess
geostructures and key issues posed by abandoned room and its suitability to the different geological settings concerned.
pillar mines has focused an important societal attention with In the salt ore basin, one of these experiment consisted,
a series of large-scale ground collapses and subsidence phe- along with Solvay mining company means and courtesy, in the
nomena that occurred in the old iron basin of Lorraine from unprecedented field monitoring experiment of the controlled
1996 to 1999, triggering emergency evacuations and reopen- collapse of a large solution mined cavern underlying 190
ing brutally socio-economic scars let by the mine industry meters deep. This resulted in the recording of a considerable
decline after-effects. Numerous sinkholes were also reported scientific databank of measurements, including thousands
during that same period of time unveiling eventually the whole of microseismic events and numerous strain, displacement
mining industry inheritance and its potential toll for the next and pressure time series. Relevant observations and results
decades to come. provided a crucial feedback experience (Fig. 1). The major
These accidents were rapidly followed by new regulatory observed phases of the phenomena appear to fit satisfacto-
reforms introduced in 1999, through the so-called “Post- rily the scenario and alarm procedure established prior to the
mining Act”. This included the reinforcing of the last existing event.
mining companies’ obligations during and after mine clo- Between 1998 and 2008, 35 local microseismic monitor-
sure, including long term monitoring whenever needed, land ing systems were deployed. Complementary field experiments
planning constraints and last but not least the setting up were carried out to calibrate the monitoring. Procedures and
of a specific public organization to undertake the expertise seamless numerical routines applied to the data processing
and the works related to these pervasive risks. Then, a con- were enhanced, while the Cloud monitoring technology was
siderable hazard assessment work was undertaken, showing introduced to improve collaborative work.
progressively numerous densely inhabited areas with rated A decade passed, 35 monitoring systems have been
risks involving public safety issues. However, considering deployed over subsidence and ground collapse prone areas
the infrequency of catastrophic ground collapses and giv- and are being run on a routine basis. Since nor catastrophic
ing careful consideration on social, technical and economical ground collapse or subsidence phenomena occurred to date,
aspects, mitigation investment strategies based on hazard or neither false alarm has been triggered along with erroneous
vulnerability reduction, as mine voids filling or expropria- evacuation, this strategy proves to be the most relevant by
tion, appeared in most cases clearly inappropriate. Then, a risk far. The cost benefit assessment related to this deployment
404

Figure 1. 3D lateral view of a portion of the microseismic swarm (789 events as dots) captured during the last two days before the final
collapse of the 3D mesh cavern with projected delineation on surface. Dense aligned microseismicity in the stiff rock strata (horizontal bar)
appears clearly, along with failure process propagating upwards. Borehole seismic probes are pictured as spheres. Scale is approximately
500 m width × 300 m height.
has been counterchecked although qualitative socio-political

aspects are rather difficult to handle. Annual cost of the mon-
itoring may be coarsely estimated to the 1/2000 of the total
overall cost if expropriation or mine voids filling solutions
had been retained only.
As concerns public threat related to sinkhole prone areas,
visual inspection implemented at first raised eventually severe
safety and technical issues. Aerial acoustic monitoring tech-
nique of the subsurface extended mine voids showed, through
field experiments and feedback experience, very efficient
to detect and measure sound and infrasound air pressure
waves caused by roof falls, spalling and breaking of super-
ficial rocks. Innovative localizations techniques have been
developed and tested successfully (Fig. 2). Visual inspection
is nowadays progressively being completed or replaced by
acoustic monitoring.
Monitoring offers clearly a smooth transition path in the Figure 2. Numerical reconstruction of aerial ray paths in a simu-
risk management of uncertain critical situations. However, lated room and pillar layout: squares are theoretical acoustic sources,
all room and pillar mines do subside or collapse with time stars are sensors, and dots are the final source location estimates. Syn-
elapsing, risk managers and decision makers must be pre- thetic scale is 350 m × 300 m, galleries are 15 m × 15 m, pillars are
pared for this. Public awareness is all the more required. Thus, 40 m × 80 m.
long term monitoring applied to short term prediction of rapid
catastrophic phenomena is a best practise challenge, requir-
ing specific know-how, long term resources, constant attention Obviously research and development must be part of
and permanent improvement of skills and tools with clearly any long term major monitoring project in order to refine
defined roles between the different actors. Operational mon- expertise, criteria and procedures based on the permanent
itoring requires high quality hazard assessment studies, good observations. Seamless data management guarantying both
understanding of the physical conditions, and relevant fore- real time and long term accessibility are also key features
seen scenario of the instability mechanisms. This calls for a that become realistic thanks to promising cloud monitoring
periodic revision relying on a back and forth process between technologies.
the observations and the models. In fact, monitoring offers a The feedback experience acquired over the past decade may
unique means to control impact of sensitive perturbations as clearly provide a useful gain of knowledge to all organizations
mine flooding, providing relevant information to be compared in mining countries facing already or in the very future similar
with presumptive evidences and hypothesis. post-mining issues at a regional or national scale.
405

Factors dominating deterioration of clastic sedimentary rocks
Pei-Chen Chan, Syuan-Jhih Lyu & Tai-Tien Wang

Institute of Mineral Resources Engineering, National Taipei University of Technology, Taipei
Fu-Shu Jeng
Tzuu-Hsing Ueng
Department of Materials and Mineral Resources Engineering, National Taipei University of Technology, Taipei
Keywords: strength deterioration; deformation reduction, petrographic parameter
1 GENERAL INSTRUCTIONS test results for the sampled rocks in dry state are referred to as
fresh ones, which are distinguished from the deteriorated sam-
Weathered condition is one of important geological investiga- ples, perhaps after some artificial weathering process, tested
tion items for rock engineering. Conventionally, the qualitative in saturated state. In this manuscript UCS dry and Edry repre-
classification defining six weathered degrees suggested by sent the uniaxial compressive strength and Young’s modulus
ISRM is wide employed for geotechnical site evaluation. And for the rock samples, and UCS det and Edet for deteriorated
the slaking and durability test is usually adopted to evaluate samples, respectively. Two indices are defined to measure the
the rock resistance for weathering. The popular used investi- deterioration of rock samples. The RUCS , the ratio of UCS det
gation approaches focus on the existed in-situ situation and the to UCS dry , measures the strength reduction, and the RE , the
mechanical resistance to short term impact on the rock sam- ratio of Edet to Edry , measures the deformation reduction.
ples. For a rock formation and thus a rock engineering projects,
the long-term weathering potential are mostly disregarded,
perhaps due to previously limited problem caused by long-
term weathering of a rock after completion of engineering 3 STATISTICAL ANALYSES AND DISCUSSION
structure, or lacking of adequate criterion for evaluation.
Focusing on clastic sedimentary rocks, the factors affecting 3.1 Multivariate statistical analyses results
the deterioration of rock samples are first investigated in this This study conducts the multivariate statistical analysis to
study. Many factors influence mechanical characteristics of a identify major factors affecting the deterioration of clastic
rock, such as the diagenesis degree, cement type and degree sedimentary rocks. In the statistical analysis, the strength
between mineral grains, water content, and stress condition, reduction and deformation reduction indices are treated as
etc. The physical indices and petrographic compositions and dependent variables, which may be influenced by the physical
geometries of a rock are related with its uniaxial compressive indices, petrographic composition and geometry factors, and
strength. Do these factors also dominate the strength deterio- fresh strength and deformation modulus, i.e. the independent
ration and deformational softening of a rock? Which factors variables. To obtain the most important factors without double
are important ones? These issues will be distinguished in this accounted variables, correlation analyses between these vari-
study. ables are performed. Accordingly, seven variables among the
initial 11 ones are remained for multivariate statistical analy-
sis, which are n (porosity), GAR (grain area ratio), Quartz
2 ROCK SAMPLES DESCRIPTION (quartz content), Gs (specific gravity), dmean (mean grain
size), FF (form factor) and Feldspar (feldspar content).
A total of 33 clastic sedimentary rocks are sampled from Statistic forward and backward selections are utilized to
Western Foothill Range in northwest Taiwan and form Coastal distinguish the major or the minor factors between the seven
Range in east Taiwan. The rock samples, mostly lithic ones influencing the deterioration of clastic sedimentary
greywacke or quartzwack, were deposited in marine, marine- rocks. Table 1 lists the major factors obtained from of multi-
terrestrial, littoral sedimentary environments in the geological ple regression analyses. For UCS dry , the seven variables are
ages from Oligocene to Pliocene. ranked by r 2 in order as Feldspar, Quartz, n, FF, Gs, dmean and
The rock samples are tested not only in dry and fresh state GAR. A regression function with an r 2 of 0.79 can be deter-
but also in saturated or weathered state to obtain the mechan- mined. The number of major factors is further selected for
ical parameters for fresh rocks and deteriorated rocks. The multiply regression analysis by if the r 2 for the regression can
rocks may be minor weathered while sampled. However, the be significantly increased or not while the additional factor is
406

Table 1. Major factors dominating the strength and module deterioration of the rock samples.
Ranking +1 +2 +3 +4 +5 +6 +7
UCS dry Feldspar Quartz nr FF Gs dmean GARr

r2 0.39 0.49 0.55 0.62 0.67 0.78 0.79
UCS det n Feldspar dmean FF – – –
r2 0.44 0.62 0.67 0.70
RUCS GAR n – – – – –
r2 0.37 0.51
E dry Quartz n Gs GAR dmean – –
r2 0.28 0.36 0.43 0.48 0.52
E det Quartz n – – – – –
r2 0.26 0.33
RE Quartz GAR – – – – –
r2 0.20 0.28
Figure 1. Variation of RUCS with n and GAR.

Figure 2. Variation of RE with n and GAR.
considered. Usually four factors are used at most for sake of Figure 2 shows the variation of RE with n and GAR. The
simplicity. variation is scatter. Again the factors considered cannot inter-
The factors influencing the deformation reduction, as well pret the deformation reduction of clastic sedimentary rocks.
as the strength and deformation in both fresh and deterio- Further study is necessary for this issue.
rated state, are also analyzed. Note the r 2 s for regressions
of deformation characteristics and associated reduction are
always low. The r 2 for the most important factors (Quartz) 4 CONCLUSIONS
only is 0.20, and is limited increased by taking into additional
factors; implying that the deformation reduction cannot be By means of multivariate statistical analysis, the factors dom-
interpreted by the factors analyzed. inating the mechanical characteristics of clastic sedimentary
rocks in Taiwan are found. Seven factors including n (poros-
ity), GAR (grain area ratio), Quartz (quartz content), Gs
3.2 Discussion
(specific gravity), dmean (mean grain size), FF (form factor)
Based on the multivariate statistical analysis results, this study and Feldspar (feldspar content) can effectively describe the
further investigates the variation of strength deterioration with UCS dry through a statistical regression function with an r 2 of
first two major factors. Figure 1 shows the variation of RUCS 0.79. And two factors, i.e. the n and GAR can interpret the
with n and GAR. The variation can be divided into two classes RUCS (strength deterioration) of the study rock samples. The
by the dotted lines. For class I the porosity n < 14–18%, RUCS variation of RUCS with n and GAR can be divided into two
rapidly decreases as n increases, no matter the value of GAR. classes with the boundary of n ranging 14–18%. The mul-
And RUCS also decreases as GAR decreases, no matter the tivariate statistical analysis results can interpret the strength
value of n. Both n and GAR play important roles for the and associated deterioration well. Nevertheless, the statistical
strength deterioration of the clastic sedimentary rocks. For results are not good enough for deformational characteristic
class II the porosity n > 14–18%, RUCS decreases slightly as and related reduction. It is obviously that the factors domi-
n increases. Especially as GAR < 50%, the increasing of n nating the deformation reduction of clastic sedimentary rocks
seems do not change the RUCS . are different from those of strength deterioration.
407

Image analysis system for areal measurement to preserve historic masonry wall, Japan
H. Chikahisa
Innovation Center, University of Yamaguchi, Yamaguchi, Japan
R. Ohta & T. Kawahara

Department of Education, Hikari City, Yamaguchi, Japan
N. Kitayama & Y. Okamoto

Technical Research Center, Fukken Co., Ltd., Hiroshima, Japan
Keywords: field measurements, monitoring
This study examined historic masonry walls called “Kogoishi”

constructed in the seventh century AD. The stones constitut-
ing the walls have been moving discontinuously, individually
and gradually due to the looseness of the surrounding ground
caused by prolonged weathering and plant root growth.
For daily preservation management by non-experts, we
created an image analysis system that combines precise
close-range photogrammetry analysis with laser profiler mea-
surement to overcome respective weaknesses. The system
corrects the distortion of the photo image and automatically
extracts to vectorize the edges of stones from photographs
as shown in Figure 1. And the image coordinates of the vec-
tors are converted into model coordinates by using the results
of laser profiler measurement. Therefore, this system allows Figure 1. An ortho photograph and vectorized edges of a masonry
us to compute the movement of the stones easily from sev- wall created from a photograph and LP measurement by the system.
eral photographs taken with time. For example, the movement
of stone and aperture distance between adjacent stones can be
computed from the results on a computer display by the image
measurement system as shown in Figure 2. This paper intro-
duces the developed image analysis system and its application
to the masonry walls.
Because the stones move individually and discontinu-
ously, the areal measurement method demonstrated here
yields useful information which is effective for understanding
the stability of the walls as well as for selecting counter-
measures. Furthermore, our evaluation of this system shows
the following:
• The areal measurement is effective for monitoring the dis-
placement behavior of the stones which are discontinuous
and local. Figure 2. Example of image measurement for masonry wall on PC
• The image analysis technique using photographs is use- display.
ful for daily preservation management by local residents
unfamiliar with civil engineering.
• The image analysis that combines photogrammetry analysis
with laser profiler measurement easily computes three-
dimensional model coordinates of stone edges and informs
overall movement of the walls.
• Visualization of the movement of masonry walls is
very effective for gaining the informed consent of local
inhabitants.
408

Property evaluation of beachrock based on field investigation in

Okinawa Island, Japan
T. Danjo
Graduate School of Engineering, Hokkaido University, Sapporo, Hokkaido, Japan
S. Kawasaki
Keywords: field measurements, lab testing, case studies, rock properties
Beachrock is a coastal deposit that is cemented by cal- this figure, the relation between qu , and formative age, t, is
cium carbonate and/or silica cement in the tidal zone of expressed with the following equation:
sandy beaches in the tropical and the subtropical zones
(Fig. 1). Considering the application of man-made beachrock
to preserve submerged-looking islands, we performed a liter-
ature survey and a field investigation to understand essential From the above data, we obtained sufficient information to
beachrock characteristics. build man-made beachrock.
The methods are as follows. The literature survey was per-
formed from the viewpoints of regional distribution, formative
age, and mineral content based on 113 papers about beachrock
worldwide. The location sites were A (Atobihamabaru), M
(Maeda) and G (Gima) on Okinawa Island, Japan. In-situ tests
included P-wave velocity measurements, and Schmidt ham-
mer and Equotip hardness tests. The collected rocks from
the sites were cored and shaped. In the laboratory, we per-
formed unconfined compression tests, density measurements,
P- and S-wave velocity measurements, and elemental analyses
using EDX.
The results of the literature survey are as follows. Over
90% of beachrocks globally exists between about 40 degrees
north and southern circle. It is found that beachrock formed
from several decades to around 7350 years ago. Furthermore,
beachrock around the world differs in composition. Moreover,
the results of the field investigation are as follows. Beachrocks Figure 1. Beachrock in Atobihamabaru, Okinawa.
from various sites are mechanically and physically heteroge-
neous. For the field tests, the P-wave velocities range from
1.58 to 6.02 km/s. The Schmidt hammer rebound hardness and
the Equotip hardness range from 13 to 45, and from 127 to
456, respectively. As for the laboratory tests, wet and dry den-
sities range from 1.96 to 2.50, and from 1.57 to 2.39 g/cm3 ,
respectively. The water content differs in rocks from different
sites. For example, in sites A, M, and G the water content is
9.3–12.5%, 25.2–26.0%, and 3.4–4.2%, respectively. Uncon-
fined compression strength, qu , values are from 5.63 to
57.80 MPa. P- and S-wave velocities are from 3.00 to 5.14
and from 2.05 to 4.16 km/s, respectively. The compositions
of the beachrock are mainly Ca and C. The concentra-
tions of CaO and C were 50.23–59.65 and 35.60–46.49%,
respectively.
In wet conditions, the relation between qu and formative
age that were researched by Omoto (2005) is shown in Fig-
ure 2. It is shown that qu increases with formative age. From Figure 2. Relation between qu and formative age.
409

An engineering view of the seismic prediction during tunneling
T. Dickmann
Amberg Technologies AG, Regensdorf, Switzerland
F. Pittard
Shimizu-Nishimatsu-UEMB-IJM (SNUI) Joint Venture, Kuala Lumpur, Malaysia
Keywords: field measurements, rock properties, risks and hazards, tunneling
A major problem for the driving of a tunnel is the knowledge of noise ratio and the least restrictive conditions for recording
the geological environment and its geotechnical parameters. and processing. The body waves travel as compression or
Significant geological boundaries, which especially cut the shear waves through the ground and are being reflected at
tunnel axis very suddenly can cause serious problems and interfaces with different mechanical properties like density or
risks during tunnel advancement. elasticity. Thus, by separation of the different wave types using
Common exploration methods like exploratory drillings three-component-sensors it is possible to deduct information
or geophysical predictions from the surface can lead to about the mechanical properties of the ground such as elastic
insufficient results due to frequent changes of geological modules.
conditions, high overburden, dense infrastructures above the To perform reliable seismic measurements and thus miti-
tunnel track etc. Besides exploratory drilling from the tunnel gating risks for tunnel construction in hard rock conditions,
face, non-destructive geophysical methods can detect litho- the Tunnel Seismic Prediction (TSP) proved its efficiency in
logic heterogeneities within sufficient distances ahead of the hard rock tunnel projects worldwide. It provides an important
tunnel face. These methods need to be robust and reliable and impact on logistic optimization that the contractor himself can
shall not disrupt the tunneling process while yielding results manage and implement in the tunneling workflow.
quickly and at moderate costs. Fuerthermore it is discussed, what are remaining problems
Seismic reflection imaging is the most effective prediction raised by the engineers on-site, when dealing with the seis-
method because of its large prediction range, high resolu- mic prediction during tunneling. Various important aspects
tion and ease of application on a tunnel construction site. It on data evaluation and interpretation are responded and will
is based on evaluating elastic body waves, which are being be explaind by real data case histories of tunnelling sites.
excited by detonation charges providing the best signal to
410

Application of acoustic emission measurement on assessment of loosened zone

around underground cavern
L. Fan, H.M. Zhou, Y.H. Zhang & Z.W. Zhong

Key Laboratory of Geotechnical Mechanics and Engineering of the Ministry of Water Resources,
Keywords: field measurements; monitoring; rock caverns; rock failure
1 INTRODUCTION measurements. However, acoustic and other geophysical test

results are ambiguous and the deformation monitoring mea-
The stress state of surrounding rocks will be disturbed by the surements can only capture the variation information when
excavation of underground cavern. And loosened zones would the displacements of the surrounding rocks reach a certain
be produced by the redistribution of stress field. The range and magnitude. Especially for brittle rocks, the deformation and
grade of the loosened zone is one of the important parame- relaxation of surrounding rocks are very fast and the displace-
ters for surrounding rock stability evaluation and supporting ments are quite small. Using routine real-time monitoring
design. The research of the loosened zone has been a key prob- methods, it is very difficult to detect the damage of surround-
lem to evaluate the stability and design the protective measures ing rocks. New effective monitoring techniques need to be
of the underground cavern surrounding rock. developed.
The stress state of surrounding rock will be disturbed by the In recent years, rock AE technique, which can show the
excavation of underground cavern. And loosened zone would interior behavior of surrounding rocks based on AE signals,
be produced by the redistribution of the stress field. The range provides new impetus for loosened zone characteristics stud-
and grade of the loosened zone is one of the important param- ies. H.G. Ji analyzed the acoustic emission pressure-coupling
eters for surrounding rock stability evaluation and supporting modes in deep rocks (Ji, 2001). L.M. Dou used AE tech-
design. The research of the loosened zone has been a key ele- niques to monitor the activity of surrounding rocks in tunnel
ment in evaluating the stability and design of the protective (Dou, 2002). X.P. Lai carried on in-situ monitoring and anal-
measures for the underground cavern surrounding rocks. ysis of the main transport roadway in Linglong Gold Mine
Since the early 20th century, many domestic and foreign (Lai, 2006). M. Cai took a back analysis of the mechanical
experts have studied the theory of loose circle of surround- parameters of surrounding rocks according to the results of
ing rocks. Many important research achievements have been AE monitoring (Cai, 2007).
obtained. The main representative accomplishments in foreign However, restricted by the field observation conditions and
countries are the natural equilibrium arch theory, the wedged equipment capability of AE technical, the research and appli-
arch body theory, the loose fracture theory and Soviet Union’s cation on monitoring in real-time and quantitative analysis
discontinuous theory (Dong, 2001; Sun, 2009). Domestic of the occurrence and development of unloading and loos-
research on the broken zone of surrounding rocks began in ening of surrounding rocks have just started. In this paper,
the 1980s. Plenty of field and theoretical works have been SENSOR HIGHWAY system has been applied to monitor
carried out. For example, Z.J. Xiao and Y. Liu used the ultra- an excavation step of the experimental cavern. The occurrence,
sonic detecting technology to determine the region of the development and evolution rules of broken zones of surround-
loosened zone (Xiao, 2006; Liu, 2007). Based on the deforma- ing rocks have been discussed. The results can be taken as a
tion monitoring results of Multi-point borehole extensometer, basis for the study of rock damage mechanism and design of
X.J. Zhang analyzed the excavation of the underground surrounding rock reinforcement.
powerhouses of Xiluodu hydropower station and predicted According to the AE monitoring data, some important
the displacement of the underground cavern (Zhang, 2008). results were obtained. The main frequencies of the AE sig-
N. Li considered the loose circles of surrounding rocks as nals are in the range of 10–20 kHz. The maximum amplitude
a number of thin layers with different mechanical proper- is 37 dB. The rock damage and stress release process inside the
ties. And elastic-plastic back-analysis on loosened zones was surrounding rocks lasts about 9 hours. The range of the loos-
carried out (Li, 2006).According to the monitoring data, Back- ened zone is from 0 to 2.7 m deep to the cavern sidewall, the
analysis method was used to evaluate the loosened zone by range from 2.7 m to 5.5 m of the surrounding rock is the dam-
S.H. Ni and Z.G. Zhang (Ni, 2009; Zhang, 2009). H.Y. Zou age zone and the region outside 5.5 m from the sidewall is the
presented a mathematical model of loosened zone based on undisturbed rock zone.In each excavation step, the relaxation
acoustic data (Zou, 2010). These theories and the monitor degree of surrounding rock exposed in the last step will be
achievements have played a guiding role in the design and further intensified. And the loosened zone will be expanded
construction of underground caverns. to the deep. A preliminary damage will be aroused by this
At present, the loosened zone was obtained generally by excavation step inside the surrounding rock exposed in the
GPR, Seismic wave and Multi-point borehole extensometer next excavation step.
411

Investigation of 3D terrestrial laser scanning techniques for potential application

to rock mechanics
Q. Feng, G. Wang & K. Röshoff

Section of laser scanning, ÅF-Infrastructure AB, Stockholm, Sweden
Keywords: field measurement, site characterization, rock mass, 3D terrestrial laser scanning, monitoring
Site characterization of a rock exposure in rock mechanics is 5) More development of hardware suitable to different sit-
one of the important steps, which is required to collect the uations, such as scanning with explosion-proof, mobile
input data for further rock mechanics analysis, rock engineer- scanning by loading a scanner on the car/train etc., remote-
ing design and numerical modeling. However, in the current controlled by a robot deployment system, cave-scanning
practice, these data is still obtained by hand including by using system with the remote-control.
compass and inclinometer for fracture mapping, measurement 6) Software development focusing more on standard data for-
with a ruler, and documentation by recording information on mat and compatible to other software, and simple and
a notebook and photographing with a camera, so the quality user-friendly to the users. Especially much more devel-
and quantity of the data are sometimes unable to meet the opment of software to create the interface to exchange
requirement in rock engineering projects. scanning data with other software.
3D laser scanning becomes now more popular to be used for 7) In order to make the whole procedure simple, the new
capturing 3D digital data to documentation and measurement. development is focusing now on the key-door solution,
This technique shows the great potential to rock mechanics which develops a system including both scanning hardware
application. In order to investigate the practical possibility, and software. A typical example is a tunnel scanning sys-
ISRM has set up a national task together with Swedish national tem, TMS, developed by Amberg AG, and specially applied
group since 2007, and aims to evaluate the state-of-the-art for to tunneling projects to control the geometry of a tunnel.
the development of laser scanning techniques and its potential
In the case study, laser scanning has been tested to different
application for rock mechanics practice. Based upon the inves-
applications related to rock mechanics concerns. With the high
tigation of ISRM-Swedish national task during 2007–2011,
resolution and accuracy scanning data, position and dimension
this paper presents the current state of laser scanning tech-
control are tested to check out the tunnel dimension, monitor
niques on both hardware and software, and summarizes the
deformation of a rock mass. Fracture mapping with two differ-
project examples related to rock mechanics, and then discusses
ent methods, both semi-automatic and full-automatic methods
the needs for further development.
are tested, which is potential to improve the input data for fur-
In current market, there are different scanning systems
ther analysis and numerical modeling. Intensity laser image
developed for different applications, but for rock mechanics,
has much information to be useful for site mapping also. Water
terrestrial laser scanner (TLS) is mostly used, and grouped
leakage is detected in intensity laser image, and possible to be
as pulse-based and phase-based scanners. By reviewing of
enhanced by image processing to identify different amount
the TLS development, the current development shows the
of water leakage. Similarly, laser image has also been tested
following features and development trend:
to identify different rock types, which is useful if no color
1) Hardware have been improved in their functions, such as scanning data available.
scanning range extended from several hundred meters to 3D laser scanning can not only be taken as a new 3D survey-
6000 m for the pulse-based scanner, and up to 187 m for ing method, but also a powerful tool for documentation and
phase-based scanner. visualization. With the infra-red laser, a high resolution 3D
2) Scanning speed has much improved also, from few thou- image can be obtained from the scanning data, which is par-
sands of points per second at the beginning, and up to ticularly important for documentation in a dark situation like
now with 122 000 points/second for the pulse-based scan- in the tunnel. These 3D information both 3D model and 3D
ner, and up to 976 000 points/second for the phase-based image have the potential to improve both quality and quantity
scanner. of site mapping data in rock mechanics.
3) The scanner become more stable and robust with working However, there are still some limits with current techniques,
temperature −10–+45, smaller and light with a size of such as colour scanning still limited to a good illumination,
24 × 20 × 10 cm with 5 kg, and also easy to operate by difficult to process the large amount of scanning data with
in-built computer, internal battery and blue-tooth control. high resolution, especially no so much development for the
4) Scanning with color become more alternative by real color application software to rock mechanics. These problems can
scanner limited for scanning small objects, and possible to be solved with the further development.
scan a large object by scanning with in-built color camera
and extra photographing.
412

Interpretation of rock mass deformability measurements using large flat jack tests
B. Figueiredo, F. Bernardo, L. Lamas & J. Muralha

National Laboratory for Civil Engineering (LNEC), Lisbon, Portugal
Keywords: field measurements, numerical modelling, rock properties, site characterization
Rock mass deformability plays a significant role in the design

of large structures such as concrete dams, large bridges, under-
ground caverns and tunnels, since their behaviour depends on
the displacements undergone by the rock mass. For the design
of these types of structures, it is not adequate to characterize
rock mass deformability just using laboratory tests on intact
rock specimens, and extrapolating their results to the rock
mass based on subjective indices such as the RMR, Q or GSI
values. In situ tests involving relatively large and representa-
tive rock mass volumes can provide reliable estimates of rock
mass deformability, and allow reducing uncertainties during
design stages. Figure 1. a) Large flat jack; b) location of the displacement
Large flat jack (LFJ) tests are one of the field techniques transducers; c) ready for testing jacks in two contiguous slots.
used to assess the rock mass deformability that allow test-
ing relatively large volumes of rock mass. LFJ tests begin
with cutting a thin slot in the rock mass using a disk saw,
and inserting a flat jack. Then, hydraulic pressure in the flat
jacks applies loads to the slot walls, while rock mass defor-
mation is measured at several locations sets of displacement
transducers (Fig. 1).
Interpretation of flat jack test results is not straightforward,
due to the development of a tensile crack along the plane
of the slots as applied pressures increase. Crack initiation
and propagation depends on the applied pressure and also on
the in situ stresses and rock mass tensile strength, which are
generally unknown parameters.
To deal with this difficulty, a three-dimensional numeri-
cal model was recently developed for the interpretation of
the large flat jack test results, enabling to simulate several
cases of test geometries with two co-planar contiguous slots.
It was developed using FLAC3D software and the mesh is a
30 m × 30 m × 15 m solid with 149,440 zones. In order to sim-
ulate with detail the crack initiation and propagation into the
rock mass, a very refined mesh around the slots was required, Figure 2. Three dimensional finite difference model.
since they are just a few millimetres thick. Figure 2 (top left)
shows a view of the mesh, and Figure 2 (top right) displays a
middle vertical cross-section. The dark zones at the top are the of LFJ tests: two loaded flat jacks in two flat jacks; one loaded
refined mesh, which is presented magnified at the bottom. The jack in two slots, and one flat jack in one slot. These calcula-
model allows simulating an actual large flat jack test protocol tions allowed obtaining the 3D surface that relates the tensile
with several loading and unloading cycles (Figueiredo et al. crack depth with the pressure and the tensile strength, both
2011), and can also be used for estimating the initial in situ divided by the initial stress normal to the slot (Fig. 3). The
stresses perpendicular to the slots (Figueiredo et al. 2010). influence of the tensile crack on the test results is assessed
Due to the importance of the tensile crack depth for the and discussed.
LFJ tests interpretation, a methodology that relates the depths Figure 4 shows diagrams of the normal stresses on the inter-
with the in situ stresses normal to the slots, the applied pres- face for a null value of the rock mass tensile strength, where the
sures, and the tensile strength of the rock mass, was developed. extent of the tensile crack around the slots for three different
Several calculations were performed for three common cases situations can be clearly seen.
413

Figure 5. Variation of the coefficients ki with the tensile crack depth.
Case A: two slots and two flat jacks.
Figure 3. Three-dimensional surface relating crack depth with the
pressure and the tensile strength (two loaded slots). deformability, at location i, for any given tensile crack depth
(Fig. 5), using the equation:
This numerical model presents several important advan-

tages. It enables to determine in situ stresses perpendicular
to the slots and to model rock mass deformation during the
loading and unloading cycles of the LFJ tests. In this paper,
a methodology for determining the tensile crack depth, a cru-
cial parameter for the correct interpretation of the tests, is
presented. Several computations were performed in order to
determine the coefficients that allow calculating the rock mass
deformability modulus for several common geometries and
loading conditions of the tests.
Studies are now being carried out that consider different
situations as regards the inclination of the slots with respect to
the rock face and different bounday conditions that simulate
different test chamber geometries.
REFERENCES
Bernardo, F. 2011. Numerical modelling of LFJ tests for the deter-

mination of rock mass deformability (in Portuguese). MSc thesis,
Figure 4. Normal stresses on the interface.
New University of Lisbon.
Figueiredo, B., Lamas, L. & Muralha, J. 2010. Determination of
in situ stresses using large flat jack tests. ISRM International
The numerical model was also used for calculating, for each Symposium 2010 and 6th Asian Rock Mechanics Symposium,
pressure increment, the tensile crack depth and the displace- New Delhi.
ments at the locations of all displacement gauges (Bernardo Figueiredo, B., Lamas, L. & Muralha, J. 2011. Numerical simulation
2011). These results enabled to obtain the values of the coeffi- of a large flat jack test with cyclic loading. Proceedings of the 6th
cients ki that are needed to calculate the rock mass modulus of International Conference on Dam Engineering, LNEC, Lisbon.
414

Abandoned roadways aged up to 50 years observed in Kushiro Coal Mine, Japan
Y. Fujii, T. Sugawara & J. Kodama

Faculty of Engineering, Hokkaido University, Sapporo, Japan
Y. Ishijima, T. Kiyama & M. Takada

Horonobe RISE (Horonobe Research Institute for the Subsurface Environment), Horonobe, Japan
Y. Ichihara
Kushiro Coal Mine, Kushiro, Japan
S. Kumakura
Docon, Sapporo, Japan
T. Narita
AIST (Advanced Industrial Science and Technology), Tsukuba, Japan
M. Sawada & E. Nakata

CRIEPI (Central Research Institute of Electric Power Industry), Abiko, Japan
Subject: Site investigation and field observations.
Keywords: case studies, coal mines, field measurements, lab testing, monitoring, nuclear repository
Excavation Damaged Zones (EDZs) appear around deep rock and 1.5 m from the deformability distribution along a bore-
repositories for geological disposal. Permeability in EDZs hole in the sidewall. In the triaxial compression test on roof
would be much higher than in intact rock masses, and the sandstone from the site, permeability increased with speci-
sealability of a rock cavern is supposed to be significantly men failure. It gradually decreased with time in the residual
compromised by high permeability. Mechanical properties of state and reached almost the same level as that for the intact
medium-hard clastic rocks in EDZs are unknown. To clarify rock specimen. Permeability decreased with the increase in
the sealability of abandoned and closed rock caverns in clastic confining pressure and then increased with its decrease. Frac-
rocks, old roadways were observed at 95 sites during the min- ture closures and poorly connected tortuous fractures were
ing of longwall panels in Kushiro Coal Mine. Rayleigh wave seen in the micrograph of the thin section. The sealabil-
velocities and permeabilities in abandoned and closed road- ity which is recovered to approx. 1/40 the virgin rock mass
ways as old as 50 years, which can be regarded as very severe would be much better than rock cavern in crystalline rocks
EDZs with zero initial sealability, and those in EDZs and in which much less recovery in sealability is expected when
EdZs (Excavation disturbed Zones) around recently excavated collapsed.
roadways in clastic rocks as deep as 300 m were measured
and compared. Most roadways were completely closed by
deflection of roof (Fig. 1) and/or floor heave due to plastic
deformation of clay and siliceous rock. There observed no
significant fractures in the closed parts. On the other hand,
a few roadways were left open by the wooden support. No
back filling was reached to the open roadways. Rayleigh
wave velocity in the closed roadway EDZ and EDZ around
the present roadway are in the range of 0.3–1.2 km/s and
1.1–1.8 km/s, respectively. Permeability in the closed road-
way EdZ, closed roadway EDZ, EdZ and EDZ around the
present roadway were 5.1 to 6.0 × 10−14 , 2 to 10 × 10−13 ,
1.4 × 10−15 and 5.3 × 10−15 m2 , respectively. It is estimated
that the rock mass within 4 m from the sidewall was an EDZ Figure 1. Old roadway observed in Kushiro Coal Mine. Roadway
from permeability distribution along a borehole in roof sand- which is denoted by the broken line was closed by the deflected and
stone. The EDZ boundary was estimated to be between 1 m sagged roof. Height of old roadways is approx. 2.5 m.
415

The role of instrumentation and monitoring system on the stability of a slope
M. Gharouni Nik
School of Railway Engineering, Iran University of Science & Technology (IUST), Iran
Subject: Slope stability and Monitoring.

Keywords: Instrumentation, Monitoring, Slope stability
Monitoring of slope movements have been considered by points of view, as well as generically of the types of equipment
researchers, since the investigation of effective factors on the and instrumentation that are available. Legal issues relevant to
start, continuation and stopping of natural and artificial slides slope monitoring have also been dealt with. From the recorded
movement as a result of various internal factors such as slope results of monitoring and detected effects of instrumentation
proportional to inherent soil conditions and the type of its in controlling the slope, one can easily show that conventional
components and also stimulant external factors such as dis- geotechnical instrumentation and survey will continue to have
ruption of forces equilibrium due to the forces of loading and their place in slope monitoring.
unloading, which practically may not be modeled easily in The inclinometer data indicated movement along distinct
laboratory conditions. shear planes whenever there was a built up in water pressure.
Principally monitoring systems are set of instruments which However, slope movement generally ceased upon completion
are capable of recording the variable parameters existing in the of blasting and excavation operations, although some incli-
nature in terms of time, and provide it in order to be used in nometers have shown several cycles of minor movement and
probable events analysis. Some of these variables are natural subsequent stabilization attributed to drained water. As the
and some others are the results of the changes made in nature instrumentation showed some changes in displacements and
due to natural changes taking place. pore pressure, specially after the construction of the slope,
The main magnitudes measured in monitoring are displace- countermeasures consisting of drainage system, rock anchors,
ments, stress variation, water pressure, load exerted on rock meshes and shotcrete against the slope surface were employed
mass and support systems. Measurement of displacements and to arrest slope movements and coming to a standstill. There-
deformations of rock mass, and support systems are the most fore, the use of an instrumentation program for this part of the
effective tools for investigation and inspection of a geotechni- project allowed the owner to make a secure condition for both
cal structure. These tools may have different types according personnel and equipment while there was no need to evacuate
to their application. the area.
In this research, the factors affecting the instability of In summary, instrumentation on its own to some extent can
slopes, the solutions for stabilization of slopes, behavioral predict the failure of the slopes. However, together instru-
variation of sliding, the necessity of behavior measurement mentation and finite element analysis can be used to assess
of slopes movements and investigation of the various tools if the work has the conditions required for a failure to occur.
related to the slopes are briefly introduced and their perfor- Once this has been established the most effective method of
mance and also the method of measurement and the general detecting the failure as it occurs are combination of inclinome-
points related to interpretation of the monitoring data have ters, crack meters and piezometers that have been correctly
been presented. installed at the appropriate locations, but the programme of
Site characterization for a surface mine in design stages monitoring is likely to be long-term. In the short term signifi-
suggested the presence of some instability. Excavation of cant savings could be made by slope monitoring as opposed to
the lower parts of the mine would aggravate the instability. site investigation with immediate remediation in mind. Moni-
Excavation has exposed rock strata that exhibit considerable toring and analysis, done in the correct way, should relatively
horizontal and vertical variation in strength and deforma- quickly identify those slopes that are most at risk of a failure
bility but with a general tendency for vertical repetition of occurring and should afford the opportunity to stagger spend
behavioral characteristics. Right-of-way restrictions limited on remediation, whilst providing the peace of mind that the
rock cut slope angles to those steeper than traditionally used problems are not being ignored.
for these slide prone and highly erodible geologic strata. Cut
slopes were designed to fit within the proposed right-of-way
by developing countermeasures to minimize weak rock degra-
dation, installing sub-horizontal drains to lower groundwater
levels and developing an instrumentation program consisting
of inclinometers, crack meters, extensometers and piezome-
ters to monitor slope movement, tension crack opening and
water levels. Contingency plans to arrest global movement
were to be employed if slope movements occurred beyond
tolerable values. The slope was monitored during and post
construction. If displacement of the material of the slope or
the pore water pressure exceeded preset tolerances, the system
was capable to alert personnel to evacuate the area.
In this paper an overview has been given of slope mon- Figure 1. Water pressure built up readings by the piezometers before
itoring requirements, from both the geotechnical and survey and after rainfalls.
416

Determination of mechanical properties of a large deformation zone
Rune Glamheden
Golder Associates AB
Isabelle Olofsson
Svensk Kärnbränslehantering AB
Anders Fredriksson
Sweco AB
Keywords: Nuclear repository, Numerical modelling, Site characterization, Rock Properties, Weak rock
1 INTRODUCTION boundary of the area considered for the repository and was
chosen as a reference object to verify the application of the
For the purpose of studying the possibilities of a repository for modelling strategy used for evaluation of rock mass proper-
spent fuel at approximately 500 m depth in saturated granitic ties for large deformation zones, as monitoring values were
rock, the Swedish Nuclear and Fuel Management Company available from tunnels penetrating the zone.
(SKB) has carried out a series of site investigations between
2002 and 2008.
Data collected from these site investigations have been 3 DESCRIPTION OF THE DEFORMATION ZONE
interpreted and analysed to achieve a site description, which
is built up of models from several disciplines of importance The Singö deformation zone is the most pronounced structures
(SKB, 2008). One of the models is related to rock mechan- in the current region. The deformation zone is a regional fault
ics, which aims to characterize the mechanical properties as characterized as a 200 ± 50 m thick steeply dipping zone that
a basis for predicting the short- and long-term stability of the exist over a length of at least 30 km (Glamheden et al. 2007a).
rock mass. The Singö deformation zone is intersected by four tunnels,
two discharge tunnels from the Forsmark nuclear power sta-
tions and two parallel access tunnels to the SFR waste reposi-
2 MODELLING STRATEGY FOR ROCK MECHANICS tory for low and medium radioactive waste (SFR operational
and construction tunnels).
During the site investigations a modelling strategy was applied The appearance of the Singö deformation zone differs
for determining the rock mass properties for each dominant somewhat between the tunnels, but a core zone with transi-
rock type. This article deals with the theoretical approach of tion zones have been observed in all four tunnels. The width
the rock mechanics modelling.The modelling strategy is based of the core zone is 15–35 m. It is characterized by a 2–12 m
on Discrete Element Method for the modelling of behaviour wide zone of crushed rock, showing a high degree of alter-
of rock masses. The rock mass is modelled as a discontinu- ation and disintegration. On one or both sides of the crushed
ous geometry in which contacts between blocks are fractures rock, several clay filled fractures are found, with a thickness
(Olofsson and Fredriksson, 2005), and the discontinuities in of centimetre up to a metre.
the rock are provided by a Discrete Fracture Network model The transition zones are defined as zones having between
(DFN). The numerical code used was 3DEC. Besides fracture two and ten fractures per metre (Glamheden, 2007a).
geometry, the mechanical properties of fractures and intact
rock are required for the modelling.
In order to catch the natural variability of rock properties
4 MODELLING OF EQUIVALENT PROPERTIES
as well as the variability related to sampling procedure many
samples of intact rock and fractures were collected and tested
4.1 Mechanical properties estimated from numerical
in laboratory. More details are provided in (Glamheden et al.
modelling
2007b). In the same way 20 realizations of a DFN were sim-
ulated for each rock type and stress level to account for the Input data required for the modelling of the zone are the DFN
variability in the fracture sets. for each sector of the zone as well as the intact rock and frac-
When looking forwards to the planning of the construction ture mechanical properties. For the host rock mean values of
of the deep repository, determining mechanical properties of mechanical properties of intact rock and fractures as evaluated
major deformation zones are an important issue. The Singö in the site descriptive model of Forsmark were used (SKB,
regional deformation zone in Forsmark defines the northern 2008). In transition and core zones no tests were available
417

Table 1. Evaluated mean values that were used in the subsequent of equivalent properties that are valid for the deformation zone
modelling of deformations in the SFR passage. as a whole. The numerical calculation was, in this case, per-
formed by a box shaped model that includes each typical sector
Parameter Host rock Transition zone Core zone of the zone and embraces the whole zone width. The model
Em (GPa) 45 16.0 2.7
was used to simulate a large direct shear test of the deformation
νm 0.36 0.43 0.43 zone. The parameters that were evaluated from the simulation
φm (deg) 65 51 40.0 are the normal stiffness, the shear stiffness, the cohesion and
Cm (MPa) 4.0 2.0 2.0 the friction angle.
Tm (MPa) −0.3 −0.1 −1.0
6 CONCLUSIONS
for intact rock and fractures, and their properties were there- The results presented in the paper demonstrate that the
fore estimated from empirical experiences and values given methodology used for simulating equivalent mechanical prop-
in the literature. Due to the high fracture density in the core of erties is an applicable and adequate method, even in case of
the zone only 5 DFN realizations for each sector were simu- large deformation zones.
lated. For the host rock 20 DFN realizations (based on fracture Typical equivalent rock mechanical parameters of the Singö
statistics of the host rock) were used as a comparison. Addi- deformation zone evaluated for use in a regional stress model
tional details regarding the fracture network generation and of the Forsmark area are: 200 MPa/m in normal stiffness, 10–
the modelling are presented in (Glamheden et al, 2007a). 15 MPa/m in shear stiffness, 0.4 MPa in cohesion and 31.5 deg
The five DFN realizations for each of the sub-sectors were in friction angle.
analyzed at three different stress levels: 5 MPa (equivalent
to the maximum principal stress σ1 ), 2.5 and 0.5 MPa. The ACKNOWLEDGMENT
mechanical models were loaded with a constant velocity in
the vertical direction while the horizontal stresses were con- The authors wish to acknowledge that this paper is based
stant during the loading test. Evaluated mean values of the on work funded by the Swedish Nuclear Fuel and Waste
mechanical properties of the deformations in the SFR tunnel Management Co (SKB).
passage are presented in Table 1.
4.2 Verification of the results REFERENCES

The results from the rock block model were verified by com-
Glamheden, R., Hansen, L. M., Fredriksson, A., Bergkvist L.,
paring the calculated equivalent properties with records from Markström, I. and Elfström E. 2007a. Mechanical modelling of
in situ load tests and properties evaluated by empirical meth- the Singö deformation zone. Site descriptive modelling Forsmark
ods. The evaluated equivalent properties of the deformations stage 2.1. Svensk Kärnbränslehantering AB. Report R-07-06.
zone were also verified by comparing the recorded and cal- Glamheden, R., Fredriksson, A., Röshoff, K., Karlsson, J.,
culated deformations in the SFR passage. The rock mass Hakami, H. and Christiansson, R. 2007b. Rock mechanics
response in the passage through the Singö fault was calcu- Forsmark. Site descriptive Forsmark stage 2.2. SKB R-07-31.
lated by 3D numerical model using 3DEC. From a general Olofsson, I. and Fredriksson, A. 2005. Strategy for a numerical Rock
point of view, the agreement between recorded and calculated Mechanics Site descriptive Model. SKB R-05-43.
response was considered to be relatively good. SKB, 2008. Site description of Forsmark at completion of the site
investigation phase. SDM-Site Forsmark. SKB TR-08-05.
5 MODELLING OF PROPERTIES VALID FOR THE

DEFORMATION ZONE AS A WHOLE
The rock mass properties that best fitted the recorded deforma-
tions in a back-calculation were used for numerical modeling
418

Spalling in a field experiment of heated boreholes in hard rock
Rune Glamheden
Golder Associates AB, Sweden
Billy Fälth
Clay Technology AB, Sweden
Rolf Christiansson
Svensk Kärnbränslehantering AB, Sweden
Keywords: field measurements, monitoring, nuclear repository, rock failure
1 INTRODUCTION
The Swedish Nuclear Fuel and Waste Management Company

(SKB) has developed the KBS-3 disposal system for spent
fuel. KBS-3 consist of copper canisters with cast iron inserts
containing spent nuclear fuel which are surrounded by ben-
tonite clay and deposited at approximately 500 m depth in
saturated granitic rock (SKB, 2006). For a repository at the
proposed depth of 500 m spalling of the deposition holes may
occur during the drilling of the deposition holes or can be
thermally-induced (SKB, 2008). The process of thermally-
induced spalling in deposition holes may impact the safety of
a repository if a continuous damaged zone is developed along
the hole.
To improve the understanding of the spalling process and
examine the possibilities of reducing its impact on the safety
of a repository, SKB has carried out a large field experiment
during 2008–2009. The objective of the experiment was to
determine if the small counterforce obtained by application of
dry bentonite pellets in the slot between the bentonite buffer Figure 1. View of the test site showing the heating holes and the
instrumentation holes, and the boundaries of a shear zone which
and the rock could be sufficient to suppress or at least reduce intersects the site. The test sequence of the field experiment is also
thermally-induced spalling. indicated in the figure.
The field experiment was carried out in the TASQ tunnel
on the 450 m level at Äspö HRL. The dominating rock type boreholes, which were heated in pairs, have a diameter of ca
around the tunnel is quartz monzodiorite to granodiorite. The 485 mm and a depth of 4 m. The centre to centre distance
quality of the rock mass in the part of the tunnel used for the between the heating holes is 1.2 m.
experiment correspond to “good to excellent” rock quality
according to the Q-index (Andersson, 2007, Glamheden et al,
2010). 2.2 Test sequence
The field experiment was accompanied by numerical sim-
The test sequence of the field experiment and designations of
ulations for estimation of the stress development in the tests.
the heating holes, are indicated in Figure 1. The first and the
The design of experiment and the results of the investigation
second heating tests were performed in open holes without
are summarised below.
any confining pressure on the borehole wall. The third and
the fourth heating tests were carried out in holes with loosely
placed LECA pellets (Light Expanded Clay Aggregates) in a
2 DESIGN OF THE FIELD TEST
50 mm gap created between an inner tube and the borehole
wall, to observe any difference in the occurrence of spalling
2.1 Layout of the experiment
compared to the previous tests. Furthermore, the second and
The test site consists of four pairs of heating holes, each the fourth tests included artificial wetting of the heating holes
surrounded by six instrumentation holes, see Figure 1. The to preserve the moisture content of the borehole wall. After the
419

final heating test, post characterization of the spalled damaged The extent of observed breakouts in the holes after the heat-
zone in the heating holes was carried out. The post character- ing phase is in close agreement with the expected interval of
ization included examination of the hydraulic transmissivity the damaged zone based on the results from the numerical
of the spalled zone in one heating hole in the final test, and modelling. However, it is mainly in the upper part of the heat-
documentation of the geometry of the spalled zone in the other ing holes that the breakouts form continuous notches. The
heating holes. breakouts in the lower part of the heating holes consisted
mainly of isolated spots of spalling.
2.3 Test equipment and instrumentation The results indicate that dry pellets cannot prevent the bore-
hole wall from cracking, whereas they will keep slabs with
The instrumentation in the field experiment included tempera- larger dimensions than the pellet fraction in place. These slabs
ture gauges, relative humidity gauges and monitoring cameras represent a sufficiently large proportion of the spalled volume
in the first test. The test equipment used in the heating holes to obtain a reduced hydraulic transmissivity of the damage
was mounted on a holder for installation as one unit into zone compared to the unsupported case where all fragments
the hole. are free to detach.
In tests 1 and 2 the equipment installed consisted of a centre The dimensions and volumes of the breakouts have been
tube (Ø76 mm) that included the heater, and three pairs of compared between unconfined and confined holes. The results
adjustable spacers for centring of the tube. The dimensional indicate equal or smaller breakouts in the tests performed with
output of the heater elements was 2,300 W. The instruments confinement compared to those carried out without confine-
were in this case mounted on the spacers. In tests 3 and 4, the ment. Considering that the numerical calculations indicate that
holder was complemented with a larger tube (Ø385 mm) that the nominal spalling strength is exceeded further into the bore-
created a 50 mm slot towards the borehole wall. In this case hole wall in the confined holes compared to the unconfined
the main part of the instrumentation was mounted directly on ones, the conclusion is that the limited confinement provided
the surface of the larger tube. Temperature gauges were also by the pellets has reduced the thermal-induced spalling.
installed on several levels in the six boreholes surrounding
each pair of heating holes.
ACKNOWLEDGMENTS
2.4 Substitute for bentonite pellet
The authors wish to acknowledge that this paper is based
The confining pressure on the borehole wall was accomplished on work funded by the Swedish Nuclear Fuel and Waste
with the use of LECA pellets instead of bentonite pellets to Management CO (SKB).
enable simulation of the counterforce of dry pellets, without
the effect of the swelling component.
Pellets of LECA were found to be a suitable substitute for REFERENCES
dry bentonite based on oedometer tests, which demonstrated
similar stiffness of both pellet materials. Calibration of the Andersson J C, 2007. Äspö Hard Rock Laboratory. Äspö Pil-
confining pressure was also carried out in compression tests lar Stability Experiment, Final report. Rock mass response to
using a pellet filled slot with the same dimensions as in the coupled mechanical thermal loading. SKB TR-07-01. Svensk
field experiment. Kärnbränslehantering AB.
The results from the oedometer tests and the tests performed Glamheden R, Fälth B, Jacobsson L, Harrström J, Berglund J and
in the pellet filled slot indicate that loosely placed LECA pel- Bergkvist L, 2010. Counterforce Applied to Prevent Spalling. SKB
TR-10-37, Svensk Kärnbränslehantering AB.
lets generate a confinement pressure on the borehole wall in SKB 2006. Long-term safety for KBS-3 repositories at Forsmark and
the range of 10–30 kPa for 0.5% deformation (Glamheden Laxemar – a first evaluation. Main Report of the SR-Can project.
et al, 2010). SKB TR-06-09. Svensk Kärnbränslehantering AB.
SKB 2008. Underground design Forsmark. SKB R-08-116. Svensk
Kärnbränslehantering AB.
3 SUMMARY OF THE RESULTS
Thermally-induced spalling occurred in the unsupported

holes at approximately the predicted magnitudes of tangen-
tial stress, i.e. at a tangential stress corresponding to about
60% of the uniaxial compressive strength.
420

About the determination of the joint set intensity
N.F. Grossmann
Foundations and Underground Works Division, Concrete Dams Department, Laboratório Nacional de Engenharia
Civil (LNEC), Lisbon, Portugal
Keywords: rock joints; site characterization
1 DEFINITION function f(ω, ε), where ω and ε are, respectively, the longitude
and the colatitude of the attitude (ω, ε), in a spherical coor-
For a homogeneous rock mass, the intensity I corresponds to dinate system whose revolution axis is normal to the mean
the ratio between the sum of the individual areas of the N attitude of this joint set.
joints of the considered set which occur in a given volume V The total intersection becomes
of the rock mass, and that volume
where (sin α) is the mean of (sin α), and the intensity
where Aj is the area of the joint j of that set.

Equation (1) can also be expressed as
3 PLANE OBSERVATION SURFACE

where n is the joint density of the considered set, and A the
mean area of the joints of that set. For a plane observation surface, expression (6) becomes
2 TOTAL INTERSECTION
Generally, neither n nor A are known, and the only data

available are the intersections (or trace lengths) ij of the dif- where S is the area of the observation surface, and
ferent joints of the considered set with a chosen observation
surface S.
In order to obtain an expression for the intensity which may
be used in the practice, the expression for the total intersection
it (the sum of the intersections ij ) of the joints of the considered
set with the surface S, is deduced.
As a first step, all joints of the considered set are supposed where ω0 and ε0 are, respectively, the longitude and the colat-
to have the same attitude (σ, δ) (with the strike σ, and the dip itude of any normal to the plane surface, in the spherical
δ) and the same area A. coordinate system whose revolution axis is normal to the mean
In this case attitude of the considered joint set.
If the attitudes of the joints of the considered set follow a
bivariate normal distribution on the tangent plane at the mean
attitude (Grossmann 1985)
where α is the angle between the normal to the surface element
dS, and any normal to the attitude (σ, δ).
In the next step, the areas of the joints of the considered
set are supposed to follow a distribution with the probability
density function f (A), and, so
where σM and σm are, respectively, the maximum and the

Finally, the attitudes of the joints of the considered set are minimum standard deviations, and ωM the longitude for which
supposed to follow a distribution with the probability density the maximum dispersion occurs.
421

Figure 1. arcsin (sin α) versus σ, for different values of ε0 – Figure 2. arccos (cos β) versus σ, for different values of ε1 –
equation (10). equation (15).
If, by chance, the distribution of the attitudes of the joints the scanline, β the angle between the scanline and any normal
of the considered set is isotropic, equation (10) becomes to the attitude (σ, δ), and (cos β) the mean of (cos β).
As one has always to use the absolute value of (cos β)
where σ is the standard deviation.

In order to give an idea of the values which are obtained where ω1 and ε1 are, respectively, the longitude and the colati-
with equation (10), Fig. 1 presents the variation of the arc sine tude of the scanline, in the spherical coordinate system whose
of (sin α) with σ, for different values of ε0 . revolution axis is normal to the mean attitude of the considered
The value of ε0 for each one of the curves, is, obviously, its joint set.
abscissa for the standard deviation 0. If the attitudes of the joints of the considered set follow a
bivariate normal distribution on the tangent plane at the mean
attitude
4 CYLINDRICAL OBSERVATION SURFACE
If the observation surface can be assumed to be a cylinder or

a prism with infinite height, the intensity becomes
If, by chance, the distribution of the attitudes of the joints

where p is the length of the intersection of the observation of the considered set is isotropic, equation (14) becomes
surface, with any plane having the attitude (σ, δ), β the angle
between the axis of the cylinder (or prism), and any normal
to the attitude (σ, δ), (p cos β) the mean of (p cos β), and dz
a line element which is parallel to the axis of the cylinder (or
prism).
5 SCANLINE
In order to give an idea of the values which are obtained with
If the observation surface is reduced to a scanline, the intensity
equation (15), Fig. 2 presents the variation of the arc cosine
becomes
of (cos β) with σ, for different values of ε1 .
The value of ε1 for each one of the curves, is, obviously, its
abscissa for the standard deviation 0.
where Nt is the total number of intersections of the different

joints of the considered set with the scanline, l the length of
422

Determination of in-situ stress orientation by thermally induced spalling
E. Hakami
Geosigma AB, Stockholm
R. Christiansson
Swedish Nuclear Fuel and Waste Management AB, Stockholm
Keywords: rock stress, field measurements, nuclear repository, rock failure, site characterisation
1 INTRODUCTION These above mentioned facts, and experiences from larger

scale boreholes, gave rise to the idea that induced breakouts
Swedish Nuclear Fuel and Waste Management Co (SKB) are using thermal loading in smaller diameter boreholes, could
conducting the planning and design of a future underground be a possible way of measuring stress orientation in situations
storage for nuclear waste based on the KBS-3 method. The where this was desired. This paper describes the first two pilot
KBS-3 method has vertical deposition holes for placing the experiments performed using a methodology based on this
canisters. In this design work one of the issues is to optimize idea.
the layout of deposition tunnels such that the rock stresses on The main requirement of the method is that the geome-
the boundaries of deposition holes are minimized. try of a borehole can be measured in detail and that such
The SKB experiences from in-situ stress estimations at measurements can be repeated with good repeatability in
depth during the site investigations is that it sometimes is diffi- the results. For this purpose an acoustic televiewer logger
cult to perform the overcoring and hydraulic fracturing stress technique.
measurement methods, and the results are not always consis- The second requirement is that a heat load can be applied
tent. However, one of the observations during the course of symmetrically in the centre of a vertical borehole, such that
these investigations, as also previously observed by several the temperature increase in the borehole wall becomes high
other researchers, has been that the orientation of the break- enough to make the thermal expansion cause breakage. Since
outs in the investigation boreholes are quite consistent and that the stresses caused by the in-situ stress is not isotropic about
this information is valuable as a support to the understanding the borehole but following the direction of the regional stress
of stress orientations at the sites. field, the total stress experienced in the direction of the mini-
The rock types in Sweden are in general crystalline and mum stress direction will reach the failure point first and the
very strong and stiff. At the site, Forsmark, now selected location of the observed spalls can be used to interpret stress
by SKB for detailed investigation and planning, the rock orientation.
type is mainly granite to granodiorite with a high compres-
sive strength. Therefore, borehole breakouts have only been
observed along some minor parts of the 500–1000 m deep
core drilled investigation boreholes. The reason for the break- 2 EXPERIMENTAL SETUP
outs where they do occur was ascribed to stress magnitudes
and rock strength in combination with the temperature con- Details of experimental setup and procedures are presented in
ditions during drilling at the particular point of the borehole the full paper (digital on the CD).
or fractures intersecting the borehole. The breakouts that have
been observed in these boreholes are small and in many cases
only occurring as non-continuous small observations (denoted 3 CONDITIONS AT THE FIELD TEST SITES
micro breakouts).
At the Äspö Hard Rock Laboratory (ÄHRL), a research The borehole tested at ÄHRL is 8 m deep, 76 mm diameter,
laboratory in south-eastern Sweden belonging to SKB, field located at the same site as another field experiment called
experiments with boreholes with the size of a future deposition CAPS (See Glamheden et al. in these proceedings for infor-
borehole (1.75 m) has been performed. When the experimental mation about this experiment.). Due to the CAPS tests the
site was thermally loaded breakouts (spalling) were induced temperatures probably have already reached about 90◦ in the
as fairly continuous thin wedges along the borehole walls. upper parts due to previous heating campaigns. The proxim-
Similarly, in another field test at ÄHRL eight boreholes of ity of the larger holes also means that the stress orientation
0.45 m in diameter were subjected to heat load, and breakouts is expected to be influenced by these holes and the tunnel
were observed in some sections in all cases. The orientations of itself. At ONKALO, the tested borehole, which is 101 mm in
these thermally induced breakouts were also quite consistent diameter, is drilled 8 m vertically in the floor of the POSE
and in agreement with the expected orientation. niche.
423

4 OBSERVATION OF BOREHOLE SPALLS The relatively smaller amount of spalling at ONKALO,
although the strength of the rock is fairly low, can possibly be
The studies of the results were performed in three steps. First attributed to the fact that the ratio between uniaxial strength
the acoustic results for the radii were studied. When there is and tensile strength is quite low (ca 8). This should be com-
an anomaly in the curves showing the maximum radius or the pared to the main rock type at the heated borehole in the Äspö
minimum radius, respectively, this can easily be detected. The Hard Rock Laboratory, where this ratio is in the order of 15.
curves from the repeated loggings at different times may be Empirical studies of spalling behaviour at tunnel scale indicate
compared, and any change in the curve shape between stages that rock types with a higher ratio between uniaxial compres-
be identified. It is notable, and expected, that the places where sive strength and tensile strength are more prone to spalling
we have larger values for the maximum radius, we also get behaviour.
lower values in the curve showing the minimum radius. This Other differences between the sites, such as difference
is explain by the fact that small pieces about to spall will in-situ stress magnitude and differences in rock type textu-
first bulge out from the wall and then loosen on one side, ral, thermal or mechanical properties may also be explaining
still being attached on the other side. The free edges around a factors but further study is needed to resolve this issue with
break will also move slightly inwards. Actually, it was found certainty.
that the minimum radius value curve was more useful than the
maximum radius value to detect the first signs of spalling.
The second step is to study the image showing the results of 6 CONCLUSIONS AND RECOMMENDATION
the acoustic amplitude at the depth indicated by the borehole
curves. By comparing the images from different stages one The pilot test with heating in slim boreholes at two differ-
can find out where some new darker areas or points appear, ent sites as a method to determine stress orientation worked
and it may be judged if the reason is spalling behaviour. out satisfactory, in view of the fact that the stress orientation
Thirdly, the result from the optical logging at this depth level interpreted from of the spalls was shown to be the same as the
is studied. In these images the real character of the rock and expected orientation based on other measurement methods at
the rock type is possible to observe in detail. In this stage the both pilot test sites.
length to width scale of the image should be kept a close to a The amount of induced spalls was small, and less than what
realistic ratio, such that the realistic picture of the fracturing was expected based on an empirical spalling strength criterion
occurring is obtained. for tunnels. This may be an indication of different spalling
In a similar way, at ONKALO it was possible to detect strength for different borehole sizes due to the fact that the
three small differences between the televiewer log results from induced spalling is in the scale of the mineral grains.
before and after the heating period. In this case there was no It is recommended that in future efforts the induced spalling
optical log to compare with. be tested with higher heater effect to increase the thermal load
and amount of spalling, and also decrease the time needed
for the test. Procedures for faster identification of the small
5 INTERPRETATION OF STRESS ORIENTATION breakouts from the logger results may also be developed. Fur-
ther, it is suggested that fragments induced are collected below
The interpretation for the orientation is simply made by study- the heater in the borehole, such that the point of time when
ing the places were anomalies are observed and the location spalling occur may be readily detected.
of the central line through the spall is estimated. The ampli- If the heating procedures are standardized for small
tude logging results are the easiest to use for this purpose. diameter boreholes there will as well be a possibility to
In our case all occurrence was found to be oriented almost use the method to directly quantify the spalling sensitivity
in the same direction 20◦ (or on the opposite side 200◦ ). for the rock type around the borehole. The potential for using
This result agrees well with the previous measurements at the drilling equipment to obtain high borehole temperatures,
Äspö laboratory using overcoring measurements and conver- and purposely induce breakouts while drilling, may also be
gence measurements. Also this agrees with the orientations of investigated.
previously observed breakouts in larger diameter boreholes.
For the three small indications at ONKALO, two of them
are located at about 260◦ and one at about 80◦ . This means REFERENCE
that a NNW-SSE orientation for the major principal stress
is interpreted and agrees well with previous overcoring References are made and presented in the full paper.
measurements made in the same tunnel section.
424

Site investigation in volcanic rock mass for building projects
L.E. Hernández & M.J. Pomares

Regional Ministry of Works, Government of the Canary Islands, Spain
J.A. Rodríguez-Losada & A. Eff-Darwich

Department of Soil Science and Geology, University of La Laguna, Spain
C. Olalla
E.T.S.I.C.C.P., Universidad Politécnica de Madrid, Spain
Keywords: rock mass, site characterization
1 INTRODUCTION Table 1. Type of building.
The geology of the Canary Islands is dominated almost Type of

entirely by a succession of volcanic materials and structures. building Description
The Regional Ministry of Public Works and Transportation C-0 Buildings of less than 4 levels and built area less
of the Canarian Government has drafted The Geotechni- than 300 m2
cal Investigation Guide for Building Projects in the Canary C-1 Other buildings of less than 4 levels
Islands, GETCAN-11 (Hernández et al., 2010). In this guide, C-2 Buildings from 4 to 10 levels
geotechnical surveys will be carried out depending on the type C-3 Buildings from 11 to 20 levels
of building (Table 1) and the following type group of terrains: C-4 Monumental, unique, or more than 20 levels
– Favourable terrains (T-1): those with little variability, and
where the usual practice in the area is direct foundation by
isolated elements.
– Intermediate terrains (T-2): those who show variability, or fragments. And 2) As domes, like a large rock mass. They
that the area does not always happen the same foundation are massifs of trachytic-phonolitic composition, generally of
solution, or where it can be assumed to have some relevance moderate to high strength. They are considered as T-1.
anthropic filled, although probable it not exceed 3.0 m. Unit III: Altered basaltic massif
– Unfavourable terrain (T-3): those who do not qualify in any Composed of basaltic lava flows of small thickness (around
of the previous types. 1 m or less) and moderate to high disturbance. The remarkable
peculiarity of these basaltic lavas is a vertical alternation of
compact basaltic levels (basalt rock) and scoria levels (gran-
2 CANARIAN VOLCANIC ROCK MASS ular material). Pyroclastic mantles and burned paleosols may
appear interspersed within these massifs. The presence of
Canarian volcanic rock mass have been grouped into follow- interspersed scoriaceous levels produces a high heterogeneity,
ing units with geomechanical behaviour more or less similar due to alternations both vertically and horizontally. Overall,
(Rodríguez-Losada et al., 2006, 2007, Hernandez-Gutierrez scoriaceous levels tend to behave like a granular soil, slightly
et al., 2007): compact or loose. But these features fade in Unit III mate-
Unit I: Basal complex rials due to the advanced state of alteration. These materials
Represented by Cretaceous sediments, submarine lavas and show problems as expansivity, high deformability, slope insta-
plutonic rocks (gabbros and syenites). This set is traversed by bilities, and there may be caverns due to water circulation
numerous dykes intrusion with a density so high that often and low compaction. They are generally soft rocks. They are
leave no trace of the rock disposition. Typically, they show a considered as T-3 terrain.
high degree of alteration and rock materials are very slippery Unit IV: Fresh basaltic lava flows
and difficult to recognition. They are considered as T-3 terrain. This is the dominant unit in the Canary Islands.
Unit II: Salic lava flows and salic massifs To this unit belongs the basaltic lava flows that retain their
This unit consists of highly resistant rock materials. There original structure due to their low state of alteration, so the
are two forms of outcrop: 1) As very thick lava flows, usu- types “pahoehoe” and “aa” can be differentiated.
ally with horizontal arrangement or as thick tabular sets, with The basalt rock massif levels in general have high bearing
slopes not too steep and large horizontal extension. Sometimes capacity. However, scoriaceous levels may show low bear-
these sets may consist of very compact breccias with Salic ing capacity and high deformation if scorias are loose and
425

Table 2. Maximum distance (dmax ) between survey points. Table 3. Minimum depth of prospecting.
dmax (m) Minimum depth of prospecting (m)

Geotechnical units/
Type of building T-1 T-2 Building type I (T-3) II (T-1) III (T-3) IVa (T-1)
C-0, C-1 35 30 C-0 5 4 5 4

C-2 30 25 C-1 8 6 8 6
C-3 25 20 C-2 12 8 12 8
C-4 20 17 C-3 16 10 16 10
C-4 20 12 20 12
Minimum depth of prospecting (m)

without matrix. Besides, they show moderate bearing capac- Geotechnical units/
ity and low deformability, if they are welded or with a weak Building type IVb (T-3) Va (T-2) Vb (T-3) VI (T-2)
cementation degree. So for geotechnical surveys purposes,
the Guide considers that within this unit, two subunits may be C-0 5 5 5 5
recognized: C-1 8 7 8 7
C-2 12 10 12 10
– Subunit Iva: “Aa” lavas with compact basalt thicknesses C-3 16 12 16 12
equal to or greater than 2 m, while retaining its lateral con- C-4 20 14 20 14
tinuity across the plot; with less than 0.5 m scoriaceous
levels, absence of cavities and a field slope less than 15◦ .
They are considered as T-1 terrain.
– Subunit IVb: Includes “pahoehoe” lavas and “aa” lavas At least three survey points are established and a maximum
with compact basalt thicknesses <2 m, interspersed with distance (dmax ) between them (Table 2).
scoriaceous levels and/or presence of cavities. They are The maximum distance can be considered as the radius
considered as T-3 terrain. circle of the influence areas of the recognition points. The
Unit V: Pyroclastic deposits influence circle areas of the research points obtained must
It consists of areas of undifferentiated pyroclastic deposits. exceed 90% of the contact area with the ground of the studied
Thickness and dip depend upon topography of the place where parcel.
they were deposited at the time of the eruption. They can be For T-3 terrains recognition points are interspersed in the
subdivided into: problem areas until they are sufficient to characterize the
ground properly. Absence of singularities under the founda-
– Subunit Va: Ignimbrites and tuffs: hard or medium hard tion level and excavation fronts must be checked in order to
rock corresponding to highly compact pumice or cinder ensure the security of the building, the construction process
pyroclastic deposits.. These are materials with characteris- and neighboring buildings.
tics more or less of a hard rock, with a degree of compact- In geotechnical units I, III and IVb (T-3 terrains), the sur-
ness and/or variable cementation. They are considered as veys will be carried out at least as Table 3 indicates for T-2
T-2 terrain. terrains. Additionally, exploration will also be required under
– Subunit Vb: Pyroclastic materials loose or weakly each load transfer element of the structure on the ground. In
cemented: Non-compact and easily collapsible. They are this case, the prospecting that exceeds the number of the appli-
considered as T-3 terrain. cation of Table 3, may be made by rotary-percussion drilling
Unit VI: Volcanic breccias system, under the technical direction and supervision of an
This unit is associated with violent eruptive episodes of high expert in geotechnical survey.
explosivity. The final result is a chaotic and brecciated mass, In the event that the maximum distance (dmax ) exceeds the
formed by blocks of different nature, generally very sharp. dimensions of the study area, distances should be decreased
They exhibit characteristics of hard rock and sometimes of until they reach the required minimum number of points.
medium hardness. They are considered as T-2 terrain. The recognition points were selected at a depth where the
settlements under the loads transmitted by the building are
not significant. This depth can be that where net increase
3 SITE INVESTIGATIONS of tension in the ground, by the building’s weight, is equal
or less than 10% of the vertical effective stress, in that level
Planning field survey should define exploration methods to be point before building construction. These considerations will
used, their distribution and intensity. Based on type of building be valid unless it has previously reached a geotechnical unit
and type of volcanic massif, number of surveys points and their resistant (bedrock) such that the pressure applied on it by the
depth are determined. foundation of the building produce significant deformations.
In GETCAN-11, surveys points correspond to boreholes However, the minimum depth surveys has been established
made by exploration diamond drilling system. In these bore- taking into account the peculiarities and problems associated
holes intact core rock extraction is possible, in order to with each of the geotechnical units and type of building, as
establish geomechanical properties of rock massifs. shown in Table 3.
426

Prediction of rock mass along tunnels by geostatistical method
K. Kaewkongkaew, N. Phien-wej & D. Kham-ai

Asian Institute of Technology, Thailand

Keywords: site characterisation, rock mass classification
In this study, the geostatistical method is tested for its goodness quality at un-sampled areas by means of geostatistical analy-
in prediction of rock mass conditions along alignment of two sis. For Lam ta Khong tunnels, anisotropy variograms RMR
tunnel projects in Thailand excavated in differential geologic of of Phu Kradung rock formation, tunnel hosting rock for-
settings. The first case was tunnels of Lam Ta Khong Pumped mation, were used in the prediction of rock mass quality of
Storage project excavated in a simple geological setting. The the two tunnels selected for study for the project because the
second case was tunnels of Tha Dan dam project which was tunnels traversed only the rock formation. The kriging estima-
situated in a more complex geological setting. In the first case, tion showed similar RMRs to field observation data, except
sample data were from vertical drillholes made in horizontally for shallow zones within 500 m from the portal as shown in
bedded sedimentary rock formations whereas for the second Fig. 1.
case they were from vertical drillholes orientated more or For Tha Dan Dam Tunnels, analysis of drillhole data showed
less sub-parallel to predominant discontinuity set of volcanic that a good variogram for kriging could not be determined
rocks. Parallel faults, shears and fractured zones in hard rocks in horizontal direction, indicating poor spatial correlation of
are characteristics of the volcanic rocks at the site. Rock mass data. This was likely due to effect of the predominant set of
quality along the tunnel alignments is interpreted based on steeply dipping weak discontinuities traversing tunnel align-
borehole data and surface mapping data. It is then compared ment at high angle. Thus the omni-directional variogram was
with the actual ones as mapped at tunnel face during construc- used in the kriging. The comparison between the actual and
tion, so that the goodness the geostatistical predictions can be the predicted RMR along two tunnel of the project as shown
assessed. Fig. 2, indicates that RMRs from field observation consis-
Core logs of drillholes made prior to construction were used tently showed abrupt fluctuations with distance along the
in determination of the rock mass quality in term of RMRs tunnel length with could not be captured by the kriging esti-
along depths of the drillholes so that statistical properties mation. For this geologic setting, the geostatistical method has
and variograms were determined for prediction of rock mass limitation.
Figure 1. Actual versus predicted RMR of Lam Ta Khong Tunnels. Figure 2. Actual versus predicted RMR of Tha Dan Dam Tunnels.
427

Developing in situ stresses estimation algorithm using a hybrid numerical method

based on local stress measurement-case study: Seymare dam
K. Shahriar, M. Gharouni Nik & Z. Khademian

School of Railway Engineering, Iran University of Science & Technology (IUST), Iran
Subject: Numerical methods
Keywords: Rock stress; Back analysis; Numerical methods; Field measurement; Case study; Simulation
1 INTRODUCTION gained from Re-simulation stage, in points of set (N − T ).

It is proposed that as initial values the parameters of a and
Recent experience with underground research has highlighted b is equaled to 904.6 and 812.1, respectively. Equation (2)
the necessity of understanding the in situ stress state for formulates the necessary provision for further investigations.
designing and constructing underground and open exca-
vations, especially with regard to excavation stability and
hydraulic suitability. Analytical solutions for stresses exist
for a range of idealized two dimensional (2D) topogra-
phies, but accurate analytical solutions are not a valuable
for an elastic-plastic with an irregular three-dimensional (3D)
surface (Savage et al. 1985). here, σci and σmi are calculated and measured stresses in point
In this paper, a numerical inverse method is presented to i , respectively.
characterize the state of in situ stresses in a rock mass. According to Eq. (2), If γ ≤ 0.1, the next stage is inves-
tigated; otherwise, it is addressed to recalculate a and b
2 OVERVIEW OF METHOD coefficients by Eq. (1). These parameters indicate a non-linear
distribution of certain stresses in terms of depth h and should
The in situ stress field can never be completely measured, and be determined using a simplex method to minimize the sum
as a result, the methods developed for evaluating it include of the differences between stress values, given by Eq. (1), and
simplifying assumptions (Peng et al. 2007). Equation (1) the measured stresses by HF.
provides a more general relationship.
3.2 Description of the second provision
After determination of local stresses by FEM or FDM, the
remaining stress values (t points) measured by HF should be
compared with the results obtained from the algorithm in the
same points using Eq. 3.
here σ0 and S stand for the in situ measured and the tectonic
stress state, respectively. Also, α is a 3-by-3 matrix of coeffi-
cients and k is referred to as a 3-by-3 matrix of coefficients for
a linear distribution of horizontal stresses (Cornet FH, et al, To be able to proceed along the algorithm, the second pro-
1984 & G. Li, et al, 2009). vision must be investigated, too. If β ≤ 0.1, the process is
terminated and we will have the values of in situ stresses in
3 CONTRIBUTION every arbitrary point; otherwise, according to the second pro-
vision, another set of measured points must be chosen from
In this section, the proposed algorithm is described step- set N to restart the algorithm.
by-step and its applicability is shown by highlighting the
most important stages. Figure (1) illustrates the suggested 4 ALGORITHM APPLICABILITY
flowchart.
To show the applicability of this algorithm, it is applied for
3.1 Description of the first provision
Seymare Dam (SD) to determine its stress field using the pro-
To increase the ability of this algorithm in estimation of in situ posed method and based on the local stresses measured by HF.
stress, it is necessary to compare the measured data with those Fig. 2 shows the location of SD in Iran.
428

Table 1. Input material properties for Hybrid method.
Young’s
Density Poisson’s modulus
Description t/m3 ratio GPa
Upper Assmary 1 2.72 0.31 1000

2 2.78 0.32 950
3 2.54 0.31 150
Middle Assmary 4 2.71 0.3 600
5 2.68 0.28 1200
Lower Assmary 6 2.69 0.29 1800
Table 2. Comparison of measured in situ stresses by HF with

calculated values using the proposed algorithm in SD. district.
σh σH σxy
Location Cal M Cal M Cal M
1 2 2.2 3.6 3.9 4.117 4.475

2 1.8 2.235 4 3.87 4.376 4.485
3 2.4 2.2 4 4.04 4.66 4.496
4 1.99 2.35 3.7 3.863 4.201 4.498
5 2.22 2.48 4.039 4.108 4.68 4.515
6 2.369 2.295 4.1 4.26 4.74 4.524
7 2.3 2.277 3.7 4.174 4.6 4.533
8 2.8 3 4 4.222 4.44 4.22
9 3.401 3.54 3.8 4.01 4.3 4.42
10 2.4 2.318 4.1 4.318 4.35 4.48
11 2.34 2.4 4.3 4.567 4.7 4.6
12 2 2.244 4.6 4.415 4.5 4.3
13 2.1 2.458 4.5 4.463 4.6 4.7
14 2.3 2.571 4.5 4.511 4.357 4.6
15 2.61 2.68 4.4 4.521 4.67 4.8
γ = 0.08, β = 0.098
Figure 1. Proposed algorithm for estimating in situ stresses.

5 CONCLUSION
A new algorithm for determination of the far-field and local

stresses based on stress measurements in some regions is
developed in this paper. The algorithm considers the tec-
tonic stresses, topography, anisotropy and heterogeneity of
the rock mass as effective parameters. Moreover, it is possible
to achieve an estimated in situ stress with allowable error by
changing the provisions. in mentioned algorithm.
Figure 2. Seymare Dam district and its location in Iran.
Table 1 shows input material properties for this method.

Finally, the comparison between measured in situ stresses
by HF with calculated values by the proposed algorithm in SD
is shown in Table 2.
429

In situ stress determination by hydro jacking tests on fractured rock mass
M.A. Kanji
University São Paulo, Brazil
Keywords: hydraulic fractures, rock stress, tunneling, rock caverns
1 INTRODUCTION that the maximum principal stress was also measured addi-
tionally to the minimum principal stress. The paper describes
This paper aims at the presentation of a new in situ test to the testing equipment, procedures and its interpretation.
determine directly the minimum principal stress (σ3 ) and the
maximum principal stress (σ1 ) by means of cheap and simple
hydraulic jacking tests performed in boreholes in fractured 2 EQUIPMENT REQUIRED AND TEST SECTION
rock masses.
Hydro jacking tests are usually performed when studying The required equipment consists of a hydraulic pump with
pressure tunnel sites under possible insufficient rock cover. a minimum flow capacity of 100 l/min under a pressure of
In the event of hydrofracturing, severe damage can occur to 20 to 25 MPa, double packers separated by about 1 to 1.5 m,
the rock mass, as shown by an actual drastic example of rock pressure and flow gages with automatic recording. A drill rig
damaged by hydrofracturing is shown in Figure 1. or a winch is also required do lower and raise the packers. If
The tests are run in boreholes on stretches of fractured rock the rock is too permeable the pump may not be able to reach
separated by double packers (usually separated by 1 to 1.5 m), the required pressure. In this case, it is recommended to select
until opening and closure of fractures are observed, indicating a new borehole stretch.
the value of the minimum stress acting in the rock mass.
Conventional hydrofracturing tests (Haimson, 1978 and
ASTM, 2008) require a section of intact rock for the test, 3 TESTING PROCEDURE AND INTERPRETATION
and the HTPF (Cornett, 1986) needs the determination of the
spatial orientation of at least six known pre-existing fractures After setting the packers in position within the borehole, the
to be testes individually. On the contrary, the test herein con- pressure is raised by steps of 2 atm up to 5 atm at the most,
cerned does not require a borehole section of intact rock but a waiting between 1 and 2 min in each pressure stage, as the
jointed rock mass and do not need inspection of the borehole joints opening pressure is to be determined as close as possible.
walls to detect the strike and dip of the joint. This is called the “slow” cycle. In these hydro jacking tests it
In carrying on hydro jacking tests in several hydro power is not necessary to spend more time in each stage, as in the
projects in the Andean region, the tests were conducted well Lugeon tests, as the response of the ground to the pressure is
beyond the usual pressure ranges for hydraulic jacking tests, very quick.
since a higher capacity pump was available. Several behav- When equaling the minimum confining pressure of the
iors during the tests were noticed, allowing the interpretation ground, plus eventual joint strength that might exist, the joints
will open.This is noticed by the change in relationship between
pressure and flow, as exemplified by the graph of Figure 2,
where the points of opening and closure of fractures are
marked respectively as “O” and “C”.
The pressure is continued to be increased in similar steps
to the maximum value or capacity of the pump, and then
decreased under the same pressure steps. The closure pres-
sure of a joint is a direct measure of the minimum principal
stress, or σ3 . In increasing the pressure, it is possible to notice
more than one changes of pressure to flow ratios, as shown in
Figure 3, indicating that other joints, with different orientation
opened up on their turn.
Following a “slow” cycle, a “fast” loading and unloading
cycle, was carried on, increasing and decreasing continuously
the pressure, within about 5 min. The results were surprisingly
very close to the “slow” cycle. Figure 3 show the curves of
both “slow” and “fast” cycles, with very good reproducibility.
Figure 1. Rock fracture opened by hydrofracturing in a pressure Several other particular behaviors may occur during the test,
tunnel with only shotcrete as support. consisting either as an oscillation (or jerking) of the flow or of
430

Figure 2. Graph of flow x pressure for a test in one cycle of loading
and unloading.
Figure 4. Graph pressure x depth of the several tests made in the

same power project site. The lower and upper limit lines indicate σ1
and σ3.
surface, which may be the neighbor hillside in case of inclined

surface.
At increased pressures a point above which no more events
are noticed and the flow rate is continuous is reached (right
dashed line of Figure 4) and represents the maximum principal
stress, or σ1 (as proposed by Kanji, 1993 and 1998). This
allows to calculate Ko.
4 CONCLUSIONS
Provided a pump with adequate high capacity is available and

the jointed rock mass permeability is limited, this type of test is
Figure 3. Graph of flow x pressure for a slow and fast cycles of the capable of determining both the minimum and the maximum
same borehole section tested. principal stress acting in a rock mass (σ1 and σ3 ), from which
the Ko coefficient may be calculated.
the pressure alone, or even both of them together. This means However, the horizontal direction of the maximum principal
that the pressure in some of the fractures of the jointed system stress is not known, which could be determined by other types
has equaled the ground pressure on it. These occurrences may of tests (flat jacks at surface, etc.). The test herein described is
repeat under different pressures in the same test as the several of low cost and of easy procedure, being useful in preliminary
hydraulically connected fractures are subjected to different design of underground workings.
ground pressures normal to each one and react when the water
pressure reaches the correspondent value. The points where
these oscillations occur in the curves are indicated in Figures REFERENCES
2 and 3 as “F” for the flow and as “P” for the pressure.
Sometimes “spontaneous jacking” may be observed, indi- ASTM (2008): Standard method for determination of In-Situ stress
cated as “J” in the same Figures 2 and 3, when the equip- in rock using hydraulic fracturing method, D4645-08, 7p.
ment behaves as a piezometer and changes the pressure Cornett, F.H. (1986): Stress determination from hydraulic tests on
spontaneously controlled by the joint pressure. preexisting fractures – The H.T.P.F. Method, Proc. Int. Symp. Rock
Each of the test curves must be interpreted marking the Stress & Rock Stress Measur., Balkema/Centek Publ., 301–312.
points of joints opening and closure. All the values may be Haimson, B.C. (1978): The hydrofracturing stress measuring method
then plotted on a graph relating the pressure to the depth of and recent results, Int. Jl. Rock Mech. Min. Sci., V15:167–178.
Kanji, M.A. (1993): Pruebas de hidrofracturamiento para deter-
the tests, as shown in Figure 4, presented as an example. It
minación del estado de tensiones en macizos rocosos, Special
can be seen that the graph has a lower limiting curve (at left) Lecture, Proc. VII Congr. Nac. Mec. Suelos e Ing. Ciment.,
corresponding to the first joint re-opening (or closure) for Lima, 17p.
each of the tests, which corresponds to the minimum principal Kanji, M.A. (1998): Experiences with hydro-jacking tests for the
stress, or σ3 . These values are the same as the calculated by state of stress determination in jointed rock masses, Proc. 5th.
the product of the rock density by the minimum distance to the South Am. Congr. Rock Mech., Santos, V.1: 99–105.
431

A case study on in situ testing of rock mass deformability
B. Kavur, N. Stambuk Cvitanovic & I. Vrkljan

Institut IGH d.d., Zagreb, Croatia
ABSTRACT: Parallel deformability tests, three rigid Plate Loading (PLT) and one extra Large Flat Jack (LFJ) were performed
side by side in slightly jointed marlstone in order to compare their results. The pressure-displacement data obtained from LFJ
and PLT tests are presented and processed using different formulas to determine overall (by the least-squares method), instant
and interval moduli. Different moduli interpreted from PLTs are compared to moduli from LFJ test data. The case shows typical
problems encountered in PLT data interpretation.

Keywords: field measurements, rock properties, site characterisation, rock mass
1 INTRODUCTION plate was 1m in diameter and maximum loading pressure of

the equipment was 10 MPa.
A case in which two types of large in situ tests: a plate load- Test areas were carefully prepared. Damaged and loose rock
ing (jacking) and a large flat jack, were used side by side to material was removed by hammer and chisel manually, as care-
determine rock mass deformability is presented. Such large fully as possible not to disturb planned area of approximately
in situ tests are rarely used due to cost and time constraints 2 × 2 m. After that two coaxial boreholes were core drilled per-
and especially rarely together, side by side in order to compare pendicularly into each prepared rock surface. Deformations of
their results. the rock mass were measured with the use of multiple position
A plate loading is probably the most often used test today borehole extensometers.
but only in larger rock engineering projects. The plate load
test (PLT) is relatively easy to perform and theoretically it
3.2 Large flat jack test
is highly suitable for determining deformability of rock mass.
However, the interpretation of PLT results is encumbered with Large flat jack test was performed with an extra large jack of
numerous factors so that confusion when calculating modulus 1.2 m in width and 1.8 m in length or 1.65 m2 of loaded area.
is rather common. The jack dimensions were proportionally, 1.2 times larger than
The main problem with the interpretation of PLT results the LNEC jack dimensions. The slot opening under the test
arises from the very principle on which it is based. The mod- load was measured at 4 points inside the jack by means of 4
ulus is calculated from displacements measured inside rock spring displacement sensors.
mass by means of multiple position borehole extensometers
(MPBX), while stress distribution is calculated according to
the theory of elasticity. 4 TEST EVALUATION AND RESULTS
Three plate load tests and one large flat jack test were carried
2 TEST LOCATION
out in total. Two PLTs were set-up in a horizontal direction and
one in a vertical direction. The LFJ test was set-up as a vertical
The test area was located on the right bank of Karun River near
single slot test in the heading wall of test adit that enables a
the inlet of the Karun 3 dam diversion tunnel in sedimentary
good comparison with the horizontaly oriented PLTs.
rocks belonging to Pabdeh formation that comprises moder-
First PLT was set up in a horizontal direction perpendic-
ately strong, slightly jointed (joints spaced from 20 to 60 cm)
ularly to the adit axis and located 4 m far from the adit end
marlstone of low durability with RMR values ranging from
(heading wall). Second PLT was carried out immediately after
40 to 45. The Uniaxial Compressive Strength (UCS) of tested
the first. It was set up in a vertical direction and located 18 m
marlstone was between 15 and 25 MPa depending on water
far from the adit end. Seven months later third PLT was set up
content. The lower UCS limit corresponds to wet (saturated)
in a horizontal direction side by side to the first PLT, 5 m far
and the higher to dry rock.
from the adit end. It was carried out in order to compare its
The deformability tests were performed at the end of an
results with the first PLT.
exploratory adit that was excavated by drill and blast tech-
As large differences between the first and third PLT were
nique. The blasting was carerfully performed in order to
noticed, it was decided to carry out a large flat jack test. A year
decrease damage of the rocks.
later then the first PLT, the LFJ test was set-up as a vertical
single slot test in the heading wall of the test adit.
3 TEST PREPARATIONS AND EQUIPMENT Tests are evaluated on the basis of the measured pressure-
displacement curves. Different deformability moduli were
3.1 Plate loading test calculated from PLT and LFJ test data. A distinction was made
Plate loading (jacking) tests were carried out by means of a between moduli of loading (Eml ), unloading (Emul ) and peak
pair of circular rigid plates as a double load plate test. The to peak moduli (Empp ).
432

The PLT data were interpreted in accordance with the for- Generally, exclusions of deformation data measured in sur-
mulae derived from the elastic theory for a rigid circular plate face zones from calculations of overall moduli did not give
loading on an elastic half space. The overall, interval and good results here as calculated PLT moduli were too high in
instant moduli were calculated according to definitions given comparison with LFJ moduli. However, in the P10V floor plate
by Unal (1997). only, where large deformations were recorded in the surface
The calculation of LFJ moduli is based on the theory of lin- zone (blast damage), the exclusion of Zo and Z1 measuring
ear elasticity and the assumption of a homogeneous isotropic points made good result.
body according to the formulae given by the ISRM (1986). The When plate load testing is scheduled to determine deforma-
LFJ moduli Eml , Emul and Empp are compared to the previously bility of rocks like marlstone which remarkably alter their
described PLT moduli. mechanical properties with changes in moisture content, great
care should be taken to preserve moisture content in the
superficial zone of the rock.
5 DISCUSSION Stress-strain data obtained by the large flat jack technique
show a linear elastic behaviour of the tested rock mass with
All pressure-displacement curves obtained in the LFJ test a low degree of hysteresis and high repeatability. The LFJ
demonstrate a linear elastic behaviour of the tested rock mass proved as a valuable and reliable in situ test for the rock mass
with a low degree of hysteresis and high repeatability. There deformability determination and we strongly recommend its
are no significant differences between loading, unloading and use when favorable rock mass conditions exist.
peak to peak moduli. The average most probable values for the Indirect estimates of deformation moduli from RMR values
loading and unloding moduli are 11.9 and 13.3 GPa, respec- based on Serafim and Pereira empirical correlation give about
tively. The average value of the peak to peak modulus is twice lower values than the LFJ measured.
10.2 GPa.All LFJ stress-modulus curves show very small vari-
ations of moduli with stress changes up to 16 MPa after the
first loading cycle is applied. REFERENCES
Indirect estimates of deformation moduli from RMR values
based on Serafim and Pereira (1983) give 5.6 to 7.5 GPa for ASTM D 4394, 2008. Standard test method for determining in situ
the tested area or about twice lower than the LFJ measured modulus of deformation of rock mass using rigid plate testing
values. method.
On the other hand, PLT pressure-displacement curves show ISRM, 1986. Suggested Method for Deformability Determination
relatively large differences in displacements of surface mea- Using a Large Flat Jack Technique. Int. Jour. of Rock Mech. Min.
suring points (Zo ). Such differences are probably caused Sci. and Geomech. Abstr. Vol. 23, No. 2, pp. 131–140.
Poulos, H.G. & Davis, E.H. 1974. Elastic solutions for soil and rock
mainly due to different degrees of surface disturbance. Signi- mechanics. John Wiley. Pp. 165–180.
ficant differences are visible between the LFJ stress-modulus Serafim, J.L. & Pereira, J.P. 1983. Consideration of the Geomechan-
curves on one side and the PLT curves on the other. Calcu- ics Classification of Bieniawski. Proc. Int. Symp. Eng. Geol. and
lated PLT moduli are widely scattered between tests but also Underground Construction, Lisbon, 1 (II) 33–44.
between plates belonging to the same test. Vrkljan, I., Kavur, B., Mehinrad, A. & Ghiassi, S. 1995. Rock Mass
In order to eliminate the effect of the surface zone, dis- Deformability by Extra Large Flat Jack (ELFJ), Plate Load (PLT)
placements measured at surface point (Zo ) were excluded and Dilatometer Testing. Proc. of 8th Congress of the Int. Society
from calculations of the overall moduli. Unfortunately, the for Rock Mechanics, Tokyo, Vol. 1, pp. 185–191.
PLT moduli from such calculations do not seem real as they Vrkljan, I. & Kavur, B. 2001. Experience gained in rock mass
are much higher than the LFJ moduli. deformability testing by large flat jacks. Proc. of the ISRM
regional Symposium, Eurock 2001, Helsinki, Finland, Balkema,
pp. 191–196.
6 CONCLUSIONS Unal, E. 1997. Determination of in situ deformation modulus: New
approaches for plate-loading tests. Int. Jour. of Rock Mech. Min.
Sci. Vol. 34, No. 6, pp. 897–915.
PLT overall moduli determined in tests P10H and P10V by Wittke, W. 1990. Rock Mechanics. Berlin: Springer, p. 754.
the least-squares method with all measured data included are
relatively in good agreements with the LFJ moduli.
433

In situ rock stress-strength comparison: Posiva’s Olkiluoto Spalling Experiment (POSE)
K. Kemppainen
Posiva Oy, Olkiluoto, Finland
M. Hakala
KMS Hakala Oy, Nokia, Finland
E. Johansson
Saanio & Riekkola Oy, Helsinki, Finland
H. Kuula
WSP Finland Oy, Helsinki, Finland
J.A. Hudson
Imperial College and Rock Engineering Consultants, London, UK
Subject: Rock material and rock mass property testing (in situ)
Keywords: nuclear repository, site characterisation, rock properties, rock stress
1 INTRODUCTION
After the Finnish Government’s favourable policy decision

in 2001, Posiva Oy has focused further investigations at the
Olkiluoto site, western Finland and has started in 2004 the
construction of an underground characterisation facility
termed the ‘ONKALO’. The ONKALO construction will be
continued until 2012. The findings of the ONKALO will pro-
vide the knowledge needed for an application, supported by a
Preliminary Safety Assessment, and to construct a final repos-
itory for spent nuclear fuel. Posiva will submit an application
for the construction licence by the end of 2012. The target is
to begin disposal operations in 2020. Figure 1. The layout of the ONKALO underground rock character-
The access tunnel and shafts of the ONKALO will be exca- isation facility and the POSE investigation niche.
vated to the main characterisation level at a depth of 437 m
(Fig. 1). The ramp of the ONKALO has already been exca-
2 POSE EXPERIMENT
vated to a chainage of ∼4600 m (March 2011), which is
equivalent to a depth level of about 430 m. The geological
2.1 Objective and plan
conditions in the ONKALO are characterised by crystalline
bedrock, dominated by migmatitic, foliated gneiss with mas- Spalling strength information for the Olkiluoto rock is
sive, coarse-grained pegmatitic granite dykes. The Olkiluoto required to accurately estimate the probability and severity of
rock mass is also characterised by sub-horizontal fractures potential rock stress-induced damage around the anticipated
and brittle deformation zones. After tunnel chainage 1300 m repository facilities. Based on current information, systemic
(depth ∼ 130 m), the average rock mass quality has been very spalling can be avoided during construction if there is an
good. appropriate design of the underground facilities but, after
A key aspect of the rock mechanics analysis is the under- canister emplacement, the additional thermal stresses can
standing of the in situ spalling strength to be able to predict the cause spalling.
spalling potential, not only during deposition tunnel and depo- The in situ spalling experiment was planned to be carried out
sition hole excavation, but also in the longer term when the in the investigation niche location off the ONKALO ramp at
rock mass is subjected to canister heating. This paper describes about the −345 m depth level, chainage 3620 m, to determine
the status of one of the main rock mechanics in situ exper- the rock spalling strength in representative rock conditions
iments ongoing in the ONKALO—termed POSE (Posiva’s (Fig. 1). To obtain favourable stress conditions below the tun-
Olkiluoto Spalling Experiment) (Aalto et al. 2009). nel floor, the 4.5 m wide and 5.0 m high niche was reshaped
434

Figure 2. The layout of the POSE experiment in an ONKALO niche
at the −345 m level. Figure 3. First 7.2 m deep hole of the POSE experiment showing
the monitoring system: the video recording system is on the right and
the glued strain gauges on the left.
to be 9 m wide and 7 m high. The niche as well the expan-
sion of the niche have been excavated using careful blasting
procedures. The POSE plan consists of drilling two near full-
scale deposition holes, ∅1.52 m (compared to 1.75 m for an
actual deposition holes), to a depth of 7.2 m and leaving a
0.9 m pillar between the holes and observing the holes’ stabil-
ity during the drilling (Fig. 2). The stresses around the holes
will be increased by additional heating in the next phase. A
third additional hole was drilled closer to the niche end.
2.2 Investigations and spalling predictions

Before the execution of the POSE work, several investiga-
tions and spalling predictions were conducted. In line with
the necessity to develop a predictive capability for design pur-
poses, the Prediction-Outcome campaign being conducted for Figure 4. Location of three excavation induced fractures in the
the ONKALO ramp conditions. Hence, predictions were made POSE holes.
using two different approaches: one based on fracture mechan-
ics with Fracod2D (Siren 2011); the other based on traditional did not induce any spalling in the pillar side walls. This is an
continuum thermomechanics with the 3DEC program. Based indication that the spalling strength is higher than anticipated,
on both predictions, damage or spalling could occur in the or that the stress state is different from that estimated, or that
pillar between the two adjacent holes after drilling the second the effect is due to a combination of both of these. The only
hole. Fracture mechanics prediction shows the cracks form- damage that was observed during the hole excavation was
ing only ‘crevices’ in one model but, in the rest of the models, three sub-vertical fractures (two in the first hole and one in
the fractures form wedges—or, in other words, spalling. The the second hole) that were formed on the wall of holes during
fracture growth starts when the rock stress is in the range drilling (Fig. 4). The fractures are located in a mica rich layer,
63–73% of the uniaxial compressive strength. which typically indicates a low strength. The fractures have
A continuum mechanics prediction performed with the not yet been propagated further.
3DEC code also shows spalling in the pillar walls with a max- The preliminary results from the POSE in situ spalling
imum depth of around 80 mm after drilling the second hole. experiment show that about 65 MPa tangential stress has not
The maximum principal stress in the rock pillar between the indicated any spalling in the holes but has caused shear failure
holes is around 75–80 MPa. In the 3DEC analyses, the effect on the wall of 1.5 m diameter full-face bored hole.
of the heating was also studied: after two weeks heating in
the pillar area, the stresses will be increased to the level of
130 MPa. 3 CONCLUSION
The ONKALO facility has enabled direct in situ measure-

2.3 Implementation of POSE and preliminary outcomes
ments and observations of the rock mass. One of the main
The POSE holes were drilled during the summer of 2010 ongoing rock mechanics in situ tests is the POSE experiment
using a full-face boring machine. The boring of the holes at the −345 m level. In addition to the in situ tests, all the
succeeded technically well. Before boring the second hole, observations made during the ONKALO excavations (having
a monitoring system, including video recording, grid marking now reached a depth of 430 m) will also give information for
and strain gauges glued on the pillar wall, was installed in the the rock mechanics property characterisation. Only prelimi-
first hole (Fig. 3). nary results are so far available from the POSE—since the
The heating phase of the POSE test is still underway at the heating phase of the test is still ongoing. The execution of
time of writing this paper, but the drilling of the two holes the test will be finally followed by post-characterisation and
435

will be completed with associated simulations using numerical REFERENCES
codes in order to understand the observed thermo-mechanical
responses and to obtain as accurately as possible an estimate Aalto, P., Aaltonen, I., Ahokas, H., Andersson, J., Hakala, M.,
for the spalling strength of the Olkiluoto gneissic rocks. Hellä, P., Hudson, J.A., Johansson, E., Kemppainen, K., Koskinen,
L., Laaksoharju, M., Lahti, M., Lindgren, S., Mustonen, A.,
Pedersen, K., Pitkänen, P., Poteri, A., Snellman, M. & Ylä-Mella,
M. 2009. Programme for Repository Host Rock Characterisation
ACKNOWLEDGEMENTS in ONKALO (ReRoc). Posiva Working Report 2009–31.
Posiva, 2011. Olkiluoto Site Description 2011. Report POSIVA 2011.
The authors wish to thank Posiva Oy, Finland, for supporting (in prep.)
the work and allowing publication of this paper. However, the Siren, T. 2011. ONKALO POSE Experiment. Spalling prediction
views are solely those of the authors. based on fracture mechanics. Posiva Working Report 2011-23.
436

Comparison of monitoring items for hazardous ground in subsea tunnels
Hyunwoo Kim
School of Civil, Urban and Geosystem Engineering, Seoul National University, Seoul, South Korea
Seokwon Jeon
Department of Energy Resources Engineering, Seoul National University, Seoul, South Korea

Keywords: monitoring, numerical modeling, risks and hazards, tunneling
Monitoring of the ground condition ahead of subsea tunnel displacement and settlement at tunnel crown, when the tunnel
face is indispensable for safety and efficiency of excavation. face approaches a zone of different stiffness. The preceding
Since a sea floor is covered with water, geotechnical investiga- displacements are continuously measured by horizontal incli-
tion is usually carried out in the limited extent prior to the exca- nometers installed at tunnel crown in front of tunnel face, and
vation, resulting in insufficient geotechnical data available for their monitoring results are able to indicate the change of rock
tunneling. Hence, geotechnical investigation should be made mass condition. Lastly, it was considered that the monitoring
continuously during excavation with monitoring items speci- of water inflow at tunnel crown and bottom could inform us a
fied especially in subsea tunnels. Probe drilling has been one change of ground under the assumption that the permeability
of the most widely used methods to determine the ground heterogeneity of rock mass would change the amount of water
condition during subsea tunnel construction. The quality and inflow entering working space immediately after excavation.
applicability of its results are however largely dependent on the In the present study, we observed the trend of vector orienta-
number of drilling holes. The continuously stable excavation tion and preceding displacement at tunnel sidewall as well as
could make a contractor be over-optimistic and decrease the at crown.
number of probe drilling for monitoring, so that the detection A three-dimensional finite element analysis program was
of unexpected hazard may be too lately. Probe drilling some- used to simulate the process of subsea tunnel excavation. With
times requires an additional surveying plan to understand the the results of numerical simulation as shown in Figure 2, three
overall structure of surrounding rock mass. monitoring items were compared in terms of the capability of
The purpose of the present study is to compare monitor- categorizing type of hazardous ground and the early-warning
ing items to find out which of them can provide reliable and time. We assessed their capability to identify the shape of
supplementary information on the ground and can reduce hazardous ground ahead of face and examined which item
uncertainty when being adopted with probe drilling. From deviated from normal trend at the earliest time by the influence
the case studies on construction of subsea tunnels, we first of hazardous ground.
extracted the main geotechnical factors affecting tunnel sta- Figure 3 presents the typical characteristics of monitoring
bility such as type and property of hazardous ground, condition items in four different ground types. It was assessed that the
of surrounding rock mass, in-situ stress and rock cover. Six vector orientation could find out the type of hazardous ground
representative hazardous ground types were then obtained in all cases. The preceding displacement was assessed to detect
by the combination of the factors. Depression and weakness ‘Weak−45’ type clearly from increase of trends both at crown
zone type of ground were selected among them as the major and at sidewall. While it could notice the presence of ‘Depres-
hazardous ground type, which are shown in Figure 1. sion’ or ‘Weak+45’ type, no distinction was made between
Displacement vector orientation, preceding displacement two ground types. The existence of ‘Weak+00’ type could
and water inflow were chosen as monitoring items. The mon- be expected from the decrease of preceding displacement at
itoring with vector orientation can detect the weaker zone crown and the slight change at sidewall. In the case of water
ahead of tunnel face by tracing the ratio between longitudinal inflow, ‘Weak+00’ and ‘Weak−45’ type could be detected
Figure 1. Schematic diagrams and side view of numerical models.
437

Figure 2. Simulation results of three monitoring items in each hazardous ground.
Figure 3. Typical characteristics of monitoring items.
clearly. The presence of ‘Depression’ and ‘Weak+45’ type

was also noticed, but it was not possible to know which of two
grounds was ahead of tunnel face, similarly to preceding dis-
placement. Consequently, the vector orientation was assessed
as the best monitoring item in terms of the capability of hazard Figure 4. Detection time of monitoring items.
categorizing.
For the comparison of the capability of early warning,
the hazard-detecting time of three items was determined by by the water inflow at −1.3D and −1.0D section, so that the
means of cumulative sum (CUSUM) control chart. Figure water inflow was considered as the best monitoring item. In
4 presents the detection time of monitoring items in four ‘Weak+00’ type where the weakness zone was not inclined,
different hazardous ground types. In ‘Depression’ type, we though the preceding displacement solely measuring verti-
were first warned by the preceding displacement and next cal displacement of rock mass detected the weakness zone
by the vector orientation. While the water inflow at crown at the latest time, the water inflow deviated from normal
adjacent to permeable soil deposit changed around the same trend at the earliest time. With ‘Weak−45’ type, the first
time as vector orientation, the water inflow at bottom changed detecting time of three monitoring items were similar in the
only when the tunnel face neared the hazardous ground. With range −1.5D to −1.4D section, but there was much differ-
‘Depression’ type, the early-warning capability of preceding ence in the second detecting time. From these comparison
displacement was evaluated as the best and that of vector results, the water inflow and the preceding displacement were
orientation as the next. With ‘Weak+45’ in which a weak- assessed as the best item in terms of the capability of early
ness zone inclined to the tunnel face, it was first detected warning.
438

Determination of principal stress orientations from wireline logging caliper data and
borehole images in deep drilling
W. Lin
Kochi Institute for Core Sample Research, Japan Agency for Marine-Earth Science and Technology (JAMSTEC)
Keywords: rock stress, site characterization, case studies
During IODP Expedition 319, a deep ocean borehole was

drilled in the center of the Kumano forearc basin in the south-
west Japan Nankai subduction zone, as a part of the scientific
ocean drilling project called NanTroSEIZE. The borehole
penetrated the Kumano forearc basin sediments and the under-
lying accretionary prism. In a depth range from approximately
700 mbsf (meters below seafloor) to the target depth of 1600
mbsf, wireline logging was carried out including caliper and
the FMI borehole images. I used both the borehole images
and caliper data to identify borehole breakouts that could be
interpreted in terms of principal horizontal stress orientations.
The results of stress orientaion determination and thier scien-
tifc interpretations have been published (Lin et al., 2010). This
report is focused on the method of stress orientation analyses
based on borehole breakouts.
Over a depth range of 1288–1578 mbsf (accretionary prism,
late Miocene, Unit IV) of the C0009 borehole, many break-
outs were recognized. I recorded the azimuth data of breakouts
from the FMI images, and yielded average SHmax azimuths and
its standard deviation of approximately N144◦ E ± 9◦ in the
depth range (1288–1578 mbsf, 42 breakouts). The FMI caliper
data showed that at depths above 1285 mbsf the two orthogo-
nal diameter profiles were the almost same, showing no clear
borehole elongations, which suggests that almost no breakout
occurred. Clear borehole elongations were recognized, show- Figure 1. Bathymetric map of the drilling sites and their maximum
ing the presence of breakouts, below 1285 mbsf. I determined principal horizontal stress (SHmax ) orientations (Lin et al., 2010). Bars
the profile of SHmax azimuths below 1285 mbsf shown by the at the drilling sites show the SHmax orientations in boreholes. Real
orientation of maximum borehole elongation in the caliper line bars show stress orientations in the accretionary prism at C0009
data. The average azimuth and its standard deviation were (this study) and C0002 (Tobin et al., 2009). Arrows in right-bottom
N135◦ E and 11◦ , respectively. The average azimuth from the show the far-field convergence vectors between the Philippine Sea
FMI images differed by 9◦ from that from the caliper data, plate and Japan; arrow in left-top shows far-field plate motion vector
based on geodetic results.
probably because poor image coverage (about 50%) resulted
in inaccuracies in picking breakouts. Therefore, I believe
that the stress orientation data from caliper should be more
oblique plate motion is being partitioned into strike-slip and
accurate than those from FMI images.
thrusting.
In the relatively shallow drilled depth range, the orienta-
tion of maximum horizontal stress at C0009 is approximately
parallel to the convergence vector between the Philippine Sea REFERENCES
plate and Japan, showing a slight difference with the stress
orientation which is perpendicular to the plate boundary at Lin, W., et al., 2010. Geophys. Res. Lett., 37, L13303, doi:10.1029/
previous drilling sites C0001, C0004 and C0006 but ortho- 2010GL043158.
gonal to the stress orientation at site C0002. These data show Tobin, H., et al., 2009. Proc. IODP, 314/315/316, doi:10.2204/iodp.
that horizontal stress orientations are not uniform in the fore- proc.314315316.111.2009.
arc basin within the surveyed depth range and suggest that
439

Experimental and numerical analysis of rock block stability using

a remotely positioned laser Doppler vibrometer
G.C. Ma, K. Sawada & A. Yashima

Department of Civil Engineering, Gifu University, Gifu, Japan
F. Uehan & O. Murata
Structural Mechanics Laboratory, Railway Technical Research Institute, Tokyo, Japan
H. Saito & Y. Ootsuka
Technology and Engineering Center, OYO Corporation, Tsukuba-shi, Japan
Keywords: field measurements, risks and hazards, rock failure, rock mass, stability analysis.
This paper examines a new method of evaluating the stabil- the effectiveness of LDV for evaluating the stability of rock
ity of a rock slope using a remotely positioned laser Doppler slopes and cleared that the scaling effects of block size have
vibrometer (LDV). We conducted an experiment using phys- to be considered for the evaluating stability of rock slope.
ical models and performed a numerical analysis to evaluate This study determined that LDV can make accurate block
the new method. vibration measurements, and that the dominant frequency and
Figure 1 shows one of the physical model, which included amplitude of vibrations in a block are related to the block
concrete blocks on an artificial soil slope with two block sizes stability. Furthermore, we found that the dominant frequency
and three slopes. Figure 2 shows one of the result of the exper- of vibrations in the block was inversely proportional to the
iments. The LDV measurements agreed with conventional size of the block.
seismometer measurements.
The experiment results have cleared that the dominant fre-
quency of the blocks varied with the stability, and dominant
frequency and the amplitude varied with the block size (Fig-
ure 3). The numerical model was used to examine a concrete
block adhered to a concrete base with different contact areas.
The dominant frequency of the blocks determined using the
numerical model agreed with those obtained from the physical
experiments (Figure 4 shows the results by numerical model).
Furthermore, we analyzed different sized blocks to examine
the scaling effects. The dominant frequency of the blocks was
inversely related to the block size. These results demonstrated
Figure 3. Examination of the scaling effects using LDV

measurements.
Figure 1. Schematic view of the model experiment.
Figure 2. Comparison of the LD and seismometer results. Figure 4. Example of the block vibration analysis.
440

Investigation of spalling failure at Nayriz Marble Mine (Iran)
M. Moosavi, H.M. Doghozlo, M. Goodarzi & N. Ghavami

School of Mining Engineering, The University of Tehran, Tehran, Iran
F. Moosavi
School of Metallurgy & Materials Engineering, The University of Tehran, Tehran, Iran
Keywords: Rock stress, Numerical modeling, Field measurements, Spalling, Under Coring Technique
Spalling is a mode of failure which occurs in brittle materials. as the average of the values, is more reliable. The measured
In this paper, the effective factors on spalling phenomenon at maximum principle stress is about 8.8 MPa. The amount of
Nayriz Marble Mine (Iran) are investigated (Fig. 1). induced stress at mine wall cannot provide essential conditions
Earlier experiments have proven that this type of failure can for this type of failure, so some other factors that are mentioned
occur when the stress is as high as one-third of the material’s above, might play noticeable roles.
uniaxial compressive strength. Anisotropy and fluctuations of The variation of temperature during a day was measured,
temperature also contributes to weakening of the rock as well and numerical models were carried out to analyze the effect of
as inducing thermal stresses which helps spalling to occur at it on redistribution of stress state at mine wall. This effect can
lower stress levels. In this study, the main emphasize is put on be as high as 15 MPa due to 20 degrees Celsius temperature
the stress measurement to investigate if only this parameter change (Fig. 2).
can explain spalling in the mine or there are other factors The results showed that the concentrated stresses at the wall
contributing to this phenomenon. Samples from the mine were due to in situ stress filed can not by itself provide necessary
taken and their mechanical parameters were determined in conditions for spalling failure to occur, and other factors (such
the laboratory, its results can be found in table 1. The in situ as temperature variations, anisotropy, and/or time dependent
stress state was determined at site by using a new modified behavior etc) should also play a role to cause spalling.
Under Coring Method which is suitable for brittle materials.
This modified approach uses 12 pins around borehole, and
instead of one sets of diagonal for stress determination, 20
states are available, so the obtained result, which is calculated
Figure 1. Spalling and cracking at slope wall.
Table 1. Mechanical properties of the marble.
E UCS Tensile Strength

(GPa) v (MPa) (MPa)
68 0.18 60 7 Figure 2. Distribution of mechanical stress (up), distribution of

thermo-mechanical stress state (down).
441

Geotechnical and monitoring challenges during excavation of

a large urban tunnel in soft ground and low overburden
B. Moritz, R. Matt & R. Heissenberger

Department Tunnelling and Geotechnics – Engineering Services, ÖBB – Austrian Federal Railways, Vienna – Graz, Austria
Keywords: tunnelling, stability analysis, numerical modeling, soft ground, monitoring, support
1 INTRODUCTION
Nowadays, tunnels in modern cities are essential providing

both efficient infrastructure and pollution control. They are
often excavated in difficult ground conditions. Hence tunnel-
ing engineers involved in such projects have to appreciate the
inherent risks when dealing with natural ground (Vavrovsky
et al. 2001). These risks depend on uncertainties in the ground
model, in the physical properties of the ground, in limited
accuracy and simplifications within the mathematical models
used for design.
These uncertainties and consequently the design of tunnels
therefore call for an observational approach in order to allow
for safe and economical tunnelling. For this purpose a geotech-
nical safety management plan has to be implemented (Moritz
et al. 2009a). Figure 1. Geotechnical safety management in tunnels – warning
and action levels of the geotechnical safety management (Vavrovksy
et al. 2001).
2 ACTION LEVELS OF THE GEOTECHNICAL
SAFETY MANAGEMENT
the connection between the western, southern and eastern rail-
The Geotechnical Safety Management – since 2000 sys- way lines in and around the city of Vienna. The project is
tematically applied by the Austrian Federal Railway (OeBB scheduled to be opened for traffic in 2012.
former HL-AG) – contains a warning system adjusted to After groundwater relaxation in advance considerable sec-
the requirements and boundary conditions of a project. For tions were excavated underneath residential areas or major
the implementation a suitable organisation on site, special traffic infrastructure (buildings, a major railway line) with
technical requirements and a management concept is nec- shallow overburden.
essary for deviations from the predicted behaviour – both
in favourable and unfavourable tendencies (Vavrovsky et al.
2001). The successful application of the geotechnical safety 4 DESIGN
management is based on a comprehensive definition of the
expected behaviour by the designer. Stability analyses are the The diameter of the double-track tunnel is approx. 13m and the
basis for the design of the shotcrete lining. During construction cross-section is around 130 m2 . In order to minimize surface
the actual conditions are verified by an on-site geotechnical settlements and to provide a stiff support the side wall drift
engineer. method (two side galleries and a center core) was designed, as
shown in Figure 2.
3 PROJECT OVERVIEW
5 MONITORING AND EVALUATION
Increasing the capacity of the main railway axes crossing
Vienna is an essential concern both at the national and the A comprehensive monitoring programme for the observa-
European level. The Lainzer Tunnel project – a 12.8 km tion of surface settlements and of displacements in the
new strategic railway link is necessary due to the increasing tunnel was executed. Tunnel displacements were systemati-
demand for passenger and freight transport and will improve cally obtained in regular intervals by absolute 3D monitoring.
442

Figure 3. Lainzer Tunnel LT 31 section “W” – example level
of loading of the shotcrete lining for a certain time and different
construction phases.
Figure 2. Lainzer Tunnel LT 31 construction phase – Cross section
with sidewall galleries and center core.
6 CASE STUDIES
A grid of surface targets were arranged in cross and longitu- Two case studies illustrate the successful application of
dinal sections and on buildings. the observational method within the geotechnical safety
The displacements of the targets in the area and along the management.
railway tracks were automatically obtained using the Online- In the first considerable surface settlements in soft ground
3D-geomonitoring system recorded continuously every 15 were measured in a section where the overburden decreases
minutes from stationary monitoring stations. Displacements to a minimum of approx. 6 m and before the excavation works
of the tunnel face, correlating with pre-face surface settle- succeeded in breaking through the concrete pile wall of the
ments, are monitored by an innovative, fully automatic face adjacent lot LT 44.
monitoring system called OBM. The second deals with unfavourable system behaviour
Advanced tools and methods for monitoring data evalua- during bench and invert excavation observed in flysh rocks.
tion assist in the continuous interpretation of the behaviour In both cases the spatial deformation analysis and the
of ground and support interaction, timely detection of devi- comparison between predicted and monitored displacements
ations and check against the warning and alarm criteria. The formed the basis for the selection and management of the
programs allow the evaluation and visualizing of displace- mitigation measures to successfully cope with the hazardous
ment versus time, distance-displacement, deflection curve situation.
diagrams, trends of displacements and ratios of different
displacement components. Furthermore contour plots of REFERENCES
surface settlements can be evaluated. A main feature was the
determination of the stress intensity in the shotcrete lining Moritz, B.; Schubert, W.: Application of the Observational Method in
by calculating the level of loading based on the measured Geotechnical Safety Management, Geomechanik und Tunnelbau
displacements (Figure 3). (2009a) No. 3, Verlag Ernst & Sohn.
Overall these methods allow a much better assessment of the Vavrovsky, G. M.; Ayaydin, N.; Schubert, P.: Geotechnisches Sicher-
geomechanical process during tunnelling, than it was possible heitsmanagement im oberflächennahen Tunnelbau, Felsbau 19
with traditional methods. (2001) Nr. 5, pp. 133–139.
443

A study of the compound evaluation for geophysical explorations by self-organizing maps
M. Nakamura & H. Kusumi

Department of Environmental and Urban Engineering, Kansai University, Osaka, Japan
T. Yamamoto
Department of Environmental and Urban Engineering, Kansai University, Osaka, Japan
Kinki Regional Development Bureau, Ministry of Land, Infrastructure, Transport and Tourism, Osaka, Japan
T. Tsuji
Department of Environment and Resource System Engineering, Kyoto University, Kyoto, Japan
Keywords: monitoring, field measurements, rock slopes and foundations, neural network, rock properties
1 INTRODUCTION
In the investigation of the soundness of aging slopes, the geo-

physical exploration that makes the underground visible in
a nondestructive way by using various physical phenomena
attracts attention. With physical information on the natural
ground by a single geophysical exploration, there is a limit to
interpret the state of the natural ground, therefore two or more
geophysical explorations are often used to make up for the
interpretation limit. Then, we propose the method of eval-
uating the ground by converting into the porosity and the
saturation fraction from the seismic velocity and the resistivity.
However, since even two physical values are not enough, it is
preferable to make a complex evaluation with different phys-
ical values in addition. Therefore, in this paper, we focused
on SOM which is widely used in the field of information pro-
cessing engineering.There were only some cases of the ground
evaluation by SOM, so, in this paper, the adaptability to the
ground properties evaluation of the aging slope was examined.
The result of the classification by SOM was able to be related
to the rock kind and the rock class division that became clear
in the boring investigation. Therefore, it was able to show that Figure 1. The comparison of the classification result and the boring
this method was effective for the improvement of the interpre- investigation.
tative accuracy of compound geophysical exploration
Table 1. The relative amount of the geophysical value in each
cluster.
2 RESULT OF THE STUDY Cluster Vp Vs Vel R
In this research, the geophysical data was characterized and 1 ••• •• •• •

classified to 4 clusters by SOM with k-means method. Figure 1 2 ••• •• ••• •••
and Table 1 show the result of this study. In Table 1, a relative 3 •• • • ••
amount of the geophysical value in each cluster is shown by 4 • •• ••• •
the number of [•]. The opinions obtained from this research
are shown as follows.
By using different dimensional geophysical value (the evaluation of many geophysical explorations because it is
P-wave velocity, the S-wave velocity, the electromagnetic- possible to read many geophysical exploration data simul-
wave velocity, and the resistivity) that had been obtained from taneously and quickly understand and evaluate the feature of
various geophysical explorations, we could cluster them by each clusters on the SOM map.
SOM with k-means method and the classification result that In the change with the passage of time of each geophysi-
was qualitatively corresponding to the degree of the rock class cal data by the regular measurement, it is thought that using
and RQD by the boring investigation. the SOM with k-means method makes it possible to evaluate
It is recognized that this technique is effective for the the change of situation of inside of the ground such as the
improvement of the interpretative accuracy of the compound weathering and the aquifer situation.
444

Measurement of exterior deformation of an earth-filled dam using

GPS displacement monitoring system
S. Nakashima, H. Kawasaki & N. Shimizu

Graduate School of Science and Engineering, Yamaguchi University, Ube, Japan
S. Kubota
Civil Engineering Division of Yamaguchi Prefectural Government, Yamaguchi, Japan
T. Nakano
Tokiwa Chika Kogyo Co., Ltd., Ube, Japan
Keywords: monitoring, field measurements
The exterior deformation of a dam body is one of the

most important measurements in the safety monitoring of
earth-filled dams. While the deformation has generally been
measured by conventional surveying techniques, namely, total
station surveys and leveling, the speeding-up and the automa-
tion of these measurements is required these days from the
viewpoints of saving labor, reducing costs, and maintaining
safety in the measurement works under emergency situa-
tions, such as earthquakes, heavy rains and floods. The Global
Positioning System (GPS) displacement monitoring system
is a new methodology for measuring ground surface dis-
placements. It can automatically and continuously measure
the three-dimensional displacements of multiple points using
GPS technology.
The present study applied the GPS displacement mon-
itoring system to measure the exterior deformation of
Majimegawa Dam, a 21.9-meter-high earth-filled dam located
in Yamaguchi Prefecture, Japan. This paper verifies the effec-
tiveness of the system based on the observational results from
a three-year period.
Figure 1 shows the changes in the reservoir level and the
measured displacements at the dam crest by the GPS system, a
conventional survey and settlement meters. Although the dam
deformation has been less than several millimeters within the
observation period, the GPS system accurately and smoothly
measured such small deformation in response to the rising and
falling of the reservoir water level and the self-weight com-
paction of the core, as shown in the figure. The combination
of highly-frequent measurements, at one-hour intervals, along
with appropriate data treatment by the trend model produced
more reliable displacement data than the conventional sur-
veying method that took measurements once a week at most.
This study has shown the high measurement accuracy in the Figure 1. Comparison of the GPS measurements, the conventional
vertical direction of the GPS system from a comparison with survey measurements and settlement meter measurement.
settlement meters.
The GPS displacement monitoring system is advantageous the effectiveness of the GPS displacement monitoring system
in its laborsaving measurements in routine works as well in the management of fill dams. This monitoring system
as in its prompt measurements under emergency situations, is expected to be developed not only as a simple safety-
because the whole measuring process, from data collection monitoring tool, but also as a non-destructive or non-contact
to processing, is entirely automated. This study has proven visualization tool for internal dam bodies.
445

An innovative method to determine the indirect deformation modulus of

rock masses in Asmari formation of Iran
A. Noorzad
Faculty of Water and Environmental Engineering, Power and Water University of Technology, Tehran, Iran
K. Ahangari & A. Haghnejad

Department of Mining Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran
Subject: Rock material and rock mass property testing [laboratory and in-situ], information system, artificial intelligence and
other advanced techniques
Keywords: field measurements, lab testing, rock mass, case studies, neural network, Asmari formation
Numerous large dams of Iran have been constructed in Asmari Table 1. Equations between CYC1 to CYC5 and E.
formation. Knowing the geomechanical characteristics of this
formation can help significantly to reduce costs and increase Equation R2 F Sig.
accuracy of parameters used in future studies. Deformation
CYC1 = E 0.886 .82 179.70 .000
modulus represents the rock mechanical behavior before fail-
CYC2 = E 1.001 .89 363.33 .000
ure and is used as the most important data for understanding
CYC3 = E 1.029 .90 433.44 .000
deformation behavior of rock mass. Two large concrete dams, CYC4 = E 1.022 .91 470.89 .000
namely Karun IV and Khersan III located in Asmari forma- CYC5 = E 0.934 .97 1.04E3 .000
tion in the south of Iran, have been examined as case studies.
The relation between elasticity and deformation modulus in
two mentioned dams has been investigated by artificial neural
network and statistical method. Dam and hydro power plant of methods for training, validation and neural network testing,
Karun IV is located in Bakhtiari region, one of the branches of are utilized in order to control the results achieved by MAT-
Zagros Mountain. Rock mass surrounded the dam site is com- LAB software. If the obtained results are in agreement with
posed of units ofAsmari formation including different types of the measured values, then the equation between the predicted
limestone, porous limestone, marly limestone, limy marl and values of CYC5 and measured values of E will be presented.
marl. Khersan III dam foundation and abutments have been The characteristics of the developed program are:
situated in Asmari formation. The formation is composed of
• Number of transfer function: Tansig, Purelin
limestone with a medium thickness, marly limestone, marl and
• Learning steps: 6862
small amounts of dolomite. Because of variety in lithology and
• Gradient descent: 0.005
stratigraphy characteristics, both formations are divided into
• Number of input layer: 1
two parts of lower Asmari and upper Asmari.
• Number of hidden layer: 1
In the present research, the data sets of elasticity and defor-
• Number of neurons in the hidden layer: 2
mation modulus have been utilized. It is common by many
• Number of neurons in output layer: 1
researchers to apply RQD, RMR and GSI to present the dif-
• Type of network: Back-Propagation Neural Network
ferences in the behavior of intact rock and rock mass in their
relations to consider the effect of discontinuities. While in this (BPNN)
• Learning algorithm: Traingdm
study only the relations between the 1st to 5th loading cycles
in plate loading test with elasticity modulus have been con- The results indicate that the predicted values of statisti-
sidered. The results indicate that the correlation coefficient cal and neural network methods have the highest accuracy
(R2) and the F values from F tests are improved by increas- by nonlinear relations and have a very good agreement with
ing loading cycles as given in Table 1 with power model. In the in-situ measurements. The accuracy of the equations
other words, increasing the loading cycles leads to decrease is expressed by R2 and F values, which are defined by
the effect of discontinuities in relations between elasticity and the correlation relation between independent and dependent
deformation modulus. Therefore, it is obvious that there is no parameters, respectively. The most desirable equation between
need to utilize RQD, RMR and GSI in order to develop equa- the measured and predicted values of CYC5 is a power equa-
tions. Thus in this paper, the relation between elasticity and tion. The R2 and F values resulted from neural networks
Asmari deformation modulus obtained from the 5th cycle of are more desirable than the obtained values from statistical
plate load test has been employed. method. To conclude the above discussion, using neural net-
In the statistical method, the relation between CYC5 and E work power relation equation is recommended. Therefore, the
is determined considering with and without intercept in the 5th equation is more appropriate as there is a real prediction
equation, in which CYC5 is a dependent parameter while E between E and the CYC5 value resulting more validity in com-
is an independent parameter. In neural network model, E and parison to other equations. On the other hand, the difference
CYC5 values are used as input and output data, respectively. between E and CYC5, which is caused by discontinuities of
Initially, all 26 data sets, which are employed in statistical the rock mass, decreases by increasing E.
446

Measurement of rock stress change by Cross-sectional Borehole

Deformation Method (CBDM)
Y. Obara, Y. Fukushima & T. Yoshinaga

Department of Environmental and Civil Engineering, Kumamoto University, Japan
T. Shin
Chiikikagakukenkyusyo, Japan
M. Ujihara
Mitsui Mineral Development Engineering Co., Ltd., Japan
S. Kimura
Kamioka Mining & Smelting Co., Ltd., Japan
T. Yokoyama
Oyo Corporation, Japan

Keywords: field measurements, monitoring, rock cavern, rock stress
The Cross-sectional Borehole Deformation Method (CBDM) Based on the estimation of stress changes, it was made
is a method that two dimensional state of stress change within clear that the stress components increased in former period
rock mass in a plane perpendicular to a borehole axis can be of excavation and decreased in latter period gradually and
measured. This method was applied to estimate stress change continuously, then that the CBDM is available for measuring
of in mediate rock mass of a rock cavern in Figure 1. This is stress change.
the plan view of measurement site in Kamioka Mine. A cavern
was excavated at a depth of 900 m within gneiss. The Young’s
modulus and Poisson’s ratio are 30 GPa and 0.2 respectively.
The dimension of the cavern is 15 m by 21 m and 15 m in
height. The borehole with a length of 5 m for measurement of
stress change was drilled from the gallery to the cavern before
the start of its excavation. The width of the rock between the
gallery and the cavern is about 7 m. The measuring point is
located at a depth of 4 m.
The measurements were performed at nine stages before,
during and after excavation. The results of stress change are
shown in Figure 2. The vertical stress change σY is zero
until stage II, but increases at stage III, then reaches the
maximum value at stage V. After that, the stress decreases
gradually with the progress of excavation. The tendency
of the horizontal stress change σX is almost the same
that of σY . This means that the rock near the measur-
ing point was damaged and became the loosened zone.
However, as the change of all stress components is con-
tinuously, it is considered that that damage did not happen Figure 1. Location of borehole for measurement and cavern in the
suddenly. plan view of measurement site in Kamioka Mine.
Figure 2. Changes of all components of stressin the X -Y coordinate system: X -axis is defined in the horizontal direction as shown in Figure
1 and Y -axis is defined vertically.
447

Rock investigations for the construction of railway tunnel through Slovenian Karst
Borut Petkovšek, Joerg Prestor, Andraž Geršak

Slovenian National Building and Civil Engineering Institute, Ljubljana, Slovenia, Geological Survey of Slovenia,
Ljubljana, Slovenia
Keywords: site characterization, rock properties, rock mass classification, Karst formations, field measurements, monitoring
Results of investigations: porosity with significant discontinuous permeability and gen-

The tunnel T1 is running from the north to the south through erally very high karstification. Permeability tests gave highly
karst plateau known as Slovenian Karst. It belongs to tec- variable results, from k < 4.5 × 10−7 m/s to k > 3 × 10−3 m/s,
tonical unit of the Outer Dinarides lying in the belt between which is not surprising for karstified limestone. Data for the
Adriatic – Apulian foreland and Dinarides. More precisely, better determination of the groundwater table are still not sat-
it starts in the unit of Tržaško – Komenski synclinorium and isfactory, since the observation period should cover at least
ends in Reški synclinorium. The route is not crossing major two very high rain events and one dry period, which has not
regional faults, though faults are quite frequently expected yet been the case. The first data from 2 months period show
features in the geological profile. They are not more than that in the worst case, the high water table could be expected
few meters wide. The expected directions of faults and major 50 to 100 m above the tunnel level in some parts of the route
fracture systems are N–S in the northern part of the tunnel and while in the others the tunnel would run above the water level.
NW–SE in the southern part. Using analytical assessment, the maximum expected water
Vast majority of the tunnel route will cross limestone, of rel- inrushes to the tunnel front have been calculated and in the
atively homogenous carbonate sedimentation in paralic basin. worst case they come up to 1.5 m3 /s, diminishing in one hour
This sedimentation which is several hundreds of meters thick to 430 l/s. The total amount of drained water through the tunnel
is of the upper Cretaceous and Palaeocene/Eocene age, rep- in use could range from 80 to 260 l/s, which means redirection
resented by alveoline-numulitic (ANA) limestone, limestone of existing flow of the underground water from the direc-
breccia and bio-micrite with Rudist shells and corals. This tion E–W to N–S. Thus, the construction of 1/3 of the tunnel
sedimentation is directly covered by the Eocene deep sea sed- route in non drained conditions (which is more demanding
iments prevailing above the middle part and in the southern due to additional hydrostatic pressure on tunnel lining) has
part of the tunnel route. It consists of mixed layers of marl been proposed. Chemical composition of the water is such
and sandstone. They are called Transient layers and Eocene that chemical aggression on concrete is not expected.
flysch, consisting of mixed layers of marl and sandstone. This The expected engineering geological conditions follow in
rock alters more quickly than limestone and therefore has a the great part the prediction of kastification of the lime-
thicker soil cover. Both the flysch rock and its cover are prone stone and hydrogeological prognosis: The tunnel route has
to erosion (depending on marl content in flysch) which results been divided into 4 quasi homogenous zones. For these, the
in steep canyons of torrents, the slopes of which are often rock was described and characterised in accordance with 2
unstable, with smaller or larger landslides valid rock mass classification systems (RMR, Austrian Matrix
The limestone is karstified, which was evident from the method). It showed up that in the main part the limestone
surface observations. Many different sinkholes, caves, chasms would be stable and only block stability problems would occur,
and ceiling-less caves and also dolinas have been detected in especially in the northern part of the tunnel. For this part,
the area. According to the density of these karst features in the stability of blocks within 3.5 m long unbolted tunnel sec-
the limestone, the tunnel route has been divided into three tion behind the face has been investigated. It showed that in
areas of different karst intensity: the first – northern part, the worst case, unstable blocks of up to 400 t might occur on
which is 2.02 km long and is extremely highly karstified. the face. But the safety factor rapidly increased with an early
The second part of 0.8 km is less karstified but the esti- shotcreeting. High stress induced failures are expected only
mated degree is still quite high. The third part, 3.1 km long within 0.25% of the tunnel route (in the thrust zone on contact
is in between the two formerly defined levels. The rest of the with flysch).
tunnel route runs through flysch, where karstification is not The first zone (2.02 km) consists of extremely karstified
expected. limestone, and when close to the surface (when less than
Hydrogeological prognosis is based on the knowledge of 50 m) quite heavily altered (red clay with debris), with possi-
the wider regional conditions. In this area a highway has ble channels for water. In major part, the tunnel runs above the
been constructed in the past and several structural studies groundwater level. The dip of fractures is very unfavourable:
of the wider area have been published by Slovenian geol- steep and running along the tunnel axis (block stability along
ogists. Limestone in the region is an unconfined aquifer, larger distances). It was proposed that the service tunnel would
more than 1000 m deep. It is characterized by typical karst advance some tens of meters ahead of the main one, and for
448

detection of karst features, continuous georadar sounding to to the tunnel, their frequency is not high. Karst features are
be used. expected, combined with occasionally high water inflows.
The second zone, 0.8 km long, is by prognosis less kars- Drilling ahead of the tunnel front is compulsory.
tified, less altered and less fractured. Faults may reach the The last zone consists of tectonic contact between lime-
thickness of not more than few meters. The groundwater level stone and flysch and further on of the flysch sediments. The
may rise above the tunnel in extreme situations. The over- water level on the contact may be very high, even more than
burden reaches here its maximum value (367 m). If no special a 100 m and crossing of this zone may become highly chal-
karst features are met, no extreme problems are expected here. lenging task, since the flysch sediments may be, in the case
Due to unclear water level conditions it is proposed that the that marl is prevailing over sandstone, very sensitive to the
front of the tunnel is secured by fore – drilling for the detection water. Intensive karstification of limestone is possible at tec-
of water bearing fault zones. tonic contact, which means that special precautions should be
The third zone, 2.2 km long, represents limestone which taken. Extensive monitoring of groundwater level and of rock
is intensively karstified (but less than in the first zone); the deformation, including drilling ahead of the tunnel front will
groundwater level is above the tunnel – in the worst case be necessary.
between 50 and a 100 m. Fracture systems run unfavourably
449

In-situ test program for fault zone characterization
T. Pilgerstorfer & W. Schubert

Institute for Rock Mechanics and Tunnelling, Graz University of Technology, Graz, Austria
Keywords: field measurements, rock properties, site characterization, tunnelling, weak rock
Cataclasite associated with tectonic faults is often assumed to

produce soil-like material, and traditional soil mechanics test-
ing procedures are normally adopted. Its mechanical behavior
is poorly understood, originating from difficulties in sam-
ple acquisition during field investigations, sample preparation
and laboratory testing. However, sections with a high content
of cataclasites form the most challenging stretches during
tunnelling, and proper geomechanical characterization is of
crucial importance. The knowledge of the rock mass proper-
ties is of tremendous relevance for the selection of appropriate
construction methods and supports.
The tectonic faults are usually composed of lens-shaped,
relatively competent rock blocks surrounded by a finely
grained cataclastic material. Hence, their properties are highly
anisotropic.
The 32.9 km long Koralm base tunnel in Austria, which
is part of the Baltic-Adriatic line, will pass a roughly 450 m
long section of the so called Lavanttal fault zone, featur-
Figure 1. Assembled press unit of the plate load tests.
ing above mentioned cataclastic material. In order to gain
reliable information about the rock mass conditions and its
behavior an exploration tunnel in Carinthia was constructed namely a numerical back analysis is presented, using all of the
between 2004 and 2010. The total length is approximately gained data and featuring remarkable consistency.
8 km. Within the scope of an extended exploration program a The performed in-situ tests and accompanying measure-
test adit was constructed in the core region of the Lavanttal ments allow an accurate determination of the rock mass
fault zone in order to conduct an extensive in-situ test- properties. The plate load tests yielded reliable data regarding
ing program. This program included the following tests and the deformability of the rock mass. Based on the results of the
measurements: in-situ shear tests in combination with the results of laboratory
tests the friction angle of the rock mass could be estimated in
• 3D displacement measurements with a dense grid of mon-
a reasonable range. Once knowing those two parameters it is
itoring points to observe the system behaviour during and possible to determine further rock mass parameters through
after excavation numerical back analysis. The generally unknown amount of
• Chain inclinometer above the crown in longitudinal direc- pre-displacements, being commonly assessed based on expe-
tion, installed prior to the excavation, in order to measure rience, was precisely measured by a chain inclinometer. The
the vertical pre-displacements complex mechanical behaviour of the shotcrete was exactly
• Extensometer installed in the centre of the cross section captured by the strain gauges, hence the mobilised support
in longitudinal direction to determine the longitudinal pre- pressure was derived directly from these measurements.
displacements The numerical back analysis of the rock mass parameters
• Strain gauges installed in tangential and longitudinal direc- was based on an extensive data base, granting a high relia-
tion in the shotcrete support to measure the strains occurring bility and soundness of final results. Meeting the following
in the lining during excavation requirements only those results were deemed plausible:
• Plate load tests parallel and perpendicular to the fault orien-
tation to identify the anisotropic deformation characteristics • Pre-displacements correspond to the inclinometer measure-
• In-situ direct shear tests at the boundary of the combined ments
ground-shotcrete structure to determine its bond properties. • Mobilised support pressure matches the evaluation of the
strain gauges in the shotcrete shell
This paper describes the design and the execution of above • Young’s Modulus and friction angle of the rock mass cor-
mentioned in-situ tests and measurements. The evaluation and respond to the results of the plate load tests and the direct
results are discussed in detail. The final evaluation procedure, shear tests.
450

The in-situ testing, when coupled with appropriate evalu- thus creating a substantial basis for the determination of the
ation methods capturing all sources of the gathered data rock mass parameters of the Lavanttal fault zone. The reliable
yields a vast amount of reliable and valuable information for knowledge has lead to a considerable risk decrease for the
geomechanical design of underground structures. intended TBM-advance through the Lavanttal fault system.
Concluding, the most important parameters for the rock It is pointed out that sometimes in-situ testing can not be
mass surrounding the test adit were determined through in-situ replaced by mere laboratory testing and numerical tools, no
tests, measurements, laboratory tests and back calculations, matter how advanced they may be.
451

Geological origin and stability analysis of the Suoduoxi accumulation body

in the upper reaches of the Jinsha River
A.W. Ren, X.G. Wang, Z.X. Jia, Y.J. Wang & Q.W. Duan
Department of Geotechnical Engineering, China Institute of Water Resources and Hydropower Research
Keywords: case studies; site characterization; stability analysis
The Suoduoxi accumulation body is an aggregation of accu- the accumulation are got. Then, on the basis of geomorphic
mulation. It can be seen everywhere along the upper reaches features, the accumulation is divided into four domains. Next,
of Jinsha river. It is a big problem. So many studies have its stability is evaluated through detail rock structure analysis.
been done to resolve the problem. However, they were all And the geological origin is analyzed by integrating all the geo-
focused on single accumulation features. The entirety charac- logical information. Finally, four results are achieved. (1) Most
ter of an aggregate cann’t be achieved by this way. Therefore, of the accumulation is composed of big rock blocks and gravel-
the integrated geological analysis method is applied to study bearing soil, dominated by big rock blocks. So there is a rock
the ancient accumulation in this paper. The study is based on a block arching in the accumulation. But domain 2 has obvi-
project of Suwalong Hydropower Station in the upper reaches ous different structure from other domains. (2) Domain 2 is
of the Jinsha River. The accumulation body is located at an composed of gravel-bearing soil and rock blocks, dominated
area strongly affected by geological structure and climate. It by gravel-bearing soil. Furthermore, there is a grading feature
lies on the right-bank of the Jinsha River and is only about from up to down. This is a typical character of debris-flow. So,
one kilometer far from Suwalong Hydropower Station. It is a domain 2 came from anti-shore. (3) The stability of the accu-
very huge geological body, almost three million tons weight. mulation is largely good. But collapse will probably happen
Its geological and origin is not known up to now, and so its at domain 2. So some control measures should be adopted to
stability will be a big problem for the Suwalong project. In domain 2 before the construction of the hydropower station.
this paper the problem is analyzed by two ways: engineering (4) The product period of domain 1 is earlier than that of the
geological exploration and rock structure theory. There are other domains. But river blocking did not happen during the
two geological thoughts through the paper, not only grasping production period of domain 1. The production of domain 1
the whole, but also doing analogy analysis in detail. First of made just the river diverting. The true river blocking happened
all, by on-site exploration, the whole geological characters of after production of domain 2, domain 3 and domain 4.
452

In situ rock stress measurement using an improved downward compact

conical-ended Borehole Overcoring technique
K. Sakaguchi, A. Kizaki & K. Matsuki

Graduate School of Environmental Studies, Tohoku University, Japan
Subject: Site investigation and filed observations
Keywords: rock stress, field measurements, lab testing
The Downward Compact Conical-ended Borehole Overcoring An in situ test was carried out in a gallery of the Kamaishi
(DCCBO) technique is a method for the in situ measurement mine about 270 m below the surface. In situ stress measure-
of stress that is based on the Compact Conical-ended Borehole ments were carried out at depths of 10.1 m (No. 1), 18.0 m
Overcoring (CCBO) technique. This measurement technique (No. 2), 18.5 m (No. 3), 19.0 m (No. 4) and 19.6 m (No. 5)
can be applied to a water-filled vertical borehole for the in situ from the gallery floor. Through the use of in situ tests, we ver-
measurement of stress. ified that this improved measuring apparatus for a HQ-size
Sakaguchi et al. pointed out that the time-dependence of the borehole works well in a water-filled vertical borehole.
mechanical properties of the adhesive used to bond the strain The measured vertical stress σ v varied in the range 2.9 MPa
cell to the bottom of a borehole filled with water decreases the to 7.9 MPa. However, σ v at No. 5 was 7.9 MPa, and this was
sensitivity of strain measurement. This decrease in the sensi- approximately equal to the overburden pressure 7.3 MPa. With
tivity of strain measurement directly reduces the accuracy of regard to the magnitude of the principal stress, the minimum
stress measurement. To avoid such a decrease in the accuracy principal stress σ 3 at No. 4 shows the influence of some tensile
of stress measurement, they stated that we must consider the stress, but the other results were consistent with each other.
elapsed time after gluing the strain cell and estimated strain The maximum principal stress σ 1 tended to have a north-
coefficients by using a simulation model that incorporated a south orientation. It has been reported that the maximum
layer of adhesive and a strain cell consisting of epoxy resin and principal stress around the Kamaishi mine shows north-
brass. Moreover, for practical use, downsizing of the DCCBO south compression, which was consistent with the present
technique, such as downsizing of the borehole diameter and results.
the measurement equipment, is necessary. We verified the practicality of the measurement method
In this study, first, we downsized the DCCBO technique by comparing our results to those obtained at nearby sites
and performed in situ stress measurement at Kamaishi mine. by JNC (Japan Atomic Energy Agency). JNC conducted in
Second, we estimated in situ stress by using strain coefficients situ stress measurement by the CCBO, Hydraulic Fracturing
that were estimated according to the procedure of Sakaguchi method and Core-based methods (AE and DRA). When we
et al. Finally, we compared the results of this study with compared the magnitude of the principal stress estimated by
those obtained at a nearby test site by other methods, and we this study with those obtained by the CCBO, the intermediate
demonstrated the practicality of the proposed measurement principal stress σ 2 and the minimum principal stress σ 3 were
method. very similar. On the other hand, the magnitude of the maxi-
The main modification points for downsizing are as fol- mum stress estimated by the proposed method was less than
lows:(1) The outer pipe of the measuring apparatus was half of that by the CCBO, and about 70% of those by core-
downsized to HQ-size. (2) The dip/azimuth device consisting based methods. With regard to the difference in comparison
of a small digital video camera and a compass was changed to the CCBO, one possible explanation was that the mechan-
to a tilt/azimuth sensor. This tilt/azimuth sensor had a built-in ical properties of the adhesive in the actual test environment
data-logger for strain measurement. could not be accurately reproduced in the numerical simula-
We selected an epoxy resin-based adhesive (Bond E250: tion. With regard to the core-based methods, this difference
Konishi Co., Ltd.) to use. We performed a laboratory experi- might be due to the difference in the measurement princi-
ment to evaluate the measurement sensitivity under the same ple for the DCCBO technique, which is based on elasticity,
conditions as at the in situ test site. The sensitivity of mea- and that for the core-based methods, which are based on the
surement clearly fell. Therefore, we estimated the strain strain of inelasticity. We concluded that in situ stresses mea-
coefficients by using a simulation model that incorporated sured by the proposed measurement method were sufficiently
both the adhesive and the strain cell. reliable.
453

Laser scanning images analysis for tunnel inspection
F. Sandrone
SBB, Lausanne, Switzerland
R. Wissler
Spacetec Datengewinnung GmbH, Germany
Keywords: case studies, field measurements, tunnelling
Regular inspections of tunnels are necessary for documenting niches, handrails, etc.) and to draw them on a separate layer.
their actual conditions, clarifying the evolution rate of detected This operation allows estimating the serviceability level of
problems and planning repairs. According to the traditional the tunnel installations and equipments. Moreover, when the
inspection procedure for tunnels belonging to the Swiss rail- tunnel is not newly constructed, it is possible to compare suc-
ways, (SBB 2007a, b) all the identified problems should be cessive analyses results as well as successive scan data, which
mapped in a specific inspection drawing once each 6 years. allows assessing the evolution rate of observed damages.
This means that according to the tunnel size, length and con- Due to the application of these new techniques, a new
ditions a visual inspection may often require a quite long inspection methodology (SBB 2011) has been developed for
time to be completed. Today, in spite of an increasing traf- improving and optimising the duration of the tunnel inspec-
fic and demand for safe operations the time for performing tion. The use of scanner systems leads to a considerable
inspections and maintenance of the infrastructure is constantly reduction of the time spent in the tunnel for a visual inspec-
reduced. Therefore, it is necessary to develop new systems for tion. As a matter of facts, according to this new process, only
tunnel survey and maintenance. Those systems should be able to the hammering and verification (i.e. measurements directly
to obtain more accurate results in a shorter time. done in the field by the inspection team, e.g. crack opening
Though the possibilities of using laser scanning techniques width, depth of empty joints) phases need to be realised in the
for detecting problems affecting the tunnel lining was already tunnel while the damages identification, mapping and tunnel
discussed by Haack et al. (1995), it is only today, with the conditions analysis phases can be done in the office. Fur-
improvements in resolution and quality of the measurements, thermore, working in the office with a scanned image in the
that laser scan can be considered one of the quickest and most drawing background strongly reduces the problems affecting
accurate systems for survey and monitoring conditions of the inspection quality observed in the past and mainly due to the
infrastructure (Yoon et al. 2009). Swiss railways currently use unfavourable working conditions in tunnel at night.
laser scan data for clearance profile analyses. Additionally
to the coordinates (i.e. tunnel profile), the used scanner sys-
tem (Spacetec Datengewinnung GmbH) provides in one single REFERENCES
recording process also the temperature and the grey value of
each point. Thus, together with the tunnel profile, the scan- Haack, A., Schreyer, J., and Jackel, G. 1995. State-of-the-Art of
ner produces a visual and a thermal image of the tunnel (i.e. Nondestructive Testing Methods for Determining the State of a
Tunnel Lining. Tunnelling and Underground Space Technology,
a record of the temperature distribution on the tunnel lin- 10(4): 413–431.
ing intrados surface). These images can be analysed with a Inagaki, T., and Okamoto, Y. 1997. Diagnosis of the leakage point on
specific tool which allows damages mapping and statistical a structure surface using infrared thermography in near ambient
analyses of recorded findings. Temperature patterns may give conditions. NDT & E International, 30(3): 135–142.
additional information if compared to visual images. In par- SBB 2007a. Conservation des tunnels (I R 20020): 37 pp.
ticular since cavities or water leaking in the structure appear SBB 2007b. Inspection principale des tunnels. Cahier des charges
in a thermal scan as temperature anomalies, thermal data can (D I-PS-IB 04/07): 16 pp.
be very helpful in identifying the main causes of observed SBB 2011. Unterhaltsrichtlinie Tunnels. Arbeitsanweisung: 39 pp.
damages affecting the shotcrete lining (Inagaki & Okamoto Yoon, J.-S., Sagong, M., Lee, J.S., and Lee, K.-S. 2009. Feature extrac-
1997). These analyses provide a clear image of the structural tion of a concrete tunnel liner from 3D laser scanning data. NDT &
E International, 42(2): 97–105.
conditions of the tunnel.
By analysing the visual images it is as well possible to iden-
tify all the installations characterising the tunnel (i.e. catenary,
454

Numerical simulations of hydraulic fracture intersecting an interbed of sandstone
M. Sarmadivaleh, V. Rasouli & W. Ramses

Department of Petroleum Engineering, Curtin University, Perth, WA, Australia
Keywords: Hydraulic fractures, Lab testing, Numerical modelling
Hydraulic fracturing is a common technique used in the oil creation of a strong bond between the mortar and the sand.
and gas industry for stimulation of tight formations in order to The samples were preserved in water bath for a period of 28
increase productivity. When a hydraulic fracture approaches days in order for the cement to reach to its final strength. The
an interface, which could be a second formation or a small blocks kept in room temperature for one month and were given
pocket of shale accumulation or sand lens, it may get arrested adequate time to loss their water gradually. The oil coated
or cross the interface. This depends on the state of in-situ interface in both blocks was separated for a fraction of mil-
stresses, rock mass properties and the interface characteristics. limeter in order to generate an open interface. A 6 mm hole
From practical point of view, sometime the induced fracture is was drilled in the centre of each block, i.e. in sandstone for
expected to be contained within the reservoir section: the cap sample I and mortar for sample II. A fracturing fluid with
rock integrity is the main objective in this case. The interaction a high viscosity of 100 Pa.s was injected at a constant flow
of an induced hydraulic fracture and an interface with different rate of 50 cc/hour using computer-controlled high pressure
rock types in each side of the fracture is the problem to be syringe pumps. The magnitude of stresses used for testing
studied in this case. However, the fracture may need to intersect sample I were σv = 2500 psi, σH = 2000 psi and σh = 500 psi.
and cross a number of sand packs in order to propagate to the Hydraulic fracture directly crossed the interface at both sam-
required length. An example of this is in tight gas formations ples except for one wing of fracturing test II that crossed the
where gas bearing sandstones are lenses distributed randomly interbed by offsetting. High normal stress on the interface
between shale formations: crossing different layers is the main (σH = 2000 psi) would be the main reason for crossing as it
objective in this case. provide enough fractional force on interface surface.
Simulation of hydraulic fracturing in an unconventional The numerical simulation carried out to investigate the
reservoir which is likely to have lenticular sedimentation, effect of different parameters on the interaction mechanism
directional anisotropy and heterogeneity would be more com- using PFC2D. PFC2D is a district element method (DEM)
plicated than conventional formations. For simplicity, such based numerical code developed by Itasca group and it uses the
formations may be considered as a combination of number of assembly of number of circular elements (2D disks) bonded
individually uniform and homogeneous sections being sepa- together to generate the model. The macro properties of the
rated with one or more interfaces. These interfaces may be samples (UCS, friction coefficient etc) used for this study were
any kind of discontinuity like fractures, interbed, or a bound- estimated from the model micro properties (bond strength,
ary across which the rock type changes. An ideal fracture is friction) which are the input to PFC models. This was done by
expected to propagate longitudinally far enough to satisfy the performing several simulations including bi-axial, direct and
production needs and cross any interfaces that may exist along Brazilian tensile tests from which the rock properties were
its propagation path. At the same time, the hydraulic frac- determined through the plot of Mohr Circles corresponding
ture should be contained within the reservoir section, i.e. does to different stress levels. Similarly, micro and macro hydrauli-
not cross the top and bottom interfaces. This indicates the cal properties were estimated using permeability simulation
multi-functionality expected from an initiated fracture. tests. A hydraulic fracture was initiated in the centre of the
This paper presents the result of numerical simulations sample and model response was monitored.
using PFC2D corresponding to two lab experiments that were Modeling the natural discontinuity may be the most dif-
performed using a true triaxial stress cell (TTSC). The inter- ficult part of this study. This is due to the limited infor-
action mechanism was simulated when a fracture approaches mation being available for the two open (oil wetted) and
an interface across which the formation properties are differ- close cement-sandstone interfaces. Modeling such scenarios
ent. Changing the location of formations with respect to each requires detailed interface characteristics including normal
other and also the interface properties the simulations result bond strength, shear and normal stiffness, friction and dila-
in different model responses. Sample I includes a sandstone tion angles, and cohesion of the discontinuity plane. The
block occupies a volume of 15 × 13.5 × 8 cm in the middle cement-sandstone bond across the closed interface was so tight
of a 15cm mortar cube. In sample II, the mortar fills in the and therefore assigning the average cement and sandstone
volume between two blocks of sandstone of 15 × 13.5 × 4 cm mechanical properties appears to be appropriate. Available
size located on both sides of the mould. The interfaces are fracture properties in the literature were used for modeling
parallel and perpendicular to the sample base. In both sam- the oil wetted (not fluid saturated) interface. This kind of
ples, before pouring the mortar in the mould, one side of the fracture is commonly referred to as gas filled fracture. The
sandstone blocks was coated with oil in order to prevent the experimental set up for estimating the normal stiffness is more
455

complicated than shear stiffness and consequently the avail- In this study no difference was observed in both lab exper-
able data is inadequate. Sayers and Kachanov (1995) proposed iments and numerical simulations between the interaction of
an analytical formula for the ratio of shear (Kshear) to normal the hydraulic fracture intersecting the soft-hard interface (i.e.
(Knormal) stiffness in relation to poisson’s ratio (ν). Liu et al. cement-sandstone interface in test II) and hard-soft interface
(2000) presented a similar relationship based on the planar (i.e. sandstone-cement interface in test I). This could be due
distribution which was considered for slippage of small and to similar mechanical properties of the cement and sandstone
isolated areas. that was used in both cases. The similarities between the sand-
In our study here, the interface is made of two walls with the stone and cement properties will result in an almost a uniform
Poisson’s ratios of 0.16 (sandstone) and 0.24 (cement). Based distribution of stresses in both bodies which was a good effect.
on the mentioned analytical relations, the theoretical range of The offsetting mechanism which was observed in the lab for
the stiffness ratio will be between 0.83 and 0.92. This is close sample II was not predicted in numerical simulations. It may
to the ratio of 1.0 being commonly used for dry gas filled be due to complexities of fluid flow in natural fracture under
fractures. Lubbe et al. (2008) suggested a ratio of 2 based on stress.
his experimental studies but in this study we use a value of 0.9 The above results demonstrate the applicability of the PFC
as the ratio of shear to normal stiffness for the unbounded in simulating a hydraulic fracturing process and also to study
interface. Shear stiffness of 9.75 GPa/m as reported for a the interaction mechanism. A number of sensitivity analy-
sandstone-marl contact by Baldovin (1970) was adopted here ses can be carried out using the developed model in order
for the open cement-sandstone interface. Considering a ratio to investigate different model parameter including formation
of 0.9, the normal stiffness will be 10.83 GPa/m. properties and stresses as well as the effect of sample size on
After fluid injection into the wellbore the induced fractures the results.
were initiated in both models along the direction of maximum
stress. It was seen that in the simulation result corresponding to
sample I the hydraulic fracture was able to cross the interface REFERENCES
in both sides. It generates an ideal double wing fracture. This
is in a very good agreement with the observations in the lab. Baldovin, G. 1970. The shear strength of some rocks by laboratory
Both of the interfaces have experienced a normal stress which tests. In Proceeding of 2nd Congress International Society of Rock
Mechanics, Belgrade, 2, 165–172.
increases their shear strength, which provides enough unity Liu, E., J.A. Hudson, and T. Pointer. 2000. Equivalent medium repre-
even across the open interface. This results in the tensile stress sentation of fractured rock. Journal of Geophysical Research 105,
which was created at the tip of propagating hydraulic fracture 2981–3000.
to be transferred to the other side of the interface. This tensile Lubbe, R., J. Sothcott, M.H. Worthington, and McCann. 2008.
stress transfer ensures the propagation of fracture from one Laboratory estimates of normal and shear fracture compliance,
side to another, i.e. the crossing mechanism is dominant in Geophysical Prospecting, 56, 239–247.
this case. The result of lab test on sample indicates that the Sayers, C.M., and M. Kachanov. 1995. Microcrack-induced elastic
fracture was able to cross the interface in opposite sides of the wave anisotropy of brittle rock. Journal of Geophysical research.
wellbore along the direction of maximum stress and propagate 100: 4149–4156.
into the sandstone blocks.
456

GPS displacement monitoring system for the precise measuring of rock movements
N. Shimizu
Department of Civil and Environmental Engineering, Yamaguchi University, Japan
T. Masunari
Furuno Electric Co., Ltd., Japan
T. Iwasaki
Kokusai Kogyo Co., Ltd., Japan
Keywords: field measurements, monitoring, rock slopes and foundations
Monitoring rock deformations is important to assessing the

stability of rock structures, i.e., slopes, tunnels, dams, foun-
dations, etc., to confirm the validity of the design during/after
construction and to assist in answering specific questions on
a project, etc. In order to achieve successful monitoring, vari-
ous instruments and systems have been developed. The Global
Positioning System (GPS) has the potential to become an
attractive monitoring tool in Rock Engineering.
In this paper, new developments in displacement moni-
toring using GPS are described. In particular, methods for
improving the accuracy and for reducing the measurement
errors caused by meteorological factors and overhead obsta-
cles are highlighted. Practical applications of the system to a
rockfill dam and slopes are introduced.
The Global Positioning System began to be used for dis-
placement monitoring in the fields of Rock and Geotechnical
Figure 1. GPS displacement monitoring system.
Engineering in the mid-1980s. Although it had the potential
to continuously monitor displacements of the earth’s surface In particular, the errors caused by tropospheric delays are
over an extensive area, it was not applied much in practice. the most difficult to correct in GPS. The system has adopted
The reasons were uncertain accuracy, troublesome handling, the modified Hopfield model to estimate tropospheric delays.
and high costs at that time. Obstacles above/beside a sensor, e.g., trees, roofs, buildings,
In order to make GPS much easier to use, and at a low cost, etc., will disturb the signals from satellites and cause measure-
a GPS displacement monitoring system was developed by the ment errors. In order to reduce the influence of such obstacles,
authors and their colleagues, as shown in Figure 1. The system the data from the satellites moving behind the obstacles are
can automatically measure three-dimensional displacements ignored in the analysis. It has been proven that both the meth-
with high accuracy and provide the results to users on the web ods for tropospheric delays and signal disturbances due to
through the Internet in real time. A user only needs to access obstacles can reduce the influence and improve the accuracy.
the home page to see the measurement results at anytime and Various practical applications of the GPS monitoring sys-
anywhere. tem related to slopes, dams, open mine quarries, tunnels,
In addition, new methods for improving the accuracy and foundations, embankments, cut and cover walls, viaducts,
for reducing the errors have been developed in this research. railway trucks, etc., have been carried out.
The accuracy of the conventional GPS monitoring is 5– The conclusions of this paper are summarized as follows:
10 mm for the horizontal direction and 10–20 mm for the
vertical direction. This is not sufficient for precise moni- • Three-dimensional displacements can be measured with
toring. The trend model, which can estimate the real values millimeter accuracy. The methods for reducing the influ-
from scattered data, is adopted to improve the accuracy. ence of tropospheric delays and obstacles have been
Through many cases of experiments and practical applica- established to improve the accuracy.
tions, it has been proven that the system using the trend • GPS displacement measurements can help us to understand
model can detect displacements of 1–2 mm and displacement the unknown mechanisms of complex rock behavior. They
velocities of 0.1 mm/day. can lead to the development of new fields of study in Rock
The next issues are to overcome three major error factors in Engineering.
GPS, namely, (1) tropospheric delays, (2) signal disturbances
due to obstacles above the sensors, and (3) multipath effects.
457

A study on prediction of squeezing and rockburst based on surface exploration for

selection of HLW disposal site in Japan
K. Shin, M. Sawada & Y. Inohara

Central Research Institute of Electric Power Industry, Japan
T. Shidahara
NEWJEC Inc., Formerly at Nuclear Waste Management Organization of Japan (NUMO), Japan
T. Hatano
J-POWER EPDC, Formerly at Nuclear Waste Management Organization of Japan (NUMO), Japan
S. Arai
Geosphere Science Lab., Formerly at Nuclear Waste Management Organization of Japan (NUMO), Japan
T. Miwa
Tohoku Electric Power Co., Formerly at Nuclear Waste Management Organization of Japan (NUMO), Japan

Keywords: nuclear repository, site characterization, tunneling, problem rocks, weak rock
For the selection of the Detailed Investigation Areas for HLW at Quarternary volcanic zone, fault crush zone and hydrother-
disposal in Japan, predicting the tunnel constructability is one mally altered zone of Green Tuff area, mudstone and fault
of the requirements together with assessing long-term safety. crush zone of sedimentary rock of Neogene and later.
This paper dealt with the geological factors relating to difficult Information useful for predicting rockburst has been gath-
tunneling such as squeezing and rockburst. Also it dealt with ered from previous reports. In the Preliminary Investigation
the prediction of rockburst and squeezing. Another paper in stage, geological survey, geophysical survey and borehole
this Congress (Sawada et al. 2011) deals with the engineering survey from the surface are the source of information. There-
characteristics of rock mass through rock mass classification fore rock type, P-wave velocity from seismic exploration
based on rock core. and in-situ rock stress from hydrofracturing have been con-
This paper about difficult tunneling has been based upon sidered. Majority of rockburst events occurred at granitic
analysis of more than 500 tunneling reports about 280 tunnel rock, excluding coal mine where different kind of rockburst
construction. The causes of difficult tunneling are related to occurred at pillars. And P-wave velocity was around 5 km/s at
(i) underground water, (ii) mechanical properties of the rock, the rock of rockburst events. Horizontal maximum and min-
or (iii) others such as gas. The geological factors for exces- imum stresses σH and σh measured by hydrofracturing have
sive water inflow are porous volcanic product of Quarternary, been tested as a criterion for rockburst. It has been inferred
fault crush zone and hydrothermally altered zone of Green that σH − σh > 30 MPa is one of a good criterion for rockburst
Tuff area, and degenerated mixed rock in accretionary com- occurrence. When rock stress data is not available, continu-
plex. The geological factors for squeezing are solfataric clay ous occurrence of borehole breakout is also a good indicator
of rockburst.
This paper also dealt with the prediction of squeezing of
rock in tunneling. For this study, 187 Japanese tunnel projects
which encountered squeezing of rock were collected. And the
relationship between the degree of squeezing (δ/D) and geo-
logical structure, results of seismic exploration and various
rock properties from laboratory tests have been analyzed. As
a result, the followings have been found. i) Many squeezing
incidents are found near anticline, syncline and thrust fault.
Within the data collected, the hanging wall had more incidents
of squeezing. ii) Many squeezing incidents occur in the region
where Vp is less than 2.5 km/s. iii) Cation Exchange Capacity
has not been found to have correlation with δ/D. The parame-
ters which show correlation with δ/D include the ratio of UCS
to overburden pressure (Gn), unit weight (γ) and sand con-
tent (S). (See Figure). Using these 3 parameters, a criterion
Figure 1. Distribution of 3 ranks of δ/D in the 3D-space. Stereo for predicting the squeezing has been established. These find-
graph by parallel method (left graph for left eye and right for right) ings can be used during the following each stage of surface
S (%), γ(kN/m3 ), Gn (-), δ/D = convergence divided by diameter of geological survey, surface geophysical survey and borehole
a tunnel (degree of squeezing). survey.
458

Evaluation of stresses based on overburden and in-situ measurements
Rajbal Singh
Central Soil and Material Research Station, New Delhi, India
Keywords: field measurement; hydraulic fractures; rock caverns; rock mass; rock stress; tunneling
The Central Soil and Materials Research Station (CSMRS), jack tests were 2.67 and 2.96 MPa, respectively with a stress
New Delhi has conducted a number of stress measurements ratio of 1.11. The overburden stress with a height of 70 m over
for many hydroelectric and highway projects being con- the testing site was 1.89 MPa which was less than the mea-
structed/completed in the Himalayan region of India, Nepal sured vertical stress. The average vertical stress as measured
and Bhutan. The methods of measurements were flat jack, by hydraulic fracturing test was 3.795 MPa which was almost
over coring and hydraulic fracturing. There is lot of scatter in two times the overburden stress. It is, therefore, clear from the
the data of these measurements. The horizontal stresses were in-situ stress tests conducted at the site that the in-situ stresses
higher than the vertical stresses. The stresses estimated based in Himalayas can not be estimated based on the overburden.
on overburden were not matching with in-situ measurements. The measured vertical stress was more than the stress due to
Results of all the tests at different project sites have been overburden and horizontal stress was higher than the vertical
analysed and will be presented in this paper. The result of the stress. It is also concluded that there was very large variation
flat jack tests should be used with cautions, as the tests are con- in the results of in-situ stress by flat jack test as compared to
ducted on the surface of the drift, which is highly influenced hydraulic fracturing.
by the disturbances caused during excavation of the testing The case study of another project in the same vicinity of the
drifts. However, flat jack tests are not being conducted these Himalayan region will be presented along with interpretation
days. of data and a comparison with estimated stresses based on
This paper deals with the comparison of flat jack tests with overburden. A comparison of in-situ stresses based on over-
the tests conducted by hydraulic fracturing tests. The aver- burden and measured by flat jack and hydraulic fracturing will
age vertical and horizontal stresses as determined from flat be discussed in the paper.
459

Stress parameter analysis in hydraulic fracturing with open fracture pattern
Joong-Ho Synn, Chan Park & Chulwhan Park

Korea Institute of Geoscience and Mineral Resources, Daejeon, Korea
Ki-Seog Kim
Heesong Geotek Co., LTD, Seoul, Korea
Keywords: field measurements, fluid flow, hydraulic fractures, rock stress
In in-situ stress measurement by hydraulic fracturing method, and pre-existing fracture. The typical pressure-time record
the values of the fracture reopening pressure (Pr ) and the obtained by hydraulic fracturing and the typical fracture pat-
fracture shut-in pressure (Ps ) are the governing parameters terns are represented in Figure 1. The fracture patterns are
in stress analysis. These pressure parameters are commonly classified according to the open-closed condition of the frac-
determined from the pressure-time (P-T) record obtained by ture plane at the unpressurized state and the flow condition
the conventional hydraulic fracturing (HF) and/or sometimes through the fracture plane at the pressurizing state. The three
the pressure-flowrate (P-Q) record obtained by the hydraulic fracture patterns are selected from the observation of the pres-
testing on pre-existing fracture (HTPF). However, it is not so sure drop tendency in the shut-in cycle of P-T record and the
easy to take the exact pressure levels in the P-T record corre- investigation of the fracture features by a borehole televiewer.
sponding to the actual reopening and shut-in behavior of the The closed fracture is defined as Frac-C, in which the frac-
fracture. According to whether the fracture plane is open or ture is completely closed at the unpressurized condition and
closed, the pressure level, which is apparently regarded as the shows the closed-loop flow within the fracture plane during
point of the reopening or shut-in in P-T record, may indicate the reopening to shut-in cycle. The slightly open fracture is
a different meaning in the actual pressure-fracture behavior. defined as Frac-S, in which the fracture is open at just tip at the
In this study, the validity of the determination of the pressure unpressurized condition and shows a closed-loop flow within
parameters (Pr and Ps ) in P-T record and the selection of the the fracture plane during the reopening to shut-in cycle. The
governing equation for stress analysis are discussed with the wide open fracture is defined as Frac-W, in which the fracture
fracture pattern and the stress estimation results. is open to some extent of the fracture plane at the unpres-
The field hydraulic fracturing is carried out in depths of surized condition and shows an open-loop flow through the
100–200 m of a vertical borehole in a mine region where rock fracture plane during the reopening to shut-in cycle.
mass is more or less fractured. The test sections have vari- In hydraulic fracturing using a vertical borehole, the hori-
ous fracture patterns, including new hydro-induced fracture zontal stresses are calculated in general by Equations 1 and 2
Figure 1. Governing parameters of the stress analysis and the fracture pattern in hydraulic fracturing.
460

based on the linear elasticity where Sh is the minimum hori- obtained from HTPF. The meaning and validity of these pres-
zontal stress, SH is the maximum horizontal stress, Ps is the sure parameters are analyzed with the fracture patterns, and
shut-in pressure, Pr is the reopening pressure and Po is the the results of stress calculation are discussed by adopting the
pore pressure. Equation 2 is used as valid for the closed frac- governing equations according to the fracture patterns.
ture pattern of Frac-C, but it comes to be not valid for open From the field test data for the wide open fracture of Frac-W,
fracture patterns of Frac-S and Frac-W. In the case of Frac-S, the Ps,a value is determined as near the pressure level of the
the governing equation for the calculation of the maximum water head at the test depth because of the outward open flow
horizontal stress is derived as Equation 3, where the tensile loop. The Pr,a value is determined to be about 20% higher
stress at the open fracture tip induced by the borehole injec- than the Pn value. The Pr,a value in the reopening cycle is
tion pressure is added to Equation 2 by replacing the pore more similar to the level of the fracture normal pressure or the
pressure (Po ). This can be explained by the linear elasticity fracture shut-in pressure rather than the Ps,a value in the shut-
and the stress distribution around the borehole. However, in in cycle. From the field test data for the slightly open fracture
the case of Frac-W, the elastic governing equation is no longer of Frac-S, the Pr,a value is determined as slightly higher than
valid and the maximum horizontal stress cannot be calculated. the Ps,a value. Ps should theoretically be not less than Pr when
the governing equation for slightly open fracture model is
adopted as SH = 3Ps − 2Pr .The relation between Pr and Ps for
the theoretical validity is represented as Equation 4 with the
stress ratio, K = SH /Sh . According to this relationship, Pr is
smaller than Ps and the ratio of Pr to Ps is linearly decreased
For the open fracture patterns of Frac-S and Frac-W, careful with increasing K. However, as shown in the result of the
consideration is needed in not only the adoption of the gov- determination of pressure parameters from the P-T record by
erning equations but also the determination of the pressure actual field testing, the point of fracture reopening in the P-T
parameters (Pr and Ps ) in P-T record. In adopting the gov- record is not clear and may be overestimated.
erning equation, SH = 3Ps − 2Pr , for the fracture pattern of
Frac-S, Pr and Ps must have the relation of Ps ≥ Pr for the
validity of the stress calculation according to the condition of
SH ≥ Sh . Also to be determined is whether the values of the
apparent reopening and shut-in pressures (Pr,a and Ps,a ) in In summary, the feature of the fracture plane in hydraulic
the P-T record correspond to the actual reopening and shut- fracturing affects the shape of the P-T record and the determi-
in behavior of the fracture or not. In case of the wide open nation of the pressure parameters in stress analysis. According
fracture pattern of Frac-W, the injection flow loop is open to the initial open or closed condition of the fracture plane,
outwards through the fracture plane and so the reopening to especially in the case of the wide open fracture pattern, the
shut-in behavior of the fracture cannot be explained well in reopening and shut-in pressures are not so clear in the P-T
P-T record. This means that although the apparent reopening record and sometimes the apparent pressure parameters in
and shut-in pressures in the P-T record can be determined in P-T record may not represent the actual reopening or shut-
the apparently similar shape of the P-T record, they are not in behavior of the fracture. The fracture reopening pressure
the actual reopening and shut-in pressure. In this wide open seems to be more sensitive to the open fracture pattern than
fracture, the fracture normal pressure (Pn ) can be obtained the fracture shut-in pressure is. Therefore, the pressure param-
from the stepwise flowrate test. This value is used as the pres- eters should be determined carefully with consideration for
sure parameter corresponding to the shut-in pressure in the the fracture plane pattern, the flow loop pattern and the initial
physical meaning of the fracture behavior. stress ratio at the test site. Especially in adopting the govern-
For three kinds of fracture patterns, the apparent reopening equation of stress analysis for the open fracture tip model,
ing pressure (Pr,a ) and the apparent shut-in pressure (Ps,a ) the validity of the fracture reopening pressure is checked and
are determined in the P-T records obtained from conventional adjusted by an analytical approach based on the relationship
HF. In addition, for the fracture pattern of Frac-W, the frac- between the reopening pressure ratio and the initial horizontal
ture normal pressure (Pn ) is also determined in the P-Q record stress ratio at the test site.
461

Nanoseismic monitoring of shallow caverns for detection of incipient failures
M. Tsesarsky
Department of Structural Engineering and Department of Geological and Environmental Sciences,
Ben Gurion University of the Negev, Beer-Sheva, Israel
G.H. Wust-Bloch
Department of Geophysical and Planetary Sciences, Tel Aviv University, Tel Aviv, Israel
Keywords: field measurements, monitoring, rock caverns, stability analysis, weak rock
1 INTRODUCTION Table 1. Classification of detected signal types.
Nanoseismic Monitoring (NM) is a new technique in applied Duration Bandwidth

seismology, which is aimed at monitoring very low energy Name (sec) (Hz) Detection Remarks
(ML > −4.0) source processes. This paper presents a NM fea- Quake >0.3 40–60 Hz All Sensors Extended coda
sibility study, carried out at the shallow caverns of the Bet Spike <0.05 50–80 Hz Partial
Guvrin National Park (BGNP), Israel. These caverns have Puff ∼0.1 >80 Hz Partial Only on vertical
been exhibiting unstable behavior over the years, including Tremor >0.3 >80 Hz Partial
recent failures. NM was applied to detect brittle failures gen-
erated within these caverns. The caverns of the BGNP consist
of a system of underground openings of intricate geometry,
with unsupported spans of up to 40m, excavated in soft chalk
in the 7th–11th centuries.
2 LABORATORY EVENT DETECTION
The ability of the NM system to detect low energy signals

was tested using “four-point bending” of limestone beams
(75 cm × 15 cm × 2 cm), monitored using a single 3-D seis-
mometer (Lennartz Electronics LE-3D) set at the mid-span
and two vertical 1-D seismometers (LE-1D) on either side,
sampled at 2000 Hz. The signals were detected within the flat
response spectrum of the sensors, by low-pass and high-pass
filtering preserving signal energy in the 30–70 Hz band.
2.1 Signal characterization

Four types of different signals were detected and observed.
The signals were classified according to their duration and Figure 1. Time distribution of event types with signal characteris-
frequency content (Tab. 1). Event duration and time decay is tics: amplitude, frequency and signal rate.
classified into three distinct stages according to energy release
and productivity type and rate. These stages are immediately monitoring. During these campaigns we were able to detect
followed by a pre-failure high productivity phase, concluded events in the range −2.4 ≥ ML ≥ −3.8 and properly locate
in a snap trough failure of the beam. Time distribution of event stronger events −2.4 ≥ ML ≥ −2.8. The data sets show no dis-
type and characteristics are presented in Figure 1. tinct event decay but rather loose clustering along a slow event
rate of 10−4 events/sec.
The analysis of active cracking events that occurred in April
3 IN SITU NANO-SEISMIC MONITORING 2009 showed a different pattern, with marked event decay. This
behavior suggests that the catastrophic failure nucleation that
Three monitoring campaigns were carried out in 2008 and led to the opening of the 1-m long crack entered a phase of
2009. Each campaign consisted of two weeks of continuous restrained micro-cracking and bulk stain accommodation.
462

The stability of an underground congress center in soft tuffs through

an integrated in-situ monitoring, experimental, analytical
and numerical methods (Cappadocia, Turkey)
R. Ulusay
Department of Geological Engineering, Hacettepe University, Ankara, Turkey
Ö. Aydan
Department of Marine Civil Engineering, Tokai University, Shizuoka, Japan
M. Geniş
Department of Mining Engineering, Zonguldak Karaelmas University, Zonguldak, Turkey
H. Tano
Nihon University, Koriyama, Japan
Subject: Rock engineering in hazardous environment
Keywords: monitoring, numerical modelling, stability analysis, lab testing, rock caverns
There are numerous antique and modern rock structures in acoustic emission (AE) devices, crack monitoring devices and
the Cappadocia Region of Turkey, which are excavated by temperature, humidity and pressure sensors were installed
mankind. In addition to historical and archaeological charac- onto the outside walls of the hall to observe the behavior of
teristics of the rock structures in the region, their long-term surrounding rock. This research project still continues and the
behavior is also an important data source in terms of rock paper describes the preliminary findings.
engineering. Even today, modern underground openings of The Cappadocia Region is mainly covered by volcanic units
different sizes and shapes in the region are being used for multi of Neogene-Quaternary period. The surrounding tuff of the
purposes. In several locations of Cappadocia, some stability Congress Center is the fall-out product (pumice flow) of the
problems of different scales were observed. One of the modern Kavak member. It is white in color and has a thickness of
underground openings in the region is the Avanos Congress about 7–8 m in the vicinity of the Congress Center. Except 2 m
Center. Avanos municipality planned to utilize a hill remnant thick deposits of thinly layered marly rocks and top soil, the
from an abandoned quarry as an underground congress center surrounding rock has no discontinuities and is very massive
in 1980s. The rough excavation of the congress center was and the bedding is widely spaced. In fully saturated condition,
done. However, the construction has not been completed due a decrease of about 40% occurs in strength of the surrounding
to financial problems and also the change of governors and tuff when compared to that in dry condition. Although the
mayors. The excavation has been completed by more than 10 RMR rating of the rock mass ranges between 70 and 75, in
years. The surrounding rock, in which the Congress Center terms of strength and deformability of the intact rock it is a
has been excavated, is a very soft Cappadocian tuff and sub- weak rock. Comparison of the wetting-drying and freezing-
jected to degradation of material properties, resulting from thawing test results indicates that freezing-thawing cycles are
cycles of wetting-drying and freezing and thawing processes. more effective on decrease in weight loss and UCS of the
Furthermore, it is known from previous experimental works tuff studied. Based on the 4-cycle slake durability tests, Id
that the tuffs of Cappadocia exhibit strong creep behavior. values from the 1st to 4th cycles are between 81.5–98, 69.3–
The observations done by the authors inside and outside of 94, 56.7–88.3 and 40.7–77.1%, respectively. Particularly after
this uncompleted congress center revealed that there are some the 2nd cycle a considerable disintegration of the samples
new fracture formation and propagations at several locations was observed. Freezing-thawing shows its effect as the flaking
in the underground halls as well as at the perimeter of the hill. and/or slabbing of rock surface as observed by the authors.
These cracks are easily recognizable during drying process of The acoustic emissions induced by fractures coincide with the
rock soon after a rainfall. observation periods
This study aims to clarify the short- and long-term stability The experimental results indicated that the surrounding
of the Avanos Congress Center in terms of rock engineering. rock was quite vulnerable to cyclic freezing and thawing and
For the purpose, experimental studies concerning short-term wetting-drying processes.The strength of rock is reduced dras-
properties as well as the effects of freezing-thawing, cyclic tically under saturated conditions and the process of freezing
wetting-drying were carried out, and the stability of the open- and thawing accelerates further degradation of rock under such
ing was assessed by analytical and numerical techniques by a condition. The simple short-term stability analyses by ana-
considering the data from experimental studies. Two complete lytical methods and finite element analyses clearly showed
463

that some tensile cracking would take place. The bending of rock mass for the fully saturated condition, implying that
stress indicated that the tensile strength of solid rock at the some tensile fracturing may occur in the roof of the opening.
top of the solid roof near the sidewalls. The pure creep anal- Furthermore, tensile stress occurs near the ground surface at
ysis with the consideration of the arching model implied that both sides of the opening. This is similar to those of the built-
the openings may become unstable 28–30 years after the exca- in beam condition. The maximum displacement occurs under
vation. However, if the degradation resulting from the cyclic fully saturated condition and greater 2.8 times than that for the
freezing-thawing and wetting-drying taken into account the dry condition.The in-situ monitoring clearly showed that some
stable duration becomes shorter. Numerical analyses indicated further crack propagation does occur especially after the rainy
that high tensile stresses occurred at the crown of the widest and freezing-thawing periods. Nevertheless, further studies on
opening in both dry and wet conditions. Especially, tensile the long-term characteristics of this tuff are necessary to check
stress at the ceiling opening is greater than tensile strength this conclusion.
464

Evaluation study of nozzle design selection for hydraulic fracturing test on concrete,
limestone and andesite with triaxial loading at the laboratory
R.K. Wattimena, S. Kramadibrata, N.P. Widodo & R.K. Uripto

Laboratory of Geomechanics and Mine Equipment, Department of Mining Engineering,
Faculty of Mining and Petroleum Engineering, Institute of Technology Bandung, Bandung, Indonesia
Subject: Others – (Influence of Spray Direction in Hydraulic Fracturing Test)
Keywords: fluid flow, hydraulic fractures, lab testing, rock stress
As it has been well known that hydraulic fracturing test is a Table 1. Physical properties of the samples.
method whereby in-situ stress at a borehole can be predicted
and this method employs the principles of hydraulic. When Sample ρnat ρdry ρsat W S n
the fluid pressure is greater than the tensile strength of the Type (gr/cm3 ) (gr/cm3 ) (gr/cm3 ) % % % e
borehole wall fractures will then be initiated. Concrete 1.72 1.54 1.95 11.60 44.65 40.62 0.69
In this research, the effect from different types of fluid Limestone 2.37 2.37 2.38 0.28 50.00 1.32 0.01
burst to the result of hydraulic fracturing was studied. The Andesite 2.20 2.13 2.27 3.62 53.89 14.13 0.17
hydraulic fracturing test in this research used different con-
figuration of nozzles. In order to create various fluid burst
directions, three different nozzles were designed and used and
each of these enables vertical spray, horizontal spray and a Table 2. Mechanical properties of the samples.
combination of vertical and horizontal sprays. The hydraulic
fracturing tests were conducted in three types of cubical sam- Sample σt σc E c φ
ples such as concrete, limestone and andesite with dimension Type (MPa) (MPa) (MPa) ν (MPa) (◦ )
of 25 × 25 × 25 cm, and triaxial loading was applied for each
Concrete 1.08 10.93 4.07 0.13 3.71 33.29
sample. Limestone 6.06 34.57 9.85 0.25 9.17 55.14
The physical and mechanical properties of the samples can Andesite 6.54 62.36 10.73 0.19 12.38 56.62
be seen at Table 1 and Table 2.
The summary of the testing results can be seen in Table 3.
There were nine tests in total, consisting of each type of nozzle
used in each type of sample once. Table 3. Summary of breakdown pressure.
The tests revealed that the most efficient type of nozzle that
could initiate fracture in all of those samples was the nozzle Breakdown pressure (MPa)
that bursts a combination of vertical and horizontal spray. The
tensile strength properties of the sample also have influence Sample Vertical Horizontal Combination
to the breakdown pressure, that the higher tensile strength will Type nozzle nozzle nozzle
have the higher breakdown pressure. The direction and shape
Concrete 6.01 4.94 0.76
of fracture induced by hydraulic fracturing test in this research
Limestone 18.37 12.17 11.86
is parallel to the maximum horizontal stress and perpendicular Andesite 22.22 18.95 18.80
to the minimum horizontal stress.
465

Seismic imaging of stressed pillars in abandoned coal mines:

Case studies from Australia
Robert J. Whiteley
Coffey Geotechnics, Sydney, Australia
Keywords: case studies, coal mines, rock stress, risks and hazards
Since mining began in Australia over 200 years ago many

underground mines have been developed and abandoned.
These now lie within the precincts of large Australian cities,
towns or high priority growth areas and are adversely impact-
ing surface developments due to fear of mine subsidence. As
many of these old mines were operated with the room-and-
pillar mining the major subsidence risks occur where standing
and/or partially collapsed workings lie within 100 m of the
ground surface.
Standing coal pillars exhibiting anomalous stress conditions
represent the major geotechnical hazard as they can fail catas-
trophically and can trigger other collapses. Failed workings
and mine voids typically have lower seismic velocities than
the surrounding rock mass while anomalous stress conditions
show increased seismic velocities due to the closure of micro-
cracks in response to stress Also the concentrations of stress in
strata underground will be greatest in the most competent rock
units. Using first arrival compressional (P) waves researchers
have observed seismic velocities increasing, non-linearly, at a
decreasing rate in direct proportion to the applied stresses until
the yield point, beyond which, velocities decrease rapidly until
failure. The advantages of P-wave seismic imaging using first
arrival travel-times to image anomalous stress fields are that
P-waves are easier to generate and detect in boreholes or
underground and they tend to concentrate in the higher veloc- Figure 1. Pre-split seismic image of the coal pillar.
ity geological units and where increased stresses are present.
Also these regions frequently transmit a higher proportion claystone floor, a very low strength, low modulus and high
of higher frequency energy that enhances object resolution. swelling material that degrades rapidly when exposed to mois-
These effects are in contrast to those at a mine void whose ture. The spanning capacity of the roof unit and the bearing
presence is inferred from attenuated and delayed P-wave capacity of the support pillars on the soft claystone floor
arrival that have been mainly diffracted around or scattered are considered critical to the success of mining and to the
by the void. prevention of large surface subsidence.
This research provides two case studies from different It was decided to split a stable pillar that had been standing
abandoned coal mines in New South Wales that demonstrate for 15 years. The pillar was at a depth of about 85 m depth and
the application of underground and borehole P-wave seismic away from any significant geological structures. The seam
tomographic imaging to detect and locate anomalous stress height is 2.4 m and the solid pillar dimensions were about
conditions above standing coal pillars. 30 m by 21 m. Seismic imaging of the pillar-roof system was
carried out by mounting accelerometers at mid-seam level
in the pillar at 1 to 2 m intervals around the pillar. Hammer
1 CASE STUDIES
impacts on a metal plate pressed against the pillar, again at
mid-seam level, supplied the P-wave seismic energy detected
1.1 Pillar split, Cooranbong Colliery
by the accelerometer array. The pillar was then split and the
Cooranbong Colliery is located about 140 km north of Sydney. seismic imaging operation was repeated on the most accessible
Currently this mine practices partial extraction of older room- split-pillar section.
and-pillar workings within the Great Northern Seam. This Figure 1 shows the seismic image in the pre-split pillar.
seam is overlain by strong roof and underlain by a weak Velocities range from about 1400 m/s i.e. less than the coal
466

Figure 2. Post-split seismic image of coal pillar. Figure 4. Full velocity range seismic image, Line 3.
Figure 3. Historic mine plan and borehole locations.
velocity to 2200 m/s, exceeding the normal coal velocity in

this area. The higher velocities in the lower half of the pil-
lar towards CT/9 also indicate that the pillar–roof system,
through which the bulk of the first-arrival P-wave have trav-
elled, is bearing a greater load that is also not being borne Figure 5. Limited velocity range seismic image, Line 3.
symmetrically by this pillar.
The post-split seismic image of the lower velocity part of
the pillar towards C/T 8 is shown in Figure 2 This may be of the workings. None of these boreholes encountered mine
compared with the pre-split image of the same region in Fig- workings. Crosshole and surface-to borehole P-wave seismic
ure 1. Even though the seismic waves in these images have imaging was completed mainly between BH 2 and BH 3 that
not strictly travelled the same paths through the pillar-roof were about 40 m apart.
systems it is apparent that seismic velocities have increased The seismic image obtained beneath Line 3, spanning the
significantly in the post-split pillar. These have increased near full velocity range from 350 to 3000 m/s, is shown on Figure 4.
the centre of the smaller pillar by about 40% i.e. from about The lower velocities in the surface soils can be clearly observed
1900 m/s to 2700 m/s. The area imaged in Figure 2 is about in addition the abrupt velocity increase in the saturated and
35% of the original pillar area so that the stresses in the post- weathered rocks at the water table. The velocities in these
split pillar have been increased by about 90%. Also the higher rocks are relatively similar apart from a higher velocity zone
velocity regions within the smaller pillar generally co-incide in the roof above the coal seam. This represents a standing
with the initially higher velocity regions in the original pillar. high-risk coal pillar and indicates that mine workings extend
beyond the locations indicated on the historic mine plans.
Figure 5 shows the same image where the velocity range dis-
1.2 Pillar and void imaging, Oaklands Basin
played has been limited to 1500 to 2000 m/s. This emphasises
The Oaklands Basin is located about 650 km south west of the lower velocity regions within the rock section and clearly
Sydney. Until 1959 coal was extracted by the room-and-pillar shows the flooded voids, fault zone intersected in BH 1 and a
method, when the flooded underground mines were closed groundwater infiltration zone near the water table at BH3.
and abandoned. Historic mine plans indicated that a proposed The underground and borehole P-wave seismic imaging
land development was near the margin of the abandoned coal field experiments completed at two abandoned coal mines
workings at about 50 m depth. A combination of drilling and in Australia have clearly demonstrated the capacity of these
borehole seismic imaging was undertaken to locate the mar- technologies to image not only mine voids but also the more
gins of the workings and to identify any higher risk standing highly stressed regions in standing coal pillars that represents
pillars. high risk areas for surface subsidence. Increasing the focus on
Figure 3 shows a portion of the historic mine plan and these more highly stressed pillars has the potential to improve
one of the seismic scan lines (Line 3) that were completed surface subsidence assessments and the efficiencies of drilling
between three boreholes (BH 1 to 3) drilled near the margin only investigations.
467

Dynamics and blasting

Damping analysis with depth and influence of geology on the transmission of

vibrations generated by blasts
M.I. Alvarez-Fernandez & C. Gonzalez-Nicieza

Department of Exploitation and Prospecting Mines, University of Oviedo, Asturias, Spain
A.E. Alvarez-Vigil
Department of Mathematics, University of Oviedo, Asturias, Spain
Keywords: blasting, case studies, risk and hazards, mine design
At the time to evaluate the environmental impact created by is a strong damping with depth. Therefore, in order to study
blasts at quarries, in general is considered the noise effect the damage on underground structures, suitable controls were
and the vibrations, which can cause damage to neighboring carried out by means of systems set out that allow register data
buildings. of the actual structure liable to be affected.
In recent decades the execution of blast vibration control Another important effect on the transmission of vibrations
measures has become general. The reason is the increase pre- is caused by the structure of the rock mass which induces
occupation shown by the administrations over environmental anisotropy to the transmission laws. The presence of an impor-
and social impact caused by civil and mining works. tant fault crossing the quarry defines the difference between
In some Quarries the proximity the road slopes or under- two areas, so it has been measured the absorption of the blast
ground structures like tunnels, water or gas conductions, even wave trough the fault. Consequently, the blasts made at each
another underground exploitations, it turns in a problem for the side of the fault have different levels of risk respect the build-
possibles complaints to these elements. Blasts complaints to ings situated in the area, and we traduce this to directional
the rock mass are: the decrease in the resistance of the joints transmission laws.
that can instabilize wedges or blocks, permebility changes, Deep vibrations due to blasts have been measured using two
without forgetting the dinamic effect of a shock wave, that three-axial geophones for boreholes. Surface vibrations have
introduces an instantaneous force that it takes part directly in been measured using three-axial stations type ETNA of high
the balance of the forces that define the wedges and blocks dynamic range.
stability. Initially, an analysis of the vibrations generated by blasting
This paper describes a set of control of vibrations produced was performed. To do so, there was a set of measurements
by the blasts in a limestone quarry in Spain. Also it is analyzed in surface, recording 17 blasts. In order to measure deep
if they possibly can affect an underground mine sited near the vibrations two geophones were placed into two boreholes
quarry. This mine is a experimental exploitation of limestone near the underground mine and at its level. An ETNA sta-
with three pillars. tion was sited too in the underground mine and another two in
By means of a comparative analysis of the measured data different places of the quarry. These equipments registered
of vibration in surface and with depth it is verified if there 31 blasts.
471

Comparison of blast-induced ground vibration predictors in Seyitomer coal mine
E. Arpaz
Vocational school of Kocaeli, Kocaeli University, Kocaeli, Turkey
Ö. Uysal
Mining Engineering Department, Dumlupinar University, Kütahya, Turkey
Y. Tola
K. Görgülü
Mining Engineering Department, Cumhuriyet University, Sivas, Turkey
M. Çavuş
Seyitomer Coal Enterprise, Kutahya, Turkey
Keywords: blasting, blast induced vibrations, vibration predictors, peak particle velocity
In sectors such as mining, quarrying, excavating tunnels and Totally, 93 particle velocity data were taken during the blast-
pipeline trenches, blasting is indispensible. Thus, these sec- ing from the three panels (Panel-1 33 recordings, panel-2 46
tors are always prone to suffer from adverse conditions caused recordings, panel-3 14 recordings).
by ground vibrations, air shock, and flying rocks. Therefore, The data obtained from peak particle velocity recordings
when planning the blasting, besides crashing performance, were exposed to evaluation on the basis of panels the whole
size distribution, and cost and other economic issues, the recordings and peak velocity prediction methods that are
likelihood of the problems mentioned above must also be found in the literature (based on 11 different methods).
reckoned. It is of utmost importance to predict and measure From the evaluation data, it can be seen that different meth-
blast-related problems in the way to reduce environmental ods have different regression values. These variations raise
concerns that may arise. doubts concerning the reliability of the methods.
The issue is being studied extensively. Nevertheless, an As a result of the evaluation, the following were found:
agreement does not seem to have been reached on the pre-
1. All methods produced different reliability rates based on
diction and prevention of blast-induced vibrations. The lack
different panels.
of agreement can be explained by the variety of conditions of
2. Scaled distance method, which is widely accepted and
blast environments.
favored in the literature, did not produce clear rates in the
This study is intended to the results of studies that have
recordings.
been carried out at TKİ Seyitomer Coal Enterprise in order
3. No method surpassed by a high regression rate in all of the
to measure the performances of equations that can be used in
panels.
predicting blast-induced vibrations
4. The conclusion that can be drawn from the results is that
The experiments were performed in the dragline panel of
no single method provides the highest reliability rate in
Seyitomer Coal Enterprise, Kutahya, Turkey. The panel height
predicting blast-induced vibrations.
was 24 m. The geological formation over the coal in the panel
5. It can also be seen that the best method must be selected
was marl. ANFO was used for rock fragmentation. The blast
after investigating the effectiveness of all predictors based
holes were 24 m long, with 228 mm diameter. The distance
on their regression value.
between the blasting holes was 10 m. ANFO, power gel dyna-
mite and nonel capsule were used for the explosion. The delay
between each hole was 42 ms and in-hole delay, 25 ms.
472

A new testing station for rock impact on full-scale structures
M. Bost, S. Loock, L. Lenti & F. Rocher-Lacoste

Department of Geotechnical Engineering, Environment and Risks, Ifsttar, Paris, France
J. Gineys & L. Dubois

Department of Laboratory of Lyon, CETE, Lyon, France
Keywords: back analysis, dynamics, numerical and physical modeling, risks and hazards
Regarding natural disaster like rock falls, protection of civil

engineering infrastructure against impacts becomes more and
more important. The global behaviour of structures under such
conditions has been poorly investigated, for the time being, and
generally at a reduced scale. As a consequence, the predicted
behaviour of civil engineering infrastructure at full-scale is
a result from wide extrapolation. The French institute of sci-
ence and technology for transport, development and networks
(Ifsttar, previously LCPC) has decided to build a new testing
station dedicated to this topic with the target to perform tests
at scale 1.
This new testing station, located in the French Alps, allows
dropping heavy loads from the top of a cliff down onto vari-
ous structures in order to test their resistance to big impacts,
and study their dynamical behaviour at this high energy level.
The experimental area at the bottom of the cliff allows test-
ing at full-scale diverse parts of building structures as well as
protection systems under impact.
Such an experimental facility needs to be equipped with Figure 1. The tested reinforced concrete beam.
a relevant instrumentation. High capacity stress sensors,
accelerometers and high speed cameras are available for
experiments. The monitoring with these experimental devices • The second falling action aimed at studying and diagnosing
is performed at a suitable high-sample frequency for a very structural damage induced by an indirect impact on ground,
short load like an impact. A radio-controlled system allows close to the structure. The approach is based on a progres-
triggering, monitoring and dropping at the same time. sive energy release thanks to block droppings from different
Two years were necessary for the construction of the test- heights.
site and its equipment. The station is now on process. Research • The third falling action (not described here) aimed at study-
works started in October 2010 with some direct and indirect ing and diagnosing structural damage induced by a direct
impacts on a reinforced beam related to the European FP7 impact of the block onto the reinforced concrete beam. In
research project ISTIMES. this second configuration, due to the high risk of failure
For this project and in order to make measurement testing of the concrete beam, only the 2500 kg steel block was
and analysis easier, a simplified structure was so designed and used, with a progressive energy release approach (two small
built. A reinforced concrete beam has been realised. It leans heights: 1 m and 5 m).
on two pillars screwed on two foundation structures realized at During trials, rapid digital video camera and geophones at
ground level (Fig. 1). Experiments addressed the progressive ground level were used to obtain complementary information
damage, at different stages, of the concrete beam submitted to on the structural damage process.
falling blocks impacting beside the structure (indirect impact) The analysis of these first results will contribute to evaluate
and in a second phase directly onto the beam. Steel blocks the effects of rock falls on structure foundations and on the
were used and three falling actions scenarios were foreseen: structure itself investigating the degree of damage.
• The first falling action aimed at studying quarry ground Such research works will allow understanding material
by impact on the ground without the structure. These trials behaviour during rock impact, in order to mitigate damage
allow characterising wave propagation in the ground. from rock falls on civil engineering structures.
473

Mechanical behaviour of rock discontinuities under seismic conditions
L. Canelli, A.M. Ferrero, M. Migliazza & A. Segalini

DICATeA – University of Parma, Italy
Keywords: dynamics, rock joints, lab testing, numerical modelling
1 INTRODUCTION
The mechanical behaviour of rock discontinuity under mono-

tonic load is a deeply investigated topic in rock mechanics and
it is well known that such topic is conditioned by several fac-
tors among which the rock morphology, the rock strength, the
acting confining stress. When earthquake is involved, how-
ever, the behaviour is also conditioned by the loading wave
frequencies, amplitudes, acceleration etc.
For these reasons the authors have set up an experimental
device able to perform cyclic load tests on smooth and rough
discontinuities.
Figure 1. Total shear displacement versus shear stress measured
during test C3 on rough discontinuity.
2 EXPERIMENTAL ACTIVITY
Tests on smooth discontinuities were performed by applying This can be considered as an ultimate cycling angle that does
900 displacement cycles. Waveform of displacement was sinu- not coincide with the ultimate monotonic friction angle since
soidal in time with each cycle characterized by an amplitude it is conditioned by the cycle amplitude. The cycle amplitude
of 2 mm, a frequency of 0.5 Hz. and a duration of 30 minutes. involves a limited discontinuity portion and consequently it
Displacement for each cycle was 8 mm, and data acquisition determines a localized damage of the asperities that it is dif-
frequency was 5 Hz. Different confining stresses were applied. ferent by the damage caused by a larger displacement induced
Normal stress was constant throughout the test. Results have by a monotonic load.
shown that the friction angle determined by cyclic tests is Based on Homand’s formulation for the assessment of the
higher than the value obtained with the monotonic tests (40◦ damage index Dw (2000) under cyclic shear tests, a new index
and 33◦ , respectively). This difference is due to a higher veloc- has been defined as follows:
ity of load application if compared to standard monotonic
shear tests.
Following the same procedure adopted for smooth dis-
continuities tests were carried out on rough samples, during
those tests the following parameters were varied as: maxi- where σ0 is the initial compressive stress and σι is the mean
mum displacement (cycle amplitude u0 ) ranged between 1 and shear stress at the i-th cycle. C0 is the uniaxial compres-
4 mm; cycle frequency (0.013–3.9 Hz); initial normal stress sive strength of the intact material, βd is the progressive
(1–1.5 MPa) and total number of cycles (112–450, so as to degradation function defined by Belem et al. (2007) as
obtain a final total shear displacement equal to 1800 mm). follows:
Generally, tests showed a normal rapid decrease in stress in
the first cycles, reaching an asymptotic value for a total dis-
placement of 900 mm (50% of the total number of the applied
cycles). Due to the geometric anisotropy of the discontinuity
along the shear direction, the shear stress showed an asym-
metric behaviour as in figure 1 (negative values correspond to where us-tot is the total shear/relative displacement, a0 is the
upwards movement of the shear apparatus, whereas positive roughness amplitude, and ka , DRr and θs are the joint geomet-
values correspond to downwards movement). rical parameters. This formulation is valid for a cycle number
The data obtained were analysed by considering the peak N > 1.
stress values, both for upwards and downwards shearing. Let us highlight that this index is strictly related to the cycle
At the end of each cyclic shear test, the friction angle was amplitude, values of ID close to 1 indicate the reach of ultimate
computed. Its average value was 45.2◦ , higher than the one dynamic friction angle, while other values indicate a partial
obtained from cyclic shear tests along a smooth discontinuity. degradation of the asperities.
474

3 A FATIGUE BASED APPROACH FOR DAMAGE where:
QUANTIFICATION UNDER LOADS OF VARIABLE
AMPLITUDE
Starting from the obtained data, the main objective of the

Authors was to develop a new methodology that allows to with ni number of cycles in the specific acceleration range
quantify the damage of joints surface induced by a cyclic load and NTot total number of cycles.
such as the one due to seismic waves in order to estimate, Knowing the equivalent acceleration obtained by the above
through back analysis, the shear strength loss. The proce- mentioned Wohler’s formulation it is possible to extrapolate
dure hereby proposed takes into account similarities between the data using log(a)–log(Nf ) curves. Given the equivalent
fatigue in steel structures and shear strength in rock joints. acceleration value, it is possible to determine an equivalent
Fatigue is a term that applies to changes in properties which number of shear cycles and, therefore, estimate the amount
can occur in a metallic material due to repeated application of of shear strength lost or the equivalent degradation index as
stress and strains and specially this changes lead to cracking function of the number of cycles.
and failure (International Organization for Standardisation, It should be pointed out that the test conducted involves
1964). artificial replicas of natural joint, in order to fully understand
Tests results have shown that on the same joint replicas the the phenomena and to validate the described procedure it is
dynamic ultimate friction angle is reached at the same total dis- important to conduct force controlled shear tests to obtain
placement. Therefore if for metallic structure it can be inferred more reliable data.
that after an established number of cyclic loads failure may
occur, it can be also stated that after an established number of
shear cycles a discontinuity could reach the ultimate dynamic 4 NUMERICAL MODELLING
friction angle. Four main steps are required to correlate the
seismic load to the cyclic load applied during laboratory tests. A UDEC numerical model of shear test has been developed
Tau normalization. For each test shear load value was following five main steps:
normalized using the initial confinement stress, σ0 . 1. Definition of geometry and boundary condition;
Determination of the asymptotic value of τ. Limit value of 2. Definition of material and joints properties;
τ after which there is not any other increase or decrease of 3. Choice of a degradation model for the discontinuity;
shear strength. 4. Implementation of a FISH function to evaluate displace-
Log (a)–log (Nf ) diagram. As tests were conducted con- ment and shear stress of the joint;
trolling the shear displacement, the characteristic acceleration 5. Application of shear velocity to the superior block.
for each test was calculated deriving twice the displacement
and using the maximum value for one cycle. It was possible The geometry of the model has been defined in order
in this way to associate the specific acceleration to each test to reproduce the dimensions of the tested discontinuity
obtaining the equivalent to the Wöhler’s curves, well known in (8 × 5 cm). Deformable block together with a triangular mesh
fatigue studies of steels and usually described by the following of 2 cm of dimension has been chosen for this model.
equation: Continuum model of yielding joints has been used.
Results obtained from the simulation have shown signifi-
cant differences if compared to the real tests. In order to obtain
the same shear behavior, the input friction angle of the model
where C and β are constant depending on geometry and needed to be significantly increased.
material strength. In this study β parameter was correlated
with JRC of the discontinuity through back analysis. ACKNOWLEDGMENTS
Seismogram analysis and determination of equivalent
acceleration. The applied acceleration history induced by a The present research has been carried out with the contri-
seismic wave can be manipulated using some cycle count- bution of the Italian Ministry of University and Research –
ing method. Rainflow cycle counting algorithm was preferred Cofin 2006–2008 titled: “Experimental and theoretical study
because easy to implement and apply to the tests. Using the on the mechanical behaviour of rock discontinuities under
method on a design seismogram, acceleration range histogram cyclic loads”.
were obtained to be used in the following expression for the
estimation of the equivalent acceleration induced by the wave:
475

Dynamic responses of tunnels subjected to Rayleigh wave
Cheng-Hsun Chen
CECI Engineering Consultants, Inc., Taipei
Tai-Tien Wang
Tsan-Hwei Huang
Keywords: dynamics, numerical modelling, tunnelling
1 INTRODUCTION
Seismically induced tunnels damages are extensively reported

recently; resulting in the urgently need to understand dynamic
responses of a tunnel subjected to various seismic impacts.
Focusing on the dynamic response of tunnels in shallow over-
burden, this study utilizes numerical simulation to investigate
the seismically induced stress and deformation of lining.
Rayleigh wave (R-wave) is considered here as the major
seismic impact.
2 NUMERICAL MODEL AND VERIFICATION
By means of the 2D finite element method this study inves- Figure 1. Idealized scheme of numerical model.
tigates the stress increment of a tunnel lining subjected to
seismic wave caused by an earthquake. Figure 1 shows model
configuration and associated setups for boundaries and seis- tunnel vault (σNθ , σVθ and σM θ ) is recorded during simulation.
mic wave input. The incident wave is applied by a harmonic The maximum and minimum values can be determined after
sinusoidal displacement right along the left boundary. The the time history of seismically induced stress has been com-
upper boundary of the model is a free surface, while the left and pleted. These maximum magnitudes of seismically induced
right boundaries are set to be “absorbent” boundaries which axial-, shear- and flexural stress are noted as (σNθ )max , (σVθ )max
are allowed to move with minimal reflecting wave from the and (σM θ )
max , respectively. Accordingly, the maximum mag-
two sides. The lower boundary is also an absorbent one. To nitudes of seismically induced stress, for instance, the axial
enhance the precision of numerical simulation, various model stress, in the liningcan be obtained by comparing (σNθ )max all
scopes for distinct tunnel depths have been tested to minimize over the profile of the tunnel. Then the maximum seismi-
the interference from two lateral boundaries. cally induced axial stress on the whole tunnel lining can be
In semi-infinite domain in case the incident angle of an inci- determined and written as σN . So as the maximum seismi-
dent S-wave is greater than the critical angle and the distance cally induced shear- and flexural stress on the whole tunnel
between the incident position and the tunnel is large enough, lining, σV and σM , respectively. These maximum seismically
the incident S-wave will reflect in a form of R-wave in addition induced stresses are further normalized by the maximum stress
to S- and P-waves. This study generates a harmonic S-wave amplitude of the incident wave σ0 , for discussion.
with the frequencies of 1 and 3 Hz as an incident wave in the Figure 2 shows the variation of maximum seismically
left boundary of the model, R-wave is then generated within induced axial- and flexural stress in the lining for various
some region in the model. The simulation results fit in with the tunnel depths caused by an incident wave with a frequency
analytical solution and the simulation model is then verified. of 3 Hz. The maximum seismically induce stresses on the
lining increases as the tunnel depth increases significantly
3 SIMULATION RESULTS AND DISCUSSION and peaks at H /λ = 0.06. While H /λ > 0.06 the seismically
induce stresses vary minor. Figure 3 shows the variation of
The seismically induced axial-, shear- and flexural stress in the maximum seismically induced axial- and flexural stress in the
point on the lining with a θ angle clockwise measured from lining for various tunnel depths caused by an incident wave
476

Figure 2. Variation of absolute value of maximum seismically
induced stress in the lining for various tunnel depths. The frequency
of incident wave is 3 Hz.
Figure 3. Variation of absolute value of maximum seismically

induced stress in the lining for various tunnel depths. The frequency
of incident wave is 1 Hz.
with a frequencyof 1 Hz.The maximum of seismically induced

stresses dramatically increase for shallow overburden condi-
tion, and get stable while H /λ > 0.04–0.06. The simulation
results match the well-known characteristics of R-wave that
its propagation decreases as H /λ > 0.2; also implying that Figure 4. Maximum seismically induced stresses of the tunnel lin-
ing subjected to Rayleigh waves. (a) Max. axial stress envelope,
the less overburden of a tunnel, the higher damage potential
(b) Max. shear stress envelope, and (c) Max. flexural stress envelope.
of its lining structure.
Figure 4 shows envelopes of normalized seismically
induced axial-, shear- and flexural stress along whole profile 4 CONCLUSIONS
of a tunnel subjected to R-wave.The seismically induced axial-
and flexural stress caused by an R-wave with a frequency of This study utilizes numerical simulation to investigate the seis-
3 Hz have maximum values at tunnel vault, which implies that mically induced stress and deformation of lining. Rayleigh
the most possibly damaged location under such an impact.This waves with frequencies of 1and 3Hzareconsidered as the major
simulation results may partly account one of the seismic dam- seismic impacts. Results of simulation indicate that the seis-
age patterns, i.e. single longitudinal cracks along tunnel vault mically induced stresses increase as the tunnel depth decrease
in shallow sections of the San-I No. 1 tunnel during the Chi- while H /λ < 0.04–0.06. The maximum seismically induced
Chi Earthquake. Additionally, the seismically induced shear axial- and flexural stresses caused by an R-wave occurs at
stress caused by R-wave have maximum values at locations of tunnel vault, and the maximum seismically induced shear
θ = ±30◦ , which also partly elucidate another seismic damage stress occurred at locations of θ = ±30◦ . These locations with
pattern, i.e. two parallel longitudinal cracks in crown of the maximum seismically induced stressesmatch some reported
tunnel in shallow overburden sections. The simulation results seismic damage patterns.The simulation results provide
of this study further supply additional scenarios for tunnel partly evidence for seismic damage mechanism of tunnels in
seismic damages under shallow overburden. shallow overburden.
477

Dynamic tensile-splitting tests of rocks
S.H. Cho, J.R. Ahn & M.S. Kang

Chonbuk National University, South Korea
Y. Nakamura
Kumamoto National College of Technology, Japan
Y. Ogata
National Institute of Advanced Industrial Science and Technology (AIST), Japan
K. Kaneko
Graduate School of Engineering, Hokkaido University, Japan
Subject: Dynamic tensile-splitting tests
Keywords: dynamic tensile strengths, underwater explosive-driven loading apparatus
Dynamic tensile properties of rocks are a factor of consid- dynamic events and the motion analysis of the digital images
erable importance in assessing the stability of underground was available to programmable operations. Kimachi sandstone
structure under dynamic loadings such as blasting, rocks was prepared as 50 mm in diameter and 26 mm thickness.
burst, earthquake etc. It is also essential to understand The density, elastic wave velocity and elastic modulus of the
that fractures initiate and propagate under different loading sandstone were 2 g/cm3 , 2710 m/s and 6.5 GPa respectively.
conditions. Tage tuff was also prepared in the same geometry. The den-
This study used the underwater explosive-driven loading sity, elastic wave velocity and elastic modulus of the tuff were
apparatus and split Hopkinson pressure bar (SHPB) system 1.76 g/cm3 , 2380 m/s and 4.5 GPa respectively. Note that the
to measure dynamic tensile strengths of rocks. The underwa- selected rocks belong to medium and soft rock respectively
ter explosive-driven loading apparatus consist of steel frame because of the capacity of the dynamic loading apparatus. The
and an electric detonator, and impact load cells and high- loading velocities were estimated from the position of upper
speed video technique were adopted to evaluate the fracture load plates with time. The average ratio of Sd /Ss (dynamic
behavior of rocks. SHPB system involving a pulse shaping strength to static strength of static tensile strength) was 1.79
technique was used to impact the same geometry rock sam- for sandstone and 1.84 for tuff.
ples under dynamic stress equilibrium and constant strain rate. Microfocus X-ray computerized tomography (CT) system
The microfocus X-ray Computerized Tomography was used to has been used to observe microstructure of the rock test sam-
visualize dynamic damages inside the impacted samples non- ples. The sandstone samples were scanned covering entire
destructively. The dynamic fracture processes of rocks under diameter of 50 mm and height of 25 mm. Figure 4 shows
indirect tensile loading were analyzed using the dynamic a schematic diagram of the micro CT system. X-rays from
fracture process analysis (DFPA) code. microfocus X-ray generator are partially attenuated by a spec-
It is also essential to understand how fractures initiate and imen that is made to rotate in equal steps in a full circle
propagate under different loading conditions in order to deter- about a single axis close to its center. The system can achieve
mine rock breakage and fragmentation in rock blasting and 2048 × 2048 pixel reconstructions and obtain data needed for
percussive excavation. The dynamic behavior of brittle mate- 3-D image in one scanning session. The slice thickness was
rials, including rocks and concrete, show complex fracture set to be 0.076 mm for 370 slices. The cone beam can be
patterns and fracture processes when they are subjected to used to scan the entire diametric sample. The specimen table
high strain rate loading. Fractures may occur from pre-existing was placed at 400 mm away from the X-ray tube. The setting
flaws or micro-cracks under applied stress condition. The condition provided a magnification of 5.21. Figure 2 shows
present study used underwater explosion-driven loading by the 3-dimensinal volume-rendering image from 370 X-ray atten-
detonation of an electric detonator in a water filed chamber, uation slice scan images obtained by cone-beam geometry of
to fracture along the diametric tuff and sandstone specimens. X-ray tube. The open arrow shows the direction of load. The
Figure 1 shows the schematic diagram of the dynamic appa- contrast in the images shows differences in the density of min-
ratus. High-speed videography system was used to observe eral; that is, the white portions indicate the places where high
the fracture processes in the rock samples. The frame rate of density minerals exist while the black indicates low density
the video can was variable from 30 to 40,500 frames per sec- such as air. The dynamic tensile-splitting behavior of the sand-
ond. The system was capable of recording 49,152 frames of stone samples was observed by high-speed video camera and
478

Figure 2. 3-dimensional images of the impacted Tage sandstone.
fracture process zone model was used to simulate the dynamic

splitting-tensile test of rock. In order to investigate the state of
dynamic stress equilibrium through the sample in the dynamic
splitting-tensile test, the analyses, which do not allow fractur-
ing in the model, were carried out. The 5 nodes were loaded by
Figure 1. Schematic diagram of the underwater driven-loading moving the position of the nodes at 5 m/s. It is noting that the
apparatus. dynamic stress field is similar to static stress field except for
the early stage of the test, thus in order to estimate the dynamic
the applied load histories were monitored by an impact load tensile strength of sandstone samples it is necessary to pre-
cell. vent the sample from compressive fracturing before the stress
In this study, pulse shape-controlled Split Hopkinson state reaches dynamic stress equilibrium at 30 µs. Note that at
Pressure Bar system was used to impact the rock samples. vd = 5 m/s, the load increased with time and the slope of the
SHPB consists of an impact (striker) bar, an incident bar, and load-time decreased after 100 µs. This time ranged between
a transmission bar. The rock sample is placed between the the times when the opening cracks connected to compressive
incident and transmitted bars. A gas gun is used to launch the failure zones. The fracture processes regarding generation and
impact bar. For an ideal dynamic test, the sample should be interaction of opening cracks and compressive failures play an
in dynamic stress equilibrium and should deform at constant important role in the entire failure of specimen in the dynamic
strain rate before the ultimate failure. The pulse shaping tech- splitting-tensile test.
nique was employed to achieve the dynamic stress equilibrium In summary, experimental and numerical studies were car-
and constant strain rate through a test sample before failure ried out to investigate the dynamic fracture process of rock in
of the sample. The tested samples were placed on the sample the dynamic tensile-splitting test. It was revealed throughout
table and scanned by cone beam geometry X-ray generator. the high speed videography observations that the tensile crack
A volume-rendering 3D reconstruction images revealed that appeared around the center of samples contributes the entire
parallel cracks along loading axis form a shear failure fracture failure in the dynamic tensile-splitting test. The ratio Sd /Ss
pattern. The shear fractures which are decomposed into ten- of static tensile strength and dynamic tensile strength was
sile cracks parallel to the loading axis, oblique wing-cracks 1.79 for sandstone and 1.84 for tuff. Shape-controlled Split
and coalescence cracks. The number of cracks increased with Hopkinson Pressure Bar system was also used to impact the
increasing impact velocity and however a number of cracks rock samples. X-ray CT images of the tested sandstone sam-
which are not connected each other are visible at lower impact ple showed that there are many tensile stress-induced cracks
velocity. With increasing impact velocity, parallel cracks along except for main tensile crack joining the load plates. The frac-
loading axis form multi shear fractures and connect both ture process analyses revealed that it will take 30 µs or more
the end side of the cylindrical samples. Ultimately, the shear to achieve the dynamic stress equilibrium though the sample
cracks led to finger-like fragments. during the test of sandstone.
The dynamic fracture process analysis (DFPA) code which
employs the microscopic strength inhomogeneity and the
479

Characterization of dynamic rock fracture parameters using Notched Semi-circular Bend

(NSCB) method and Cracked Chevron Notched Brazilian Disc (CCNBD) method
F. Dai & K. Xia

Department of Civil Engineering and Lassonde Institute, University of Toronto, ON, Canada
Keywords: Dynamics, Lab testing, Rock failure, Rock properties
1 INTRODUCTION
Dynamic fracture plays a vital role in various geophysical pro-

cesses and engineering applications involving rocks. Accurate
determination of dynamic fracture parameters is crucial for
understanding mechanisms of dynamic fracture and is also
beneficial for hazards prevention and mitigation. The Interna-
tional Society of Rock Mechanics (ISRM) also proposed short
rod (SR) and chevron bending (CB) tests in 1988 (Ouchterlony
1989) and cracked chevron notched Brazilian disc (CCNBD)
in 1995 (Fowell et al. 1995). In this study, two dynamic frac-
ture techniques are proposed using notched semi-circular bend
(NSCB) and cracked chevron notch Brazilian disc (CCNBD)
specimens loaded with a 25 mm split Hopkinson pressure bar
(SHPB) system. A laser gap gauge (LGG) is used to monitor
the crack surface opening displacement (CSOD) of the spec-
imen during SHPB testing. With a pulse shaping technique
employed, the dynamic forces on both ends of the sample Figure 1. The SHPB system and the NSCB and CCNBD specimen.
are balanced and the fracture initiation toughness is calcu-
lated with a quasi-static data analysis. Based on the first law
of thermodynamics, the fracture energy and the propagation
toughness are also obtained.
2 EXPERIMENTAL SETUP
where Pmax is the peak load, S is the distance between the
A 25 mm diameter SHPB system is used in this study (Fig- two loading pins, a is the crack length, B is the thickness and
ure 1). Assuming one-dimensional stress wave propagation, R is the radius of the specimen. The key factor Y in the two
the forces on the incident end (P1 ) and transmitted end (P2 ) equations can be calibrated using finite element analysis.
of the sample can be calculated (Kolsky 1953). The fracture energy and fracture toughness can be quanti-
The LGG system is developed to monitor the CSOD. By fied using the energy conservation principle. The total energy
recording this voltage output of the photodiode, the gap open- absorbed by the specimen (W) can be estimated from the
ing distance (i.e., CSOD) ant thus the opening velocity can energy difference during the stress wave propagation. Part
be obtained with pre-calibration. Laurentian granite (LG), is of this energy creates new crack surfaces, the total fracture
selected in this study to demonstrate the two methods. The energy (WG ); the remaining is the residue kinetic energy (K)
geometry of the NSCB and CCNBD specimen is shown in of cracked samples. With the aid of LGG, the total rotational
Figure 1 as well. kinetic energy for NSCB and CCNBD tests can be measured.
Thus, the average propagation fracture energy is Gc = WG /Ac ,
where Ac is the area of the crack surfaces created. The average
3 MEASUREMENT PRINCIPLES dynamic propagation fracture toughness is:
The mode-I initiation fracture toughness can be calculated
using equation 1 for NSCB specimen and equation 2 for
CCNBD specimen following ASTM standard E399-06e2.
where E and ν are the Young’s modulus and Poisson’s ratio of

the specimen respectively.
480

Figure 2. The variation of propagation fracture toughness with
fracture velocity. Figure 3. Comparison of the initiation toughness from the CCNBD
method and the SCB method.
4 RESULTS the dynamic CCNBD method are comparable with those from
dynamic NSCB tests as shown in Figure 3. We are thus con-
The pulse shaping technique has been employed for all fident that the measured fracture initiation toughness results
dynamic tests and the dynamic forces on both sides of the from both dynamic tests are reliable. It is noticed that we can-
samples are almost identical (Frew et al., 2000). The temporal not compare the fracture energy values obtained from both
derivative of the CSOD history is the crack surface opening methods. This is because of the difference in failure mode
velocity (CSOV) history. CSOV increases with time and then of the two specimen configurations: unstable fracture for
approaches a terminal velocity at a turning point. The separa- NSCB tests and stable-unstable fracture transition for CCNBD
tion velocity of the two fragments (normal to the bar axis) is tests. The fracture energy or fracture propagation toughness
approximately the terminal velocity in CSOV, and doubles the is strongly influenced by the mode of failure. As a conse-
fragment velocity, from which the crack propagation time can quence of this difference in failure modes, the variation of
be calculated. Given the crack distance is known for this test, fracture velocity observed in dynamic NSCB tests is missing
the average crack growth velocity can thus be determined. for the dynamic CCNBD results, where the stable and unstable
Both of the measured initiation and propagation fracture fracture velocities are almost constant.
toughness from NSCB tests increase almost linearly with
increasing loading rates. The fracture loading rate is deter-
mined from the slope of the loading curve before fracture 5 CONCLUSIONS
initiation. The propagation fracture toughness also increases
with the fracture velocity (Fig. 2). At the highest fracture We proposed two SHPB fracture testing techniques with
velocity (∼850 m/s), the fracture toughness is 9.48 MPa m1/2 , NSCB and CCNBD specimens for determining the dynamic
about twice of those observed at slower fracture velocities mode-I fracture parameters of rocks with the aid of a laser
(near 300 m/s). gap gauge system. The two methodologies are demonstrated
For SHPB-CCNBD tests, a strain gauge is mounted on the with Laurentian granite. Both the initiation and propagation
sample surface to detect the fracture initiation and propaga- toughness of studied rock are loading rate dependent. The
tion. The dynamic fracture process of the CCNBD specimen propagation toughness is larger than the initiation toughness
in a SHPB test can be divided into four stages. Stage I is the in a given test. These techniques are readily implemented
elastic deformation of the CCNBD specimen. At the end of the and thus can be applied to investigating dynamic fracture
stage I, the crack initiates from the notch tip, and propagates mechanics of rocks.
until the propagating crack reaches the critical crack length,
am (stage II). In stage II, the crack propagates stably while in
stage III, the crack propagates in an unstable manner. Finally, REFERENCES
the sample is cracked completely into two half fragments in
stage IV. In view of these four stages, both stable and unsta- Fowell, R.J., et al. 1995b. Suggested method for determining mode-I
ble crack propagation velocities can be estimated accordingly. fracture toughness using cracked chevron-notched Brazilian disc
(CCNBD) specimens. IJRMMS. 32: 57–64.
The measured average stable and unstable fracture veloci-
Frew, D.J., et al. 2002. Pulse shaping techniques for testing brit-
ties are shown to be weakly dependent on the loading rate. tle materials with a split Hopkinson pressure bar. Experimental
The unstable fracture velocity is always larger than the stable Mechanics. 42: 93–106.
fracture velocity for each test (two to three times). Kolsky, H. (1953), Stress waves in solids, 212 pp., Clarendon Press,
The measured initiation and propagation fracture tough- Oxford.
ness from CCNBD tests also increase with increasing loading Ouchterlony, F. 1988. Suggested methods for determining the
rates. The measured fracture initiation toughness values from fracture toughness of rock. IJRMMS. 25: 71–96.
481

Numerical simulation of train-induced vibrations in rock masses
Andreas Eitzenberger, Ping Zhang & Erling Nordlund

Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, Sweden
Keywords: dynamics, numerical modelling, rock joints
1 INTRODUCTION depth and the unit weight of the overlying rock, and the ratio
between horizontal and vertical stress was set to two.
A train moving along an underground railway will generate The bulk modulus was set to 30 GPa, the shear modulus to
vibrations in the rail and the underlying track structure. These 18 GPa, and the density to 2700 kg/m3 . The friction angle of
vibrations will radiate into the surrounding ground as elas- the discontinuity was set to 30◦ .
tic waves and will, in densely populated areas, reach nearby The dynamic load used in the analyses was one period of
buildings and its residents. a sinusoidal wave with a duration of 0.01 sec and a particle
The propagation of train-induced vibrations is complex due velocity of 100 mm/s at a frequency of 100 Hz.
to the scale and the many uncertainties involved. In the past, The UDEC-model was 100 m high and 100 m wide. The
empirical methods were used to analyse problems related to tunnel was shaped as a square with a height and width of
train-induced vibrations, while today the analyses are per- 10 m. Due to the tunnel shape, symmetry was used in order to
formed with the aid of numerical models in combination with reduce the model size. The damping ratio was set to 2%.
field measurements. During the analyses history points on the ground surface
A common assumption in numerical analyses is to repre- were used to monitor the maximum and minimum (most neg-
sent the ground by a homogenous and isotropic material. For ative) vibration levels in y-direction (vertical). The largest
situations where the ground consists of clay and soil assuming absolute value was determined as the vibration level at each
homogenous and isotropic material may be correct, but for a point.
discontinuous rock mass it is not. The vibration levels on the ground surface near a tunnel was
In a rock mass the presence of discontinuities will have a analysed for the following parameters:
significant impact on the transmission of waves, since a wave
• overburden – four different overburdens were studied; 10,
propagating across a discontinuity will be partly reflected and
25, 40, and 55 m, respectively. No discontinuities were
partly refracted. How much that is refracted and reflected,
included in the model during these analyses.
respectively, depends mainly on the normal and shear stiff-
• position of a single horizontal discontinuity near the
ness of the discontinuity and wave frequency, but also on
tunnel – three different positions were studied: 5 m above
the angle of incidence and the number of discontinuities. The
the tunnel roof, 5 m below the tunnel floor, and 5 above
propagation of waves in discontinuous rock masses has mainly
the tunnel floor (i.e. middle of the wall). The normal and
been studied with regard to blasting and explosions, i.e. high-
shear joint stiffness was 1 and 10 GPa/m and the overburden
frequency waves. Hence, there is lack of knowledge regarding
was 55 m.
the influence of discontinuities on low-frequency wave prop-
• stiffness of discontinuity – three values of the normal and
agation in discontinuous rock masses (i.e. for the frequency
shear stiffness were studied; 1, 10 and 100 GPa/m. The joint
range associated with train-induced vibrations).
was positioned in the middle of the wall and the overburden
This paper addresses how (i) overburden, (ii) position of
was 55 m.
a horizontal discontinuity in relation to the tunnel, and (iii)
the normal and shear stiffness of a discontinuity influence the
wave propagation in a rock mass near a tunnel. The model,
methodology, and input data are described as well as the results 3 MODELLING RESULTS AND DISCUSSION
and the conclusions of our findings.
Figure 1 shows the vertical vibrations levels at different posi-
tions on the ground surface for different the four overburdens.
2 MODEL SETUP AND ANALYSES The highest vibration levels occur above the tunnel (0 m). The
vibration levels decreases with increasing horizontal distance
Wave propagation analysis was conducted using the Univer- from the tunnel until they reach a minimum and then increases
sal Distinct Element Code (UDEC). Input data used in the to levels similar to those observed above the tunnel. It is evi-
analyses were (i) rock stresses, (ii) rock mass properties, (iii) dent that the overburden has a great impact on the vertical
mechanical properties of discontinuities, and (iv) dynamic vibration levels on the ground surface and that increasing
load. overburden will reduce the vibrations.
The virgin rock stresses were based on Swedish conditions, The distance from the centre of the tunnel to the point
where the vertical stress was assumed to be the product of the where the minimum occurs on the ground surface increases
482

It can be seen that if a horizontal discontinuity is located
in the middle of the wall the vibration levels decreases,
while if a horizontal discontinuity is located below the tun-
nel the vibration levels increases compared to the model
without discontinuity. If a horizontal discontinuity is located
above the tunnel, the vibration levels decrease near the tun-
nel while becoming higher than the reference levels (i.e. no
discontinuity) at greater distances.
Another observation is that the difference in vibration lev-
els between models with and the model without discontinuity
is small for a normal and shear stiffness of 10 GPa/m. By
lowering the stiffness from 10 to 1 GPa/m the influence of
the discontinuity becomes more evident. The results implies
that the discontinuity prevents propagating waves from pass-
ing through, especially when the normal and shear stiffness is
Figure 1. Vertical vibration levels on the ground surface for low, i.e. <10 GPa/m.
different overburdens.
4 CONCLUSIONS
With the aid of numerical simulation of low-frequency vibra-

tions the influence of some rock and discontinuity parameters
has been studied. Based on the results presented in this paper,
the following conclusions can be made:
• The overburden has a large impact on the vibration levels
on the ground surface. The vibration levels will decrease
with increasing overburden.
• The addition of a discontinuity in the vicinity of a tunnel
will influence the vertical and horizontal vibration levels
on the ground surface. A horizontal discontinuity posi-
tioned above the dynamic source (e.g. above tunnel or in
the wall) will reduce the vibration levels while a discon-
Figure 2. Vertical vibration levels on the ground surface for three tinuity located below the dynamic source (e.g. below the
different positions of a horizontal discontinuity, where the normal tunnel) will increase the vibration levels.
and shear stiffness is varied and the overburden is 55 m.
ACKNOWLEDGMENTS
with increasing overburden. The position of the point seems to
coincide fairly well with the depth at which the tunnel floor is This work was funded by the Swedish Transport Administra-
located, and can be estimated using an angle of approximately tion (Trafikverket), formerly known as Swedish Rail Admin-
45◦ from the corner between tunnel floor and wall. istration (Banverket), and Centre of Advanced Mining &
Figure 2 shows the vertical vibrations levels monitored at Metallurgy (CAMM) at Luleå University of Technology.
different positions on the ground surface when a horizontal
discontinuity is placed either 5 m above, 5 m below, or at the
middle of the wall of the tunnel.
483

Effect of micro-defects and macro-joints on stress wave propagation

in rock mass
L.F. Fan & F. Ren

School of Civil and Environmental Engineering,
Nanyang Technological University, Singapore
G.W. Ma
School of Civil and Resource Engineering,
The University of Western Australia, Crawley, WA, Australia
Keywords: Dynamics Rock dynamics; Stress wave propagation; Rock mass; viscoelastic behavior
A rock mass is a material with various discontinuities different the dynamic complex viscoelastic modulus (Figure 2) of the
from other traditional structural materials in civil engineering. sedimentary rock was also derived. Measured stress waves
Ideally, a rock mass is composed of a system of rock blocks show that micro-defected rock is viscoelastic.
separated by discontinuities forming a material in which all The effect of micro-defect on the viscoelastic response
elements behave in mutual dependence as a unit. There- of rock is further investigated using numerical manifold
fore, the dynamic properties of rock mass are characterized method (NMM) (Figure 3). Complex viscoelastic modulus of
by mechanical and geometrical properties of discontinuities.
Investigating the mechanical effect of rock blocks and discon-
tinuities on the stress wave propagation is important in solv-
ing problems of seismology, mining, tunneling, underground
constructions, etc.
An investigation of the effects of discontinuities on stress
wave propagation through rock mass is carried out. The dis-
continuities in the rock mass are divided into two groups. The
primary discontinuity set is the one with relatively large or
the same order spacing to the wavelength, known as “macro-
joint”, while the secondary discontinuity set, characterized by
high density and relatively small spacing to the wavelength,
known as “micro-defect”.
The rock with micro-defects is modeled as an equivalent
viscoelastic medium. The effect of micro-defects on the wave
propagation is evaluated by an impact test on defected sedi-
mentary rock using a pendulum (Figure 1). The sedimentary
rock bar was cored from an underground cavern construction
site. The strain pulses were measured at the middle of the bar Figure 2. Complex viscoelastic modulus of sedimentary rock.
(Figure 1). Wave propagation coefficients (attenuation coef-
ficient and wave number) and phase velocity were obtained
from the frequency spectra of the recorded strain pulses and
Figure 1. Scheme for impact tests. Figure 3. NMM model for simulation of micro-defected rock.
484

Figure 5. Scheme of extended displacement discontinuity method.
Figure 6. Transmission coefficient for wave propagation through

rock mass with different joint spacing.
by the crack length, quantity and stiffness, and the model

satisfies the uniqueness principle to the loading. The results
also show that NMM is suitable in solving static and dynamic
problems in highly defected rock, which gives sufficient
accuracy in the simulation of stress wave propagation.
The macro-joints in the rock mass are modeled explicitly as
physical discontinuities. Displacement discontinuity method
(DDM) was commonly used to investigate the effect of macro-
joint on the wave propagation. Although the deformational
behavior of joints can be considered as linear or nonlinear, the
rock at opposite sides of the joint is intact and linear elastic.
Therefore, the effect of the inherent rock viscosity on wave
propagation is always ignored in such methods. An extended
displacement discontinuity method (EDDM) (Figure 5) was
proposed in the present study to evaluate the stress wave
propagates across the joint in the viscoelastic rock mass.
The transmission coefficient and reflection coefficient for the
stress wave propagation across the joint in the viscoelastic
rock mass are analytically deduced.
A numerical example is performed using present equivalent
medium-discontinuity combination approach (Figure 6). The
Figure 4. Effect of micro-defect on viscoelastic response of rock results show that the overall mechanical response of rock mass
mass. is viscoelastic, both the micro-defects and macro-joints have
significant effects on the wave propagation through rock mass,
sedimentary rock was obtained and the effect of micro-defect the NMM can be used to analyze the effect of micro-defect
quantity, stiffness and length are discussed (Figure 4(a), (b), (c) on the wave propagation efficiently, and the approach using
and (d)). The NMM simulation results indicate that dynamic equivalent viscoelastic medium with explicit macro-joint is
viscoelasticity of micro-defected rock is significantly affected valid in the dynamic analysis of complex jointed rock system.
485

Progress in study of rockburst experiments and control countermeasures
M.C. He, X.N. Jia, W.L. Gong & G.J. Liu

School of Mechanics and Civil Engineering, China University of Mining and Technology, Beijing, China
J.T. Zhang
School of Mechanics and Civil Engineering, China University of Mining and Technology, Beijing, China
Keywords: rock bursts; lab testing; bolts and anchors; rock support
1 INTRODUCTION
For opening up a new way for the rockburst studies involving

in physical modeling the rockburst phenomena, categorizing
rockburst phenomenon according to the Hoek-Brown crite-
rion and establishing the criterion, from an energy perspective,
specifically for each type of the rockburst, the research team
led by Prof. He Manchao have devoted to the innovative work
towards the goals over years. This article introduces the state-
of-the-art advancement of anchor bolt technology, i.e. the large
deformation cable with constant-resistance (LDCR), devel-
oped by the prof. He Manchao in 2009 and patented both
at home and abroad. The newly developed LDCR have the
ability to accommodate larger deformations of the adjoining
Figure 1. Sketch of dowel steel and pressure head.
rock masses of a roadway tunnel at great depth and output a
constant-drag resistance in response to the external perturbed
forces. The energy release characteristics with different samples are
also analyzed. AE energy release curve can be divided into two
types based on the experimental results, namely single steep
2 NOVEL ROCKBURST TEST SYSTEM unloading type, multi-line steep unloading type. The first type
normally associated with continuous non-linear increase after
Physical modeling of rockbursts under the conditions found unloading corresponding to instantaneous rockburst; The sec-
in deep underground engineering is one of our major goals ond type usually associated with several damage production
in our research. Based the original idea and initial design by and lead to sudden rockburst or corresponding to the rockburst
Prof He Manchao, the Deep Underground Rockburst Ana- of several scales.
logue Test Machine (DURATM) were further developed by
Prof. He and his research team at China University of Mining
and Technology, Beijing (CUMTB) at 2004 (He et al., 2007). 4 CONTROL MEASURES FOR ROCKBURST
Figure 1 shows the Sketch of dowel steel and pressure head.
The experimental results demonstrate that the novel bolt,
LDCR, can accommodate large deformations of the in situ
3 AE CHARACTERISTICS OF ROCKBURST rock masses and also absorb the deformation energy of the
surrounding rock masses. Therefore, LDCR can prevent the
Figure 2 shows the whole frequency bandwidth which large-scale deformed surrounding rock masses from breaking
is divided into four sub-bandwidths, i.e., 30–150 kHz, and failing. The novel bolt suited for the large-scale road-
150–230 kHz, 230–330 kHz, and 330–400 kHz respectively way and tunnels at great depths and deep ground engineering
according to the frequency distribution features of each projects.
waveform. The four sub-bandwidths were highlighted in the
following four colors: the purple for 30–150 kHz (low fre-
quency, LF), the green for 150–230 kHz (low-and-middle 5 CONCLUSION
frequency, LMF), the blue for 230–330 kHz (middle-and-
high frequency, MHF), the red colors for 330–400 kHz (high (1) The Deep Underground Rockburst Analogue Test Machine
frequency, HF). found under the conditions in deep underground engineering
486

Figure 2. Relationship between AE waveform files sequence and frequency.
include the loading/unloading device, high speed data acquisi- (LDCR), can accommodate the large deformations of the rock
tion system and AE detecting system. The loading/unloading masses and absorb the impact by the sudden release of the
device of the DURATM is composed of the main stand; During rockburst energy while output a constant resistance subjected
the test, one surface of the specimen can be unloaded immedi- to the external loading. Therefore, LDCR can prevent a large-
ately from the true triaxial compression condition, simulating scale deformed rock mass from breaking and failing, and suit
the stress condition for rock mass at the free excavation bound- for preventing and mitigating the engineering disasters in deep
ary in underground excavations. (2) The distribution law of the ground such as coal bumps and rockbursts.
dominant frequency bandwidth (DFB) for different rock spec-
imens under varied stress paths during rockburst simulation
tests was obtained based on the time-frequency analysis and ACKNOWLEDGEMENTS
the discrete Fourier transformation (DFT) from the AE wave-
form data. The rock specimens with different lithology under This work was supported by the National Basic Research
rockburst tests have their unique frequency features of the AE Program of China (2006CB202200), the Major Program of
signals, including the DFB distribution in the LF, LMF, MHF, National Natural Science Foundation of China (50490270),
and HF bandwidths. (3) Laboratory experiments and field and Innovative Team Development Project of the Ministry of
engineering applications demonstrate that the novel anchor, Education of China (IRT0656).
i.e. the Large Deformation Cable with Constant Resistance
487

Evaluation of dynamic behaviors of bedrock foundation based on

FEM and DEM simulations
Y. Jiang, L. Yang, Y. Tanabashi & B. Li

Faculty of Engineering, Nagasaki University, Nagasaki, Japan
S. Li
Research Center of Geotechnical and Structural Engineering, Shandong University, Jinan, China
Keywords: dynamics, numerical modelling, rock joints, stability analysis, rock mass, rock slopes and foundations
Seismic behaviors of the bedrock foundation concern the sta-

bility and safety of nuclear plants. Simultaneously, there are
lots of discontinuities like joints and faults per-existing in
bedrock, weakening the mechanical properties of bedrock.
The evaluation of seismic behaviors of bedrock with pre-
existing discontinuities remains as one of the most funda-
mental and promising problems. The dynamic FEM (finite
element method) was utilized commonly in analyzing the
seismic responses of bedrock. However, the DEM (distinct
element method) has a better capability of simulating the slid-
ing and separation of the pre-existing discontinuities in the Figure 1. Comparison of response waves obtained from FEM and
bedrock, which affect the propagation of seismic waves. In this DEM simulations (depth: 30 m).
study, the FEM and DEM simulations were adopted to inves-
tigate the seismic behaviors of the bedrock under a nuclear
power plant located in Japan, and the differences between aspects: the amplitude variation at peaks and troughs of waves
those two methods were illuminated. is more obvious; the time corresponding to peaks and troughs
The bedrock of the nuclear belongs to Mesozoic cumulates, exhibits deviation. With the upward propagation of seismic
containing three major faults with the dip angles of 72◦ , 56◦ waves, the amplitudes of response waves decrease. In addi-
and 63◦ respectively. The EW component of the seismic wave tion, the time deviation for waves reaching their peaks was
obtained at the Ojiya observation site during the M6.8 Niigata also obvious.
earthquake (July 16, 2007) was selected in this study. The code To make a further comparison of the numerical results
of k-SHAKE was used to investigate the seismic motion at the from FEM and DEM simulations, the response waves at the
deep rock formation under the proposed nuclear power plant same monitoring points were analyzed. The results indicate
with a depth of 200 m. A period of 10 seconds (20.8–30.8 s) in the amplitudes of response waves from DEM simulations are
the velocity and acceleration waves at the deep rock formation smaller than those of waves from the FEM simulation (see
with the maximum amplitude were selected as the input waves Fig. 1). In addition, the difference in wave amplitudes becomes
to the DEM and FEM models, respectively. more obvious when the depth of monitoring point decreases.
Numerical simulations on the seismic behaviors of the Moreover, the divergence of waveforms obtained with DEM
bedrock were carried out, by using the FEM code of Super- is more obvious.
FLUSH/2D and the DEM code of UDEC. The FEM results The differences between the FEM and DEM simulation
indicate that despite of the difference in depth, the response results were resulted from the treatment of discontinuities.
waveforms at different monitoring points with varying depth The FEM can not present the large deformation behaviors
exhibit good consistency. During the propagation process of of faults, underestimating the fault’s weakening effect on the
seismic waves from the deep rock formation to the ground propagation of seismic waves. Relatively, the DEM with the
surface, the amplitudes at wave’s peaks and troughs increase. capability of simulating the sliding and separation of faults can
Therefore, the bedrock has an amplification effect on the evaluate the effect of discontinuities well in seismic simula-
amplitudes of the seismic waves in the FEM model. For the tions. Therefore, the DEM can be regarded as a better method
DEM results, the velocity waves at different monitoring points than the FEM in seismic response evaluations of bedrock with
have similar waveforms comparing with the input waveform. pre-existing discontinuities. The main results may provide
However, the response waves from the DEM simulation show suggestions to the numerical studies and safety evaluations
different characteristics with the FEM results, mainly in two of bedrock subjected to seismic loads.
488

Application of discrete element method for validity evaluation of rock dynamic fracture
toughness measured by semi-circular bend technique
T. Kazerani & J. Zhao

Laboratoire de Mécanique des Roches (LMR), Ecole Polytechnique Fédérale de Lausanne (EPFL), Switzerland
Keywords: dynamics, numerical modeling, rock failure, rock properties
1 INTRODUCTION the specimen surface friction influences the SCB test mea-
surements. The results of the RD model, labeled as RD (φ: 5),
This study investigates the influence of fracture constitutive suggest the necessity of introducing rate-dependency into the
law, loading rate and frictional boundary condition on the contact law, as the RD model can reproduce the test data.
rock dynamic fracture toughness, KICd , measured by the semi- Figure 3 presents the variation of predicated stress intensity
circular bend (SCB) method. Using UDEC, rock material is factor, KId , versus time and the observed fracture patterns.
modeled as an arbitrarily produced assemblage of elastic trian- Like the laboratory observations, fracturing begins from the
gular particles interacting at their edges. The size of particles is crack tip, then propagates straight, and eventually causes
taken as the average grain size of rock. The interface between the specimen to break into two pieces. However, the frac-
the particles is numerically represented as a contact surface ture starts propagating 55 µs after √ the loading bar touches
formed between two particle edges. the specimen, when√ K Id = 1.40 MPa m. In fact, the peak of
KId (=2.40 MPa m), which is experimentally reported as the
dynamic fracture toughness, does not correspond with the
2 CONTACT STRESS-DISPLACEMENT LAWS instant of fracture initiation. This is a point of concern as
it means that the dynamic fracture toughness of material is
The contact models developed are presented in Figure 1, overestimated by the test.
where σ = stress acting on contact surface, δ = contact dis-
placement, kt = contact initial stiffness, tc = contact strength,
δct = contact critical tensile displacement beyond which cohe-
sive softening happens, and δut = contact ultimate displace-
ment at which contact entirely loses its strength. Two models
are adopted: in the rate-independent (RI) model, δut is held
fixed, whilst in the rate-dependent (RD) model, it is assumed to
change with the contact opening rate through being multiplied
by a factor of α.
3 CALCULATION RESULTS
Two RI simulations are carried out that are labeled as RI (ϕ:

0) and RI (ϕ: 5). The values in the parentheses indicate the Figure 2. Dynamic fracture toughness versus loading rate for RI
friction angle assumed for the interface between the loading and RD models.
plates and the specimen. The results are presented in Figure
2. The difference between the fracture toughness predicted by
the RI model and the test data is apparent. It is also seen that
Figure 1. Contact behavior in RI and RD models. Figure 3. Stress intensity factor versus time.
489

The mechanism behind seismic damage to railway mountain tunnels and

assessment of their aseismic performance
Y. Kojima, K. Yashiro & T. Funahashi

Tunnel Engineering Laboratory, Structure Technology Division, Tokyo, Japan
H. Miyabayashi & T. Iura

Design and Technology Department, Japan Railway Construction, Transport and Technology Agency, Yokohama, Japan
T. Asakura
Department of Civil and Earth Resources Engineering, Kyoto University, Kyoto, Japan
Subject: Others – Earthquake
Keywords: case studies, lab testing
1 INTRODUCTION a model experiment was conducted to clarify the damage

mechanism and assess tunnel aseismic performance. Accord-
Previous studies have shown that mountain tunnels may some- ingly, the experiment was conducted using a two-dimensional
times suffer damage such as cracking and spalling caused shear box. It was confirmed that earthquake damage such as
by flexural compression failure depending on the magnitude that seen in this tunnel can be reproduced by applying shear
of an earthquake and the distance from its seismic source. deformation on to a tunnel.
The mechanism behind earthquake damage to tunnels has
not yet been fully analyzed, and there are many unknown
conditions related to such damage. This paper outlines the 4 MECHANISM BEHIND SEISMIC DAMAGE AND
results of a model experiment conducted to clarify the damage TUNNEL ASEISMIC PERFORMANCE IN GROUND
mechanism and levels of aseismic performance of tunnels. WITH POOR GEOLOGICAL CONDITIONS
Presumable causes of Type II damage include the fact that

2 ANALYSIS OF EARTHQUAKE DAMAGE CASES ground with poor geological conditions is generally soft,
which is likely to exacerbate the deformation caused by
According to case studies, the modes of earthquake damage earthquake motion, and that loads such as loosening ground
to mountain tunnels can be classified into the three types pressure and squeezing ground pressure may have also been
(I to III). present before the earthquake. To investigate Type II damage
Type I: Cracks at the arch shoulder part characterize damage for tunnels in ground with poor geological conditions, a model
to shallow tunnels. Generally, such tunnels are often found in experiment was conducted to clarify the damage mechanism
soft ground. and assess tunnel aseismic performance.
Type II: Flexural compression failure and spalling at the The experiment was conducted using the loading-model
crown characterize damage to tunnels in ground with poor experimental apparatus. It was confirmed that earthquake
geological conditions. damage such as that seen in this tunnel can be reproduced by
Type III: Damage caused by fault displacement, includ- applying displacement or loading in the horizontal direction
ing lining fracture resulting from local forced displacement to a tunnel with a void above the lining.
accompanied by fault displacement.
The experiments were conducted with an emphasis on
seismic damage Type I and Type II in order to reproduce 5 CONCLUSION
earthquake-related failure and investigate the related damage
mechanism. The experiments were conducted with an emphasis on seis-
mic damage to mountain tunnels in shallow cover and in
ground with poor geological conditions in order to reproduce
3 MECHANISM OF SEISMIC DAMAGE AND earthquake-related failure and investigate the related damage
ASEISMIC PERFORMANCE IN SHALLOW mechanism. The experiments clarified the aseismic perfor-
TUNNEL COVER CONDITIONS mance of lining, and revealed that a tunnel with an invert and a
sound lining with no void above it and no lack of thickness sus-
It is presumed that earthquake-related shear deformation in the tains very little damage, while a tunnel without an invert, with
ground induces shear deformation in the tunnel, which causes defects and under conditions of local displacement is likely to
an increased bending moment at the lining arch shoulder part sustain damage such as heaving, flexural compression failure
and results in failure. For Type I damage to shallow tunnels, and spalling.
490

Destress rock blasting as a rockburst control technique
Petr Konicek, Petr Konecny & Jiri Ptacek

Institute of Geonics, Academy of Sciences of the Czech Republic
Keywords: blasting, coal mines, rockstress, rockburst, case study
1 INTRODUCTION 4 DESIGN OF DESTRESS BLASTING
Rockburst in the Upper Silesian Coal Basin (USCB) has rep- Destress rock blasting is designed according to natural and
resented a very dangerous phenomenon for a long period mining conditions. Boreholes with a diameter of 75–105 mm
of time (e.g. Staube 1972, Petuchov & Zamarski 1990, are usually drilled from the gateways, with a borehole incli-
Takla & Ptacek 1990, Holecko et al. 1999, Konecny 2005, nation of up to 30◦ and length of 30–120 m. The spacing of
Dubinski & Konopko 2000, Drewiecki & Kabiesz 2008). The parallel boreholes is usually from 5 to 12 m. Explosives are
rockburst problem relates primarily to the existence of compe- pneumatically charged into boreholes (in cartridges). The total
tent rocks in rock mass and the irregular stress fields induced amount of explosive charges ranges from 400 to 4781 kg. The
by long-term excavation. Destress rock blasting in overlying predominant charge percentage of the length of boreholes
competent rocks is a very important rockburst control tech- varies from 45 to 85% (increasing with an increase in hole
nique especially in the most complicated stress conditions. length).
The presented system of destress rock blasting as a modifica-
tion of the South African method (e.g. Comeau et al. 1999) is
a unique European system used in the Czech part of the USCB 5 MAIN GOALS AND THEIR EVALUATION
in difficult stress conditions in rock mass.
Properly designed and realized destress rock blasting (location
and spacing of boreholes, diameter of boreholes, length of
2 NATURAL AND MINING CONDITIONS charge, number of fired boreholes, total explosives charge,
etc.) reduces the strength and deformation properties of rocks
The USCB is a carboniferous multiseam sedimentary deposit. and releases a high stress concentration in the area surrounding
The sequence of competent sandstone beds and other much the blasting.
softer rocks and the network of faults form a very complicated Stress release is evaluated according to calculated seismic
set of natural conditions which results in an irregular natural effect (SE). The SE of destress rock blasting is the ratio of
stress state. seismic energy released in the rock mass when blasting to
Coal mining in the USCB started in the eighteenth century. the considered energy of the particular detonated charge. The
The irregular stress fields induced by long-term excavation are evaluation of stress release from the rock mass due to SE was
very complicated and in many cases they are very dangerous proposed by Knotek (1985) and verified by Konicek (2009).
for mining (increasing the rockburst risk).
6 CONCLUSIONS
3 CONCEPTUAL MODEL AND HISTORY OF
More than 30 years of experience with destress rock blasting
DESTRESS BLASTING
in hardcoal longwall mining in the Czech part of the USCB
have shown that it is a very useful method of rockburst control
A rockburst model has been formed since the second half of
in difficult stress conditions.
the twentieth century. The predominant types of rockbursts in
More than 2000 cases of destress rock blasting have been
the Czech part of USCB are rockbursts with a focal area in the
realized since the 1980s. Many of the cases which have been
main roof which originates from the high stress concentration
carried out show high SEs which document the required stress
in competent rocks. Destress rock blasting seems to be the
release in the rock mass. In the future, problems such as the
best measure against these. The competence of rock layers is
distribution of stress fields in the vicinity of the boreholes after
influenced by the application of this control technique.
blasting, the range of the resulting rock deformation, and more
The system of destress rock blasting has been developed
precise determination of the stress concentration area must
since the 1980s. The technique of destress rock blasting has
be solved. Nevertheless, destress rock blasting contributes to
improved (drilling, length of boreholes, diameter of boreholes,
safe longwall mining of hardcoal in conditions of substantial
weight of blasted charges, design, evaluation of effectiveness,
rockburst risk.
etc.). Since it began there have been more than 2000 destress
rock blasts and more than 3.5 million tons of explosives have
been used.
491

Dynamic test of a high energy-absorbing rock bolt
C.C. Li
Department of Geology and Mineral Resources Engineering,
The Norwegian University of Science and Technology (NTNU), Norway
Subject: Others – Rock support
Keywords: bolts and anchors, lab testing, rock support, rock bursts
The D-bolt is a type of energy-absorbing rock bolt. It absorbs

deformation energy through mobilization of the load and
deformation capacities of the bolt material (steel) along its
entire length of the bolt. The bolt is made of a smooth steel
bar that has a number of integrated anchors evenly or unevenly
spaced along its length. Figure 1 shows some D-bolts of three
anchors. The section between two adjacent anchors is designed
1 m long, but it can be adjusted in accordance to rock condi-
tions. The first anchor is positioned closely to the thread/plate
for the purpose of reducing the load on the thread/plate. The
bolt is fully encapsulated in a borehole with either cement or
resin grout. The anchors are either firmly fixed in the grout or
are allowed to plough at a predefined load level. The smooth
bar sections between anchors have very weak or no bonding
to the grout due to the smooth surface of the bar. When the
rock dilates, the anchors tend to restrain the rock dilation and
the in-between bar sections become stretched so that a tensile
load is induced in the bar sections. The sections elongate elas-
tically in the beginning and soon become yielded after a small
amount of rock dilation. The sections elongate plastically at
the yield/ultimate load levels until the ultimate strain of the
steel bar is reached. For example, a 1-m long section of the
22 mm steel bar can elongate about 150 mm and absorb about
40 kJ of energy prior to failure under static pull load.
The dynamic performance of the bolt was studied through
a series of drop tests. There is no means currently to apply
dynamical loads to all sections of a D-bolt at the same time.
Therefore, the tests were carried out by applying impact load
only to one section of the bolt. In the tests, boreholes were
simulated with split steel tubes. Resin cartridges were slid
into two tubes that were axially aligned. The bolt specimen
was then spun into the tubes. The split of the two tubes was
located approximately in the middle of the stretch section of Figure 1. D-bolts; 2.2 m long.
the bolt specimen. The installation-ready specimen was hung
up at the end of the upper tube. Every test consists of dropping
a mass of 893 kg–2679 kg from a height of 1.5–2.1 m onto a 220 kN and every drop brought about an elongation of about
plate connected to the lower tube. The impact load is then 50 mm. For the specimens of 22 mm × 0.9 m, the input energy
transferred from the tube to the D-bolt through the resin grout was 30–39 kJ. The stretch section did not fail in the first drop
and the anchor in the lower tube. The impact velocity was in a when it was impacted with an input energy less than and equal
range from 5.42 m/s to 6.42 m/s. The potential energy of every to 36 kJ. It failed when the input energy was increased to 39 kJ.
drop, so called the input energy, was between 13 and 60 kJ. The maximum elongation of the 0.9 m long stretch section was
The bolt specimens were 20 mm or 22 mm in diameter and about 150 mm. For the specimens of Ø22 mm × 1.5 m, the
had a stretch section of 0.8, 0.9 or 1.5 m. For the specimens of input energy varied from 20 kJ to 60 kJ. The stretch section
20 mm × 0.8 m (diameter × length of the stretch section), the did not fail in the first drop when they were impacted with
impact input energy was 13 kJ. Three drops had to be done to an input energy less than and equal to 50 kJ. It failed when
break the bolt shank. The impact load was at a level of about the input energy was increased to 56 kJ. The elongation of the
492

Figure 2. Impact and plate loads versus displacement for a 22 mm specimen with an input energy of 56 kJ. The test section of the specimen
is 1.5 m long.
1.5 m long stretch section was approximately 220 mm. The The test results of all the specimens have shown that the load
impact load for the 22 mm specimens was at a level of about transferred to the plate/thread of the D-bolt is only a small
280 kN. Figure 2 shows the test results of a Ø22 mm × 1.5 m portion of the impact load, indicating the anchor provides a
specimen. It was tested with an input energy of 56 kJ (i.e. a good protection to the thread/plate of the bolt.
mass of 2900 kg falling from a height of 1.97 m). The input In conclusion, the D-bolt fully utilizes the load and defor-
energy was just beyond the critical input energy (55 MPa) so mation capacities of the bolt steel material so that it is both
that it failed. Another specimen that was tested with an input strong and deformable. The high strength of the bolt provides
energy of 50 kJ did not fail. a high resistance to rock dilation so that the rock deformation
For conventional rock bolts using thread/plate, it is often is profoundly restrained. The high deformability of the bolt
observed in the field that the thread/plate fails when the bolt makes it possible that the bolt would not prematurely fail when
is heavily loaded. The thread is the weakest portion of the bolt subjected to large seismic bursts. A D-bolt can absorb up to
since the effective area of the cross section of the threaded 50–100 kJ of deformation energy, depending on the length of
portion is smaller than the solid bar. For the D-bolt, one of the the bolt. The energy absorption of the D-bolt is repeatable and
anchors is positioned close to the thread. It is seen in Figure 2 predictable since it is purely dependent upon the mechanical
that the plate load is very low compared to the impact load. properties of the bolt material, i.e. the steel.
493

A 1D equivalent medium method for wave propagation across parallel

jointed rock mass
J.C. Li & H.B. Li

G.W. Ma
School of Civil and Resource Engineering, University of Western Australia, Crawley, Australia
J. Zhao
Ecole Polytechnique Federale de Lausanne (EPFL), Laboratory for Rock Mechanics (LMR), Lausanne, Switzerland
Keywords: dynamic equivalent method, rock joints, rock mass, wave propagation, viscoelastic model
Rock mass usually consists of multiple, parallel planar joints, to be a completely elastic medium, which is effective only if
known as joint sets, which govern the mechanical behavior the frequency-dependence and the discreteness of joints, or
of the rock mass. The dynamic behavior and wave prop- multiple reflections among the joints are negligible.
agation across jointed rock mass are of great interest to The one-dimensional equivalent medium method proposed
geophysics, mining and underground constructions. It is also in the present paper is an extension of the study by Li et al.
significant to assess the stability and damage of rock struc- (J.C. Li, G.W. Ma and J. Zhao. An Equivalent Viscoelas-
tures under dynamic loads. Because of the discontinuity by tic Model for Rock Mass with Parallel Joints. Journal of
the joints, the dynamic response of jointed rock mass is a Geophysical Research, 115(B03), 2010). By analyzing the
complicated process. It is of significance to develop an effi- normal and shear equivalent property of a parallel jointed
cient and explicit method to analyze longitudinal- or shear- rock mass, the improved method is mainly to investigate
(P- or S-) wave normally across a rock mass with parallel P- and S-wave propagation across parallel jointed rock mass.
joints. For P- or S-wave, the method includes two portions: a vis-
Currently there are two common methods to analyze the coelastic medium model which is shown in Figure 1 and a
wave propagation in jointed rock masses; namely, the dis- concept of “virtual wave source (VWS)”. The viscoelastic
placement discontinuity method (DDM) and the effective medium model is to describe the normal and shear property
moduli method. Generally, the displacement discontinuity of the representative volume element (RVE) for a rock mass,
method (DDM) treats joints, particularly the dominant sets as shown in Figure 2(a). The VWS is proposed to take into
as discrete entities. The effective moduli methods predict the account the wave reflections between the joints. Figure 2(b)
aggregate effects of many joints or systems of joints within
a representatively elemental volume (REV), which creates a
continuum analysis of the problems. However, The conven-
tional effective moduli methods ignore the discreteness of
joints in rock masses and simplify the discontinuous rock mass
Figure 1. Equivalent mechanical model of an auxiliary spring in Figure 2. Scheme of jointed rock mass and equivalent medium
series with Voiget model. (k = p for incident P-waves and k = s for incident S-waves).
494

Figure 3. Comparison of transmitted waves obtained from displacement discontinuity method (DDM) and equivalent medium method
(EMM) with different joint number (S = 1/10λ).
shows the one-dimensional equivalent medium for RVE in of transmitted waves obtained from displacement discontinu-
Figure 2(a). From the viewpoint of all, the normal or shear ity method (DDM) and equivalent medium method (EMM)
equivalent property of a rock mass is properly expressed from with different joint number when the joint spacing is one
the proposed method. Meanwhile, the derived wave propaga- tenth of the wave length. The results from the equivalent vis-
tion equation is explicit, which simplifies the analytical study coelastic medium method are very similar to those determined
of wave propagation. from the displacement discontinuity method. By analysis, it
The 1D equivalent medium method can be effectively is found that the P- or S-wave propagation normally across
applied to analyze P- or S-wave propagation normally an equivalent medium is related to the VWS number and the
across discontinuous rock mass with a set of parallel joints. ratio of the VWS spacing to an incident wave length. For a
Parameters in the viscoelastic medium model are derived given VWS number, if the wave length of the incident wave
analytically based on P- or S-wave propagation across a RVE. is sufficiently long compared to the joint spacing, the effect
The proposed method is then verified by applying identical of the VWS on wave propagation across rock mass is promi-
incident waves to the discontinuous rock mass and the cor- nent. Finally, frequency dependence and joint spacing effect
responding equivalent medium at one side to compare the on the wave propagation across parallel jointed rock mass are
output waves at the other side. Figure 3 shows the comparison discussed.
495

Effects of decoupled charge blasting on rock fragmentation efficiency
Wei-min Liang
School of Civil Engineering, Henan Polytechnic University, Jiaozuo, China
H.Y. Liu
School of Engineering, The University of Tasmania, Australia
Xiao-lin Yang
Luoyang Institute of Science and Technology, Luoyang, China
D.J. Williams
School of Civil Engineering, The University of Queensland, Australia
Keywords: blasting, dynamics, lab testing, numerical modelling, rock failure
1 EXTENDED ABSTRACT that the gap between the explosive and the borehole wall in
decoupled charge blasting decreases the stress wave loading
Rock blasting has been the popular method for rock break- rate on the wall and extends the action time of stress wave and
age in mining and civil engineering. In this paper, laboratory gas, which correspondingly decreases the size of the crushed
experiments are first carried out in cement mortar samples zone formed, as shown in Figure 3, results in better fragmen-
with various air-decoupled charges to investigate effect of tation efficiency and improves explosive energy utilisation
decoupled charge blasting on rock fragmentation efficiency. ratio. Throughout this study, it is concluded that the rock frag-
The geometry of the specimen used in decoupled charge mentation efficiency can be greatly improved by adopting the
blasting is depicted in Figure 1. decoupled charge blasting with the optimal decoupled coeffi-
Two types of charge structures are adopted: 1) air-decoupled cients, which can be determined using the formulas proposed
charge with decoupled coefficients of 1.33, 1.60, 2.00 and in this study.
2.67, and 2) coupled charge. The resultant fragments from the
experiments were screened and statistically analysed using the
so-called Gates-Gandin-Schumann fragment model to obtain
fragment size distribution and then calculate the mean size
of fragments K50 and the boulder fragment ratio Bfr. The
obtained K50 and Bfr for the different tests with various decou-
pled coefficients k are plot in Figure 2. As it can be seen from
Figure 2, the fragmentation efficiency in decoupled charge
blasting can be improved if the decoupled coefficient is in
an optimal decoupling coefficient range (ODCR). After that,
analytical formulas are derived to predict ODCR in decouple
charge blasting from the lab to the field.
Numerical simulations are finally conducted to investigate
the mechanisms in decoupled charge blasting and it is found
Figure 2. K50 versus k and Bfr versus k for the various tests.
Figure 3. Comparison of fracture patterns resulting from decoupled

Figure 1. Geometry of sample for decoupled charge blasting. charge blasting and coupled charge blasting.
496

Modelling earthquake-induced rock failures using NMM and DDA
Y.J. Ning, X.M. An & Z.Y. Zhao

G.W. Ma
School of Civil and Resource Engineering, University of Western Australia, Crawley, Australia
Keywords: Rock failure; Risks and hazards; Numerical modelling
The numerical manifold method (NMM) employs two inde-

pendent cover systems, i.e., the mathematical cover and the
physical cover, to model a physical problem. Continua and dis-
continua are represented in a unified form. The transition from
continua to discontinua can be conveniently modelled with a
fracturing algorithm as well. Comparatively, the discontinuous
deformation analysis (DDA) is a discrete numerical method
especially developed for the modelling of discontinuous block
system behaviours. In this paper, earthquake induced rock
fracturing as well as the transition from continuous rock to
discrete rock blocks are modelled with the NMM. To avoid
possible rotation-induced material domain area change in the
NMM and improve the calculation efficiency, after discrete
rock blocks are created in the NMM modelling, an NMM
model is converted into a DDA model and the motions of
the generated rock blocks are continued to be modelled with Figure 1. A rock slab with non-persistent joints on a slope.
the DDA method. An algorithm based on the Mohr-Coulomb
failure criterion with a tension cutoff is used in the NMM
fracturing modelling. The transition from NMM to DDA mod-
elling is based on strain and kinematic energy conservation. Table 1. Parameters of the intact rock.
The calculation efficiency can be improved considerably due
Young’s tensile
to the transition because of a smaller number of contacts and a Density modulus Poisson’s friction cohesion strength
smaller scale of the system equilibrium equation in the DDA (Kg/m3 ) (GPa) ratio angle (◦ ) (MPa) (MPa)
as compared with that in the NMM.
The complete failure process of a rock slab with pre-existing 2550 20 0.25 45 2 0.5
non-persistent joints located on the crest of a rock slope (Fig. 1)
under earthquake loading is modelled. The non-persistent
joints in the rock slab and the joint between the rock slab and
the slope crest have a friction angle of 35◦ and 5◦ , respectively,
with zero cohesion and zero tensile strength. The external sur-
faces of the slope and the rock slab have a friction angle of
35◦ . The parameters of the intact rock are given in Table 1.
The earthquake acceleration time history applied in the hori-
zontal direction is shown in Figure 2, in which g = 9.8 m/s2 is
the acceleration of gravity. The sampling frequency is 200 Hz
and it lasts for 50 s. The earthquake loading is applied in
the form of constraining seismic displacement time history
(Fig. 3) along the bottom and the left and right boundaries of
the model (see Fig. 1). The slope is supposed to be located far
from the epicentre thus the vertical acceleration is not taken
into consideration. A trial calculation approach based on the
strength reduction method is used to drive the non-persistent
joints to propagate, i.e., the friction angle, cohesion and ten-
sile strength of the intact rock are reduced simultaneously at Figure 2. Earthquake acceleration time history.
497

Figure 5. Absolute trajectories and relative trajectories of the
Figure 3. Earthquake displacement time history. centres of the falling blocks.
instances within the simulation. After the coalescence of the

propagation of J1 and the pre-existing joint between the rock
slab and the slope crest, the collapsed rock volume is created
and slides down along the slope crest. Consequently, a rock-
fall process follows and the propagation of J2 and J3 cut the
collapsed volume into three sub-blocks. The modelling is con-
verted from the NMM to the DDA at 8.42 s before the falling
blocks collide with the slope waist. The follow-up motions of
the falling blocks are continued to be modelled with the DDA.
The modelling by the DDA method finishes in two minutes,
while the corresponding modelling by the NMM may take
more than half one hour. Figure 5 gives the absolute trajec-
tories of the centres of the falling blocks and their relative
trajectories to the instant slope base position. After the disap-
pearance of earthquake loading, the blocks get still at 50.54 s
in front of the slope toe. The distances between the block cen-
tres and the final slope toe are 1.15 m, 2.47 m, and 4.52 m,
respectively.
In this paper, a rock failure process, including the fracturing
in intact rock, the transition from continuous rock to dis-
crete rock blocks and the subsequent motion of the collapsed
rock mass under earthquake loading, is completely modelled
using the NMM and DDA method. The earthquake accelera-
tion amplitude which can lead to the initiation of the failure
could thus be derived. Thereafter, the trajectories and the final
motion distances of the collapsed rock blocks are obtained.
The numerical approach adopted is promising for the mod-
elling of earthquake-induced rock failures and deserves to be
further studied.
Figure 4. Failure process modelled with the NMM followed

by DDA.
each trial calculation. When the tensile strength is chosen as

0.09 MPa, joint J1 starts to propagate at 5.46 s, when the ampli-
tude of the loaded acceleration is −0.61 g and the horizontal
displacement of the slope is −0.084 m. The negative acceler-
ation of the slope induces tensile stress status at the front of
the joint, thus crack propagation occurs. Figure 4 gives several
498

Prediction of rockburst based on an accident database
Ana Peixoto & Luis Ribeiro e Sousa

Department of Civil Engineering, University of Porto, Porto, Portugal
Rita Leal e Sousa

Department of Civil Engineering and Environment, Massachusetts Institute of Technology, Cambridge, USA
Xia-Ting Feng
State Key Laboratory for Geomechanichs and Geotechnical Engineering, Wuhai, China
Tiago Miranda & Francisco Martins

Department of Civil Engineering, University of Minho, Guimarães, Portugal
Keywords: risks and hazards; numerical modeling; case studies
The rockburst is an event that is caused by high stresses

that occur in intact brittle rocks, located generally at great
depths, during the excavation of an underground work. This
phenomenon can occur in tunnels for transport systems,
hydroelectric projects and in mining operations. The rock-
burst may take such proportions, causing the sudden rupture
of rock, with serious consequences in the building process and
the people involved in the work (He, 2009; 2010).
The event is caused by the overstressing of massive or
intact brittle rock, i.e. the stresses developed in the ground
that exceed the local strength of the material. Rockbursts are
violent and sudden ruptures of bedrock and can cause serious,
and often fatal, injuries (Sousa, 2010).
The experiences from deep level mining have contributed
significantly to the understanding of the rock mechanics
involved in the phenomenon of rockburst. Methods for predic-
tion of rock stress problems that cause spalling and rockburst
have been developed based on experience and the develop- Figure 1. Rockburst experimental test system (He et al., 2011).
ment of rockburst vulnerability indexes (Hudson, 2009; He,
2009). Rockbursts are not easy to predict. Investigations using
acoustic emission monitoring are sometimes recommended. In order to evaluate and understand what types of undesir-
Acoustic emissions allow one to monitor the accumulation able events may occur after the rockburst, we tried to gather
of cracking and evaluate the tendency for the rock to suffer and store a certain amount of information in a database. This
rockburst (Tang, 2010). would improve the knowledge about what kinds of accidents
Influence diagrams, containing the major parameters that occur due to rockburst and identify in what circumstances are
influence the rockburst can be built (Sousa, 2010). For pre- at the massive surrounding the opening of the tunnel.
diction of rockburst special reference is also made for the To complete the database, it resorted to an extensive survey
development of a rockburst experimental system at laboratory of several cases in which they used a form that covered all
(Figure 1). aspects mentioned above and which constituted the entire 62
The organization of a database of selected cases of rock- cases in the database. Each record in the database is based on
burst occurred during the construction of several tunnels was the interpretation of articles as well as through a questionnaire
performed. To develop the database, it resorted to an extensive completed by an expert. The database has a small number of
research from technical literature, publications and reports cases. However, it is a fundamental first step in order to create
of experiences in tunnels. Data were stored and analyzed on a more complex model in the future (Peixoto, 2010). More
the database being classified the rockburst by its proportions than half of the accidents are due to excessive load, where the
and dimensions, its causes and consequences. Then, Artificial rockburst is very strong (Figure 2).
Intelligence (AI) techniques were applied to this database, DM techniques were applied to the database. The environ-
including Data Mining (DM) techniques. ment R (R Development Core Team, 2010) was applied using
499

Figure 2. Distribution of cases by accident type.
Figure 4. Learned Rockburst Bayesian Network (K = σh /σv ,
TS = Tunnel Shape, ORIENT = Orientation/location of the rock
burst, ADV = Advance Rate, SUPP = Type of Support; D = depth;
EXC = Excavation method; ERRB = Equivalent Radius of the
Rockburst; TO = type of occurrence).
Table 1. Confusion matrix.

Slight Moderate Strong
Overbreak Rockburst Rockburst Rockburst
Overbreak 100.0% 16.7% 0.0% 0.0%

Slight Rockburst 0.0% 66.7% 50.0% 28.6%
Moderate Rockburst 0.0% 16.7% 50.0% 14.3%
Strong Rockburst 0.0% 0.0% 0.0% 57.1%
Figure 3. Comparison of accuracies for different techniques.
The rockburst develops due to the high stresses in rock

masses. There appears, therefore, the need to develop appro-
RMiner software. Different DM techniques were applied, priate computational tools that serve as methods of prediction
being obtained the accuracy to each group of data, for both and control of the rockburst. 3D numerical models were devel-
type of data (training and test). In the test phase, the Artificial oped in order to study the damage induced in the primary
Neural Network (ANN) technique presents more percentage support by the rockburst using the software FLAC3D (Peixoto,
of accuracy as illustrated in Figure 3. 2010).
Bayesian Networks (BN) were also applied. Despite the lim-
ited amount of cases from the rockburst database, a BN was
learned, trained with 30 of the most well documented cases REFERENCES
available. The algorithm used for the learning of model 31
was the “greedy thick thinning” with a uniform prior (Sousa, He, M. (2009). The mechanism of rockburst and its contermeasure
2010). Note that because no cases where no rockburst occurred of support. Int. Consultation Report for the Key Technology of
where collect one is not able to learn a network that will Safe and Rapid Construction for Jinping II Hydropower Station
allow you to predict whether or not a rockburst will occur High Overburden and Long Tunnels, 2/03/2009, pp. 23–28.
under certain geological and construction related scenarios. He, M.; Xuena, J.; Coli, M.; Sousa, L. (2011). Experimental study on
rockburst in underground quarrying of Carrara. Journal of Rock
However with the data available one was able to learn a BN Mechanics and Mining Science (submitted).
that has the goal of predicting what type of event (overbreak, Hudson, J. (2009). Predicting rockburst occurrence and develop-
slight rock burst, median rockburst and strong rockburst), and ment of Rockburst vulnerability index (RVI). Int. Consultation
where, would occur under certain circumstances (geological, Report for the Key Technology of Safe and Rapid Construction for
construction method, etc). The learned BN is presented in Jinping II Hydropower Station High Overburden and Long Tun-
Figure 4. nels, 2/03/2009, pp. 29–35.
It is important to also check, in the cases the model does not Peixoto, A. (2010). Prediction of rockburst in underground works
accurately predict the type of occurrence, which type of occur- (in Portuguese). MSc Thesis, University of Porto, 135p.
rence is predicted. For that a confusion matrix is presented R Development Core Team (2010). R: A language and environment
in Table 1 for training data set. The results are promising; for statistical computing. R Foundation for Statistical Computing,
Vienna, Austria. http://www.R-project.org.
despite the fact of the limited amount of cases, and the fact Sousa, R.L. (2010). Risk analysis for tunneling projects. MIT,
the database contains only cases where events occurred. If Cambridge, 589p.
cases where no event occurred were also documented and Tang, C. (2010). Preliminary engineering application of microseis-
available, one could train a BN that would predict whether mic monitoring technique to rockburst prediction in tunneling
or not an event (and what type) will occur on a certain section of Jinping II project. J. of Rock Mechanics and Geotechnical
of a tunnel/mine. Engineering, 2 (3): 193–208.
500

An investigation on stress wave propagation in micromechanical models of rock
R. Resende
Concrete Dams Department, National Laboratory for Civil Engineering, Lisbon, Portugal,
currently at ISCTE-Lisbon University Institute, Portugal
L.N. Lamas & J.V. Lemos

Concrete Dams Department, National Laboratory for Civil Engineering, Lisbon, Portugal
R. Calçada
Faculty of Engineering of the Porto University, Porto, Portugal
Keywords: blasting, dynamics, numerical modelling, rock joints, rock properties
This work is dedicated to the study of stress wave propaga-

tion in two-dimensional Bonded Particle Models (BPM) of
rock and to the development of the techniques necessary to
carry out dynamic modelling of stress wave propagation in
rock masses. Its purpose is to investigate the capabilities of
this numerical method on modelling wave propagation in rock
masses and rock fractures at the microscopic level. The focus
is on the simplest case, i.e., propagation of low amplitude,
high frequency (kHz range) plane compressive waves across a
synthetic rock core. Stress wave loading is studied and absorb-
ing boundaries and hybrid static and dynamic boundaries
are implemented. Attenuation relations are defined, setting
the minimum requirements for further studies of stress wave
interaction with rock masses.
Micromechanical models aim to predict the mechanical
response of rock through the simulation of the microscopic
mechanisms of deformation and failure in assemblies of cir-
cular or spherical particles. The philosophy behind this class Figure 1. Complete model and sections were wave velocity was
of models is that no complex mathematical formulations are measured (left), top and lateral boundaries of the model (right).
needed if the fundamental hypotheses that drive the micro-
scopic behaviour of the media are correctly and efficiently
implemented. As a result, the macroscopic response of the in dynamic problems since it is more versatile, allowing for
material under a broad range of situations is covered through the application of several kinds of conditions in the same
the interaction of a few simple mechanisms. boundary.
A synthetic rock specimen was created to simulate the To convert kinematic loadings (velocity or acceleration
stress wave propagation. It is 90 × 600 mm, with approxi- time-histories) into forces that can be applied at boundaries,
mately 35,000 particles with radii that vary uniformly between two steps were taken. First, velocity was converted into stress.
0.50 and 0.83 mm. Particle density is 3150 kg/m3 and given Then, stress must be transformed into equivalent forces that
that the porosity is 14.3%, the continuum-equivalent poros- are applied at the boundary balls (Fig. 1), via boundary depth
ity becomes 2700 kg/m3 . Normal contact stiffness was set to correction factor.
62 GN/m2 and the shear stiffness to 24.8 GN/m, resulting in a This procedure was implemented in the Itasca code PFC2D,
young modulus of 43.5 GPa and a Poisson’s ratio of 0.2. The permitting the utilization of mixed boundary conditions, in
tension and shear strength of the contacts were set to very high which static and dynamic loadings coexist with a viscous
values, to allow for a separation of the different phenomena absorbing boundaries. The hybrid boundary was tested with
that influence stress wave propagation. pure shear and compressive Ricker waves hitting straight
Boundary conditions can be kinematic or static. In the first boundaries at normal angles. The viscous boundaries pre-
kind there is a prescription of particle movement, while in sented an efficiency (defined as the ratio between absorbed
the second forces and/or moments are applied to the centre and incoming wave amplitude) superior to 99%, as shown in
of mass of the particles. The latter case is more advantageous Figure 2.
501

Figure 2. Wave velocity recorded at the centre of the model. The
first wave is going upward; the second wave is the downward coming
wave that was reflected (a), reflected (b) or absorbed (c) at the top
boundary.
Figure 4. Relative variation of peak wave velocity measured at the
bottom and top end of the model. Plot (b) zooms in the axes origin
and the dashed horizontal line marks 5% loss.
was recorded along the model. Figure 4 provides a compact

way to compare the performance of all models by representing
the relative loss of velocity amplitude between the start and
the end of the model for the studied wavelength-ball diame-
ter relations. If an amplitude loss threshold level of 5% at the
end of the model is set, the wave frequency and associated
wavelength/particle diameter relation that complies is 82 for
the BPM, 24 for the Rectangular, 31 for the Hexakn=ks and
42 for the Hexakn=2.5ks .
The BPM model is much more demanding than all orga-
nized packings. Of these, the one that preserves better wave
Figure 3. Relative variation of compressive wave velocity with quality is the Rectangular packing, followed by the Hexakn=ks
peak frequency. and then the Hexakn=2.5ks .
This paper studies the propagation of compressive stress
waves on rock bonded particle models. The tests will be
Parallel to the BPM of rock, three other assembly types repeated with shear waves in order to completely characterize
(rectangular and hexagonal arrangements) with the same the dynamic properties of the model: shear and bulk mod-
external dimensions, ball radius equal to the average radius uli. Other waveforms may also be used and the consistency
of the BPM model and similar microproperties where gen- of the results checked. The expansion to three dimensions
erated to compare and frame the dynamic response of the will allow for greater realism of the waves, which can then
BPM. Measured compressive wave propagation velocities are be polarized. Finally, the implementation of rock fractures in
4191 m/s, 4819 m/s, 6279 m/s and 5750 m/s in BPM, Rect- micromechanical models (described in Resende & Resende
angular, Hexakn=ks and Hexakn=2.5ks assemblies. Predictably, et al., 2010) allows for the simulation of rock masses more
the lower wave velocity occurs in the BPM since this model complex dynamic behaviour.
has the lowest degree of organization and waves follow less
direct paths.
Figure 3 displays the relative variation of propagation veloc- REFERENCES
ity with wave frequency. It can be seen that wave velocity
variation across the considered frequency range is small and Resende, R., Lamas, L., Lemos, J., Calçada, R. (2010). Microme-
higher frequencies travel at slightly higher velocities. Having chanical modelling of stress waves in rock and rock fractures.
a non-organized structure, BPM is the most dispersive assem- Rock Mechanics and Rock Engineering, 43(6): 741–761.
bly: from 5 kHz to 200 kHz the wave propagation velocity Resende, R. 2010. An investigation of stress wave propagation
increases a little more than 3%. The variation is smaller in the through rock joints and rock masses. PhD thesis, Faculty of
other assemblies. Engineering of the Porto University.
To measure amplitude attenuation, waves of increasing fre-
quencies were injected at the base and the peak amplitude
502

A comparison of empirical methods and evolutionary programming to predict

blast-induced ground vibration
A. Sadeghee & H. Khoshrou

Mining Engineering, Amirkabir University of Technology, Tehran, Iran
Keywords: blasting, risks and hazards, case studies
Drilling and blasting are two necessary stages in mining engi- and β = charge constant and transmitting medium constant
neering and construction works.Although drilling and blasting respectively.
prove useful, have a few detrimental impact on structures Although these formulae are accepted in literature to predict
located at the neighboring township and residential spots. PPV from available data, in most cases they result in unaccept-
To name a few, these problems are fly rock, air blast, back able values compared to posterior measured ones i.e. in nearly
break and ground vibration. Amongst all problems originated all cases predict erroneous values. Also these formulae actu-
in blasting, because of stringent statutory regulations blasting ally are obtained from setting a correlation between the charge
induced ground vibration can lead into early mine closure. per delay value, the distance value and PPV value through sta-
When an explosive charge is detonated in a solid medium tistical analysis. So human intervention on statistical analysis
like rock, a family of waves is generated. These waves have results in a predictor which its general formula is arbitrarily
different particle motions and travel at different wave veloc- chosen and inevitably will predict different PPVs compared
ities through the medium (Sanchidrian, J. A. et al., 2007). to measured ones. However, nowadays computer based soft-
As the seismic waves travel through the rock, they excite wares prove viable and accurate computations on which PPV
particles to move. The motion of the ground particles (vibra- values can be obtained with maximum accuracy in prediction
tion) occurs in three directions which are vertical, radial and results.
transverse. Using a seismograph the particle velocity in three New studies on soft computing techniques offer good
directions are captured and maximum velocity of particles in approaches such as artificial neural networks (ANN), fuzzy
either direction is referred to peak particle velocity (PPV). regression and evolutionary programming. Some researchers
Several researchers have suggested that peak particle velocity proposed ANN to predict ground vibration components (PPV
(PPV) manifests the most appropriate and accurate indicator and frequency) to an acceptable level (Khandelwal & Singh
of possible damage to environment and structures. 2007).
During the evaluation and optimization process of a blast- In the present study the possibility of creating a model
ing operation, it is important to predict vibration levels (via through soft computing techniques and especially evolution-
PPV) at a specific point according to a given blasting design. ary programming is examined to find a predictor and its
The magnitude of PPV at a determined point varies with the constants to overcome aforementioned problems.
explosive charge amount that is detonated per delay (not total In GP, individuals are represented as trees. Elements of the
charge) and the distance of a specific point from the charge trees are functions and terminals. Terminals are the variables
center (Sanchidrian, J. A. et al., 2007). Certain empirical for- and the functions are operations applied to these variables
mulae were proposed by the different researchers to predict forming the model together. For example, Fig. 1 shows the
PPV using the charge per delay and the distance to shot center representation of the simple expression, x − y + x/5.
quantities (Duvall & Foghelson 1962, Langefors & Kihlstrom In applying genetic programming to a problem, there
1963, Ambreseys & Hendron 1968, Indian Standard Institute are five major preparatory steps. These five steps involve
1973). determining:
Table 1 lists four frequently used formulae called herein
1. predefined set of terminals,
empirical predictor equations.
2. predefined set of primitive functions,
where Q = charge weight (kg); R = distance between the
3. a fitness measure (called a “fitness function” to specify
charge center and the PPV measurement point (m); PPV =
what needs to be done)
peak particle velocity of ground vibration (mm/s); and K
Table 1. Different predictor equations mostly used in literature.
Name of predictor equation Formula
USBM (Duvall and Fogelson, 1962) PPV = K(R/Q1/2)−β

Ambraseys–Hendron (1968) PPV = K(R/Q1/3 )−β
Langefors–Kihlstrom (1978) PPV = K[(Q/R2/3 )1/2 ]β
Indian Standard Predictor (1973) PPV = K(Q/R2/3 )β
Figure 1. An example tree representation in genetic programming.
503

4. the parameters for controlling the run, such as population The equation generated by GP method has is
size, reproduction operators, probabilities of the operator
and so on, and
5. A method for designating a result and the criterion for
terminating a run. Where C is a constant and here it equals to 742.5.
Genetic programming is an iterative method. The algo- Fig. 4 demonstrates the graph between measured PPV and
rithm used by genetic programming in finding solutions is predicted PPV by GP on 1:1 slope line.
summarized in Fig. 2. Based on the current study, it is concluded that the evolu-
To be able to compare the results of GP with empirical tionary programming approach is marginally the best option
formulae, average proportional percentage error (APPE) is for accurate and appropriate prediction of PPV. It and the
used. APPE equation is given in equation 1. USBM approach are similar in their ability to match mea-
sured data over a wide range of scaled distances. This study
indicates that all predictors used in the paper are either over
estimating or under estimating the safe explosive charge to
some degree to keep the PPV under the safe limit. No mat-
ter which prediction approach is taken, prediction equations
need to be chosen with care and should be employed only
Where PVi = predicted value of PPV (mm/s) for ith test data in the same geology and topography from which they were
set; MVi = measured value of PPV (mm/s) for ith test data derived.
set; and n = number of data set.
62 data sets are divided into two data sets called training
data sets and test data sets. Full description of research conduct
in this study is depicted in figure 3.
Figure 2. Algorithm used in genetic programming to obtain a PPV

predictor. Figure 4. Measured and predicted PPV by GP obtained equation.
Figure 3. Research algorithm in the present study.
504

Effective explosive energy utilization for engineering blasting – initial results

of an inventive stemming plug, SPARSH
Md. Sazid & M.R. Saharan

Central Institute of Mining & Fuel Research
T.N. Singh
Indian Institute of Technology, Mumbai
Keywords: blasting, case studies, mining
Drill-and-blast cycle is still the most economic and convenient of uniform size and well below the handling limit of mining
method of obtaining ore / mining / tunneling. Much advances machineries (Figure 3 and Figure 4).
related to the blasting technology has been studied which is In another case study, enhancement in the explosive energy
termed as ‘Blasting Engineering’. It is, however, an efficient utilization through application of SPARSH is clearly demon-
application of the technology makes a blasting as ‘Engineering strated. The mine management for a rock phosphate mine
Blasting’ which shall result into minimal deleterious impacts undercharged their blastholes due to proximity of habitats and
of blasting and maximum energy utilization from the blasting.
Further, little studies have done to understand impact of proper
stemming for obtaining ‘Engineering Blasting’.
We have invented a stemming plug named SPARSH
(A Stemming Plug Augmenting Resistance to Stemming in
Holes) which has many unique features and phenomenally
improves the blasting. This paper discusses initial results
of SPARSH and it is demonstrated that the application of
SPARSH results into NO FLY ROCK BLASTING (Figure 1
and Figure 2). The explosive energy utilization increases
by 30% and more with the use of SPARSH which results
into fewer requirements of boreholes drilling and lesser con-
sumption of explosive thus ground vibrations are lower in
magnitude than the conventional blasting. Another major
advantage of the application of the SPARSH is minimal
requirement of secondary blasting as the muck produced is
Figure 2. Borehole ejection position from a comparative blast in

the limestone mine.
Figure 1. NO FLY ROCK BLAST with SPARSH in a limestone Figure 3. Fragmentation before SPARSH application in a
mine. manganese mine.
505

Figure 4. Fragmentation after SPARSH application in the man-
ganese mine. Figure 6. Rock breakage from the same blast in the rock phosphate
mine with SPARSH application.
The direct benefits of using SPARSH is increase in the pro-

ductivity as burden and spacing pattern increases to a value
of 30% or more. Indirect benefits of SPARSH application
involves improvement in the mining machineries performance
(shovel and dumper), less pressure on milling operations due
to smaller fragment sizes and lesser environmental nuances.
Immense benefits from NO FLY ROCK BLASTING tag are
an invaluable impact of SPARSH application. Further, applica-
tion of SPARSH also assists in reducing carbon footprints for
the mine management through reduced explosive consump-
tion, lesser energy consumed for drilling, vehicular movement
and during milling operations. The paper will elaborate the
above mentioned outputs with scientific measurement of the
parameters from three case studies.
Figure 5. 120 m long face without fractures after blasting in a rock
phosphate mine.
some critical structures for a particular blast. The blasthole

those were not charged with SPARSH could not produce a
single crack after the blasting for a 120 m long face length
(Figure 5) and 45 m long face resulted into only fissures gen-
eration while the blasting with SPARSH with the same charge
amount in the same blast for a 120 m long face resulted into
a good displacement of muck pile and fragment size were
adequate to be loaded by loading equipment (Figure 6).
506

Rock support subjected to dynamic loading: Field testing of ground support using
simulated rockburst
S. Shirzadegan, E. Nordlund, U. Nyberg & P. Zhang

Division of Mining and Geotechnical Engineering, Luleå University of Technology, Luleå, Sweden
L. Malmgren
Mining Technology R&D, LKAB Kiruna Mine, Kiruna, Sweden
Keywords: Dynamics, field measurements, rockbursts, rock support
Increasing the mining depth at LKAB Kiirunavaara mine support systems and the interaction with the rock mass can
located in the northern part of Sweden is leading to higher be investigated. A number of studies using explosives to gen-
stress magnitudes, resulting in increased seismic activity and erate the dynamic load have been carried out by for example
more rockburst damage. The effectiveness of various ground Milev et al. (2001), Espley et al (2004), and Heal & Potvin
support systems under dynamic loading conditions has there- (2007). A total number of twelve tests are planned to be car-
fore become of prime interest to LKAB for successful and ried out. Tests 1 and 2, called Zero tests-1 and 2, were planned
safe mining at deep levels. Therefore series of rockburst sim- in order to design the charge density, the instrumentation and
ulations will be conducted using explosives to generate the the influence of blast gases. This paper covers the set up and
dynamic load, on a number of support systems. In large-scale results from the first trial test called Zero test-1. In this test, the
dynamic tests of ground support the performance of ground explosive used to generate the dynamic load was of military
type, NSP711 with a high VoD compare to other commer-
cial explosives. The blasthole was approximately 15 m long
with the diameter of 115 mm. A nominal burden of 3 m was
selected between the blast hole and the test panel (see Fig-
ure 1). Two different charge diameters (76 mm and 45 mm)
were used in the blasthole to simulate different magnitudes
of the seismic event. The LKAB Standard support system for
seismically active mining areas was installed in the test drift.
The field monitoring included ground motion measurement
with a number of accelerometers, fracture investigation with
observation holes, ground and support motion imaging with
high speed camera, as well as the deformation measurements
(see Figure 2). Results from accelerometers and gas pressure
gauges indicate that the wall was loaded by a shockwave first
and then a gas pressure, which proves that the methodology
Figure 2. Location of measurement instruments on the test panel.

Figure 1. a) Site, support, gas holes and blasthole location. b) Blast a = accelerometers, s = displacement gauges, o = observation holes,
design. t = coaxial cables connected to TDR system.
507

REFERENCES
Espley, S. J., Heilig, J. & Moreau, L. H. 2004. Assessment of the

dynamic capacity of liners for possible application in highly
stressed mining environments at Inco limited. In Potvin, Y.,
Stacey, T.R. & Hadjegeorgiou, J (eds). Surface support in Mining,
187–192, Nedlands: Australian Centre for Geomechanics.
Heal, D. & Potvin,Y. 2007. In-situ dynamic testing of ground support
using simulated rockbursts. In Potvin Y (ed). Deep Mining 07;
Proceedings of the 4th International Seminar on Deep and High
Stress Mining, Perth, 373–394, Nedlands: Australian Centre for
Geomechanics.
Milev, A.M., Spottiswoode, S.M., Rorke, A.J. & Finnie, G.J. 2001.
Seismic monitoring of a simulated rockburst on a wall of an under-
ground tunnel. The Journal of the South African Institute of Mining
and Metallurgy: 253–260.
Figure 3. The range of PPV measured by surface mounted

accelerometers on the test panel.
used to simulate rockburst is suitable. The PPV measured by

surface mounted accelerometers on the test panel shows fair
agreement with the charge density distribution (see Figure 3).
The results obtained from other monitoring instruments in
this test were satisfactory. However, since the designed blast
did not generate severe damage to the ground support system
except for some cracks on the surface of the shotcrete, a larger
amount of explosive (compared to Zero test-1) will be used in
Zero test-2.
508

Effects of loading rate on compressive strength of Maha Sarakham salt
T. Sriapai, P. Samsri & K. Fuenkajorn

Geomechanics Research Unit, Institute of Engineering Suranaree University of Technology, Thailand
Keywords: weak rock
Uniaxial and triaxial compression tests have been performed by the Office of Atomic Energy for Peace. This study is part
to assess the influence of loading rate on the compressive of a research effort on determination of the safe maximum
strength and deformability of the Maha Sarakham salt. The withdrawal rate of compressed air stored in salt caverns. The
tested specimens were prepared from 60 mm salt cores drilled salt specimens are prepared to have a nominal dimension of
from depths ranging between 140 m and 250 m by Pimai Salt 5.4 × 5.4 × 5.4 cm3 and compressed to failure using a polyax-
Co. in the northeast of Thailand. The salt cores belong to the ial load frame. The lateral confining pressures are maintained
Middle Salt member of the Maha Sarakham formation. This constant at 0, 3, 7, 12, 20 and 28 MPa while the axial stresses
salt member hosts several solution-mined caverns in the basin. are increased at constant rates of 0.001, 0.01, 0.1, 1.0 and
The Maha Sarakham formation is also being considered as a 10 MPa/s until failure occurs.
host rock for compressed-air energy storage caverns by the Table 1 shows the test results. The measured stress-strain
Thai Department of Energy, and for chemical waste disposal curves indicate that the salt behaves as a nonlinear material
Table 1. Salt strengths and elastic properties under various loading rates and confining pressures.
Specimen ∂σ1 /∂t σ3 σm εm τoct,f γoct,f E

No. (MPa/s) (MPa) σ1 (MPa) (MPa) (10−3 ) (MPa) (10−3 ) (GPa) ν
31 10 0 25.1 8.4 – 11.8 – – –

7 1 22.6 7.5 – 10.7 – 21.4 –
6 0.5 20.4 6.8 – 9.6 – 19.9 –
9 0.1 19.2 6.4 – 9.1 – 19.5 –
11 0.05 19.1 6.4 – 9.0 – 19.3 –
10 0.01 17.8 5.9 – 8.4 – 18.2 –
8 0.005 17.8 5.9 – 8.4 – 17.8 –
12 0.001 16.5 5.5 – 7.8 – 15.7 –
35 10 3 49.5 18.5 – 21.9 – – –
33 1 45.3 17.1 −3.2 19.9 25.8 22.5 0.25
32 0.1 42.5 16.2 −1.0 18.6 27.9 21.8 0.26
38 0.01 41.2 15.7 −4.8 18.0 33.3 19.6 0.24
16 0.001 37.5 14.5 −3.1 16.3 37.9 15.6 0.31
34 10 7 69.0 27.7 – 29.2 – – –
41 1 64.8 26.3 −0.6 27.2 30.1 21.7 0.29
29 0.1 60.3 24.8 −3.2 25.1 33.1 17.6 0.32
27 0.01 57.4 23.8 −6.3 23.8 35.6 17.9 0.27
20 0.001 48.1 20.7 −6.0 19.4 37.1 16.9 0.32
43 10 12 88.9 37.6 −0.8 36.3 23.3 22.6 0.24
39 1 81.8 35.3 −1.7 32.9 27.3 22.7 0.33
17 0.1 72.8 32.3 −4.1 28.7 35.1 19.8 0.38
26 0.01 66.0 30.0 −6.3 25.5 36.1 19.0 0.35
15 0.001 54.8 26.3 −6.0 20.2 36.8 17.8 0.38
40 10 20 106.4 48.8 −0.3 40.7 23.8 22.0 0.23
18 1 96.2 45.4 −0.5 35.9 27.2 21.9 0.39
42 0.1 87.5 42.5 −1.4 31.8 30.7 19.1 0.43
30 0.01 78.3 39.4 −0.8 27.5 36.6 19.9 0.40
23 0.001 64.8 34.9 −0.4 21.1 42.1 17.5 0.42
22 10 28 119.7 58.6 −0.2 43.2 25.1 25.2 0.24
44 1 109.0 55.0 −0.4 38.2 28.7 23.0 0.36
37 0.1 99.2 51.7 −0.5 33.6 32.7 19.8 0.42
45 0.01 87.4 47.8 −0.5 28.0 39.0 17.2 0.43
46 0.001 73.6 43.2 −2.9 21.5 45.4 16.7 0.43
509

Figure 3. Distortional strain energy at dilation (Wd,d ) as a function
Figure 1. Octahedral shear stress (τoct ) at failure as a function of mean strain energy at dilation (Wm,d ).
mean stress (σm ) for various confining pressures (σ3 ) and loading
rates (∂σ1 /∂t).
The strains induced at failure decrease as the loading rate
increases. Various multi-axial formulations of loading rate
dependent strength and deformability are derived. The vari-
ation of the octahedral shear stresses and strains induced at
dilation and at failure with the applied shear stress rates can
be best described by power relations. The distortional strain
energy at dilation and at failure from various loading rates
varies linearly with the mean normal stress (Figures 2 and 3).
Potential applications of the proposed empirical relations
are demonstrated by simulating the reduction of the internal
pressure of salt cavern under various rates using a finite differ-
ence code FLAC. Two cases with different minimum pressures
are studied: 1.1 MPa and 2.2 MPa (0.1σcs and 0.2σcs ). Each
case is simulated for four different rates of cavern pressure
withdrawal: 17.3, 8.6, 0.6 and 0.3 MPa/day (equivalent to pres-
sure schemes of 1 cycle/day, 15 cycles/month, 1 cycle/month
and 6 cycles/year, respectively). The Maha Sarakham salt is
Figure 2. Distortional strain energy at failure (Wd ) and at dilation assumed to behave as a Burgers material. The octahedral shear
(Wd,d ) as a function of mean stress. stresses and strains around the salt cavern are determined and
compared against the proposed empirical criteria. The safe
under the range of the parameters used here (Figure 1). The maximum pressure reduction rate of the compressed-air stor-
salt strength and elasticity increase with the loading rates. age cavern can be determined. The Wd,d −Wm,d criterion gives
The tangent elastic modulus measured at 50% failure stress the most conservative results. It shows the FS values lower
varies from 15 to 25 GPa, and the Poisson’s ratio from 0.23 than 1.0 for the withdrawal rates of 15.1 and 7.6 MPa/day
to 0.43. The elastic parameters tend to be independent of the (1 cycle/day and 15 cycles/month). This criterion is perhaps
confining pressures. The cohesion ranges from 4.0 to 5.3 MPa, the most appropriate for use in the design of the withdrawal
and the friction angle from 40.4 to 46.6 degrees. They are rate because it incorporates both shear and mean stresses and
calculated at the confining pressure of 7 MPa for all loading strains. It implicitly incorporates the rate effect, and requires
rates. Both increase with the loading rate, suggesting that the complete stress-strain relations that must be obtained from
salt is stronger under higher loading rates. compression testing under various loading rates.
510

Effect of artificial discontinuities on blast induced vibrations in open pit mines
Ö. Uysal
Mining Engineering Department, Dumlupinar University, Kutahya, Turkey
E. Arpaz
M. Cavus
Seyitomer Coal Enterprise, Kutahya, Turkey
S. Beyhan
Y. Tola
S. Yuvka
Keywords: Blasting, field measurements.
In the case of blast induced vibrations it is very common that 10 holes were drilled at 2.5 meters distances and 10 meters of
mine managers come across with complaints from local res- length. The distance between the barrier holes and the blast
idents. This may lead to decrease or total cessation in the holes is 360 meters. This was intended to provide a discontinu-
production. Therefore, there are a lot of studies that focus on ity that comprises of holes, which was expected to eliminate or
eliminating the effects of blast induced vibrations. The first reduce the vibrations. The holes were 225 mm in diameter and
method that comes to mind for the elimination of blast induced 10 meters in length. For the recordings two different devices
vibrations is to reduce the amount of explosive. However, were used (Instantel mini mate plus, Instantel Blasmate III).
decrease in the amount of the explosive leads to a decrease Within the scope of the study, 395 vibrations were recorded
in the pre-planned production. Studies apart from this focus from 150 instances of blastings. Recorders were placed in front
on pre-split, barrier holes, and trench methods. While some of and 25, 50, and 75 meters at the back of the discontinuity in
of these researchers claim that blast induced vibrations are panel 1 and panel 2. For panel 3, the recorders were placed in
largely eliminated by the help of these methods, some others front of the discontinuity line and 5, 10, 15, and 25 meters at
claim that they have no positive influence on the issue. It can the back of it. Thus, the changes in the front line and backline
be said that there does not exist a common agreement on the of discontinuity were recorded according to distance.
effects of these methods on the elimination of blast induced Findings were evaluated based on the location of recorders
vibrations. as to the discontinuity lines. The values in panel 1 seem to
The primary purpose of this study is to eliminate the adverse have been unevenly distributed due to the fact that they have
effects of blast-induced vibrations at Seyitomer Coal Enter- not been subjected to any categorization. When categorized,
prise by forming artificial discontinuities. In order to do this, it can be seen that the values recorded in the front of the
an artificial discontinuity was formed in the area between discontinuity have higher particle velocity compared to the
blasting area and measurement point in the dragline panel. The ones taken from the back of the discontinuity.
panel height was 24 m. The rock type existing over the coal in In the light of the evaluation of the values in panel 2, it can
the panel was marl. ANFO was used for rock fragmentation. be seen that the data gathers around a narrow line when they
The blast holes were 24 m long, with 228 mm diameter. The are not grouped. After the grouping, it can be seen that the
distance between the blasting holes was 10 m. ANFO, power values taken from the front of the discontinuity are higher
gel dynamite and detonator were used for the blasting. The than the other values and the values taken from behind the
delay between each hole was 42 ms and in-hole delay, 25 ms. discontinuity have slower particle velocity.
The area has been separated into three sections comprising of In the light of the evaluation of the data from panel 3, the
panel 1, panel 2, and panel 3. At the end points of each panel data is distributed unevenly when they are not grouped. If they
511

are grouped, it can be seen that the values taken from the front cannot be the only solution and thus it becomes of utmost
of the discontinuity appear to have higher particle velocity importance to investigate techniques other than using lesser
compared to others. amount of explosive. The studies in the literature indicate that
In the light of the evaluation up to now, it can be seen that the most of the complaints stem from surface vibrations. Differ-
solid line representing the recordings taken from the front of ences in the area may cause surface vibrations to go different
the discontinuity is at the top and the broken lines that belongs directions or change their properties. In the present study, an
to the recordings taken from behind the discontinuity is below artificial discontinuity was formed between the blasting point
in the graphics. It can be seen that as the distance increases and the measurement point. By recording numerous instances
behind the discontinuity, the values taken from behind the dis- of blasts either in front or behind the discontinuity point. The
continuity become closer to the ones taken from front of the effect of discontinuity on ground vibrations was investigated.
discontinuity. It is speculated that the reason for the diversity As a result of the evaluation of the recordings, a significant
in the results taken from the 75th meter is the insufficiency decrease was observed in the values that were obtained around
of data. This idea is also supported by the total data analysis the discontinuity. But it was found that the decrease in the
and the value distribution patterns. In order to affirm the data vibrations reduces as the distance behind the artificial barrier
above and document the changes in vibrations in the front increases. It would not be wrong to suggest, in regard to the
of and the behind the discontinuity, 6 blasting were recorded features of waves, that the decrease was caused by the depth
from behind the discontinuity and out of the range of the dis- of the discontinuity.
continuity by setting up devices. Vibration values at the back In this study, the effect of artificial discontinuities on blast
of the discontinuity fell as much as 38%. induced vibrations was investigated. To this end, 3 different
The issue of eliminating the adverse effects of blast induced panels in Seyitömer Coal Mine in Turkey were formed, and
vibrations is one of the most hotly debated and investigated a total of 395 vibration records were taken in front of and
topic by researchers. Most of the studies on the issue focus on behind the discontinuities from different places pertaining to
decreasing the amount of explosive and using more controlled 150 blasts. Analysis of data has revealed that a decrease in
techniques. However, such practices lead to a decrease in the PPV of up to 80% has been detected right behind the barrier
production. Therefore, decreasing the amount of the explosive holes.
512

Wave propagation in inhomogeneous orthotropic rocks
C.D. Wang & D.S. Wu

Department of Civil and Disaster Prevention Engineering, National United University, Miao-Li
J.J. Liao
Department of Civil Engineering, National Chiao-Tung, Hsin-Chu
Keywords: dynamics, rock properties
1 INTRODUCTION hence, Cij (i, j = 1–6) can be presented in terms of the nine
independent elastic constants as:
Wave propagation in anisotropic materials is of great inter-
est in geophysics and other branches of applied sciences
such as rock mechanics/rock engineering, earthquake engi-
neering, and petroleum engineering, etc. It is known in the
present day that for most rocks, such as foliated metamor-
phic rocks, stratified sedimentary rocks, and regularly jointed
rocks, their properties might exhibit both anisotropic and
inhomogeneous characteristics. The mechanical responses of
anisotropic media with spatial gradients in composition are
called anisotropic functionally graded materials (FGMs), and
their characteristics are widely employed in various fields of
applied mechanics. Hence, in this study, the solutions of body-
wave velocities for continuously inhomogeneous orthotropic
rocks with Young’s moduli (Ex , Ey , Ez ), shear modulus (Gxy ,
Gyz , Gxz ), and the medium density (ρ) varying exponentially
with depth are derived.The governing equations for the present
solutions can be combined with the generalized Hooke’s
law, the strain-displacement relationships, and the equilibrium
equations. Then, supposing the direction of wave propagation
is on the x-z plane, the solutions of three quasi-wave veloci-
ties can be yielded. The proposed solutions are identical with
those of Tsvankin (1997) when the inhomogeneity parameter Based on the inhomogeneity parameter, α, the following
(α) is equal to 0. situations exist:
• α > 0 indicates a hardened surface; all Young’s and shear
moduli decrease as the depth increases.
2 THE PHASE VELOCITY SOLUTIONS
• α = 0 is the conventional homogeneous condition.
• α < 0 indicates a soft surface; all Young’s and shear moduli
The expression of the stress-strain relationship for a continu-
increase as the depth increases.
ously inhomogeneous orthotropic medium is:
Then, the strain-displacement relationship for a small strain
in a Cartesian co-ordinate system, and the differential forms of
equations of motion can be presented in terms of displacement
vector components, ux , uy , uz as:
where α is the inhomogeneity parameter (unit: length−1 ); Cij

(i, j = 1–6) are the elastic moduli or elasticity constants of
the orthotropic medium. Since νij /Ei = νji /Ej (j = x, y, z), and
513

in which
Figure 1. Definitions of phase angle and group angle. Eventually, the three body waves, VSH , VSV , and VP , are
derived for an inhomogeneous orthotropic medium as:
where:
The direction of wave propagation is assumed on the x-z 3 CONCLUSIONS

plane; thus, the displacement components for each particle
can be assumed as (Wang et al., 2010): The solutions for body wave velocities, VP , VSV , and VSH , for
a continuously inhomogeneous orthotropic medium, in which
Young’s moduli (Ex , Ey , Ez ), shear modulus (Gxy , Gyz , Gxz ),
and the medium density (ρ) varied exponentially as depth
increased are derived by solving the eigenvalue problem in
this article. However, three Poison’s ratios (υxy , υyz , υxz ) of
the orthotropic material are remained constants regardless of
depth. The yielded solutions (Equations (5)–(7)) can be uti-
lized to perform a series of parametric study to show the
influenced factors including (1) the inhomogeneous param-
eter; (2) the type and degree of rock anisotropy, (3) the phase
where A1 –A3 are non-zero constants, θ is the phase angle angle, and (4) the medium density. The proposed solutions are
between the normal wavefront and the unique axis, as shown compared with homogeneous orthotropic solutions given by
in Figure 1, and V is the desired body-wave velocity. Tsvankin (1997) when the inhomogeneity parameter, α is 0;
An anisotropic medium generally has three body waves all solutions are in good agreement. Results strongly reveal
propagating with velocities that vary as the direction of that the inhomogeneous characteristics should be considered
phase propagation varies. Their polarizations are orthogonal for waves propagating in the orthotropic rocks.
and fixed for a particular phase propagation direction. The
waves are called quasi-waves as polarizations may not be
along dynamic axes. Hence, the following matrix equation REFERENCES
is presented as:
Tsvankin, I. 1997. Anisotropic parameters and P-wave velocity for
orthorhombic media, Geophysics, 62, 1292–1309.
Wang, C.D., Lin, Y.T., Jeng, Y.S. & Ruan, Z.W. 2010. Wave propaga-
tion in an inhomogeneous cross-anisotropic medium, International
Journal for Numerical and Analytical Methods in Geomechanics,
34, 711–732.
514

Experimental investigation on dynamic mechanical performances of granite
Wanpeng Wang, Yongle Hu, Xintao Ren & Yibo Xiong

Northwest Institute of Nuclear Technology, Xi’an, Shanxi, China
Keywords: dynamics, rock failure, lab testing, rock mass
ABSTRACT: In mining, blasting, earthquakes, rock bursts by mechanical drilling, cutting and grinding. For impact flat-
and defence engineering involving penetration of missiles, tened Brazilian disc specimens, two parallel flat ends which
rock failure pattern, strength and breakage are often affected central angle is 25◦ are introduced to the disc for load bearing.
by loading rate. Therefore, it is of great importance to inves- This loading manner is superior to the original line loading of
tigate the effect of the loading rate on rock failure parameter the Brazilian test in avoiding local cracking, breakage or yield-
and fragmentation. It is a well-known fact that rock is an brit- ing around the loading point caused by stress concentration.
tle porous medium compositing of mineral grain, microcrack A SHPB apparatus which is 74 mm diameter was used to test
and pore space, this is mainly due to difficulty in theoreti- dynamic mechanical properties of granite. Brass pulse shapers
cal analysis and experimental study on rock at high loading with 12 mm diameter and 2 mm thickness were designed to
rates. So it is active demand on systematically dynamic testing improve incident wave forms, so that granite specimen has
of granite. The split Hopkinson pressure bar technique orig- enough time to obtain stress equilibrium, at same time.
inally were developed by Kolsky (1949), as one of method Besides, brass pulse shapers can smooth wave forms and
for testing dynamic mechanic character, the Split Hopkinson eliminate wave form oscillations. For test of impacting flat-
Pressure Bar (SHPB) technique which based on assumption of tened Brazilian disc, incident bar and transmission bar close
one-dimensional stress wave propagation and the assumption contact with specimen on two parallel flat ends, the specimen
of stress uniformity along the specimen thickness, has been is subjected to diametral impact. Strain gauges are mounted at
widely used in studies on the dynamic mechanical behav- the center of two sides of the disc respectively, and the average
ior of rock materials at high strain rates. Shan (2000), Liu of the two measured values is taken for analysis.
(2000), Li (2005) and Zhai (2006) have reported the result Before impact testing, the quasi-static compressive tests of
of granite under dynamic compression. The tensile strength granite specimens are performed with MTS810 test machine.
of rock material is typically an order of magnitude smaller Five kinds strain rate are selected in impact compression, and
than the compressive strength, tensile failure often occurs in four specimens at least are used in each kind strain rate impact
rock masses. However, a direct tension test of rock is very compression. The experiment results showed that the dynamic
difficult to perform, so that an indirect tension, such as the strength raised quickly and specimens damage significantly
Brazilian test, also called the split tensile test, has been a pop- contest with static compression. The stress increases linearly
ular choice. This splitting test method has been widely used initially and then decreases after the stress achieves the peak
in static indirect tension for rocks, concrete and many other value. The strain rate effect on the peak stress and strain can
materials. WANG and WU (2004) proposed introducing two be seen clearly. That is to say, granite is strain rate depen-
parallel flat ends to the disc specimen for loading, in this way a dent materials, the dynamic increase factor (DIFC) is directly
BD becomes the Flattened Brazilian Disc (FBD). The FBD is depend on the strain rate, and increase approximately lin-
favorable for reducing stress concentration effect, thus ensur- early with the increase of average strain rate. The compression
ing crack initiation from the center of the disc instead of from experiments show that the failure of the granite specimen was
the loading point. typical axial cleaving failure mode under the direction of uni-
However, as we know, few systematically dynamic testing axial compressive loading at different strain rates. Besides,
results about granite at relatively high strain rates in compres- it was concluded not only the strength of granite increased,
sion and tension were ever reported, despite it is very important but also the fragment size decreased and fragment numbers
for design of underground structure. In this study, the impact increased with the increasing strain rate, the specimens will
compressive and impact flattened Brazilian behaviors of gran- be crushed under higher strain rate.
ite at different strain rates are performed by split Hopkinson The impact splitting tension experiment results showed that
pressure (SHPB) techniques, the dynamic strength, failure the dynamic tension strength, peak strain will raise quickly
strain, failure pattern and other parameter at strain rates of with strain rate increment, the ratio of tension strength of
2.3 s∼108.4 s are obtained from impact test. dynamic and static is about 1.82∼2.82, the tension strength
The reddish-brown granite taken from northwest of China increase approximately linearly with the increase of average
were tested; its Poisson’s ratio is 0.24, Young’s modulus is strain rate. It shows that granite specimen will split into two
50 GPa, and density is 2650 kg/m3 . The cylindrical samples almost complete halves along the loading diameter after test
with 60 mm in diameter and 25 mm in thickness are obtained when specimen was subjected to a relatively low strain rate.
515

The application of smooth blasting for freezing rock in vertical shaft
X.G. Wang
China University of Mining and Technology, Beijing, China
Jizhong Energy Group Company Limited, China
E.X. Gao, Z. Wu, L.L. Zhang & L. Li

China University of Mining and Technology, Beijing, China
State Key Laboratory for Geomechanics and Deep Underground Engineering, Beijing, China
Keywords: blasting; coal mines; mine design
1 INTRODUCTION Considering the factors of frozen and due to the relevant exper-
iment, the empirical formula concerning the calculation of the
The vertical shaft in Xuandong is about 832.0 m in depth and Protodikonov’s hardness coefficient is made as follows:
6.0 m in diameter and its frozen foundation rock is 36.5 m
high.
The method of staggered freezing is adopted in order to
freeze to the rock of shaft. In the method of staggered freezing,
the long tube is 85 m and the short tube is 70 m, and the single Here, f is frozen Protodikonov’s hardness coefficient and f1
hole and the double hole are arranged respectively. There are is Protodikonov’s hardness coefficient before freezing. T1
26 frozen holes altogether, placed around the circle diameter is absolute value of the frozen stone temperature and T0 is
of 9.8 m. absolute value of the freezing point temperature for that stone.
The modification of frozen Protodikonov’s hardness coef-
ficient is as fallows: andesites: f = 6, orthophyre: f = 8,
2 GEOLOGY AND HYDROGEOLOGY dolerite: f = 7.
The blasting parameters are designed basing on the radius
In the 41.5 m depth of the main shaft, there are the andesites of 3800 mm of the net section. When the compound cut is
with growing cranny in length of 27.2 m, covered with the executed, depth of the burn cut is L1 = 3.2 m, with the diam-
watermark of iron rust in the joints. It also includes the amphi- eter of arranging for burn cut D1 = 1.4 m. The number of
bole and calcite which fills the cranny. Down to the 68.7 m holes arranged is N1 = 6. With the distances between holes
of the shaft, a yellow orthophyre of 4.8 m appears in thick is E1 = 733 mm. Every hole packs 6 volumes of water-gel
layers structuring in cryptocrystalline. The phenocryst is com- explosives. Depth of the second cut hole is L2 = 3.2 m. With
posed of the orthoclase, filled the cranny with calcite. In the the diameter of arranging is D2 = 2.20 m. The line of least
depth of 73.5 m, there is diabase composed by the pyroxene, resistance is W1 = 400 mm. The number of holes arranged is
anorthose and chlorite. The thick-layered celadon diabase is N2 = 9, and the distance between holes is E2 = 768 mm and
4.5 m in length. Water inflow of the bedrock before freezing is 5 volumes of water-gel explosive are packed every hole.
207 m3 /h, while no inflow water is there after freezing except Depth of the first, second and third satellite hole is
the “dead water” in the shaft. After frozen, the shaft walls are L3 –L5 = 3.0 m.With arranging diameters are D3 = 3.60 m,
also in good shape. D4 = 5.20 m, D5 = 6.60 m respectively. The line of least resis-
According to the “Geological Survey Report on Shaft tance are W2 = 700 mm, W3 = 800 mm, W4 = 700 mm. The
Review of Xuandong Shaft”, uniaxial compressive strengths number of arranged holes are N3 = 12, N4 = 15, N5 = 24.The
of the andesites, orthophyre and dolerite are 52.1 MPa, distances between holes are E3 = 942 mm, E4 = 1089 mm,
70.1 MPa and 65.9 MPa respective, while the rock E5 = 864 mm respectively. The first and second satellite holes
Protodikonov’s hardness coefficients are 5.2, 7.1, and 6.6 are packed 4 volumes of water-gel explosives and 7 volumes
respectively. of ammonium nitrate explosives for the third satellite holes.
Depth of the periphery holes is L6 = 3.0 m, with arranging
diameter of D6 = 7.40 m. The tilt angle is 81◦ 10 . The line
of least resistance is W5 = 400 mm. The number of arranged
3 CONSTRUCTION METHOD IN FROZEN ROCK
holes is N6 = 45.The distance is E3 = 942 mm and 6 volumes
of water-gel for every hole. Since the distance between the
3.1 Analysis of blast-ability
periphery holes is larger than the line of least resistance, so that
It shows some increasing in rock compressive strength and the empty holes arrayed in the proper places. In the whole sec-
Protodikonov’s hardness coefficient when the rocks are frozen. tion periphery, the number of holes is N = 111 and 156.87 kg
516

T220 water-gel explosive and NO.3 water-resistance ammo-
nium nitrate explosives are used, the water-gel explosive
is 45 × 400 mm. Each volume is 0.8 kg and density is
1.0∼1.25 kg/cm3 . Brisance is 16∼18 mm and explosion
strength is 350 ml. Detonation velocity is 4100∼4600 m/s.
The NO.3 water-resistance ammonium nitrate explosives
is 32 × 170 mm. Each volume is 0.15 kg. Density is
1.10 g/cm3 , detonation velocity is 3996 m/s and explosion
strength is 290 ml.
There are no the fracture for frozen tube and the deforma-
tion phenomenon happen. The freezing sections are without
accidents.
The construction progress is accelerated and project engi-
neering cost is reduced. This valuable experience of the
construction of shaft in Xuandong No.2 coalmine is benefit
for the similar projects construction henceforth.
4 CONCLUSIONS
During the construction of shaft frozen foundation rock in

Xuandong 2 Shaft, the depth of blast hole is to 3.2 m, which
Figure 1. The disposal of borehole. is 1.4 m deeper than specification requirements.
Designed distance between the blast hole and the freezing
of water-gel explosives, 65.70 kg of ammonium nitrate explo- pipe is 1.2 m; the use of the millisecond delay electric detona-
sives for the blasting, charge quantity is 222.57 kg and the tor and water gel explosive is successful regarding the full-face
powder factor is q = 1.92 kg/m3 . blasting.
3.2 Rock drilling machines and blasting materials

FJD—6 Umbellate drill and 6 YGZ—70 Guide rail type
independent cyclotron rock drill are used. The drill pipe is
25 × 3300 mm hollow hexagon bar, with 52 mm crossed
alloyed steel broach.
517

Fluid and gas flow

Experimental and numerical study on the behavior of supercritical CO2 injected to

low permeable rocks
A. Arsyad, Y. Mitani, H. Ikemi, K. Kuze & S. Oura

Department of Civil and Structural Engineering, Graduate School of Engineering, Kyushu University, Fukuoka, Japan
Keywords: fluid flow, rock properties, lab testing, numerical modeling
Over the past several hundred years, CO2 emissions in the A new developed flow pump permeability test was set up
atmosphere has increased steadily and risen over the pre- in a laboratory system where pressure and temperature can
industrialized level due to intensified human activity produc- be controlled. The entire laboratory system was isolated from
ing greenhouse gas emissions such as burning fossil fuels for external temperature change. All the equipments were placed
electricity generations, industries, transportations and domes- in a greenhouse chamber controlled with air conditioning.
tic uses. The increase of atmospheric CO2 emission has In case of internal temperature effect induced by laboratory
become a major contributing factor of a gradual raise of measurement heat, the pressure vessels and pipes includ-
the earth’s temperature, popularly known as global warming. ing syringe pumps were double insulated and connected to
Recently, the methods of reducing CO2 emission have been hemathermal circulation tanks and controller, so theirs tem-
developed including carbon capture and geological storage perature changes can be minimized as low as possible. The
(CCS). So far, CCS in depleted oil and gas reservoirs is the rock specimen used in this study was isotropic Ainoura sand-
most readily applicable since the technique is similar to the stone obtained from Nagasaki Prefecture Japan. The porosity
enhanced oil recovery (EOR) commonly applied in petroleum and specific gravity of the sandstone is 12.6% and 2.88 respec-
industries (IPCC, 2005). Even though oil and gas reservoirs tively. At the initial condition of 35◦ C temperature, 20 MPa
remain worldwidely available, these reservoirs are inequitably confining pressure and 10 MPa hydraulic pressure, supercrit-
distributed around the world. Therefore, deep saline aquifers ical CO2 was injected to the rock specimen saturated with
are recently investigated since their estimated storage capac- water. The upstream and downstream pressures were mea-
ities is far largest compared to the other formations. sured. To investigate the effect of CO2 state on the behavior of
A number of researchers have conducted the study of CO2 CO2 injected to the specimen, we also conducted the injection
storage in deep saline aquifers. In a core scale, Perrin and of CO2 in liquid state in the same initial condition, except the
Benson (2009) undertook the injection of CO2 and brine temperature of 20◦ C.
to sandstone obtained from the CO2CRC-Otway Project in The experimental results of supercritical CO2 injection
Australia. By employing a CT-scan, they observed the migra- show that the hydraulic pressure increases gradually. In com-
tion of CO2 along the sandstone pores and estimated CO2 parison to the liquid CO2 injection, the magnitude of the
saturation that changed during the injection. They found that hydraulic pressures generated by the supercritical CO2 injec-
heterogeneity of sandstone porosity has significant impact tion is lower than that generated by the liquid CO2 injection.
on the efficiency of CO2 migration. Shi, et al., (2010) per- However, the supercritical CO2 injection can yield a higher
formed CO2 injection to Tako Sandstone saturated with brine. differential pressure across the specimen than that by liquid
They also utilized a CT-scan to map the saturation profiles CO2 injection.
during the injection. Their findings show the heterogeneity Based on the pattern of differential pressure obtained from
effect of porosity is insignificant for the case of high flow rate the experiment, it can be suggested that there are four stages
employed in injecting CO2 . However, their studies do not proof conditions occurred during the injection of supercritical
vide adequate information about the relation of the increase CO2 into the specimen. At the pre-injection condition, the
of hydraulic pressure and the rise of CO2 saturation driven by water occupies most of the specimen pores and no CO2 exists
very low hydraulic gradient as representation of laminar flow (initial stage). After the injection started, CO2 begin to flow
in deep underground. and penetrates the bottom of the specimen (Stage 1). As the
In this paper, we present a new developed flow pump perme- injection continuing, more fractions of CO2 penetrate and
ability test to investigate the behavior of CO2 in low permeable occupy the specimen pores while more saturated water dis-
rocks. The method was originally introduced by Olsen et al., placed out (Stage 2). In this period, supercritical CO2 has
(1985), which has been able to determine permeability char- been able to penetrate the top of the specimen indicated by a
acteristics of low permeable. However, several developments drop of hydraulic pressure measured in the downstream gauge.
are needed in order to enhance the method’s ability of creating Finally, most of CO2 occupies the specimen pores and just a
similar physical conditions of geological site for storing CO2 , few amounts of water trapped (Stage 3).This stage is the period
expected to located in the depth of 800–1200 meters, where of the differential pressure declines and stabilizes to a steady
CO2 performs as supercritical fluid. Moreover, the develop- state. It seems to take a long time to accomplish due to very
ment of the method is conducted so as to work in very low low flow rate of the injection employed in the experiment.
hydraulic gradient of CO2 flow, due to groundwater flow in
deep underground is generally laminar.
521

A fully coupled thermoporoelastic model for drilling in HPHT formations
A. Diek, L. White & J.-C. Roegiers

The University of Oklahoma, Norman, OK, USA
D. Blankenship
Sandia National Laboratories, Albuquerque, NM, USA
Keywords: Geothermal, oil reservoir, physical modeling, numerical modeling, fluid flow
Coupled processes in porous media impact drilling and bore- evolution of the potential energy is obtained by combining the
hole stability. The coupling of the matrix deformation, and localized internal energy and entropy balance equations which
fluid and heat diffusions results in a time dependent response satisfy the first and second law of thermodynamics, respec-
in the formation. During the drilling of an oil or gas well, tively. The rock constitutive equations describe the response
temperature gradients between the drilling mud and the for- of the rock to mechanical and thermal loading. They express
mation can significantly change the pore pressure and stress the temporal evolutions of the total stresses, the variation of
distributions around the borehole. A general theory of ther- the fluid content, and the rock entropy in terms of the above
moporoelasticity is developed that fully couples the three pro- five independent variables. The transport relations that couple
cesses: mechanical, hydraulic, and thermal in porous media the influx and efflux of fluid and heat to their driving forces
saturated by a fluid. The theory satisfies the first and second are derived from the definition of the internal entropy produc-
laws of thermodynamics and is based on concepts of irre- tion rate and its associated generalized forces. The transport
versible thermodynamics, a novel rock constitutive relation, fluxes of fluid and heat are established through Onsager’s phe-
and Onsager’s transport phenomenology. Fully coupled field nomenology and concepts of irreversible thermodynamics.
equations are developed in terms of five independent vari- The temperature is locally equilibrated between the fluid and
ables, namely the three solid displacements, the pore pressure, the solid.A finite element model is developed for the fully cou-
and the absolute temperature. These equations are obtained pled processes consisting of: thermoporoelastic deformation,
by implementing the rock constitutive and transport equations hydraulic conduction, thermal osmosis, heat conduction, pres-
into the momentum, fluid mass, and energy balance equations. sure thermal effect, and the interconvertibility of mechanical
The rock constitutive equations are derived from a state func- and thermal energy. The model is used to analyze the problem
tion that describes the manner in which the potential energy of of a wellbore subjected to a hydrostatic in situ stress field.
the porous media changes with respect to time. The temporal Distributions of total stresses, pore pressure, and temperature
Figure 1. Induced pore pressure distribution around the borehole Figure 2. pressure distribution around the borehole for the cooling
for the heating case due to thermal loading at times of 0.05 (blue), case due to thermal loading at times of 0.05 (blue), 0.1 (green), 0.5
0.1 (green), 0.5 (red), and 1 (cyan) day. (red), and 1 (cyan) day.
522

are determined around the borehole for various cases. The of heating and less compressive in the case of cooling rela-
results indicate that the mud or borehole temperature has an tive to the isothermal or poroelastic case. It is also shown
important impact on stability, namely heating the mud increase that the lower the permeability/fluid viscosity ratio, the larger
the pore pressure, and the radial and tangential total stresses the thermal loading effect on the pore pressure changes in
around the borehole relative to the isothermal or poroelas- the vicinity of the borehole. Moreover, the partially coupled
tic case, and vice versa. That is, heating induces an increase model overestimates the pore pressure for the heating case, and
in the pore pressure, while cooling induces a decrease in underestimates the pore pressure for the cooling case. From
the pore pressure. The change of the pore pressure is large the numerical results, it could be concluded that the full cou-
at short times; however, it decreases and moves deeper into pling, and the mud temperature and rheology are important
the formation with longer times (Figures 1 and 2). The total criterions of wellbore stability analysis in tight formations.
radial and tangential stresses are more compressive in the case
523

Hydraulic stimulation of geothermal wells: Modeling of the hydro-mechanical behavior of

a stimulated fractured rock mass
S. Gentier, X. Rachez, M. Peter-Borie & A. Blaisonneau

Department of Geothermal Energy, BRGM, Orléans Cedex 2, France
Keywords: geothermal, numerical modeling, hydraulic fractures, fluid flow, case studies
1 INTRODUCTION
Experiments were initiated at the Soultz-sous-Forêts site

(France) in 1987 with the goal to produce electricity from the
heat available in the granite basement. Three boreholes were
drilled down to 5 km depth. For such a system, the heat extrac-
tion efficiency depends not only on the flow pathways within
the pre-existing fractures but also on our ability to connect
with efficiency the wells to the pre-existing fracture network.
Understanding the mechanisms involved in hydraulic stimu-
lation is a key issue in the optimization of an EGS reservoir.
Up to now, modeling was limited to the pressurization of the
wells, but the return to the initial pressure in the well has
not been taken into account yet. That is only by consider-
ing the up and down pressurization that the real impact of
the hydraulic stimulation in terms of irreversible permeability Figure 1. Pressure vs. Flowrate calculated in each fault zone in the
could be evaluated. well GPK4.
2 THE NUMERICAL MODEL The shapes of the curves are slightly different. Three of
the six fault zones in which most of the flowrate is injected
To simulate the behavior of the fractured rock mass in the (3 to 5 l/s), are roughly parallel to the maximum horizontal
vicinity of the wells during their hydraulic stimulations, a 3D stress. But, among the three others, one is oblique (at around
Distinct Element Method code was used. 30◦ ) and the two others are sub-perpendicular to the maximum
The numerical model is a 400 m × 400 m × 1000 m paral- horizontal stress. The three fault zones which react first to the
lelepiped volume centered on the stimulated open hole and stimulation, between 6 and 10 MPa, are the fault zones sub-
in which a network of nine fault zones has been defined from parallel to the maximum horizontal stress. For overpressures
various well data. The fault zones are roughly sub-vertical and greater than 10 MPa, the permeability increase is higher in two
four of them are sub parallel to the maximum horizontal stress. other fault zones whose one is sub perpendicular to the max-
The distribution of the initial fluid pressure in the frac- imum horizontal stress. The maximum shear displacement is
tured network is assumed to behave like a hydrostatic field. around 2 cm. The analysis of the evolution of the shear and
Zero displacements are imposed at the North (x = z = 0), West normal displacements shows that the stimulation cannot be
(x = z = 0) and bottom faces (y = 0) and stresses are applied considered in each fault zone independently of the other. The
on the South, East and top faces. combination of shearing in some fault zones can lead to irre-
All fault zones have the same mechanical constitutive law. versible opening of another fault zone which intersects them
The normal mechanical behavior is elastic linear, while the and not necessary sub-parallel to the maximum horizontal
fault zone is in compression, and tensile strength is null. The stress.
tangential mechanical behavior is elasto-plastic. It follows
a Mohr Coulomb failure criterion with associated dilation 3.2 GPK4 Well-shutdown modeling
effects. The effects of dilation appear as soon as the maxi-
mum shear strength is reached. The normal displacement and After the application of several stages of increasing pressure
the hydraulic apertures are directly related on the basis of an in the well, the pressure is shut down by returning to a zero
initial hydraulic aperture. flowrate at the wellhead.
At the end of the stimulation test, only a specific part of
the fault zone network shows a significant irreversible perme-
3 RESULTS
ability increase (Fig. 2), but the locations of these areas do
not systematically permit a connectivity increase. Above the
3.1 GPK4 Hydraulic stimulation test
potential local permeability increase which depends on the
A plot of the overpressure in the open hole versus the flowrate fault zone orientation regarding to the maximum horizontal
calculated in each fault zone is shown in Figure 1. stress, the whole network must be considered.
524

Figure 2. Shear displacement contours [m] and hydraulic aperture contours [m] before and after well-shutdown in some fault zone.
Figure 3. Calculated pressure at wellhead vs. flowrate in some fault Figure 4. Pressure vs. total flowrate at wellhead during stimu-
zones during stimulation test and post-stimulation injection. lation test and post-stimulation injection test (model and in-situ
experimental curves of the site).
3.3 GPK4 Post-stimulation injection test modelling

a 12 MPa overpressure, no more real gain in permeability can
After the pressure is shut down in the well, a post-stimulation be observed. This point has been usually noted on the in-situ
injection test is performed by applying again increasing over- experimental curve.
pressure stages at the wellhead. Plots of pressure at the
wellhead vs. the injected flowrate during the stimulation test
and during the post-stimulation injection test are shown in 4 CONCLUSION
Figure 3. The comparison between the response during the
stimulation test and post-stimulation shows very clearly how The adopted numerical approach successfully simulated the
the irreversible increase of the permeability in the fault zones behavior of a fractured rock mass during hydraulic stimula-
sub-parallel to the maximum horizontal stress modifies the tions of a deep well and post-stimulation injection tests after
ratio pressure-flowrate for overpressures lower than 10 MPa. the well was shut down. The results obtained show that we
Indirectly, the shearing of those fault zones permits a simulta- can explain the shape of the experimental in-situ stimulation
neous irreversible opening of a fault zone sub-perpendicular curves and the irreversible permeability increase due to dila-
to the maximum horizontal stress. tancy in the fault zones in relationship with the stress state.
Consequently, for the 6 MPa overpressure stage, the total However, a more complex behavior can result from the geom-
injected flowrate into the well is about 10 times greater than etry of the global network and from the resulting intersections
the flowrate injected during the stimulation test (Fig. 4) and is like irreversible normal opening and channelization in fault
still about 5 times greater for a 9 MPa overpressure stage. Over plane.
525

Investigation of thermal effect on stability of deep wellbores in

low and high permeable formations
Z. Hou
Institute of Petroleum Engineering, Clausthal University of Technology, Clausthal-Zellerfeld, Germany
Xuan Luo
Energy Research Centre of Lower Saxony, Goslar, Germany
Keywords: geothermal stability analysis
1 INTRODUCTION
The stress distribution around the borehole is one of the most

important factors affecting the stability of wellbores, e.g. deep
geothermal wells. Thermal effect in form of thermal stress
will change the stress distribution near the borehole, because
the heating expansion or cooling shrinkage is restrained in
the near-field of deep wellbores during a drilling process.
Furthermore, the thermal stress affects the wellbore stability.
Temperature change under conduction and advection con-
dition can be calculated, among others, by using the finite
difference method. With a known temperature distribution the
thermal stress and a suitable mud pressure can be determined
e.g. by using numeric integration method.
Figure 1. Temperature distribution in the near-field of a deep
2 TEMPERATURE DISTRIBUTION IN CONDUCTION wellbore (z = 5000 m) under the conduction condition.
AND ADVECTIONS PROCESS
2.1 Thermal conduction

The differential equation of temperature conduction is well
known and can be described in a cylindrical coordinate system
(Tang & Luo 1998). The finite difference approximation of the
temperature distribution is realized in this paper by writing a
program with C#.
Figure 1 shows that the temperature in the formation at a
depth of 5000 m decreases with time under cooling condi-
tion in conduction process. Far from the borehole and at the
borehole wall the temperatures stay constant because of the
boundary conditions (Tmud = 100◦ C, TRock-initial = 150◦ C).
Figure 2. Temperature distribution in the near-field of a deep
wellbore (z = 5000 m) under the advection condition.
2.2 Thermal advection
The thermal advection is the heat transport through a moving
fluid (Chen & Russel 2005). In advection process the temper- 3 THERMAL STRESS UNDER CONDUCTION AND
ature distribution depends strongly on the rock permeability. ADVECTION CONDITON
Figure 2 shows the temperature distribution in the near-field
of the same deep wellbore as in Figure 1, but under the advec- According to Tang & Luo (1998), the radial thermal stress
tion condition. It is found that after 860 s, the bigger is the equals zero at the borehole wall and in the remote distance.
permeability; the larger is the formation area cooled to the The thermal stresses in conduction (TM) and conduction-
mud temperature. advection process (THM) are shown in Figure 3. Negative
526

Figure 3. Tangential thermal stress distributions in the near-field
of a deep wellbore (z = 5000 m) under the conduction (TM) and Figure 5. Minimum required mud pressure minpi. depending on the
conduction-advection (THM) process. Biot’s coefficient α (z = 5000 m).
or cooling shrinkage is restrained. Thermal stress will affect

the stress distribution near the wellbore and furthermore the
wellbore stability. In this paper, the temperature changes and
the corresponding thermal stress in the near-field of a deep
wellbore (z = 5000 m) are analyzed in consideration with
the conduction and advection processes by using the own
developed program based on the finite difference method.
Furthermore, the minimum mud pressure is determined for
the same wellbore under the same conditions.
It is found that the thermal stress distribution in the
conduction-advection process does not differ from it in the
conduction process in case of a very low permeability (e.g.
10−20 m2 ), while the advection process dominates in case of
a relatively high permeability (e.g. >10−15 m2 ). A thermal
Figure 4. Stress distributions in the near-field of a deep wellbore tensile stress in the tangential direction is induced under the
(z = 5000 m). cooling condition (Tmud < TRock-initial ), leads to a reduction
of the effective tangential stress and thus to a stabilization of
tangential thermal (tensile) stresses near the borehole are the borehole during the drilling process. This cooling effect
induced under the cooling condition with different rock per- reduces furthermore the minimum required mud pressure and
meability (Fig. 3). In case of a very low permeability (e.g. makes the drilling more economic.
10−20 m2 ), the thermal stress distribution in the conduction- All of the above results are based on the cooling condition,
advection process does not differ from it in the conduction which presents the normal case during a drilling process. With-
process. If the permeability bigger than 10−15 m2 , the advec- out mud circulation, the reduced temperature in the near-field
tion process dominates in heat transport. Considering the of the wellbore will be increased as well as recovered until to
thermal stress, the stress distribution is altered (Fig. 4), and the the initial formation temperature after an enough time. Vice
minimum required mud pressure minpi is also reduced clearly versa, the heating effect must be considered.
under cooling condition (Fig. 5). We can conclude that a ther-
mal tensile stress in the tangential direction is beneficial for
the borehole stability during the drilling process. It reduces REFERENCES
the minimum required mud pressure and makes the drilling
more economic. Chen, G. & Russel T.E. (2005). Thermoporoelastic Effect on Well-
bore Stability, SPE 89039, March 2005.
Tang, L. & Luo, P. (1998). The Effect of the Thermal Stress on Well-
4 CONCLUSION bore Stability. SPE 39505, presented at the 1998 SPE India Oil and
Gas Conference and Exhibition held in New Delhi, India, 17–19
Thermal stress is significant for the wellbore stability, espe- Feb. 1998.
cially during drilling of deep wellbores (e.g. deep geothermal
wells). Thermal stress occurs only when the heating expansion
527

Modelling and optimization of multiple fracturing along horizontal wellbores

in tight gas reservoirs
Z. Hou
L. Zhou
Keywords: hydraulic fractures, numerical modeling
1 INTRODUCTION
By use of the horizontal boreholes, a remarkable increase of

the gas and oil productivity has been achieved comparing with
vertical boreholes. However, in order to maximize the produc-
tion capacity, it is still necessary to operate the HF (hydraulic
fracturing) in the horizontal boreholes, especially in tight gas
reservoir. In late 80’s, hydraulic fracturing was applied in hor-
izontal boreholes and has been further developed since then.
The pre-calculation is very necessary for planning and opti-
mizing a multiple fracturing operation. However individual
fractures can be treated by the known analytical simulators
only separately, so that the interactions on fracture geometry
and permeability between the neighboring fractures can not Figure 1. Geometric model with stratigraphy and the position of
be considered. By contrast the full 3D numeric code FLAC3D each injection point in the model.
is possible to simulate the multiple fracturing, because in
FLAC3D all of the fractures can be integrated in one model,
therefore such interactions can be considered.
2 SIMULATION OF THE TRANSVERSE MULTIPLE

FRACTURES AT LOCATION A
The model generation in FLAC3D begins with the stratigra-

phy (Figure 1). All of the layers lie between −4290 m and
−4500 m. Because of the y-axis-symmetry only ½ of the total
model is built, which has a dimension of 930 m (x) × 200 m
(y) × 210 m (z).
After the model generation and parameter input the Figure 2. Propagations of eight fractures at the end of the multiple
hydraulic calculation will be firstly carried out during an fracturing.
injection time interval t. The change of pore pressure is
induced by fluid injection. Then the mechanical calculation a loop computation until the accumulative time reaches the
will be turned on. Due to the new pressure distribution the injection duration.
total system is no longer in a mechanical equilibrium state. In Figure 2 all eight fracture propagations at the end of
The stresses must be redistributed to achieve equilibrium. The the injection are shown. Each one has a different fracture
redistribution is divided into two parts. One part occurs under shape and geometry compared to the others. The difference is
an elastic state and the other under a plastic state. The crite- caused by different in-situ stresses, pore pressures, injection
ria used to describe the plastic behavior are Mohr-Coulomb rates, injection volumes and also breakdown pressures. The
with strain softening for shear failure and the tensile fracture breakdown pressure depends strongly on the in-situ stresses,
model with strain softening for tensile failure. If an element injection rate and reservoir pressure. It could be calculated by
fails, then its strength will be reduced and the permeability using FLAC3D, but in our calculations the breakdown pres-
will increase. After this step the mechanical calculation will sure comes from the measured data (GdF SUEZ 2009) in order
change to the hydraulic calculation. The two processes build to match the pressure development at bottom hole. It is found
528

From fracture 1 to 8 the fracture average half length is
between 96 m to 152 m, the height between 89 m to 117 m and
the width between 13 mm to 49 mm. The difference of frac-
ture height between each fracture is obviously smaller than
the fracture length difference, because the fracture barriers at
bottom and top limit the fracture propagation in z-direction. In
comparison with the analytical pseudo-3D simulator FracPro
most results match well and are comparable. The only differ-
ences are the length of fractures 2 and 8, the height of fracture
8 as well as the width of fractures 2 and 5. In our calculation
the fracture half length and height depend mostly on in-situ
stress especially σh , breakdown pressure, initial pore pressure,
Figure 3. Average half length of each fracture at the end of injection injection volume and rate. At fracture 2 the breakdown pres-
and in comparison with FracPro. sure and the initial pore pressure are very low. So it is difficult
to increase the pore pressure in reservoir to break the rocks.
That is the reason why the fracture propagation is so slow.
Meanwhile the small fracture area and pore pressure lead also
to a small fracture width. The geometry of fracture 8 is just
on the opposite compared with fracture 2 and 5.
3 CONCLUSION
Multiple fracturing along a horizontal wellbore in a tight gas

reservoir in North Germany is simulated and optimized by
using a newly developed method in the FDM code FLAC3D. In
comparison with the analytical pseudo-3D simulator FracPro,
the numerical code FLAC3D with the own developed method
Figure 4. Maximal fracture height at the end of injection and in has the following advantages: all fractures in one calculation
comparison with FracPro.
model, irregular fracture geometries in different rock for-
mations, both shear and tensile failure mechanisms etc. The
results show that horizontal wells should be drilled along the
direction of the minimum horizontal stress σh , because the
transversal fractures have totally a much bigger stimulated
volume than the corresponding longitudinal fractures. In this
study, all of the eight transversal fractures are generated in just
one model. From fracture 1 to 8 the modeled fracture average
half length is between 96 m to 152 m, height between 89 m
to 117 m and width between 13 mm to 49 mm. It is found
that the layers, in which the fracture propagates much faster
than other layers, are the same in all eight fractures. In these
layers the minimal horizontal stress σh is lower and the pore
pressure higher than the corresponding value in their neigh-
Figure 5. Maximal fracture width at the injection point of each bors. Three parameters (Young’s modulus, Poisson’s ratio and
fracture at the end of injection and in comparison with FracPro. fluid viscosity) are varied additionally, in order to investigate
how much these parameters affect the fracture propagation
that the layers, in which the fracture propagates faster than and geometry.
other layers, are the same in all eight fractures. In these layers
the minimal horizontal stress σh is lower and the pore pressure
higher than the corresponding value in their neighbors. REFERENCE
The fracture geometry of each transversal fractures at the
end of injection will be shown in Figures 3–4. Fracture 2 has GdF SUEZ (2009). Project data and reports (not published).
the smallest geometry and fracture 8 the biggest. GdF SUEZ E&P Deutschland GmbH.
529

Dependence of fracture geometry and spatial variation in pressure on hydraulic

conductivity in rock fractures
K. Kishida & T. Hosoda

Department of Urban Management, Kyoto University, Kyoto, Japan
A. Sawada & H. Sato

Japan Atomic Energy Agency, Tokai, Japan
S. Nakashima
Department of Civil and Environmental Engineering, Yamaguchi University, Ube, Japan
H. Yasuhara
Department of Civil and Environmental Engineering, Ehime University, Matsuyama, Japan
Keywords: fluid flow, rock joints, numerical modelling, lab testing
In order to discuss fracture flows, we must obtain the exact

aperture and the geometry of the rock fracture and conduct a
precise simulation, including the effect of the inertia term,
the pressure term, and the shear resistance from the frac-
ture walls. In this study, the aperture and the geometry have
been measured using an optical measurement system and the
flows through a single fracture have been calculated. Apply-
ing this precise numerical technique, the fracture flows have
been discussed for conditions under which the cubic law is
applicable.
In this work, the aperture measurement data conducted by
Sato and Sawada (2010) are used to calculate the flows in a sin-
gle fracture. Sato and Sawada (2010) made an artificial single
fracture in a granite sample and a replica specimen using the Figure 1. Contour map of aperture distribution obtained through
granite facture and a transparent epoxy resin. They applied optical measurement system.
an optical measurement system upon measuring the frac-
ture aperture distribution of the transparent replica specimen.
Figure 1 shows the distribution of the fracture aperture.
Hydraulic flow tests have been carried out under constant
hydraulic gradient conditions, such as several hydraulic head
differences of 0.25, 0.50, 0.75, 1.00 and 1.30 cm, respectively.
The experimental conditions are satisfied with Darcy’s law.
Moreover, the Reynolds number (Re) in each hydraulic head
difference is observed to be less than 2.0 in these experimental
results. It is thought that the experimental conditions, at least
those with a hydraulic head difference of less than 0.75 cm,
are satisfied with the cubic law condition.
Using the aperture distribution and the geometry of the
fracture surface obtained through the optical measurement
system, flow simulations through a single fracture have been
carried out. In this research work, the 2D model developed Figure 2. Comparison of experimental and numerical results for
the discharge-hydraulic head difference relation and discussion with
by Mgaya, et al. (2004, 2005) is applied. This model consid- influence of inertia term.
ers the effects of the inertia term, the pressure term, and the
shear stress of the fracture surface in Navier-Stokes equations. their results are also plotted in Figure 2. A good agreement
Figure 2 shows the discharge–hydraulic head difference rela- between the experimental results and those of the 2D model
tion obtained from both the experimental and the simulation can be confirmed in Figure 2. In addition, the 2D model is
results. Simulations through LCL have been conducted and more advantageous than LCL.
530

Analytical solutions for poroelasticity within a finite two-dimensional domain
P.C. Li
College of Mechanical Engineering, Shanghai University of Engineering Science, Shanghai, China
D.T. Lu
Department of Modern Mechanics, University of Science and Technology of China, Hefei, China
Keywords: fluid flow; rock stress
1 INTRODUCTION settlement when the height of the two-dimensional saturated

soil foundation was limited. Huang & Xiao (1996) proposed
In recent years, poroelasticity has been increasingly drawing the analytical solutions of consolidation when the lower sur-
our attentions. It has been widely applied to many engineer- face of the two-dimensional saturated soil foundation with
ing fields, such as consolidation of soft soils under loading, finite height was assumed to be impermeable and rigid.
municipal waste filling and nuclear waste disposal, reservoir- Ai & Wu (2009) took the effect of permeability anisotropy
induced earthquake, stability analysis of slopes, coal bed gas and derived the analytical solution of plane strain consoli-
explosion, land subsidence due to subsurface fluid withdrawal dation of saturated soil foundation with limited height into
(e.g. undergroundwater, oil and gas, geothermy), industrial account. Hu et al. (1998) presented the exact solution of plane
filtration in chemistry, biological soft tissue deformation, etc. strain consolidation in the integral form for the multi-layered
Poroelasticity originated from the research of Biot (Biot saturated soil foundation. Zhu et al. (2004) gave the function
1941a; Biot 1955; Biot & Willis 1957) on the three- solutions for the two-dimensional dynamic consolidation of
dimensional consolidation of saturated soft soils under load- single-layered saturated soil foundation subjected to periodic
ing. Since then, many studies concerning flow-deformation loading.
coupling models has been carried out, and numerous papers It is worth noting that the above analytical solutions were
have appeared. For brevity, only a very few of them are listed obtained under the assumption that the horizontal dimension
here (Lewis & Schrefler 1978; Zienkiewicz & Shiomi 1984; of the formation was infinite. However, in engineering prac-
Cheng 1997; Settari & Walters 2001; Gutierrez & Lewis 2002; tice, the horizontal dimension of specific physical problems
Li et al. 2003; Kihm et al. 2007). The overwhelming major- (such as saturated soil consolidation, deep excavation dewater-
ity of them are based on Biot’s consolidation theory although ing) is usually limited. Besides, using analytical solutions for
minor differences exist due to different assumptions or approx- infinite domains to verify numerical solutions is hard to avoid
imations. Biot’s consolidation model can solve pore water the error, because numerical solutions are always confined to
pressure and soil deformation simultaneously. It has been limited domains, and theoretically impossible to reach infinite
generally accepted and widely used by scientific workers. domains. Therefore, it is quite necessary to study analytical
Therefore, Biot’s consolidation theory is directly applied to solutions of Biot’s consolidation within a two-dimensional
model coupled fluid flow and rock (or other porous media) finite domain.
stress within a two-dimensional finite rectangular domain in Barry & Mercer (1999) first presented the analytical solu-
this study. tion of poroelasticity due to a point sink within a two-
A number of researchers have succeeded in obtaining dimensional finite rectangular area, which is applicable for
analytical solutions to the Biot’s consolidation within a two- the first type of boundary conditions of pore pressure field.
dimensional domain. Biot (1941b) presented the analytical For pore pressure field with the second type of boundary
solution of vertical displacement when the upper surface of conditions, Li & Lu (2011) derived the analytical solution
a semi-infinite plane domain was subjected to a uniform in detail. This paper will focus on the simplified analytical
loading with a certain width. Gibson et al. (1970), Booker solution and related validation when pore pressure satisfies
(1974) obtained complex function solutions of the surface closed boundaries and the point sink flux is a constant.
531

Mechanical effect of the presence of gas on faults modeled as a

sandwiched Cam-Clay material
V. Maury
IFP School Rueil-Malmaison and Université de Montpellier, France (formerly with Total)
J.-M. Piau
IFSTTAR, formerly Laboratoire Central des Ponts et Chaussées, France
D. Fitzenz
Universidade de Evora, Centro de Geofisica, Portugal
Subject: Earthquake initiation
Keywords: fluid flow, gas flow, physical modeling, risks and hazards, rock failure, stability analysis
EXTENDED ABSTRACT while the fault core is contractant during interseismic periods.
This dilatancy in undrained regime is prone to pore pressure
The role of fluids on earthquake initiation is now fully rec- fall-off and degassing of the fluid of these surrounding zones
ognized, but the effect of gas is still disputed. In particular, and of the much less thick fault core. In addition, the fault
contrary to what was hoped for in the seventies from Chi- core can be isolated or in relationship with an external source
nese observations, we now know that the observation of gas of over- or under-pressurized fluid, also a possible cause of
leakage around faults before a large earthquake is far from variation of the pore fluid pressure and its gas content. These
systematic and that using gas leakage as earthquake precur- mechanisms make the presence of gas a frequent issue in active
sor would lead to missed earthquakes as well as false alarms. fault settings, as confirmed by numerous observations.
Modeling efforts addressing explicitly both the mechanical Based on known properties of usual rocks and porome-
role of compressible fluids in faulting, but also the role of chanic consideration (Part II), we focus on the fact that the
fluid-assisted interseismic processes, and the changes in the presence of a tiny percentage of gas makes the Kempton’s
hydraulic structure of the fault through time (e.g., Fitzenz and coefficient decrease to zero. Thus porous rocks exhibit quasi
Miller, 2003) are still scarce. Here we want to go further in this drained geomechanical characteristics when saturated with a
direction and investigate in details the effects of the presence fluid including a tiny gas content (few per thousands), even if
of gas in the pore fluid, both on the stress paths and on the they are in undrained conditions (no exchange of fluid with
stability of faults. outside of the rock volume).
The paper builds upon the recent detailed descriptive work During interseismic periods (Part III), assuming the fault
on fault zone structure and deformation mechanisms to inves- core undrained, saturated with liquid and submitted to tec-
tigate several aspects and consequences of gas appearance tonic compressive (and shear) stress components, the pore
and/or disappearance on the geomechanical behaviour of pressure increases limiting the effective normal stress induced,
active faults. Some conclusions are drawn from an analyti- and hence the shear stress.
cal principle model aiming at describing and investigating the For faults in elastic behaviour, the presence of gas induces
behaviour of a fault within two bedrocks submitted to far-field greater normal effective stresses without inducing pore pres-
relative displacements. sure increase leading to greater possible shear stress and
The paper reviews some features of the fault zone struc- storage of elastic energy in surrounding and virgin zones.
ture and deformation mechanisms in Part I. The effects of If the fault is already active (in plastic contractant
gas on fluid compressibilities and geomechanical properties behaviour), the presence of gas in the fault core leads to a
of porous rocks are given Part II. The role of fluid and more more advanced compaction, with decrease of porosity and
particularly gas appearance and/or disappearance on the stress improvement of geomechanical properties. In addition the
paths acting on faults during interseismic period is presented permeability decreases. That again allows the active fault to
Part III. To evaluate the occurrence of instability and the effect sustain larger shear tectonic stress.
of gas, we propose an analytical model assuming a Cam-Clay If a communication between the fault core panel and an
type behaviour for the fault core, briefly described and dis- external over-pressurized fluid source is possible and fast
cussed in Part IV. Conclusions and perspectives are presented compared to the tectonic loading rate, the stress path is still dif-
in Part V. The considerations of this paper would also apply to ferent. The effective normal stress on the fault now decreases,
situations such as landslides, mine roofs, borehole stability. leading to conditions much more prone to rupture triggering
Seismological field observations (Part I) show that in some but at smaller shear stress values. Small overpressures are then
cases, zones surrounding the fault core may exhibit dilatancy sufficient to reach unstable conditions (Maury 1989, 1993).
532

To get a deeper insight into the initiation of fault insta- decrease under brutal dilatant conditions and therefore any
bility (Part IV) and more specifically investigate the role of effective stress increase, able to stop the onset of fault slip.
gas and of an external fluid pressure source, we introduce a It is believed that such conclusions also apply in the realty.
simplified 2D model of a fault, made of two compact elastic Main conclusions appear as follows (Part V):
pads (P), separated by a poro-plastic non-cohesive frictional
– during interseismic periods, in case of link with external
interface (I ).
over-pressurized fluid, the stress path on the fault core is
Taking advantage of the phenomenological considerations
changed, leading to instabilities over small shear stress
from part I, II and III, this interface is modelled by a constitu-
– in case of no link with external fluid sources, the pore
tive law which can feature either contractance and hardening
pressure in the fault core is driven by the Skempton’s effect
(improvement of mechanical interface properties with fault
resulting from tectonic deformation: if the pore fluid is
slip) or dilatance and softening (decrease of fault character-
fully liquid saturated, the Skempton’s effect is important
istics with fault slip). The first case is considered to happen
with a small increase of the normal effective stress, rapidly
for high intensity of the effective normal stress σn , whereas
going closer to the failure criterion. If the pore fluid is
the second one is supposed to happen for relatively low (abso-
now gas invaded, the Skempton’s effect disappears, the
lute) values of σn . The model is derived from the 3D Cam-Clay
normal effective stress increases significantly, insuring a
constitutive law of Soil Mechanics which exhibits similar fea-
temporary stability, but resulting in greater shear stress.
tures in a different context. Thus the fault criterion, expressed
in terms of the shear and effective normal stresses acting on As regards the fault deformation mode and occurrence of
the fault edges, is taken as an ellipse, which size (cf. consoli- instability, as shown by some observations, the model predicts
dation pressure) changes with the normal fault displacement, that, according to the actual stress paths:
considered as a yield variable.
– when the deformation is contractant, this deformation is
Regarding the pore pressure evolution within the fault
progressive and stable,
core, it is assumed to result from two processes : one is the
– when the deformation is dilatant, the deformation can be
Skempton’s effect linked to the contractant/dilatant behaviour
either progressive (stable) or unstable (sudden); the limits
of the fault; the other one is the possible presence and
of the domain of instability can be calculated,
communication with an external over- or under- pressurized
– the fluid compressibility increase (gas presence) enlarges
fluid reservoir. Terzaghi’s law is used to relate the total and
the domain of instability.
effective normal stress. The whole model (P + I ) is consid-
ered driven by far field displacement conditions imposed to Especially the model shows that for stress, pore pressure and
the pads. geomechanical properties being constant, a fault in condition
The discussion is focused on the incremental response of the of dilatant progressive deformation can switch to instability
fault, which can be either inactive, active and contractant or (rupture) when pore fluids are invaded by gas.
active and dilatant, depending on the resulting state of effective As regards the use of gas observations at the surface as
stress acting on the fault. a precursor to earthquake, the problem must be considered
In particular the model puts in evidence the existence under accounting for the local structural geology and gas reservoir
some conditions of an unstable domain, linked to the dila- engineering aspects. Notably case histories about gas leakage
tant side of the fault criterion, for which the fault exhibits above and around gas reservoirs show that gas may migrate
some sudden shear slip with the generation of seismic waves, underground over very large distances, redissolve, disappear
before reaching a new stable position. Among the conditions or be stacked in unexpected geological reservoirs. Therefore
for such a possibility, the fluid compressibility is found to play in the same way the observation of gas at the surface around
a major role. As shown by simulations, a fluid with a high com- an active fault may not be systematical before an earthquake.
pressibility and a low Skempton’s coefficient (cf. presence of Nevertheless when gas appears at the surface, it is surely an
gas) is more prone to fault instability than the opposite case indicator that it may have induced detrimental effects on the
(liquid phase only). This can be interpreted through the fact stress paths leading to large shear stresses and that it can also
that a zero Skempton’s coefficient impedes any pore pressure play a direct role on the earthquake initiation.
533

Impact of fluids and formation anisotropy on acoustic, deformation and failure

characteristics of reservoir shales and pure clay minerals
A.N. Tutuncu
Colorado School of Mines, Golden, CO, US
A.I. Mese
Geomechanics Engineering & Research, Houston, TX, US
An experimental investigation have been conducted on the

relationship between swelling, swelling pressure, compres-
sional and shear wave velocities, plastic behavior and defor-
mation characteristics of Barnett, Haynesville, Woodford gas
shales, Pierre and North Sea shales and reconstituted ben-
tonite. Swelling characteristics, surface areas, cation exchange
capacity and tensile strength for several pure clays, seal shales,
gas and oil shaleshave been measured in addition to the stress-
strain, failure properties and compressional and shear wave
velocities to understand the storage capacity and rock-fluid
interaction impact on the deformation and failure charac-
teristics of the unconventional resources. Directional wave
velocities for several shale samples were also collected at
varying in situ stress conditions to obtain intrinsic and stress
induced anisotropy. The clay fraction and in particular the Figure 1. Tensile strengths as a function of residual strain in
reconstituted bentonite and native shale samples of this study.
type and amount of the respective clay minerals determines
the chemical and deformational properties of the rock.
Clay minerals are very reactive because of their large sur-
face area and their charged surfaces. The existence of the
charge is the basis for the exchange capacity and swelling that
provides key understanding of these properties for petroleum
engineering applications. Adhesion and friction model have
been used to incorporate attractive and repulsive interactions
taking place between clay platelets. Initial and equilibrium
separation distances were calculated for different fluids at sev-
eral stress conditions. Relations obtained between clay platelet
thickness, swelling and fluid composition changes calculated
from initial and equilibrium separation distances agree well
with experimental swelling and swelling pressure measure-
ment results for the shales and pure clay minerals studied in
this study. The relationship between compressional and shear
velocities, plastic behavior, stress state and various ionic con- Figure 2. Tensile strengths as a function of swelling pressure in
centrations has been investigated using triaxial compression reconstituted bentonite and native shale samples of this study.
experiments.
Well integrity studies benefit utilization of surface area,
swelling pressure, and cation exchange capacity measure- as evidence that the surface area and cation exchange capacity
ments in addition to classical deformation and wave veloc- dominates the swelling of clays and shales and impacts sig-
ity measurements to characterize the shale-fluid interaction nificantly the tensile strength and plastic deformation of shale
impact on these properties. While plasticity, brittleness and formations. The fluids used in unconventional resource frac-
ductility of shale are critical, smallest measured expansion turing operations particularly slick water operations should
does not always implement full stability and control of the take into consideration the impact of the chemicals used in
shales (Figure 1 and 2). Thus, measurement of expandability fracturing fluid not only from environmentally safe operations
should not be used alone to predict shale stability. for not contaminating the groundwater, but also to reduce the
An excellent agreement obtained between experimental and instability problems in the wells and to ease the breaking of
theoretical results for all the comparisons made is interpreted the formation.
534

Change of joint aperture during formation of ice-ring around

underground LNG storage cavern
Eui-Seob Park, So-Keul Chung & Joong-Ho Synn

Geologic Environment Division, Korea Institute of Geoscience and Mineral Resources, Daejeon, Korea
Woo-Cheol Jeong & Taek-Kon Kim

Geotask Team, SK Engineering & Construction, Seoul, Korea
Seong-Ho Bae
Geogeny Consultants, Seoul, Korea
Keywords: fluid flow, ground freezing, rock joints, cryosuction
In general, either above-ground tanks or in-ground tanks been The phenomena involved in phase change during formation
used worldwide to store LNG. These types of tanks have a of an ice-ring can be divided into the followings: cryo-suction
number of problems such as the need for a large site for storage and swelling. Cryo-suction refers to suction pressure occurred
tanks, ground settlement, high operating costs, an excessive during phase change in the coexistence zone of ice and
boil-off ratio due to radiant heat from the sun, and so on. Many water. The zone where water and ice coexist is from 0◦ C
attempts have therefore been made in the past to store LNG isotherms up to the penetration zone of groundwater inside
underground in unlined containment, though without success. of the 0◦ C isotherms. In previous studies of Jeong et al.
The new system for storing LNG in rock caverns had been (2007), the temperature of rock mass frozen completely by
developed and verified through the design, construction and the groundwater infiltration was identified as groundwater
operation of a LNG pilot plant in Daejeon, Korea. Among freezing temperature. According to Lee et al. (2004), it was
the key technologies of the LNG storage system, drainage of known that the groundwater freezing temperature is in the
groundwater and formation of an ice-ring are very important range of −3◦ C to −2.5◦ C. However, it can be varied depending
ones. These are strongly related to the complex mechanism on the recovery rate of groundwater and average aperture of
between thermal and hydraulic characteristics of rock mass joints.
and groundwater. Another phenomenon that happens during freezing of water
Chung et al. (2005) performed a numerical calculation is swelling effect, which increases 9% in volume after phase
for the thermo-hydraulic modeling of the ice ring using the change from water to ice. The phase change to ice in a very
FLOW3D code. The coupled analysis reveals that an ice-ring tightly closed state can generate enormous swelling pressure
can be easily formed by controlling the groundwater drainage due to volume expansion. However, it is expected that the ice
system; this is based on the assumption that the average dis- in rock joints swell into a cavern at an atmospheric pressure
tance of the 0◦ C isotherm reaches 3 or 4 m from the cavern rather than opens joints during the phase change.
wall in the rock mass with the hydraulic conductivity of 10−7 The effect of phase change of groundwater on rock mass
to 10−6 m/s. Through the previous researches, the core of ice- near LNG storage caverns can be summed up the two fol-
ring design is summarized as follows: propagation distance of lowings. First, the residual water near the cavern is frozen,
0◦ C isotherms after the injection of LNG; groundwater pen- and frost heave pressure is generated after injection of LNG.
etration distance within 0◦ C isotherm; and change of joint Second, when an ice-ring is formed due to the recovery
aperture during/after ice-ring formation. of groundwater after the stop of drainage, cryo-suction is
It was confirmed that the propagation distance of 0◦ C occurred in rock joints.
isotherms after the injection of LNG affects the operation In this paper, thermo-hydraulic-mechanical coupled anal-
period of the drainage system, and can be predicted from the ysis with UDEC code is performed to find out the effect
thermal properties of rock mass, and initial temperatures of of cryo-suction on joint apertures. The following two mod-
LNG and surrounding rock mass. Also the following facts els were used for the T-H-M analyses: simple joint and pilot
were verified that in the sense that the stability of contain- plant models. In order to make cryo-suction effect clearer, it
ment system and rock mass around caverns can be affected by is assumed that phase change occurs in a temperature range
frost heave pressure, the penetration distance of groundwater of −20 to 0◦ C. In addition, the permeability of joints decrease
during/after ice-ring formation is an essential checklist for the linearly to zero from 0◦ C to −2.5◦ C.
design of the drainage system, and is mainly affected by the From the numerical results of a simple joint model, it was
recovery rate of groundwater and aperture of rock joints after verified that the phase change occurred and resulted in cryo-
the drainage work. suction in a temperature range of −20 to 0◦ C. Furthermore,
535

Figure 2. Analysis results for the pilot plant model (Joint displace-
Figure 1. Hysteresis of joint separation with temperature on the ment vs. time).
excavation surface during cooling and thawing stages.
the aperture of the joint at the center of the model was reduced As temperature dropped, the joint aperture was gradually
by about 0.4 mm. increased by shrinkage of rock mass as like the results of mon-
During the operation of the LNG pilot plant, joint meters itoring. However, the aperture was reduced in an instant after
were installed and measured the change of aperture of joints at the temperature falls below 0◦ C. The aperture was decreased
the cavern. As a result, the suction phenomenon was observed to 0.01∼0.035 mm in a temperature range of −2.5 to 0◦ C. It
by the joint meter at the top of the cavern around 0◦ C. As can be thought that this result comes from the suction in the
shown in Figure 1, joint apertures were opened in a linear joint.
manner during cooling, although their magnitudes were small, The change of aperture by numerical analysis was smaller
and they were closed as temperatures were raised during the than that by direct measurement from LNG pilot plant. This
thawing stage. It means that groundwater flowed into a joint difference is caused by the following reasons: There are
after the stop of drainage, and the phase change of groundwater different measurement points between LNG pilot plant and
caused the cryo-suction in joints. It is important for the joint numerical analysis. And operating LNG pilot plant at the
opening to show the elastic hysteresis even at the temperature time, the groundwater penetrated deep into the cavern during
cycling, as this implies that the rock around the cavern is elastic ice-ring formation.
even at the low temperature of −30◦ C. In summary, the effect of cryo-suction on aperture of rock
Numerical analysis for LNG pilot plant was performed by joints was verified by the analysis results of the single joint and
the following four stages: First, Excavation of a cavern, sec- pilot plant model. The generalized Clapeyon equation defined
ond, start of drainage for drawdown of groundwater, third, in porous media was used to quantify the phenomena of cryo-
Build-up of containment system and start of LNG supply, suction during ice-ring formation. Since then, it was applied
and finally, Stop of drainage and formation of ice-ring. The to discrete element modeling code such as UDEC and simu-
change of joint aperture was measured at the top, bottom, lated to find out the change of joint aperture during ice-ring
and wall of storage caverns shown in Figure 2. It can be seen formation.
that the aperture of the joint was increased slightly and then Through these results, it was revealed that the cryo-suction
decreased again after the stop of drainage (June 10th). The occurs as groundwater is recovered and changed to ice for
result is caused by the following sequential processes: the forming an ice-ring, and can affect the joint aperture instantly.
increase of pore pressure due to penetration of groundwater Furthermore, the joint apertures around the cavern were
and cryo-suction in the joints. However, it can be expected decreased up to 0.01∼0.035 mm, and the change of the joint
that the increment of pore pressure and cryo-suction occur at aperture had little effect on the stability of the cavern.
the same time in the LNG underground storage cavern.
536

A parametric study to investigate the effect of injection and production well operation on
the CO2 sequestration process in deep unmineable coal seams
M.S.A. Perera & P.G. Ranjith

Department of Civil Engineering, Monash University, Clayton, Australia
S.K. Choi
CSIRO Earth Science and Resource Engineering, Clayton, Australia
Keywords: numerical modeling, CO2 injection, deep coal seam sequestration
Injection of CO2 into a coal seam can cause changes to both the The model results show that when injection is introduced
chemical and physical structures of the coal. It has been shown from a single injection well at the bottom mid-point of the coal
that CO2 adsorption can cause swelling of the coal (Ates and seam, there is a higher rate of CO2 spread at the beginning of
Barron, 1988) and this may lead to reductions in permeabil- injection and the rate decreases with time. This is due to the
ity which will cause difficulties in injection of the CO2 into fact that the CO2 injection rate greatly depends on the pres-
the coal. This swelling phenomenon introduces complexity to sure gradient at the injection well and the gradient decreases
the design of injection systems for CO2 sequestration in deep with time as the CO2 is sequestered.The CO2 injection rate can
coal seams. Drilling an injection well into a deep coal seam be significantly increased by increasing the injection pressure.
(at depths on the order of 1000 m) is an expensive and time This may be related to expansion of coal cleats with increasing
consuming process. Therefore a good understanding of the injection pressure and/or due to increased permeability with
influence of design parameters on the effectiveness of CO2 the formation of new fractures. When more than one injection
sequestration in deep coal seams prior to drilling is impera- well are introduced into the coal seam, CO2 injection capacity
tive. Based on existing knowledge and current understanding is influenced by the pressure distribution around the injection
of the major processes, numerical modeling, which takes into wells. For the modeling scenario used in this work, the max-
account the interaction of the various processes, can be used imum injection capacity was observed for the two-well case
to conduct parametric studies in order to determine the oper- and a reduction in CO2 storage capacity was observed when
ational parameters to achieve optimal response from the coal the number of injection wells was increased. The CO2 storage
seam in the sequestration process. In this study the COMET3 capacity can be greatly increased by incorporation of a pro-
numerical simulator (Sawyer et al., 1990) has been used to duction well. Where a production well is included the optimal
develop a model to study the flow behavior inside a coal seam distance between the two wells will be influenced by a trade-
following CO2 injection. off between the effect that fluid migration toward a proximal
The main objective of this study was to conduct a paramet- production well from the injection well has on the storage
ric study using a 3D numerical model to optimize the CO2 capacity of the coal seam in the vicinity of the injection well,
sequestration process in a 540 m × 500 m × 20 m coal seam, and the time required to achieve pore pressure reduction in
located at 1000 m below the ground surface. The effect of the vicinity of the injection well when the production well
three model parameters on the CO2 storage capacity of the is farther away. From the model results, it is clear that just
coal seam was considered. These were injection pressure, the increasing the injection pressure or the number of injection
number and location of the injection wells and the location of wells will not necessarily increase the CO2 storage capacity
the production well. Firstly, one injection well with 0.1 m inter- of a deep coal seam and that the location of injection and
nal diameter was introduced to the bottom of the coal seam production wells can heavily influence CO2 storage capacity.
from the midpoint and CO2 was injected at 10 MPa pressure A numerical model can be used to assess the effects of oper-
for ten years. Then the injecting pressure was changed from ational parameters such as injection pressure and the number
10 MPa to 20 MPa and corresponding possible CO2 storage of wells and their location on the performance of a deep coal
capacities for ten years of injection were determined. Next, seam CO2 sequestration system.
the number of injection wells was increased to four and cor-
responding storage capacities were observed. Following this,
an injection pressure of 20 MPa was used for the two injec- REFERENCES
tion wells scenario and the distance between the injection
wells was changed to investigate the response in CO2 stor- Ates, Y., and Barron, K. 1988. Effect of gas sorption on the strength
age capacity. Finally, a production well was added and the of coal. Mining science & technology, 6(3): 291–300.
Sawyer, W. K., Paul, G. W., and Schraufnagel, R. A. 1990. Develop-
corresponding storage capacity was observed for a condition ment and application of a 3D coalbed simulator. Paper CIM/SPE,
whereby the water was pumped from the production well at Petroleum Society CIM, Calgary: 90–119.
a rate of 15 m3 /day. The distance between the injection and
production wells was changed and the corresponding storage
capacity variation was observed.
537

Development of the confined pressure type of multi-purpose rock flow system and its
application to the evaluation of permeability of various geomaterials
A. Sato, H. Yoshitake, T. Shiote & K. Sasa

Graduate School of Science and Technology, Kumamoto University, Japan
Keywords: fluid flow, lab testing, monitoring, rock properties
1 INTRODUCTION
Authors developed a new apparatus, confined pressure type of

multi-purpose rock flow system, and its applicability is veri-
fied. This system is based on the flow pump type of permeation
system, and permeation tests can be easily conducted under
the high confining pressure condition and peculiar conditions
like as super critical CO2 . Moreover, X-ray CT tomography-
ing during permeation tests is also possible. Here permeation
tests are conducted towards several kinds of geomaterials such
as sedimentary rocks and bentonite, and the applicability, Figure 1. Confined pressure type of multi-purpose rock flow
accuracy, and the usefulness of the system are verified. system.
2 CONFINED PRESSURE TYPE OF MULTI-PURPOSE

ROCK FLOW SYSTEM
Newly developed confined pressure type of multi-purpose

rock flow system is shown is Figure 1. This system consists of
two syringe pumps and a pressure vessel. The maximum pres-
sure of the pumps is 25 MPa. The characteristic point of the
system is that the pressure vessel is made of carbon fiber rein-
forced plastic (CFRP) tube and the flanges made of aluminum
are fixed by four shafts at both ends. The density of CFRP is
lower than metals and it is suitable as a material of the pres-
sure vessel for X-ray CT tomographying (please refer the full
paper). Moreover, the tensile strength of the material is very
high (approximately 700 ∼ 1,000 MPa), and 5 mm thickness
of the tube is strong enough to hold the 20 MPa pressure.
Figure 2. Relation between the intrinsic permeability of Kimachi
3 APPLICATION TO BENTONITE sandstone and effective confining pressure.
The permeability measurements of bentonite during the with time, please refer the full paper. The X-ray CT image dur-
swelling process are conducted. The bentonite used in this ing swelling process is also shown in Figure 2. These images
study consists of large and small particles. Using the devel- are taken 10 days after commencement of swelling and taken
oped flow system, the intrinsic permeability k are evaluated. in the installed condition in the developed pressure vessels
The value of k is converged around the order of 10−19 m2 during permeation tests. Some original shape of bentonite
after 3 days. This value is almost same as the low perme- particles is observed, it is found that the bentonite has been
ability Granite and bentonite shows enough performance as a almost homogenized at the region around inlet surface, and it
water barrier. As for the relation between the evaluated values is thought that this region works as water barrier.
538

Near-field application of aperture back calibrated distinct fracture network
M. Sharifzadeh & M. Javadi

Faculty of Mining & Metallurgical Engineering, Amirkabir University of Technology, Tehran, Iran
Keywords: fluid flow, hydraulic fractures, back analysis, numerical modelling, field measurements
Ground water inflow causes several difficulties such as

unsafely, impairing the project schedule, equipment damaging
and altering the groundwater regime as well in the construction
phase of any underground constructions as in the operation
phase. To evaluate the related problems, the possibility and
the probable water inflow into excavation must be some-
how predicted in advance. In many geological structures, the
matrix permeability is negligible compared to permeability
of fractures and rock mass hydraulic behavior is controlled
by fractures. In such highly heterogeneous fractured media,
fluid flow takes place along preferential pathways within the
fractures. In such situations, the discrete fracture network
(DFN) concept may appear much more adapted for fluid
flow simulation. Although DFN provides a powerful means Figure 1. Variation of error value as a function of hydraulic aperture.
of investigating rock mass hydraulic behavior, field applica-
tion of DFN models is restricted by input data, especially for
hydraulic aperture of fractures. This paper discusses the appli- Consequently, 450 and 500 DFN realizations were generated
cability of realizations of aperture back calibrated discrete around the caverns for inflow simulation in S1 and S2 sections,
fracture network (ABCDFN) for fluid flow analysis through- respectively.
out near-field scale domain in rock mass. Due to the negligible Following geometrical analysis of fracture network around
matrix permeability of rock mass surrounding Siahbisheh cav- caverns in section S1, three different states of percolation are
erns, DFN model was selected to analysis of water inflow. observed. In some DFN realizations both PHC and TC are
Detailed measurements of the inflow into Siah Bisheh power- percolated simultaneously. In some realizations neither is per-
house cavern (PHC) and transformer cavern (TC) provided an colated and in some other realizations only one of the two is
opportunity to determine the hydraulic aperture through back distinguished as percolated cavern. 27 out of 448 total DFN
calibration in the discrete fracture network model. In order to realizations are not percolated at all while the remained 421
reach this goal, a new computational code, FNETF, has been models (94% of the total) are partially percolated. Among the
developed for generating DFN and laminar fluid flow analysis. percolated realizations, in 56 models (12.5% of the total), only
The main input data for discrete fracture fluid flow modeling PHC is percolated, in 90 realizations (20% of the total) only
are classified in two different classes: (1) the geometric data TC is percolated and in 275 realizations (61.4% of the total)
of the fractures such as orientation, trace length, and density; both are percolated. All in all, 331 realizations (74% of the
and (2) the hydrology data that are used for boundary condi- total) have shown percolation for PHC and 365 models (82%
tions assessment, which are captured from site investigations of the total) have shown percolation for TC.
and measurements. The inflow simulation in fractured media Direct back calibration for PHC and TC was used on the
around both PHC and TC has been studied within two dif- basis of water inflow rate values in order to find the optimum
ferent practice cases through two different types of vertical value of the effective hydraulic aperture. In this method, the
section i.e. S1 and S2. The type S1 crosses over both PHC optimum value of the effective hydraulic aperture has been
and TC, hence inflow could be simulated around both caverns determined by trial and error method. The error value ε(p)
simultaneously. The results of simulation in this section were has been calculated for three different states. For the first and
utilized for calibration. On the other hand, the type S2 only second states only results of water inflow into PHC and TC
crosses TC and as a result, it could only be used for inflow have been used respectively. In the third state, all the water
simulation around TC. Results of the latter were utilized for inflow results to both caverns have been used to calculate the
assessment of calibration. With regard to the statistical nature error value. Variations of error value as a function of different
of fractures geometry, large number of DFN realizations shall values of mean hydraulic aperture are shown in Figure 1, for
be generated to ensure that the results are not dependent on every three aforementioned states. For water inflow into PHC
specific fracture geometry arrangement and to produce more and TC, the error value has reached its minimum in hydraulic
representative of stochastic behaviour of the fractured media. mean apertures of 0.46, 0.625 and 0.452 millimeters for three
539

Field data of water inflow into the end part of TC in the
time period between December 2007 and December 2008
accompanied by simulation of fracture network in S2 have
resulted in validation of ABCDFN model. For this purpose,
simulation of inflow in fracture network around TC and cal-
culation of water inflow into the end part of TC have been
implemented through consideration of equivalent hydraulic
aperture value of 0.452 mm. The field data of water inflow
into TC in sections of S2 type have shown percolation in the
distances between chainages 0+00 to 0.010 (RI), 0+010 to
0+020 (RII) and 0+020 to 0+030 (RIII). A comparison of
modeling results obtained from back calibrated flow simula-
tion, and the real quantity of water inflow into TC during the
time interval from December 2007 to December 2008 has been
illustrated in Figure 2. Comparison of results indicates that,
there is an appropriate correspondence between flow simula-
tion through ABCDFN models and the real quantity of water
inflow into TC. Therefore, ABCDFN model can be utilized
for fluid flow analysis in the near-filed domains and shows
appropriate results of rock mass hydraulic modeling.
Figure 2. Comparison between ABCDFN prediction and measured

inflow into TC for different sections: (a) RI (b) RII and (c) RIII.
states respectively. In addition, the error graphs for three dif-

ferent states cross each other in the hydraulic mean value of
0.49 mm. The equivalent hydraulic aperture is considered to
be 0.452 mm according to the third state.
540

A combined Bayesian-Wavelet-Data fusion workflow for breakout zone

identification in oil and gas wells
H. Soroush
Well Engineering Centre of Excellence (WECOE), Weatherford Oil Tool M.E., Dubai, UAE
V. Rasouli
Department of Petroleum Engineering, Curtin University of Technology, Perth, Australia
Keywords: oil reservoir, rock failure, rock properties, stability analysis
1 INTRODUCTION the identification of BO from nBO can be achieved with higher

accuracy if de-noised logs are used. Bayesian classifier was
The integrity of the wellbore plays an important role in then applied to the logs to classify each depth to either BO or
petroleum operations including drilling, completion, and pro- nBO classes. 70% of the data were used for training the clas-
duction. To complete a successful wellbore stability analysis, sifier and the remaining 30% were used for testing. Finally,
building a reliable geomechanical model is a basic require- data fusion techniques (MV and OWA) were utilized to fuse
ment. Breakouts provide valuable information with respect the data from different analysis and increase the accuracy of
to evaluation of maximum horizontal stress magnitude and the model accordingly.
also calibration of the model. In addition, they provide final
borehole shape which is a critical factor in completion and
production optimization. This paper presents a novel workflow 4 RESULTS
for identification of breakout zones in a wellbore from com-
mon petrophysical logs using some data processing techniques The results on the de-noised data showed a total accuracy of
such as wavelet decomposition, Bayesian classification, and 77 to 89% which is a significant achievement. To examine
data fusion. the model generalization, each study well was considered as
a target well and BO-nBO zones were identified using data
from the other wells. This resulted in four different classifi-
2 LIMITATION OF ROUTINE METHODS cations for each target well in each depth. The results showed
an accuracy of 74 and 82%. These accuracies increased to 82
While caliper data are most often used in regional and field to 98% after data fusion using OWA.
studies, acoustic imaging tools are considered the best devices
for identifying breakouts. Unfortunately, acoustic image logs
are not usually available, especially in older fields. Caliper and 5 CONCLUSIONS
electrical image logs are, however, associated with several lim-
itations including not having a full coverage of the borehole A new methodology was presented to identify wellbore
wall, being influenced by different parameters, pad width lim- breakout zones. The approach was applied to the data from
itation, low resolution, complicated processing procedure, and carbonate and its applicability and generalization capabil-
unavailability in old wells. These limitations required estab- ity was confirmed through a sequence of single-well and
lishing a new approach by processing the commonly available multi-well analyses. The capability of OWA data fusion tech-
logs. Our investigation suggests the implication of a multi- nique in improving the accuracy was demonstrated. The major
variable approach to include the effect of several parameters. conclusions of this study are:
In this study, analysis was done in a 10-D space using CGR, • Providing that good quality image data are available at
SGR, Dt, ILD, NPHI, RHOB, PEF and CALI logs in addition least in one well in a formation, the proposed approach
to mud weight and vertical stress, in a carbonate field. enables identification of breakouts in other wells in the same
formation with a reasonable accuracy.
• Availability of more wells with good quality data increases
3 METHODOLOGY
the accuracy of the method if OWA data fusion technique
is used.
In this study, 1-D wavelet approach was used in order to ana-
lyze petrophysical logs and energy matching strategy was It is noted that the applicability of the workflow was
used to select the OMW. The OMW was then applied to also approved shale through a similar study, however, its
remove noises from the logs. Our investigation revealed that performance in other type of rocks yet to be studied.
541

Hydraulic stress effect on 3 dimensional pore network and permeability change
M. Takahashi
National Institute of Advanced Industrial Science and Technology, Tsukuba, Japan
C. Ahn
Saitama University, Urawa, Japan
M. Kato
Hokkaido University, Sapporo, Japan
Keywords: case studies, fluid flow, lab testing, oil reservoir, rock mass, rock properties
In this paper, we use the 3DMA method to quantify the confining-pressure region, permeability of Berea sandstone
flow-relevant geometrical properties of the pore structure in decreased with increasing effective confining pressure. The
Berea sandstone, We measure distributions of pore size, throat rate of change of the permeability weakens gradually with
size, channel length, coordination number, and correlations increasing effective pressure. In horizontal direction, X and Y
between pore and throat geometry of Berea sandstone under axis, the tortuosity under the stressed condition were greater
zero and pressurized conditions for vertical and horizontal than those under zero pressure. In vertical direction, Z axis,
to the bedding plane. In addition, we present the quantita- tortuosity distributions under the stressed condition were com-
tive characterization of tortuosity distribution for intact Berea plex, but its variation distributed within narrow region. These
sandstone and Berea sandstone pressurized to 25 MPa. We phenomena are attributed to the decrease of the number of
discuss relation between permeability reduction by pressuriza- connecting paths or the increase of tortuosity for individual
tion and pore connecting or tortuosity information for different directions.
3 mutually perpendicular directions. In the low effective
Table 1. Connecting information and tortuosity distribution for X, Y and Z directions under zero and 25 MPa pressures.
0 MPa 25 MPa
X start face 226 end face 324 start face 279 end face 371
35816 connecting paths 51072 connecting paths
tort – min 1.50954 med 1.79514 max 2.34023 tort – min 1.54513 med 1.79524 max 2.57653
Y start face 337 end face 358 start face 278 end face 358
Z start face 422 end face 373 start face 357 end face 330
Figure 1. Tortuosity distribution in X, Y and Z axial directions for Berea sandstone.
542

Algorithm of coupled seepage and normal stress processes in fractured rock mass by
composite element method
L.L. Xue
School of Civil Engineering and Transportation, South China University of Technology, Guangzhou, China
S.H. Chen
State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University, Wuhan, China
I. Shahrour
Laboratoire de Mechanique de Lille (UMR 8107), Université des Sciences et Technologies de Lille, France
Keywords: fluid flow, rock stress, hydraulic fractures, rock mass
ABSTRACT: There is an important and complicated cou- equivalent continuum) approach (Long, etc.,1982; Hsieh &
pling between seepage and stress-strain fields in fractured Neuman, 1985; Oda, 1985, 1986), which takes into account the
rock mass: the stress-strain field is influenced by the seepage; deformation and permeability characteristics of fractures but
however, the stress-strain field affects the permeability char- neglects their exact positions, therefore, leading to some calcu-
acteristics and in turn the seepage field as well. In the seepage lation errors. Another one is the explicit (or discrete) approach
and stress-strain coupling system, fractures play a fundamen- (Schwartz, etc. 1983; Dershowitz & Einstein, 1987; Cacas,
tal role because the seepage and deformation of fractured rock etc. 1990; Andersson & Dverstorp, 1987), which considers the
mass are mainly dominated by them: stresses acting on the geological and mechanical properties of each fracture deter-
fracture surface lead to the change of fracture aperture, which ministically. Usually there are a large multitude of fractures
causes the change of conductivity matrix of the fractured rock with small aperture. Therefore, the explicit approach often
mass and the seepage field. According to this coupling mecha- makes the pre-process work time costly and tedious, even
nism, seepage and stress-strain coupling analysis of fractured makes the generated mesh with bad topological configuration,
rock mass can be achieved by applying the iterative algorithm. which brings some calculation errors or makes calculation
It is also widely recognized that in the geotechnical engineer- nonconvergent.
ing, shear influence on fracture is less important than tension Composite element method (CEM) has been proposed by
or compression influence (Zhang, 2005; Raven & Gale, 1985; Chen (2003). One composite element is defined to cover sub-
Tsang & Witherspoon, 1981). domains with different shapes and characteristics, and these
Seepage and normal stress coupling in fractured rock mass sub-domains are named as sub-elements. Unknown variables
has been studied from various aspects including experimental within each sub-element (e.g. φ, {u}) can be interpolated
technology, mechanic parameters, numerical models, and so from the mapped nodal variables defined on the composite
on. Snow (1968) built the relationship between permeability elements (e.g. {φ}, {δ}), and the shape function is the same
coefficient and fracture aperture for horizontal fracture. Louis as that of conventional FEM defined over the whole com-
(1974) obtained a famous semi-empirical formula between posite element. The governing equation for the solution to the
permeability coefficient and normal stress through the packer mapped nodal variables can be established by applying Virtual
test in borehole with different depth in the homogeneous frac- Work Principle or Variational Principle. Based on the solved
tured rock mass. Gale (1982) proposed experiential formula variables on the mapped nodes on the composite elements,
between permeability coefficient and normal stress through the unknown variables within the relevant sub-element can be
the laboratory experiments on three kinds of rock mass. Bar- interpolated through the shape function. The proposed CEM
ton (1985) proposed joint roughness coefficient (JRC), and can discrete the fractured rock region regarded as one homo-
founded one empirical formula for seepage and stress-strain geneous medium, then the fractures can be embedded within
coupling. Malama and Kulatilake (2003) proposed a general the elements but be simulated explicitly. In this way the com-
exponential model by defining half-closure stress σ1/2 , in putation mesh generation based on the CEM is more simple
which fracture deformation is a function of the maximum frac- and convenient, especially for the complicated fractured rock
ture closure and σ1/2 . Tsang and Witherspoon(1981), Chen problems.
(1989) proposed physical models to simulate the coupling Since flow rate is proportional to the third power of frac-
behavior between seepage and normal stress-strain. ture aperture, and most deformation of fractured rock masses
The numerical methods used in simulation of rock frac- is generated by fracture deformation, therefore, fracture defor-
tures contain two categories. The first one is the implicit (or mation governs the flow capability of fractured rock masses.
543

Because the behavior of fracture is more dependent on the nor- 2. The proposed algorithm has the advantage of simplicity
mal stress than the shear stress, only the seepage and normal in the preprocess, because the existence of the fractures
stress coupling is taken into consideration in this paper. needn’t be considered in the computation mesh genera-
Rock fractures are classified into filled fractures and non- tion, but it will be considered explicitly in each composite
filled fractures. Non-filled fracture can be represented as two element.
parallel plates in contact with convex parts, and the asperities 3. The proposed algorithm considers not only the flow in the
are considered as a layer of granular material which has high fractures, but also the flow exchanges between fractures
porosity and is clipped by the two parallel plates. Accordingly, and the adjacent rock masses, and it also can be used for
Chen (1989) has proposed “filled model”, which considered both the filled and non-filled fractures.
the asperities as an evenly “filled” medium with deformation 4. The uneven hydraulic behavior of the fractured rock masses
and permeability characteristics. In this way, a uniform model resulting from the normal stress is remarkable when con-
for both the filled and non-filled fractures can be established. sidering the seepage and stress coupling. Therefore, the
This paper presents a composite element algorithm of the importance of the coupling analysis for fractured rock
seepage and normal stress coupling in fractured rock mass, masses should be emphasized.
which is an important extension to the CEM. The coupling
relation between the fracture seepage and stress makes use of
“filled model”, and the coupling algorithm has been achieved
by applying the iterative algorithm. The verification of the
proposed algorithm has been carried out, and conclusions are
demonstrated as follows:
1. The results obtained from the proposed algorithm and the
conventional finite element method (FEM) exactly agreed
with each other, through which the reliability and the
advantage of the proposed CEM coupling algorithm were
well verified.
544

Evolution of fracture permeability in granite under temperature, confining

pressure and pH conditions prescribed
H. Yasuhara & N. Kinoshita

Department of Civil and Environmental Engineering, Ehime University, Matsuyama, Japan
S. Nakashima
Department of Civil and Environmental Engineering, Yamaguchi University, Ube, Japan
K. Kishida
Department of Urban Management, Kyoto University, Kyoto, Japan
Keywords: fluid flow, lab testing, nuclear repository, rock joints
EXTENDED ABSTRACT
A safe, long-term isolation of high level radioactive wastes

(HLW) apart from our human life has to be secured. Under
high temperature, induced by heat from the wastes, and stress
conditions, geochemical reactions such as mineral dissolution/
precipitation may be rather active. Those geochemical reac-
tions may change the flow and transport properties of fractured
rock masses that are natural barriers on HLW repositories.
Therefore, the evolution of the properties should be examined
under such high stress and temperature conditions.
In this work, we present the results of flow-through exper-
iments in a single induced fracture in granite at temperatures
of 25 and 90◦ C, at confining pressures of 5 and 10 MPa, and at
pH’s of 7 and 11 and under fluid differential pressures of 0.01–
0.5 MPa. The experiments follow the progress of the fracture
aperture or related fracture permeability using the continuous
measurements of the evolving flow rates under the prescribed
fluid differential pressures, and of element concentrations
periodically measured by ICP-AES.
As shown in Figure 1, generally, the fracture permeability
decreases with time and reaches a quasi-steady state within a
few hundred hours. However, after the temperature is raised to
90◦ C, the permeability again resumed decreasing throughout
the rest of the experimental periods. The higher tempera-
ture, confining pressure, and pH are prescribed, the faster
evolutions of the fracture permeability are observed.
The effects of the prescribed confining pressure, tempera-
ture, and pH on the concentrations measured can be examined
by standardizing the measured concentrations with the resi-
dence time – the standardized concentrations of Si for E-1, Figure 1. Evolution in fracture permeability and element concen-
E-3, and E-6 are shown in Figure 1. For E-6, which is tration for E-1, E-3, E-6.
conducted under the alkaline condition of pH 11, the Si con-
centrations are higher than those of E-1 and E-6, which are and pH’s being prescribed and less on stresses exerted over
done under the neutral conditions. This is clearly due to the contacting asperities.
higher pH prescribed for E-6. Specifically, at 90◦ C the con- The observed reduction in fracture aperture or permeability
centration difference between E-6 and the others are roughly can most likely be attributed to the mechanical crushing of the
one order of magnitude. From the concentration analyses, propping asperities and/or the chemical dissolution prevailing
the mineral dissolution is dependent strongly on temperatures over the contacting asperities and free walls within a fracture.
545

Physical clogging model using stick rate of calcite on CO2 mineral trapping
S.Y. Yoo, Y. Mito & T. Matsuoka

Department of Urban Management, Graduate School of Engineering, Kyoto University, Japan
Y. Kuroda
International Petroleum Exploration Co., Japan
A. Ueda
Graduate School Division of Science and Engineering, University of Toyama, Japan
Keywords: fluid flow, field measurements, lab testing, rock properties, geothermal, numerical modeling
Flow experiments through porous and fractured media were upward. According to results, an order of magnitude differ-
performed to examine the clogging phenomena by carbonate ence of stick rate between the column test and the in-situ
mineralization at high temperature. Flow rates of the column test is distributed to 2 orders at the similar temperature. This
test decreased gradually due to calcite precipitation. The clog- discrepancy might create different transportation that deposit
ging model with a stick rate of calcite explains the observed particles, once migrated and settled in pores, remain under
permeability changes on the in-situ test. On the other hand, constant head conditions of the column test, whereas under
a numerical simulation suggests an alternative how to eval- constant flow rate conditions of the in-situ test they might
uate the deposition rate of mineral due to CO2 -water-rock undergo re-entrainment. If we accelerate the stick rate of
interaction. This carbonate mineralization on geologic seques- calcite precipitation during injecting carbonated water, the
tration will decrease the possibility of CO2 leakage risk at permeability reduction of reservoirs will occur.
hydrothermal conditions and prevent buoyancy from moving
Figure 1. (a) Schematic experimental setup for column tests (b) schematic experimental setup for in-situ tests.
Figure 2. Column test: (a) Fitting flow rate curve. In-situ test: (b) head and flow rate curve near the bore hole (c) fitting permeability curve.
546

Risks and hazards

The SEALEX in situ experiments: Performance tests of repository seals
J.D. Barnichon, P. Dick & C. Bauer

Institut de Radioprotection et de Sûreté Nucléaire, Fontenay-aux-Roses, France
Keywords: field measurements, bentonite core saturation
As a part of the overall IRSN R&D program that provides the Five performance tests using wireless intra-core instru-
bases for scientific expertise on disposal safety, the SEALEX mentation allow exploring conditions that may impact the
project is focused on sealing systems efficiency. Thanks to long-term performance of a clay-based seal, changing a single
limited-size in situ experiments on long-term performance parameter at a time with respect to a base case:
(i.e. under isothermal and water-saturated conditions) of
• The first test is a base case test itself, with a core build of
clay-based seals, SEALEX is dedicated to:
monolithic precompacted disks (70/30 MX80/sand mix).
• Test the long-term hydraulic performance of sealing sys- • Three tests designed to quantify, with respect to this base
tems (in normal conditions, i.e. non altered), for different case, the impact of the technological choices retained
core compositions (pure MX80, sand/MX80 mix) and for the engineered barrier. Modifications concern the
conditionings (pre-compacted blocks or in situ compacted); intra-core geometry (jointed disks), the core composition
• Quantify the impact of intra core geometry (construction (MX80/sand ratio) and the core conditioning (precom-
joints in the case of pre-compacted blocks) on the hydraulic pacted vs. in situ compacted):
properties of sealing systems;
– The intra-core geometry (jointed vs. monolithic disks,
• Quantify the effect of altered conditions (an incomplete
both precompacted);
saturation of the swelling clay or an incidental decrease
– The core composition and conditioning (core made of
of the swelling pressure caused by a failure of the concrete
pure MX80 pellets/powder compacted in situ vs. 70/30
confining plugs) on its performance, which tests the concept
mix precompacted monolithic disks);
robustness with respect to the hydraulic characteristics of
– The core composition (low MX80/sand ratio vs. reference
the system.
one 70/30);
This paper presents the main specifications of this test pro-
• Finally, a test to quantify the influence of altered conditions
gram, simulation results of the full hydration phase and the
(saturation defect and confining plug failure) with respect
first experimental results of the initial hydration phase.
to this base case.
All the experiments are emplaced in the Tournemire
URL (Toarcian argillites), via horizontal boreholes (diameter Pre-design simulations were carried out to optimize the
60 cm) excavated from recent drifts (2008). Each experiment dimension of the clay core seals with respect to the hydra-
represents a generic seal mock-up (i.e. either at a relevant tion phase duration and to the tests representativeness. Main
scale, or with relevant characteristics), except the presence of results are presented.
an artificial resaturation system. In parallel, a laboratory characterization program was ini-
Regarding the intra-core instrumentation, it was chosen to tiated to optimize the clay core properties with respect to the
limit as much as possible the disturbance induced by instru- specified target swelling pressure and permeability values.
mentation on the core performance, thus to avoid the potential This laboratory program also aims to characterize the long-
occurrence of flow pathways along signal wire paths. Accord- term intrinsic behavior of the clay core that will be mobilized
ingly, two types of tests have been designed: reference tests especially in the confinement loss test.
and performance tests. Finally, the first results of the initial hydration stages are
The reference tests aims at quantifying the hydro- briefly presented.
mechanical fields (swelling pressure, pore pressure, relative The SEALEX in situ experiments is an original testing pro-
humidity) within the core and their time-evolution as the core gram that should give valuable results for long-term safety
saturation proceeds, using classical wired sensors. evaluation. Preliminary results are encouraging.
549

Post-closure hydromechanical behaviour of a backfilled cavity
N. Dufour & H. Wong

Université de Lyon, Département Génie Civil et Bâtiment, Ecole Nationale des Travaux Publics de l’Etat, Vaulx-en-Velin, France
F. Deleruyelle
IRSN, DSU, SSIAD, BERIS, Fontenay-aux-Roses, France
Subject: Underground disposal facility
Keywords: nuclear disposal facility, numerical modelling, fluid flow, mine design
The handling drifts of an underground nuclear waste disposal Firstly, under simple geometric hypotheses, a complete
facility will be backfilled at the end of their operating time. simplified life cycle of the cavity is assumed (Fig. 2): (1) exca-
The lining of these tunnels will thus no longer be maintained. vation; (2) waiting for the liner; (3) lining support installation
Inevitably, they will lose their mechanical integrity with time and exploitation; (4) post-closure stage after backfilling.
and in a creeping rock, this may induce an inward, converg- Initial and boundary conditions are also presented.
ing movement of the surrounding rock mass. After filling up Secondly, fundamental equations describing coupling phe-
all the voids left behind by the construction and backfilling nomena are shown. From the fluid mass conservation equa-
operations, the rock mass would compress the backfill until a tions, the hydraulic diffusion equation is derived, linking pore
new equilibrium state is reached. In saturated porous media, pressure to volumetric strain. Poroelasticity equations are then
this could induce a water outflow, which could accelerate the recalled, giving notably incremental stress and porosity versus
dissemination of radioactive elements. This important issue incremental strain and pore pressure. Viscoplastic damage-
needs to be investigated for the purpose of long term safety able behaviour is finally described. From these equations are
assessment. derived the characteristic times which allow us to analyse
The present study aims mainly at estimating the hydraulic hydromechanical phenomena from a qualitative point of view.
disturbances inferred by this load redistribution between the Moreover, within the framework of saturated poroelasticity,
liner and the backfill around a typical gallery of an under- a quasi-analytical solution of this problem is developed using
ground nuclear waste disposal facility. The evolution kinetics Laplace transform. Solution in space and time domains can
of different hydromechanical variables (water pressure, dis- be obtained via numerical computations using the Stehfest’s
placements…) depends on a number of factors: rate of lining algorithm. It will be used as a benchmark test in subsequent
deterioration, hydraulic diffusivities of the ground and the stages. It shows for example the dissipation phenomenon dur-
backfill, initial backfill compaction as well as the general ing phases (2) and (3) (Fig. 3) or the outward displacement of
behaviour of the rock mass. The latter may exhibit a com- the host medium after backfilling (Fig. 4).
plex hydromechanical behaviour involving creep, hardening, Finally, a numerical solution is highlighted using finite ele-
and even damage. ments method via Cast3M software. It allows considering
To analyse this difficult problem, a progressive step-by- more realistic but complex behaviour of the host rock, sup-
step approach has been adopted from analytical modellings to posed to be a porous medium, such as creep (with or without
numerical simulations. hardening), coupled or not with damage effects. Comparisons
between numerical results and previously obtained quasi-
analytical solutions or published results show consistency and
allow to gain confidence in this numerical modelling.
Considering influence of liner damage or backfill com-
paction on the post-closure hydromechanical behaviour of the
cavity, principal results are presented.
Figure 1. Typical layout of an underground radioactive waste

disposal facility in a clay geological formation (ANDRA 2005) Figure 2. Simplified life cycle of a backfilled cavity.
550

Figure 3. Pore pressure profiles in the host rock at different instants
between t2 and t3 . Analytical/numerical comparison at t = 0.17 and
150 years. Figure 5. Temporal evolution of water pressure at the inner rock sur-
face starting from backfilling at t3 , in the cases of a non-degradable
or brittle liner.
Figure 4. Displacement profiles in the host rock at different instants

after t3 . Figure 6. Temporal evolution of water pressure at the inner rock sur-
face starting from backfilling at t3 , in the cases of a non-degradable
or progressively leached liner.
Different failure modes are taken into consideration: (1) an
undamaged liner, (2) a sudden, brittle failure and (3) leaching
process. Figures 5–6 show a 1 Mpa maximum overpressure in
case (2) and no predicted overpressure in case (3).
Impact of backfill compaction is also highlighted though
different backfillYoung’s modulus. A well compacted backfill
could generates a lower overpressure than a stiff one.
These results constitute the first step to more sophisticated
numerical analyses considering more realistic behaviours,
loadings and more complex geometries.
551

Risk assessment of carbon dioxide storage in carboniferous formations
He Manchao
Stake Key Laboratory for GeoMechanics and Deep Underground Engineering, Beijing, China
Rita Leal e Sousa

Department of Civil Engineering and Environment, Massachusets Institute of Technology, Cambridge, MA, US
Ana Gomes & Luis Ribeiro e Sousa

Department of Civil Engineering, University of Porto, Porto, Portugal
Euripedes Vargas Jr.

Department of Civil Engineering, Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil
Zhang Na
Stake Key Laboratory for GeoMechanics and Deep Underground Engineering, Beijing, China
Keywords: risks and hazards; coal mines; monitoring
There are several ways of mitigating greenhouse gases (GHG) safely injected and stored at well characterized and properly
emissions to the atmosphere, however, the storage of large managed sites. Injecting carbon dioxide in deeply geological
quantities of carbon in geological formations is presented formations can store it underground for long periods of time.
today as one of the most effective and visible results. Carbon At depths below about 800–1000 m, CO2 has a liquid-like
dioxide (CO2 ) capture and storage (CCS) is a process consist- density that permits the potential for an efficient use of the
ing of the separation of CO2 from industrial and energy-related underground reservoirs in porous of sedimentary rocks.
sources. Fig. 1 brings together, in schematic form, the main Geological storage requires constructing facilities to cap-
sources and some of the possible storage sites. ture in large emission sources of CO2 such as power for
Storage of CO2 in deep geological formations uses many electricity production or cement, steel, ethanol, among others.
of the technologies developed by oil and gas industry and has It is also possible to act on fuel, that is, instead of proceeding
been proved to be economically feasible under specific condi- to the capture of CO2 after combustion of coal or natural gas
tions in oil and gas fields and saline formations (IPCC, 2005). can be drawn primarily engaged in industrial units, the car-
CO2 can also be stored in carboniferous formations, either in bon present in them. The captured CO2 is then transported by
unminable coal seams or in abandoned coal mines. CO2 can be pipelines or in ships, for underground storage sites. Most of
the mechanisms related to this technology are not new, since
they are already employed by the oil industry by contract-
ing for management and distribution of natural gas, for some
industries in the food sector, among others.
Currently CO2 capture is a costly and energy consuming.
The costs obviously depend on the dimension of the industrial
unit and of the type of fuel used. There are four basic systems
for capturing CO2 from use of fossil fuels and/or biomass
(IPCC, 2005): capture from industrial process streams; post-
combustion capture; oxy-fuel combustion capture; and pre-
combustion capture.
The environmental impacts from geological storage of CO2
can be integrated into two types of categories, i.e. local envi-
ronmental effects and global effects from the release of stored
CO2 to the atmosphere. Global effects may be viewed as the
uncertainty in the effectiveness of CO2 storage. Local hazards
arise from causes like direct effects of elevated gas-phase CO2
concentrations in the shallow surface or near surface, effects of
dissolved CO2 on groundwater chemistry and effects caused
Figure 1. Capturing and storage of CO2 (NETL, 2007). by displacement of fluids by the injected CO2 .
552

Figure 2. Potential escape mechanisms (Benson, 2005).
Figure 3. BN for Risk Analysis of storage of CO2 .

Table 1. Hazard identification scenarios.
Hazard Description
H1 Leakage of CO2 from pipelines or pumping stations

H2 Leakage of CO2 from shipping
H3 Slow and steady leakage of CO2 from geological storage
H4 Fast and large discharge of CO2 from geological storage
H5 Leakage from geological storage to groundwater
H6 Leakage of CO2 from geological storage to fossil fuel
assets
H7 Leakage of CO2 that eliminates the benefits of geological
storage
H8 Induced fracturing or seismicity
H9 Leakage from abandoned coal mines
Different potential escape routes for CO2 injected into car-

boniferous formations exist. Risk assessment should be an
integral element of risk management activities, like site selec- Figure 4. Diagram for hypothesis A.
tion, site characterization, storage system, design, monitoring
and if necessary remediation. The methodology to be devel- Development of methodologies for risk evaluation is briefly
oped for risk assessment will be developed using Bayesian presented at section 5 and applications of BN are outlined for
Networks (BN). BN can be used at any stage of a risk analy- different hazard scenarios in section 6. This is followed by the
sis, and provide a good tool for decision analysis. Furthermore, conclusions drawn in section 7.
they can be extended to influence diagrams, including deci- For the risk analysis due to CO2 the developed BN is pre-
sion and utility nodes in order to explicitly model a decision sented at Figure 3, as well as an application for one of the
problem. hypothesis admitted (Figure 4).
This paper reviews literature published on geological stor-
age of carbon dioxide in deep saline aquifers and carbonifer-
ous formations, including abandoned coal mines with special REFERENCES
emphasis in the problematic associated to risk assessment.
Section 2 gives details on the technical aspects of injection and Benson, S. (2005). Carbon Dioxide Capture and Storage: Overview
safety storage. Figure 2 shows potential escape mechanisms with an Emphasis on Geological Storage. Tutorial Presented at the
of CO2 . AGU Annual Meeting, 5/Dezembro/2005, San Francisco.
In section 3 the risk assessment and risk concepts and the IPCC (2005). Carbone dioxide and storage. IPCC Special Report,
different methodologies used with special reference to BN Cambridge University Press, 431p.
are referred. After reviewing and discussing known hazards NETL – National Energy Technology Laboratory (2007). Carbon
Sequestration Technology Roadmap and Program Plan 2007.
for carbon sequestration, section 4 analyze preventing risks by United States Department of Energy.
monitoring. In Table 1 are referred nine hazard identification
scenarios that were characterized.
553

Optimisation of a borehole-tunnel concept design for HLW disposal in granite using a

3D coupled THM modelling
H. Kunz
Federal Institute for Geosciences and Natural Resources (BGR), Hanover, Germany
H.-G. Zhao
Beijing Research Institute of Uranium and Geology (BRIUG), Beijing, P.R. China
H. Shao & J. Hesser

J. Wang
Beijing Research Institute of Uranium and Geology (BRIUG), Beijing, P.R. China
V. Bräuer
Keywords: nuclear repository, numerical modelling, case studies, stability analysis
Within the Chinese-German co-operation program between

BGR (Germany) and the Beijing Research Institute of Ura-
nium Geology (BRIUG, China) the Chinese vertical design
concept for vitrified HLW is intensively studied in a compar-
ative modelling exercise (Kunz et al, 2009). According to this
concept (Fig. 1), 150 l vitrified HLW will be filled in each
carbon-steel canister. The cylindrical canister with a diameter
of 820 mm and a length of 1790 mm will then be emplaced in
a vertical borehole with a diameter of 1920 mm. The borehole
and tunnel will be filled with bentonite and backfill material
respectively.
In China, granitic rock in the Beishan site in North-western
China is considered as host rock for a potential repository area.
The Gaomiaozi bentonite is selected as a favourable buffer
material for the HLW repository. Site characterisation includ-
ing regional geological settings, crust stability, geological Figure 1. Borehole-tunnel concept design for HLW disposal in
characterisation, hydrogeological, and methodological stud- China.
ies are currently in progress. Simultaneously a comprehensive
laboratory program for investigating the THM behaviour of
bentonite is also being carried out by BRIUG. disposal, the thermal properties of the buffer material of the
A two dimensional model based on the data from the site engineered barrier system (EBS), the gaps around the canister
and laboratory investigations was used for the construction in the EBS, the initial temperature and thermal properties of
design studies using the programme code ANSYS (Zhao the surrounding host rock, as well as the repository design
2009). Mechanical stability as well as hydraulic and chemical layout.
isolation are the key issues in the repository design. Safety The calculations show that the optimised design for a verti-
analysis is done for the repository construction and operation. cal disposal concept of the canister is the distance of 9.5 m
The same is also applied for the long-term stability under the for the disposal tunnel and 9.5 m for the deposition holes
coupled thermo – hydraulic – mechanical – chemical (THMC) (Fig. 2), which are favourable in comparison with those of
conditions in the closure phase. the horizontal disposal tunnel design.
As a crucial criterion for the dimension of a repository the On the basis of 2D studies, a 3D model (Fig. 3) taking a
temperature on the canister surface is set to be less than 100◦ C. quarter domain including three disposal tunnel and 30 depo-
The factors that affect the highest temperature on the canister sition holes into consideration is used to verify the 2D results
surface are the initial power release after a 30-years interim and to optimise the dimension of the disposal. Due to the
554

Table 1. Maximal temperature.
2D 3D 3D 3D
Distance T-Model T-Model TM-Model THM-Model
9.5 m 87◦ C 67.4◦ C 69.22◦ C 74.22◦ C
With an optimised dimension (both distances between

tunnels and holes are 9.5 m) from 2D calculation, the three
dimensional model has 236522 tetrahedral elements. All
parameter values (Kunz 2009) for the THM modelling are
checked and compared by both partners for consistency.
Three runs have been done. One model considers only
Figure 2. Optimisation of dimension of borehole and tunnel. thermal process, one model couples thermal and mechani-
cal processes, and the third one couples thermo-hydraulic-
mechanical processes taking the drying due to high tempera-
ture and therefore vapour diffusion in the bentonite buffer into
consideration. The maximal temperature reached in the total
system from different model is different (Tab. 1).
The comparison between 2D and 3D modelling shows that
due to the lower maximal temperature in case of a 3D model,
the dimension between disposal tunnels and deposition holes
may be optimised. In other words more than ten percent
of the construction cost can be reduced based on the same
thermo-mechanical criterion. The modelling results may serve
as a basis for further construction design. Therefore, the 3D
fully coupled modelling has an economical and engineering
meaning.
REFERENCES
Kolditz, O. & Shao, H. (2010): Developer benchmark-book, OGS

5.0, UFZ-Publisher, Leipzig, Germany.
Kunz, H., Zhao, H.-G., Nowak, T., Shao, H., & Wang, J. (2009):
Figure 3. Temperature field at the year in which the maximal power Comparative coupled thm modelling of Chinese HLW disposal
decay reached. concept using codes rockflow and ANSYS, Sinorock – ISRM
International Symposium on Rock Mechanics, Hong-Kong, 19–
22th May, 2009.
complexity of the system and processes considered, the gaps Zhao, H.-G. (2009): A study of thermal conductivity property and
in both models are neglected. The air- and water filled gaps tunnel stability of high-level radioactive waste repository – the
were considered in the 3D coupled THM modelling for single case of Beishan preselected site, Gansu Province, Dissertation,
borehole analysis (Kunz 2009). Briug, Beijing, China.
555

Design and development of a large-scale THMC experiment of compacted bentonite for

geological disposal of high level radioactive waste in China
Y.M. Liu, J. Wang, X.G. Zhao, D. Ke, J.L. Xie, S.F. Cao, L.K. Ma, W.J. Jiang & L. Chen
CNNC key laboratory on geological disposal of high level radioactive waste, Beijing Research Institute of Uranium Geology, Beijing, China
Keywords: nuclear repository; lab testing; numerical modeling; thermo-hydro-mechanical-chemical (THMC)
1 INTRODUCTION
The buffer/backfill material is designed to stabilize the

repository excavations and the coupled thermo-hydro-
mechanical-chemical (THMC) conditions, and to provide
low permeability and long-term retardation. A bentonite-
based material is often proposed or considered as a possible
buffer/backfill material for the isolation of the HLW.
In order to study the behavior of the GMZ bentonite under
relevant repository conditions, a mock-up experiment, named
China-Mock-up, was proposed based on a preliminary con-
cept of HLW repository in China. The experiment is intended
to evaluate thermo-hydro-mechanical-chemical (THMC) pro-
cesses taking place in the compacted bentonite-buffer during
the early phase of HLW disposal and to provide a reliable
database as input information for numerical modeling and
further investigations.
The main objectives of the China-Mock-Up include:
(1) to study the property of GMZ Na-bentonite under coupled
THMC conditions; (2) to study the bentonite-canister reac-
tion under coupled THMC conditions; (3) to simulate vertical
placement of a container with radioactive waste; (4) to monitor
the behavior of GMZ Na- bentonite barrier at high tempera- Figure 1. Sketch of the China-Mock-up facility (unit: mm).
ture and in special water; (5) to experiment the installation
method and validity of sensors; and (6) to provide data for
future design for engineered barrier system. The computer-controlled triaxial experiment machine in
combination with specially designed steel molds was used
to make the GMZ-Na-bentonite into compacted blocks with
2 EXPERIMENTED MATERIAL five different shapes. High-density pellets of bentonite in
different grain sizes were filled in the space between the
The GMZ bentonite deposit has been selected as the most bentonite blocks and the heater / steel tank walls. The total
potential buffer/backfill material supplier for China’s HLW dry density of compacted bentonite blocks and pellets is
repository. Comprehensive studies have been conducted on the 1600 kg/m3 .
GMZ bentonite. The previous study on GMZ bentonite shows
that the bentonite is characterized by high content of Mont-
morillonite (70%) and low impurities. Various experiments 3 DESIGN AND EXPERIMENT OF THE
have revealed that the GMZ bentonite has cation exchange CHINA-MOCK-UP T-H-M-C EXPERIMENT
capacity 77.30 mmol/100 g, methylene blue exchange capac-
ity 102 mmol/100g, and alkali index 1.14. The main properties Figure 1 shows the layout of the China-mock-up experi-
of the bentonite compacted to a dry density of 1.8 g/cm3 are: ment. It was constructed with compacted bentonite-blocks in
thermal conductivity of around 1.0 W/m·K at water content of a large steel tank of 900 mm internal diameter and 2200 mm
8.6%, hydraulic conductivity of 1×10−13 m/s, and swelling height. An electric heater of 300 mm diameter and 1600 mm
pressure of 10 MPa. Those characteristics indicate that the length, which is made by the same stainless carbon steel as
GMZ bentonite has the very similar properties as those of the the substitute of a real HLW container, is placed inside the
mostly investigated MX-80 and FEBEX bentonites to be used bentonite-buffer. The EBS system is heated by the heater from
as buffer/backfill material. ambient temperature to 90◦ C and then will be cooled down.
556

The groundwater flow is simulated by injecting the formation The water will be slowly injected to hydration system. First
water (taken from the host granite rock in the Beishan site / the pellets around blocks will be filled with water and after-
URL, NW China) around the outer surface of the barrier. It wards the water pressure was gradually increased to 1 MPa
can be expected that complex T-H-M-C processes will occur after 1 year. Then keep the water pressure for about four years.
in the bentonite-buffer, which will be monitored by a number After experimenting, the Mock-up facility will be disman-
of sensors to be installed at various locations in the buffer. The tled and samples will be extracted from the buffer. Possible
main parameters to be measured in the EBS include temper- changes of the bentonite properties will be investigated by
ature, water inflow, relative humidity (suction), swelling and determination of geochemical, physico-chemical and miner-
total pressure, as well as displacement of the heater inside the alogical components of the buffer, and special attention will
buffer. The experiment is conducted at the BRIUG laboratory be given to the interfaces of bentonite/canister.
in Beijing. The mock-up will be numerically modeled to verify the con-
In order to design the power of the heater and the parameter stitutive models and the computer codes will be used, such as
of the heat insulation layer, as well as the evolution of the heat, CODE-BRIGHT, LAGAMINE and FLAC3D, for THM pro-
water flow and stress in the buffer material, numerical mod- cesses modeling; and EQ3/6 and Crunch-flow for modeling
eling was conducted. Figure 4 shows the 3D geometry model THC processes. The modeling work includes blind predictions
of China-Mock-up established by FLAC3D. The power of the at the beginning of the experiment, calibrated predictions in
heater is suggested to be 1000 W. If the thermal conductivity the mid-term of the monitoring time, and final simulations at
of the heat insulation material is 0.04 W/m◦ , the thickness of the end of the monitoring period.
the heat insulation layer is 35 mm.
For the China-Mock-up, 44 bentonite block sections, each
with a thickness of 50 cm, are installed across the entire length 4 CONCLUSION
of the tank. Each complete buffer section consists of two or
three concentric rings, which are dependent on the space in The buffer material is one of the main engineered barriers for
which the heater is placed, formed by compacted bentonite the HLW repository. In order to study the behavior of the GMZ
blocks with different shapes and numbers. bentonite under coupled THMC conditions, a large-scale
The China-Mock-up is equipped with 10 different types mock-up facility, China-Mock-up based on a preliminary con-
of sensors to monitor the comprehensive performances of cept of HLW repository in China, has been designed and
GMZ Na-bentonite under coupled THMC conditions. The 6 constructed in the laboratory of BRIUG.
sensor types inside the China-Mock-up include stress sen- The experiment procedures of China-Mock-Up include
sor, hydraulic pressure sensor, LVDT displacement sensor, experiment design, preparation, installation, conduction (heat-
temperature sensor, RH sensor and electrochemical corrosion ing and cooling), dismantling and post-experiment, evaluation
sensor. In addition, a series of metal corrosion samples are and report. The modeling work will be performed parallel to
placed inside the bentonite blocks and crushed pellets to inves- the experiment.
tigate the influence of internal environment of the Mock-up The China-Mock-Up experiment is an important milestone
on metal corrosion behaviors. Another 4 sensor types consist- in the buffer material study for HLW disposal in China. The
ing of Coriolis mass flowmeter, fiber Bragg grating (FBG) parameters and evaluation of THMC processes taking place
strain/temperature sensor, resistance strain gauge and dial in the compacted bentonite-buffer during the early phase of
gauge are located outside the Mock-up. HLW disposal can provide a reliable database for numerical
The assembly works started in March 2010.After the assem- modeling and further investigations of EBS, and the design of
blage and sealing, the heater started to work. The temperature HLW repository.
level was gradually increased during the first six months
experiment. The maximum of temperature will be reached,
with a temperature at the central tube of about 90◦ C. Then
keep the temperature for about four years.
557

Is hydraulic jacking a threat to the long-term safety of a nuclear waste repository?
M. Lönnqvist & H. Hökmark

Clay Technology AB, IDEON Research Centre, Lund, Sweden

Keywords: nuclear repository, fluid flow, hydraulic fractures
1 INTRODUCTION AND BACKGROUND fashion. The glacially induced pore pressure is influenced by
the coupling in two ways: 1) parts of the glacially induced
The Swedish Nuclear Fuel and Waste Management Co. (SKB) mechanical load may be transferred to the pore pressure and
are planning to construct a repository for permanent storage of 2) variations in effective stress may alter the hydraulic con-
high-level nuclear waste in the Forsmark area at approximately ditions in the rock. An assessment, in one dimension, of the
450 m depth in crystalline rock. impact of the mechanical load during SKB’s reference glacial
During its life-time, the repository host rock will be sub- period has shown that with a coupled analysis, the pressure is
jected to a number of different loads, which, under certain cir- increased more quickly during the advance of the ice sheet and
cumstances, will be accompanied by an increase in pore pres- reduced more efficiently during the retreat compared with an
sure. If the pore pressure in a fracture exceeds the sum of the uncoupled analysis. In the upper kilometre of rock the influ-
fracture’s normal stress and tensile strength, the result would ence is small. However, at very large depths, the pore pressure
be a significant and difficult to quantify increase in the frac- is dominated by the mechanical load. Typical variations in
ture’s transmissivity – a phenomenon known as hydraulic jack- transmissivities during the glacial phase are estimated to be in
ing. Therefore, if hydraulic jacking was likely to occur in the the range 0.4–3 of present-day values (Hökmark et al. 2010).
near vicinity of the repository, this would have to be included These should be compared with the large natural variability
in the risk assessment required by the regulating authorities. among fractures at Forsmark (Follin et al. 2007). However,
Hydraulic jacking is unlikely to be initiated at other times without explicit modelling, it is difficult to estimate the impact
than during a glacial period. After excavation, there will be on the effective large-scale permeability. Based on these anal-
atmospheric pressure within the excavated openings leading yses, hydro-mechanical couplings are not further considered
to a reduction in the pore pressure of near-field fractures. The in the assessment of hydraulic jacking.
hydraulic conductivity (Follin et al. 2007) at all depths is suf-
ficiently high to prevent large pore pressures from building up 3 MEANS TO ADDRESS HYDRAULIC JACKING
during the heated phase (Selvadurai & Nguyen 1996). How-
ever, during combined glacial and permafrost conditions, the The pore pressure is estimated by use of a combination of
low permeability of the frozen rock will prevent an upwards analytical solutions and numerical models in two dimensions
flow of water and increase the pore pressure. At the end of based on a continuum representation of the rock mass. The
the glaciation, a large proportion of the pore pressure may be ice sheet is represented by a generic steady state ice profile
retained by the rock. When the ice sheet covers the site, the (Paterson 1994) with a maximum thickness of 3 km. Frontal
glacially induced stresses are sufficiently high to balance the advance and retreat rates are set according to typical values in
increase in pore pressure. Several studies of the potential for SKB’s reference glaciation scenario (SKB 2010). The hydro-
hydraulic jacking during a glacial cycle have been conducted static water pressure at the ice/bed interface is set to 90% of
over the years (e.g. Chan et al. 2005, Vidstrand et al. 2008) the ice thickness. For combined glacial and permafrost condi-
but there is no commonly agreed model of the maximum jack- tions, seasonal variations in the pressure beneath the ice sheet
ing depth. The need to include hydro-mechanical couplings in are also considered. These are only noticeable within a dis-
the assessment of hydraulic jacking beneath a warm-based ice tance of a few kilometres from the meltzone. Further away,
sheet has been debated. Although it is theoretically possible the pore pressure is governed by the average pressure over a
for hydraulic jacking to be initiated at large depths, the Fors- seasonal cycle.
mark site does not appear to have the necessary geological The jacking criterion will first be fulfilled in fractures ori-
or hydrological conditions (Hökmark et al. 2006). Observa- ented perpendicular to the least principal stress, which, at
tions of sediment-filled sub-horizontal fractures, which may the Forsmark site, has been shown to be the vertical stress
be indications of hydraulic jacking events in the past, have (Hökmark et al. 2010). Therefore, it is sufficient to consider
only been made to depths of a few tens of metres. This paper the potential for hydraulic jacking of horizontal fractures. It
will illustrate the impact of hydro-mechanical couplings and is assumed that the tensile strength is zero such that hydraulic
establish bounding estimates of the maximum jacking depth jacking occurs when the pore pressure in a fracture exceeds the
during SKB’s reference glacial cycle. normal stress. The glacially induced vertical stress at all depths
is set equal to the mechanical load at the ground surface.
2 IMPACT OF HYDRO-MECHANICAL COUPLINGS
4 RESULTS
In order to determine the hydro-mechanical (H-M) evolution
of the rock mass, it is generally required that equations of During combined glacial and permafrost conditions as the ice
fluid flow and stress or deformations are solved in a coupled front is advancing, the hydraulic conditions of the rock mass
558

in regions without permafrost are approximated by two lay- REFERENCES
ers with different hydraulic diffusivity. The results are very
sensitive to the assumption regarding the hydraulic diffusiv- Chan, T., Christiansson, R., Boulton, G.S., Eriksson, L.O.,
ity (10−2 –10−1 m2 /s) in the uppermost 400 m of the rock at Hartikainen, J., Jensen, M.R., Mas Ivars, D., Stanchell, F.W.,
Forsmark. For the higher value, the maximum jacking depth Vistrand, P. & Wallroth, T. 2005. DECOVALEX III/BENCHPAR
is approximately 350 m, whereas for the lower value, it is PROJECTS. The Thermal-Hydro-Mechanical Responses to a
approximately 225 m. Considering seasonal variations in the Glacial Cycle and their Potential Implications for Deep Geologi-
cal Disposal of Nuclear Fuel Waste in Fractured Crystalline Rock
hydrostatic pressure at the ice/bed interface, the maximum Mass., SKI 2005:28. Stockholm, Sweden: Strålsäkerhetsmyn-
jacking depth could be reduced by 40–50% to about 200 m. digheten (Swedish Radiation Safety Authority).
These results are based on the assumption that there is 100% Follin, S., Levén, J., Hartley, L., Jackson, P., Yoyce, S., Roberts, D. &
permafrost coverage in front of the ice sheet. Open taliks Swift, B. 2007. Hydrogeological characterisation and modelling
(unfrozen parts of the permafrost) or other discontinuities of deformation zones and fracture domains, Forsmark mod-
within the permafrost body would efficiently dissipate the elling stage 2.2, SKB R-07-48. Stockholm, Sweden: Svensk
pore pressure. Kärnbränslehantering AB (SKB).
During the retreat phase, the hydraulic conditions are Hökmark, H., Fälth, B. & Wallroth, T. 2006. T-H-M couplings in
approximated by uniform bulk values of the hydraulic diffu- rock. Overview of results of importance to the SR-Can safety
sivity. In reality the hydraulic conductivity decreases rapidly assessment, SKB R-06-88. Stockholm, Sweden: Svensk Kärn-
bränslehantering AB (SKB).
with depth (Follin et al. 2007). The low permeability at Hökmark, H., Lönnqvist, M. & Fälth, B. 2010. THM-issues in
large depths reduces the vertical flow of water at large repository rock. Thermal, mechanical, thermo-mechanical and
depths and contributes to a more efficient drainage of the hydro-mechanical evolution of the rock at the Forsmark and
upper part of the rock (Lönnqvist & Hökmark 2010). For Laxemar sites, SKB TR-10-23. Stockholm, Sweden: Svensk
the bulk hydraulic diffusivity relevant for the Forsmark site Kärnbränslehantering AB (SKB).
(≥10−2 m2 /s), the maximum jacking depth is of the order of Lönnqvist, M. & Hökmark, H. 2010. Assessment of Potential for
50–100 m and practically independent of the ice frontal retreat Glacially Induced Hydraulic Jacking at Different Depths, SKB
rate (200–500 m/year) and assumptions regarding time-frame R-09-35. Stockholm, Sweden: Svensk Kärnbränslehantering AB
of the approach (40 m/year for 10,000 years or steady state (SKB).
conditions). Paterson, W.S.B. 1994. The physics of glaciers. Elsevier Science Ltd.,
Oxford, UK.
Selvadurai, A.P.S. & Nguyen, T.S. 1996. Scoping analyses of the
5 CONCLUSIONS coupled thermal-hydrological-mechanical behaviour of the rock
mass around a nuclear fuel waste repository. Engineering Geology
47: 379–400.
Results from all models support the view that hydraulic jack- SKB 2010. Climate and climate related issues for the safety
ing will be confined to the near-surface part of the rock. assessment SR-Site, SKB TR-10-49. Stockholm, Sweden: Svensk
Locally, hydraulic jacking can occur in suitably oriented frac- Kärnbränslehantering AB (SKB).
tures close to repository openings (Hökmark et al. 2010), but Vidstrand, P., Wallroth, T. & Ericsson, L.O. 2008. Coupled HM
this does not influence the large-scale potential for jacking. effects in a crystalline rock mass due to glaciation: indicative
The estimated maximum jacking depths are likely to be over- results from groundwater flow regimes and stresses from an FEM
estimates since the potential for hydraulic jacking is greatest study. Bulletin of Engineering Geology and the Environment 67(2):
near the ground surface and the associated increase in fracture 187–197.
transmissivity, if hydraulic jacking does occur, will reduce the
pore pressure in the fracture and prevent hydraulic jacking
from occurring elsewhere.
559

Modelling of rockfall impact using spring-dashpot-slider system
T. Nishimura
Department of Civil Engineering, Tottori University, Japan
Keywords: numerical modelling, risks and hazards, rock slopes and foundations
1 THREE MODES OF BOUNCING MOTION AND THE

COEFFICIENT OF RESTITUTION
Rockfall poses serious hazards to roads and facilities in

regional areas. It is particularly important to have the best
understanding of rockfall trajectories in order to evaluate accu-
rate zone at risk and to design adequate measure to reduce
risk. The use of computer programs is one of the most pop-
ular approaches for the development of the understanding.
The bouncing phenomenon, occurring when boulders impact
on the slope, is the most difficult to model because of its
complexity and very limited knowledge. In order to perform
rational analyses and gain more accurate predictions, it is cru-
cial to improve the rebound modeling by introducing more
mechanical system with quantifying the input parameters in
the system. Figure 1. Comparison of the coefficient of restitution Ret obtained
This study has adopted a spring-dashpot-slider system. The by the experiment with by the numerical simulation.
normal and tangential components (Ren , Ret ) of restitution
coefficient are expressed using the micro-properties (kn , kt The partial behavior is the intermittent stick with some
ηn , ηt ) of the system and the kinetic conditions (Omachi et al. slip or the intermittent slip with some stick behavior. For this
1986, Nishimura 2009). The tangential behavior during the behavior, Ret will be numerically given as a value which exists
impact can be separated into the following three behaviors between the two extreme behaviors.
depending on the forces in the springs (en , et ) and the frictional
angle φ:
(i) stick (et < en tan φ) 2 FREE-FALL-REBOUND TEST AND SIMULATION
(ii) partial slip (et < en tan φ, et = en tan φ)
(iii) full slip (et = en tan φ) A small-scale free-fall-rebound test using a sphere boulder in
laboratory was conducted. The free fall and rebound motions
Assuming that et = en tan φ is satisfied during the whole were captured using a high-speed video camera. The captured
period of the impact and neglecting the damping forces, Ret video was replayed frame-by-frame in a computer program to
for the full-slip behavior is given by: get the position of the boulder.
Figure 1 shows the comparison of the experimental obser-
vation for Ret with the numerical results using the numerical
code implemented with the expressions. When α1 > 50◦ , the
where α1 is the incoming angle and µ = tan φ. experimental result fits with Eq. (1) and the behavior must be
For the stick behavior, the force in tangential spring never identified the full slip. In the lower range of α1 , the change in
reaches at the maximum frictional resistance. The solution for Ret is not obvious and almost the same value of Ret is obtained
the equation of motion produces the following: for α1 = 10◦ and 20◦ . This response is supposed to be the
stick behavior. This paper provides a very important data for
the coefficient of restitution and a possible numerical method
which can capture the behavior not only free fall but also
during impact.
where ωdn and ωdt are the damping circular frequency.
560

Micro-mechanical study on the failure process of clay formations during

excavation of galleries
S. You
School of Civil and Environmental Engineering, University of Science & Technology Beijing, Beijing, China
G.F. Zhao
School of Civil and Environmental Engineering, The University of New South Wales (UNSW), Sydney, Australia
V. Labiouse
Laboratory of Rock Mechanics, Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland
Keywords: nuclear repository; lab testing; numerical modelling; rock failure
Within the 6th EURATOM Framework Program of the Euro- excavated in these two mediums. However, mechanism of why
pean Community, the TIMODAZ project (Thermal Impact the two apparent different failure patterns are formed is still
on the Damaged Zone Around a Radioactive Waste Disposal not clear, it also cannot be directly obtained from existing labo-
in Clay Host Rocks) aims at studying the fracturing and ratory tests or analytical solutions. This paper is to understand
self-sealing processes that develop in the Excavation Dam- the mechanism of the failure process of two different clay for-
aged Zone (EDZ) around disposal galleries for heat emit-ting mations during excavation. In that case, the micro mechanics
radioactive waste and at assessing the impact of the ther- by means of numerical modeling based on the DLSM with a
mal phase on their evolution. In this framework, simulation two phase model is performed. The numerical modeling are
tests are performed at the LMR-EPFL on hollow cylinders try to reproduce different failure patterns observed in exper-
with mechanical loadings fairly similar to the reality that will iments through choosing different micro failure laws of the
occur around galleries for heat emitting radioactive waste. Two micro structure. From the modeling results, it shows that the
different potential geological formations for deep radioactive micro tensile failure between particles is the main reason of
waste repositories are considered, i.e. Boom Clay (a kind of anisotropic failure for Opalinus Clay, whereas the micro shear
plastic clay) and Opalinus clay (an indurated clay) are promis- failure is the main failure mechanism for Boom Clay. Finally,
ing options for deposit nuclear wastes in Switzerland and the role of mechanical anisotropy on the failure process in the
Belgium respectively. The results from laboratory tests show near field of the underground galleries in those clay formations
that different failure patterns will be formed for the galleries are analyzed and discussed.
561

Criterion and classification for strain mode rockbursts based on

five-factor comprehensive method
J.J. Zhang
North China Institute of Water Conservancy and Hydroelectric Power, Zhengzhou, Henan, China
B.J. Fu
Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
Z.K. Li
Tshinghua University, Beijing, China
S.W. Song
Chengdu Investigation and Design Institute, China Hydropower Consulting Group Co., China
Y.J. Shang
Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China
Keywords: case studies; problem rocks; risks and hazards
1 INTRODUCTION Table 1. Rockburst criterion and classification based on five-

factor comprehensive method.
The earliest rockburst activity in the world was recorded at
Leipzig Coal Mine, affiliated to Stanford Coal Field in 1738. No Slight Moderate Heavy
Factor rockburst rockburst rockburst rockburst
Since then, the research on rockburst has been continuing
without slackening in underground construction. σ1 /Rc <0.15 0.15–0.20 0.20–0.40 >0.40
A lot of results from research and engineering practice relat- σθ /Rc <0.20 0.20–0.30 0.30–0.55 >0.55
ing to rockburst showed that rockburst may occur under the Rc /Rt <15 15–18 18–22 >22
following conditions (Zhang and Fu, 2008). Wet <2 2–3.5 3.5–5 >5
Kv <0.55 0.55–0.70 0.70–0.80 >0.80
(1) The major principal in-situ stress should be higher than
certain percentage of the uniaxial compressive stress of Note: σ1 is the major principal stress at the tunnel periphery, σθ
rock material. is the maximum tangential stress at the tunnel periphery, Rc is the
(2) The rock material should be hard and brittle. uniaxial compressive strength of rock material, Rt is the uniaxial
(3) The rock mass should be intact or comparatively intact. tensile strength of rock material, Wet is the elastic energy index of
(4) The elastic strain energy should be much higher than the rock material and Kv is the intact index of rock mass.
dissipation strain energy of rock material.
Otherwise, no rockburst occurs.
five-factor comprehensive method for rockburst criterion and
classification as shown in Table 1.
2 SUGGESTED FIVE-FACTOR COMPREHENSIVE
The following points shall be noted when using the five-
CRITERION AND CLASSIFICATION OF STRAIN
factor comprehensive method:
ROCKBURST
(1) For construction sites being prone to rockburst, during
For this reason, the rockburst activity is subjected to multiple the design stage, the in-situ stress condition at the tunnel
factors. However, the available criteria were derived mainly periphery could be obtained by using numerical analysis
by single factor or a few factors. For refraining from the lim- method from the regional initial geostress field. In addi-
itation caused by single factor or by other artificial factors tion to the laboratory and field tests, we can get some
in using fuzzy mathematics for determining the weighting, preliminary information of σ1 , σθ , Rc , Rt and Kv . During
with reference to different methods of rock burst criteria the construction stage, it is suggested to carry out some
and classification commonly used worldwide, in connection additional tests in-situ to get more accurate date of σ1 , σθ
with engineering practices in China, this paper suggests a Rc , Rt , Wet and Kv .
562

(2) The terms σ1 and σc are necessary external causes for
rockbursting. Therefore, the terms σ1 /Rc and σθ /Rc are
essential mechanical condition for rock bursting. As
internal causes, the terms Rc /Rt , Wet and Kv are inherent
properties of rocks. They decide the rockburst intensity
under external causes.
(3) Firstly, when the rockburst class determined by using
σ1 /Rc is greater than that from σθ /Rc , the rockburst class
should be adopted by σ1 /Rc . Otherwise the rockburst
class should be adopted by σθ /Rc . Secondly, the rock-
burst classification should be determined by terms of
Rc /Rt , Wet and Kv respectively. Finally, the classification
is determined by comprehensive analysis. If any item of
Rc /Rt , Wet and Kv is lower or higher by one class, the
final comprehensive rockburst classification should be
on the low side or high side respectively. If any item of
Rc /Rt , Wet and Kv is lower or higher by two classes, the
final comprehensive rockburst classification should be
decreased by one class or increased by one class.
(4) As long as any one item is classified to be no rockburst, Figure 1. Diagrammatic explanation of the five-factor comprehen-
even if other four items satisfy the rockburst requirement, sive rockburst criterion and classification.
no rockburst will occur. It is possible to observe some Notes: NR-no rockburst, SR-sight rockburst, MR-moderate rock-
cracks or fractures at the tunnel surface. burst, HR-heavy rockburst, A-steel fiber shotcrete or C30 Nano
shotcrete; B-rock bolt and rock cable or dense rock bolt and long
rock bolt; C-lattice steel frame and C25 secondary shotcrete.
3 VERIFICATION OF FIVE-FACTOR
COMPREHENSIVE ROCKBURST CRITERION AND
CALSSIFICATION IN ENGINEERING PRACTICES 6 CONCLUSIONS
For verifying the suggested method mentioned above, 5 case At present, different methods for rockburst criteria and clas-
histories relating to underground works of Jinping I and II sification put forward by different scholars are available.
Hydropower Stations, Cangling Highway tunnel, diversion With reference to different methods available, combining
tunnels of Tianshengqiao No.2 and Taipingyi Hydropower sta- engineering experiences in China, this paper presents a five-
tions in China are provided. The results obtained seem to be factor comprehensive method for criterion and classification
satisfied. of strain-mode rockbursts as well as some case histories for
Especially, in Jinping I Hydropower Station, the average further discussion.
value of Rc /Rt is 11 (<15), so there was no rockburst, but a
series of failure happened in the study area (Chengdu Investi- ACKNOWLEDGEMENT
gation, Design and Research Institute, 2009; Chinese Society
for Rock Mechanics and Engineering, 2009a,b). Sincere thanks are due to Prof. QIAN Qihu, Fellow of Chinese
Academy of Engineering for his kind direction and revision.
4 DIAGRAM FOR FIVE-FACTOR ROCKBURST

CRITERION AND CALSSIFICATION REFERENCES
The five-factor rockburst criterion and classification can be Chinese Hydroelectric Consulting Group, Chengdu Investigation,
explained through diagram as shown in Fig. 1. Design and Research Institute. 2009. Report on Deformation
It is indicated that the heavy rockburst region is located at and Failure Mechanism as well as Reinforcement Measures of
the left upper part with inclined parallel lines when all the five Surrounding Rock Masses in Underground Caverns Complex of
Jinping No.1 Hydroelectric Power Station on Yalongjiang River in
factors belong to heavy rockburst class simultaneously. Sichuan Province. Sept. 2009.
The area with vertical parallel lines at the middle upper Chinese Society for Rock Mechanics and Engineering, Chengdu
part shows that only 3 among the 5 factors belong to heavy Investigation, Design and Research Institute, Chinese Hydro-
rockburst class. electric Consulting Group. 2009a. Consultation Report for the
Stability and Support Technology of Surrounding Rock Masses of
Underground Complex for Jinping I Hydropower Station.
5 DISCUSSIONS ON SOME PROBLEMS Chinese Society for Rock Mechanics and Engineering. 2009b. The
International Consultation Report for the Key Technology of Safe
and Rapid Construction for Jinping II Hydropower Station.
Some key problems concerning hardness, brittleness of rock
Zhang Jingjian & Fu Bingjun. 2008. Rockburst and its Criteria and
material, intact index of rock mass and so on are discussed. Control. Chinese Journal of Rock Mechanics and Engineering,
The term RC /Rt is commonly used to evaluate the brittleness Vol. 27 No. 10, pp 2034–2042.
of rock in the world. Obviously, if RC /Rt = 1 the rock should be
equivalent to metal, having no brittleness, but large ductility.
Judging from this, to adopt the brittle index RC /Rt for
evaluating the brittleness of rock is acceptable.
563

Numerical simulation of micro-earthquakes induced by reservoir stimulation in the deep

heat mining project Basel
L. Zhou
M.Z. Hou
Keywords: case studies, dynamics, geothermal, hydraulic fractures, numerical modeling, risks and hazards
1 INTRODUCTION tensile strength left after the initial tensile failure. So the sec-
ond tensile failure behind the fracture front contributes hardly
In the deep heat mining (DHM) project Basel, a seismic to energy releasing. On the contrary certain residual shear
event with a local magnitude of 3.4 was measured after about strength always exists in the rock after the initial shear failure.
12,000 m3 water injection, which caused building damages in There is probably a big shear failed volume (e.g. 1,000,000 m3 )
the city of Basel, the stop of the project for ever and suspicion behind the fracture front during the injection, which could be
against deep geothermy. For this reason it is very impor- the mechanism for the large amount of energy release in deep
tant to investigate the mechanism and to simulate seismic geothermy.
events induced by reservoir stimulation of an EGS (enhanced
geothermal system). Through the numerical modeling, the
process of energy release and the influencing factors on the 3 COMPUTATIONAL PRINCIPLE IN FLAC3D TO
energy release can be better understood, so that the procedure CALCULATE KINETIC ENERGY AND LOCAL
of the stimulation can be optimized and the geo-risks can be MAGNITUDE
minimized.
The kinetic energy of an object is the energy which it possesses
due to its motion. It is defined as the work needed to accelerate
2 MECHANISM OF INDUCED SEISMIC EVENTS BY a body of a given mass from rest to its current velocity. If the
RESERVOIR STIMULATION velocity of the object is known, then the kinetic energy is also
known. To calculate the kinetic energy the velocity must be
The induced seismic event by reservoir stimulation in principle firstly computed, which is the native function in FLAC3D, if
belongs to a thermal-dynamic process, in which the energy the dynamic option is turned on. The theoretical background
conservation rule must be obeyed. In other words the energy of dynamic calculation in FLAC3D bases on linear elasticity.
cannot be lost but transform from one type to another. The The key is to solve the equations of motion. During the calcu-
transformation can be described by eq. (1). lation the velocity of all grid points will be computed at every
calculation step. So the kinetic energy can be calculated with
the known velocity and mass. By use of an empirical equation
the kinetic energy can be converted to local magnitude.
Where dE k = increment of kinetic energy; dW = increment

of deformation energy; dA = increment of external work; and 4 NUMERICAL SIMULATION OF
dQ = increment of heat energy. MICRO-EARTHQUAKES INDUCED BY RESERVOIR
Seismic wave is strictly speaking a form of kinetic energy. STIMULATION IN THE DEEP HEAT MINING
In reservoir stimulation the most kinetic energy comes from PROJECT BASEL WITH FLAC3D
the deformation energy, which is released by rock failure due
to reservoir stimulation. There are two types of rock failure: According to the x- and y-axial symmetry a ¼ model with a
tensile and shear. Both failures lead to release deformation dimension of 700 m × 500 m × 1,179 m in x, y and z direction
energy. But a big amount of deformation energy cannot be is generated, which is shown in Figure 1.
released by tensile failure. There are two reasons. Firstly, the Figure 2 shows the fracture propagation at t = 6 d. At this
tensile strength is normally much smaller than shear strength. time a volume of 850 m (x) × 83 m (y) × 779 m (z) is stim-
Therefore the stored deformation energy before tensile fail- ulated. As can be seen from the legend, there is only shear
ure is much smaller than before shear failure. Secondly, the failure, because the maximum pore pressure does not exceed
most energy releasing by tensile failure happens at the front the minimal principle stress at any time. On the contrary there
of the fracture propagation because there is almost no residual is mostly tensile failure by hydraulic stimulation in gas and
564

Figure 4. Comparison of the measured seismic events and cracked
volume in the numerical modeling.
Figure 1. Calculation model in FLAC3D.
the biggest seismic event of both happens after shut-in. In the
field monitoring in the DHM project Basel the number of
seismic events per hour was also registered. In the numerical
simulation the failure of one element can be treated as one
seismic event. Figure 4 shows the compared results, which
matches well with each other.
5 CONCLUSION
In this paper the mechanism of seismic events induced by

reservoir stimulation in EGS are studied and compared with it
in tight gas reservoirs. The necessary and sufficient conditions
must be satisfied simultaneously, in order to release suddenly
the stored deformation energy. But for releasing of an enough
amount of deformation energy to induce a damageable earth-
quake there must be a big shear failed volume, which appears
Figure 2. Fracture propagation at t = 143 h.
not only at the front of fracture propagation but also behind
it. A THM coupled dynamic simulation of micro-earthquakes
induced by reservoir stimulation in the DHM project Basel
is carried out by using the FDM code FLAC3D. The calcu-
lation is based on the history match of the time dependent
bottom hole pressure. The kinetic energy and local magnitude
are continuously calculated and registered. The kinetic energy
has a good relationship with the injection rate and the viscos-
ity of the injection fluid, while the stimulated volume in the
rock mass depends strongly on the injection volume. Through
the numerical modeling, the process of energy release and the
influencing factors on the energy release can be better under-
stood, so that the procedure of the stimulation can be optimized
and the geo-risks can be minimized. We conclude that the mod-
ern multiple fracturing technology and the horizontal wellbore
Figure 3. Comparison between the numerically calculated and the technology could be a better solution.
measured local magnitudes.
oil boreholes (Hou & Zhou 2009). The reasons are the differ- REFERENCES
ent in-situ stress state, shear and tensile strength, as well as
Häring, M.O., Schanz, U., Ladner, F. & Dyer, B. C. 2008. Character-
softening behavior of rocks.
isation of the Basel 1 enhanced geothermal system. Geothermics
During the injection the kinetic energy in the model at each 37 (5): 469–495.
time step is continuously registered and evaluated as local Hou, Z. & Zhou, L. 2009. Vorausberechnung der gespeicherten und
magnitude, which is shown in Figure 3. The biggest calculated freizusetzenden Deformationsenergie bei Frac-Behandlungen in
value is 2.95, which is smaller than the biggest measured value der Tiefengeothermie. Presentation at the Geothermal Congress
3.4. However the tendency of both is comparable. Furthermore 2009, Bochum, 17–19 November 2009.
565

Mining and rock support

New grouting materials using calcium phosphate compounds
M. Akiyama
Advanced Technology Development Project Group, Geoscience Research Laboratory, Co., Ltd., Yamato, Japan
S. Kawasaki
Faculty of Engineering, Hokkaido University, Sapporo, Japan
Keywords: rock properties, lab testing
To develop novel rock grout materials containing calcium

phosphate compounds (CPC), which are one of the most
fundamental of biological substances, we explored suitable
conditions for CPC precipitation in vitro, conducted uncon-
fined compressive strength (UCS) tests of Toyoura sand test
pieces cemented with CPC, and carried out observations and
elemental analysis of precipitated CPC crystals by scanning
electron microscopy (SEM) (Fig. 1). CPC precipitation was
recognized in all reaction mixtures made from two kinds of Figure 2. Effect of pH on the precipitation volume of CPC.
phosphate stock solution (monoammonium phosphate and
diammonium phosphate) and two kinds of calcium stock
solution (calcium nitrate and calcium acetate). The volume
of CPC precipitation in the reaction mixture increased with
pH, rising from weakly acidic to around neutrality (Fig. 2).
Toyoura sand test pieces cemented with a reaction mixture
containing 1.5 M diammonium phosphate and 0.75 M cal-
cium acetate showed a tendency for the UCS to increase
with time, reaching a maximum of 87.6 kPa at a curing time
of 84 days (Fig. 3). CPC precipitation via calcium nitrate
treatment induced the formation of plate-like crystals with
diameters of about 10 µm, whereas that via calcium acetate
treatment induced whisker-like crystals (Fig. 4).
This study examined the feasibility of a new rock grout
material containing CPC. The results of this study showed
sufficient potential for utilizing the material in the future as:
(1) chemical grout using the self-setting property of CPC, and
(2) biogrout using the pH dependence of precipitation and
crystal structure of CPC.
Figure 3. Temporal variations in unconfined compressive strength

(UCS) of sand test pieces cemented with CPCs.
Figure 1. Flow chart of the study. Figure 4. SEM images of test piece segments.
569

Design of room & pillar exploitations of non-expensive minerals

and ornamental rocks in Spain
L.R. Alejano, F. García-Bastante & J. Taboada

Natural Resources & Environmental Engineering Department, University of Vigo, Spain
R. Migliazza
Dipartimento di Ingegneria Civile, dell’Ambiente, del Territorio e Architectura, Universitá de gli Studi de Parma, Italy
Keywords: mine design, site characterization, case studies, mining, numerical modeling, rock caverns
1 INTRODUCTION
In the recent times, the high cost of land and the stringent
environmental legislation, together with the good prices of
some minerals and rocks, such as marble and roofing slate,
has forced miners to foster the underground mining (Figure 1).
Even if the mining costs increased in comparison to quarrying,
by means of a cheap method like room & pillar, mining costs
can be kept at low levels (10–20 a/ton) in such a way that the
mining of these minerals is still profitable.
We have been involved in the design and stability study of
some of these mines, and in the preset article we seek to present
some of the basic methodology for the design of these mines.
This includes the selection of empirical simple methods for
initial design and the application of numerical techniques to Figure 2. Design and detail of numerical models of a quartz and a
simulate the proposed models (Figure 2). magnesite room & pillar exploitation in Spain.
The combination of empirical, analytical and numerical
techniques result quite useful for these design processes, since
when one single method is applied, some topics maybe over- 2 BASIC DESIGN CONSIDERATIONS
looked, so the combination of methods provide more confident
designs. To provide a good design the field and laboratory investiga-
tions are of relevance.A basic knowledge of the geology is first
needed. Since the underground mines we are talking about are
excavated near open pit mines of the same mineral, we have
available excellent outcrops to measure a good number of dis-
continuities following standards. These data are interpreted to
obtain joint sets and their features.
In our case, we usually have core-drilling in the areas to be
mined, since it is important to locate the walls of the seam to be
mined and to control the mineral grade or quality. Rock cores
are used to perform laboratory characterization tests. Labo-
ratory and field information is gathered to obtain the basic
figures of empirical approaches including Q, RMR and GSI.
Lab tests are interpreted to obtain peak strength envelopes,
deformability pre-failure data and, if necessary, post failure
parameters, following published characterization techniques.
In the room & pillar mining method, the extent of the rooms
is directly related to the mining costs. So in our designs, rooms
tend to be as large as possible. In this way, our initial room
Figure 1. Pictures of some room & pillar exploitations of designs, including the support needs, have been based on the
non-expensive minerals and ornamental rocks in Spain. application of the stability graph method (Potvin, 1988). This
570

method has been useful in different geological environments one or more strata detached from the upper vault and its sta-
to select room orientations, estimate room widths and provide bility is a major safety concern. Roof mechanics have been
support needs. traditionally analyzed by means of analytical voussoir tech-
Once an initial design of the room is proposed, based niques. These techniques have been applied by the authors to
on empirical methods, we recommend to routinely studying back-analyze roof instability problems in a magnesite mine
wedges or structurally controlled instabilities, stress-strain (Alejano et al., 2008). Very often in wide rooms, the roof is
related instability issues and analysing roof bed stability in not self stable and support is needed. The support needs can
case of bedded rock. be quantified according to the indicated voussoir approaches.
Pillar basic design is based on traditional approaches. The Discrete numerical models can be also used for this task.
load on the pillars is preliminary estimated according to the As far as our experience is concerned, low horizontal
tributary area analysis and the operational pillar strength is stresses, relaxation due to the opening of parallel contigu-
estimated starting from the estimated compressive strength ous rooms and the presence of badly oriented joints might
of the pillar taking into account shape and size corrections contribute to the occurrence of local problems. It is helpful
(Hedley & Grant, 1972). Then, final pillar designs are based to analyze these cases to carry out back-analysis follow-
on standard safety factor (SF) over 2. Since we design mines ing different approaches (empirical, analytical, numerical. . .).
at shallow depth, less than 300 m, no pillar stability problems These studies enable a better understanding of failure and to
have arisen so far. fine-tune the most significant parameters for further design.
3 DESIGN CONTROL 4 CONCLUSIONS
The initial designs are based as illustrated on empirical We currently have available a number of sufficiently devel-
approaches and traditional rules of thumb. Once this initial oped rock mechanic tools to design room & pillar mines
design is proposed, and it has been proved that the mine is in shallow mineral or ornamental rock deposits. These tech-
economic, this design is checked out in detail by analysing the niques can foresee varied instability mechanisms, which may
different geomechanical issues which could produce problems occur in this type of exploitations. A proper rock mechan-
during the exploitation stage. ics approach should start with careful field data recovery and
painstaking laboratory testing. Empirical methods are conve-
3.1 Structurally control instabilities nient to propose a basic design, but analytical and numerical
techniques are needed to fine-tune these designs. Moreover,
With the field information and according to block theory or different types of control and monitoring are needed, to check
standard underground wedge analysis the factors of safety of the viability of the design implementation. Some of these
the wedges formed around the rooms are estimated. The initial mines have largely shown to be economically feasible.
support may be increased in particular zones of the rooms if
needed. It is important to correctly quantify the actual persis-
tence of discontinuities for this is a highly relevant issue in ACKNOWLEDGEMENTS
wedge failure control. The projection of joint traces derived
from field data in the roofs and walls of the room, with the The authors acknowledge the Spanish Ministry of Science &
help of codes able to represent joints can be helpful to quantify Technology, Spain, for financial support of the project under
continuity. reference number BIA2009-09673, which has been of help in
some of the research topics presented in the paper.
3.2 Stress-strain behavior
Stress-strain instability studies can be performed with the help REFERENCES
of the various continuous numerical methods available in the
market. In the mines we are designing, due to its low depth, Alejano, L.R., Taboada, J., Bastante, F.G., Rodríguez, P. 2008. Multi-
plastic zones do not usually extend far from the room surface, approach back-analysis of a roof collapse in a mining room
however this stress strain analyses are convenient, for they excavated in stratified rock. Int. J. of Rock Mech. Min. Sci. &
can also put forward the occurrence of tensile zones and reveal Geomech. Abstr. 45: 899–913.
relaxation problems. The role of faults can also be enlightened Hedley, D.G.F. & Grant, F., 1972. Stope and pillar design for the Elliot
by performing numerical models. Lake uranium mines. Can. Inst. Min. Metall. Bull 65:37–44.
Potvin, Y. 1988. Empirical open stope design in Canada. Ph.D. The-
sis. Dept. of Mining and Mineral Processing, University of British
3.3 Study of stability of rooms in stratified media Columbia, 1988.
For the design of rooms in stratified rock masses, particu-
lar attention is given to the immediate roof, which includes
571

Double shearing of rebar steel and cable bolts for effective strata reinforcement
Naj Aziz & Jan Nemcik

Department of Civil Mining and Environmental Engineering, University of Wollongong, NSW, Australia
Hossein Jalalifar
Department of Mining and Petroleum Engineering, Shahid Bahonar University of Kerman, Iran
Subject: Rock engineering in hazardous geo-environment
Keywords: bolts and anchors, laboratory testing
An experimental study was undertaken to examine the shear- These bolts were tested in 20, 40, 50 and 100 MPa con-
ing behaviour of both steel rebar and hollow strand cable bolt crete blocks, subjected to different pretension loads of no axial
across joints and shear planes. Double shearing testing using tension (0 kN), 5, 10, 20 50 and 80 kN respectively.
double shear box as shown below, were made on rebar steel A particular aspect of this study was focused on the influ-
and cable bolts in concrete blocks. Five steel rebar bolt types ence of rebar (bolt) profile configuration on the load transfer
were used for the study. Three bolts were 22 mm core diam- capacity between bolt, resin and rock, under shear. It was found
eter bolt types with different profiles, the forth was 11 mm that; the level of maximum shear load and displacement were
threaded bars and the fifth was a 12 mm construction bolt. different because of different pretension loads; the strength
of the concrete had significant influence on the shear load
behaviour and shear deformation of the concrete in the vicin-
ity of the sheared joint planes; the shearing load across the
joint planes were less in 20 MPa concrete in comparison with
higher strength concrete of 40 and 100 MPa concrete. The
strength of the concrete and the level of bolt pretension were
found to influence the elastic limit and stiffness of the bolt.
The testing of the 28 mm diameter cable bolt was carried
out in a 50 MP concrete with the cables being tested under
two initial axial loading of 50 and 100 kN respectively. The
cable consisted of a hollow mild steel core surrounded with
nine 7 mm diameter wire strands.
Two tests were carried out using two 1.5 m long cable sec-
tions. The first test was limited to a 50 mm travel of the testing
machine loading plates and produced a shear load of 900 kN
(92 t) at the maximum 50 mm displacement. The axial load
generated on the cable bolt was 238 kN (24.3 t). In the second
test the machine travel was increased to 75 mm, and first strand
failure of the cable, due to shear loading, occurred at 1 354 kN
(138 t) vertical load at a vertical displacement of 59 mm. The
cable axial load was in the order of 385 kN (39.3 t). Analysis
of the failure mode and loads achieved showed that the cable
strands bent and the concrete crushed along the shear plane,
the shear loading across the concrete and grouted cable then
Figure 1. Double shear testing of bolts and cable bolts in concrete. reached the tensile strength of the steel wires.
572

Numerical evaluation of stress distribution along grouted rockbolts
S.A. Boulamatsi, P.P. Nomikos & A.I. Sofianos

Keywords: bolts and anchors, numerical modelling
The aim of this paper is to present the numerical evaluation of There, an internal pressure pi = 10 MPa is applied at the tun-
the load distribution along fully grouted rock bolts. In order nel boundary, stresses are initialized in the finite differences
to achieve that, three series of models are prepared using the zones and the model is allowed to reach an initial equilibrium
explicit finite difference codes FLAC2D and FLAC3D (Itasca, state. After the initial reproduction of the stress conditions,
2009). four stages follow which simulate the excavation and the rock
Firstly, plane strain and three dimensional models are pre- reinforcement process. The tunnel excavation is simulated by
pared, which intend to simulate a rockbolt installed in the rock reducing the internal pressure at the tunnel boundary. At the
walls of a circular tunnel with radius R = 5.0 m. Secondly, two second stage, the internal pressure is reduced to pi = 3 MPa
dimensional axisymmetric models are also prepared, which and the model is once again let to reach an equilibrium state.
simulate a rockbolt installed in the rock walls of a spherical At this point the rockbolt is installed along with the grout and
cavity of the same radius. the face plate. In the third and forth stages the internal pres-
Four different rock masses are considered, with the proper- sure is reduced to pi = 2 MPa and pi = 1 MPa respectively. In
ties shown in Table 1. The intact rock properties correspond the last stage no internal pressure is applied.
to a rock of medium to high strength. The rock mass is In the plane strain model series, the rock bolt is simulated
assumed to be elastic-perfectly plastic following the “general- using the “cable” model built in the FLAC2D code. The influ-
ized” Hoek-Brown strength criterion (Hoek et al. 2002), which encing area of the bolt in the rock is modeled as a sector of
is implemented in the FLAC code. For the present simulations circular annulus.
associated flow conditions are approximated. The axial load distribution along the rockbolt is shown in
A rockbolt pattern of 1 m2 is assumed. The length of the Figure 1. These diagrams indicate that the anchoring part of
rockbolt is 3.0 m and its diameter is 25.0 mm. The bolt is con- the rockbolt is reduced with a movement of the neutral point
sidered to be an elastic material with Young’s modulus value towards the rockmass with decreasing rock mass quality. Also,
equal to 200 GPa. The grout has an annulus of 11.5 mm. A the distribution of the load along each bolt is monotonically
Young’s modulus of 10 GPa is assumed and the material is con- decreasing, as expected, either side of the neutral point. Out-
sidered to be either elastic-perfectly plastic, with cohesion and side the plastic zone, it is clear that the bolts practically cease
friction angle 7.0 MPa and 20◦ respectively, or elastic-brittle to carry load.
plastic with the residual cohesion being reduced to 10% of the In the three-dimensional model series, the rock bolt instal-
peak. lation is simulated using the FLAC3D code. The influencing
The in-situ stress field is considered hydrostatic with field area of the bolt in the rock is modeled as a prism and due
stress equal to 10 MPa.This value is applied at the outer bound- to the symmetry conditions only one quarter of the prism is
ary of all models. In all three model series, the simulation is simulated. The z-axis of the model coincides with the axis of
performed in five stages. In the first stage, the model is allowed the rockbolt which is simulated by 1 zone along its radius, 10
to reach an initial equilibrium state before any excavation. zones along its quarter circumference and 60 zones along its
length.
Table 1. Rock mass properties used in the numerical models. Three zones along the grout annulus are used to simulate the
grout. The first two zones near the bolt are assumed elastic and
Rock mass properties the zone close to the grout-rock interface elastoplastic with a
strain softening material model. Peak strength parameters are
σ ci (MPa) 50 the same as those used in the plane strain models, while the
mi 10 residual cohesion is reduced to 10% of the peak.
Ei (GPa) 55 In Figure 2 the results from the three-dimensional models
GSI 90 70 50 30 show that when the quality of the rockmass is low, the axial
mb 6.997 3.425 1.677 0.821 load along the bolt is higher and that in very good quality
s 0.3292 0.0357 0.0039 0.0004
a 0.500 0.501 0.506 0.522
rockmasses the rockbolts are practically inactive.
Erm (GPa) 52.8 40.3 8.55 4.48 Finally, two-dimensional axisymmetric models, such as
*σ cm /p0 2.84 1.41 0.87 0.55 shown in Figure 3, are prepared with the FLAC2D code.
The influencing area of the bolt in the rock is modeled as
*σ cm is calculated with the method of Hoek et al. (2002). a truncated cone.
573

Figure 1. Axial load distribution along the rock bolt obtained from
the FLAC2D models for various values of GSI and the respective Figure 3. 2D axisymmetric simulation of grouted rockbolts: con-
length of the plastic zone. ceptual model.
Figure 2. Axial load along the rock bolt for various values of GSI. Figure 4. Axial load along the rock bolt for various values of GSI.
The bolt, the grout and the rock mass are modeled using In Figure 4 the axial load along the grouted rockbolt for the
finite differences zones. The y-axis of the model coincides first model configuration is shown for the four cases of GSI
with the axis of the rockbolt which is simulated by one zone values. It is observed that the neutral point moves towards to
along its radius and 120 zones along its length. rock mass for lower rock mass qualities.
Two different model configurations are used. In the first To conclude, the maximum load in the rockbolt is evaluated
configuration, interface elements are used to simulate the bolt- to be significantly affected by the quality of the rockmass,
grout and the grout rock interface. These are assigned very being higher for a poorer rockmass. For very good quality
high values of normal and shear stiffness equal to 1000 GPa/m. rockmasses the rockbolts are practically inactive.
The grout is modeled as an elastic material with aYoung’s mod- Further, the bolts outside the plastic zone practically cease
ulus value Eg = 10 GPa. The bolt-grout interface is assigned a to undertake load. This illustrates their rather ground rein-
Coulomb slip criterion with 7 MPa cohesion and 20◦ friction forcement action, than any suspension from deeper more
angle. The grout-rock interface is given the same properties competent strata.
as the rockmass (using equivalent Mohr-Coulomb strength Finally, the neutral point of any rockbolt is found to move
parameters). In the second model configuration, interface ele- deeper with decreasing the quality of the rockmass. Moreover,
ments are not used. The two zones of the grout near the bolt this point does not go much deeper than half a metre for most
are assigned an elastic material model. The third zone near the of the cases examined. Such findings are in accordance with
rock is assigned the properties of an elastoplastic brittle mate- in situ measurements.
rial, with peak strength parameters cp = 7.0 MPa, ϕp = 20◦
and residual strength parameters cr = 0.7 MPa, ϕr = 20◦ .
574

Rock support for deep tunnels in highly stressed rocks
M. Cai
School of Engineering, Laurentian University, Sudbury, Ontario, Canada
P.K. Kaiser
CEMI, Centre for Excellence in Mining Innovation, Sudbury, Ontario, Canada
Keywords: rock support; bolts and anchors; rock bursts; dynamics; tunneling; mine design
1 INTRODUCTION
As the depth of mining and underground civil construc-

tion increases, stress induced rock fracturing is inevitable
and in some cases rocks can fail violently, leading to seis-
mic events and rockbursts, which in some cases can cause
fatalities and injuries to workers, damage to mine infrastruc-
ture and equipment, and substantially increase investment
risk. In the mining and civil tunneling worlds, a few min-
ing induced rockbursts (Beaconsfield gold mine, Australia;
Kidd Mine, Canada; Jingping-II hydropower intake tunneling
project, China; Olmos Trans-Andean tunneling project, Peru)
occurred recently have reached the media due to either tragic
nature or damage severity.
Rockburst risk can often be reduced by selecting appro-
priate mining or excavation methods and sequences, and by
strategically placing developments and infrastructures. How-
ever, due to uncertainties in rock mass properties and boundary Figure 1. Rockburst damage mechanism, damage severity, and
conditions, all engineering design, calculations, and seismic- required support functions (modified from Kaiser et al. (1996)).
ity monitoring will have to rely on ground control measures
and burst-resistant rock support as an important line of defense
to ensure underground safety. Therefore, it is imperative
to design proper burst-resistant support when mining and
tunneling at depth.
2 ROCKBURST PHENOMENON AND ROCKBURST

DAMAGE
2.1 Rockburst phenomenon and types of rockbursts

A rockburst is defined as damage to an excavation that occurs
in a sudden or violent manner and is associated with a seis-
mic event (Kaiser et al., 1996). As mining migrates to deeper Figure 2. Main factors that influence rockburst damage.
ground, in-situ stress becomes high and the likelihood of rock-
burst increases drastically. Rockbursts are mostly associated
with hard rocks and some geological structures such as faults to seismic energy transfer, and rockfalls induced by seismic
and dykes, and in mining are often related to high extraction shaking (see Figure 1). Rock bulking due to rock fracturing
ratio and associated with some particular mining methods. can be caused by both a remote seismic event and the bursting
Three types, strain burst, pillar burst, and fault-slip burst, can event itself. Rock ejection can be caused either by the burst
be classified. event itself, or by a remote seismic event. Seismically induced
rockfalls, as the name suggests, are caused by large remote
2.2 Rockburst damage mechanism
seismic events induced by pillar burst or fault-slip. There are
Kaiser et al. (1996) classified rockburst damage into three many factors that influence rockburst damage and damage
types, i.e., rock bulking due to fracturing, rock ejection due severity and Figure 2 summarizes the main factors.
575

3 ROCKBURST SUPPORT DESIGN PRINCIPLES AND
DESIGN ACCEPTABILITY CRITERIA
3.1 Rock support functions

Kaiser et al. (1996) summarized three key support functions
as: (1) reinforce the rockmass to strengthen it and to control
bulking, (2) retain broken rock to prevent key block failure
and unraveling, and (3) hold key blocks and securely tie back
the retaining element(s) to stable ground.
The reinforcement, retaining, and holding functions do not
act independently. These support functions are achieved by
various rock support elements that are well connected, form-
ing an integrated rock support system. Figure 1 illustrates that
all three support functions are needed in an effective rock-
burst support system no matter what the rockburst damage
mechanism or damage severity is.
Figure 3. Summary of seven rockburst support design principles.
3.2 Rockburst support design principles and design

acceptability criteria for rockburst conditions is conducted by combining different
Realizing the importance of understanding rockburst support holding, reinforcing, and retaining elements and ensuring the
design guiding principles, Cai and Champaigne (2009) sum- integrity of the support system.
marized some of their and other practitioners’ experiences This is achieved by considering compatible support ele-
into seven easy-to-understand principles (Figure 3). The first ments to form an integrated rock support system, and elim-
principle is to avoid rockburst. The second principle advo- inate the weakest link in the system. A design tool called
cates the use of yielding support in bursting grounds. The BurstSupport is currently under development at Laurentian
third principle is to address the weakest link in the support sys- University in Canada. This tool will facilitate the interactive
tem. Other principles are using integrated system, simplicity, process of rockburst support design.
cost-effective, and anticipating and be adaptable. By under-
standing the seven principles, the ability to safeguard workers 4.2 Design verification
and company property can be increased. These core principles Observational methods utilize monitoring as an integral part
must guide support design. in the rock support system design process. Field monitoring
Rock support in burst-prone ground differs from con- provides feedback loops to the design process. A rationale
ventional rock support where controlling gravity induced design combined with field observation and monitoring is the
rockfalls and managing shallow loosing zones is the main key to success for rockburst support design in burst-prone
concern. In addition to these design issues, rock support in grounds.
burst-prone grounds needs to resist dynamic loading and
large rock dilation due to violent rock failure. In burst-
prone grounds, the following four design acceptance criteria 5 CONCLUSION
need to be considered: force criterion, displacement criterion,
energy criterion, and system compatibility criterion. In design, Rockbursts are complex natural and mining-induced phenom-
the load, displacement, and energy capacities must exceed ena occurring in deep underground construction. As mining
the load, displacement, and energy demands, respectively, progresses to deeper grounds, violent rock failure cannot be
and the load, displacement, and energy capacities of surface avoided and it will have to be dealt with on a routine basis by
support must be compatible to these of the reinforcement/ implementing a rockburst resistant support strategy.
holding elements. Rockburst support is not a mystery, nor a skill that only a
few can master. The first important step that leads to mastering
the science and art of rockburst support design is to understand
4 ROCK SUPPORT DESIGN AND VERIFICATION rockburst mechanisms, identify major factors that influence
rockburst damage, understand the support functions, design
4.1 Rockburst support design procedure and tools principles, and design acceptability criteria. We present an iter-
Rockburst support design is to meet the load, displacement, ative design procedure and a design tool for conducting rock
and energy demands with appropriate support capacities, support design in highly stressed and burst-prone grounds.
under given ground and excavation conditions. In each design
domain, one needs to estimate the anticipated seismic event
magnitude and event location as well as potential rockburst REFERENCES
damage mechanisms, and calculate the load, displacement,
and energy demands on the rock support for the dominant Cai, M., Champaigne, D., 2009. The art of rock support in burst-
prone ground. Keynote Lecture. In: RaSiM 7: Controlling Seismic
rockburst damage mechanism.
Hazard and Sustainable Development of Deep Mines, Tang, C. A.
Next, one will have to examine all available rock support (editors), pp. 33–46.
elements and pick the best combination of the support ele- Kaiser, P.K., Tannant, D.D., McCreath, D.R., Canadian Rockburst
ments to form an integrated rock support system with the Support Handbook. Canadian Rockburst Support Handbook.
desired support capacities exceeding the load, displacement, 1996. Geomechanics Research Centre, Laurentian University,
and energy demands previously determined. Support selection Sudbury, Ontario, 314p.
576

Analytical derivations of bolt profile shear strength
Chen Cao, Jan Nemcik & Naj Aziz

Subject: Rock engineering in hazardous geo-environment
Keywords: bolts and anchors, shear load transfer
Current developments in theoretical calculations of stress surface profile geometry indicate a significant variation in
distribution within the resin encapsulating roof bolts in under- peak bolt performance. Both profile height and spacing have
ground mines are presented. Monitoring of steel bolt perfor- been found to affect the load transfer mechanism of bolts
mance indicate that, amongst many parameters, bolt profile (Aziz et al., 2003, 2008). In particular, and as shown in
plays an important role in load transfer between the bolt and Figure 1, wider spacing of 37.5 mm bolt was found to be the
the surrounding strata. To optimise the bolt profile geometry most effective profile in bolt resin encapsulation interaction
a better understanding of the failure mechanisms, that occur when compared with similar height profile bolts of different
within the resin and bolt interaction, must be achieved. The spacing of 12.5 mm, and 50 mm bolts respectively.
short encapsulation pull out tests of roof bolts with various The analytical calculations supported by the numerical
modelling are used here to predict a probable mode of resin
failure while the laboratory and the in situ short encapsula-
tion pull out tests enable direct comparison to the theoretical
investigations.
Particular emphasis was directed to the study of:
• The stress distribution in infinite elastic media,
• Modelling of fully grouted bolt profiles,
• Stress distribution in the resin, and
• Mohr-coulomb failure study along the plane of weakness.
The important outcome of the study was to show that there
is another way to examine resin failure for different pro-
file configurations that can be compared with the laboratory
tests and numerical modelling. The ultimate aim and objective
of the theoretical predictions is to evaluate and complement
the experimental studies, enabling research to the best profile
geometries and to reach the optimum load transfer between
the bolt and the surrounding strata.
Figure 1. Pull out tests of bolts with various profile configuration Figure 2. A schematic drawing of a single spacing between two bolt
and spacing. profiles showing parameters used in calculations.
577

Layout assessment of underground tunnel for the stability of gangway in non-coal mine
Sung O. Choi
Department of Energy & Resources Engineering, Kangwon National University, South Korea
Keywords: mining, mine design, numerical modelling, rock mass classification
A lot of non-coal mines in Korea are planning to change first stage of its development. Furthermore, the gangway in
their mining method from the open-pit system to the under- non-coal mine have become bigger and more irregular, so the
ground system, not only to reduce expenses for environmental stability of gangway have become poor with the stress concen-
services but also to resolve an economic problem by low tration on the corner of safety pillar and the shear displacement
productivity. In underground mine operation, however, the along the rock blocks in safety pillar.
essential mechanical issues lie in keeping the stability of gang- In this study, the parametric analysis has been performed
way, and therefore an optimal design of gangway is most for the optimal design of safety pillar in underground non-coal
important to secure a mine stability as well as its productivity. mine. For this purpose, the theoretical solution by Obert and
For the optimal design of gangway space, the rock mass Duvall (1967) has been evaluated numerically, and the numeri-
classification method for mine has been adopted in this study. cal analysis with changing the width and height of safety pillar
It was modified from the conventional rock mass classifica- has been done to consider the correlation of gangway stability
tion method, considering the stand-up time and the equivalent with the space of safety pillar. Through the numerical anal-
support ratio in mine gangway. Because unlike most trans- ysis, the optimal space of safety pillar has been decided to
portation tunnels, mine gangway should reflect its direction minimize the stress concentration phenomena at the corner of
change, duration time, number of open space, dimension gangway.
change, and so on. The rock mass classification for mine can Figure 1 shows the numerical model used in this study. To
be suggested by RMR and Q , which are the simplified forms compare results with the theoretical solution by Obert and
of RMR and Q, respectively and namely elements related to Duvall (1967), the UDEC-2D program has been adopted for
rock mass only are considered neglecting the direction of this numerical model. The dimension of numerical model is
discontinuities and gangway. From the field investigation for 10 m × 10 m in case of single opening, and 20 m × 10 m in
two limestone quarries, the minimum width of unsupported multiple openings. In every numerical case, the diameter of
span was obtained by Eq. (1). opening is 5 m and the roller boundary condition has been
applied to the left and right boundary. Also the stress bound-
ary condition has been applied to the top and bottom boundary
with applying the 2 MPa compressive stress on the top bound-
where W is the minimum width of unsupported span in meter, ary. The numerical model was assumed to be elastic, and two
and RMR is basic RMR. joint sets were reflected in direction of 45◦ and 135◦ counter-
Also, the correlation of RMR with the stand-up time was clockwise from the horizontal line. The mechanical properties
investigated because the total stand-up time without any for rock mass as well as joints are summarized in Table 1.
supports can act as a negative factor on the stability of
gangway.
where Tp is the total stand-up time of gangway in day.

From Eq. (1) and (2), the general equation for unsupported
span of gangway can be obtained as follow;
Consequently, the stability of each gangway can be evalu-

ated using Eq. (3) and their supporting plan can be established
according to the purpose of gangway utility. However the
construction of support in mine gangway during the mine
operation is not easy from the limiting work space. As a result,
the best way to secure the gangway stability with increasing Figure 1. A schematic diagram of numerical model to compare with
the mine productivity is the optimal design of gangway in the the analytical solution.
578

Table 1. Mechanical properties used in numerical analysis.
Properties Value
Rock mass Density (kg/m3 ) 2,400

Bulk modulus (GPa) 8.0
Shear modulus (GPa) 5.0
Joint Cohesion (MPa) 1 × 104
Friction angle (◦ ) 35
Normal stiffness (GPa/m) 2.0
Shear stiffness (GPa/m) 2.0
Joint distribution Set 1 Dip (◦ ) 45
Space (m) 1.0
Set 2 Dip (◦ ) 135
Space (m) 1.0 Figure 3. Stress concentration pattern in safety pillar, with the array
of mine gangway.
than 1.5. This result remarks that the horizontal thickness of

safety pillar should be at least one and half times the width
of gangway. Another numerical analysis have been done with
changing the array of gangways and changing the width of
safety pillar, as shown in Figure 3. The lowest stress concen-
tration occurred in case 3 which means that the vertical and
horizontal width of safety pillar is one and half times gangway
width and the array of gangways are a zigzag pattern.
In conclusion, the optimal unsupported span of gangway
can be suggested reasonably through the rock mass clas-
sification method for mine in their first stage. In order to
achieve the optimal safety pillar design in an ongoing mine,
the numerical analysis has been made using the UDEC-2D.
Figure 2. Numerical result of stress concentration pattern in safety Comparison of UDEC analysis on a circular hole with the
pillar, with increasing the pillar width. theoretical solution has been performed to secure the accu-
racy of numerical analysis. In the following numerical analysis
Figure 2 shows one example of numerical analysis on the considering the various array pattern of gangway and the var-
gangway stability according to various thickness of horizontal ious width of safety pillar, the stability of safety pillar can be
safety pillar. As shown in this figure, the stress concentration assured when the width of safety pillar is at least 1.5 times
acting on the safety pillar, which denotes σ p /Pz , converges the width of gangway and the array pattern of gangway is
into a certain value when a value of Wp /Wo has become larger zigzag.
579

Longer-term performance of grouted bolts in different lithologies in the

Illinois coal basin mines
Yoginder P. Chugh, H. Gurley & W. Bell

Mining and Mineral Resources Engineering, Southern Illinois University Carbondale, IL, US
Keywords: bolts and anchors, coal mines, mining, mine design
This paper assesses anchorage performance of fully grouted not effective. These coupled with well-documented studies
bolts in different lithologies in two coal mines in Illinois on moisture sensitivity of coal seams associated shales led
using the Short Encapsulation Pull Test (SEPT). These tests to the current study whose goal is to assess the long-term
were done at two different times (0 and 120 days) after bolt performance of different bolts used as primary supports.
installation. Additional testing is planned after much longer A roof fall data base for the period January 1, 2005 through
periods (180 and 270 days). A borehole was also drilled May 15, 2010 was also developed for the two cooperating
in each test area about 2.6 m above the coal seam prior to mines. At Mine 1 most falls occurred in 4-way intersections.
bolts installation and examined with a camera to document Most falls had stand-up time of less than 60 days or greater
changes in de-laminations. Load cells monitored changes in than 120 days. Most falls with stand-up time of over 180 days
few selected bolts. In addition to the Grip Factor (GF), anchor- were supported by 1.8 m long fully-grouted, passive #6 rebar,
age stiffness data is also developed that provides interesting Grade 40 bolts. Lower number of falls for the same time
insights. Anchorage stiffness is defined as the slope of the period supported by 1.2 m long fully-grouted passive bolts
load-displacement curve at 50% of the bolt anchorage capac- was most likely due to the presence of substantial limestone
ity or “Anchorage Stiffness around 50% of the failure load or in the immediate roof. Mine 2 experienced most falls in 3- and
AS (50)”. GF values for all lithologies are similar to those 4-way intersections in isolated areas of transitional lithologies
reported in literature. They typically increased after time. and they occurred between 0 and 90 days or after 180 days.
Several findings here corroborate the observations in the field Very few falls occurred where limestone was present within
on roof falls. 1.8–2.5 m above the coal seam.
The principal author analyzed the US Mine Safety and The results of SEPT show that 1) # 6 rebar installation in
HealthAdministration (MSHA) reportable roof-falls data base 35 mm boreholes should be avoided; 2) Longer bolts seem
of 642 falls for Illinois mines for the period 2004 to 2008. to result in lower GF values, 3) AS values vary much more
The data accounted for about 100% of the underground coal than GF values; 4) Authors think that performance testing
production. Although normalization of data was not possible, of grouted bolts should be based both on GF and AS val-
the relevant observations and conclusions made were: 1) Fall ues because AS values represent the rock mass reinforcement
height is typically larger than bolt length by 0.3–0.6 m for potential while GF values represent the load carrying poten-
all types of bolts used, 2) About 35% of the falls had stand- tial, 5) Mine 2 primary roof support should be reassessed since
up time less than 90-days, and 40% less than 120-days, 3) the AS values for larger size bolts with longer grouted length
Fully-grouted resin bolts may be more effective than other are similar to those for smaller size bolt and smaller grouted
bolt types, 4) Longer fully-grouted bolts (beyond 1.8 m length) length at Mine 1. Additional data will be developed over the
should be avoided, and 5) Point-anchored bolts were generally next six months.
580

A new approach to simulate crush pillar behaviour in a tabular layout
M. du Plessis
Lonmin Platinum, South Africa
D.F. Malan
Department of Mining Engineering, University of Pretoria, South Africa
Keywords: mine design, back analysis, monitoring
1 INTRODUCTION been reported in South Africa. The displacement discontinu-

ity boundary element method (DDM) is a numerical technique
Crush pillar mining appears to be a method unique to South that is very well suited to the analysis of large-scale tabular
African mines with the pillar system being applied to shallow layout problems in which the plan projection of the mined area
and intermediate depth gold and platinum ore bodies. Min- may cover several square kilometres. The DDM technique has
ing methods using elastic non-yield pillars should in general not been applied to crush pillar layouts yet. When very small
only be considered for shallow operations. At greater depth, crush pillars are simulated, it requires very small element sizes
crush pillars become a more attractive option as it allows for a which make the simulation of a large-scale layout impracti-
greater extraction ratio. Ozbay and Roberts (1988) suggested cal. A DDM code developed by Napier (Napier & Malan,
that crush pillars should be implemented at a depth greater than 2007), which overcomes this problem by allowing the use of
400 m below surface. The crush pillars are usually designed triangular shaped elements, was therefore used in this study.
for a width to height ratio (w:h) of approximately 2, (Ryder & The TEXAN (Tabular EXcavation ANalyser) code
Jager, 2002). This ensures that the pillars fail as they are being (Napier & Malan, 2007) utilises a general revision of this
cut at the face. Once the pillar has failed in a stable manner, the approach in which triangular or quadrilateral element shapes
residual strength of the pillar contributes to the required panel are introduced in conjunction with higher order variations of
support by resisting the deadweight load to the height of the the displacement discontinuity shape functions. This facili-
uppermost parting on which separation is expected to occur. tates an accurate evaluation of detailed stress and displacement
In comparison to stable pillar layouts, failure of crush pillars components close to excavation surfaces and allow for the
is in fact desired as long as it occurs in a controlled manner. assessment of tabular layouts which includes a large number
Pillar failure and the resulting load shedding should ideally of pillars.
be continuous to prevent accumulation of dangerous levels of It must be noted that while local failure and deformation
elastic strain energy. Occasionally these pillars fail violently mechanisms may be analysed using non-linear continuum
in the back areas, resulting in damaging seismicity. A bet- plasticity finite element or finite difference models, these
ter understanding of crush pillar behaviour and an appropriate methods are themselves difficult to apply uniformly and are
methodology to design these pillars is therefore required. This usually not suited to the large-scale three-dimensional analysis
study describes a numerical study of a crush pillar layout. of interacting pillar systems.
1.1 Historic use and design of crush pillars

2 MODELLING OF A CRUSH PILLAR LAYOUT
RPM (Rustenburg Section) was the first platinum group
to implement the use of crush pillars as early as 1974, Stope closure data for the simulated experimental area was
(Korf, 1978). On most mining operations, the design of the collected during September 2004. The area simulated was
crush pillars is based on trial and error. As the pillar strength therefore limited to the face positions as it appeared at that
is unknown, the pillar sizes are adjusted to obtain the cor- time. The pillar and mining outlines are approximated using
rect behaviour. During the past 4 decades, several parameters straight line polygons (Figure 1) to enable the area to be easily
affecting crush pillar behaviour were studied in an attempt discretised using triangular elements. Two constitutive mod-
to better understand and predict the behaviour of the pil- els were used for the reef material. The first run was for an
lars. Crush pillar layouts have, however, remained essentially elastic rock mass were the pillars were simulated as a “rigid”
unchanged over this period. material that did not allow any failure. For the second run, the
limit equilibrium constitutive model (Malan & Napier, 2006)
was used. This model allowed failure and load shedding of the
1.2 Numerical simulation of crush pillar behaviour
pillars.
To date, no numerical modeling of a mine-wide tabular layout, Closure and pillar stress profiles were compiled based on
which explicitly included a large number of crush pillars, has the modeling results. Pillars 15 to 19 are the pillars which were
581

Figure 3. Closure profile for the limit equilibrium model. Block 1
defines the mining block exposed by mining step 1.
less compared to a pillar in close proximity to the face. This

highlights a very important principle regarding crush pillar
behaviour, namely, that it should be designed to crush in the
face area. If the pillar does not crush and moves into the back
area, the change in stress becomes less and the pillar may
therefore never crush, especially if it is significantly oversized.
Alternatively, pillar failure in the back area may occur violently
Figure 1. Extent of the area digitised for simulation with the
as the stresses on the pillars are much higher and the loading
TEXAN code. environment has become much softer as the abutment is no
longer in close proximity. All of the pillars, when being formed
at the face, have an APS of around 400 MPa. This far exceeds
the pillar strength, implying that these pillars would have been
crushed at the face.
The panel closure for both models was analysed. Figure 3
indicates the closure profile for the limit equilibrium model.
The average closure (average of all the collocation points in
the entire mining region) for Step 0 was 48.3 mm. The rate of
closure decreases as the face advances.
At a distance of approximately 80 m from the face, the total
closure in a specific mining block approaches an asymptotic
value. The rigid model has the same closure profiles, but has
smaller values. Of interest is that the results obtained from
the TEXAN modelling compared well with the in-situ closure
measurements conducted during 2004.
Figure 2. Average pillar stress per mining step for the elastic model
with rigid pillars. Each mining step corresponds to a face advance of REFERENCES
approximately 20 m.
Korf, C.W. 1978. Stick and pillar support on union section,
cut for the five mining steps (P15 is cut in the first mining step). rustenburg platinum mines, Assoc. of Mine Managers of S.A.,
The other pillars were randomly chosen in the mining area. pp. 71–82.
The average pillar stress (APS) values in the elastic model Malan, D.F. & Napier, J.A.L. 2006. Practical application of the
with rigid pillars are very high. The purpose of this analysis TEXAN code to solve pillar design problems in tabular exca-
was to determine if pillar crushing would occur at the face and vations, South African National Institute of Rock Engineer-
how the pillar stress evolves as the mining faces move away. ing (SANIRE) Symposium, Facing the challenges. Rustenburg,
The pillar stress is influenced by both pillar dimension and pp. 55–74.
pillar location. Napier, J.A.L & Malan, D.F. 2007. The computational analysis of
shallow depth tabular mining problems, J. S.Afr. Inst. Min. Metall.,
Figure 2 illustrates the evolution of average pillar stress
vol. 107, pp. 725–742.
(APS) for each pillar after each mining step. The stress starts Ozbay, M.U., & Roberts, M.K.C. 1988. Yield pillars in stope sup-
off at a relatively low value when the pillar is still part of the port. Proceedings SANGORM Sympoium in Africa, Swaziland.
face abutment. This stress rapidly rises after the mining step pp. 317–326.
in which the pillar is formed. If the pillar is far away from Ryder, J.A. & Jager, A.J. 2002. A textbook on Rock Mechanics for
the face, the change in stress caused by a mining increment is Tabular Hard Rock mines. SIMRAC, Johannesburg.
582

Experiences with tunnel support in a South African platinum mine
L.J. Gardner
Mining Projects Department, Impala Platinum Limited, Rustenburg, South Africa
Keywords: case studies, rock support, tunneling
Tunneling in South African platinum mines has traditionally while reducing their exposure to unsupported rock by using
been conducted using hand-held drilling and blasting meth- areal coverage with the tendon support. While presenting a
ods, with tunnels being supported using tendons installed minimum of technical detail, this paper attempts to explain
on at regular intervals. Rocks can and do fall out between how Impala has continually optimized its tunnel support sys-
the individual tendons, often with fatal results. To reduce tem to reduce rock-related risk in a challenging environment.
levels of rock-related risk in line with a “Zero Harm” pol- Although this story is by no means unique, it provides an exam-
icy, Impala Platinum has adopted a two-pronged strategy for ple of how such changes should be ongoing, and the resulting
mining project tunnels. This includes limiting the number of improvements that can be brought about in both safety and
personnel in tunnels by the increased use of mechanization, productivity.
583

A new method on extraction of two thick seams simultaneously on extremely

difficult ground condition – a case study
C.N. Ghosh, Prashant, P.K. Mandal, John Loui P. & A. Sinha

Central Institute of Mining & Fuel Research, Dhanbad, India
Keywords: coal mines, case studies, field measurements, mine design, monitoring, numerical modelling
In Indian coal mines, blasting gallery method of extraction lower seam extraction leading the upper seam extraction by
has emerged as a very successful method of extraction for 30–60 m. With this technology the simultaneous extraction of
thick seams above 7 m. The paper deals with an innovative the two seams with upper seam using blasting gallery method
technology where two seams were extracted simultaneously and the lower seam by bord and pillar method of depillaring
with the upper seam by blasting gallery method.The two seams with stowing were tried at one panel.
exist with a parting of 9 m. The upper seam is 10 m thick and While extracting by this method the panels were instru-
the lower one is 3.5 m thick. The upper seam was extracted mented with different rock mechanic instrumentation like
with caving using blasting gallery method.The lower seam was multipoint borehole extensometer, load cell, convergence sta-
developed by bord and pillar method and it was experiencing tions etc. The instruments were monitored regularly and data
severe ground control problem due to the blasting and caving were analysed. The critical convergence value for predicting
effect of the upper seam. An attempt was made to extract the the roof fall of the blasting gallery panel could be determined
lower seam with stowing and extraction of upper seam by experimentally. All predictions related to the stability of the
blasting gallery method over the stowed pack of lower seam. upper seam and lower seam extraction could be made well in
But due to sub-surface subsidence cracks were generated at advance based on the instrument data.
the floor of the upper seam and it became extremely difficult The method has emerged as a very successful method for
to extract the upper seam using blasting gallery method. As a the mine and all nearby mines have accepted this method as
result lot of coal were lost. the sole method for high percentage of extraction in similar
Then an innovative technology of simultaneous extraction condition.
of both the seams were thought of. A thorough numerical The paper deals in detail the method used, the detailed anal-
modeling analysis using FLAC 3D software were attempted. ysis of data obtained from numerical modeling studies and the
It has been observed that the extraction is possible with the analysis of data obtained from rock mechanic instrumentation.
584

Interpretation of pull-out tests on instrumented rock bolts by

means of 3D numerical modelling
R. Giot, C. Auvray & A. Giraud

LAEGO-ENSG, Vandœuvre-lès-Nancy, France
B. Gatmiri & A. Noiret

ANDRA, Châtenay-Malabry Cedex, France
Keywords: back analysis, bolts and anchors, field measurements, lab testing, numerical modelling, rock support
The study deals with the interpretation of in situ pull-out tests argillite/grout and grout/bolt interfaces. In this first attempt,
on instrumented grouted rock bolts, which took place at the the argillite was modelled with a simplifying Mohr-Coulomb
ANDRA underground research laboratory of Meuse/Haute- constitutive law. The elastic behaviour of the interfaces is char-
Marne (France). It aims at characterizing the behaviour of a acterized by a normal and a shear stiffnesses, while the plastic
single rock bolt in argillite, taking into account both the differ- behaviour, of Coulomb-type, is characterized by a cohesion,
ent materials (bolt, argillite, grout) and the different interfaces a friction angle and a dilatancy angle. Both the grout and
(argillite/grout and grout/bolt). 6 tests were carried out in situ, the bolt were considered as elastic materials, on the basis of
on two types of bolts: 3 tests on HA-25 rock bolts, and 3 the lab tests on grout. The whole sequence of excavation of
tests on CT-Bolt rock bolts. For each kind of rock bolt, one the gallery, then drilling, installation and grouting of the bolt,
short term and two long term pull-out tests were performed. and of the pull-out test is taken into account in the numerical
The paper focuses on the interpretation of the long term tests. modelling.
During these tests, the end of the bolt is pulled with different Even if the orders of magnitude of experimentally mea-
successive load stages, each load being maintained constant sured and calculated displacements and strains are similar, the
during various times. Two different loading paths were con- numerical modelling did not allow to carry out a precise quan-
sidered in this work. During the tests, the displacement of the titative interpretation of the in situ tests so far, that is to say that
head of the rock bolt is classically measured. In addition, in the parameters set that best fits the experimental curves was
the tests considered in this work, the bolts were instrumented not identified until now. Nevertheless, these numerical mod-
with axial strain gauges, fixed to the rock bolts every 15 cm, ellings allow a qualitative interpretation of the tests, enabling
so that the axial strains are recorded during the whole test (as the understanding of a part of the experimental observations.
well as before the test) along the grout/bolt interface. In addi- The elastic properties of the grout/bolt interface seem to be
tion to the in situ pull-out tests, laboratory tests were carried of fundamental importance. The quantitative analysis should
out in order to characterize the mechanical behaviour of the then focus on these parameters. The quality of the grout seal-
grout. ing seems to be heterogeneous, with a rather good quality in
The analytical models developed by several authors (Cai the far end of the drilling (from where the grout is injected)
et al., 2004; Farmer, 1975; Li and Stillborg, 1999; Ren et al., and a rather poor quality at the head of the drilling, maybe due
2010) were used to try to interpret the in situ tests but did not to a more intensive fracturation due to the development of the
allow to explain the experimental results, due to several simpli- EDZ at the wall of the gallery.
fication hypothesis. So, numerical modelling was considered From these results, the future developments should focus on
to interpret the tests, with the finite differences code FLAC3D the numerical modelling, and particularly on the calibration of
(Itasca). We made use of three-dimensional modelling in order the parameters of the grout/bolt interface. Moreover, labora-
to take into account all the possible geometrical configura- tory tests should be performed to improve the characterisation
tions (respective orientation of the transverse isotropy planes of the two interfaces, which could lead to the development of
and the principal stresses, orientations of the bolts and the a more suitable constitutive law. In parallel, a more appropri-
strain gauges). The numerical modelling takes into account ate constitutive law for the argillite will be considered, on the
suitable constitutive laws for the bolt (steel), the grout (charac- basis of the knowledge acquired on this material from other
terized through the lab tests) and the argillite, as well as for the numerous studies on it, among others to account for the EDZ.
585

Estimation of support requirement for large diameter ventilation shaft at

Chuquicamata underground mine in Chile
E. Hormazabal & C. Soto

SRK Consulting, Santiago, Chile
A. Russo
CODELCO (National Copper Corporation), Santiago, Chile
C. Carranza-Torres
Department of Civil Engineering, University of Minnesota, Duluth Campus, MN, US
Keywords: mine design, rock support, stability analysis, numerical modeling
The Chuquicamata Underground Mine project is the under-

ground extension of the existing Chuquicamata open pit mine
(one of the largest open pit mines in the world). The mining
project, which is located in the Atacama Desert in northern
Chile (Figure 1), contemplates using the method of block
caving with macro-blocks caving option to mine copper ore.
The project is to become as well one of the largest underground
caving mining operations in the world.
VP-CODELCO (Vice-President Office of the National
Copper Corporation of Chile) is finishing a pre-feasibility
engineering evaluation of the project, which considers the con-
struction and operation of at least two macro-block mining
units to be operated independently from each other.
Among the critically important elements of the permanent
mining infrastructure to be designed and constructed in a first
phase, are two large air ventilation shafts. These ventilation
shafts are required to have an internal diameter of 11 meters
and a depth of approximately 970 meters. The ventilation
shafts are also required to have a permanent liner that will
guarantee the safe and continued operation of the shafts for a
period of 50 years or more.
The objective of this paper is to present general aspects of
the design of one of the shafts, including the gathering and
interpretation of geotechnical data from site investigations,
and the use of empirical, analytical and numerical methods to
determine the appropriate temporary and permanent support
to be considered for the construction and operation of the shaft.
Figure 1. Chuquicamata mine location in relation to Antofagasta

and Calama cities in northern Chile.
586

Load distribution in inner anchor segment of prestressed cable with time
F. Jing, W.S. Li, C.H. Chen & Y.K. Liu

Key Laboratory of Geotechnical Mechanics and Engineering of Ministry of Water Resources,
Keywords: rock mass; rock support; bolts and anchors; case studies; monitoring
1 INTRODUCTION Through installing strain meters at inner anchor segment

and anchor load cells at porthole of the prestressed cable and
Because prestressed cables can fully take advantage of the based on the in-site pulling test and monitoring of prestressed
self-supporting ability of rock mass and the self-strength of cable after locked, the deformation and load distribution of
steel, adjust the self-strength of rock mass, reduce the material the test cable during pulling test, the variety law of effective
consumption and the construction cost estimate, and can also prestress of cables and the load distribution and variety law of
guarantee the safe and stability of engineering and construc- inner anchor segment of the prestressed cable with time after
tion, it has been widely applied in almost geotechnicals(Cheng locked, and the relationship between the load distribution vari-
et al., 1994). But because of the high cost and difficulty of ety law of inner anchor segment and the change law of effective
in-site experiment for the prestressed cables, the design and prestress of the prestressed cable are studied systematically.
calculation analysis methods about the prestressed cables now
mainly borrows ideas from the research findings of bolts, and
there only have very minute quantity in-site experiments study 3 TEST RESULT AND ANALYSIS
for prestressed cables, especially for the load distribution vari-
ety law of inner-anchor segment with time and its relationship Based on the test result, it is found that stress concentra-
with the cable effective prestress are almost seldom studied. tion of inner-anchor segment of classic concentrated tension
So performing above in-site experiments can study the vari- prestressed cable is very serious, the load distribution curve
ety laws of effective prestress of cables, the load distribution of inner anchor segment shows negative exponent shape and
and variety law of inner anchor segment of prestressed cable extremely nonlinearity, especially the forepart of inner anchor
with time, and the relationship between the load distribution segment from 0 to 3 m almost bears about 90 percent tension
variety law of inner anchor segment and the effective prestress loading of the test cable, which shows the test anchor cable
variety law of cable, which are important for the long term sta- has a certain emergency capacity length.
bility analysis of anchoring engineering and improving design According to the effective prestress variety characteristic
method of the prestressed cables. of the test cable after locked from more than 75 days moni-
The fall of rapid flood passage of one large hydropower toring, the change process can be divided into three stages,
station, located at Hubei province China, is more than 170 m namely speedy decline stage, slow decline stage and rela-
high, the wave wall is located along base of landslide and its tively stable stage (Zhang et al. 2002, Tang & Li 2005). The
maximal depth among 23 and 48 m. The rock mass of slope first stage is mainly influenced by wirerope relaxation, outer
are mainly sandrock and shale and the strength of rock mass anchorpier displacement, inner-anchor segment and rock mass
is very low, the stability of the slope and wave wall is very displacement etc. During the second stage, the effect of
low, and 2000 kN grade systematic prestressed cables must anchor wirerope relaxation and time-dependent deformation
be adopted for the stability of the landslide and wave wall. on anchor prestress becomes smaller, mainly influenced by
Because of the low strength of rock mass, the bearing capacity displacement of anchorage rock mass and inner-anchor seg-
of the rock mass is relative not enough for the 2000 kN grade ment as well as by environmental temperature and atmospheric
prestressed cables, and in-site pulling test and monitoring of water. The third stage is mainly influenced by underground
prestressed cables should be carried out. water, temperature variation, wirerope relaxation, the stress
adjustment of rock mass anchored and anchor cable, the
reducing degree is rather small compared with that of the
2 TEST PROGRAMME second stage.
According to the monitoring data of the load distribution
Test contents mainly includes: displacement and the rules change of inner-anchor segment among early stage after the
of load distribution of inner-anchor segment during anchor cable locked, it is obvious that the stress of inner-anchor seg-
pulling, initial prestress and the changing rules of load dis- ment appears transference backwards and adjustment from
tribution of inner-anchor segment by monitoring after anchor initial point to end. But with continue decreasing of cable
locked and the relationship between load distribution vari- effective prestress, the stress transference backwards of cable
ety of inner-anchor and initial prestress changing after anchor inner-anchor segment from initial point to end point and the
locked and so on. stress of whole inner-anchor segment reduces, and at last the
587

change law of interior bond section and that of cable effective different tension loads, influenced by partial geological
prestress become similar. difference of inner-anchor segment and strain inconsis-
According to the stress variety characteristic of the test tency etc., axial stress variety doesn’t amplify slowly as the
cable inner-anchor segment from monitoring, the axial stress same function mode, but is a complicated process.
changing of inner-anchor segment after cable locked can be 3. During anchor pulling of experimental anchor, the maxi-
divided into three stages which are similar to that of anchor mum transference depth of inner-anchor segment is 7.5 m,
prestress changing. The first stage is speedy changing stage, stress concentration is serious from 0 to 3 m which bearing
and only lasts several days, and the whole trend of axial stress about 90% of effective loads, and it shows anchor cable has
is quickly declining; the second stage is slowly adjusting stage, a certain emergency capacity length.
the changing speed of axial stress is smaller than that of the 4. The rules of prestress variety after cable locked can be
first stage, but there are signs of stress transferring backwards divided into three stages: The first stage is speedy decline
which is closer to the initial inner-anchor segment; the third stage; the second stage is slow decline stage; the third stage
stage is relatively stable stage, which shows the whole trend is relatively stable stage.
of axial stress at different depth is slowly declining, but the 5. The axial stress changing of inner-anchor segment after
changing speed is already obviously smaller than that of the cable locked can be divided into three stages. The first
first two stages. stage is speedy changing stage, and only lasts several days,
After cable is locked, effective prestress and load variety of and the whole trend of axial stress is quickly declining; the
inner-anchor segment all go through three similar stages, but second stage is slow adjusting stage, the changing speed
the respective durations of every stage is different. It shows the of axial stress is smaller than that of the first stage, but
loss of initial prestress is influenced synthetically by wirerope there are signs of stress transferring backwards which is
relaxation, outer anchor pier deformation, rock deformation closer to the initial inner-anchor segment; the third stage is
and inner-anchor segment deformation etc. According to both relatively stable stage.
long-term changing relationship between effective prestress 6. Anchor relaxation and rock creep is main factor of anchor
of anchor and load of inner-anchor segment, both influence prestress loss, based on anchor prestressed prediction
each other. model of rock creep and anchor relaxation coupling the
prestress changing can be fitted better.
4 THE ANALYSIS OF ANCHOR LONG-TERM The above conclusions drawn in this paper can supply refer-
INFECTIVE PRESTRESS ence for the stability study and reinforcing treatment of other
similar engineering.
Many factors influencing prestress loss are studied systemat-
ically. Anchor relaxation and rock creep are main factors of
anchor prestress loss, based on anchor prestressed prediction ACKNOWLEDGEMENTS
model of generalized Kelvin rock creep model and anchor
relaxation coupling the prestress changing can be fitted bet- The work presented here was supported by public welfare
ter. This model consider the main influence of anchor prestress trade research special aiding project of ministry of water
changing, the fit of middle and final prestress is more accurate resources of China (201001009).
as well as at last effective prestress can be estimated.
REFERENCES
5 CONCLUSION
Cheng Liangkui, Han Jun & Zhang Peiwen. 2008. Long-term per-
1. By measuring the stretch of wirerope during anchor formance and safety assessment of anchorage in geotechnical
pulling, the deformation condition of inner-anchor segment engineering. Chinese Journal of Rock Mechanics and Engineering,
27(5): 865–872.
is directly understood, which is significant for understand- Tang Ping & Li Duanyou. 2005. Monitoring and Research of Pre-
ing the characters of anchor bearing capacity during anchor stressed Anchoring on Rock Slopes of TGP Lock. Journal of
pulling. Yangtze River Scientific Research Institute, 22(6): 34–40.
2. The whole load distribution of inner-anchor segment is Zhang Faming, Liu Ning & Chen Zuyu. 2002. Analysis of factors
mode of negative exponential function and is highly non- affected on load losses of high capacity and long rock anchors.
linear, especially in the initial inner-anchor segment. Under Rock and Soil Mechanics, 23(2): l94–197.
588

Reinforcement technique and its application in complicated roadways

in underground coal mines
H.P. Kang, J.H. Wang & J. Lin

Coal Mining and Design Branch, China Coal Research Institute, Beijing, China
Keywords: coal mines, problem rocks, bolts and anchors, case studies, stability analysis
1 INTRODUCTION theory has been proposed. The main functions of bolting are
recognized as controlling the separation and slippage of sur-
Most coal mines in China operate by underground mining, rounding rock in bolted zone, crack opening and development
the geology and production conditions are complicated in associated with dilatant displacement, limiting the discontin-
many coal mines. The total length of roadways driven per uous displacement, and keeping the integrity and continuity
year amounts to 10000 km in Chinese coal mines, and 80% of of surrounding rock, making it in the compressive state, and
those are driven in coal seams. reducing the strength loss of rock mass. The pretension on
With increasing mining depth and extending mining area, bolts and its spreading to surrounding rock play a key role, and
the proportion of roadways with complicated conditions surface protecting parts are very important for the spreading of
becomes higher and higher, which poses a tough challenge bolt pretension. There is a critical reinforcement stiffness for
to ground control, and makes greater and strict demands pretension bolting system. For complicated roadways, the high
on reinforcement techniques. Therefore, reliable reinforce- pretension and intensive bolting system should be applied to
ment techniques for roadways with complicated conditions control the deformation and damage of rock mass once and for
have a special significance for the safety and high efficient all, avoiding the secondary reinforcement and maintenance.
production of coal mines.
4 REINFORCEMENT TYPES FOR COMPLICATED
2 TYPES OF COMPLICATED ROADWAYS ROADWAYS
Based on the natural and physical properties of surrounding There are many reinforcement forms for complicated road-
rock, the stress states in it, and the production conditions, ways. In recent years, the high pretension and high strength
complicated roadways in Chinese coal mines can be divided reinforcement system with bolts and cables as basic parts and
into 8 types: sometime complemented by grouting, has been the first choice
for complicated roadways.
1. Roadways with soft and swelling rock;
2. Roadways with extremely fractured rock mass;
3. Roadways with high stresses in deep mines; 4.1 High pretension and high strength bolting techniques
4. Roadways with large section; A series of high strength threaded steel rods used for bolts
5. Roadways affected by violent mining activities; were developed (see Tab. 1) to meet the needs of the compli-
6. Roadways with coal roof or all around the coal seam; cated roadways. The structures and shapes of the rods were
7. Roadways driven along next gob; optimized to improve the anchorage ability of bolts; and spe-
8. Roadways retained along gob-side or in the gob area. cial steel materials were developed to reach the level of high
or extra high strength. In addition, a series of resin capsules,
W-shaped and M-shaped straps were also developed to form
3 SUPPORT AND REINFORCEMENT THEORIES the high strength resin bolting system.
FOR COMPLICATED ROADWAYS
4.2 High pretension and high strength cable bolts
With the fast development of high strength bolting tech-
niques, the rock bolting mechanism for complicated roadways High extension and intensive cables were developed specially
has been studied deeply and extensively. The nature of rock for complicated roadways. The cables are made from 1 × 19
bolting has been recognized as improving the mechanical per- strands of steel wire, which increase the broken strength
formance of surrounding rock and the stress state in it, and and extension rate of cables substantially. To match with the
controlling the displacement and damage of surrounding rock. strength of rock bolts, the diameter of cables is divided into
It has been realized that it is very important to bolt surrounding three types: φ18 mm, φ20 mm and φ22 mm respectively. The
rock immediately after excavation, exert enough pretension broken load of φ22 mm cables amounts up to 550 kN, and the
on bolts and increase the stiffness of bolted body. Based extension rate is about 7%, obviously higher than that of 1 × 7
on these new insights, high pretension and intensive bolting strands. The cable pretension can amount up to 300 kN.
589

Table 1. Mechanical performance of steel rods for bolts.
Nominal Yield Tension Extension

diameter strength strength rate
Type /mm /MPa /MPa /%
Q235 14∼20 235 380 25

BHRB335 16∼22 335 490 22
BHRB400 16∼22 400 570 22
BHRB500 16∼22 500 670 20
BHRB600 16∼25 600 800 18
4.3 Combined reinforcement by bolting and grouting

Combined reinforcement with bolting and grouting can pro- Figure 1. Bolting pattern for an ingate (unit: mm).
vide a good way to tackle complicated roadways with fractured
rock mass. A variety of bolts with function of grouting
was developed. For extremely fractured coal and rock mass, are analyzed. The bolting pattern and parameters, monitor-
combined reinforcement technique with drilling, bolting and ing data, and reinforcement performance are introduced. The
grouting was introduced and developed to tackle the problem roof-to-floor convergence was 281 mm, the side-to-side con-
of borehole caving on this geological condition. Combined vergence was 173 mm, the roof sag was 40 mm and the floor
resin and grouting anchorage technique for cables was also heave was 241 mm. These displacements were reduced respec-
developed. tively by 69.8%, 77.8%, 79.5%, 67.2% compared with those
from the original support pattern. The roof separation was only
5 CASES STUDIES 5% of its original value, the surrounding rock mass was in
a state of good integrity. The high pretension and intensive
5.1 Bolting for roadways in soft rock mass bolting system had effectively controlled the strong defor-
mations and provided an effective support approach for deep
The study and experiment of the high pretension and high roadways.
strength bolting system at the Hongmiao coal mine of the
Pingzhuang coal district, a typical soft rock coal district in 5.3 Combined reinforcement with grouting and bolting in
China, are introduced. The contents include the deformation soft and fractured chambers
and damage features of the soft rock, bolting patterns and
parameters, field practice and monitoring, and the evaluation Taking the chamber group on the underground station in the
of reinforcement performance. Tunliu coal mine, the Luan coal district, as an example, the
During heading period of the tailgate, the side-to-side con- deformation and damage characteristics of soft and broken
vergence was 79 mm, and the roof-to-floor convergence was surrounding rock are introduced. The synthetic reinforcement
281 mm. During the period of retreat mining, the side-to-side technique was adopted with grouting under high pressure at
convergence was 256 mm and the roof sag was 110 mm. As a first step, and then the high pretension and intensive bolting
whole, the surrounding rock mass was of integrity and stable, followed. The bolting pattern of the north ingate is shown in
and met the requirements of production safety. Figure 1.
After this reinforcement pattern was carried out, the ingate
5.2 Bolting for high stress entry with depth over 1000 m did not show evident displacements and was in the state of
stability all the while. The side-to-side convergence was below
The high pretension and intensive bolting techniques were 30 mm, and the floor heave was less than 30 mm. The synthetic
conducted in deep roadways in the Xinwen coal district, reinforcement measure effectively controlled the deformation
a typical deep coal district in China. The in-situ stresses, and damage of the soft and fractured chamber groups, and
the rock deformation and damage modes in deep roadways kept their stability in long term.
590

Determining the dimensions of combined supports in roadways
P. Konečný & R. Šňupárek

Institute of Geonics ASCR, Ostrava, Czech Republic
Keywords: coal mines, bolts and anchors, stability analysis
The predominant amount of European coal deposits exploited of lifetime of a mine working. The practical experience in
underground is extracted by the longwall method with con- the application of the combined supports leads to opinion that
trolled caving. Experience shows that roadways, which ensure even in in the conditions of disturbed rock mass roof bolts play
all transport and ventilation in coalfaces, restrict both out- a positive role, which can be expressed in the methodology of
put and safety, as well as economy in coal production. The determining the dimensions for standing supports.
problem is urgent especially in deep mining. The contribution The suggested and simple geo-mechanical model permit to
describes procedures for designing the supports and stabiliza- determine approximate dimensions of the combined support
tion measures using combined support in the already planned in a long mine working, so as the standing support is designed
roadways on the base of system developed in the Ostrava with regard to the favourable residual co-effect of roof
Karvina Coalfield (OKR) in Czech Republic. bolting.
The combined support in roadways combines roof bolting From the calculations carried out, it is apparent that the
and standing (conventional) supports. The calculation of a so called bonus of bolts with the application of the com-
load bearing capacity of roadway supports contains period of bined support – i.e. the reduction of the normal load of the
drivage and period of minig in the vicinity of roadways. The standing arch support, is 15–45% of the original load in OKR
load bearing capacity of supports during drivage is based on mine conditions. The most frequent values for load reduction,
the size of the natural arch, which comes into existence in the corresponding to the width of 5–6 m of a mine working, vary
vicinity of a roadway. between 20–30%.
The supports in roadways are dimensioned to the load The effect of roof bolting, i.e. the reduced requirements for
applied during the roadway’s life. In the first period of drivage bearing capacity for the standing support, rises with a reducing
of a roadway, the minimum bearing capacity of the supports in width of a mine working.
must correspond to the load of the loosened rock in its vicin- The greater efficiency of roof bolting is shown in condi-
ity or, as the case may be, to a portion of this load, which the tions of firm overburden, which is usually associated with the
support must transfer with respect to a certain coherence of manner of disturbance of rock mass – the firmer rock mass
the loosened rock. The input data for the calculation method is is disturbed in larger clusters, which can be better stabilized
deduced from external measurements of convergence and dis- with roof bolting.
placement in roadways. The loading of supports during mining Monitoring, together with the behaviour of the combined
comes out from a stress wave in rock mass in the forefront of support, now more frequently installed in OKR mines, primar-
the coalface and from the caving area of the mined out panel. ily in connection with the dual application of roadways, will
The great advantage of the application of the combined enable further verification and obtaining more precise results
supports is the smaller convergence in the important period from the theoretical solution presented.
591

Leachate characteristic of sulphidic coal mine waste rock

due to various water pouring interval
G.J. Kusuma, K. Hiroto, H. Shimada, T. Sasaoka & Kikuo Matsui

Department of Earth Resources Engineering, Faculty of Engineering, Kyushu University, Fukuoka, Japan
R.S. Gautama & B. Sulistianto

Department of Mining Engineering, Institute Technology Bandung, Bandung, Indonesia
Keywords: coal mines, risks and hazards, weak rock, problem rocks
Acid Mine Drainage (AMD) generation from sulfide min- The XRD analysis conclude that the samples consist of clay
eral containing rock constitutes a significant environmental mineral such illite and kaolinite as well as secondary mineral
problem in coal, and poly-metallic sulphide mine due to the such as ferrihydrite, jarosite (KFe3 (SO4 )2 (OH)6 ) and gypsum
generation of sulfate and iron salts and possibly mobilizing (CaSO4 · 2H2 O), while Pyrite (FeS2 ) and siderite (FeCO3 ) are
trace metals present in the rock. AMD generation rate were not detected. XRF analysis result shows that the sample has
depend on many factor, both of inherent stoichiometry of acid high sulfur content as much as 3.6% and low alkali content
and base mineral and external factor such as climate which which is concordant with the result of static geochemical test
make the AMD characteristic would be site specific. (paste pH, paste EC, Acid Neutralizing Capacity test, and also
Water plays an important role in acid mine drainage, other- Net Acid Generating test).
wise as reactant as well as contaminant carrier (Bussière et al. Five columns with diameter 50 mm and 150 mm in height
2004, Hollings et al. 2001). On non-durable slaking rock mate- were used in the experiment. Amount of 150 gram sample with
rial such as coal measure rock, the existence of water in pore certain grain size distribution (d10 = 0.2 mm, d50 = 0.85 mm)
space will control the oxygen diffusion, slaking, and further- was used. Leachate from the columns following flushing were
more the reaction rate, both of sulfide oxidation and alkali dis- collected and sampled which was then measured for pH, EC
solution, of mineral contained in rock (Gosselin et al. 2007). and Eh as well as major sulfate and metal content such as
At the surface layer of uncovered waste rock dump, rocks calcium, magnesium, sodium, potassium, iron, aluminum and
will interact with climate components involving rainfall, solar manganese.
light, temperature and humidity ambient that will control the As result of various water pouring interval, each column
weathering and water balance in the surface material. With has different water content fluctuation behavior. In general,
the assumption that the average temperature and humidity are maximum water content of each cycle in each column was
relatively same, the interval of rainfall will be the biggest decreased by time due to self compaction of rock sample in
influence that will determine the wetting and drying cycles regards with re-packing of the loose grain which can observed
of material, and furthermore will affect to the weathering rate by the increase of the bulk density.
and water balance of material. In relation with pouring interval, certain interval of water
This study is intended to investigate AMD release behavior pouring has resulting in lower pH of leachate water while
of sulfide mineral containing rock with different water content the EC has the opposite order. In addition, the reduction –
condition affected by different water pouring interval by using oxidation potential (Eh) were increase along the experiment
a column leach test in the laboratories. cycle for all sample meant the increase of the oxidation-
In this study, we used sample which taken from a coal reduction activity. Similar to the pH, sulfate and metal content
mine, PT Berau Coal, which is located in Berau Regency, (Al, Mn and Fe) were higher in the column with certain (one
East Kalimantan Province, Indonesia. Sample was identified days interval) pouring interval.
as partially weathered conditions since it was taken from the This experiment result suggests that acid generation is
surface of existing uncovered waste dump with more than one controlled by several factors that influence one another, both
year old. physically and geochemically.
592

Stresses and stability of a mine paste fill
C. Laín Huerta, P. Ramírez Oyanguren & R. Laín Huerta

Polytechnic University of Madrid, Spain
F. Carnero & C. Santana

Mina de Aguas Teñidas S.A.U., Spain
L. Alejano Monge
University of Vigo, Spain
Keywords: numerical modeling, stability analysis
The Aguas Teñidas mine is located in the municipality of exceeds 600kPa. The remaining properties obtained in the
Almonaster La Real, Huelva. laboratory are, on average, as follows:
The mining method uses large sublevel stopes and fan blast- • Elastic module, E = 2–30 MPa
ing. The stopes are 20 m wide, 30 m high and between 20 m • Poisson’s ratio, µ = 0.2–0.4
and 60 m long. The method used to exploit the area consists • Density, 2400–2860 kg/m3
in extracting the primary stopes with a height of 30 m, and • Permeability, 2.38–4.83−6 cm/s
then extract the secondary stopes, of the same size, located
between the primary ones, after these have been filled with A FLAC 3D software programme was used for the numeri-
paste. Paste is a mix of mill tailings and cement that can be cal modelling. The model was constructed using paralepipedic
considered as non-segregating slurry which means that it has blocks of the same length and width, both variable, and a con-
negligible excess water when stationary and remains essen- stant height of 2.5 m. The total number of blocks is 36864.
tially as a homogeneous single phase product (Henderson, A. Two primary stopes have been successively excavated and
et al., 2005). filled into the model, each of which has been exploited in
First of all the ore of the first primary stope is fun blasted three sections. There is a secondary stope, located between
and extracted, afterwards the stope is filled with paste; when the primary ones, which has also been exploited in three sec-
the stope is too long so that the stability of its back can not tions. The height of the primary and secondary stopes is 30 m.
be guaranteed, it is convenient to extract and fill it in several The width varies between 12 m and 28 m and the length varies
sections of shorter length; every section must be filled before between 36 m and 84 m.
the next one is blasted. The other primary stope is extracted in The model has been firmly secured to its base, while the
the same way. In the primary stopes the walls are in ore and sides have only been fastened in the direction perpendicular to
in the secondary ones are in paste. the face. Before emptying and filling the stopes in the model,
Paste fill is a mix of mine tailings, with between 78 and the prior stresses were introduced and were assumed to be
85% solids by mass, and 3 to 7% cement. It uses some water hydrostatic, i.e. equal in vertical and horizontal directions.
for cement hydration and retains the rest within its matrix, so Extraction and filling of the stopes was carried out in 18
that it will not create any bleed water. phases as defined below:
The stability analysis of paste fills has been undertaken Phase 1: Extraction of first primary section.
by several authors. Mitchel et al. (1982, 1988) developed a Phase 2: Filling of first primary section.
three dimensional limit equilibrium solution for the stability Phase 3: Extraction of second primary section.
of exposed vertical face. Winch (1999) proposed an analytical Phase 4: Filling of second primary section.
solution to the total vertical stress within a three dimensional Phase 5: Extraction of third primary section.
back fill mass. Bloss (1992, 2001) used TVIS modelling pack- Phase 6: Filling of third primary section.
age to study the underground stability of cemented hydraulic Phases 7 to 12: Extraction and filling of primary sections, 4,
fills at Mount Isa mines. Numerical modelling with FLAC3D 5 and 6.
was used by Rankine et al. (2001, 2003, 2004) to examine the Phases 13 to 18: Extraction and filling of secondary sections,
stability of back filed stopes. 1 to 3.
Trials have been carried out using paste samples made with
The properties allocated to the paste are as follows:
flotation tailings and quantities of cement, usually varying
between 3% and 7%, with a slump between 7 and 10 . • Cohesion, C = 300 kPa (different cohesions have also been
Uniaxial compressive strength of the paste, after 28 days, applied)
593

• Angle of friction, φ = 0 an arc effect in the paste, with the resultant reduction in the
• Tensile strength, 80 kPa vertical stresses. In the second example this is not the case.
• Elastic module, E = 15000 kPa In the numerical simulation, the paste strength, i.e. the cohe-
• Poisson’s ratio, µ = 0.3 sion, required for the wall to be stable was obtained by trial
• Density 2800 kg/m3 and error. The cohesion of the paste was varied in the model
A zero friction angle has been established to the paste as until the fill broke. The following results were obtained:
the stability study presented below is in the short term, i.e. • For the primary stope of 20 m × 30 m, c = 140 kPa
assuming that the water doesn’t have time to drain out of the • For the secondary stope of 20 m × 30 m, c = 230 kPa
paste wall which is exposed after extracting the mineral from
the stope. According to the formula established by Mitchell, R. J. et al.
As mentioned earlier, the 18 phases required to extract and (1982):
fill the primary and secondary stopes have been simulated with
the model. The vertical stresses on the paste faces exposed
during the emptying of the third section of the primary stope where:
and the third section of the secondary stope, i.e. in phases 5 B: stope width = 20 m
and 17 of the mining operations simulated with the model, H: stope height = 30 m
were recorded. When the third primary section was emptied, γ: specific weight of paste = 27.5 kN/m3
the vertical stresses on the paste wall were recorded in three c: paste cohesion in kPa
vertical lines. In accordance with Mitchell et al. (1982) the wall would
Before starting to simulate the above-mentioned extraction be stable beyond cohesion of 165 kPa, a value bigger than the
and fill phases, a so-called “gravity pass” was carried out, to 140 kPa obtained with the numerical model for the primary
balance the prior stresses in the model. stope. Therefore, this formula can be used when referring to
On the whole, the behaviour of the model was elastic, i.e. the primary stopes. However, when calculating the stability of
stresses did not exceed the strength of the paste. The vertical paste walls in secondary stopes the use of this formula could
stresses in the paste face which was exposed when the third lead to risky situations, since, according to the model, a value
section of the first primary stopes was extracted (phase 5), for of 220 kPa is required for a stable outcome.
a width of 20 m. In the latter case, the classic formula (Terzaghi and Peck,
The stresses in a vertical line and another horizontal line 1967) shown below, which is the most appropriate for the two-
which pass through the centre of the paste wall which is dimensional analysis of the stability of saturated soil slopes in
exposed on removing the third secondary section. The ver- the short term, i.e. assuming that the paste has a zero friction
tical stresses in the central vertical line of the paste face are angle, should be used:
somewhat greater than the lithostatic load at the top part of the
wall and somewhat lower at the lower part of the wall. When
this stope is filled with paste, the stresses in what was the wall
exceed the lithostatic load in the majority of the wall, except From this formula, the cohesion required for a 30 m high
close to its base. slope to be stable is 214 kPa, in line with the 220 kPa obtained
The vertical stresses are almost equal along the width of the using numerical simulation.
wall and very close to the lithostatic load value. From this we It is worth to highlight that both formulae give similar
deduce that the arc effect is hardly ever produced in the 20 m results for stopes which are slightly wider than their height.
wide secondary stopes. From the formula established by Mitchel, R. J. et al. (1981),
We have also studied the variation in the vertical stresses in given a constant height, as the stope width decreases the
the paste wall of a primary and secondary stopes for different stability of the paste wall increases. This hypothesis has been
stope widths: 12, 16, 20, 24 and 28 m, with a standard height analysed using the numerical model described and has been
of 30 m. confirmed to be valid.
The same study has been carried out in the paste wall The numerical modelling carried out has demonstrated that
exposed when the third secondary section is excavated. For an arc effect is produced in the paste fills of primary stopes,
stopes which are wider than 24 m, that is, with a width/height i.e. those with both lateral walls in mineral or rock, due to
ratio bigger than 0.8, there is hardly any arc effect in the the shear strength which is mobilised in the contact between
secondary stopes. the lateral walls and the paste. This effect relieves the vertical
The FLAC 3D software programme and the above models stresses and increases the stability of the exposed paste wall.
were used to carry out stability calculations of the paste walls From these investigation, it is deduced that even for stopes
of the primary and secondary stopes measuring 30 m high and with a width/height ratio close to 1, there is an arc effect in
20 m wide. These calculations were repeated for 16 and 12 m the primary stopes.
wide stopes, in order to determine the influence of stope width In secondary stopes, that is, stopes in which there is paste
on stability. on both lateral walls, the arc effect begins to appear for
The calculations were carried out to determine the paste width/height ratios less than 0.8 in 30 m high stopes.
strength at which there is a break in a wall of the dimensions The stability of the paste walls in primary stopes can be
shown: 20 m wide by 30 m high, in a primary and in a sec- calculated using the formula established by Mitchel, R. J. et al.
ondary stope. The break is assumed to be short term, that is, for (1982). However, to study the stability of the paste walls in
a zero friction angle. The difference between a primary and a the secondary stopes, it is more appropriate to use the classic
secondary stope with respect to the stability of the paste wall is stability formula for two-dimensional vertical soil slopes in
that in the primary stope the paste fill is located between rock the short term (Terzaghi and Peck, 1967). However, in stopes
or mineral bodies, and in the secondary stope it is between with a width/height ratio of less than 0.5, this formula may
paste bodies. As explained above, in the first case there is give somewhat conservative results.
594

Establishing mining limits for partial undermining of surface infrastructure

at Union Mine

Anglo American Platinum, Johannesburg, South Africa
Keywords: stability analysis, mine design, mining, numerical modeling, surface protection, undermining
1 INTRODUCTION Pipes. For this study, the categories for buildings will be con-
sidered. Stacey and Bakker (1992) further suggested that of
1.1 Background the available parameters such as vertical subsidence, horizon-
tal strain, tilt, radius of curvature and differential settlement,
Union Mine is located in the Limpopo Province of South
only two (horizontal strain and tilt) should be adopted for
Africa on the western limb of the Bushveld Complex. Two
subsidence guidelines. The reasons that Stacey and Bakker
reefs are currently being extracted, namely the Merensky reef
(1992) suggests for the choice of these two parameters is
and the UGII reef. The current mining depth extends from
that they are both readily measured, both can easily be cal-
surface to approximately 1.5 km below surface. The plan is to
culated from mining simulation stress analysis programs and
extend mining to a depth of approximately 2 km below surface.
that they have been commonly used. The horizontal strain and
A portion of the UG2 reef has been left un-mined at the
tilt criteria limits suggested by Stacey and Bakker (1992) were
Declines Section, Union Mine, Anglo American Platinum,
used for this analysis. A conservative approach was adopted
possibly because of the overlying surface structures includ-
for this study by using the Building Category 1 superficial
ing buildings (closest being an engineering workshop) and
damage limits.
railway tracks. Union Mine is planning to mine the UG2 reef
in the vicinity of the existing surface infrastructure.
2.2 Pillar design criteria
1.2 Objectives The Factor of Safety (FOS) approach for pillar design is well
The primary objectives of the study are to ensure that the stabil- established for South African hard and soft rock mines and is
ity of the underground excavations and surface infrastructure not repeated in this report. For normal operating conditions
is maintained and in so doing ensure the safety and health of (i.e. no protection of surface infrastructure required) it is typ-
workers and to satisfy the requirements of the aforementioned ical for mines to use a FOS = 1.6 for pillar design. When
legislation by: designing pillars in workings where subsidence control is a
consideration, higher FOS’s are advocated since:
(i) Determining the optimum extraction depth below the
surface structures with due cognisance taken of: – The life of the surface infrastructure typically far exceeds
that of the underground work space
– damage to any of the surface infrastructure (workshop – The pillars are required as far as possible to be permanent
and railway tracks), and unchanging with the passage of time
– the ability of the systematic pillar system to maintain – The decay of pillar strength with time must be considered
its integrity for the length of period that subsidence – The influence of substandard mining controls such as
control is required, drilling and blasting on FOS’s must be considered
– permissible mining spans to ensure appropriate pillar
loading and tolerable surface settlement. A FOS of 2.5, suggested by Bakker (1992) for ‘buildings or
places where people congregate’ was used for this analysis.
(ii) Recommending procedures to be adopted prior to extrac-
tion which will include but not be limited to:
3 UNDERGROUND GEOTECHNICAL ASSESSMENT
– instrumentation and monitoring,
– pillar and panel design. The rock mass rating system (RMR1976 ) introduced by
Bieniawski (Bieniawski, 1976) and Norwegian Geotechnical
Institute’s Q-system (Barton et al., 1973) was applied to clas-
2 CRITERIA FOR ASSESSMENT
sify the rock mass conditions. No major geological structures
were mapped in this area. The RMR1976 rating for the rock
2.1 Structural damage criteria
mass conditions observed in the ASD was approximately 90,
Stacey and Bakker (1992) described deformation limits for which could be described as Very Good. The Q-rating for the
four main groups of surface structures applicable in the South rock mass conditions observed in the ASD was approximately
African mining context as Buildings, Roads, Railway lines and 180, which could be described as Extremely Good. The ASD
595

is classified as a life-of-mine excavation, for which an Exca- For the purpose of this analysis, this is not regarded as a
vation Support Ratio (ESR) of 1.6 was assumed. The ratio factor as:
of Span/ESR is then estimated at approximately 2. Apply-
1. There are no buildings directly above the UG2 reef where
ing, the Q-system Support Design Chart, spot bolting may be
the middling between surface and UG2 reef is greater than
sufficient for this excavation.
105 m.
2. The UG2 below a depth of 105 m has already been
4 NUMERICAL MODELLING extracted.
An additional 5 m wide pillar was recommended between
4.1 Structural damage analysis results the planned area and the old mined out area to the east.
Horizontal strains were obtained from the models by placing a
surface parallel line at a depth of 5 m below surface. The depth
6 CONCLUSIONS
of the line was chosen to avoid artificially induced surface
edge effects. A line along the central axis of pillars parallel to
The horizontal strain and tilt deformation limits are not
the UG2 seam was used to obtain average pillar stresses. The
exceeded for the workshop. Mining of the UG2 is permissible
results indicate the following:
under the following provisos:
• Unstable zone may extend up to approximately 8 m into the
– panel spans measured skin-to-skin from pillar to pillar must
hanging wall above the top panel just below the weathered
not exceed 30 m for mining depths between 40 m and 105 m
zone for panel spans of 30 m.
(this is less than the 33 m panel spans which is being used
• Neither the horizontal strains nor the tilt exceeds the damage
currently at Union Mine),
criterion when mining in the vicinity of the workshop.
– pillar dimensions must be 5 m × 10 m so that a FOS of at
least 2.5 is maintained. These pillar dimensions are con-
4.2 Pillar design analysis results sistent with the current mine standard for mining at these
The pillar strengths were calculated using the Hedley & Grant depths,
(1972) power law formula and the pillar loads using results – an instrumentation and monitoring programme to record,
from the numerical models. The exercise was done solely to validate and manage surface subsidence must be instituted
provide FOS’s as a function of mining span. For this assess- prior to commencement of mining.
ment of pillar strengths, a k-value equivalent to 30% of the
UCS of the rock containing the pillar was used. Pillar strength
ACKNOWLEDGEMENTS
is defined as:
The authors would like to thank Anglo American Platinum for
allowing us to publish this work.
Where h is the pillar height and weff the effective pillar width. REFERENCES
Assuming a UGII strength value of 75 MPa, a stoping width
of 1.5 m and a pillar width of 5 m, the pillar strength equate to Anon. 2000. UDEC User Manual. ITASCA Consulting Group, Inc.
∼41 MPa. The FOS, in turn is simply the ratio between pillar 1st Edition, Minnesota, USA.
strength and average pillar stress (APS). Bakker D. 1992. The undermining of surface structures and construc-
tion/erection over undermined ground. Symposium on construc-
tion over mined areas. Pretoria. South Africa. 1992.
5 SUMMARY Barton, N., Lien, R. & Lunde, J. 1974. Engineering classification of
rock masses for the design of tunnel support. Rock Mechanics. 6:
The analysis indicates that a FOS design limit of 2.5 for pil- 4: 189–236.
lars is achieved for mining depths down to a depth of 105 m
below surface, for a 5 m × 10 m pillar system and mining spans
of 30 m.
596

Recent developments in the monitoring of excavation stability

in South African mines
D.F. Malan
Department of Mining Engineering, University of Pretoria, South Africa
Subject: Monitoring in excavations
Keywords: monitoring, mining, case studies
1 INTRODUCTION
The objective of this paper is to highlight the importance of

using appropriate monitoring in mining projects. In contrast
to mining and rock engineering, routine monitoring is used
to a greater extent in many of the other engineering disci-
plines. In general, the rule applies that that if uncertainty
is associated with an engineering design, monitoring will be
used to reduce the risk associated with any failures. The ques-
tion should therefore be posed whether sufficient monitoring
of the rock mass behaviour is being conducted in mining
operations.
2 EXAMPLES OF HISTORIC MONITORING
Historic examples of monitoring in mines are useful as it illus- Figure 1. An area of a collapse in a platinum mine. Note the failed
trates that the early miners realised the value of measurements elongate support.
and that they were able to implement monitoring programmes
without the sophisticated electronic systems available today.
The earliest recorded example of in-stope monitoring in the 4 CASE STUDY ILLUSTRATING THE VALUE OF
South African mines was the use of “sag” meters at Crown MONITORING
Mines in 1927. The meters were originally manufactured from
timber. These were installed in all remnant areas. Systematic A case study is included in the paper to illustrate the value of
measurements of convergence (“sag”) were conducted at the monitoring programmes. A Platinum Mine in the Marikana
mine and the data was recorded on special forms. As described area of the Western Busheld Complex recently experienced
in the paper, other measurements of convergence were also a number of large panel collapses. The workings are rela-
conducted by various workers. tively shallow and current mining focuses on the UG2 reef
horizon. A number of monitoring sites were established at the
mine and important observations regarding the large collapses
were made. It should be noted that these observations are not
3 THE NEED FOR MONITORING applicable to small falls.
Large collapses are often preceded by time-dependent
Ryder & Jager (2002) summarises the need for rock engineer- deterioration of the hanging wall and an increase in the rate
ing monitoring as follows: of convergence. Broken elongate support (Figure 1) has tra-
“In geotechnical engineering work, there is inevitably a dis- ditionally served as early warning of the collapses, but these
crepancy between prediction and performance. …The instru- elongates only fail after approximately 30–50 mm of conver-
mentation and monitoring of real underground behaviour is, gence. The falls do not occur instantaneously, but appear to
therefore, an essential component of the rock engineering be preceded by a lengthy period of deterioration.
discipline.”
4.1 Panel convergence measurements
Even after many years of research, there are still a large num-
ber of unknowns in rock engineering. Some of the issues are Elongate deformation is not a very useful means of providing
described in the paper and these are pillar strength, under- early warning of the instabilities as the rate of convergence
ground de-rating of elongate strength and determination of preceding collapses at the mine was very small. Too much
stable spans. movement needs to occur before the support gave a clear
597

Figure 2. Low rate of convergence observed in stable sections of
the mine.
Figure 4. Example of a recent convergence-type telltale installed

in a stope panel.
Figure 3. An increase in the rate of convergence in areas that
became unstable. This area eventually collapsed. The Convergence Loggers used in the initial monitoring in
the case study described above were nevertheless considered
to be impractical and too expensive for mine-wide applica-
visual indication of instabilities. In order to precisely mea-
tion. A suitable instrument, the Convergence Telltale (termed
sure panel convergence on a continuous basis, Convergence
a closure robot in South Africa), was developed some time ago
Loggers were installed in various locations in this mine. It
and implemented in some mines in the industry. A recent ver-
soon became clear that the convergence measurements are
sion of this Convergence Telltale is illustrated in Figure 4. The
extremely useful to give early warning of the collapses, pro-
Telltale consists of two spring-loaded telescopic plastic tubes
vided the instruments are installed in the correct locations.
to hold the instrument in place and a central sleeve containing
This confirms the findings from previous case studies. Owing
the movement sensor and other electronics. Relative move-
to the relatively shallow nature of the excavations at the mine,
ment between the hanging wall and footwall is detected by
it was found that the rate of convergence is very low in the
the sensor and converted to a visible traffic light signal trans-
stable areas. This is illustrated by the graph in Figure 2.
mitted by two LED lights situated on either side of the sensor
Similar graphs were obtained for all the stable areas where
sleeve. The Convergence Telltale translates rock movement
convergence monitoring was conducted.
into green, amber or red LED ‘traffic light’ signals, which can
In comparison, measurable rates of convergence were
be easily seen and interpreted by panel workers. This imme-
recorded in unstable areas and these rates persisted for many
diate information enables the workers to be more aware of
days and even weeks before the large collapses occurred.
potential rock hazards which would not be seen or detected
Figure 3 illustrates convergence data from a panel which
by normal methods. A database of continuous convergence
eventually collapsed. Note that the rate of convergence has
information needs to be collected to serve as calibration of
increased significantly compared to the low rates in the stable
the telltales. A value commonly used in the South African
areas (Figure 2).
mines is 10 mm as the maximum alert level.
Based on these observations, the mine used convergence
instruments in the high risk areas and these were abandoned
if the rate exceeded a particular value. REFERENCES
Ryder, J.A. & Jager, A.J. 2002. A textbook on Rock Mechanics for
5 THE DEVELOPMENT OF CONVERGENCE tabular hard rock mines. SIMRAC, Johannesburg.
TELLTALES
Of the various types of in-stope instrumentation, routine

monitoring of panel convergence on a large scale is probably
the most viable. Compared to extensometer type telltales, it is
relatively easy to install monitoring equipment and no expen-
sive drilling is required. Moving the convergence instruments
forward as the panel faces move away can also be done with
relative ease.
598

The influence of shear stress and weak contacts on pillar behaviour
J.A. Maritz & D.F. Malan

Department of Mining Engineering, University of Pretoria, Pretoria, Gauteng, South Africa
Keywords: numerical modelling, mining design
1 INTRODUCTION
Although pillar behaviour has been studied for many decades

(e.g. Hedley & Grant, 1972; Ozbay et al., 1995; Martin &
Maybee, 2000), the methodologies used to design pillars are
nevertheless still imperfect and it requires additional research
(Malan, 2010). The difficulty of pillar design arises as pillar
failure occurs according to a number of complex local mech-
anisms imposed by the heterogeneity of the pillar geological
structure.
Regarding pillar design, the Hedley & Grant (1972) empir-
ical method of estimating pillar strength is commonly used in
South Africa.
2 THE EFFECT OF WEAK PARTINGS ON PILLAR Figure 1. Scaling facilitated by a weak contact at the pillar/hanging
STRENGHT wall contact.
The inherent dangers of using empirical design formulas for

4 EXCESS SHEAR STRESS ACTING ON A PILLAR
rock masses in which they were not originally designed for
is the possibility of large scale failure. Three case studies of
In the South African hard rock mines in the Bushveld Com-
recent mine collapses in South African is discussed in Malan
plex, a weak parting forming the contact between the pillar
(2010). The failure in all three cases was facilitated by the
and the hanging wall is frequently encountered (Figure 1). As
presence of weak partings (sometimes with a clay infilling)
the reefs typically dip at approximately 9◦ , some concern has
which substantially weaken the pillars. Figure 1 illustrates
been expressed that the shear stress acting on these pillars
such a weak parting at the pillar/hanging wall contact.
will cause a mobilization of these partings. This may cause
pillar failures at loads less than the expected failure load or
even cause regional failure as the hanging wall “ride” over
3 FACTOR OF SAFETY CALCULATIONS AND the pillars. This possible effect has not been quantified yet.
PILLAR STRESS The measure that will be used to investigate this effect in the
modeling below is the excess shear stress (ESS). The equa-
Traditionally in the South African mining industry, the factor tion for calculating the ESS on the hanging wall/pillar parting
of safety (FOS) is calculated as the ratio between the pillar plane is:
strength, as typically determined using the Hedley & Grant
empirical formula and the pillar load. The pillar load is calcu-
lated using either tributary area theory or numerical modeling.
Only stresses acting normal on the pillar (σn ) are considered where τmax are the simulated shear stress acting on the pillar,
when calculating factors of safety on the pillars. The inher- µ = tan φ (φ is the friction angle of the pillar hanging wall
ent danger when following this approach is that in dipping parting) and σn the normal “clamping” stress acting on the
stopes, a significant shear stress component may be acting on pillar.
the pillars and this may or may not weaken the pillars. When the ESS value is positive, shear failure on the parting
The effect of shear stresses on pillar strength needs to be will be possible. To simulate the shear stresses on the pillars,
further investigated to obtain an improved understanding of theTEXAN numerical modeling code (Napier & Malan, 2007)
pillar strength in complex stress environments. As a first step, was used.
a preliminary investigation of the shear stresses acting on A number of TEXAN runs (Table 1) were conducted to
the pillars in a simple bord and pillar layout is illustrated in determine the effect of the dip of the reef plane as well as
this paper. the depth below surface on the ESS on the pillar partings.
599

Table 1. List of models simulated with the associated parameters.
Model number Dip of reef plane (◦ ) Depth below surface (m)
Base (BC) 0 400

C1 0 600
C2 0 800
C3 30 400
C4 30 600
C5 30 800
C6 60 400
C7 60 600
C8 60 800
Figure 4. Pillar strength as a function of friction angle on the parting
at the hanging wall/pillar contact.
5 SIMULTING THE EFFECT OF A WEAK PARTING ON

PILLAR STRENGHT
Although the simulation above illustrates that for certain com-

binations of friction angles and dip of the reef, the parting
will not be subjected to shear slip, failure is still frequently
observed on these pillars and the parting appears to weaken
the pillar. This is even observed for very small reef dips. This
may be caused by different failure mechanisms. To illustrate
this, some inelastic numerical modeling results (FLAC) are
discussed in the paper. A general pillar model with partings
was built to investigate the effect of an inhomogeneous pillar
with weak interfaces. The presence of a weak internal had vir-
tually no effect, even if the friction angle of interface 1 was set
as low as 6◦ . In contrast, low friction angles on the hanging
wall contact had a powerful effect, reducing the peak strength
of the pillar significantly.
Figure 2. A portion of the pillar geometry simulated.
REFERENCES
Hedley, D.G.F. & Grant, F., 1972. Stope-and-pillar design for Elliot
Lake Uranium Mines. Bull. Can. Inst. Min. Metal., 65, pp 37–44.
Malan, D.F. 2010. Keynote Lecture – Pillar design in hard rock
mines: Can we do this with confidence? In: Hagan, P and Saydam,
S., (eds.) Proc. Second Ground Control in Mining Conference,
Sydney, November 2010, pp 15–30.
Martin, C.D. & Maybee, W.G., 2000.The strength of hard rock pillars,
Int. J. Rock Mech Min. Sci, vol. 37, pp 1239–1246.
Napier, J.A.L. & Malan, D.F., 2007. The computational analysis of
shallow depth tabular mining problems, J. S. Afr. Inst Min Metall.,
vol. 107, Nov 2007, pp 725–742.
Ozbay, M.U., Ryder, J.A. & Jager, A.J., 1995. The design of pillar
Figure 3. ESS values on a plane at constant 400 m depth as a systems as practiced in shallow hard-rock tabular mines in South
function of dip with a friction angle of 20◦ . Africa, J.S. Afr. Inst. Min. Metall., Jan/Feb 1995, pp 7–18.
A portion of the base layout is presented in the Figure 2. The

stresses of the pillars highlighted in red are presented in the
graph below.
The modeling results were used to calculated the ESS values
according to Equation 3. The magnitude of shear stress acting
on the pillars was determined with a set of models simulated
at various degrees of dip. Figure 3 illustrates the results.
600

Performance of polymer skin spray-on liner in coal mines
Jan Nemcik, Ian Porter & Ernest Baafi

Current development in roof bolting automation requires fast of integrity where the polymer was not damaged. The type of
and effective skin support of underground coal mine roadways. failure indicated a substantial toughness and strength of the
A strong and tough fibre reinforced polymeric thin spray-on TSL material.
liner (TSL) is emerging as a logical substitute for the old steel
mesh support system. This paper investigates the effectiveness
of a reinforced polymer skin liner in coal mines subject to high
horizontal stress. Investigations using numerical modelling
and laboratory experiments suggest that a polymer based TSL
can outperform steel mesh and significantly improve stability
of the strata skin between the bolts. Strength tests performed
on 5 mm thick polymer sheets indicate load bearing capacities
comparable to the heavy duty steel mesh currently used as a
skin support underground. Puncture tests and polymer adhe-
sion to the substrata indicate that the polymer TSL can provide
additional benefits to the skin support.
A puncture test was performed using a series of steel discs to
predict whether the standard steel bolt plate system that is cur-
rently used underground does not place excessive loads onto
the polymer TSL. Discs of various diameters were placed onto
a 5 mm thick fiber reinforced polymer sheet and loaded until
the polymer surface was compromised. The summarised test
results in Figure 1 show that a 5 mm thick reinforced polymer
sheet can withstand substantial loads, e.g. the typical bearing
capacity of the polymer when loaded using a 120 mm disc was
up to 3300 kN before permanent damage occurred. The plates,
which are commonly used in the underground bolting system,
yielded the polymer sheet at approximately 3300 kN a load Figure 1. Summary of the load bearing capacity tests using round
comparable to the 120 mm diameter steel disc load shown in discs and steel plates.
Figure 1.
In order to determine the ultimate strength of the reinforced
polymer a series of tests were performed to load the poly-
mer TSL sheets to failure as shown in Figure 2. During the
first test the ultimate load was not reached due to the com-
pressibility of the airbag that was used for loading, absorbing
the 100 mm loading stroke of the Avery loading machine. A
load of 68 kN at a deflection of 35.3 mm was reached without
any visible damage to the polymer sheet. On unloading, the
sheet deflection recovered 23.9 mm with 11.4 mm of perma-
nent deformation, indicating that the sheet was mostly within
the elastic region of deformation during the loading stage.
Another similar test reached 100 kN without polymer failure.
Since the polymer sheet did not fail, the loading method was
modified and two sheets were reloaded using a 150 mm steel
spherical seat underlined with a dense rubber mat (Figure 2) to
minimise the stress concentrations at the plate edge. The sam-
ple was loaded at 2 mm/minute until the polymer failed at a
load of 45 kN and an average deflection of 52 mm in both tests.
As expected the load at failure was lower due to the smaller
loading area that was applied to the sheet. The location of
failure was directly under the point of loading. The polymer
sheet failed in a non-brittle manner with a tear and no loss Figure 2. Loading of glass reinforced polymer sheet to failure.
601

Rock slope reinforcement in open cut mining using grouted rock pins
G. Ren & S. Kerris

School of Civil, Environmental and Chemical Engineering, RMIT University, Melbourne, Australia
Keywords: bolts and anchors, mining, problem rocks, mine design, rock failure
1 INTRODUCTION masses two symmetrical plastic hinges are formed; one above
and one below the slip surface. These hinges act to focus axial
Reinforcement of rock slopes is a common practice in both force in the bolt until tensile failure occurs in the bolt at some
civil and mining applications. The reinforcement methods point between the hinges. The failure is primarily axial with
employed are generally based on the fundamental mechan- little shear force evident. The other mode occurs in stiffer less
ics although the design goals may differ. These differences deformable rock and also softer weaker rock. In this mode,
are reflected largely by the required longevity of the design. axial and shear force act at the joint intersection. The bolt
Civil applications often require design lives of 50 to 100 years fails through shear. Research conducted on the effects of soil
whereas mining applications can be anywhere from a few nail inclinations on a reinforced soil mass suggests that certain
weeks to a decade or more. The difference in the designed inclination angles can be detrimental to the normal stress on
lifespan can greatly affect the chosen reinforcement scheme. the slip plane. This has direct relevance to rock masses as the
Current reinforcing schemes utilized in the mining industry mechanics involved are similar in application. The method of
heavily rely on empirical data and experience gathered by the reinforcement pin installation currently used at the Cadia Hill
engineers in charge of the design and as such do not always rep- Gold Mine results in pin inclinations with greater dip than the
resent the best economical and efficient solution. Research has intersecting discontinuity plane. With reference to the typi-
been conducted on a number of different factors that influence cal shear pin installations most encountered at the Cadia Hill
the performance and efficiency of rock slope reinforcement Gold Mine and the angles at which the reinforcing elements
using grouted rock anchoring system. Traditional empirical meet the discontinuity, the research has indicated that some of
approaches for rock reinforcement design are examined in these angles may lead to a reduction in strength of the shear
conjunction with case studies. plane due to the mobilization of compression forces within
the element. The compression forces act to reduce the normal
stress on the plane and it is through this action that the overall
2 DESIGN PRACTICE AND ANALYSIS resisting force of the system is reduced. It is generally agreed
that the reinforcement pin increases the strength of the system
There are two types of common methods for slope rein- through two primary mechanisms. These mechanisms are the
forcement: one involves the use of active grouted anchors dowel effect and the friction effect. This increase in strength
installed into the face and pre-tensioned and the other utilizes is dependent on rock deformability and surrounding material.
un-tensioned steel bolts or shear pins. Existing knowledge on The friction effect on the other hand is the contribution of the
the consequences of rock bolt inclination in relation to the slip pin to the normal force on the discontinuity plane.
surface is rather limited. The mechanism of the grout – rock –
steel interaction on the slip surface is prohibitively complex
for straight forward closed-form analytical solutions. Current 3 DISCUSSION
research indicates that the complexity of stress interaction
problem can only be analyzed by computer programs such The assessment of the shear pinning techniques used at the
as Finite Element or Discrete Element Modeling techniques. Cadia Hill Gold Mine involved the analysis of the different
A number of authors have had a various degree of success with factors and mechanisms that interact to provide the required
modeling the problem utilizing the Finite Element model in resistance to an unstable geological structure.
three dimensions. The models used different failure criterions The results produced indicated that compression induced
for steel, grout and surrounding material. within the reinforcing element during the early stages of mobi-
Computer models are also able to identify that the bond lization will lead to a reduction in the normal stress acting on
between the grout and the steel at the intersection of the slip the shear plane. This reduction in normal stress will weaken
surface broke at low values of load. The model was able to the resistive force generated from friction between the two
accurately predict the size of the gap between the grout and the shear surfaces. A design methodology for pullout failure of
steel with the laboratory testing confirming this effect without the pinning installations was developed. The developed rela-
exception. It is observed that the maximum strength at failure tions used to redesign an existing setup at the Cadia Hill Gold
is generally achieved in deformable and weaker rock masses Mine could potentially lead to approximate 25% saving in
due to the larger plastic deformations occurring, with greater required grouted steel. Further investigation will be required
steel ductility also adding to the effect. There are two possible to determine the need for such relation or if it in fact holds
modes in which the steel reinforcing element can fail once true for the types of rock encountered.
the mass has mobilized. In weaker and more deformable rock
602

Geotechnical design of the sublevel stoping method for the Poder and
Trabajo Veins in the San Genaro Mine – Peru
A. Samaniego & H.I. Pehovaz

SVS Ingenieros SAC & Mining Department of the Pontificia Universidad Católica del Perú
Keywords: case studies, mine design, stability analysis
1 INTRODUCTION Table 1. Summary of rock mass properties.
The study consists of determining the dimensions of the stable E c ø

open stopes according to the geological and geomechanical Lithology (GPa) (MPa) (◦ ) mb
conditions of the various sectors analyzed in the Poder and Andesite 5.30 1.49 46.9 1.79
Trabajo Veins in the San Genaro mine (Peru). The San Genaro Pyroclast 5.80 0.82 30.8 0.40
mine is located in the eastern slope of the Western Cordillera Breccia 5.93 0.96 37.0 0.72
of the central Peruvian Andes. From a metallogenic perspec- Ore 10.13 1.33 44.0 1.18
tive, the mine is located within the silver mining district of
Castrovirreyna, which politically belongs to the jurisdiction
of the Santa Ana district, in the province of Castrovirreyna,
department of Huancavelica. testing, allow for the characterization and geomechanical
classification of the rock mass and the ore body. The RMR,
2 GEOMECHANICAL MODEL Q and GSI rock mass classification systems are used in the
geomechanical characterization. Rock mass strength is esti-
2.1 Geology mated using the Hoek and Brown (2002) failure criterion. The
rock mass properties are summarize in the table 1.
The morphology is the typical of a high Andean plateau, with
rounded hills, large plains and lagoons formed in depres-
sions. The main lithological units in the mine are: a) the 3 STABILITY ANALYSIS AND DIMENSIONING OF
Caudalosa Formation, b) the Diorite Sub Volcanic and c) Qua- MINING STOPES
ternary colluvial and fluvioglacial deposits covering the rock
outcrops. The main structures follow a prevailing East-West 3.1 Stability analysis and dimensioning of openings
structural pattern, followed by the North-West Andean align- Stope stability assessment is carried out by means of the
ment system and, to a lesser degree, the North-East system. Mathews (1980) stability graph method. For this study, the lat-
The alignment of the Poder and Trabajo veins corresponds est version, updated by C. Mawdesley & R. Trueman (2000),
to the first East – West system, and they dip 70 to 87◦ to was used.
the north and to the south, respectively. Globally two sets
of discontinuities are shown, with dip and dip direction of 3.1.1 Stability verification of current stopes
87◦ /168◦ and 85◦ /096◦ . The analysis shows the following dis- In order to know the current stability conditions for mining the
continuity sets: Sector 1, 82◦ /168◦ , 83◦ /089◦ and 83◦ /360◦ ; open stopes that are being developed in the Poder vein area,
sector 2, 85◦ /159◦ , 77◦ /136◦ , 77◦ /053◦ and 72◦ /015◦ ; sector stability was calculated using the Mathews graph method,
3, 77◦ /171◦ , 76◦ /006◦ , 71◦ /143◦ and 59◦ /080◦ and sector 4, with probability of failure values in the order of 20% and
90◦ /102◦ and 83◦ /335◦ . Some levels in the Poder vein such as 40% being obtained (dilution due to rock falls). With this,
Lv690, Lv665 and Lv640 show sporadic dripping in the vein the rock fall and instability levels are determined to be high,
and footwall area and occasionally in the drifts and bypass. In particularly for a 40% probability of failure.
the Trabajo vein, flows were observed only at Lv 680 in the At Level 640 of the Poder vein, there is an open stope
form of dripping and slight dampening of the vein fill mate- (Stope B) 3.2 m in width, 41 m in height and 100 m in length,
rial, but in general these remain dry. The water table surface with a 20% probability of failure, which remains relatively
should be below Lv. 590. stable. In the same vein, Stope C is 3.2 m wide, 65 m high and
115 m long, with a 40% probability of failure, showing higher
unstable and failure conditions. In accordance with the results
2.2 Geomechanical evaluation and characterization
obtained using the Stability Graph method and the observa-
The analysis of the data from geological mapping and discon- tions to the current stopes in the Poder vein, it is possible to
tinuity characterization, as well as the results of the laboratory determine stope dimensions for stable conditions.
603

Table 2. Parameters of the model materials. when stopes remain empty 25 m in height. The maximum
deformations in the hanging walls, footwalls and roofs are
Andesite Andesite Waste rock
Parameter Sector 1 Sector 4 Ore Backfill
in the order of 0.1 a 0.2%, indicating minor to slightly signifi-
cant damage levels according to Beck’s classification (2005).
z (m) 300 245 300 300 From the analysis, it was found that the barren rock and/or
γ (MN/m3 ) 0.025 0.025 0.0255 0.023 mine waste backfill works as a good confinement element for
K 0.48 0.48 0.48 the stope walls, helping to maintain the stability of the stopes
E (MPa) 5296 5296 10127 150 when mining narrow veins.
ν 0.32 0.285 0.345 0.30 It is concluded that the factors of safety found are just
GSI 58 56 55 slightly greater than 1.0, that is, they are at the stability limit
σc (MPa) 99.3 93.2 119.2 with a minor probability of failure (spalling and/or fall of
mi 18.0 18.0 14.0 rock fragments), affected by the induced stresses of the mined
D 0.70 0.70 0.70
zones.
c (MPa) 1.49 1.22 1.33 0
ø (◦ ) 46.9 47.1 44.0 45.0
mb 1.79 1.60 1.18
s 0.0023 0.0017 0.0015 5 CONCLUSIONS
a 0.503 0.504 0.504
The sublevel stoping mining method using waste rock backfill
is appropriate for narrow ore bodies, as it adapts to the char-
acteristics of the Poder and Trabajo vein design sectors, with
3.1.2 Dimensioning of Crown pillars and Sill pillars using average stope widths of 3.2 and 3.3 m, respectively. The suit-
the Carter method able mining sequence is to stope from the lower levels towards
Rib pillars were designed using the empirical method, which the upper levels, with stope lengths of 250 m to 1000 m. In this
involves designing inclined pillars based on their strength, part of the mining process, it is recommended to leave in situ
first estimating the induced stress field acting on the pillar rock bridges with minimum thicknesses of 1.75 m between the
with the Obert and Duvall method, and then estimating pillar main levels, if any on site. This would improve the stability
strength using the Lunder and Pakalnis methods (1997). of the mining stopes, and at the same time allow for mining
The results obtained using factors of safety of 1.94 and using stoping heights greater than 25 m. At the last mine levels
1.98 for thicknesses of 1.75 m of the sill pillars between the near the surface, in turn, it is recommended to leave a crown
mine levels (for a semi-permanent condition) and the factors of pillar 2.5 m in thickness, so as to guarantee surface stability,
safety of 2.31 and 2.37 for thicknesses of 2.5 m of the surface in accordance with the current national regulations on mine
crown pillars (for a final condition) are considered to be stable. closure. In any case, if there are no crown pillars, the cavity
will have to be completely filled. Finally, it has been proven
4 STRESS-STRAIN ANALYSIS that the results obtained by finite element numerical modeling
match the results from the empirical method.
A stress-strain analysis of the different excavation stages was
performed in order to confirm the results of the empirical
analysis of the opening dimensions for the mining stopes. REFERENCES
The geomechanical parameters required for the modeling have
been obtained from the rock mechanics laboratory testing and Carter, T.G & Miller, R.I. 1995. Crown Pillar risk assessment –
investigations conducted for this project. The failure criterion planning aid for cost-effective mine closure remediation. Trans.
used in this analysis was the empirical modified Hoek-Brown Inst. Min. Metl, vol 104, pp. A41–A57.
(Hoek, 1992) failure criterion, the parameters of which are mb , Lunder, P.J. & Pakalnis, R.C. 1997. Determination of the strength of
hard rock mine pillars. Bull. Inst. Min. Metl., 90 (1013): 51–55.
s and a. The parameters of the values adopted for the materials Mawdesley, C., Trueman, R. & Whiten, W.J. 2000. Extending the
are shown in Table 2. Mathews stability graph for open stope design. IMM transactions
The results show factors of safety greater than 1.0 under all section A, vol 110, number 1, pp. 27–39.
the stability conditions simulated. For dimensioning mining SVS Ingenieros SAC, 2010. Estudio geomecánico del método de
widths of 3.5 m, factors of safety above the limit equilibrium minado de las vetas Poder y Trabajo. Informe Final. Reporte
were obtained in the hanging wall and footwall host rock area, 1-M-284-005. 63 pages.
604

Numerical study on disturbance of longwall mining to stress state in the rock mass
Mahdi Shabanimashcool & Charlie C. Li

Department of Geology and Mineral Resource Engineering, Norwegian University of Science and Technology (NTNU), Norway
Keywords: coal mines, numerical modeling, field measurements
In longwall mining, stope voids are filled by caving of the

roof strata. Caving causes enormous stress concentrations in
the host rocks and results in rock fracturing and large con-
vergences in maingates, Nowadays, it is obvious that stress
concentration is a significant controlling factor in many of
the longwall mining collapses. Stress concentrations in main-
gates, chain and barrier pillars are of particular concern.
Stresses can not transmit through unconsolidated caved mate-
rials. Therefore stress concentrations in maingates and pillars
are associated with the consolidation degree of the caved rock
in goafs and as well as the geometry of the caved space in the
rock mass.
Redistribution of the vertical stress in longwall mines has
been studied by a number of authors using numerical mod-
elling and in-situ measurements. However, redistribution of
the horizontal stresses has not been appropriately studied.
Study site, Svea Nord mine, is one of the biggest coal mines
in the arctic area. The mine is located at 78◦ N in the west-
ern part of the Spitsbergen, Svalbard. Coal is extracting by
longwall mining method with an annual coal production of
3 millions tones. Longwall panels are 250 wide and 2500 m
long. Overburden varies from 10 to 400 m. In addition, most Figure 1. Numerically simulated and in-situ measured horizontal
of the ground surface of the mine is covered by a glacier with stresses in the roof of a maingate.
a thickness up to 250 m. The aim of this study is to investigate
how the longwall mining changes the stress distribution in the
host rock. Caving is different from failure or yielding of the strata.
Stress measurements conducted in the mine revealed that Caving is a process that pieces of failed rocks move downward
the in-situ major principal stress σ1 is horizontal and approxi- under gravity, accompanying large deformations. A novel
mately parallel to the longwall panels; the moderate principal numerical approach was developed to simulate longwall min-
stress σ2 is also horizontal and approximately perpendicular ing. In the proposed method, progressive caving and fracturing
to the longwall panels; and the minor principal stress σ3 is of the roof strata was modelled in the first step. Then, the
vertical.The horizontal to vertical stress ratios, Kmax and Kmin , consolidation of the caved materials and stress changes were
are 2 and 1.7, respectively. simulated. Boundaries of the caved zone were outlined by
In the study site, a monitoring station was setup in a main- a specific criterion based on the maximum principal strain.
gate to monitor how the in-situ horizontal stresses in the roof of The consolidation of the caved materials was simulated by the
the maingate would change during mining in the two longwall double yield constitutive model. Salamon’s formula for loose
panels in the two sides of the maingate. materials was used to calculate the cap pressure parameters of
FLAC3D code was used for the three-dimensional simu- the double yield constitutive model.
lations. In the numerical model nonlinear strain softening The numerical investigations show that the mining in the
constitutive models were used to simulate the behaviour of neighbour panel does not cause any significant stress changes
the rock masses. and furthermore yielding in the roof strata of the maingate.
Caving of the roof strata is a dynamic process so that it can- The numerical results may provide useful information for
not be accurately modelled with ordinary simulation codes. chain pillar designs and rock support in the mine.
605

Ground support system at Leeville underground mine
Changshou Sun
Newmont Mining Corp, Elko, NV, USA
Keywords: Geotechnical data collection, Ground support system, Rock bolts, Shotcrete
The Leeville underground mine has been in operation since The value of ESR is related to the intended use of the exca-
2006 and currently produces 3176 tons (3,500 short tons) of vation and to the degree of security which is demanded of
ore per day with an average gold grade of 10 g/ton (0.32 opt). the support system installed to maintain the stability of the
The mine has a large horizontal extent of 762 m (2,500 feet) excavation. (Barton et al., 1974). Then, a support system is
by 152 m (500 feet) and a vertical extent of 6.1 to 61 m (20 to determined by a chart, the equivalent dimension De versus
200 feet). Based on the characteristics of the ore bodies, the the value of Q. Q values in Leeville are 0.1 to 4. The average
deposits are divided into several mining zones. The geotech- Q value is 0.4. The spans of drifts are between 4.5 to 12 m (15
nical properties of mining zones are also different from one to 40 ft). So, for short term excavations, De is 1.52 to 4.06 m,
zone to another. The typical ground condition at Leeville is and long term excavations, De are 2.86 to 7.62. All numbers
poor to very poor. So, an appropriate ground support system is are plugged into Grimstad and Barton’s chart. The support
key for the successful production at the Leeville underground system can be determined. The spacing of bolts is 1 to 1.4 m
mine. (3 to 4.6 ft) for both long term and short term structures. The
Based on the available geotechnical data, the empirical and shotcrete thickness is 2 inches for short term structures and
numerical methods are used to construct the ground support 50 to 100 mm (2 to 4 inches) for the long term structures.
system at the Leeville underground mine. The main ground Several types of rock bolts were used at the Leeville mine,
support system includes in the following. including Split Sets, Swellex, Resin bolt, Cable bolt, and MAI,
Geotechnical data collections, core logging, site investiga- a self-drive grout bolt. Based on the geotechnical investiga-
tion and extra geotechnical drills. Based on the characteristics tion in Leeville, the rock condition in Leeville is poor to very
of the ore bodies, the deposits are divided into several mining poor. Although the intact rock is hard, it is extremely fractured.
zones. The geotechnical properties of mining zones are also The average RQD of rock in Leeville is only 21. The ground
different from one zone to another. To classify and evaluate condition is weak. To select an appropriate support system,
the rock mass condition, a complete geotechnical data must the concept of the interaction of ground reaction curve and
be collected. support load-displacement characteristics is applied. As rock
mass in Leeville is very weak, a yield support system should
• Data collected by core logging. Most geotechnical data is
be selected. The yield support system can allow the ground to
collected by this method from the exploration drills. Param-
deform more and decrease the support pressure requirement.
eters of RQD (Rock Quality Designation), FF (Fracture
To select the appropriate rock bolt, the anchorage character-
Frequency), Compressive strength, and RMR (Rock Mass
istics of rock bolts are evaluated. Stillborg (1994) carried out
Rate) are estimated
a number of tests on rock bolts and dowels installed across
• Data collected by site investigation. More detail geotechni-
a simulated ’joint’, using two blocks of high strength rein-
cal data and geo-structure data are collected by this method.
forced concrete. Based on Stillborg’s test results, the Swellex
Parameters of RQD, FF, Compressive strength, RMR, Q
and Split Sets are good for the yield support system. However,
(Rock Tunneling Quality Index) and geo-structure data are
the anchorage of the Split Sets is too low (1.5 to 4.5 ton/m)
estimated in the underground drifts and stopes.
and also it is hard to be installed for the extremely fractured
• Data collected by extra drill holes for the critical infras-
rock. So, the Swellex bolt is selected in Leeville. The Swellex
tructure facilities. To know more detail geotechnical infor-
bolt has a high anchorage (12 to 18 ton/m), easy to be installed
mation in some areas in which the critical infrastructure
in the fractured rock, and installation quality control is easy
located, this method is used.
to be guaranteed by operators
Leeville ground support systems are established based on Based on the Hoek and Brawn, Stillborg I, modified
the Tunnelling Quality Index (Q) for the determination of rock Barton, and Lang methods, a new equation to determine the
mass characteristics and tunnel support requirements, (Barton bolt length in Leeville is proposed. L is the length of the
et al., 1974). In relating the value of the index Q to the stabil- bolt and B is the span of the drift:
ity and support requirements of underground excavations, an
additional parameter is defined called the Equivalent Dimen-
sion, De , of the excavation. This dimension is obtained by
dividing the span, B, diameter or wall height of the excava- Bolt lengths in current in use at Leeville are based on
tion by a quantity called the Excavation Support Ratio, ESR. the new method. The new method is verified by a numeri-
Hence: cal method. In the numerical approach, the minimum stress
method is used to determine the length of the rock bolt. The
analysis also was proven by a ground control monitoring
system in Leeville underground.
606

Quality control (QC) system includes materials quality con- • Geotechnical data collections, the collections consist of
trol and installation quality control. The materials quality core logging, site investigation and extra geotechnical drills.
control is conducted by manufacturer. An annually ASTM • Based on the analysis of geotechnical information, appro-
F432 certification is provided by producers for the rock bolts priate stiffness of support system is selected. Flexible and
products. Backfill and shotcrete samples are tested at the lab active support system seems to fit the ground condition at
after curing 28 days. The installation quality controls for the the Leeville underground mine.
support system are conducted quarterly on the site. The site • Rock bolts selection. Using characteristics of the ground
quality controls include the pull test for the rock bolts and deformation and anchorage capacity of rock bolts to select
Schmidt hammer test for the shotcrete. the suitable rock bolt in the ground support system.
Two types of pull test methods are used. One is 0.6-m • Determination of the length and the pattern of the rock
(2-ft) anchor test (also called test with sleeve) and the other is bolts, using empirical method determine the length and the
a non-destruction test. For the sleeve test, a 1.8-m (6-ft) sleeve pattern of rock bolt in ground support system.
steel tube is used for the confined non-anchored with hole and • Construct the ground support standard, based on the vari-
only 0.6 m (2 ft) in the toe of hole is anchored in the hole so ety of ground condition at the different mining zones and
that the anchorage capacity per feet can be determined. For service time of the drifts, the ground support standard is
non-destruction test, when the load capacity for the bolt is sat- established.
isfied (maximum 11 tons) the test is passed. The site shotcrete • Quality control, to secure the ground support system
is tested with Schmidt hammer quarterly. being workable, the quality control is conducted to check
Leeville underground mine started to operate in 2006. The anchorage of the rock bolts, strength of the shotcrete, and
initial ground support system is modified from the similar backfill.
underground mine, Carlin East. However, when more geotech-
After several years application for the new ground support
nical data were available and more drifts were developed the
system at Leeville, good safety and production results have
ground condition at Leeville is completely different from that
been achieved which include the double wide and double high
at Carlin East. So, the new ground support system at Leeville
stopes and the large span drifts.
is constructed by following steps.
607

Surface settlement due to deep mining with the Numerical Manifold Method
Y. Tal & Y.H. Hatzor

Department of Geological Environmental Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel
Keywords: mine design, numerical modelling
This paper examines the expected settlement at the ground Table 1. Maximum HD before surface heave, and settlement
surface due to circular cavities in layered rock. We begin with profile width in four different cavity configurations.
a verification of the Numerical Manifold Method (NMM) for
the case of an infinite plate with a circular cavity (Kirsch solu- Cavity depth [m] 40 60 80 60
Cavity diameter [m] 20 20 20 40
tion). The relative errors obtained for the vertical displacement Maximum HD [m] 40 80 120 80
change from 7.65% to 0.64%, depending on the position of the B [m] 80 120 140 120
measurement point. The Kirsch solution assumes constant far
field stresses and ignores the influence of gravity. We there-
fore verify the NMM for the case of a layered plate subjected
to gravity but with no cavity under plane stress conditions,
using a self-developed analytical solution for this problem.
A very good agreement is obtained, with relative errors never
exceeding 0.32%.
This set of verification experiments allows us to proceed
safely with exploration of boundary effects on the com-
puted surface settlement. The amount of surface settlement
is controlled by two opposite factors: 1) stress concentra-
tions due to cavity which increases the amount of settlement,
2) elastic rebound due to stress relaxation of the underlying
strata as a response to removal of material in the formation
process of the cavity, which decreases the amount of settle-
ment. Considering item 2 above, the influence of the distance
between the center of the cavity and the lower boundary of
the model (HD ) on the surface settlement is explored here Figure 1. Profiles of surface settlements above a 20 m diameter
in four different configurations (Table 1). In each config- circular cavity embedded in an elastic medium at depth of 60 m
uration, the distance HD is increased until surface heave is (E = 10 GPa, ν = 0.2, γ = 18 kN/m3 ) for various HD distances. In
obtained. this case surface heave is obtained when HD is greater than 80 m.
The results indicate that the distance HD has a significant
influence on surface settlements above the cavity. The total cavity depth as well as the settlement profile width (B)
surface settlement decreases significantly with increasing HD (Table 1). The diameter of the cavity does not seem to influ-
due to the elastic rebound effect (Fig. 1). The maximum HD ence the maximum HD or the settlement profile width B, but
before surface heave is obtained increases with increasing only the total amount of surface settlement V (Table 1).
608

Numerical analysis of rock support interaction in tunnels in seismic regions
Tashi Tshering
University of Oslo, Oslo, Norway
Rajinder Bhasin & Roger Olsson
Norwegian Geotechnical Institute, Oslo, Norway
Valerie Maupin
University of Oslo, Oslo, Norway
Keywords: Tunnel, seismicity, rock support, Q-system, Maximum Axial Force on Lining (MAFL)
1 INTRODUCTION increasing Q, the deformation modulus, peak friction angle,

and peak cohesion also increases. Other parameters such as
Although it was a common notion among engineers that unit weight, Poisson’s ratio, tensile strength, and dilation angle
underground structures such as tunnels are invulnerable to are kept constant.
earthquakes, some underground structures have undergone The deformation moduli for different Q values are obtained
severe damages during recent large earthquakes. from the relationship in Barton (2002) and the peak friction
The factors determining the impact of earthquakes on tun- angle and cohesion were determined from tables in Waltham
nels are, among others, the quality of the rock mass in which (2004), with peak cohesion modified as per Singh & Goel
the tunnel is placed, depth and dimension of the tunnel, and (2006).
the intensity of the earthquake felt at the tunnel location As it is important to specify post-failure characteristics
(Dowding & Rozen, 1978). This paper presents numerical while modeling elastic-plastic materials, elastic-perfectly-
studies to investigate the effect of earthquakes, due to shaking plastic (EPP), applicable for weak rocks, was chosen as the
or vibration, on tunnels placed in rock masses with different post-failure characteristics for all the rock classes used here.
rock mass quality Q. The effect of dimension of tunnel and The plane-strain two-dimensional model of a circular tunnel
depth are also included. excavated within a rock mass with different Q values, obtained
by specifying the elastic and strength parameters discussed in
2 MODELLING Q VALUES previous section, is considered for this study. The gravity field
stress, with lithostatic stress, was used as the initial stress on
The NGI Q-system (Barton et al., 1974) is a rock classification the model.
method used to determine the support for a tunnel placed in In order to study the rock mass-support interaction, a 10-
a particular rock mass quality. In order to study the effect of cm thick shotcrete liner, modeled as standard Timoshenko
rock mass quality on impact of earthquakes on tunnel support, beam, was installed immediately after excavation of the tunnel.
five rock classes with different rock mass quality index Q As the load on the support also depends on the properties
(see Table 1) are simulated in Phase 2 by specifying different of the liner, the properties of the liner was fixed for all the
elastic and strength parameters. experiements so that the impact of earthquakes, irrespective
The finite element modeling (FEM) program Phase 2, of the properties of the liner, can be studied.
commonly used for underground designs, can be used to This procedure was achieved in three stages such as: (1)
investigate the effect of seismic loading through pseudo-static Stage 1: the tunnel is still unexcavated and equilibrium at
approach. During pseudo-static seismic analysis, an additional this stage merely represents consolidation due to self weight,
load, equal to a given percentage of the body force, or self (2) Stage 2 includes excavated tunnel with installed shotcrete
weight, of the finite element, which is determined by the spec- liner, and (3) Stage 3 is just seismic loading added to Stage 2.
ified seismic coefficient, is added to simulate seismic loading. Thus, pseudo-static seismic loading is added after the static
The five rock classes used in this study are simulated using equilibrium has been reached, representing that the seismic
respective elastic and strength parameters – namely defor- loading due to earthquakes occur only at a later stage after the
mation modulus, peak friction angle, and peak cohesion. For construction of the tunnel.
The maximum axial force on the liner (MAFL), the largest
Table 1. The description and range of Q values for the five rock axial force experienced around the circumference of the liner,
classes modeled in this study. Adapted from Waltham (2009). is used as a proxy for support pressure. The difference between
MAFL for static loading and MAFL for seismic loading is the
Class I II III IV V additional force due to seismic loading and is referred to as
seismic force on the lining.
Very Very A series of numerical experiments, for Q from 1 to 40, depth
good poor
from 30 m to 200 m and tunnel diameter from 5 m to 20 m,
Description rock Good rock Fair rock Poor rock rock
were conducted to investigate the effect of seismic loading as
Q value >40 10–40 4–10 1–4 <1 a function of rock mass quality Q, and depth and dimension
of the tunnel.
609

be independent of depth of the tunnel. On the other hand, the
seismic force on lining decreases (for both percentage and
net increase) as the depth of the tunnel increases for elastic-
perfectly-plastic model for both poor (Q = 1) and good quality
(Q = 40) rocks.
For a 20-m diameter tunnel at depths ranging from 30 m to
200 m, the seismic force on the lining decreases from about
80% of the static MAFL at 30 m depth to 2% of static MAFL
at 200 m depth for both Q = 1 and Q = 40.
4 CONCLUSIONS
For elastic good quality rocks, the MAFL, for both static and
seismic loading, is independent of depth (or in-situ stress) and
Figure 1. MAFL for both static and seismic loading (h = 0.5) for tunnel dimension. On the other hand, for EPP models (weaker
the four rock classes used for this study. The MAFL for elastic and rocks), the MAFL under static and seismic loading appears
EPP models overlap for good quality rocks with higher Q. to depend on tunnel dimension, depth of the tunnel and rock
mass quality Q. Consequently, the seismic force on the lining
For this study, a fixed pseudo-static seismic loading with is also dependent on depth and the surrounding rock mass
horizontal coefficient = 0.5 and vertical coefficient v = 0 is quality Q.
used for all the experiments. As the tunnel diameter increases from 5 m to 20 m, the
absolute magnitude of the MAFL increases with increasing
3 RESULTS tunnel dimension, but the net increase in MAFL due to seis-
mic loading (seismic force) does not vary significantly with
3.1 Effect of rock mass quality tunnel dimension.
Figure 1 shows the variation in the axial force on the shotcrete Although the absolute magnitude of MAFL for both static
liner of a tunnel in the four rock classes at 60 m depth. It can and seismic loading increases with depth, the seismic force
be seen that the MAFL, for both elastic and EPP models, is on the lining decreases with depth even though the seismic
dependent on the rock mass quality, where the MAFL for both coefficient is fixed for different depths, thereby simulating
static and seismic loading increases as Q decreases. the same level of shaking at different depths.
As the rock mass quality decreases from Q = 40 to 1, the For a 20-m diameter tunnel at depths from 30–200 m, the
absolute MAFL for both seismic and static loading increases seismic force on the lining decreases from about 80% of the
by about 3–4 times. static MAFL at 30 m depth to 2% of static MAFL at 200 m
For a particular depth, it is observed that the MAFL, at depth for Q = 1 and Q = 40; not only the percentage, but the
least for static loading, for elastic and elastic-perfectly-plastic net increase also has a decreasing trend with depth. This indi-
models overlap for good quality rocks (Q ≥ 4 for 60 m depth). cates that although the level of shaking is same for different
Conversely, for rocks with lower Q, the MAFL values deviate depths, there is still some reduction in effect of earthquakes
for the two models. This is because, within the elastic limit, at depth compared with shallower tunnels. It is generally
the rock mass behaves like elastic materials irrespective of the accepted that the major reason for the reduction of under-
choice of the post-failure characteristics. ground damage at depth is due to lower level of shaking with
For EPP materials, the seismic force on lining increases with respect to motion at the surface (Sharma and Judd, 1991).
decreasing rock mass quality. For the four rock classes, there However, from this study it can be inferred that there could be
is an about 12% increase in MAFL due to seismic loading at least another, probably stress-controlled, phenomenon that
for elastic models, and the % increase due to seismic loading contributes to the reduction of impact of ground shaking at
for EPP materials is about 20–26%. depth, in addition to the commonly accepted “lower level of
shaking at depth”.
3.2 Effect of tunnel dimension
For elastic-perfectly-plastic models, the MAFL increases with REFERENCES
tunnel diameter for Q = 1, whereas on the same scale, the
increase in MAFL with tunnel diameter for Q = 40 is insigni- Barton, N., R. Lien and J. Lunde. 1974. Engineering classification
ficant. Similarly, the increase in MAFL for elastic models is of rockmasses for the design of tunnel support, Rock Mechanics,
lower than for elastic-perfectly-plastic models. This shows that Vol. 6, No. 4, pp. 189–236.
for good rocks, behaving elastically, the MAFL does not vary Barton, N. 2002. Some new Q-value correlations to assist in site
significantly with tunnel dimension, and whereas the MAFL characterization and tunnel design. Int. J. Rock Mech. & Min. Sci.
increases with tunnel dimension for poor quality rocks, such Vol. 39/2:185–216.
as for Q = 1. On the other hand, the seismic force on the lin- Dowding, C. & Rozen, A. 1978. Damage to rock tunnel from earth-
quake shaking, Journal of Geotechnial Engineering ASCE, vol.
ing seems independent of the tunnel dimension. The increase
104: 175–191.
in MAFL due to seismic loading is ∼0.1 MN for Q = 1 and Singh, B. & Goel, R.K. (First Edition) 2006.Tunnelling in weak rocks.
∼0.02 MN for Q = 40. Oxford, UK: Elsevier.
3.3 Effect of depth Sharma, S. & Judd, W. R. 1991. Underground opening damage from
earthquakes, Engineering Geology, vol. 30: 263–276.
The magnitude of the seismic force on lining is insignifi- Waltham, T. 2009. (Third Edition). Foundations of Engineering
cant for elastic models when compared with EPP models. Geology. Taylor & Francis.
In addition, the seismic force for elastic models appears to
610

Investigation into the strength of iron ore pillars in the Lorraine basin
J. Nielen van der Merwe

University of the Witwatersrand, South Africa
Christophe Didier & Frederic Poulard

INERIS, France
Keywords: risks and hazards, stability analysis, back analysis, case studies, rock failure
1 INTRODUCTION where k = number related to material strength

we = effective pillar width
Iron ore mining was carried out in the Lorraine district in the h = pillar height
north-east of France for several centuries. The predominant Safety factors were subsequently calculated for each of the
mining method was bord and pillar, with pillar extraction in cases in the two data sets of failed and stable pillar cases by
some areas. There is no evidence today that a formal design varying α and β between the limits of 0 and 2 in increments
method for pillar size was in use. Pillars appear to have been of 0.2. There were thus 100 sets of safety factors for the two
sized by trial and error. data bases, each calculated with different combinations of α
Mining ceased in 1975. Although knowledge of the pillar and β.
strength is not required now for mining purposes, it is impor- The next step entailed identifying the values of α and β
tant for the evaluation of pillar stability and hence surface for which the overlap area was the smallest, the underlying
stability as the region has been extensively built up. philosophy being that the best formula is the one that results
The method of analysis was essentially to create data bases in the smallest possible area of overlap.
of failed and stable pillars and then to develop a pillar strength The method that was used, was based on the procedure
formula that resulted in the best separation between the dis- described by Harr (1987) to determine the area of overlap
tributions of the data bases of collapsed and intact pillar between two normal distributions.
cases. For each of the 100 data sets obtained with the different
combinations of α and β, the area of overlap between the safety
factors of failed and stable cases was subsequently calculated.
2 DATA BASES
Following careful scrutiny, 17 cases of collapsed pillar cases

(where a number of pillars in a cluster had collapsed, either 4 RESULTS OF THE METHOD TO DETERMINE
indicated as such on mine plans or indicated by surface α AND β
subsidence) were identified
For each collapse case, it was attempted to include two sim- The results of the attempt to optimise α and β are shown
ilar cases where there was no evidence of collapse, in the graphically in the form of a contour plan, see Figure1.
data base of stable pillars. The final data base of stable pillars The centre of gravity of the minimum area in Figure 1 occurs
consisted of 33 cases. where α = 0.8 and β = 0.6. At that point, the relative area of
Pillar loads were obtained using a simple Tributary Area overlap is 16.3%. However, the area of overlap for the combi-
approach, modified by a factor based on the surrounding nation of α = 1.0 and β = 1.0 (thus a linear strength formula)
mining geometry. is only minimally worse, being 16.8%.
The frequency distributions for the failed and stable cases
are compared in Figure 2.
3 METHOD OF DETERMINING OVERLAP BETWEEN
DISTRIBUTIONS
The best measure of stability is the concept of a safety factor, 5 DETERMINATION OF k

simply being the quotient of the strength of a pillar and the
load imposed on it. Pillar strength is widely accepted as being The final step is to determine the value of k. If one accepts that
a function of the material strength and the width-to-height ideally, failure should occur when the load on a pillar exceeds
ratio of a pillar, or the strength, then a safety factor of 1.0 can be deemed the
critical value. The ideal value of k is therefore that value where
the mean of the safety factors in the failed data population, is
equal to 1.0.
611

Figure 3. Cumulative normalised frequency distribution plot for
the failed data base, with k = 3.7.
Figure 1. Contour plot of the areas of overlap between the distribu- and
tions of safety factors of failed and stable pillar populations.
6 CONCLUSION
The method used to determine the optimal strength formula

for coal pillars in South Africa has been used successfully for
the iron ore pillars in the Lorraine basin in France.
In the case of the Lorraine pillars, the linear formula is
preferred even though it has a slightly lower capability to
distinguish between failed and stable pillars. Therefore, the
proposed strength formula for the iron ore pillars in the
Lorraine basin is Equation (3) above.
Figure 2. Frequency distributions of the populations of failed and

stable pillar cases, for α = 1, β = 1 and k = 3.7. REFERENCES
Harr, M.E. (1987). Reliability-Based Design In Civil Engineering.

For the combination where α = 0.8 and β = 0.6, the value McGraw-Hill Book Company 1987. p. 47, pp. 130–134.
of k was found to be 3.5 MPa and for the combination where Van der Merwe, J.N. (2003). Linear coal pillar strength formula for
South African Coal. J. S. Afr. Inst. Min. Metall. June 2003.
α = β = 1.0, it was found to be 3.7 MPa, see Figure 3.
Wagner, H. (1980). Pillar Design in Coal Mines. J. S. Afr. Inst. Min. &
The resulting equations for the two combinations are thus: Metall. January 1980.
612

On the fractal characterization of mining-induced crack network in

overburden rock strata
Zhi-guo Wang
Institute of Rock Mechanics and Fractals, China University of Mining and Technology (Beijing), Beijing, China
College of Mining Engineering, Hebei United University, Tangshan, China
He-ping Xie
Sichuan University, Chengdu, China
Hong-wei Zhou
Keywords: mining induced crack network, deep mining, distinct element method, fractal dimension, evolutional law
With the increasing of coal mining depth, a better understand- spatial distribution of mining-induced crack network in over-
ing of evolution of mining-induced crack network plays an burden rock strata under the condition of deep mining at
important role in deep safe mining. The spatial distribution different coal seam thickness, according to the factors of engi-
of mining-induced crack network in overburden rock strata neering geology and mining technology on the coal seam
affect the stability of overburden rock strata to some extent No. 11-2 in Panyi Coal Mine, based on results of physical
under the condition of deep mining. So it is very important simulation experiment, taking the mined rockmass in over-
to research the spatial distribution of mining-induced crack burden rock strata as the crack rockmass of continued broken
network and its evolutional behavior for understanding the and fracturing. The rockmass material constitutive model is
evolution characters of overburden rock strata structure, the defined as Mohr-Coulomb plastic model, the coal seam is
subsidence law and stability of mined rockmass. The crack defined as strain softening model, and the joints is defined
network of rockmass can be described by the fractal geometry as joint area contact model with Coulomb slip failure in the
method, and the studies shows that the spatial distribution of numerical model. The original stress in the numerical model is
mining-induced crack network has the fractal characters, the defined by the actual measured stress in the mine. For model-
evolution law of the fractal crack network may be researched ing the mining-induced crack network better, the setting of the
with numerical simulation method for understand the break- horizontal joint set is according to the borderline in the hori-
age and movement law in the overburden rock strata. This zontal rock strata, the virtual joints is added in thick rock strata,
paper researches the statistical self-similarity and the evolu- and the space between every two vertical joints is defined as
tional law of mining-induced crack network by adopting the 9m base upon the pace of roof periodic caving in the phys-
fractal geometry theory in combination with the experiment ical simulation and field measurement. The software UDEC
results of the numerical simulation in the coal seam No. 11-2 is employed to simulate the mining-induced crack network
of Panyi Coal Mine. by considering the main fault Fe7 and it’s branches, and sim-
The study object of the paper is based on the working face ulated a series of models with different coal seam thickness
No. 2171(1) of the coal seam No. 11-2 in Panyi Coal Mine which equal to 1.8 m, 4 m, 6 m, 8 m for comparatively study the
of Huainan coal district. The coal seam is a near horizontal coal seam thickness effects to mining-induced crack network
coal bed, with an average thickness of 1.8 m. The dimension distribution.
of working face is 1780 m long in strike, 206 m in dip. The Box-counting dimension is one of the most used fractal
advancing velocity of the working face is 3.6 m/d by longwall dimension, so it is used to calculate the fractal dimension of
coal mining method with integrated mechanization coal cut- mining-induced crack network. The software Fractal Dimen-
ting system. The coal field is located in the plain, the terrain sion adopted may calculate the fractal dimension of all kinds of
is flat and the average mining depth is 742 m. The geologic two-dimension digital image quickly. Firstly the open mining-
structure in the area is middle complicated, belongs to a mon- induced crack network can be output from the numerical
oclinal structure, and is crossed at an angle by the fault Fe7. simulation experiment, then the images with different min-
The immediate roof of the coal seam No. 11-2 is mudstone, ing width is processed by the software Photoshop to extract
sandy mudstone and coal seam No. 11-3, the main roof is a mining-induced crack network, and finally the processed
medium to fine grained sandstone, immediate bottom is mud- images is inputted in the software Fractal Dimension with
stone, sandy mudstone, mudstone, carbon mudstone and coal box-counting dimension to calculate the fractal dimension of
seam No. 11-1, and the main bottom is fine grained sandstone. mining-induced crack network.
The DEM (discrete element method) was applied to estab- By studying of mining-induced crack network characters
lish discontinued medium models for the simulation on the in different thickness, different mining width with fractal
613

geometry, the study results is shown that mining-induced crack increases with increasing of mining width in general. The
network has statistical self-similarity very good, the relation evolutionary curve of the fractal dimension may be divided
between its fractal dimension and crack network is one-to-one into three stages: increasing dimension, falling dimension and
correspondence, the fractal dimension may response the distri- re-increasing dimension. The fault makes the subsidence area
bution characters of mining-induced crack network preferably. of overburden rock strata become saddle shape and separate
Base on the above results, the fractal geometry theory is into two zones, change the spatial distribution law of mining-
adopted to research the evolutional law that the fractal dimen- induced crack network, and decrease the fractal dimension.
sion of mining-induced crack network varied with mining The fractal dimension of mining-induced crack network has
width, underground pressure, rock strata sedimentation coef- nonlinear relation with advance supporting pressure, rock
ficient, and analyze the effect of fault on the evolutional law strata subsidence coefficient, and is affected by the thickness
and coal seam thickness on the fractal dimension. of coal seam to some extent, increases with thickness before
It is shown that the results of the numerical simulation by the work face reach to the fault, decreases with thickness after
UDEC are in good agreement with that of physical simula- the fault, but this law is indistinctive and should be further
tion modeling, and the fractal dimension of crack network studied.
614

Safety evaluation of an underground excavation for crusher operation
Hyung-sik Yang
Department of Energy and Resources Engineering, Chonnam National University, Gwangju
Myong-Hwan Jang
Korea Resources Corporation (KORES)
Won-Beom Kim, Hyong-Doo Jang, Jung-Gyun Kim & Seung-Jun Kim

Department of Energy and Resources Engineering, Chonnam National University, Gwangju
Keywords: mining, numerical modeling, rock joints, rock support, stability analysis
With the development of industry and increased awareness of Paddies and ordinary fields are located on the upper part of
environmental issue, eco-friendly operation of mine is gain- the facility, and the surface water may flow into the pit along
ing importance. The process of rock crushing generates noise, the joints from the upper part to the lower part during the rainy
vibrations and enormous dust, which pollutes the surround- season. Therefore, water must be cut off to properly guide the
ing environment. Mitigation of these environmental nuisances water in the pit.
generated by the crusher is essential. In this aspect, the present The stereographic-projection analysis results showed that
study focuses on geo-mechanic based safety investigation for pillars 1, 2, and 4 were exposed to planar failure, but that
operating crusher in an underground mine. pillar 3 hardly was. Risk of wedge failure was found in pillars
Geo-mechanical data of rocks (Q, RMR value) were inves- 1-3 and was low for pillar 4.
tigated and fed to appropriate FLAC3D analysis for evaluating According to the RMR analysis results for the pillars and
safety of the given underground excavation for installation of excavation site, the engineering properties of the rock was
crusher. Safety evaluation was undertaken in 3 steps and the “ordinary” or “satisfactory,” with an average RMR of 58 and
salient outcome is given as follows. a Q-system value of 4.4.
The stability test results based on the numerical analysis
i) Safety evaluation of existing underground excavation
showed that after the application of the recommended sup-
without removing any pillars to provide the benchmark-
port amount, pillars 1 and 3 had high safety factors above 3,
ing results; This step revealed that the underground
but pillars 2 and 4 still had low safety factors (0.97 and 1.0,
excavation was safe.
respectively). The numerical analysis for the extended pillars
ii) Safety evaluation of an underground excavation assuming
2 and 4 increased the safety factors to 1.83 and 1.92, respec-
some part of the pillars to be removed, this step showed
tively. This indicates that the pillar size must be increased
the generation of stress concentration in the span between
to improve the safety factor. In addition, the pillar must be
the pillars and in the pillar sidewalls.
supported according to the rock class.
iii) Safety evaluation of an underground excavation was done
assuming some pillar removed. However, after installation
of suitable supports the situation was safe.
615

Experimental study on reinforcing effect of bolt on jointed rock masses

under uniaxial tension
N. Zhang
Institute of Hydroelectric and Geotechnical Engineering, North China Electric Power University, Beijing, China
S.C. Li
The Research Center of Geotechnical Engineering, Shandong University, Jinan, Shandong, China
A.Z. Lu
Institute of Hydroelectric and Geotechnical Engineering, North China Electric Power University, Beijing, China
M.T. Li
Department of Civil Engineering, Shandong Jiaotong University, Jinan, Shandong, China
L. Yang
Keywords: bolts and anchors; lab testing; rock failure; rock support
1 INTRODUCTION
Fractures in the forms of joints and microcracks are commonly

found in natural rocks, and their failure mechanism strongly
depends on the crack coalescence pattern between pre-existing
flaws (Wong et al., 2001). It is commonly accepted that a typ-
ical mechanism of fracture in brittle rocks and heterogeneous
materials under uniaxial compression consists in growth and
interaction of wing cracks initiated from various pre-existing Figure 2. Fixing device for pre-
defects (Dyskin et al., 1999). Therefore, the understanding of Figure 1. Fixed mould. existing crack and shape of crack.
crack growth in rocks is crucial to the reliable performance
of many geotechnical structures. In the actual project, bolt is
usually used to prevent failure of jointed rock mass. But the
reinforcement mechanism is not fully understood by people.
Many experimental and theoretical studies have been con-
ducted to reveal the mechanism (Guo & Ye, 1992; Ge & Liu,
1988; Li et al., 2003).
Note that most of the previous studies were conducted
under uniaxial compression, there were extremely rare results
obtained under uniaxial tension. Although rock mass is gen-
erally under compressive pressure, there exist exceptions of Figure 3. Layout of pre-existing fracture and GFRP.
tensile stress zone in rock mass. Besides, rock is a brittle
material with tensile strength being far lower than compres-
sive strength. Sudden failure may happen when a small tensile
pressure acting on rock mass. Hence, it is crucial to conduct 2.2 Test mould and the method of fracture
researches on the reinforcement mechanism of bolt on jointed manufacturing technique
rock mass under uniaxial tension.
The mould used in this test is shown in Fig. 1. The bolts can be
fixed by holes in the side plates and the pre-existing fracture
can be manufactured by a hard resin sheet fixed on the bottom
2 EXPERIMENTAL STUDY plate by the fixing device (see Fig. 2(a)). The hard resin sheet
is about 0.2 mm thick. The shape of hard resin sheet is shown
2.1 Rock-like material and bolt material in Fig. 2(b).
Rock-like material used in this study was cement mortar, Fig. 3 shows the layout of the pre-existing fracture and
mixed with 40 mesh modified rubber powder. The similarity GFRP in specimens of different anchorage modes. The dimen-
ratio of stress and density were chosen as 10 and 1, respec- sion of specimens is W × T × H = 70 mm × 70 mm × 140 mm.
tively. In this paper, the glass fiber reinforced plastics (GFRP) The pre-existing fracture is half-through with the width
was chosen as the bolt material. c = 35 mm (see Fig. 2(b)).
616

Figure 6. Stress-strain curve of
Figure 5. Two types of typical specimen with anchorage being at
failure modes. T/2.
Stage A: Linear elastic deformation stage (stage oa). In

this stage, the stress-strain curve shows almost a straight line
shape. Stage B: Pre-existing fracture initiation-propagation
Figure 4. Typical stress-strain curves of specimen with different
stage. This stage can be subdivided into two stages, stage ab,
anchorage modes.
the initiation & sudden propagation stage corresponding to
➀➁ in Fig. 5 (I), and stage bc, the slow propagation stage
corresponding to ➂ in Fig. 5 (I). Point c is the critical point.The
3 EXPERIMENTAL RESULTS
crack expands throughout the specimen and dislocation occurs
between bolt and specimen after point c. Stage C: Slipping
3.1 Stress-strain curves and strength of specimens
stage (stage cd). In this stage, bolt is pulled out gradually
Fig. 4 indicates that residual strength was found for all the form one end of specimen. Stress decreases with the increase
reinforced specimens at the post-failure stage. According to of displacement. All the axial load is borne by the bolt.
the values of average tensile strength, it is found that the bolt
can improve the tensile strength by 7∼15% with the mode (c)
(see Fig. 3) having the highest uniaxial tensile strength. 4 CONCLUSIONS
Modified cement mortar is selected to simulate the rock and

3.2 Path line of fracture on surfaces glass fiber reinforced plastics is selected to simulate the real
bolt. Mould used for fixing bolt and installing pre-existing
Based on the observation on the failure surface, the following
fracture and direct tensile device with higher success rate are
phenomena were observed:
designed. Tensile tests on specimens with different reinforced
1) All path lines on specimen surfaces were almost perpen- modes are carried out. Based on the experimental analysis,
dicular to the axial load and every specimen was split into the corresponding conclusions have been obtained.
two parts.
2) Path lines on non-reinforced specimen appeared from two
wings of pre-existing fracture on the front surface and ACKNOWLEDGEMENT
expand rapidly throughout the whole specimen. For the
reinforced specimens, path lines firstly appeared on one The study was financially supported by the Natural Sci-
surface and expanded rapidly toward the region near the ence Foundation of China (Grant Nos.: 50874047, 40872203,
bolt position, and then the path lines propagated towards 40902083).
the opposite surface with very slow growth rate until they
expanded throughout the specimen. REFERENCES
3) Two types of propagation mode were observed.
I) Crack initiated from two wings of pre-existing fracture Dyskin, A.V., Germanovich, L.N. & Ustinov, K.B. 1999. A 3-D
and propagated towards the side surfaces (Fig. 5 (I) ➀) model of wing crack growth and interaction, Engineering Fracture
then towards the back surface (Fig. 5 (I) ➁➂), in the Mechanics, 63: 81–110.
direction perpendicular to the axial load. Guo Y.L. & Ye J.H. 1992. Study on reinforced effect of frac-
II) Crack initiated from the tip of pre-existing fracture and tured rockmass. Water Resources and Hydropower Engineering, 7:
propagated towards the back surface (Fig. 5 (II) ➀). 41–44.
Then the crack propagated in the opposite direction Ge X.R. & Liu J.W. 1988. Study on the shear resistance behaviour of
(Fig. 5 (I) ➁). bolted rock joints. Chinese Jounal of Geotechnical Engineering,
10(1): 8–19.
The reason for the two types of propagation modes is that Li S.C., Chen W.Z., Zhu W.S., et al. 2003. Catastrophe model of crack
bolt played a bridge role, preventing the pre-existing frac- propagating in bolt-supported jointed rockmass. Chinese Journal
ture from initiating. That is, the propagation mode of the of Rock Mechanics and Engineering. 22(10): 1661–1666.
pre-existing fracture was related with the position of bolt. Wong, R.H.C., Chau, K.T. Tang, C.A. & Lin, P. 2001. Analysis of
crack coalescence in rock-like materials containing three flaws-
Part I: experimental approach. International Journal of Rock
3.3 Analytical failure process of reinforced specimens Mechanics and Mining Sciences, 38: 909–924.
Take the mode (c) (see Fig. 3) as an example, the failure pro-
cess of reinforced specimen can be divided into three stages,
i.e., stage A, B and C as shown in Fig. 6.
617

Tunnelling, rock caverns, and underground space

For a correct interpretation of ground reaction curves
F. Amberg
Lombardi Engineering Ltd., Minusio, Switzerland
Subject: Analysis technique and design methods
Keywords: Rock support, Numerical modeling, Tunnelling
For a given underground excavation geometry, the ground in particular if a rigid lining is placed. The most evident cases
reaction curve represents the displacement of a selected point are swelling and creep, but other possible conditions might be
on the excavation boundary as a function of the support pres- contact grouting with high pressure, excavation in a low per-
sure. The classical method refers to the simple case of axial meable and soft ground or placement of a rigid lining near the
symmetry in an infinite elasto-plastic medium, but in fact a tunnel face in case of squeezing ground. In fact, also the loos-
ground reaction curve can be defined for any type of geom- ening up of rock beyond the roof may lead to an equilibrium
etry and loading condition. It is thus possible to include any above the ground reaction curve.
desired effect, e.g. any shape of the cross section, the tunnel Recent developments have shown that even the usual con-
face, the presence of a fault beside the tunnel, a shallow tun- struction procedure of a tunnel tends to lead to an equilibrium
nel, the gravity, the schistosity, the stratification of the rock or lying above the ground reaction curve. This behavior can be
the anisotropy of the initial stress state. The great advantage of pointed out by simulating the tunnel advance in 3D-analyses.
the ground reaction curve is that the result is not limited to a Initially, this statement might sound peculiar, since it puts
single equilibrium, but it allows immediately to verify how the the classical theories aimed to define the equilibrium in tun-
equilibrium of the excavation can change for different support neling somehow into question. However, in every tunnel the
conditions. This is an essential aspect to analyze and to opti- radial pressure at excavation radius is in fact never only always
mize the design of a tunnel in case of expected convergences. decreasing. From the initial stress ahead of the face the radial
It allows to dimension the tunnel lining with respect to the pressure decreases to the minimum value in the excavation
construction procedure and to assess the safety condition of zone just behind the face. Subsequently, an increase of radial
the excavation. stress occurs while support measures are put in place. Any
The ground reaction curve has also some restrictions: in additional support measure, if placed at a distance from the
fact it must be considered that it does not represent all pos- face where convergences still occurs, produces a certain radial
sible equilibrium points, but only the lower boundary. The pressure increase causing a re-compression of the surrounding
ground reaction curve is determined with a progressive reduc- ground. In case of squeezing ground this phenomenon might
tion of the support pressure and thus the stress state within lead to a very sensitive rising of the radial pressure acting on
the failure zone around the tunnel is at failure limit. In case the lining
of an increasing support pressure after lining placement, a After a brief review of the convergence-confinement
re-compression of the ground in radial direction can be method, the paper presents various possible applications of
observed. The final equilibrium will be above the ground the ground reaction curve showing its usefulness for the anal-
reaction curve, i.e. the support pressure is greater than the ysis of underground excavation problems. Finally, the concept
expected one for a certain displacement. The mentioned of the ground reaction curve as boundary for a region of the
re-compression of the ground might arise for several reasons, possible equilibriums is also presented and discussed.
621

Creep closure rate of a shallow salt cavern
P. Bérest, J.F. Béraud & V. de Greef

Ecole Polytechnique, Palaiseau, France
B. Brouard
Brouard Consulting, Paris
E. Hertz & C. Lheur

CSME, Levallois-Perret, France
Subject: Analyses techniques and design methods
Keywords: numerical modelling, rock mass classification, rock failure, rock support, stability analysis, tunnelling
1 EXTENDED ABSTRACT
In the long term, salt behaves as a viscous fluid and salt cav-
erns gradually shrink. Creep closure rates in shallow caverns
are slow. They must be assessed through shut-in pressure tests,
which consist of closing the cavern and measuring the pres-
sure evolution at the wellhead as a function of time; or through
brine outflow tests, which consist of opening the cavern and
measuring the flow of fluid (brine or hydrocarbon) expelled
from the wellhead. In this paper, we describe a brine outflow
test and a shut-in pressure test performed in the 250-m deep
SG13-14 cavern of the Gellenoncourt brine-field operated by
Compagnie des Salins du Midi et Salines de l’Est (CSME) in
Lorraine, France. The SG13 and SG14 7”-wells were operated
as brine-production caverns from 1976 to 1980. After some
time, the two caverns coalesced, and, in 1980, SG13-SG14 vol-
ume was V = 240,000 m3 . At SG13-14 depth, creep closure
rate can be expected to be ε̇cr ≈ −10−5 /yr. Brine thermal- Figure 1. Cumulated expelled mass as a function of time.
expansion coefficient is αb = 4.4 × 10−4 /◦ C. A brine tem-
perature decrease rate of Ṫc = −0.02◦ C/yr would generate a
relative brine volume decrease rate of αb Ṫc ≈ −10−5 /yr: tem- The cavern was shut-in from May 25, 2009 to November19,
perature evolution must be carefully assessed. By December 2009. During a shut-in test, cavern convergence rate is pro-
2008, a temperature gauge was lowered into the SG13 well. portional to pressure build-up rate, ε̇cr = −βṖc . As build-up
The cavern temperature remained perfectly constant during rate due to cavern creep closure is Ṗwh = 47.1 Pa/day, cav-
the period December 2008–June 2010. In 2000 the cavern ern closure rate is ε̇cr = − 0.93 × 10−5 /yr (Cavern complete
had been shut-in after a sonar survey. Eight years later, before closure is reached after more than 100,000 years). Wellhead
the test began, wellhead pressure had built up to approxi- pressure experience small fluctuations due to Earth tides and
mately 0.08 MPa. On July-3, 2008, the cavern was opened atmospheric pressure changes, which are transmitted to the
and wellhead pressure dropped to zero. Cavern compress- cavern through the rock mass. Creep closure rate is slower
ibility was measured to be βV = 130 m3 /MPa. The outflow during the shut-in test. In fact transient effects triggered by the
test began on July 23, 2008 and was completed by May 25, initial Pc = −0.08 MPa pressure drop are important at the
2009, Figure 1. As temperature rate is exceedingly slow, the beginning of the outflow test. They mainly include transient
average brine outflow rate or q = 9.5 liters/day is governed creep and salt crystallization. The steady-state creep closure
by cavern-creep closure and the relative creep closure rate is rate, as observed during the shut-in test, is slightly slower than
ε̇cr = −q/V = −1.45 × 10−5 yr−1 . Large fluctuations can be 10−5 /year or 2 m3 /year. This value proves that even in the long
observed: periodically, the brine flow rate is several hundreds term (several centuries) subsidence and possible brine leaks
of liters per day. Conversely, for most of the time, the flow from the cavern should have negligible impact from the point
rate is nil and the air/brine interface drops down into the well. of view of environmental protection.
In fact, the opened cavern behaves as an extremely sensitive
barometer.
622

Convergences and core extrusions during construction of the Gotthard Base Tunnel
L. Cantieni & G. Anagnostou

ETH Zurich, Switzerland
R. Hug
Ernst Basler + Partner AG, Zurich; formerly ETH Zurich, Switzerland
Keywords: tunnelling, weak rock, case studies, monitoring
1 INTRODUCTION 2 GEOLOGY
Squeezing intensity in tunnelling often varies over short The tunnel crosses the northern TZM formation and the
distances, even where there is no apparent change in the Clavaniev zone for about 1150 m at a depth of 800 m. Both the
excavation method or lithology. This variability makes the TZM formation and the Clavaniev zone are characterized by
prediction of the conditions ahead of the face very diffi- alternating layers of intact and more or less kakiritic gneisses,
cult, even after some experience has been gained with a slates, and phyllites. The term “kakirite” denotes a broken or
specific geological formation during excavation. However, intensively sheared rock, which has lost a large part of its
reliable predictions of the ground conditions ahead of the face original strength.
are essential in order to enable adjustments to be made dur-
ing construction to the temporary support or the excavation
diameter. 3 CONSTRUCTION METHOD
An assessment of the behaviour of the core ahead of
the face provides some indications concerning the mechan- The tunnel was excavated full-face. Squeezing was tackled
ical characteristics of the ground and seems to hold out through a yielding support system consisting of two rings of
promise with regard to ground response predictions (Cantieni sliding steel sets (TH 44/70) lying one upon the other and
2011). connected by friction loops (Figure 3, 2f and 2g).
The present case history investigates the convergence and
extrusion data monitored during construction of the western 4 MONITORING AND DATA ANALYSIS
tunnel of the new Gotthard Base Tunnel, which crosses the
northern intermediate Tavetsch-formation (the so-called TZM The core extrusion was monitored by reverse-head-
formation) and the adjacent Clavaniev zone (Figure 1). The extensometers (Thut et al. 2006). The position of the measur-
aim of this study is to find out whether there is a correlation ing head and the six measuring points of each extensometer
between the extrusion of the core and the convergences of with respect to the tunnel alignment are shown in Figure 2i.
the tunnel and, accordingly, if it would have been possible to The convergences of the opening were monitored optically
predict the convergences solely on the basis of the monitored (Figure 2h).
extrusions. In order to assess the behaviour of the ground along the
tunnel it is necessary to compare deformations that occur
under similar conditions. Therefore, the analysis considers,
as a measure of the extrusion, the strain that develops due to
the advance of the face from a distance of 6 m to a distance of
2 m in respect of each ground interval. The radial displacement
of the tunnel crown is used as a measure of convergence. The
analysis takes into account only the monitored displacements
occurring due to a face advance of 25 m.
Figure 2a shows the axial strain and radial displacements of
the crown along tunnel reaches 1 and 2. The convergences in
tunnel reach 2 exhibit a weak correlation with the extrusions.
No correlation can be observed between the convergences and
the extrusions in tunnel reach 1.
5 CONCLUSIONS
Figure 1. a) Gotthard Base Tunnel: schematic representation of the
longitudinal geological section which includes the squeezing TZM The case history of the Gotthard Base Tunnel shows that
formation (after Kovári 2009). there is a weak correlation between the axial extrusions and
623

Figure 2. (a) Axial strain εy and radial displacement ur of the crown plotted along the alignment of reach 1 and 2, including information
about the actual geology (b, c, d), the tunnel cross-section and the support measures applied (e, f, g) and the monitoring setup (h, i) (from
Cantieni 2011).
ACKNOWLEDGMENTS
The authors wish to thank the AlpTransit Gotthard AG,

Switzerland for the permission to use the data for this research.
REFERENCES
Cantieni, L. 2011. Spatial effects in tunnelling through squeezing

ground. Dissertation ETH No. 19508, ETH Zurich, Switzerland.
Ehrbar, H., Pfenninger, I. 1999. Umsetzung der Geologie in tech-
nische Massnahmen im Tavetscher Zwischenmassiv Nord. In:
Vorerkundung und Prognose der Basistunnels am Gotthard und am
Lötschberg, Symposium Geologie Alptransit, Zurich, Switzerland,
pp 381–394.
Figure 3. Longitudinal section of the yielding support system (after Kovári, K. 2009. Design Methods with Yielding Support in Squeez-
Ehrbar and Pfenninger 1999). ing and Swelling Rocks. In: World Tunnel Congress, Budapest,
Hungary.
convergences of the tunnel. A correlation might have been Thut, A., Nateropp, D., Steiner, P., Stolz, M. 2006. Tunnelling
especially helpful as the convergences changed even without in Squeezing Rock – Yielding Elements and Face Control. In:
a significant variation in the geology, construction method 8th Int. Conference on Tunnel Construction and Underground
or overburden. However, there was no clear correlation that Structures, Ljubljana.
would make it possible to predict convergences with suffi-
cient reliability on the basis of extrusion monitoring alone.
The extrusion data should be evaluated in combination with
other information, such as advance probing.
624

Investigation of the tunnel penetration to the interaction of

soil/rock-machine and its indices of cuttability
Li-Hsien Chen
Department of Civil Engineering, National Taipei University of Technology, Taipei
Yao-Chung Chen & Chee-Nan Chen

Department of Construction Engineering, National Taiwan University of Science & Technology, Taipei
Zhe-Xun Yang
Keywords: mechanical excavation (TBM/RBM), case studies, tunnelling, site characterisation
To speed up construction of mass transit subway and the Meanwhile, in-situ experimental results from shield tunnel-
popularity rate of sewage, Taiwan’s underground excava- ing and pipe jacking construction were used to examine this
tion works, especially for mechanical cutting cases, show an model and it shows a nice agreement between both. From
ascendant tendency. This study presents a generalized solution this analytical approach, a proposed “oval-shaped cutting
for underground geological-mechanical interaction. By using ellipsoid”, including its center (O), area (A), and long/short
dimensional analysis, This model generalizes geological char- axis (ax /by , or ay /bx ), can be used not only to estimate the
acteristics grouped into three categories: (1) brittle (rock-like), functionality and efficiency of cutting machine adopted for
(2) the ductile (soil-like), and (3) brittle-ductile (gravel-like) tunnel project, but also to offer a warning information for
type with respect to two cutting forces: (1) thrust and (2) torque inadequate cutting strategy.
to evaluate their excavation/penetration rate. Furthermore, the
leading cuttability indices can be obtained to enable to assess
the underground excavation.
625

On the generalized thrust system for underground tunneling
Li-Hsien Chen
Yao-Chung Chen
Department of Construction Engineering, National Taiwan University of Science & Technology, Taipei
Keh-Jian Shou
Department of Civil Engineering, National Chung-Hsing University, Taichung
Gou-Lung Lin
Keywords: tunnelling, rock failure, Mechanical excavation (TBM/RBM), case studies, site characterisation
In recent years, the construction projects increase the cases of situations for risk assessment. In addition, the in-situ data of
underground tunneling by mechanical cutting such as tunnel in trust in shield machine (Taoyuan tunneling project) is pre-
the mountain, rapid transit system in the city and sewer system, sented to confirm with. It depicts a favorable agreement for
etc. However, the complex geological conditions make exca- the estimation of thrust in this study.
vating difficulties in Taiwan. This paper presents an analytical The results shows that the total thrust for upper bounds
estimation to meet different mechanical cutting methods and lower bounds are: (401%, 37.8%) and (258.2%, 31.7%)
(tunnel boring machine, shield tunnel and pipe jacking), types compared with normal condition in gravel and weathered
(earth pressure balance, slurry pressure balance, thick-mud), sandstone cases respectively, which normalized boundary val-
and geological conditions (soil, gravel and rock) by gener- ues are (13 × 10−4 ∼82 × 10−4 ) and (0.97∼4.98) for cutting-
alizing their total thrust system. The straight-line thrust is head resistance respectively. It also found that the cutting-
calculated for either wedge- or conical-typed cutters of tun- head resistance take about 28% of the total trust resistance
neling machine. In this generalized work, the upper bound (F = 4773 kN) in the gravel case by taking cutters’ forces into
and lower bound of trust are highlighted for the warning account.
626

Numerical simulation of ice-ring formation in the jointed rock mass around

underground LNG storage cavern using thermo-hydraulic model
So-Keul Chung, Eui-Seob Park & Joong-Ho Synn

Korea Institute of Geoscience and Mineral Resources, Daejeon, Korea
Woo-Cheol Jeong & Taek-Kon Kim

SK Engineering and Construction, Seoul, Korea
Keywords: fluid flow, ground freezing, rock joints, cryo-suction
Experience on the underground storage for the crude oil and This paper focuses on the freezing temperature and penetra-
various types of hydrocarbons proved to be an economical tion length of groundwater investigated by a thermo-hydraulic
means for the storage of LNG. Capitalizing the knowledge coupled analysis and CFD analysis of the groundwater flow
from our past experience, attempts were made to develop the with in the joints at the time of ice-ring formation. The pre-
LNG underground storage system but were not successful due vious research related to the formation of ice-ring showed
to its failure in controlling the temperature and groundwater that the freezing temperature of water flowing through rock
without containment. From the past trials, it was observed that joints are affected by the latent heat generated during freezing,
the thermal stress and freezing of groundwater nearby to the aperture size of the joints and the flow rate of groundwater
cavern generated cracks in the host rock mass. However, these (Jeong et al, 2006).
thermal cracks contributed to induce gas leakage leading to In general as the flow rate of groundwater through rock
the increase in heat flux exchange between the stored LNG joints is not very fast and is controlled by Darcy’s lawexhibit-
and rock mass. Eventually, this causes excess of boil-off ratio ing a laminar flow with a Reynolds number less than 2000.
in the stored LNG. The laminar flow has the following relationship according to
A new technology has been developed for the storage of joint apertures.
LNG in a lined rock cavern by combining the well-proven
technologies and a new concept named “formation of ice-
ring”. The formation of ice-ring, one of the core technologies
in the LNG storage system can be described as a complex
mechanism underlying between the thermal characteristics of where d = joint aperture, µ = fluid viscosity, γ = unit weight,
rock and hydro-geological characteristics of groundwater. The P = difference in pressure, H = difference in hydraulic
process of ice-ring formation as a second barrier impervious head and L = travel length of groundwater through joints.
layer functioning against the leakage of contained LNG can From V = K · i,
be explained via each stage of the construction of a lined rock
cavern. Groundwater is temporarily removed from the rock
surrounding in the cavern during the first phase of the con-
struction. This preliminary de-saturation of the host rock mass
aims to prevent unacceptable hydrostatic pore pressure and ice where K is permeability and i is hydraulic gradient.
formation behind the cavern lining. After the completion of Considering equations (1) and (2), the flow rate of ground-
the excavation of the cavern and installation of the contain- water can be determined by joint aperture and hydraulic
ment system, the LNG is stored in the cavern, thereafter the gradient.
cold front from the cavern starts to propagate immediately. The analysis procedure using eqn (2) and equivalent per-
When the cold front advances far enough from the cavern meability were used for the calculation of the flow rate of the
wall, drainage can be stopped to allow groundwater progres- in-situ rock mass with infilling materials and joints. Table 1
sively to rise up and quickly form a thick ring of ice around the shows the range of the input values and the results of the anal-
cavern. Once the ice-ring formation is completed, the opera- ysis. The lowest freezing temperature of water was found to be
tion of the drainage system stops. The drainage period during −0.233◦ C in case of a fault or fracture zone (permeability of
LNG storage will last for several months or years depending 1.0 × 10−5 m/s and aperture of 5 mm). Therefore, the analysis
on the thermal properties of rock masses and hydro-geological result demonstrates that the ice-ring formation temperature
characteristics of the site. The key technologies of the LNG can be increased from −3◦ C depending on the safety issues.
storage system have been demonstrated through the design, As shown in the Figure 1, ice begins to form at the joint
construction and operation of the pilot plant in Daejeon, surface contacting with the groundwater. And it can be seen
Korea. that the completely frozen area is generated at the joint
627

Figure 1. Distribution of ice in the joint with wide aperture.
Table 1. Analysis results of the large joint considering the real characteristics of rock mass.
Hydraulic Interval of Aperture Permeability Flow Penetration length Freezing

gradient isotherm (m/◦ C) (mm) (m/s) velocity (m/s) of water (m) temperature (◦ C)
1.8 0.3 0.5 1.00E-06 1.80E-06 1.67E-02 −0.055

1.0 1.00E-07 1.80E-07 2.16E-02 −0.072
1.00E-06 1.80E-06 2.53E-02 −0.084
1.00E-05 1.80E-05 3.96E-02 −0.123
5.0 1.00E-07 1.80E-07 2.23E-02 −0.074
1.00E-06 1.80E-06 3.11E-02 −0.104
1.00E-05 1.80E-05 6.98E-02 −0.233
surface first, and the water temperature above 0◦ C also exists factors were observed to be affecting each other too. In case of
in the center of the joint aperture. It can be explained as fol- laminar flow, it is possible to estimate the freezing temperature
lows: Because the groundwater flow is very slow, groundwater considering only two of these three variables as the recovery
contacting with the joint surface having lower temperature rate is affected by the hydraulic gradient and the joint aper-
than freezing one gradually turns into ice. In addition to the ture. The freezing temperature of groundwater in both narrow
increase of fluid viscosity, the flow velocity is much slow. and wide aperture of a joint can be applied with a value higher
However, groundwater may continue to rise through the center than −3◦ C, which was assumed as freezing temperature in
of the joint aperture. the past. However, these numerical analyses were conducted
In summary, the results of numerical modeling demon- in a limited range with several assumptions. Therefore, further
strated that when the ice-ring is formed, the factors that affect analysis with a wide range of varied conditions would be ben-
freezing temperature of the groundwater are the recovery rate eficial to determine the freezing temperature of groundwater
of groundwater and aperture of joints. In addition, these two in the joints of the rock.
628

Underground storage of natural gas and CO2 in salt caverns

in deep and ultra-deep water offshore Brazil
A.M. da Costa, C.S. Amaral & E. Poiate

Technical Adviser to the E&P Director, Senior Consultant & Technical Consultant, R&D Center, Petrobras, Brazil
A.M.B. Pereira
Department of Civil Engineering, Fluminense Federal University (UFF), Rio de Janeiro, Brazil
L.F. Martha, M. Gattass & D. Roehl

Computer Graphics Technology Group, Pontifical Catholic University of Rio de Janeiro (PUC-Rio), Rio de Janeiro, Brazil
Keywords: rock caverns, mine design, numerical modeling, stability analysis, oil reservoir
1 INTRODUCTION other criteria were established, such as the maximum depth

from the top of the rock salt dome and the distance of the salt
The rock salt has negligible porosity when compared to other domes in relation to the oil fields. Based on interpretation of
geomaterials, which ensures excellent sealing to most fluids 3D seismic and 2D seismic it was selected a cluster of salt
and gases, even under high pressures. Rock salt is also subject domes 10 km away from one of the major presalt oil fields
to the phenomenon of visco-plastic creep deformation which in Santos Basin. Figure 1 shows the isopaque map of the salt
develops in the time domain the relaxation of the deviatoric or domes and the vertical position of the caverns in relation to
shear stresses, to the condition of a steady-state equilibrium the top of the salt dome selected.
with constant creep strain rate and can tolerate high levels
of strain without develop structural damage of its mineral
skeleton. 3 GEOMECHANICAL PROJECT OF SALT CAVERNS
This phenomenon can be observed in nature in the sedi-
mentary layers intercepted by salt domes or other structures Due to the large volume of salt dome selected, the distance
associated with the natural movement of salt. Another phe- between the caverns was calculated in order to eliminate the
nomenon associated with the salt rock is the process of group effect. The redistribution of stresses induced by the dis-
self-healing, where cracks and faults are self-healed with time. solution of one cavern will not influence the stability of the
Taking advantage of these physical-chemical and structural neighbor cavern in the cluster. In this condition the simulation
properties of rock salt, caverns developed by solution mining of the structural behavior of the caverns is being conducted by
in salt domes have been used for storage of hydrocarbons and the application of axisymmetric structural models. The evalu-
other products. ation of the structural behavior of the salt caverns is conducted
Considering the great regional thickness and continuity of by applying the finite element method by simulations in the
rock salt in the Santos Basin (a basin of the giant oil fields time domain using the computer code ANVEC developed by
known as Pre-Salt reservoirs), it is under study the strategy Costa (1984).
and technical and economical feasibility evaluation for the use On the geological section, resulting from the interpretation
of underground offshore caverns opened by solution mining. of 3D seismic data, is generated the axisymmetric structural
The Pre-Salt reservoirs in Santos basin are located in water model, which represents the three-dimensional behavior of an
depth varying from 150 m up to 2200 m, being necessary to isolated cavern. For discretization of the rock mass by finite
drill through 2000 m of salt rock (mainly halite, and in some elements is used a total of 13,812 isoparametric quadratic 8
places, tachyhydrite and carnallite) to reach them. nodes elements and 42,017 nodal points. Figure 2 shows the
lithologic section used as the structural geomechanical model.
In the evaluation of the salt caverns stability are adopted
2 LOCATION SELECTION OF OFFSHORE some assumptions based on the experience of the authors
SALT CAVERNS (Costa 1984).
Considering the cavern to operate as a contingency for gas
In the process of selecting potential areas for the development storage, to ensure the production of oil in the event of stop-
of salt caverns by solution mining salt domes were selected ping the flow of gas, it must be kept in most of the time at
in order to minimize the presence of interbedded non soluble minimum pressure. It is considered one contingency per year,
rock layers like shale and anhydrite and also to avoid the more with pressure going from the minimum value for the maxi-
soluble salts, carnallite and tachyhydrite. In addition, some mum value in a period of 30 days during filling and 30 days
629

Figure 1. Isopaque maps of the salt domes and vertical position of
the salt caverns.
Figure 4. Effective creep strain distribution in t = 30 years, at the

minimum pressure and caverns pattern in the salt dome.
from the maximum to the minimum during discharge. The

cavern will be at the minimum pressure for a period of 300
days per year. Figure 3 shows the evolution with time of the
vertical and horizontal displacements of the cavern wall at the
minimum pressure in 30 years of operation.
Figure 4 shows the effective creep strain at the wall of the
cavern after 30 years of operation at the minimum pressure and
the caverns pattern that is possible to be built in the chosen
salt dome.
Figure 2. Lithologic section used as the geomechanical model of

simulation. 4 CONCLUSIONS
Based on these results the caverns will be stable in 30 years of

operation at the minimum pressure of 50% of the initial stress
at the top of the caverns. Considering 12 caverns in the salt
dome it is possible to store a net volume of natural gas of 2.7
billion m3 at the minimum pressure of 19,227 kPa.
REFERENCE
Costa, A.M. 1984. Uma aplicação de métodos computacionais e

princípios de mecânica das rochas no projeto e análise de escav-
ações subterrâneas destinadas à mineração subterrânea.Tese. Rio
de Janeiro: COPPE/UFRJ.
Figure 3. Evolution of cavern closure with time at the minimum

pressure.
630

Qualitative evaluation of the influence of blocky grounds on TBMs

performances in deep Alpine tunnels
A. Delisio & J. Zhao

School of Architecture, Civil and Environmental Engineering, Laboratoire des Mécanique des Roches (LMR),
Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Keywords: back analysis, mechanical excavation (TBM/RBM), rock burst, rock stress, tunnelling
High in situ stresses strongly influence the stability of the The case of the Southern TBM-excavated part of the
excavation face in deep tunnels. When hard rock fails under Lötschberg Base Tunnel represents a very good example of
high acting stresses, it disintegrates and massive rock is such a condition. During construction a major phenomenon
transformed to blocky ground. Strong, instable rock blocks was observed, reported by the workers as “blocks formation”
may appear at the tunnel face and this condition may pre- in front of the TBM head. It consisted of blocks with a predom-
vent normal TBM operations When these instabilities appear, inant slab-like shape loosening from the tunnel face. These
instead of showing a flat shape with clear marks of cutters, blocks jammed conveyor intake, provoked frequent disc and
the tunnel face is irregular. This condition causes high fre- bearing failure, caused extensive wear of the cutterhead and
quency and great magnitude of variations on cutter force. As cutters.
a consequence, high loading and heavy dynamic impacts of This paper introduces the ongoing research on blocky
the cutters and the cutterhead against the hard rock blocks ground tunnelling at the Rock Mechanics Laboratory (LMR)
occur. Due to these effects abnormal cutter ring wear, such of the Ecole Polytechnique Fédérale de Lausanne (EPFL). The
as pitted, chipped and flatted wear appears, and the cutter aim is to show the problem by focusing on the principal aspects
bearing failure is enhanced. Furthermore, extensive cutter- involved in mechanized tunnelling in blocky rock masses and
head vibration and the cutter loading process may lead to to define the main effects of such a condition on machines per-
the instability of the rock blocks at the face. In return these formances. To this purpose, the TBM performances in some
falling blocks may destroy the cutters. To minimize the vibra- “representative” sections of the Lötschberg Base Tunnel are
tion of the cutterhead and the dynamic loading and impacts analyzed. The variation of the machine utilization factor and
on cutters and cutterhead, the thrust force and the cutter- total advance rate with rock mass conditions varying from
head rotational speed have to be decreased below the normal massive to very blocky is studied.
level. Finally, problems can be expected to the muck conveyor The results show a clear correlation between TBM perfor-
system which can be jammed or damaged by irregular rock mances and rock mass structure: the more blocky the rock
fragments. mass, the larger the downtimes, the lower the advance rate.
In this context, the required replacements of the cutters Therefore, retards in the construction of the tunnel and an
and frequent reparations of the cutterhead increase main- increasing of the construction costs must be expected when,
tenance requirements; the TBM utilization factor decreases due to high stresses, a massive to moderately jointed rock mass
while excavation times and costs increase. becomes blocky ground.
631

Impact assessment of water head on stability of a tunnel in India
R.D. Dwivedi & R.K. Goel

CIMFR Regional Centre, CBRI Campus, Roorkee, India
Subject: Site investigation and observation
Keywords: Water problem; Tunnelling; Weak rock; Stability analysis; Rock support; Monitoring; Pressure cells; Tunnel closure
Jammu Udhampur Srinagar Baramulla Railway Link (JUS- wall. It is understood that once the rock pressure is stabilized,
BRL) project is one of the prestigious projects of India to these pressure cells can also provide information of the pres-
connect Kashmir valley of J&K state with rest of the country. sure exerted by the fluctuating water head during and after the
The railway link starts from Jammu and, when completed, will rainy season. Closure studs were also installed to monitor the
travel for 345 km including tunnel route of 100 km to the city tunnel roof and wall deformations. In addition to this, to mon-
of Baramulla on the northwestern edge of the Kashmir Valley. itor the ground water table fluctuation throughout the year,
The route crosses through very difficult Himalayan terrains number of bore holes were also drilled from ground surface
making it an extremely challenging engineering project. The to the tunnel invert level on both sides of the tunnel alignment
project envisages construction of a single broad gauge railway at various offsets. Data of three years have been analysed and
line with maximum speed potential of 100 km/h. The ruling following results were obtained.
gradient was limited to 1 in 100 whereas maximum degree of The maximum water discharge (421-1125 l/s) was recorded
curvature was restricted to 2.75◦ . in the month of August every year whereas minimum value
Jammu to Udhampur section (53 km) of the JUSBRL of discharge (below 25 l/s) was recorded in the month of June
project has been already completed in the year 2005. every year. Pressure cells installed in the problematic water
Udhampur – Katra section of this rail link is 25 km long charged zone, in general, showed increased support pressure in
and involves construction of 7 tunnels having total length of rainy season (July to September) only. However, no deforma-
10 km. The paper deals the water problem faced in 2.5 km tion in walls was recorded, which indicated that the pressures
long Tunnel No. 3 between Udhampur and Katra. Tunnels in are within the capacities of installed supports. The analysis
this section has the rock masses of the Siwalik group like further revealed that there is an increasing trend of support
clay and soft sandstone and Pleistocene to recent river borne pressure at chainage 12831 m and 13125.5 m every year. A
deposits. The D-shaped Tunnel has finished width of 5.2 m and pressure cell installed on the right wall at the chainage 12831 m
height 7.97 m. Tunnel depth or the overburden above tunnel indicated the pressure of 2 kg/cm2 in the month of August
roof varies from 30 m to 40 m throughout the tunnel alignment 2009 and the pressure decreased by about 0.5 kg/cm2 after
between chainage 12200 m and 14700 m. To support the weak the rain was over. It indicates that the pressure contributed by
tunnelling strata, steel rib supports ISHB150 and ISHB200 seepage of rain water is 0.5 kg/cm2 which is also indicated
have been used for roof and vertical legs respectively. by the analysis of water table data which gives the maximum
This tunnel was facing severe water inrush problems dur- water pressure at this chainage as 0.6 kg/cm2 . But, the pres-
ing construction in the stretch of 436 m i.e. between chainage sure cell at this chainage showed increasing trend of pressure
12831 m and 13267 m. To release the water pressure, weep every year. It was considered to be due to resettlement of the
holes were drilled on both the tunnel walls below springing side rock mass comprised of silt with cobles & pebbles and
level during the excavation of tunnel. Even after the construc- clay. It seems that the clay has washed away by running water
tion of the tunnel, water seepage problem remains continued. through drainage holes and the cobles and pebbles have filled
The zone affected by water inrush was supported by double the empty space created by clay behind the wall of the tun-
ribs system. Outer ribs were installed at the spacing of 1 m and nel resulting in increase of lateral pressure. It was therefore
inner ribs at the spacing of 0.5 m from centre to centre. The recommended to divert the water through another weep hole
rest part of the tunnel was supported by single rib supporting nearby and strengthen the area by grouting.
system. Final support of 30 cm thick PCC lining is yet to be The full length paper deals with detailed analysis of the
placed in this problematic zone of the tunnel. instrumented data like annual fluctuation in water table,
To observe the pressure on the steel rib supports by the water discharge through weep holes, pressure cell indications,
surrounding rock, vibrating wire type pressure cells were tunnel-closure data and suggestions for solving the water
installed on the steel ribs at various chainages on the tunnel inrush problem.
632

Development of moisture curing system for tunnel concrete lining
Y. Furukawa & M. Suzuki

Civil Engineering Division Hazama Corporation, Tokyo, Japan
Y. Mitani
Department of Civil and Structural Engineering, Graduate School of Engineering, Kyushu University, Fukuoka, Japan
Subject: Others (tunnel concrete lining)
Keywords: tunneling, case studies, rock support
1 INTRODUCTION
Concrete linings are considered not to be subjected to any load,

since shotcrete and rock bolts are generally used as support
members for mountain tunnels. Since stringent construction
conditions and environmental conditions have been imposed
on mountain tunneling methods in recent years, however, in
some cases the concrete lining is also made to hold a part of
the load and provide support for the tunnel. In such cases the
concrete lining becomes an important aspect as part of the
support structure of the tunnel.
Early deterioration of the concrete lining has been reclas-
sified as a problem in Japan. One of the causes for the Figure 1. System configuration for tunnel (example) AC system
deterioration of concrete lining quality is imperfect curing. schematic diagram.
Form is removed from a concrete lining in the early stages,
within just 15 to 20 hours after the concrete is placed.
2 MOISTURE CURING SYSTEM
Water supplied curing can be considered a perfect curing

method, as far as the curing of concrete is concerned. The
recently developed moisture curing system (AC: Aqua Cur-
tain) is one of extremely effective water supplied curing
methods available as curing method that is capable of sus-
taining film of water formed on the surface of lining concrete
at all times after forms are removed. Figure 1 show a summary
of AC, as well as an arrangement example of the facility.
AC is comprised of a curing sheet, a suction port for air and
curing water, a suction fan for air and curing water, a curing
water supply tube for the upper section of the tunnel, etc.
Figure 2. Relationship between material age and compression
strength according to difference in curing methods of cylinder
3 VERIFICATION OF CURING EFFECTS specimen.
An experimental test of this system was conducted by curing the concrete was compared to verify the effectiveness of the
a concrete lining for an actual road tunnel. As a result it was moisture curing for the lining, it clarified that when compared
possible to ensure that moisture curing of the concrete lining to cases when moisture curing was not used (air curing), the
was achieved. Furthermore, when the compression strength of compression strength was 2.0 to 2.5 times stronger (Figure 2).
633

The response of an argillaceous rock to ventilation: Process

identification and analysis of an in situ experiment
B. Garitte
CIMNE-Universidad Politécnica de Cataluña, Barcelona, Spain
A. Gens
Universidad Politécnica de Cataluña, Barcelona, Spain
Keywords: field measurements, fluid flow, nuclear repository, numerical modelling
1 INTRODUCTION
During the construction and operation phases of a radioac-

tive waste repository, the underground drifts will be subjected
to a more or less intense ventilation period, which could
produce partial desaturation of the rock around the drifts.
Design of future storage facilities for radioactive waste in
deep geological media requires a thorough understanding of
the mechanisms occurring near the installations. This under-
standing is achieved by the parallel development of in situ
tests, mimicking storage solicitations as mechanical loading,
ventilation and heating, and theoretical THM-C formulations. Figure 1. Water extracted in the laboratory drying test. λ is a shape
One the one hand, the exploration of the THM-C formulations parameter of the permeability-degree of saturation relationship.
allows the identification of possible processes involved near
storage galleries. On the other hand, the analysis of in situ the three samples are presented in Figure 1. After 142 days,
tests establishes the validity of the formulations based on the sample A had lost about 150gr from the initial 300 gr. Simu-
comparison of numerical results with data collected during lations were run to calibrate some parameters and to validate
the tests. the formulation. These simulations showed the importance
In order to investigate the effects of ventilation on Opalinus of 1) the value of water permeability and its dependency on
clay in the Mont Terri Underground Rock Laboratory, a 10 m saturation and 2) water transport through vapour diffusion.
long section of a 1.3 m diameter non-lined microtunnel was
sealed off – July 2002 – and a ventilation system was installed –
March 2003 (Mayor & Velasco, 2008). The microtunnel had 3 ANALYSIS OF THE FIELD MEASUREMENTS
been raise-bored in February 1999. Between May to July
2002, a 2 m radius around the test section was instrumented The ventilation system includes an air blow device, with con-
with hygrometers, for measurement of relative humidity (RH), trolled air flow, temperature and air relative humidity on entry
piezometers and extensometers. Several drilling campaigns and exit. This allows the calculation of the water mass bal-
were also run at several stages of the experiment to estimate ance of the test section. The parameters calibrated on the basis
the water content around the microtunnel. of the laboratory drying experiment and the penalty coeffi-
cient used lead to a satisfactory reproduction of the measured
water balance (Figure 2). A parametric study of the in situ
2 PARAMETER DETERMINATION case also showed the important influence of the permeability
value and its dependency on saturation. Vapour diffusion was
The HM formulation used herein is a particular case of the found less significant than in the laboratory experiment but
general THM-C formulation developed by Olivella (1995). not negligible. The reason is probably that in the in situ test,
The parameters determined for the analysis are based on some an inexhaustible water reservoir is available and desaturation
reference works (Bock, 2001 and Gens, 2007) and on some is thus less intense than in the laboratory.
others specific to the VE project. Drilling campaigns were organised before the experiment
A laboratory drying experiment was set up before the start (borehole BVE_82 – parallel to the bedding), after the first
of the in situ VE (Floria et al., 2002). Three Opalinus Clay sam- desaturation period (borehole BVE_85 – perpendicular and
ples were placed on balances in a drying chamber. All sample borehole BVE_86 – parallel) and after the second desaturation
walls, except the top, were isolated in order to create conditions (borehole BVE_104 – perpendicular and borehole BVE_108 –
as close as possible to 1D flow. The measured water losses in parallel). The borehole cores were cut in slices and water
634

observations. Heavy ventilation conditions applied during one
to two years were found to produce a partially desaturated zone
of quite a limited extent, much smaller than thought initially.
The limit of the partially desaturated zone was estimated to
about 25–30 cm and the suction limit is about 2 m (or even
more) after a ventilation period of 20 months.
The interpretation of the experiment has been performed
with the aid of a numerical tool and has allowed the calibration
of important rock parameters. The simulation achieves a close
quantitative agreement with the experiment in many instances
Figure 2. Extracted water from the test section. and showed its ability to explain qualitatively most obser-
vations. A critical analysis allowed the confirmation and/or
identification of a number of relevant desaturation processes
of varying degree of importance. A hierarchical classification
is proposed.
REFERENCES
Bock, H., 2001. RA Experiment. Rock Mechanics analyses and

synthesis: Data Report on Rock Mechanics. Technical Report
2000-02. Mont Terri Project.
Fernández, A.M., Melón, A., Turrero, M.J., Villar, M.V. 2006. Geo-
chemical characterisation of the rock samples for the VE-Test
before a second cycle of drying. Ventilation Test Phase II Deliver-
Figure 3. Water content profiles at several times. able D4.3.5.
Floría, E., Sanz, F.J. & García-Siñeriz, J.L., 2002. Drying test: evapo-
ration rate from core samples of “Opalinus clay” under controlled
content was measured in each slice. This allows plotting the environmental conditions. Deliverable D6, FIKW-CT2001-00126.
water content as a function to the tunnel wall at different Gens, A., Vaunat, J., Garitte, B. and Wileveau, Y., 2007. In situ
times (Figure 3). Measurements of water content loss are sys- behaviour of a stiff layered clay subject to thermal loading.
tematically higher in the bedding direction. The simulation Observations and interpretation. Géotechnique Symposium “Stiff
results added in the two directions for each drilling campaign Sedimentary Clays. Genesis and Engineering Behaviour”.
show that this difference may be attributed to permeability Mayor J.C. and Velasco M. 2008. The ventilation Experiment
anisotropy. Phase II (Synthesis report). NF-PRO deliverable 4.3.18.
In boreholes perpendicular to the bedding plane, the mea- Olivella, S., 1995. Nonisothermal multiphase flow of brine and gas
sured water content exceeds the initial water content (espe- through saline media. Doctoral Thesis, Technical University of
cially in BVE_85). Results from preliminary computations Catalonia (UPC), Barcelona, Spain.
Traber, 2003. Geochemical characterisation of samples from drill
suggest that this abnormally high water content may be a con- core BVE82. Project Deliverable D5b EC contract FIKW-CT2001-
sequence of the stress redistribution around the tunnel and the 00126.
anisotropy of the stress state and of the material stiffness. Traber, 2004. Geochemical characterisation of samples from drill
core BVE85 and BVE86. Project Deliverable D5c&d/D22 EC
contract FIKW-CT2001-00126.
4 CONCLUSIONS
The four years long controlled ventilation conditions applied

to an unlined microtunnel has provided a set of important HM
635

Prediction of convergences in rock tunnels excavated by conventional methods
J.A. González del Álamo

PAYMACotas (BUREAU VERITAS Group), Madrid, Spain
R. Jiménez
ETSI Caminos, C. y P., Technical University of Madrid, Spain
Keywords: tunneling, time-effects, rock rheology, tunnel support, back analysis, case studies
1 INTRODUCTION From Equation (1), a parameter m that represents the

relative importance of the rheological term on the overall con-
The design of tunnels using conventional methods requires a vergence can be defined. However, our predictive capabilities
previous estimate of convergences in the excavated sections. of what parameter values should be employed in the model to
Often, the evolution of convergences with time is also of inter- best represent the future deformations in a real tunnel are still
est; in such cases, simple analytical and/or numerical tools are limited. We have employed the convergence model to back-
not always feasible due to the lack of data (mainly rheolog- calculate the set of parameter values that produce a better fit
ical parameters) or to the (lengthy) times usually involved between predicted and observed convergences for a number
with preparation of the calculation model (usually in 3D) and of case histories. We have also computed the corresponding
processing of the results. Empirical methods can be used to m values, hence extending the initial database presented by
predict the evolution of future tunnel convergences; since they Kontogianni’s et al. (2006).
are simpler and faster to use, they therefore allow to conduct
calculations in almost “real time” so that decisions can be
made on site. 3 RESULTS
This article focuses on the application of such empirical
methods to predict convergences in real tunnels. In that sense, To illustrate the flexibility of this model to predict future tunnel
it is related to previous works (see e.g. Sellner et al. 2002; convergences, we have fitted the model using real data from
Schubert et al. 2007) at TU Gratz who have proposed a soft- Cabrejas tunnel. Results show that the method provides a quite
ware tool (GEOFIT) to conduct a similar task. The empirical good approximation to predict future tunnel convergences. It
method employed is based on Sulem et al. (1987) (see also is therefore expected that, in future steps of this investigation,
Panet et al. (1985)), who considered the effect of the advance it will be possible to improve the quality of predictions by
of the excavation front face and also the influence of the rhe- constructing a more extensive database to be able to select
ological behavior of the rock mass on the evolution of the good a-priori estimates of model parameters.
convergences with time. In particular, we use the formula To that end, we have back calculated the model parame-
proposed by Kontogianni et al. (2006) to back analyze con- ters for a number of case histories. As an illustration of the
vergence data corresponding to case histories with different results (see more details in the paper), Figure 3 shows the
conditions. computed m parameters (i.e., the relative importance of rhe-
ological convergences) vs. the type of support for the case
histories considered. (Lithology is also indicated using differ-
2 METHODOLOGY ent symbols.) Four general types of support (Type I to IV) were
considered (from smaller to greater stiffness). (The paper also
We use the empirical equation proposed by Kontogianni et al. presents other similar analyses of computed m parameters vs.
(2006), in which convergences are computed as: tunnel depth, and computed m parameters vs. rock type.)
Results are further discussed in the paper (see also Gonzalez
del Alamo, 2010) and, for the sake of space, are not reproduced
herein.
where ur is the convergence of a section after time, ti , and at a 4 CONCLUSIONS

distance to the front, xi ; dxi indicates the unknown total con-
vergences produced by the face advance; dti are the unknown This paper presents an application of empirical methods for
total displacements due to rheological effects; and X and T prediction of convergences in tunnels excavated with tradi-
are unknown adimensional parameters that depend on ground tional techniques. The modelling framework employed is the
behavior. equation proposed by Kontogianni et al (2006) (see also Sulem
636

stiffness reduces the influence of rheological convergences in
real tunnels. Clearer correlations have been observed, how-
ever, with the depth of the tunnel (rheological effects tend to
increase as the depth of the tunnel increases), especially for
tunnels constructed in similar lithologies. This emphasizes the
importance of conducting careful monitoring of the evolution
of convergences with time in rock tunnels that also consider
the position of the face at each time. It also emphasizes the
interest in publishing such data, so that additional research
(e.g., more extensive databases and more advanced statistical
analyses) can be performed on this important topic of rock
mechanics. The objective is that, in the future, parameters to
be employed in real cases can be selected prior to tunnel con-
struction, therefore increasing the predictive capability of the
model.
REFERENCES
Figure 3. Computed m parameters vs. support type. Gonzalez del Alamo J. A., 2010. Prediction of Convergence in Rock
Tunnels Considering the Rheological Behavior. MSc Dissertation,
et al. 1987). The fitting process is based on minimizing the ETSI Caminos, C. y P., Technical University of Madrid, Spain.
(squared) errors between observations and predictions of [In Spanish]
the model. Results indicate that the predictive capabilities of Kontogianni V. A., Psimoulis P. y Stiros S. 2006. What is the con-
the model are reasonable, and we presented an example of a tribution of time-dependent deformation in tunnel convergence?.
Engineering Geology 82 (2006) 264–267.
tunnel section in which the model closely reproduces the evo- Panet M., Guenot A., Sulem J. 1985. A new aspect in tunnel closure
lution of observed convergences (both with respect to time interpretation. 26th US Symposium on Rock Mechanics, Rapid
and with respect to distance to the face) for a case history in City.
Spain. Schubert, W., Grossauer, K. Lenz, G. 2007. Innovative approaches
In addition, to construct an increased database with real in monitoring data evaluation and interpretation. In: CD-ROM
tunneling case histories, we have back analyzed the model Ribeiro et. al. (eds). Proc. of the Congress of the 11th Interna-
parameters for a number of unpublished case histories in Spain tional Society for Rock Mechanics; specialized session. Lisbon,
and also for several case histories published in the literature. Portugal.
Such results have been employed to study the influence that Sellner, P., Grossauer, K., Schubert, W. 2002. Advaced analysis and
different factors have on the relative importance of rheolog- prediction of displacements and system behaviour in tunneling.
Proceedings of the ISRM, Eurorock 2002, Portugal, 25–28.
ical convergences with respect to convergences produced by Sulem, J. Panet, M. Guenot,A. 1987. Closure analysis in deep tunnels.
the advance of the face in real tunnels. Results are still not International Journal of Rock Mechanics and Mining Sciences &
completely conclusive in this case and, for instance, it was not Geomechanics Abstracts, Vol. 24, No. 3, pp 145–154.
possible to completely reproduce previous (reasonable) sug-
gestions (see e.g., Kontogianni et al., 2006) that the support
637

Reliable method for controlling tunnel convergence based on analytical results and
monitoring results in squeezing rocks
H. Hayashi, K. Aoki & H. Yoshioka

Department of Environmental Engineering, Osaka Institute of Technology, Osaka, Japan
K. Yanagawa
Kanazawa Office of River and National Highway, Ministry of Land Infrastructure and Transport, Ishikawa, Japan
N. Kitou
Nanao Tunnel branch office, Nishimatsu Construction, Ishikawa, Japan
Keywords: field measurements, monitoring, stability analysis, tunneling, weak rock
After excavating a tunnel in squeezing rocks, it is important to successfully by immediately closing the cross section of
close its cross section immediately to prevent the rocks from the tunnel. This result was verified by numerical analysis
being relaxed. However, because of the immediate tunnel ring (Fig. 2).
closure of a tunnel cross section, it is extremely difficult to 2) From the execution example, it was found that the immedi-
estimate an appropriate supporting scale and cross-section ate closure of the cross section is effective in securing the
closure timing corresponding to a ground condition. stability of the tunnel cross section in the rock mass whose
Currently, no guideline for the supporting scale and closure competence factor is below 1.0.
timing has been established. In addition, few studies describe
the mechanism and influences of difference in tunnel closure
timing on the tunnel surrounding.
Accordingly, this paper shows an example of the construc-
tion of the Nanao tunnel on the basis of information obtained
by feeding back measurement results about deformation dur-
ing the tunnel excavation to determine supporting scale and
closure timing at a subsequent period (Table 1). This paper
also considers the effect of deformation suppression during
implementation of early closure and the validity of the clo-
sure timing. Further, this paper considers influences of the
difference in tunnel closure timing on the tunnel surrounding
by using elastoplastic three-dimensional numerical analysis Figure 1. Excavation method.
using FLAC3D (Fig. 1).
Consequently, the following results are evident.
1) During the information-oriented execution of the Nanao
tunnel, the stability of the cross section was obtained
Table 1. Execution rank in Nanao Tunnel.

Rank 1 Rank 2 Rank 3 Rank 4
Rock bolts (m) 4 4 4 4

Shotcrete (mm) 200 250 250 250
(σck = (σck = ) (σck = ) (σck = )
18 N/mm2 ) 18 N/mm2 ) 18 N/mm2 ) 36 N/mm2 )
Invert strut – – ◦ ◦
Steel supports H-150 H-200 H-200 H-200
Distance from 8 8 6 5
face to invert (m)
Figure 2. Percent strain.
638

Probabilistic prediction of squeezing in tunneling under high-stress conditions
R. Jiménez
ETSI Caminos, C. y P. Technical University of Madrid, Spain.
D. Recio
Acciona Infraestructuras I+D+i, Madrid, Spain
Keywords: case studies; problem rocks; tunnelling; weak rock
1 INTRODUCTION Number N (i.e., the Q value computed for SRF = 1) as an

indicator of “capacity” (N has a logarithmic scale, and it is
Rock squeezing typically occurs in weak overstressed rock therefore convenient to perform the analysis working with
masses, producing difficulties during construction associated log N ); and H (in meters) as an indicator of “demand”. (Note
to excessive convergences and/or to the need for non-standard that we also use H · B0.1 as an alternative demand indicator; in
excavation and support methods. In this work, we use sta- that case, we would have X = [1, log N, H · B0.1 ] as the input
tistical classification to define a new empirical equation for parameter matrix.) That is, X = [1, X 1, X 2] = [1, log N, H ] in
squeezing prediction (i.e., squeezing vs. no-squeezing) and the first model and X = [1, X 1, X 2] = [1, log N, H · B0.1 ] in
for estimation of the associated uncertainties (in other words, the second. Similarly, we construct a vector Y of squeezing
probabilities). We work with an extensive database case his- observations (of dimensions N × 1), where y = 1 corresponds
tories available from the literature, in which case histories to squeezing and y = 0 otherwise.
were selected so as to assure that the data are reliable (using The result of such analysis is presented in Figures 1 and 2,
data from well-documented case histories only); original (i.e., where we have also included additional lines (in light grey)
avoid several uses of the same case history); and indepen- for other probability values. (For comparison, Figure 2 also
dent (i.e., only using data from direct measurement and/or presents a dashed line with the criterion for squeezing predic-
estimation). tion proposed by Goel et al. (1995) based on their research on
squeezing in tunnels in the Himalayas.)
2 LINEAR CLASSIFICATION TO PREDICT Based on such results, as expected, the squeezing proba-
SQUEEZING bility increases as the depth of the tunnel increases (or, for
constant depth, as the tunnel span B increases). Similarly, the
The methodology is based on the use of a linear classifier for squeezing probability increases as the quality of the rock mass
statistical analysis of the database of selected case histories. decreases. As an illustration, for a tunnel with H ≈ 200 m, the
The objective is to assign labels to observations based on the
computed conditional probabilities of the class labels Y given
the observed data X , P(Y = y|X = x) = µ(x)y (1 − µx))1−y .
(In our case, Y is a random variable with values of 1 for squeez-
ing and 0 for no squeezing). The parameter µ is assumed
to depend on x as ν(x) = θ T x, and ν is transformed into a
probability scale by means of the logistic function.
Maximum likelihood estimates of the θ parameter vector
are “learned” using the Iteratively Re-weighted Least Squares
Algorithm (IRLS) (see e.g., Wasserman 2004), that leads to
the following solution:
where X is the matrix of (observed) input parameters of the

model; Y is the vector of (observed) outcomes; and W is a
diagonal matrix with weights that are a function of θ.
Following Goel et al. (1995), the classifier is developed using Figure 1. Probabilistic squeezing predictions lines (log N –H
an input vector of observations composed of the Rock Mass space).
639

squeezing criteria follows the common threshold of 1% (nor-
malized) convergence, and no intend is made to model the
time-dependant behavior of the ground. (We can only expect
that predicted squeezing cases produce “large” convergences
in a “relatively short” time, thereby producing difficulties dur-
ing construction and/or operation of the tunnel.) Our approach
is based on the use of statistical analysis, using the theory of
linear classification, of a database of well-documented case
histories compiled from the literature.
The newly proposed squeezing classifier shows that (as
expected) the probability of squeezing increases as the depth
of the tunnel increases; similarly, the squeezing probability
has been found to significantly increase as the quality of
the rock mass as measured by the Rock Mass Number N
(i.e., the Q value computed for SRF = 1) decreases. Results
also show that the tunnel span (as considered by the prod-
uct H · B0.1 ) does not seem to have such a strong influence
on the squeezing behavior in this case. In addition, the prob-
abilistic linear classifier presents an unbiased behavior, and
it also produces reasonable estimates of squeezing probabil-
Figure 2. Probabilistic squeezing predictions lines (log N –HB0.1 ity so that, for instance, no observation is miss-classified for
space).
squeezing (or not squeezing) assignments with more than 90%
computed probability of squeezing is of approximately 0, 8 probability, at the same time that produces fewer cases of
for a rock mass with N ≈ 0.1, and of approximately 0, 2 for a miss-classification than previously available criteria.
rock mass with N ≈ 1. Finally, it is important to indicate that the presented solution
Results also show that the predictor is unbiased (i.e., cannot be considered as “final”; indeed, it should be improved
it presents approximately the same number of miss- as more case histories are included in the training database.
classifications), and that also show that equi-probability In addition, the reader should note that the newly proposed
squeezing lines seem to provide reasonable results so that, criteria for estimation of (probabilities of) squeezing occur-
for instance, no miss-classification is produced outside the rence is an empirical method and, therefore, it is only valid
[10%,90%] probability range. In addition, the predictive capa- within the range of the variables for which there is data avail-
bilities of the model do not seem to improve much when the able. That means that, at least until more case histories can
dimensions of the excavation are considered. (This may be due be included in the database, our empirical squeezing criterion
to particularities of the database and more research is needed.) should be primarily employed for tunnels under high-stress
Furthermore, by comparison in Figure 2, the squeezing criteria tectonic conditions, and that it should not be employed for
by Goel et al. (1995) presents a total of 13 (11-2) miss- tunnels deeper than, say, approximately 600–800 m, since that
classifications with this database, as eleven observations are is the maximum depth range of observations available within
assigned a no-squeezing label when they actually presented the database.
squeezing, and two observations are assigned a squeezing
label although they did not present squeezing. (Note also that
the “bias” could be unconservative in this case.) REFERENCES
Goel, R. K., Jethwa, J. L., & Paithankar, A. G. 1995. Indian expe-

4 CONCLUSIONS riences with Q and RMR systems. Tunnelling and Underground
Space Technology, 10(1): 97–109.
This paper presents new solutions for empirical prediction Wasserman, L. 2004. All of Statistics: A concise course in statistical
of squeezing conditions in rock tunnels. The considered inference. Springer texts in statistics. New York: Springer-Verlag.
640

Shaft sinking and lining design for a deep potash shaft in squeezing ground
G. du T. Judeel
Mining & Engineering Technical Services, Shaft Sinkers (Pty) Ltd, Sandton, South Africa
G.J. Keyter
GeoStable SA cc, Randburg, South Africa
N.D. Harte
Mining & Engineering Technical Services, Shaft Sinkers (Pty) Ltd, Sandton, South Africa
Keywords: numerical modelling, rock support, stability analysis, weak rock
1 INTRODUCTION The initial design of the shaft liner was based on a ground
pressure assuming hydrostatic in situ ground stresses that are
The main challenges of shaft sinking in weak sedimentary equal to the overburden pressure. Cast iron tubbing and con-
rocks in potash deposits are that of dealing with groundwa- crete backing thicknesses were selected to withstand such
ter when excavating through water bearing formations, creep external pressure with an adequate factor of safety.
or squeezing in weak rock, and the build up of significant A limit state design approach was later adopted as design
swelling pressures behind the shaft lining in anhydritic shales verification. The main limit state design load case considered
and marls when brought into contact with water. For these rea- included the rock load plus the maximum external hydrostatic
sons, thick cast iron tubbing linings with concrete backing are pressure with swelling pressures in swelling rocks added to
often required in certain sections of shaft excavated in such this load case. For this case study though, an extensive pro-
rock formations. gramme of grouting planned meant that external hydrostatic
Long term creep or squeezing in excavations in weak rock is pressures could be excluded from the liner design. Swelling
well known, the usual assumption for very weak rock masses pressures were also only introduced in the numerical models
is that they are incapable of sustaining significant differential to check whether further work will be required to quantify
stresses and that rock mass failure will occur around an exca- swelling potential of rock formations in which the shaft will
vation in such ground until the horizontal ground stress and be constructed.
vertical overburden stress have been equalized. For design, The main design criteria were as follows:
any external groundwater pressures acting on the shaft lining
• The minimum factor of safety against buckling of the cast
must also be taken into account. Significant swelling pres-
iron tubbing had to be more than 2
sures can furthermore develop in anhydritic shales and marl
• Working stresses in the lining had to remain below permis-
when confined during the swelling process, this also need to
sible stresses in compression and tension.
be considered during design.
• Working stresses in the concrete backing were similarly not
The approach adopted in designing the shaft lining in
allowed to exceed permissible stresses at any stage.
squeezing ground at a potash mine in Russia, including numer-
ical modelling verification of liner performance, is presented 3 NUMERICAL MODELLING
as case study.
A model calibration using unfactored material strengths was
2 LINER DESIGN carried out to ensure accurate modelling of the lining response
to ground loading. A typical calibration result is shown in
Parameters reviewed during liner design included shaft geol- Figure 1. From these results, limiting circumferential axial
ogy, intact rock and rock mass strength and modulus, ground stress σN to less than 80% of the characteristic compressive
squeeze expected, groundwater regime, the swelling poten- strength of the cast iron material was found to be critical to
tial of anhydritic shales and marls, horizontal stress, struc- ensure elastic stability of the cast iron material in compression.
tural properties of specified liner components, construction As next modelling step, characteristic ground reaction
sequence, and watertightness of the liner. curves were developed for use as input in 2D models of the
Geotechnical parameters and numerical modelling work various design sections. Ground reaction curves like Figure 2
described relate to specific shaft sections in some of the were obtained from models in which shaft closure as a func-
weaker sedimentary rock formations where severe squeeze tion of internally applied pressure as well as incremental shaft
was expected. Sections of shaft in halites, anhydrites and sinking, was studied at design sections.
sylvinites were specifically excluded from this part of the As final design verification, the cast iron tubbing with
design work. concrete backing was modelled as a 2D ring on a horizontal
641

Figure 3. Support capacity plot for 90 mm tubbing.
Figure 1. Typical axisymmetric model calibration result.
In the third model, a 4 MPa swelling pressure was added to
the hydrostatic ground load already acting on the cast iron tub-
bing concrete backing liner. The resultant loads and moments
in the 90 mm thick cast iron tubbing in the model indicated
that the liner will become overstressed by swelling pressures
of this magnitude at this depth. This final check demonstrated
the need for further field and laboratory test work to better
quantify the swelling potential of the weak sedimentary rocks
that will be encountered during shaft sinking, especially at
sections where the shaft liner will already be performing near
maximum capacity due to severe ground squeezing.
Figure 2. Typical ground reaction curve.
4 CONCLUSIONS AND FURTHER WORK
cross-section through the shaft at design sections in squeezing
The approach adopted in designing the lining of a shaft
ground. Three verification models were developed for each
in squeezing ground at a potash mine in Russia has been
design section. In the first model, using the internal pressure
described. Detailed numerical modelling of expected ground
reduction technique, shaft sidewall convergence was modelled
reaction-support interaction resulted in design changes being
up to the point where the cast iron tubbing with concrete back-
adopted in the final liner design.
ing was installed, to check temporary support requirements in
Notwithstanding this, further work is required to better
the excavated shaft below the tubbing.
quantify the swelling potential of sedimentary rocks that will
As example, for the design section in aleurite at 370 m
be encountered during shaft sinking at this site given the poten-
depth, the modelling showed that temporary support spec-
tial impact of swelling pressures on the long term performance
ified will provide sufficient resistance to adequately control
of the shaft liner specified.
convergence of the shaft sidewall during excavation. The mod-
It is anticipated that extensive grouting will be required
elling however also showed that the 400 mm thick concrete
below shaft bottom during sinking operations, partly to control
backing will become overstressed during construction as the
groundwater inflows during construction but also to ensure
full ground load transferred onto the shaft liner. Increasing
that groundwater is pushed far enough away from the liner to
the strength of the concrete to 50 MPa resulted in forces and
prevent the build up of external water pressure against the cast
moments in the concrete backing that fell within permissible
iron tubbing in the long term. Grouting will also consolidate
limits and a 50 MPa strength for the concrete backing was
the ground through which the shaft is being sunk, thereby
therefore adopted in the final design.
reducing invert heave as well as improving overall sidewall
In the second model, a hydrostatic ground load was applied
stability. In ground with high swelling potential, the effective-
to the cast iron tubbing concrete backing liner. This model
ness of grouting as a method of reducing swelling pressures
showed that the 60 mm thick cast iron tubbing specified in the
also has to be investigated.
initial design was unable to cope with ground stresses in the
Instrumentation and monitoring is required during shaft
aleurite at 370 m depth due to long term squeeze. A 90 mm
construction, to check the adequacy of the specified shaft liner,
thick cast iron tubbing was therefore selected instead, a better
thereby verifying assumptions made during design.
long term performance was obtained as shown in Figure 3.
642

Feasibility to control the permeability of clay

barriers using geophysical methods
K. Kadiri
Tuc Rail, Belgian Rail Engineering
C. Schroeder
Laboratoire J. Verdeyen, Building Architecture and Town Planning Dept (BATir), Université Libre de Bruxelles
J.C. Verbrugge
Laboratoire J. Verdeyen, Building Architecture and Town Planning Dept (BATir), Université Libre de Bruxelles
Service Mécanique et Construction, Gembloux Agro-Bio Tech (GxABT), Université de Liège
Keywords: field measurements
The economic progress that most industrialized countries continuous and not destructive. The principal objective of the
have experienced in the twentieth century as created a sig- work presented here is to test the feasibility to use resistivity
nificant problematic in both waste and water management to measurements to control the permeability and the resistivity
protect the environment. In order to satisfy the demand of in laboratory. Two clays of different mineralogy were tested
environment protection, a clay barrier is mostly used. This in a laboratory experiment using the adapted oedometer. The
barrier is designed to comply with permeability and thick- results demonstrate that under given conditions, a linear rela-
ness requirements. Nowadays, the permeability measurements tionship exists between permeability and resistivity for the
of the barrier are expensive, destructive and in particular two tested clays. An extension of Archie’s law to soils with
long. That is why these measurements and the earth works partly conductive matrix is also proposed and shows results in
progress in the site are incompatible. Meanwhile, geophysics agreement with the experimental data.
methods used for resistivity measurements of soil are fast,
643

Evaluation of the stability of a rock septum between two large derivation tunnels
under high hydrostatic pressure
M.A. Kanji
Keywords: case studies, stability analysis, tunneling, fluid flow, rock mass
1 INTRODUCTION
The two twin derivation tunnels of a large hydroelectric project

in Brazil were 1 km long, 14 m wide and 16 m high in size
each one, 20 m apart one from the other, separated by a rock
septum. The failure of the gate at the inlet of one of the tunnels
promoted the sudden draw down of the 180 m high reservoir,
causing heavy erosion of the tunnel. The maximum velocity of
the water was calculated as being of some 50 m/s. The erosion
along the first 50 m of the tunnel was of some 8 m at the roof,
15 m at the floor, 8 m at the left wall and up to 4 m at the
right wall, as shown in Figure 1. As a result, the rock septum
between the tunnels was reduced to only 12 m. Figure 1. Shape or the eroded section of the tunnel after rapid
The Project intended to build a concrete plug at the mid- reservoir draw down, under a 50 m/s velocity flow.
dle portion of the damaged tunnel and then fill the reservoir
again, but maintaining the neighbor tunnel closed and empty
for some time, before it also receives a concrete plug. The
problem consisted in the assessment of the strength and sta-
bility of the rock septum and its stability when subjected to a
hydrostatic pressure of 18 MPa, which could cause the failure
of the rock wall, leading to a disaster with the emptying of the
reservoir again.
The paper describes the studies performed, comprising
geological and geomechanical characterization of the rock
septum, adoption of parameters for the rock mass, finite ele-
ment analysis under diverse conditions, the conclusion derived
and the treatment employed.
2 GEOLOGIC SETTING
The geologic characterization of the highly eroded stretch of Figure 2. Analysis with the program PHASE2, showing “Strength
tunnel was made by direct geologic mapping and by analysis Factors” at the rock septum lower than one.
of the geomechanical mapping during construction. The rock
is basaltic flow consisting of sound dense basalt except at the
upper one third of the tunnels where a layer of basaltic breccia The project designer run elastic FEM analysis with the
occurs. SAP and with the PHASE2 programs, utilizing the follow-
There are vertical joints occur at angles normally to the ing parameters: c = 1 MPa and φ = 35◦ for the breccia and
tunnel axis, plus horizontal joints allowing the formation of c = 2 MPa and φ = 50◦ for the dense basalt, and considering
blocks which could fail under the lateral pressure. The UCS of Ko = 1.
the rock is estimated as about 75 MPa for the Basaltic Breccia The result of the SAP analysis indicated that high compres-
and 100 MPa or more for the Dense Basalt. sion occurs at the walls of both tunnels and in the rock septum
in between, and tension occur at the roof of the eroded cavern.
3 ANALYSIS AND COMPUTER PROGRAMS USED The result of the PHASE2 analysis is presented in Figure 2,
showing that the Strength Factor (SF, representing the ration
Different analyses were run, with different programs and between the UCS of the rock and the maximum stresses) is 1
parameters, requiring judgment for their comparison. or less than 1 in the major part of the septum.
644

Table 2. Results of the analysis with the ADINA program under
parametric variation, in percentage of rock septum thickness that
undergoes plastification.
c = 1 MPa c = 2.5 MPa c = 2.5 MPa

φ = 30◦ φ = 30◦ φ = 35◦
Uplift Uplift Uplift
Ko No Yes No Yes No Yes
1 100 100 5 75 0 5
2 100 100 10 10 5 75
Furthermore, the cases varied with respect to Ko (equal 1 or

2) and with or without uplift pressure applied in hypothetical
Figure 4. Model of the ADINA analyses and results for maximum horizontal joints in the rock mass (with triangular distribu-
stresses for conservative parameters and Ko = 2. tion). The interpretation of the analysis took into account the
amount of plastification that occurs, expressed in percentage
Both analyses were also run with improved parameters of the distance between the tunnels.
showing much better condition, which clearly indicates the The results are summarized in Table 2.
importance of the adequate choice of parameters. It can be seen that the uplift has a predominant role and
For the present study FEM analysis with the elasto-plastic that if it is avoided the Ko value is not so much relevant. The
computer program ADINA were run by Prof. M. Bucalem variation of the rock friction angle taken as 30◦ or 35◦ is also
(University of S. Paulo), in six different phases simulating not very significant. Due to the reasonable good quality of the
the initial condition, the tunnel excavation, situation after ero- rock seen in the cavern walls and by the rock cores of drillings
sion (Phase 4), partial filling with concrete in the bottom (see previously made, it was decided to accept the greater cohesion
Figure 1) and finally hydrostatic pressurization with 18 MPa value considered for the rock mass.
(Phase 6).
A first set of analysis were run with conservative parame-
ters (c = 1 MPa and φ = 30◦ ). Moreover, although there were 4 CONCLUSION
not determinations of the rock state of stress, and taking into
account the valley is deep and narrow, it was decided to run The practical recommendation was then to avoid the uplift
the analyses both with Ko = 1 and 2. pressure in that stretch of the tunnel strongly eroded and rec-
A typical example of the model and results for maximum ommend a minimum cover of 10 cm of shotcrete reinforced
stresses at Phase 6 is shown in Figure 3 for the Ko = 2 case. with steel fibers and shotcrete ribs rings of at least 25 cm
The analyses have shown that a tension zone occurs in the of thickness around the tunnel, which was accepted by the
rock septum but it decreases with pressurization of the tunnel; contractor.
there are zones of low compression or even tension at the After the installation of the shotcrete lining and ribs, and
tunnel walls; an inclined zone of high compression occurs in construction of the concrete plug in the middle of the tun-
the rock septum. The ratio of rock strength to the maximum nel, the other derivation tunnel was closed and the reservoir
stress was less than 1 for both Ko 1 and 2. was filled, with progressive pressurization of the tunnel herein
Recognizing that the above parameters are very low, it was considered. The full filling of the reservoir was successful and
decided to run analysis with parametric variation, selecting the project has being operating normally for several years.
parameters limited by somewhat conservative and the other
optimistic assumptions.
REFERENCES
The adopted ranges of values for the Dense Basalt and the
Basaltic Breccia were the following:
Hoek, E. & Brown, E.T. (1980). Underground excavation in rock.
c = 1–2.5 MPa; φ = 30◦ –35◦ ; E = 9–10.5 GPa; φ = 0. The Inst. Min. Metall., London.
645

Geological assessment of underground storage cavern sites
G. Kannan, R. Rath & A. Nanda

Engineers India Limited, New-Delhi, India
P. Vaskou
Geostock, Rueil-Malmaison, France
Subject: Underground storage of petroleum, gas, CO2 , and nuclear waste disposal
Keywords: rock caverns, geological model, BHTV, dyke alignment, back analysis, critical joints, support optimization
Engineering geology plays a key role in planning, design and The paper presents two case studies of underground storage
construction of underground cavern storage facilities. Dur- cavern sites:
ing investigation, geological setting of the site is assessed
through an integrated investigation approach such as satel- Case1:
lite image lineament studies, seismic and electrical resistivity During initial investigation of an underground rock cavern
surveys, geological surface mapping, core drilling, bore hole storage site, a major geological feature (a mafic dyke with
tele viewer etc. Based on the above assessment, a geological hydrothermal alteration zone on either side) of about 50 m
model of the site is developed highlighting major geological width was reported. Though, no specific surface exposure
features and discontinuities. The geological model along with was available, based on the interpreted orientation reported
other engineering geological parameters forms the basis for in two widely spaced boreholes, alignment of the dyke
design of the underground facilities such as access tunnel, was assumed for basic engineering purpose. However, dis-
water curtain tunnel, shafts and the main storage caverns. The position of the dyke along with its attitude was planned
layout is finalized in such a way that major geological features to be ascertained through a mandatory investigation cam-
are as far as possible avoided. However, limited unavoidable paign prior to undertaking the detailed engineering of the
geological features which are having the potential to inter- facilities.
sect the underground facilities are predicted in advance and During the mandatory investigation stage, the dyke was
identified as probable geological hotspots. studied in detail by undertaking two more core holes and
The geological assessment of the site is made not only usage of BHTV so as to estimate the thickness of hydro
during investigation but also corroborated and updated dur- thermal alteration zone. Thus, this stage wise investigation
ing construction. In case of cavern storage project, relatively approach helped in scheduling the design stages viz. initial
smaller dimensioned water curtain tunnels, which lie above the investigations for basic design and bid engineering purpose
main storage cavern, are always excavated in advance before and mandatory investigation for detailed design and execu-
excavation of the main large storage caverns. This water cur- tion purpose. Thereby, the geological model of the site was
tain tunnel acts as a pilot tunnel and helps assessment of the finalized along with all the major features/discontinuities and
likely geological setting in advance. As part of the water cur- the probable geological hotspots were identified.
tain system, large numbers of boreholes are drilled from the The site is presently under construction and excavation of
water curtain tunnel and are hydraulically charged to main- the water curtain tunnels is underway. During the excavation
tain a saturated rock mass surrounding the storage caverns. of water curtain tunnels the dyke zone was probed/cored in
Selected boreholes from the water curtain system are cored or advance. The requisite pre-grouting, applicable construction
logged using Bore Hole Tele Viewer (BHTV) or Optical Bore- sequence and special design support requirements of the water
hole Imaging (OBI) revealing additional information about curtain tunnel were finalized. Further based on above active
the geological and hydro-geological setting above the cav- design approach, the geological and geotechnical model is
ern. Thus a 3D geological (structural) model of the site is proposed to be updated and design and construction of the
updated through an active design process with focus on geo- Cavern will be updated and finalized.
logical hot spots intersecting the cavern. During construction
of caverns, a careful approach is adopted along these hot spot Case 2:
segments. These adverse rock mass conditions are probed, In case of another underground storage site, which is near-
assessed and supported in accordance with special design ing completion, a predictive geological model of the site was
supports. prepared based on investigation data and further updated dur-
The geological model is further updated during excavation ing construction of water curtain tunnel and water curtain
of cavern heading and becomes useful for excavation of bench- boreholes through an active design approach. The model was
ing, for verification of the rock support in heading and in further updated during excavation of the top heading of the
further optimization of rock support during bench excavations. cavern. Prior to undertaking the bench excavation, a detailed
646

back analysis was performed to validate the applied rock List of Keywords
support in the top heading of the cavern and to further opti- • Rock caverns
mize the rock support requirements for the planned bench • Geological model
excavations. The back analysis included detailed unwedge • BHTV
analysis using the available joint data derived from the exca- • Dyke alignment
vation mapping of the top heading. The typical wall wedges • Back Analysis
which could affect further bench excavation and the associated • Unwedge analysis
critical joints configurations were identified in the unwedge • Critical joints
analysis. Further, the geotechnical monitoring data collected • Support optimization
during the heading excavation was used to verify the design • Geotechnical monitoring
assumptions using Finite element analysis. • Finite element analysis
647

Development of low speed self-rotating nozzle system for drilling
A. Kizaki
K. Yokoi
Aichi Prefectural Government, Japan
K. Sakaguchi & K. Matsuki
Keywords: mechanical excavation (TBM/RBM), mining, tunneling, blasting, bolts and anchors
1 INTRODUCTION
Waterjet technology has been expected for drilling small

and branch holes, since waterjet does not require weight for
drilling. However, the rotational speed of the nozzle system of
conventional self-rotating nozzle system is too large to drill the
rock. Since a sufficient duration of impingement is required
for the pressure of the waterjets to penetrate into rock, a new
nozzle system for drilling with smaller rotational speed of
nozzle system has been required.
In this paper, a self-rotating waterjet nozzle system that has
a vane pump for braking the rotational speed of the nozzle
system was proposed, and a prototype of a nozzle system was Figure 1. Self-rotating nozzle system for waterjet drilling with vane
developed. pump.
2 NOZZLE SYSTEM WITH VANE-PUMP TYPE

BRAKE SYSTEM
Figure 1 shows the self-rotating nozzle system for waterjet

drilling with the vane pump. The nozzle system consists of
a conventional self-rotating nozzle system and a vane-pump
type brake system. The diameter and the length of the nozzle
system are 30 and 155 mm, respectively.
To change the brake torque of the vane-pump type brake
system, two choke valves were situated in the flow line that
was connected between the discharge and suction ports in the
vane pump.
To clarify the effect of vane-pump type brake system on the
relation between the rotational torque and the rotational speed
of the nozzle system, a measurement of the rotational speed Figure 2. Relation between rotational torque and rotational speed
of nozzle system for various choke valves.
of nozzle system was conducted.
4 CONCLUSION
In this paper, a self-rotating waterjet nozzle system that has
Figure 2 shows the relation between the rotational torque and a vane pump for braking rotational speed of nozzle system
the rotational speed of the nozzle system for various choke was proposed, and a prototype of a nozzle system was devel-
valves. The rotational speed of the nozzle system obtained oped. The results of the measurement of the rotational speed of
for the nozzle system with the vane-pump type brake system nozzle system show that the developed vane-pump type brake
was much smaller than that obtained for the nozzle system system is effective for decreasing the rotational speed of the
without the brake system. In addition, the rotational speed self-rotating nozzle system.
of the nozzle system was decreased with the diameter of the
choke valve. Accordingly, it is concluded that the developed
vane-pump type brake system is effective for decreasing the
rotational speed of the self-rotating nozzle system.
648

Challenges and progress designing deep shafts and wide-span caverns
C. Laughton
Laughton Associates, Austin, TX, US
Subject: Underground Research Laboratories
Keywords: blasting, bolts and anchors, monitoring, risks and hazards
During the latter part of the 20th century, a number of under- Mine located in Lead, South Dakota. The mine produced gold
ground research laboratories (URL’s) were built at depth. for over one hundred years and contains over five hundred
These facilities are accessed via deep shafts or through- kilometres of tunnel and vertical shafts that provide access to
mountain tunnels. As a part of these facilities, large-span depths in excess of two kilometres. At the DUSEL site various
caverns have been constructed. Within these caverns par- engineering and science communities interested in conduct-
ticle detectors, with multiple layers of detection systems ing research underground will be able to take advantage of
mounted around them,have been installed to study the behav- rehabilitated mine excavations (shafts, winzes and decline tun-
ior of sub-nucleic particles such, as the neutrino, proton, and nels) and installed infrastructure to support access to depth and
astrophysics phenomena. provide the requisite operational services.
These sophisticated detectors were necessarily built at depth The paper will review various excavation engineering con-
underground, where rock overburden provides the earth cover cepts being promoted by physics collaborations around the
needed to shield the sensitive detection systems from the dele- world and discuss the key rock engineering criteria by refer-
terious effects of cosmic radiation. Such facilities have been ence tocase history data sets, with an emphasis place on setting
built and operated successfully in Asia, Europe and North realistic design and performance expectations for laboratory
America. In particular, research performed at the Homestake construction at depth. In addition, the paper will review expe-
Mine in the United States of America and the Kamioka Mine rience gained in the access, design, construction, installation,
in Japan, have led to major advances in the fields of particle operation, and decommissioning of deep laboratory facilities
and astro-physics. The value of these experiments was recog- with a specific focus on key design aspects of building larger,
nized through the award of the 2003 Nobel Prize for Physics to deeper caverns.
Dr. Davis and Prof. Koshiba for their pioneering contributions The scale and sensitivity of this new generation of physics
Given the success of this recent suite of deep sited experi- apparatus will place demanding performance requirements on
ments, a great deal of interest has been generated in developing the cavern excavation, reinforcement, and liner systems. In
new experiments in the area of “underground physics”. The particular, experiment proposals for new neutrino experiments
international community is now developing plans for major call for the excavation of caverns ranging in span from 30 to
new particle physics experimental programs. These programs 60 m. The paper will promote the need for rock mechanics
will probe the fundamental behaviours and properties of sub- and engineers to actively support the physics program, and
nucleic particles. The experiments will be long-term,with data support their colleagues in the development of low-risk cav-
collection and analysis scheduled to continue fordecadesafter ern designs that can deliver stable and watertight structures.
commissioning of the detectors themselves. The detectors,and These caverns will need to provide for safe, long-term occu-
auxiliary equipment associated with them, will need to be built pancy of the underground facilities and protect the surface and
at depth underground within large, stable rock caverns. groundwater environments.
Experimental options currently under consideration call Construction of laboratory facilities is expensive, with cost
for the excavation of rock caverns ranging in span from 20 for creation of a cubic metre of laboratory space often run-
to over 60 meters. Some of the more sensitive experimental ning between five hundred to over a thousand dollars per cubic
equipment would need to be housed at depths approaching yard. For the large-scale experiments to be viable, it will also
two kilometres. The larger experiments call for the excavation be critical for the design to incorporate the use of the most
of bank rock volumes on the order of a half a million cubic cost-effective construction techniques. The paper proposes
metres. These dimensions of span and depth are at or beyond that the large, multiphase construction scope of the planned
the conventional limits of rock engineering practice and the new laboratory facilities justifies an early emphasis on state-
design, construction and operation of such facilities will be of-the-industry technologies and the identification of possible
a major undertaking. In the US, these large caverns would research opportunities, implemented project long, from site
be constructed as part of a major new initiative to build a investigation to decommissioning. An early focus on the opti-
multidisciplinary, underground laboratory complex within the mization of engineering practices and integration of research
boundaries of the US Deep Underground Science and Engi- tasking could be of particular benefit to the later phases of
neering Laboratory (DUSEL). DUSEL is being designed to construction most notably with regard to the construction of
fit within the old mine workings of the Homestake Mine Gold the larger caverns, sited at depth underground.
649

Mechanism and key controlling technology of water inrush in tunnel construction
L.P. Li, S.C. Li, Z.H. Xu, B. Sui, S.H. Shi, Y. Zhou, C. Hu, Z.Q. Zhou & S.G. Song
Geotechnical and Structural Engineering Research Center, Shandong University, Jinan, Shandong, China
Keywords: karst formations; risks and hazards; tunneling
1 INTRODUCTION inrush and combinations of the above two forms (Li-ping

LI, 2010).
Karst areas are widely distributed in China. In the statistics Substantially, the principle of division is closely related to
of super-large tunnel accidents, large-scale water inrush ranks the protective structure and the water inrush channel.
among the highest in terms of the number of deaths and occur- 2) Criterions of water inrush in fractured rock mass
rences, and brings huge security risks and economic losses to Water inrush pattern of tensile-shear crack failure
underground engineering of tunnels.Due to the complexity of
karst development and regularities of water-bearing structure
occurrence, the key problem that the karst water inrush is diffi-
cult to control lies in the poor understanding of mechanism of Water inrush pattern of compression-shear crack failure
the water inrush and system methods of water inrush control.
Based on the characteristics of karst water inrush, this paper
first briefly puts forward the conditions and influence factors,
and the occurrence criterions of water inrush are established 3) Criterion of water inrush induced by hydrologic barrier
according to the classification of water inrush. The hazard instability
control technology system of water inrush in karst tunnel is
established based on risk assessment, advanced prediction of
hazard source and grouting technology. It was successfully
applied in the Qiyueshan Tunnel of Hurongxi highway.
2 INFLUENCE FACTORS OF WATER INRUSH

4) Criterion of water inrush induced by fillings-fault
As shown in Fig. 1, based on the statistics of significant acci-
dents of water inrush in recent 50 years, nearly half of the
water inrush in statistical cases is mainly related to geological
defects and geology prediction, and 16% for the underground 5) Criterion of water inrush induced by filling-type pipe
water.
3 TYPES AND CRITERIONS OF WATER INRUSH
1) Types of water inrush

In terms of the failure modes of water inrush, karst water
inrush can be divided into three generalized patterns, 4 KEY CONTROLLING TECHNOLGY FOR
which are geological defects inrush, non-geological defects WATER INRUSH
From the perspective of prevention and controlling, the pre-

vention technology system for water inrush hazard in karst
tunnel is established, consisting of three parts: risk assess-
ment system, comprehensive prediction methods and grouting
treatment system. The comprehensive geology prediction
system is established based on the risk level of water inrush.
And water inrush hazard management system is consist of
(1) Principles of hydrodynamic treatment, (2) Geology detec-
tion before grouting treatment, (3) Controllable drainage with
regional grouting, (4) Drilling bores upstream and drain-
Figure 1. Factors and their proportion for water inrush. ing in depth, (5) Selecting key bores and grouting slurry
650

6 CONCLUSIONS
1) The occurrence conditions of water inrush are put forward

from four aspects, and influence factors of water inrush
have been classified as karst hydrogeological factors and
engineering disturbance of human factors.
2) Wter inrush types are divided into geological defects
inrush, non-geological defects inrush and the combination.
Mechanical models and critreions for each type of water
Figure 2. Gushing holes. inrush is established.
3) A fuzzy AHP model for risk of water inrush which can
improve the accuracy of risk assessment is constituted by
the comprehensive weight combination method, and an
optimal system is probed that prediction methods vary with
the risk ranking of water inrush, and a new grouting mate-
rial with matching technology for hydrodynamic grouting
has been developed based on the newly treatment principle
of water inrush with high pressure.
4) The hazard prevention & treatment technological system
Figure 3. TSP’ result. of water inrush is composed by the three parts, and it was
successfully used in Qiyueshan Tunnel.
REFERENCE
Li-ping LI, 2009. Study on catastrophe evolution mechanism of karst

water inrush and it engineering application of high risk karst tunnel
[D]. China, Jinan, Shandong University Doctoral Dissertation.
Figure 4. GPR’ result.
Figure 5. New hydrodynamic grouting technology and its

application.
in key bores, (6) Real-time monitoring and information

construction.
5 ENGINEERING APPLICATION
By Analyzing the risk of water inrush of section

YK329+700∼YK330+106 for Qiyueshan Tunnel, TSP (see
Fig. 2) was adopted for the prediction of unfavorable geo-
logical bodies. There are some karst fissures in section
YK329+271∼YK329+247. According to TSP prediction
result and the risk assessment of YK329+271∼YK329+247,
GPR (see Fig. 3 and ABD methods are introduced. And
according to geological analysis and comprehensive geology
prediction results, karst conduit and fissures are grouted and
blocked (see Fig. 4 & Fig. 5).
651

Key technology to treat defective geological data-based comprehensive prediction

during tunnel construction in karst areas
S.C. Li, B. Liu, Q.S. Zhang, S.C. Li & L.C. Nie

Geotechnical and Structural Engineering Research Center, Shandong University, Jinan, Shandong, China
Keywords: karst formations, risks and hazards, tunnelling
With the high-speed development of infrastructure construc- construction. The challenging problems involved in are as
tion in China, more and more tunnels are and will be built follows: (i) identification and positioning of faults, frac-
in the fields of transportation, hydropower and mining. At tures, karst cavities and underground rivers; (ii) prediction
present, China has become the country in which the largest of water bearing stratum; and (iii) interpretation of prediction
number of tunnels has been built. With the development of result in the context of multiple explanations of geophysi-
infrastructure construction, more tunnels will be built in karst cal prospecting result (Yi-guo Xue, 2008, Qing-song Zhang,
areas in southwest of China, which have the characteristics of 2009).
long tunnel-line, great buried depth and complicated geologi- For the above problems, three measures as follows are taken
cal condition. In addition, the geological condition can not be in this paper. First, the method of risk evaluation for geologi-
detected clearly during engineering prospecting stage. There- cal hazards is presented, which lays a foundation for scheme
fore, the geological hazards such as groundwater inrush and establishing for geological prediction. Second, land-sonar
rock mass collapse are often encountered in construction of method is proposed as a new seismic prediction technique
tunnels in karst areas. in tunnel. the GPR signal response characteristics to water-
On one hand, a lot of tunnels in railway and highway under bearing structures are obtained by theoretical derivation. And
complicated geological and terrain condition is being in con- the judgment criterion to geological anomaly of TSP detection
struction. For example,Yiwan Railway is the most challenging is proposed, by which the level of single prediction technical
railway project in mountain area, which is known as the world- is improved. Third, a system for comprehensive prediction
famous puzzle in aspect of high risk of water inrush hazard of geological hazard sources is developed, which emphasizes
and high karstification level. Serious water inrush hazards on the role of geological analysis and the detection of water-
have taken place in Maluqing Tunnel and Yesangua Tunnel of bearing structures and provides an effective way for geological
Yiwan Railway with serious casualty and economic loss. In prediction during tunnel construction in karst areas.
addition, Qiyueshan tunnel and Wuchiba Tunnel in Hurongxi In this study, the concept that programming prediction is
Highway are constructed in the karst geo-environments. In the based on quantitative risk analysis is put forward. The risk
rock mass of the tunnel, karst cavities and underground rivers rank partitioning principle of karst tunnel should be oriented
are widely distributed. The excavation of tunnel induced the to save time, save costs, reasonable economic and controlling
groundwater inrush, along with mud and fractured rocks. All and humane construction. It don’t add unnecessary workload
of the above mentioned geological hazards can cause great and prevent the water and mud inrush (Yi-guo, 2009; Einstein,
loss of life and property to the stakeholders. 1996; Christian, 2002).
In the hydropower field, the auxiliary tunnel of Jinping According to petrofabric and lithology, geological struc-
hydropower station in Sichuan province is buried 2,375 m and ture, karst, groundwater and other factors, the risk of water
has a length of 17.5 km. The buried depth takes the construc- inrush are divided into 4 levels in karst tunnel construction,
tion of the tunnel into the condition of extremely high ground considering the rainfall and the positive and negative slope
stress level. In this circumstance, the construction of tunnel and other factors. Level I is the highest risk, while level IV is
can be influenced by the possible failure of rock mass, which the lowest risk. The purpose is to refine the range of the geo-
may be triggered by the relief of ground stress in unfavorable logical disasters and take effective prediction methods, which
geological conditions, such as the presence of faults, poor rock may occur in tunnel. The appropriate control systems and early
mass and groundwater (Shu-cai Li, 2008; Tian-bin Li, 2009). warning mechanisms are established, so construction safety is
One of the challenging issues in these hazardous conditions ensured.
is the prediction of the geological hazard sources in front Currently, the methods used in the prediction of geolog-
of tunnel face during tunnel construction. Extremely com- ical hazard sources can be classified into two categories:
plicated geological conditions and higher risk of geological geological investigation and geophysical detection. Geolog-
hazards put forward higher requirements to bad geological ical investigation includes engineering geology analysis, pilot
prediction and controlling technology in tunnel construction. excavation, and core drilling, while geophysical detection
The key technical problems of the theory, methods, tech- covers seismic, electromagnetic and geological radar methods.
niques and equipment of defective geological prediction and these prediction methods are compared in terms of prediction
controlling technology need to be solved urgently in tunnel objective, distance, advantage and disadvantage.
652

Accuracy of single method of advanced geological fore- electromagnetic. So the frequency of the radar wave decreases.
cast is not quiet reliable, because of geological multiplicity. On in-phase axis, when the radar wave encounters filled-
Results that different defective geological are detected by dif- fissure, water conductivity karst channels and filled water
ferent method are different. There isn’t one method in which cave, there are obvious anomalies or even discontinuous on
all kinds defective geological are detected accurately. The pre- the in-phase axis.
diction technical system is established in high-risk karst tun- The TEM prediction of tunnel face is different from that on
nels, based on many engineering practices and summarizing the ground. Water-filled faults and cave, TEM is sensitive to
geophysical methods. the position and size of fractured zone filled with water and
TSP is one of the elastic waves reflection methods, which mud, with distance of 50 meters .The ground prediction is half-
is sensitive to rock type, lithology changes, faults, caves and space, while tunnel face prediction is a whole-space problem.
water-rich zone location, with distance of 150 meters in front Using the equivalent conductance plane method, the theories
of tunnel face. Faults and fractured zones can be identified and methods for the 3D whole-space should be used in tunnel.
according to the following observations: the velocity of the In addition, the apparent longitudinal conductance has a sen-
vertical reflection wave increases slightly; depth migration sitive response to the low-resistivity conductive thin-layers,
begins with strong negative refection and ends with strong which helps to detect low-resistivity geological formations
positive reflection, while there are many unorganized posi- filled mud and water.
tive and negative reflective layers in the reflection bands; and A system for comprehensive prediction of geological haz-
the velocities of the vertical and horizontal reflection waves ard sources for tunnel construction is proposed. In the system,
decrease in general, but frequent changes in the level. the role of geological analysis is emphasized. The essential
Land-sonar is a good method by which middle or small objective of prediction is the geological hazard sources. The
karst caves are well detected. The third generation land- advantage of each method is utilized, but the disadvantage
sonar (LDS-3) with high performance of anti-interference, is avoided. In this manner, the geological hazard sources
dustproof and waterproof is development. Land-sonar can can be effectively predicted in front of tunnel face during
well met tunnel environment and hammering is used as seis- tunnel construction and the risks of geological hazards are
mic source. Detectors receiving reflected wave are placed minimized. Moreover, two case studies using this system are
in seismic source point, which detecting depth is between presented in this study. From the case studies, the system
100 meters and 200 meters. Signals between 10 and 4000 Hz is proved to be an effective means to minimize the risks of
can be received without distortion and freely extracted by the geological hazards during tunnel construction. Long-range
receiving system. rapid identification of the geological structure is realized
GPR is sensitive to faults, caves, the location and size of with technology of tunnel seismic advanced geological fore-
water-rich zone, with distance between 10 and 25 meters. cast. And it implements accurate detection of water-bearing
On energy aspect, the difference of the dielectric coefficients structure is realized with electromagnetic prediction tech-
between water and surrounding rock is significant. The energy nique. The effectiveness and efficiency of defective geological
of the reflection wave becomes larger when the radar wave prediction is improved greatly with the technology system,
arrives at interface of water and surrounding rock. So ampli- which provided an effective way to solve the problem of
tude increases. On frequency aspect, the conductivity of water prediction on defective geological and early warning on geo-
increases, resulting in increases of attenuation coefficient and logical disaster during construction period of high-risk karst
the attenuation speed of the high-frequency wave in the role of tunnel.
653

Geomechanical model test of underground caverns in salt rock
Y.R. Liu, B. Li, F.H. Guan & Q. Yang

State Key Laboratory of Hydroscience and Hydraulic Engineering, Tsinghua University, Beijing, China
ABSTRACT:This paper simulated the injection-production cycles of 4-cavern reservoir group under different internal pressure
with small block masonry. The variation of the displacement and strain of the bedrock and the cavern group is automatically
recorded by the displacement meters and the strain gauges buried in the model. Cavern stability under different gas production
process, the failure pressure of the caverns and the impact of mudstone interlayer on cavern stability are analyzed.
Keywords: oil reservoir; physical modeling; stability analysis
1 INTRODUCTION In the geomechanical model test, both the mechanical

parameters of a single small block and the integrated deforma-
In recent years, the world’s demand for natural gas has tion characteristics of the masonry need to be tested to ensure
increased significantly, which made energy security an impor- the similarity. Tests show that the deformation modulus of the
tant issue. Underground natural gas storage construction is an model is lower than that of the continuous rock by 30% to 50%.
effective method to ensure the storage security and regulate In order to avoid excessive deformation modulus reduction,
the seasonal supply-demand contradiction of natural gas. Salt the block size should be large enough. E. Fumagalli proposed
rocks in China have shallow buried depth, small strata thick- that the masonry block size should be checked with mechani-
ness and plenty of insoluble intercalations. To improve the cal tests of masonry including at least 100 blocks. The size of
security and space utilization of energy reserve, intensive cav- blocks used in this experiment is 2 cm–5 cm.
ern groups must be used. Thus, it is of great significance to Non-continuous structural surfaces, such as muddy inter-
research the temporal evolution of storage group damage and calation, etc., are simulated with dehydrated gypsum.
the corresponding protection theory.
At present, researches on the overall stability and temporal
evolution of underground caverns are relatively few. The sta-
bility of caverns is mainly analyzed with numerical methods. 3 EXPERIMENTAL SIMULATION
Experimental study mainly focused on the mechanical proper-
ties of rock salt and the surrounding rock stability of the single This test simulates 4 underground reservoirs; each one is sim-
cavern. Experimental study on overall stability and chain ulated as an ellipsoid with the same size: the length of the
destruction of cavern groups has not been reported yet. The 3D semi-major axis is 84 m, and the length of the semi-minor axis
geomechanical test in this paper simulates gas injection and is 36 m. The ellipsoid centers are in the same horizontal plane,
production of 4-cavern group under different pressure. The located at the vertices of a square with side length 144 m. At
variation of the caverns’ displacement and strain under differ- the bottom of the cavern is sediment area with a height of 56 m.
ent pressure is automatically recorded by displacement meters So the actual height of the cavern is 112 m. 24 m above the
and strain gauges buried inside the model. The allover stability center of the cavern is a muddy intercalation with a thickness
and the possibility of chain damage are analyzed based on the of 2 m. the buried depth of the caverns is 1000 m. The size of
experimental results. the test cell is 1.5 m × 1.5 m × 1.4 m. Schematic diagram of
the whole test system is shown in Figure 1.
The test load is divided into two parts: First, simulate the
weight load of 776 m of rock; second, simulate the pressure
variation during the gas injecting and production.
2 SIMILITUDE THEORY OF RUPTURE TESTING The internal pressure is simulated with silicone balloons.
AND SIMULATION OF SALT ROCK MASS The balloons are put into the cavern before capping. The bal-
loon is connected to the filling equipment outside with the
A geomechanical test is a non-linear destruction test. It must high-strength hose buried in the model. The balloon is con-
meet the requirement of similarity: geometry similar and nected with a balance cylinder. With the help of the high
stress-strain similar. precision barometer connected with the balance cylinder, the
Similar material is pressed into small blocks, which will be internal pressure of the balloon can be precisely controlled.
glue into a model. The small blocks are to simulate the defor- The relative displacement between the cavern walls, the
mation characteristics of the rock. And the bonding among absolute displacement and the strain on the cavern wall are
the small blocks is to simulate the strength properties of frac- measured. There are also some internal displacement meters
tures. Thus, small block masonry method can simulate both arranged in the mudstone interlayer to measure the interlayer
deformation and strength characteristics of the rocks. dislocation.
654

Figure 2. Curve of mudstone interlayer relative displacement and
the internal pressure.
Figure 1. Schematic diagram of experimental device.
The curve of mudstone interlayer relative displacement and
The self-made displacement meter is made from the man- the internal pressure (at the position of displacement meter 9)
ganese copper, with strain gauges pasted on both sides. The is shown in Figure 2. The dislocation of the mudstone inter-
absolute displacement is fixed outside the model with dis- layer with the maximum pressure 36 MPa is very small. Thus
placement rod. To avoid the impact of the model on the the impact of mudstone interlayer on cavern stability is limited.
displacement rod, the rod is wrapped with casing pipe, and
the contact of the rod and the casing should be avoided. The 5 CONCLUSION
displacement meter rod and casing are made of plastic pipe
and steel pipe. This experiment simulated the injection-production cycles of
5 kinds of cycles with different maximum pressure were 4-cavern reservoir group with different internal pressure. The
carried out, that is, 12 MPa, 18 MPa, 24 MPa, 28 MPa and variation of the displacement and strain of the bedrock and
36 MPa. For each cycle, the loading process was as follows: the cavern group is automatically recorded by the displace-
(1) weight simulation; (2) 4 caverns production; (3) 3 cav- ment meters and the strain gauges buried in the model. The
erns production; (4) 2 adjacent caverns production; (5) Single following conclusions can be drawn:
cavern production; (6) 2 diagonal caverns production.
1) The risk level order of different gas production processes in
the sequence of descending was: single cavern, 2 diagonal
4 RESULTS ANALYSIS caverns, 2 adjacent caverns, 3 caverns, 4 caverns.
2) The curves of the internal pressure and the displacement
The displacement curves of displacement meters under differ- during the production showed that the critical pressure of
ent internal pressures showed that gas production of single the model is 28 MPa.
cavern is the most dangerous gas production process. The 3) The dislocation of the mudstone interlayer with the max-
risk level order of 5 different gas production processes in imum pressure 36 MPa is very small. Thus the impact of
the sequence of descending was: single cavern, 2 diagonal mudstone interlayer on cavern stability is limited.
caverns, 2 adjacent caverns, 3 caverns, 4 caverns.
The internal pressure displacement curves show that dis-
placement 2, 3, 9 and 10 were not damaged, while displace-
ment 1, 4, 5, 7 and 8 were damaged at 28 MPa.
655

4D seismic monitoring of small CO2 injection – Results from the Ketzin pilot site
(Germany)
S. Lüth & A. Ivanova

GFZ German Research Centre for Geosciences, Centre for CO2 -Storage, Potsdam, Germany
C. Juhlin
Uppsala University, Department for Geosciences, Uppsala, Sweden
A. Kashubin
Uppsala University, now at Schlumberger Cambridge Research Centre, Cambridge, UK
N. Juhojuntti
Uppsala University, Department for Geosciences, Uppsala, Sweden
P. Bergmann & J. Götz

GFZ German Research Centre for Geosciences, Centre for CO2 -Storage, Potsdam, Germany
Keywords: case studies, field measurements, monitoring
1 THE KETZIN SITE The storage site is located near the town of Ketzin in the
state of Brandenburg, about 20 km west of Germany’s capital
The geological storage of CO2 is debated as one option to Berlin. At this site, natural gas has been stored for thirty years
reduce greenhouse gas emissions and to mitigate the severe gas at a depth of 250–400 m until the operation was abandoned
consequences of climate change. Geophysical monitoring of due to commercial reasons. The Rupelian mudstone served as
storage operations is a key component assuring the safety of cap rock and has demonstrated its suitability for gas storage
the reservoir and providing essential observations enabling operation successfully. The former gas storage formation is
to simulate and predict reservoir behaviour during ongoing separated by an additional 200 m thick clay-rich caprock from
injection and storage. Commercial CO2 storage has been the CO2 storage formation at 600 m depth. The storage forma-
operational at gas field sites like Sleipner and In-Salah for tion is located in the sandstone layers of the Triassic Stuttgart
several years, where time-lapse seismic surveys were able to Formation which consists of sandy channel facies and muddy
image the stored CO2 , amounting to several million tons in flood-plain facies of highly variable reservoir quality. The top
the resepective reservoirs (Michael et al., 2010). seal of the storage formation consists mainly of mudstone,
At the Ketzin pilot site, since June 30, 2008, approximately clayey siltstone and anhydrite, including a persistent, ca. 20 m
18 ktons have been injected per year into a saline aquifer at thin anhydrite layer showing in the seismic measurements as
650 m depth (Würdemann et al., 2010). The Ketzin site is the a clear and bright reference horizon (K2 in Fig. 1). The site
first pilot project for on-shore geological CO2 storage into included industrial land and infrastructure and a gas storage
a saline aquifer in Europe. The site has been developed in permission that enabled an easy development of a testing site
order to enhance the understanding of the underlying physi- for underground injection of CO2 . The operation of the CO2
cal, chemical and biological processes involved in geological underground storage is regulated under mining legislation of
CO2 storage in saline aquifers, to gain practical experience the state of Brandenburg. A maximum of 60,000 t of food
with monitoring and verification of CO2 storage, to test the grade CO2 is targeted to be injected, with the total amount to
predictability of different coupled models, and to verify tech- be adjusted according to scientific and site-specific require-
nologies required for safe long-term underground storage ments. A project internal and an external evaluation of the
of CO2 . The main focus of the project is the development, storage activities will be carried out after injection of every
testing and benchmarking of monitoring techniques, using a 20,000 t CO2 as part of a risk assessment review.
broad range of geophysical, geochemical and microbiological
methods.
Geophysical monitoring tools are deployed to observe CO2 2 RESULTS OF 4D SEISMIC MONITORING
migration in the subsurface. We apply surface and down-hole
measurements to test and optimize the resolution of different Repeated 3D seismic surface measurements are performed
methods and to follow the developing CO2 -plume. in order to observe time-lapse signatures indicating reservoir
656

survey. Repeatability analyses of the data revealed significant
changes in static corrections and the frequency content of the
data. The re-processed baseline and repeat volumes were then
analysed in terms of amplitude changes (Fig. 1) near the top of
the reservoir (sandstones of Triassic Stuttgart Formation). In
order to remove time-lapse efffects unrelated to CO2 injection,
the amplitudes of the 3D repeat volume were scaled using the
amplitudes of the K2 anhydrite horizon in the top seal.
Figure 1 shows the lateral extension the CO2 plume as of
October 2009 using the normalized time-lapse amplitudes at
the top of the storage formation. The CO2 which had been
injected into the sandstones of the Stuttgart formation replaces
the saline formation water. Due to the smaller bulk modulus
the injected super-critical CO2 , compared to the formation
water, the effective bulk modulus of the reservoir sandstone is
reduced. This bulk modulus reduction enhances the reflectiv-
ity at the top of the formation which had been characterized
Figure 1. In-line and cross-line sections of the 3D seismic mea- by a negative (but small) reflection coefficient before CO2
surements at Ketzin and the CO2 signature, observed at the top of injection.
the Stuttgart Formation. The comparison of seismic amplitudes, travel time changes
in deeper horizons and independent observations from bore-
changes due to the propagation of the CO2 in the reservoir. hole logging allows for a quantitative estimation of the CO2
A baseline survey was acquired in Autumn 2005, covering distribution in the reservoir and thus provides an invaluable
an area of approximately 14 km2 around the location of the basis for determining the detection threshold of geophysical
injection site (Juhlin et al., 2007). The acquisition was carried repeat surveys for monitoring CO2 storage.
out using a template scheme with the aim of having the same
acquisition geometry for each template. One template consists
of 240 geophone and 200 source point positions, respectively. REFERENCES
Due to logistical restrictions such as villages or environmental
protection directives, some templates were acquired with less Juhlin, C., Giese, R., Zinck-Joergensen, K., Cosma, C., Kazemeini,
acquisition points. The nominal subsurface coverage of 25 on H., Juhojuntti, N., Lüth, S., Norden, B., Förster, A. 2007. 3D
12 m × 12 m CDP bins was reached on a large portion of the baseline seismics at Ketzin, Germany: The CO2 SINK project.
axcquisition area. Geophysics 72(5), pp. B121–B132.
After approximately 15 months of CO2 injection into the Michael, K., Golab, A., Shulakova, V., Ennis-King, J., Allinson, G.,
reservoir, the first repeat of the 3D seismic measurements Sharma, S., Aiken, T. 2010. Geological storage of CO2 in saline
was acquired. The acquisition was focused to a smaller area aquifers – A review of the experience from existing storage sites.
International Journal of Greenhouse Gas Control 4, pp. 659–667.
around the injection site (ca. 7 km2 , 20 templates) as due to
Würdemann, H., Möller, F., Kühn, M., Heidug, W., Christensen, N. P.,
the small amount of CO2 injected after 15 months (∼22.000 Borm, G., Schilling, F. R., and the CO2 SINK Group 2010. CO2
tons) the CO2 was expected to have propagated only 200 m– SINK – From site characterization and risk assessment to monitor-
300 m away from the injection well. A comprehensive analysis ing and verification: One year of operational experience with the
and re-processing of all data, including the baseline mea- field laboratory for CO2 storage at Ketzin, Germany. International
surements, were performed after the acquisition of the repeat Journal of Greenhouse Gas Control 4, 938–951.
657

The effect of pick orientation on rock fragmentation
Seyed Saleh Mostafavi, Qingyu Yao & L.C. Zhang

School of Mechanical and Manufacturing Engineering, The University of New South Wales, NSW, Australia
Jason Lunn & Craig Melmeth

Bradken, Mayfield West, NSW, Australia
Keywords: mechanical excavation, mining
Efficient rock breakage in terms of energy consumption and The results were verified by relevant experiments on con-
dust generation has drawn significant attention. Extensive crete. Three different orientations were chosen. It was found
research has been done to find the optimized geometrical that the radius of the crushed zone along the direction of the
parameters of the cutting picks and to develop more efficient cracks and the maximum compressive stress in the crushed
mechanical fragmentation techniques. Edge chipping test has zone for Orientation (a) where an edge of the pyramid indenter
been regarded as a good fragmentation process because it is is perpendicular to the free surface of a specimen is the high-
highly efficient to generate fracture. Rock cutting can be con- est compared with those other indenter orientations. Therefore
sidered as a series of successive cycles of edge chipping, of this orientation has the worst energy efficiency. The best ori-
which each is composed of a crushing and chipping phase. entation is when a face of the indenter is in parallel to the
This research investigates the effect of pick orientation in the sample free surface.
holder on the efficiency of breaking rocks in terms of cutting
force and energy. The study was carried out by simulating a
series of edge chipping tests with a pyramidal indenter.
Figure 1. Development of a penny-shape crack and deviation of

the parallel cracks toward the free surface in an Edge chipping
simulation.
658

The soundness diagnosis system of plane concrete lining for railway tunnel in Japan
N. Okano, K. Tsuno & Y. Kojima

Structure Technology Division, Railway Technology Research Institute, Tokyo, Japan
Keywords: risks and hazards, tunneling
Railway tunnel inspections are classified into initial, general, the causes of deformation and deciding on countermeasures,
individual and occasional types. The standard maintenance which are general procedures in maintenance and control. To
and control procedure for railway tunnels is shown in Figure 1. do this, it uses deformation data on cracks, water leakages, etc.
Among these, general inspections are periodically executed and detailed data on tunnel interior displacement rates, land-
every two years to identify any deformation and the status of forms, geological conditions, design and construction details,
its progression by means of visual inspection and hammer test- etc. obtained during inspections.
ing. Therefore, such inspections are highly resource-intensive The overall system configuration is shown in Figure 2. After
in terms of time and effort. However, many railway companies inputting basic information such as the tunnel’s specifications,
are now facing a high rate of retirement among their skilled the system indicates the necessary positions for hammer test-
engineers and a reduction in the number of staff with suffi- ing based on an electronic deformation development diagram
cient knowledge of tunnel technologies. Accordingly, there is into which information on cracks and water leakages obtained
a strong need for a method to perform inspections effectively from visual inspection, etc. during a general inspection has
while also ensuring safety. been input. It then diagnoses soundness against flaking (rank
Accordingly, the authors have used past research achieve- α, β or γ) based on the results of hammer testing performed
ments to develop TUNOS (tunnel diagnosis system) – a diag- thereafter. Soundness diagnosis (rank A, B, C or S) for safety
nostic resource that allows automatic execution for most of considerations other than flaking is performed automatically
the work involved in diagnosing tunnel soundness, presuming where possible using the electronic deformation development
diagram, and the results of diagnosis are input manually for
other cases.
The system eliminates issues stemming from differences
between individual abilities, such as variations in the results
of soundness judgment (a common problem in tunnel inspec-
tions), and further improves inspection accuracy as well as the
efficiency of maintenance and control.
Figure 1. Standard maintenance and control procedure. Figure 2. Overall configuration of the system.
659

Case study of huge squeezing behavior in expressway tunnel in service
Y. Okui, T. Tsuruhara & H. Ohta

OYO corporation, Japan
M. Tada, S. Sakuma & C. Nakata

East Nippon Expressway Company Limited, Japan
M. Sueoka
Nexco-Engineering Tohoku Company Limited, Japan

Keywords: case studies, weak rock, back analysis
The Sakazuki-yama Tunnel on Yamagata Expressway which In order to understand above phenomena, we calculated
was built in 1991 is 1234 m long. The maximum overburden is deformation and lining stress using FLAC with several con-
about 100 m. The geology of this tunnel is rhyolitic tuff which ditions which represent the behavior.
received the hydrothermally alteration of the tertiary period. Consequently, we conclude the following result of this
At August 13, 2008, there occurred a sudden and huge heav- study.
ing of the road surface at the center of the tunnel, and the
1) The damage of the invert is caused by hygroscopic swelling
transportation of this tunnel was difficult. Uplifting height of
pressure of the surrounding rock.
the road surface reached to approximately 400 mm in ten days
2) Large displacement of the lining and invert heave was
from August 13.
occurred after the destruction of invert.
We investigated road surface leveling, convergence and
3) It is assumed that the cause of the above deformation of the
stress of lining, and evaluated the property of rock material
tunnel is stress-relief with squeezing after swelling.
around the tunnel.
4) It is possible to represent quantitatively using elastic-
From the results of investigation, the amount of reduced
perfect-plastic model that considered the swelling pressure
length of spring line of this tunnel is about 170 mm and lin-
and the reduction of shear strength around the tunnel.
ing compressive stress was −12 MPa and invert concrete was
damaged with open cracks.
Figure 3. Comparison between convergence and lining stress (near

Figure 1. Damaged invert of the Sakazuki-yama Tunnel. 36.601).
Figure 4. Comparison between the analysis result and measure-

Figure 2. Design of the section of the new invert. ments (Horizontal displacement (S.L. position).
660

San José tunnels: Reparation of a ground collapse
C.S. Oteo
University of Coruña, Spain
J.M. Maya
Canary Regional Government, Spain
J. Vicente
Acciona Infraestructuras, Madrid, Spain
Subject: Rock engineering in hazardous geo-environments. Others: Ground treatments.
Keywords: case studies, problem rocks, rock failure, tunnelling.
The San José tunnels (two parallel road tunnels which are both In the Project it was intended to use, in some cases,
19 m wide) were constructed as a road link of the third ring jet-grouting pre-crown (Fig. 2) which were swapped for
road of Las Palmas de Gran Canaria (Canary Island, Spain) micropiles (in view of the presence of pebbles in the Las
with the coastal Avenue. The tunnel (137 m2 section) is devel- Palmas Formation) and several levels of compensation
oped perpendicularly to the island’s coast. The excavation grouting under the houses.
involved (Fig. 1): Bench excavation was commenced, then leaving the arch
The land side mouth did not have any problem (conglom- supported on some small verges (Fig. 3). Slightly before
erate area of the Las Palmas Formation). Then the tunnel was the collapse of the right-hand tunnel, the tunnel construction
excavated using the Bernold method in the advance with a situation was the following:
central bench excavation and carrying out the side walls using
– Advance from the two mouths (after carrying out the sea
the alternate side wall system and concreting each excavated
side tunnel mouth (Fig. 1).
panel (Fig. 2).
– The central bench excavation was excavated along a very
– Antrophic filles and colluvials formations. extensive length.
– Las Palmas Formation: Quaternary conglomerates, – The arch support was being carried out with small benches
cemented, with some loose pebbles (which have average (Fig. 3).
diameters of 1.5 m). – The digging of the open bench excavation allowed the dry-
– Brown and whitish or volcanic tufa (with esmectites). ing of the site in the verges area where the earth was clayey
– Basalt rocks at some inverted arch points (ignimbites) (and expansive) tufs, opening up important cracks therein.
Figure 1. Stratigraphical profile of the San José tunnels.
661

Figure 4. Instability in the ground (crown foundation).
Figure 2. Tunnel section, including possible jet-grouting crown

reinforcement (Initial Design).
Figure 5. Collapsed zone (Longitudinal view).
Drillings of ∅ 133 mm were carried out, reinforced with

steel tube ST-52 with ∅ 88.9 mm and 6.3 mm thick. The max-
imum grouting pressure applied was 15 bars and, at each
Figure 3. San José Tunnel. Standard Advance Cross-section.
grouting stage, 150 l were provided per sleeve.
The surface movements recorded on the collapsed tunnel
With this situation there was a verge failing in one of the reached almost 30 mm. In other end areas they did not attain
tunnels by way of the cracks opened in the clayey tufa and the 15 mm. As a precaution, the houses were emptied which were
collapse of the arch occurred (Fig. 4). immediately above the “chimney”.
This collapse affected around 90 m in total in the right-hand Having carried out the filling of the chimney (part was car-
tunnel (Fig. 5). To limit the problem, concrete plugs were built ried out from above and using mortar once the heavy protection
quickly in order to limit the unstable area. basket has been done and the area to be excavated had been
The solution designed considered: subject to secondary grouting), the tunnel boring work was
supported using the Bernold method, taking bench excavation
– Treatment of the collapse area with reinforced grouting and and side wall near the advance.
filling. Investigations subsequent to the collapse allowed it to be
– Movements control (tunnel and of surface). established that, in addition to the failing in the support
– Further arch reinforcement in the other tunnel with 8– arch resting on a small verge on clayey tufa which dried
10 cm of sprayed concrete and wire mesh. out, the presence of a paleo-valley had an impact, having
– Strengthening of the arch support of the other tunnel with advanced transversally and in skewed fashion in the tunnels,
self-perforating grouted micropiling (every 2 m), forming contributing water and additional thrusts.
5◦ and 25◦ with the horizontal.
– Construction of invert in the altered tufa area.
The treatment to be carried out which was undertaken with

several ranges of pouted drills. Some of the cones which
remained on the area to be excavated were undertaken with
reinforced grouting with metallic tubes.
662

Observational construction management by field measurement of large scale

underground railway station by urban NATM – Railway Bosphorus
tube crossing, tunnels and stations
I. Otsuka, H. Taki, T. Aoki, M. Shimo, T. Kaneko & M. Iwano

Taisei corporation, Tokyo, Japan
S. Sakurai
Construction Engineering Research Institute Foundation, Kobe, Japan
Keywords: back analysis, monitoring, tunneling, weak rock
This project is named ‘Marmaray’which is a coined word com- To excavate tunnels while keeping safety for the buildings,
bining the sea of Marmara and railway. The Marmaray Project observational construction management is indispensable. It is
is the upgrading of approximately 76 kilometers of commuter important to grasp the occurred phenomena based on the mon-
rail to improve the railway system in Istanbul metropolitan itoring results. The phenomena are traced by means of back
area. Railway tracks will be connected between Asia and analysis and then the ground surface settlement and inclination
Europe through a tunnel under the Bosphorus Strait. Tun- are predicted by calculation for future tunnel excavation based
nels and stations have been and will be constructed by some on the result of back analysis. To reduce the settlement, some
methods such as immersed tunnel, shield tunnel, mountain countermeasures are considered. By analyzing the result, it is
tunnel, and cut and cover in 13.6 km length of this project. able to verify the effect of the countermeasure.
In this report, it is described that the observational construc- In this paper, the control of ground surface settlement
tion management by field measurement is performed in large was paid attention in observational construction management,
scale underground railway station (Sirkeci station) by means which is presumed of the two phenomena for the subsidence.
of NATM. One is the extension of shearing band to the ground surface.
Buildings above Sirkeci station are overcrowded as shown The other is the decrease in pore water pressure of the rock
in Figure 1. around the tunnel. Then, as the countermeasure for the sec-
Bird’s-eye view of Sirkeci station is shown in Figure 2. This ond phenomenon, recharge well was considered and it could
station has two platform tunnels that are abbreviated to PF reduce the decrease in pore water pressure. To confirm the
(about 270 m in length, about 90 m2 in cross-sectional area) effect of recharge well, site test was carried out. As a result,
and center walkway tunnel that is CE (about 152 m in length, recharge well was recognized to be adequate method for the
about 60 m2 in cross-sectional area) between two PF tunnels. control of ground surface settlement based on monitoring
The remaining width of rock between each PF and CE tunnel and analytical results. Therefore additional recharge wells
is just only 10 m. Moreover, platform tunnels connected with have been executed and can result in reducing ground surface
two shafts (about 510 m2 in cross-sectional area) are to be settlement.
equipped for the ventilating system and therefore they have The observational construction management will be contin-
large cross section, which called as PFV-L (about 190 m2 in ued in future.That means that the geological features condition
cross-sectional area). The excavation of this station is one of will be confirmed, the monitoring results will be analyzed,
the most difficult construction works in the world because and the adoption of an appropriate countermeasures will be
of its complicated structure in an urban with many old and executed according to the approrch that described on this
tumbledown buildings. paper.
Figure 1. Condition of ground surface above Sirkeci station. Figure 2. Bird’s-eye view of Sirkeci station.
663

Comparison of field and numerical modelling data for the steel pipe
umbrella arch applications
S.C. Ozer
Department of Mining Engineering, Istanbul Technical University, Istanbul, Turkey
V. Ozacar & C.O. Aksoy

Department of Mining Engineering, Dokuz Eylul University, Izmir, Turkey
Keywords: Tunnelling, Numerical modelling, Monitoring, Weak rock
1 INTRODUCTION modelling and deformations are given in Figure 2. Results

of numerical modelling studies and field measurements are
Umbrella arch application is generally applied by using given in Figure 3 (Aksoy & Onargan 2011b).
4 inches ϕ steel pipes. This application requires special expe-
rience and special machine for drilling holes to embed the
4 inches ϕ steel pipes. In such a case, a machine to make 2.2 Case 2
umbrella arches in tunnel face, a machine to embed bolts at Within 2nd phase of Izmir Metro construction, 3 inches ϕ
umbrella arch overthrust and also a machine to make exca- steel pipes umbrella arch application is done at km: 3 + 380,
vation are required. As an alternative to this application, both as one of the many cross-sections examined and analyzed on
excavation and umbrella arch application can be made with the tunnel route, a comparison for the cases with and without
just one machine when 2 inches ϕ steel pipe umbrella arch umbrella arch technique was made for settlements, conver-
application is sufficient. Hereby, both a positive effect made gences and in-situ measurements and the results from the
on project costs by increasing machine usage rate and less
amount steel pipes are used.
2 COMPARISON DATA OF FIELD AND NUMERICAL

MODELLING
2.1 Case 1
Within the scope of this studies, evaluation of numerical mod-
elling results and field measurements during 2. and 3. Phase
tunnels in İzmir Metro System. Although they have same tun-
nel cross-sections, rock mass properties vary. Plaxis 3DTunnel
V2 is used in numerical modelling. Figure 2. Mesh structure of numerical modelling and deformation
condition.
In the first section of the study, it is investigated whether
2 inches steel pipe umbrella arch application is sufficient in
Ýzmir Metro 2. Phase Construction on Üçyol-Üçkuyular route
at area between Poligon and Güzelyalı. Numerical Modelling
data is given in Tables 1 and 2. Mesh structure of numerical
Figure 3. Numerical modelling results and in-situ measurement

Figure 1. Umbrella arch application (Onargan et al. 2008). results.
664

various rock mass conditions. The results of numerical mod-
elling studies were compared with the field settlement and
convergence measurements. It is found that the results of the
numerical modelling and the measurements of settlement and
convergence are so close.
Generally, an artificial crust is constituted in front of the
tunnel face with 4 inches ϕ steel pipes application. So, a sig-
nificant amount of settlements are reduced on tunnel route
and building’s damage on the route is prevented. However,
it is determined by analyses made with numerical modelling
that it is possible to apply umbrella arch with smaller diameter
steel pipes and it is seen that numerical modelling results and
field measurements are close to each other. It is determined
Figure 4. Ground settlement values obtained from in-situ measure- that healthy results can be obtained with these analyses via
ments and numerical model. numerical modelling and 2 and 3 inches steel pipe umbrella
arch application according to rock mass condition contributes
to project economy.
REFERENCES
Aksoy, C.O. & Onargan, T. 2011a. The Role of Umbrella Arch and
Face Bolt as Deformation Preventing Support System in Pre-
venting Building Damages, Tunneling and Underground Space
Technology, doi:10.1016/j.tust.2010.03.004.
Aksoy, C.O. & Onargan, T. 2011b. Determining The Steel Pipe
Umbrella Arch Size by Numerical Modelling in Weak Rock
Conditions: İzmir Metro Case, Madencilik (in Turkish in Press)
Gibbs, P.W., Lowrie, J. Kieffer, D.S. & McQueen, L. 2002. Design of
a shallow soft ground shotcrete motorway tunnel. In Proceedings
of the 28th ITA-AITES World
Hoek, E. 2004. Numerical modelling for shallow tunnelling, Dis-
cussion Paper, p.10, www.rocscience.com/library/rocnews/Spring
2003/ShallowTunnels.pdf.
Kim, C.Y., Kim, K.Y., Hong, S.W., Bae, G.J. & Shin, H.S. 2004. Inter-
Figure 5. Vertical tunnel convergence values obtained from in-situ pretation of field measurements and numerical analyses on pipe
measurements and numerical models. umbrella method in weak ground tunnelling. In Proceedings of the
ISRM Regional Symp. EUROCK 2004 and 53rd Geomechanics
Colloquy (ed. W. Schubert), Salzburg, Austria, pp. 167–170.
models are shown in Figures 4 and 5 respectively. The proper- Miura, K. 2003. Design and construction of mountain tunnels in
ties of rock mass and structural elements are given in Tables Japan. Tunneling and Underground Space Technology, 18 (2–3),
3 and 4, respectively. pp. 115–126.Tunneling Congress, Sydney, Australia.
Onargan, T., Aksoy, C.O., Küçük, K. & Kun, M. 2008. Pre and Final
Support Project of 3rd Stage of Izmir Metro System, 2008.
3 CONCLUSIONS Volkmann, G.M. 2004. A contribution to the effect and behaviour
of pipe roof supports. In Proceedings of the ISRM Regional
In this research, we analysed the different pattern and dif- Symp. EUROCK 2004 and 53rd Geomechanics Colloquy (ed.
ferent size of pipes via numerical modelling technique for W. Schubert), Salzburg, Austria, pp. 161–166.
665

Assessment on stress induced instability in a tunnel project of the Himalaya
K.K. Panthi
Department of Geology and Mineral Resources Engineering, Norwegian University of Science
and Technology (NTNU), Trondheim, Norway
Subject: Analysis technique and design methods
Keywords: rock burst, rock cover, rock stress, back analysis
There are mainly three engineering geological uncertainties through four categories of rock sequences consisting of biotite
that directly influence on the stability of a tunnel or an under- schist, carbonaceous phyllite, Manikaran quartzite and schis-
ground cavern. These are rock mechanical properties, in-situ tose granite gneiss (Figure 1). The biotite schist is intercalated
stress conditions and groundwater inflow through fractures with small bands of quartzite schist. The carbonaceous phyl-
and weakness zones. The priorities given in dealing with these lite is deformed and folded. The schistose granite gneiss is
uncertainties may vary from project to project. In regards occasionally intercalated with small bands (1–10 meters in
with tunnelling in the Himalaya, due to high tectonic activ- thick) chlorite/talcose mica schist. However, the Manikaran
ity, the rock mass in the region are fractured, faulted, sheared quartzite is relatively fresh and massive in most of the places
and deeply weathered. Tunneling through numerous zones of excluding fractures and weakness zones. The headrace tunnel
weakness, fractures and faults is thus a matter of reality. Major- crosses several major and minor weakness zones.
ity of long tunnels meet several difficult stability problems According to the Norwegian rule of thumb (Selmer-Olsen,
and excavation challenges. Hence, selection of right types of 1965) rock spalling or rock burst is likely to occur once
excavation methodology based on sufficient engineering geo- rock cover above the tunnel exceeds approximate threshold
logical investigations during planning is crucial for successful of 500 m. The extent of this type of failure is even severe
completion of tunnel projects and also to reduce uncertainty if tunnel runs parallel to the valley side with a slope angle
associated with delay in completion and cost overruns. exceeding 25 degrees. In case of Parbati II, the rock cover
One of the major areas where instability problems are faced along 40 percent headrace tunnel length exceeds this thresh-
while tunneling through Himalayan rock mass is the instability old. Approximately 9.5 km headrace tunnel between chainage
related to rock stresses. One of such tunnel cases is the Parbati 4.5 km to 16 km exceeds 500 m rock cover. The main rock
II hydroelectric project under construction in the Kullu district formations along this tunnel reach are biotite schist, phyllite
of Himachal, India. This is a run-of-river scheme under con- and Manikaran quartzite.
struction since 2002 and has an installed capacity of 800 MW Even though headrace tunnel runs through steeply dipping
and utilizes a gross head of 862 m. The river water is being valley side slope and rock cover is exceeding the threshold
planned to be diverted from near village Pulga by construct- of 500 m in many locations up to chainage 7.5 km (Figure 1)
ing a 85 m high and 113 m long concrete gravity dam. The no noticeable rock spalling or rock burst occurred. The exca-
water will then be conveyed through a 31.37 km long head- vation went smoothly with no major stress related instability
race tunnel and through twin inclined penstock pressure shafts observed excluding minor squeezing. The headrace tunnel up
(1.54 km length each) to the semi-underground powerhouse. to this chainage mainly passes through biotite schist and a
Geologically, the project area is located in so called ‘Kulu small band of carbonaceous phyllite between chainage 7.4 km
Window’ that represents the Lesser Himalayan rock forma- to 7.5 km.
tion and the area is bounded by a major fault system in the However, after chainage 7.5 km brittle and massive
Himalaya called “Main Central Thrust (MCT)”. The Kulu Manikaran quartzite appeared. In this quartzite severe rock
Window mainly comprised by rocks from crystalline to meta- burst occurred continuously in valley side roof of the headrace
sedimentary sequence. The headrace tunnel passes mainly tunnel after almost every alternate blasting round.
Figure 1. Geological cross-section of the headrace tunnel giving two study sections (modified from Panthi, 2009).
666

Table 1. Mean values of some lab tested mechanical properties of The failure in Manikaran quartzite at chainage 15.7 was
biotite schist and Manikaran quartzite. elastic-brittle and major principle stress (σ1 ) is oriented ver-
tically. Hence, major principle stress (σ1 ) is represented by
Manikaran quartzite gravity alone and is a product of specific weight (γ) and
Mechanical Biotite
Properties Unit Schist Adit 1 DS Adit 2 US
rock cover (h). At chainage 15.7 km the rock cover is 775 m
and Manikaran quartzite has specific weight (γ) amounting
Specific weight MN/m3 0.028 0.026 0.026 0.026 MN/m3 (Table 1). This gives vertical gravity stress (σ1 )
Intact rock strength MPa 75 150 175 equal to 20.15 MPa. So, one set of principle stress is known.
Elasticity modulus GPa 12 60 65 Concerning horizontal stress, in this case minor principle
Poisson’s ratio – 0.13 0.15 0.20 stress (σ3 ), the only known fact is that it is oriented perpen-
Strength anisotropy – 2 1.05 1.1 dicular to the major principle stress (σ1 ) and is horizontal.
Mode of failure – Plastic Brittle Brittle However, the magnitude of this horizontal stress is not fully
known, but it may be expressed by the following equation;
Similarly,TBM excavation fromAdut 2 upstream (19.46 km

chainage and upstream) went smoothly up to headrace tunnel
chainage 16.02 km (Figure 1). No severe stress induced insta- where; ν = Poisson’s ratio, σv = σ1 = gravity stress (in this
bility was registered excluding at few locations with minor case also major principle stress), and σtec = tectonic compo-
scale popping of rocks and minor squeezing between chainage nent of horizontal stress.
16.02 km to 15.99 km. However, at chainage 15.99 km very The Himalayan region is tectonically active. The magnitude
abrasive, brittle and massive quartzite (similar rock forma- of this tectonic horizontal stress vary considerably depend-
tion as from 7.5 km downstream) appeared and continued ing upon geographical location, geological environment and
until chainage 15.56 km. At this tunnel segment rock splitting distance from main tectonic faults, in particular, the distance
(spalling) along spring line of headrace tunnel occurred. from the Main Boundary Thrust (MBT) and the Main Central
The headrace tunnel upstream Adit 2 is excavated using Thrust (MCT). According to World Stress Map (2008) the ori-
TBM, which gives tunnel a circular shape. Minor rock split- entation of horizontal tectonic stress at this far western part
ting occurred at around chainage 17.5 km along the spring of the Himalaya is oriented about N080–090E to S260–270W.
line of the headrace tunnel indicated failure along the spring This means, horizontal tectonic stress makes an approximate
line of the tunnel, which suggests that the major (σ1 ) and orientation of 70 to 80 degrees with headrace tunnel alignment
the minor (σ3 ) principle stresses are vertical and horizon- upstream Adit 2.
tal, respectively. In addition, the lab tested rock mechanical With this understanding and the use of data sets presented in
properties of Manikaran quartzite are available. This is an Table 1 and calculated using relationships proposed by Hoek
ideal condition for the back calculation of in-situ principal et al (2002) and Panthi (2006a) assessment on minor principle
stress magnitudes using numerical finite element code such stress (σ3 ) was done using Phase 2 finite element code. With
as Phase 2. This is what is being intended to be carried out in the use of mean values of rock mass properties the yield-
this paper. ing occurred for all three horizontal to vertical stress ratios
Table 1 below gives laboratory tested mechanical proper- representing 0.6, 0.55 and 0.5, respectively. Simulation result
ties for biotite schist and Manikaran quartzite at headrace also indicated that the strength factor close to or even below
tunnel Adit 1 downstream and Adit 2 upstream. The intact one. The yielding occurred for horizontal to vertical stress
rock strength (σci ) and elasticity modulus (Eci ) given in Table ratio representing 0.5 found to be symmetric and yielding
1 are based on the results of 32 mm diameter cores drilled concentration is also observed to be along the spring line of
perpendicular to the foliation and converted them to ISRM the headrace tunnel. The strength factor in this case was just
suggested 50 mm equivalent core diameter. below one (0.96). Hence, this stress ratio is considered rep-
Before carrying out numerical assessment for back calcu- resentative for the splitting (failure) to occur at this headrace
lation of in-situ stresses, representative input variables have tunnel chainage.
to be estimated. For this purpose rock mechanical proper- From above finding it is conclude that the horizontal to
ties given in Table 1; relationships proposed by Hoek et al vertical stress ratio at chainage 15.7 km (rock cover 775 m)
(2002) to calculate reduced material constant (mb ), constant is 0.5 or very close to this value. With stress ratio 0.5 the
(s), cohesion ©, peak frictional angle (φ) and tensile strength horizontal minor principle stress (σ3 ) is close to 10 MPa
(σt ); relationship proposed by Panthi (2006a) (Equation 1) to (20.15/2 MPa). The tectonic stress component (σtec ) perpen-
calculate deformation modulus (Ecm ); and actual rock mass dicular to the headrace tunnel alignment at chainage 15.7 km
conditions observed in the tunnel were used. thus may now be calculated as 4.4 MPa (10.07 – 5.68) using
Equation 2. To detail readings the reader is referred to the main
article.
where; Ecm = deformation modulus for rock mass, Eci =

young’s modulus for intact rock (lab tested), and σci = uniaxial
compressive strength for intact rock (lab tested).
667

Integrated modelling and design of reinforced concrete linings for tunnels
K.P. Paschalidis & S.C. Bandis

Department of Civil Engineering, Aristotle University, Thessaloniki, Greece
Keywords: numerical modelling, tunnelling, rock support, rock mass
The structural design of permanent reinforced concrete lin-

ings of tunnels involves significant uncertainties, principally
related to the simplistic representation of the concrete struc-
ture interaction with the surrounding geological medium and
the inherent limitations pertaining to direct application of
standard limit state design concepts. As a result, no standard
engineering codes are available for the design of permanent
reinforced concrete linings for tunnels.
The majority of geotechnical engineering analysis software
possesses limited capabilities for simulating the behaviour of
reinforced concrete structures. The effects of cracking are
commonly simulated by assigning globally reduced elastic
moduli or stiffness values in concrete members. Such an
approach fails to represent realistically the re-distribution of Figure 1. Comparison between the Ground Reaction Curves
internal forces due to localised cracking and may even lead to derived from explicit discontinuum (UDEC-BB) and implicit con-
favourable effects. tinuum (FLAC/M-C material and -Ubiquitous Joint) analyses of a
Structural analyses software provides adequate tools for horizontally bedded weak rock mass.
simulating the post-elastic behaviour of reinforced concrete,
but lack of capabilities for the realistic simulation of the inter-
action of the lining structure with the surrounding geological
medium.
In this paper, a rigorous model for representing the contin-
uously changing stiffness behaviour of reinforced concrete at
different stages of cracking will be described. Creep effects
are also modelled.
The interactive mass – lining response was considered in
connection with discontinuous rock and the cracked con-
crete behaviour model was implemented as machine language
(FISH) sub-routine to the Universal Distinct Element Code
(UDEC), as the state-of-art implementation of the Distinct
Element Method (DEM).
Achieving realistic and meaningful rock – lining interac-
tions on the basis of discontinuum rock mass responses is
fundamental for deriving loading mechanisms and design-
ing temporary or permanent support systems. An example Figure 2. Example application of the variable stiffness model of
modelling case will be presented referring to a weak, bed- cracked concrete illustrating the predicted reduction in the 2nd
ded rock mass to demonstrate the erroneous predictions that moment of inertia of a reinforced concrete section due to flexure
might arise, if unjustified homogenisation of the rock mass without and with creep.
was adopted. (Figure 1).
It is well established, that reinforced concrete sections
exhibit non-linear behaviour under flexural loads. As tensile Sensitivity analyses were executed to demonstrate the
stresses become greater than the tensile stress capacity fctm importance of the correct modelling of cracking of reinforced
of concrete, cracks begin to form and propagate along the concrete linings to both the bending capacity design envelope
section height, with the reinforcement bars undertaking the and the calculation of deflections.
tensile forces required for the internal cantilever to balance The maximum observed differences in Figure 3 were
the external loads. The variation of flexural stiffness of a rect- obtained by comparing discontinuum (DISCR_) vs. pseudo-
angular reinforced concrete section due to cracking is shown continuum (CONT_) rock mass models with the elastic (EL)
in Figure 2. vs. the cracked (CRK) concrete behaviour models.
668

The profound differences in the predicted bending moments
of the lining were attributed to the divert responses obtained
from the explicit discontinuum and the implicit continuum
models. A typical example of the contrasting behaviour
observed in the two modelling cases is presented in Figure 4.
The engineering implications of the cracked concrete
behaviour model were manifested by the significant changes
observed in the predicted structural behaviour of the lining.
The resultant effects on the lining response included enhance-
ment of bending moments at both the crown and sidewall
sections due to stress redistribution across the lining length,
Figure 3. Bending moments – comparison of maximum observed
differences. even though the ground reaction remained relatively stable
(as evidenced from the negligible differences in the crown
deflections and axial forces).
When creep was introduced, the loads on the lining were
decreased and no stage II cracking was exhibited. However,
the proposed method correctly modelled the decrease in axial
stiffness and led to almost doubling of the bending moments.
Load factoring is a conceptual approach allowing for plau-
sible variations in the loads acting on a structure. In the case of
tunnels, both the vertical and horizontal rock loads contribute
to the axial loads (N) and bending moments (M). The incorpo-
ration of concrete cracking effects in the structural behaviour
of a R/C lining indicated potentially significant increase in the
design envelope of the lining forces.
The increase in the design envelope surpasses commonly
used safety factors. Furthermore, only the bending moments
(redistribution due to varying flexural stiffness along the
perimeter), and not the axial forces (related to the ground
reaction curve) show significant variability. Thus typical load
factoring, already limited in nonlinear analyses, comes into
question, requiring further investigation and a different design
methodology.
Figure 4. Observed differences in the re-distribution of vertical
stresses as predicted from the analyses of the implicit continuum
and the explicit discontinuum models of the bedded rock mass.
669

System behaviour in weak ground: comparison of yielding elements
N. Radončić & W. Schubert

Institute for Rock Mechanics and Tunneling, Graz University of Technology, Graz, Austria
Keywords: numerical modelling, rock properties, rock support, weak rock
There is almost general consent in the engineering community

that our ability to resist the loads developing while tunnelling
in weak ground under high overburden is limited, at least
within the boundaries imposed by financial and risk control
considerations. The concept of allowing for deformation to
occur while retaining control over the support pressure mobi-
lization is not new (Figure 1), and has proven itself time and
time again in many tunnels.
The most frequent principle relies on a segmented shotcrete
lining and yielding elements installation in the deforma-
tion gaps. Over the past decades, several types of yielding
elements have been developed and successfully applied in
aforementioned conditions. The design of ductile linings has
to incorporate numerous influences: the interactions between Figure 1. The basic concept of a deformable lining as proposed by
Rabcewicz, utilizing wooden “squeezing elements”.
the time- and face position dependent ground deformation
development, load-displacement characteristic of the yield-
ing elements, shotcrete rheology and overall cross-sectional
layout have to be integrated into a sound overall picture.
On the other hand, the associated design requirements are
straightforward:
1. Compliance to the clearance profile,
2. Compliance to the maximum thrust, dictated by the time-
dependent shotcrete strength development,
3. Maximizing the utilisation of the available shotcrete
capacity and
4. Support resistance higher than the weight of the potentially
loosened portion of the ground.
Unfortunately, in a stark contrast to the relatively high
frequency of yielding element application, attempts of an
objective analysis and comparison of their influence on the
system behaviour are rather seldom. Hence, the goals of the
efforts presented in the publication are straightforward: after Figure 2. Results obtained from the numerical model, compared to
validating the numerical model, different yielding element the test data and the original relationship after Schubert.
types are implemented, and their effect on the displacement
development and shotcrete stresses is investigated.The numer-
ical model used for the comparison is calibrated on the ground and lining utilisation, a considerable effort has been
measurement data from the exploratory tunnel Paierdorf. It invested both in sophisticated modelling of the mechanical
crosses the Lavanttal fault zone over a length of 450 meters, behaviour of the ground and in modelling the shotcrete rhe-
mainly composed of intensely fractured gneiss shales and ological behaviour. The strain-softening constitutive law of
strongly heterogeneous cataclastic rock mass. The support FLAC3D has been applied to depict the ground. The param-
measures applied in this area are composed of 35 cm of eters have been derived from extensive laboratory testing,
shotcrete with two Strenger-Type yielding elements (thrust in-situ tests and numerical back analysis.A FISH routine steers
at yield: 1000 kN), dense rock bolting and a temporary top the creep parameters of the Norton’s Power law, shotcrete elas-
heading invert at a relatively low distance to the face. tic properties and shrinkage and re-calculates the respective
In order to be able to perform a rigorous analysis of the inter- parameters for every excavation round, depending on the strain
action between yielding elements’ characteristics, deforming history and shotcrete age.
670

The applied approach has been validated on uniaxial
creep tests of shotcrete and shows very satisfactory results
(Figure 2).
After the entire numerical model has been validated by
comparison to the displacement monitoring data, the yielding
elements type has been changed according to the currently
commercially available types. The findings imply that the
hiDCon and LSC elements’ high thrust capacity is very
favourable with regard to the final displacements and support
Figure 3. Influence of the yielding element capacity on the dis- resistance mobilisation (Figure 3).
placement development. However, hiDCon elements’ high initial stiffness leads to
significant stress peaks in the early loading stages (while
the shotcrete is still “young”), limiting this element type to
conditions of very slow face advance or swelling ground
(Figure 4).
The Lining Stress Controller is the only element type fulfill-
ing both the requirement of an appropriate yielding element
stiffness and high final capacity.
Figure 4. Comparison of the shotcrete utilisation development

observed at the crown of the numerical model.
671

Gypsum crystal growth in tunnels
A. Ramon & E.E. Alonso

Department of Geotechnical Engineering and Geosciences, UPC, Barcelona
Keywords: case studies, field measurements, numerical modelling, problem rocks, rock stress, tunnelling
Severe swelling phenomena have been observed in tunnels

excavated through sulphated formations from the Triassic to
the Paleogene period. Expansions generally occur in the mate-
rial located below tunnel invert. The geometry, characteristics
of the tunnel sections and the geotechnical properties of the
rock masses involved have been described in detail in the liter-
ature (Krause (1976), Nagel (1986), Wittke (2006) and Alonso
et al. (2007)). However, the factors that lead to swelling phe-
nomenon aren’t well known. The growth of gypsum crystals
under the floor of the tunnel is the main reason of the observed
expansions in tunnels. The excavation of a tunnel through a
sulphated formation will not necessarily result into the devel-
opment of swelling due to gypsum crystal growth. A study of
several tunnels excavated through gypsiferous formations has
been carried out in order to investigate the factors that explain
the initiation and development of swelling in tunnels. The tun-
nels analyzed belong to the Spanish high speed railway line
Madrid-Barcelona. The analysis reveals that swelling never
develops when there is not groundwater crossed by the tun-
nel. Tunnels which had experienced swelling phenomena have
in common that the sulphate content found in the rock is high
(40% in weight of solid constituents), the rock is tectonized
and there exists groundwater with a high sulphate content.
Lilla tunnel is a railway tunnel excavated through Eocene
anhydritic gypsiferous claystones. Severe swelling phenom-
ena occurred at tunnel floor level from the beginning of its
excavation. Gypsum crystal growth in open discontinuities Figure 1. Photographs taken of samples recovered from boreholes
was at the origin of the swelling phenomena. The instrumen- performed under Lilla tunnel floor at depths within the active layer.
tation installed recorded an accumulation in time of vertical
heave in an active layer about 4–5 meters thick located under Gypsum crystal growth is a consequence of supersatura-
the floor of the tunnel. Neoformation of gypsum crystals was tion of water in sulphates. Supersaturation can be achieved by
found in samples recovered from depths corresponding to the two mechanisms. The first one is the evaporation of sulphated
active layer where swelling occurs (Figure 1). water. The second mechanism leading to supersaturation is
The magnitude of the heave occurred in Lilla tunnel was not induced by the previous dissolution of anhydrite in water.
homogeneous along the total length of the tunnel (Figure 2). In The modelling of heave in Lilla tunnel should consider that
a continuous stretch of 600 meters immediate to the southern swelling at floor level is a result of gypsum crystal growth
portal no important expansions were detected. This singularity in open discontinuities. Swelling deformations due to gyp-
motivated the analysis of the characteristics of the rock mass sum crystal growth will be given by the precipitated mass
at different locations along the tunnel and the comparison of of gypsum crystals. These considerations and the evaporation
properties in locations where high swelling phenomenon was mechanism have been taken into account in the thermo-hydro-
observed with the properties in locations where no expansions mechanical model proposed by Alonso & Olivella (2008).
occurred. The analysis of the extracted samples from bore- However, the calculated accumulated vertical displacements
holes performed at different locations in Lilla tunnel, showed at floor level are about ten times lower than the values
higher proportion of striations and slickensided surfaces in measured in the field (Figure 3). This is indicating that an
the borehole position of high swelling. Moreover, the sulphate additional mechanism is provably taking part in the crystal-
proportion observed in the rock under the invert level of the lization process. Gypsum crystal growth may occur also in the
tunnel was higher in the location where swelling developed. saturated zone in the rock mass, without invoking evapora-
Finally anhydrite dominates in this location. tion as a supersaturation mechanism. The second mechanism
672

Figure 2. Measured heave at floor level along Lilla tunnel in time.
Positions of boreholes B1 and B2.
Figure 4. Reinforced section of Lilla tunnel.
the stresses in the steel reinforcement are far from the elastic
limit of the steel.
REFERENCES
Alonso, E. E., Berdugo, I. R., Tarragó, D. & Ramon, A. (2007).

Tunneling in sulphate claystones, Proc. Of the XIV European Con-
ference on Soil Mechanics and Geothecnical Engineering, Madrid.
Cuéllar et al (editors), Millpress, Rottherdam, Vol 1: 103–122.
Alonso, E. E. & Olivella, S. 2008. Modelling tunnel performance in
Figure 3. Development of heave displacements in time (Alonso & expansive gypsum claystone. Proceedings of the 12th Conference
Olivella, 2008). of International Association for Computer Methods and Advances
in Geomechanics, Goa, India, October 1–6.
Krause, H. (1976). Sulphate rocks in Baden-Württemberg and their
leading to supersaturation conditions mentioned before may importance in relation to civil engineering. Bulletin – Association
play an important role in gypsum crystal growth and in the of Engineering Geologists (13): 45–49.
development of swelling in Lilla tunnel. Marí, A. & Pérez, G. A. 2003. Refuerzo del revestimiento de los túne-
A resisting support consisting in a massive circular cross les de Lilla y Camp Magré para la Línea de Alta Velocidad Madrid-
section was built for the reinforcement of Lilla tunnel. Figure Zaragoza-Barcelona-Frontera Francesa. Tramo Lleida-Martorell.
4 describes the circular cross section defined for the reinforce- Provincia de Tarragona.
Nagel, D. (1986). Sohlhebungen in den Keupertunneln von Baden-
ment of Lilla tunnel. Concrete with a characteristic strength Württemberg, Tunnelbau, Kontakt & Studium (184): 110–125.
of 80 MPa and steel B500S reinforcement and confinement Wittke, M. (2006). Design, construction, supervision and long-term
bars were used (Marí & Pérez, 2003). Recent results of the behaviour of tunnels in swelling rocks. Proc. of the Interna-
instrumentation installed in Lilla tunnel after its reconstruc- tional Symposium of the International Society for Rock Mechanics,
tion have recorded swelling phenomena. Despite the high Eurock 2006, Liège. Van Cotthen, Charlier, Thimus & Tshibangu
values of radial pressure measured at the invert (6.1 MPa), (editors), Taylor & Francis Group, London: 211–216.
673

Geomechanical aspects of CO2 sequestration and modeling
Jonny Rutqvist
Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, US
Subject: CO2 sequestration
Keywords: Numerical modeling, Site characterization, Fluid flow
Deep underground injection of carbon dioxide (CO2 ) has At Krechba, satellite-based interferometry (InSAR) is used
emerged in recent years as an important option for seques- to monitor ground-surface deformations with remarkable
tering CO2 and thereby reducing the emission of greenhouse precision (a few millimeters), because of favorable ground-
gases to the atmosphere. Because CO2 is less dense than surface conditions, including hard desert sediments and bare
water, the targeted CO2 injection reservoir should be overlain rock. Inverse semi-analytical strain analysis as well as coupled
by low-permeability caprock formations that can sufficiently fluid flow and geomechanical numerical modeling are being
limit upward buoyancy-driven flow of the injected CO2 . Rock employed to interpret observed ground-surface deformations
mechanics plays an important role in the performance assess- in terms of underground fluid movements and structures, such
ment of such a CO2 injection site, including assessment of as faults and fracture zones. Significant modeling effort is cur-
caprock sealing performance, ground-surface deformations, rently being dedicated to analyze ground-surface uplift at one
and the potential for fault reactivation and induced seismicity. of the injection wells, where a double-lobe uplift pattern indi-
As a practical example of geomechanical aspect related to cates opening of a deep fault or fracture zone—a feature that
CO2 injection, this paper reviews recent results on model- has also been detected in a recent 3D seismic survey. Finally,
ing of geomechanical aspects associated with the In Salah coupled fluid flow and geomechanical modeling has been
CO2 storage project, at the Krechba gas field, Algeria. The In applied to estimate the potential for injection-induced micro-
Salah CO2 storage project can be considered an ideal case for seismicity. The analysis showed that the highest potential for
studies of rock mechanical aspects, because injection takes injection-induced microseismicity occurs along the horizontal
place at an injection pressure substantially higher than ambi- sections of the injection wells as a result of the stress changes
ent hydrostatic fluid pressure, and ground-surface conditions caused by the combined effects of injection induced cool-
are very favorable for monitoring of ground-surface defor- ing and pressure. However, for the best-estimated present-day
mations. Moreover, the targeted injection zone is a fractured strike-slip stress regime at Krechba, the analysis indicated a
sandstone layer intersected by minor faults and fracture zones. relatively low potential for injection-induced microseismicity.
This paper summarizes and reviews recently published work, The geomechanical models are being updated and recali-
including modeling of ground-surface deformations, potential brated as new field data becomes available, and the long-term
fault reactivation, and induced seismicity. geomechanical responses will be assessed.
674

The integrated prediction system for geological conditions ahead of tunnel faces
S. Shirasagi & T. Yamamoto

Kajima Corporation, Japan
Y. Mito
Department of Urban Management, Graduate School of Engineering, Kyoto University, Japan
M. Ashraf
School of Civil Engineering, University Sains Malaysia, Malaysia
Keywords: tunnelling, site characterization, rock properties
It is essential to employ the observational design and con- short-interval prediction subsystems for the several engineer-
struction method to manage tunnel projects efficiently. The ing purposes in the framework of observational design and
original tunnel design is immediately verified or modified construction in the tunnel project. Geostatistical method is
based on the in-situ data sequentially obtained in the con- employed to obtain not only the spatial distribution of geo-
struction stage. The authors develop the integrated prediction logical conditions but also the precision of prediction. The
system for geological conditions ahead of a tunnel face. The applicability of the proposed subsystems is verified through
system is composed of the long-interval, middle interval, and the field application.
Figure 7. The predicted rock strength index, RSI (point) at 0.5 m

of the tunnel face, and the corresponding observed value (line) by
short-interval prediction.
Figure 4. Frequency distribution of the calculated expected con-

struction period (200 realizations) by long-interval prediction.
Figure 9. The predicted RSI at 0.5 m ahead of the tunnel face,

and the corresponding observed value. The mean value and the
range of distribution (mean ± 2*standard deviation) are plotted as
the prediction.
Figure 5. The predicted geological score (point) at 5 m ahead of
the tunnel face, and the corresponding observed value (line) by
middle-interval prediction.
675

Stability analysis of concrete plugs in a pilot cavern for

compressed air energy storage
W.K. Song & D.W. Ryu

Department of Underground Space, Korea Institute of Geoscience and Mineral Resources, Daejeon, South Korea
Y.K. Lee
Department of Coastal Construction Engineering, Kunsan National University, South Korea
Keywords: compressed air storage, concrete plug, finite element analysis, contact pressure, factor of safety
Compressed Air Energy Storage (CAES) is a way to store

energy generated at one time for use at another time. Presently,
a pilot test of CAES in an underground lined rock cavern
is being conducted in Jongson, Korea. Lined rock caverns
have both merits and drawbacks compared to rock salt domes,
excavated caverns and groundwater reservoirs. Although its
economics are weaker than the other types, the lined rock
cavern type has the strong advantage of “siting flexibility”.
In underground CAES caverns, a concrete plug need to be
installed to seal the compressed air stored in the container;
therefore, the selection of the shape and dimension of the
concrete plug could present a critical design factor. In this
study, a stability evaluation of two types of plugs was carried
out in a numerical way by investigating the distribution of the
factor of safety in the plugs and the distribution of contact
pressures over the contact surface between the rock mass and
the concrete lining.
Two different types of pilot caverns, PC1 and PC2, are to be
constructed at a depth of 100 m below the surface (Figure 1).
The shape of the caverns is cylindrical, with an inner diameter Figure 1. Perspective view of pilot caverns for CAES.
of 5 m. The length of the caverns is 10 m for PC1 and 20 m
for PC2. The differences in the caverns come from the sealing
and lining materials, as well as plug configuration. Rubber of 14 MPa, corresponding to the weight of the overburden lay-
sheets and steel plates are used as sealing material in PC1 and ers, is applied to the upper face of the model. The initial stress
PC2, respectively. In PC1, iron rods are also embedded in the assigned to the left and rear surfaces of the model is 15.4 MPa.
concrete lining for reinforcement purposes, while they are not During the simulation, air pressure pi , with a maximum of
used in PC2 in order to compare structural behaviors with PC1. 5 MPa is applied to the inner walls of the storage caverns.
For comparing structural efficiency, a wedge-shaped plug is Sliding and separation are allowed on the interface between
installed in PC1 and a taper-shaped plug in PC2. During oper- concrete and rock. Sliding may occur on the interface where
ation, pressurized air at a maximum of 5MPa is inserted into coincident nodes are separated as shear force exceeds fric-
the cavern through an access hole 0.7 m in diameter installed tional resistance force. The coincident nodes may also be
in the plug. separated when tension is put on them and contact force
3D FEM code SolidWorks software was employed in this becomes zero. In this analysis, it is assumed that the cohe-
study for analyzing the mechanical stability of the CAES pilot sion on the interface is null and the residual friction angle
tunnels. SolidWorks can make reduce the time required to con- is 40 degrees. Unit weight and Poisson’s ratio of rock mass
stitute complex numerical elements of the model, as it is based were obtained from laboratory tests. Young’s modulus of rock
on CAD. The numerical models are composed of three parts: mass was determined from GSI estimation in site. Reason-
container access tunnel, plug and storage cavern. The bot- able properties are assigned to concrete. Parameters related to
tom, right and front sides of the model are restricted to only strength are not needed, as the materials are assumed elastic,
tangential movement by rollers, and the rest sides are free while uniaxial compressive and tensile strength are utilized to
to move in both normal and tangential directions. The ratio determine factor of safety based on the Mohr-Coulomb fail-
of vertical and horizontal stresses, K, is set to 1.1, which was ure criterion. Three cases of concrete model with different
obtained from the hydro-fracturing test in place. Normal stress compressive strengths are simulated in this study.
676

plug and the lining. When the compressive strength is 40 MPa,
failure regions are observed only in part of the roof and the
floor of the concrete lining. This result indicates that the taper-
type plug model is more stable from a structural point of view
than the wedge-type plug model
It is important to know the stress distribution in the inter-
face between the rock mass and the concrete, because that
area is probably the weakest section in the whole structure.
In this study, interface elements that could possibly separate
and slip are assigned to that area to analyze contact pressures.
For a conservative analysis, it is assumed that the models have
40 degrees of friction angle and null cohesion. In the case of
PC1, contact pressures to a great extent act on the roof and
the bottom, and not on the side. This results from a separa-
tion between the rock mass and the concrete in the side wall.
Figure 2. FOS distribution in the wedge- and taper-shaped plugs For the taper-shaped plug, however, relatively uniform contact
and lining; blue region (FOS > 1), red region (FOS ≤ 1). pressures are distributed over the whole surface. This indicates
that a separation of the interface does not occur and the shear
resistance force is well mobilized on the entire structure.
The safety factor for a failure of rock mass and concrete Variation in contact pressures on the middle of the roof and
materials was analyzed as the Mohr-Coulomb failure criterion the edge of the bottom along the longitudinal axis of the plugs
was applied to the stress distribution obtained from elastic are plotted with respect to the distance and depicted in Figure
analysis. According to the Mohr-Coulomb criterion, the factor 9. The distance starts from the left end of the plugs. In the
of safety, FOS, is defined by: PC1 model, contact pressures on the middle of the roof and
the edge of the bottom are 20 MPa and 40 MPa, respectively
and they increase abruptly at the left end of the plugs. It can be
seen that local separations of the interface are also produced
at some points in the middle of the roof. It should be noted
that contact pressures are zero along the entire length of the
where σ 1 and σ 3 are the maximum and the minimum principal side edge.
stress, respectively and σ t and σ c are the compressive and In the PC2 model, contact pressures with an order of 20 MPa
tensile strength, respectively. are uniformly distributed over the entire area except for the
When the value of FOS determined by the above equation left end area, as in the PC1 model. It is notable that local
with a stress state at a point in the cavern that is obtained separations of the interface occur at both ends of the side edge.
from numerical analysis does not exceed 1.0, a failure can be This implies that a separation of the materials could occur in
expected to occur. Numerical simulations were conducted for the wedge-shaped plug model. If this were to happen during
three models with different strengths of concrete: 24, 30 and operation, not only could the mechanical stability of the plug
40 MPa. 5 MPa of air pressure was directed on the inner wall not be secured, but air leakage through this opening would
of the concrete lining. Figure 2 shows FOS distribution in the also be expected. Therefore, reinforcement using rock bolts
plugs and linings of the two caverns. In the figure, the blue and resin would be required to improve the connection ability
regions indicate that FOS is superior or equal to 1.0, and FOS is between the plug and the rock mass.
inferior to 1.0 in the red regions. In the PC1 model, the regions In this study, we concluded that the taper-shaped plug is
where a failure may occur are widely spread both in the plug structurally more stable than the wedge-shaped plug. The con-
and in the lining, although the compressive strength increases tact point between the plug and the lining is the most unstable,
up to 40 MPa (Figure 2-a). Therefore, it can be seen that the as the factor of safety is lower in that area than in the other
wedge-shaped plug model has low stability. It is revealed that areas. High compressive strength concrete should be used to
especially the connection area between the plug and the lining, overcome this weakness. For the wedge-shaped plug, rein-
as well as the roof and the bottom area of the lining, is unstable. forcement efforts should be made within the contact surface
In the PC2 model, when the compressive strength of con- between the rock and the plug, as a separation could occur
crete is superior or equal to 30 MPa, no failure region occurs due to a non-uniform distribution of contact pressures on the
in the plug, except in part of the connection area between the surface.
677

Salt and rock mechanic challenges at the verification of drift seal systems at the
Morsleben repository, Germany
J. Stahlmann
TU Braunschweig, Germany
R. Mauke
Bundesamt für Strahlenschutz, Germany
M. Mohlfeld
Zerna Ingenieure, Germany
C. Missal
TU Braunschweig, Germany
Subject: Underground storage of petroleum, gas and nuclear waste disposal
Keywords: nuclear repository, case studies, field measurements, fluid flow, monitoring, problem rocks
1 ABSTRACT
The Morsleben repository for low and intermediate radioac-

tive waste is located in a former salt mine. After an operational
phase of about 20 years and several years for developing
the closure concept it is now under licence for closure. The
main functional requirements for the closure of the repository
are the stabilization of the mine, the limitation of leaching
processes and the sealing of the disposal areas.
In line with this procedure drift seal constructions are
planned and tested which will constrain the possible infiltra-
tion and in far future the migration of contaminated brine. In
lack of generally accepted codes of practice, there many com-
plex engineering performances are necessary dealing with the
construction materials, the behaviour of the host rock as well as
the excavation damage zone (EDZ) and finally the interaction
between them. Due to the functions of the dams, the investi-
Figure 1. Overview of the construction site of the prototype drift
gations cover the geo-mechanical behaviour of the host rocks seal test in the Morsleben repository (grouting tubes: green col-
and the sealing materials as well as their hydro-mechanical ored, construction of a pressure chamber for flow-through testing:
behaviour. In the recommended paper or presentation respec- background).
tively the basic conditions of the repository are described.
The geology of the salt structure and the design of the mine in the interface between dam and host rock is relevant to
lead to sealing locations in different formations of salt rock and guarantee the stability and the serviceability of the dams.
anhydrite. Based on the differences of the geo-mechanical and This verification is in particular much more difficult for the
hydro-mechanical behaviour of the host rocks the designs of anhydrite locations than for locations in salt rock. Taking all
the drift seals have to fulfil various requirements. The dams in these aspects into account many numerical and especially
salt rock profit by the creeping behaviour of this material. Con- experimental investigations have already been and are still
sequently, a small shrinkage of the construction material could carried out.
be accepted if the short-term functioning is guaranteed by As a very important issue the verification on a large scale
injections. Also the EDZ reduces its permeability by the pro- in situ experiment was identified. Consequently, the BfS
cess of self-healing. In comparison to this the sealing materials announced a prototype test construction comparable to the
of dams in the anhydrite formations have to keep the volume or drift seals which have to be built in the future to realize
have to swell because of the absence of creeping. Furthermore, the closure concept of the Morsleben repository. A number
the permeability of the EDZ when indicated has to be sealed of small-scale and medium-scale experiments prior to proto-
by a grouting material with an adequate long-term behaviour. type testing have started and finished partially. At present the
If an immediate inflow of brine has to be taken into account, prototype test-dam is under construction (Fig. 1).
the resulting hydrostatic pressure at the dams has to be trans- The recommended presentation will depict the determined
ferred in the host rock. For this case the resulting radial stress processes and represent the existing results.
678

TBM vs. D&B – a difficult choice in mountain terrain – some geotechnical guidelines
M. Verman, T.G. Carter & L. Babendererde

Golder Associates, Delhi, India; Golder Associates, Toronto, Canada & Babendererde Ingenieure GmbH, Bad Schwartau, Germany
Keywords: blasting, mechanical excavation (TBM/RBM), risks and hazards, rock bursts, rock mass classification, site
characterisation, tunnelling
Tunnelling at depth through high mountainous terrain poses complex geology and stress conditions when mining at these
difficult challenges not just for tunnel boring machine (TBM) significant depths. Extremes of ground conditions present
drivages but also for application of traditional drill and blast major contrasts to tunnelling and they often demand use of
(D&B) methods. Dealing with adverse geology at any depth flexible rock engineering solutions in order for the tunnel to
can be problematic and can lead to significant tunnelling progress. This need to adopt flexible solutions is often seen as
delays if not adequately foreseen; but geological problem con- being at variance with the constraints imposed by the rigidity
ditions, which might be tractable at shallow depth, with either of design elements incorporated into the fabrication of a typ-
TBM’s or D&B approaches, when encountered at significant ical TBM. Traditionally, with some exceptions, TBM’s have
depth (>1000 m) can prove disastrous depending on stress only been utilized on tunnel projects where the risks associated
state, rock competence and prevailing groundwater inflows. with deep tunnelling are low, or expected to be low. This risk-
Mitigating delay problems associated with exceptionally bad aversion to use of machines in these conditions has generally
ground at depth requires considerable foresight and advanced been due to the perceived adverse consequences of entrap-
planning. The challenge is not just one of tackling the adverse ping or damaging a TBM under such conditions; in some part
ground condition, by modifying the excavation and support due to the relative difficulty of dealing with adverse ground
processes in order to deal with the specific problem zone, in the confined working area of a TBM, in comparison to a
stress state and groundwater conditions, it is also often about D&B/NATM heading.
logistics, as all too often for deep tunnels in mountainous Hard rock machine designs are, however, moving forward
regions the problem geologic zones are at significant dis- to more hybrid, universal concepts that encompass ability to
tance from the nearest portal, and at such significant depth execute full umbrella forepoling and soft rock machine face
that surface pre-treatment is generally impractical. conditioning, pre-grouting and ground treatment philosophies
Traversing faulted and disturbed ground at significant depth in an attempt to combat some of these problems by making
requires that tunnelling procedures be able to cope with a huge the machines sufficiently robust and at the same time flexible
range of difficult geological conditions. Investigating, evaluat- enough to be capable of safely and successfully excavating
ing and assessing anticipated geology ahead of tunnelling, and through extremely bad ground.
dealing with encountered difficult ground conditions requires Extremes of bad ground experience are cited from tun-
that better understanding be gained of the interaction between nelling in the Himalayas, Andes and Alps where a variety
of difficult ground conditions have had to be traversed, rang-
ing from rock-burst-prone conditions with hydrothermal water
inrushes through to coping with squeezing, very soft, soil-
like fault zone infills. Based on case records of tunnels,
geotechnical guidelines are presented for aiding the choice
of tunnelling method and for guiding the selection of appro-
priate equipment for dealing with the wide range of varying
ground support requirements expected at significant depth in
mountain terrain.
Geotechnical guidelines are presented for characterizing
and evaluating of the three main elements that limit our
ability to execute trouble free tunnels at significant depth –
stress state, groundwater conditions and rock itself (Figure
1). As explained with case examples in Carter et al, 2008,
based on Hoek and Marinos, 2000 for squeezing ground and
Diederichs, 2007 for spalling ground we now have methods
available to characterize each. Adverse characteristics of any
one of these elements can, on its own, compromise D&B or
TBM tunnelling but it usually takes a combination of all three
Figure 2. Primary Risk Elements – Water Jw, Stress SRF and Rock being adverse to trap a machine or halt a drill and blast. Rec-
Quality Q, GSI or RMR and Geotechnical Definition Measures. ommendations are thus included in the paper aimed towards
679

improving our understanding of tunnelling problems at depths be appropriate to conduct detailed analytical or numerical
through application of more in-depth structural geology tech- evaluations. Some cases are discussed where sophisticated
niques as measures for fault and structure interpretation.These analytical and numerical methods need application not only as
are recommended at several scales of geological appreciation, a back-analysis tool but also for predictive use to gain some
allowing improving decision making and risk evaluation. Ter- appreciation of magnitudes of squeezing or closure, but notes
rain analysis and geological structural fabric appreciation are that limited usage only is generally all that is justified at an
seen as key aspects that are often not done well enough such early project stage.
that problems of significant importance in route planning and Recommendations are given regarding limitations on D&B
alignment definition are missed. Structural geological eval- versus TBM usage, and on factors to consider relating to
uation at various scales provides: (i) clues to the likelihood making informed decisions about machine type, the first
of encountering certain types of fault structures, (ii) indica- requirement being to produce a complete geological section
tions of current and previous likely stress regimes, useful for along the alignment, and a geomechanics characterization of
defining orientations of principal stresses for design, and (iii) that geology. Following the methodology outlined by Carter,
inferences on most probable locations and orientations for 2005, the suggestion is made that using spreadsheets and
deep tectonized zones and for zones where deep geothermal good geological mapping, simply dividing the tunnel align-
gradients could lead to encountering hot water. All too often, ment up into subsections and defining rock quality and
these facets of gaining a complete geological understanding assessing propensity to stress induced problems (spalling,
are missing, with the inevitable consequence of running into bursting or squeezing) provides a quick and ready method
problems. In cases where cover depths are so high that spalling for appropriately zoning the proposed tunnel works.
or bursting is inevitable, or squeezing is to be expected over The paper concludes that for all deep tunnels there is a
significant lengths then realigning the tunnel to shallower key need to fully understand the potential for squeezing and
depths might be the best solution. spalling/bursting. Based on review of previous problems the
The paper concentrates on examining facors important to paper stresses that more emphasis is needed to be given
improving decision making. as lack of foresight is frequently to focussed geological investigation, stating that in previ-
observed to be at the nub of many of the problems encountered ous projects because of incomplete understanding early on of
in deep tunnels, and often is a critical factor clouding under- possible ground conditions and variability and likely ground
standing, that in turn may introduce errors and uncertainties behaviour, sometimes erroneous decisions have been made,
into the decision making process related to D&B versus TBM which incurred large costs and lead to time consuming delays
selection, and even more so related to selection of machine and in some case resulted in slow TBM progress and, in
type, if a machine option is favoured. A further complication extreme cases, loss of a TBM. The comment is alos made
to the decision making process relates to the timing when mak- that the reverse situation has also occurred where a TBM has
ing this key decision, as it needs to be made some 12-18months been ruled out for all the wrong reasons, and acceptably good
in advance of actually wishing to start tunnelling, so that suffi- rock conditions were, in fact, found. In such cases progress
cient lead time is available for building the machine. However, now becomes limited by D&B capability which, even in good
all too often detailed project site investigations are incomplete, ground, with the fastest performance crews, will typically only
still ongoing or in some cases not even started when this key achieve a third of the advance achievable by a machine drive.
decision is required to be made. This adds an extra level of
totally unnecessary risk to an already difficult decision. Pre-
planning to get comprehensive enough investigations done in REFERENCES
time is key to risk minimization.
The paper stresses attention needing to be paid not only Carter, T.G., Diederichs, M.S. and Carvalho, J.L. 2008. Application
to geotechnical risks, but it also critically looks at water and of modified Hoek-Brown transition relationships for assessing
access as these also constitute major risks, commenting that strength and post yield behaviour at both ends of the rock
geothermal gradient effects can also create problems of high competence scale. J.S.Afr. Inst. of Min. Met. Vol. 108: pp 325–338.
temperature rock and hot water. While excavation of deep Carter, T.G., Steels, D., Dhillon H.S., & Brophy D. 2005. Difficulties
rock tunnels can pose several unique challenges which can of Tunneling under High Cover in Mountainous Regions, Int’l.
be daunting, all of these must be at least identified, if not AFTES Cong., Tunnelling for a Sustainable Europe, Chambery:
pp. 349–358.
fully quantified as early and as well as possible when con- Diederichs, M.S. 2007. Mechanistic Validation and Practical Appli-
sidering making decisions about potential TBM applications. cation of Damage and Spalling Prediction Criteria for Deep
Suggested approaches using geotechnical and geomechanics Tunnelling. 2003 Can. Geotech. Colloq. Can. Geotech. Journal.
sectorization and simple spreadsheet evaluation of the vari- Hoek, E. and Marinos, P. 2000. Predicting Tunnel Squeezing. Tunnels
ability of geology along a given tunnel alignment are outlined; and Tunneling International, Part 1—November 2000, Part 2—
along with suggestions as to when and/or when not it would December.
680

Underground research laboratory network
Joseph S.Y. Wang

Lawrence Berkeley National Laboratory, Berkeley, CA, US
Peter H. Smeallie
American Rock Mechanics Association, Alexandria, VA, US
Xia-Ting Feng
Institute of Rock and Soil Mechanics, The Chinese Academy of Sciences, Wuhan, China
John A. Hudson
International Society for Rock Mechanics and Imperial College, London, UK
Keywords: case studies, field measurements, nuclear repository, rock caverns, site characterization, tunneling
Scientists and engineers in underground research laboratories in heterogeneity and fracture network characterizations, in
(URLs) have shared many common objectives and exchanged coupled process testing in controlled blocks, in quantification
information on experimental designs, sensor improvements, of induced deformation and seismicity potential, to coupling
field procedures, and research findings. The enhancement of underground signals to atmospheric and ionospheric pro-
and development of URL networks can further encourage cesses. While there are clearly site specific characterizations
multidisciplinary and cross-disciplinary research aimed at and experiment specific requirements unique for each URL,
technological invention and innovation to meet energy and there are common and new approaches that network exchanges
environmental challenges. Discussions in symposia of the can contribute. In addition to fundamental challenges, we
International Society for Rock Mechanics, American Rock also address practical aspects of organization, procedure, and
Mechanics Associations, and in sessions of the American incentive needed to develop and formulate URL networks.
Geophysical Union meetings, led to development of the URL In this paper, we first summarize the establishment and evo-
Workshop associated with the 2011 Beijing ISRM Congress. lution of URLs for radioactive waste research. Concurrently,
In this presentation, we summarize broad overviews and spe- there are underground laboratories with historical focus on
cific findings from lectures on URLs and panel discussions physics experiments with detectors in mines and along tun-
on what we have learned from existing URLs, what the main nels. In this paper, we use URL for both radioactive waste
problems are facing us today that can use existing URLs and and physics experiments. We then discuss examples of ongo-
require new URLs, and how valuable are virtual URLs linked ing efforts in URLs that have the potential for innovations
to real URLs. Lectures are planned for the URL workshop from interdisciplinary studies enhanced by URLs. The URL
on URLs in hard rock with varying degree of fracturing, in environments are quieter than laboratories on and above the
soft rock with plastic deformation, in shallow and in deep ground. We then discuss questions of URL networking. This
laboratories for physics and multidisciplinary research, in use paper presents an update of a review and evaluation on URLs,
of URLs for underground injection assessment, in search for and discusses trends and consensus from some recent papers
origins of life from microbes found in isolated water pockets, and presentations.
681

The state-of-the-art and research advances on the underground natural gas storage in
bedded salt rocks in China
C.H. Yang
Key Laboratory for the Exploitation of Southwestern Resources and the Environmental Disaster Control Engineering,
Chongqing University, Chongqing, China
Y.P. Li
D.A. Qu
West-east Gas Pipeline Company Gas Storage Project Department, Petro China Company Limited, Beijing, China
Key Laboratory for the Exploitation of Southwestern Resources and the Environmental
Disaster Control Engineering, Chongqing University, Chongqing, China
Keywords: rock caverns; stability analysis; rock properties; mining; numerical modeling
1 THE CURRENT STATUS AND DEVELOPMENT

TREND OF NATURAL GAS PIPELINES AND
STORAGE IN CHINA
The total length of natural gas pipelines in China is over 40

thousand kilometers and the total gas transportation capac-
ity is over 100 billion cubic meters per year by the end of
2009. At present, the gas network system is gradually formed,
which is characterized by the main trunk lines as the skele-
ton, connecting the trunk lines with branch lines and covering
the nationwide consumption markets. The commissioning of Figure 1. 3D perspective drawing of 6 reconstructed brine caverns
a group of natural gas pipelines such as the West-East Gas and 2 observation caverns.
Pipeline, Central Asia Gas Pipeline, and so on, signifies that
the development speed of the pipeline industry in China has
stridden into the world advanced level.
2 THE FEASIBILITY STUDY ON GAS STORAGE IN
To ensure safety of the long transportation gas pipelines,
DEPLETED SALT CAVERNS
a group of natural gas storages have been built or building.
Usually, natural gas is stored in a number of different ways:
According to the requirement of the first West-East Gas
(1) depleted reservoirs in oil and/or gas fields, (2) aquifers,
Pipeline Project, the six depleted salt caverns, Xi-1, Xi-2,
and (3) salt cavern formations. In this paper, taking the Jintan
Dong-1, Dong-2, Gang-1 and Gang-2 in Jintan Salt Mine
Salt Cavern Project as an example, the underground gas stor-
(shown in Figure 1) were selected to be natural gas storages
age in bedded salt rocks formation is introduced briefly.
after appropriate reconstruction. Based on the experimental
This project, the supporting project of the first West-East
analysis of the mechanical and creep characteristics of salt
gas pipeline, belongs to the West-East gas pipeline company
rocks and mud rocks, numerical study on deformation and
of PETROCHINA. Now the total storage gas is about 0.05
stability of these caverns, and brine-pressing cavern tests on
billion cubic meters, stored in 5 salt caverns which are con-
two salt caverns, it was concluded that the six salt caverns are
verted from existing depleted brine production caverns. In
suitable to be utilized as the gas storages, and that
June, 2010, the JZ-1 salt cavern began to be injected with gas,
which is the first one leached for the purpose of gas storage • The creep characteristics of the salt rocks is close to that
rather than for brine production. There will be 57 salt cav- of the salt rocks mixed with minor mud, while the steady
erns to be built by PETROCHINA in this area in the next state creep velocity of the mud rocks mixed with minor
10 years. salt is lower, only one eighth of that of salt rocks. This
682

The rock samples were divided into three types. The first
sample type is rock salt samples, which is grey-brown and
coarse-grained, containing a little of glauberite inclusion. Its
main chemical compositions are NaCl, Na2 SO4 and CaSO4 .
The second sample type is from anhydrite interlayer, con-
taining 17.50%∼23.62% NaCl and 36.24%∼42.62% water
insoluble matter. The third type is rock salt samples contain-
ing anhydrite interlayer of various thicknesses. The photos of
three kinds of typical core samples are shown in Figure 2.
The compression tests showed that the presence of anhydrite
interlayer affects strongly the deformation and failure charac-
teristics of bedded salt rocks. A ‘stress drop’ phenomenon of
stress-strain curves was observed for the low confining pres-
Figure 2. Three kinds of typical core sample. sure cases, denoting the anhydrite layer fractures prior to salt
rock.
The creep tests on salt rock, anhydrite and composite rock
indicated that the salt rock layers control mainly the creep samples implied that the steady-state creep ratio of interlayer
deformation of salt caverns. was one order lower than that of the salt rock. The creep
• The breakthrough pressures of the six salt caverns are deformation of surrounding rocks of storage cavern will be
among 6.7 to 21.3 MPa. This denoted that the sealing abil- controlled by the salt rock layer, while the anhydrite layers
ity of the caverns in bedded salt rocks formation is good will restrict the long-term volume shrinkage.
enough to be utilized as gas storages. For investigating the mechanical properties of the inter-
• The numerical analysis showed that the reduction of the cav- faces between anhydrite and salt rock, the direct shear tests,
ern volume can be restricted by the increase of the internal Brazilian disk tests and SEM tests were carried out. These
pressure. Therefore, for lengthening the serviceable lives of experiments showed that the interfaces between anhydrite
the salt caverns, it is suggested that the continuous running and salt rock are bonded perfectly and are not a weak one
time of caverns under lower pressure should be as short as as expected. This unique property of the bedded salt rocks
possible. would benefit greatly the sealability and stability of the energy
• During the running process, the pressure difference between storage caverns in bedded salt rocks.
two adjacent caverns will affect considerably the stability The Cosserat-like medium constitutive theory and its appli-
of the caverns and the range of plastic zone. The damage cations to the stability analysis of the salt caverns in bedded
zone of the mine pillar will be enlarged and the serviceable salt rock formation are introduced. The program of the new
lives of salt caverns will be reduced when the adjacent cav- constitutive model has been validated through a test sample,
erns run with a higher-pressure difference. So, some means so it is reliable to apply it in practical engineering. The asymp-
should be adopted to reduce the pressure difference, for totic damage-failure caused by the sequent damage of the
example, simultaneous injecting gas into or extracting gas individual layer with different elastic modulus can be con-
from the closely spaced caverns. sidered while the element number needed for computation is
• The safe distance between the tube-in-tube shoe and the very small. This provides a base for numerical simulation of
cavern top are 8 m for Xi-1, 12 m for Xi-2 and 6 m for the the stability of the gas/oil storage caverns built in the bedded
other four respectively, according to design age limit, 20 salt rock mass.
years.
• The brine-pressing cavern tests on Xi-1 and Xi-2 shafts
validated the sealing ability of the two shafts, and the steady
creep parameters by invention analysis agree well with the
ones from the indoor tests on salt rock cores.
3 RESEARCH ADVANCES ON THE MECHANICAL

BEHAVIOR OF DEEP BEDDED SALT ROCKS
The engineering mechanical properties of bedded salt rocks

were investigated for the purpose of the energy storage in deep
bedded salt rocks formation.
683

Probabilistic assessment of stability of underground rock caverns and cavern shape

optimization
W.G. Zhang & A.T.C. Goh

School of Civil and Environment Engineering, Nanyang Technological University, Singapore
J.Y.K. Wong
Defence Science & Technology Agency, Singapore
Subject: Analysis techniques and Design methods
Keywords: neural network; numerical modeling; risks and hazards
An underground cavern complex is usually built in a com- of Universal Distinct Element Code (UDEC) in this paper.
plex geological environment and may be subjected to different The following stochastic variables are considered: the fric-
loading conditions. In view of this, the major factors influ- tion angle, the cohesion, the deformation modulus of the rock
encing the cavern performance such as the stiffness of the mass and the in-situ stress ratio. The overburden thickness,
surrounding rock, properties of the rock mass, overburden the unit weight of rock materials, the Poisson’s ratio and the
thickness, the magnitude and direction of the in-situ geo- joint strength are assumed as deterministic. The cavern width
stress, the size and shape of the underground cavern complex, and height are also assumed as uncertain variables in order
the type and magnitude of the loading as well as the sequence to optimize the shape of the rock cavern. Most cavern shapes
of excavation and support must be taken into consideration. are horse-shoe or bullet-head shaped. It is widely argued that
Modern design codes emanate from a probabilistic approach, the use of a flat-arch cavern would make the best use of cav-
for which either the probability of failure (or reliability index) ern space. Thus for this study, the initial cavern width and
can be used to quantify these uncertainties. Advanced numer- height are set as 30 m and 18 m, respectively. The influences
ical methods, which may be continuous or discontinuous, of the flattening process on cavern stability can be investi-
are commonly used in the design of rock caverns. Once the gated through incremental increases in the cavern width and
performance functions are obtained through numerical exper- reduction in the cavern height through six design levels to
imentations and distribution types and characteristic values assess the changes of safety factor and probability of failure.
are determined, several reliability-based design methods can For each configuration of the six levels, the probability of fail-
be used to assess the probability of satisfactory performance. ure is determined by Monte Carlo simulation incorporated by
Combining stochastic approaches with advanced numerical neural network results. The configuration satisfying the criti-
methods will not only provide a systematic evaluation of the cal safety factor and the expected performance level with the
uncertainties but also a means for assessing the impact of these flattest cavern roof can be termed as the optimal design. It
uncertainties on the likelihood of satisfactory performance of is also suggested that the critical factor of safety and the tar-
rock caverns. geted performance level be used together, as complementary
The global stability of underground rock cavern exca- measures of acceptable design.
vated in a Mohr-Coulomb material is investigated by means
684

Deformation and failure mechanism of surrounding rock mass around

underground caverns in Jinping hydropower station I
X.P. Zhou
School of Civil Engineering, Chongqing University, Chongqing, China
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Chongqing University,
Ministry of Education, Chongqing, PR China
Q.H. Qian
Engineering Institute of Engineering Crops, PLA University of Science and Technology, Nanjing, China
Keywords: stability analysis, rock support, tunneling, rock failure, numerical modelling
During the excavation, deformation and failure of surround- have not converged. From the current situation in excavation
ing rock masses around underground caverns of Jinping I site of surrounding rock masses, in the powerhouse, including
hydropower station are analyzed by the numerical simulation main transformer chamber lumbar arch, the sidewall of the
as well as the internal force of rock bolt and presstress anchor traffic tunnel and other parts, there exists the circumferential
cable. Effect of cracks and joints on deformation and failure of deformation and failure characteristics of surrounding rock
surrounding rock mass around underground caverns of Jinping masses after excavation from a high in-situ stress.
I hydropower station is investigated. The stress and displace- The deformation and failure mechanism of crack-
ment fields of surrounding rock mass around underground weakened rock masses around underground caverns in Jinping
caverns of Jinping hydropower station I is obtained as well hydropower station I, which located in a high in-situ stress
as the internal force of rock bolt and presstress anchor cable. area, are numerically simulated by the Extended Finite Ele-
Deformation and failure mechanism of surrounding rock mass ment Method in this paper. The deformation and failure
around underground caverns of Jinping hydropower station mechanism of surrounding rock mass around the under-
I are revealed. Comparison between the numerical results ground caverns is investigated. Finally, comparisons between
obtained from numerical results and monitoring data is carried results from numerical simulation and the monitoring data
out. It is shown that numerical results are in good agreement are made.
with the the monitoring data. Jinping hydropower station I underground powerhouse is
Jinping hydropower station I underground powerhouse located in right bank of the mountain with about 350 m far
mainly consists of the main and auxiliary powerhouse, main away from Dam downstream, in which there are complete
transformer chamber, tailrace surge chamber, penstock, traffic exposed bedrock and steep terrain near the river. Below the
tunnels and vertical shaft. These underground tunnels directly height above sea-level of about 1770 m, the slope angle is
impacts on project investment and construction period. Since about 70∼90◦ . However, above the elevation of about 1770 m,
late April 2008, some spray layered cracks have been detected the slope angle is about 40◦ . Underground caverns are char-
at the downstream side of the crown in the powerhouse, and acterized as intensive, huge and high walls of the main
these cracks continuously develop. In August 2008, crack- cavern, large span and complex construction conditions. The
weakened area has extended to the whole lower part of roof underground powerhouse is located between the exploration
arch. When the fourth layer of the main transformer chamber lines I and VI near PD27 exploration adit whose horizontal
is excavated, cracks occurred in the downstream lower arch depth is about 100∼380 m, and whose vertical depth is about
area, and the cracks developed rapidly. In December 2008, the 160∼420 m. Maximum principal stress near the underground
spray layered cracks appeared in almost all of the downstream powerhouse is 20∼35.7 MPa, the intermediate principal stress
side of main transformer chamber (including the exhaust fan near the underground powerhouse is 10∼20 MPa, the min-
room). Tension cracks are found at the downstream sidewall imum principal stress near the underground powerhouse is
of the traffic tunnel with the length of 3∼4 m. In March 2009, 4∼12 MPa. The direction of the maximum principal stress
the largest open displacement of cracks reaches 10 mm. In varing from N28.5◦ W to N71◦ W with an average value of
addition, in each hole with the vertical layout of underground N48.7◦ W is found. Dip angle of the maximum principal
powerhouse, such as the traffic tunnel, circumferential cracks stress is about 20∼50◦ with an average angle of 34.2◦ . The
are also found. According to the monitoring data and com- underground powerhouse is located in high stress area.
prehensive test data of acoustic emission, rock deformation Underground caverns of Jinping hydropower station I
and failure range were significantly greater than other similar passes through the second, third and fourth layer of mar-
depth and scale of the underground powerhouse. Moreover, ble. The second layer not only contains medium thickness
high proportion of rock bolts and anchor cables are overload- or thickness marble, but also contains thin marble and
ing. Till now, some important parts of the deformation of rock green schist. The main part of third layer contains mainly
685

thickness marble, and other parts of third layer contains 1∼3 m. the third joint set with strike N25∼40◦W, dip NE
medium thickness marble. The fourth layer contains thickness (SW) and the dip angle 80∼90◦ , the fourth joint set with
marble. strike N60∼70◦ W, dip NE (SW) and the dip angle 80∼90◦
Except the f13, f14, f18 and other faults, there are the first in layers of marble. The first bedding fracture set develops
bedding fracture set with strike N30∼60◦ E, dip NW, the dip mainly in the second layer of marble with most of the spacing
angle 30∼40◦ and spacing 1∼3 m, the second joint set with being 20∼50 cm. In the third and fourth marble layers, the
strike N50∼70◦ E, dip SE, the dip angle 60∼80◦ and spacing spacing of cracks is generally larger than 50 cm.
686

Rock slopes and foundations

Importance of anisotropy in dam foundation, estimated by in-situ dilatometer tests
B. Abrah, M. Karami & L. Faramarzi

Department of Mining Engineering, Isfahan University of Technology, Isfahan, Iran
Subject: Rock material and rock mass property testing (laboratory and in-situ)
Keywords: field measurements, anisotropy, deformability, dilatometer, modulus of deformation, joint sets
1 INTRODUCTION Table 1. Specification of geotechnical drilled hole for dilatometer

tests.
An application of the borehole dilatometer test, for the eval-
Number of
uation of the deformation characteristics of the foundation tests and
rock for the design of dam is studied. Then, the directional Engineering interest Rock type depth range
character of the modulus of deformation of anisotropic rock
mass was measured along circumference of a hole during in Roudbar-Lorestan Recent river 29 & 7–48 m
situ expansion tests. These analyses are used to show how dif- rock fill dam alluvium, limestone
ferent degrees of rock anisotropy affect the variation of the and dolomite
modulus of deformation around the walls of a hole in which Zagros Alluvial deposits, 15 & 4–36 m
expansion tests are conducted. soil dam green shale and
Borehole dilatometer tests were performed at the three marly limestone
Azadi Old alluvium, 14 & 8–20 m
dam sites (Roudbar-Lorestan dam, Azadi dam and Zagros rock fill dam conglomerate and
dam) in order to evaluate the deformation characteristics of marlstone
the foundation rock for the design of these three earth dam.
Roudbar-Lorestan HPP is located on the Roudbar river in the
Zagros mountains approximately 100 km south of the town
of Aligoodarz in Lorestan Province, Iran, about 454 km away
from Tehran, the capital of Iran. The Azadi rock fill dam in were conducted in left embankment, 10 tests were carried out
Kermanshah is located in the west of Iran. The Zagros earth in right one and 13 tests were performed in valley bottom
dam of 64 m height and 737 m crest length in Kermanshah along dam axis.
province is located between Ghasreshirin and Gilanegharb The geology of dam site is mostly limestone to dolomite-
towns. limestone which is referred to Dalan formation.The geological
Dilatometer used in this study was the one developed by map and anisotropy variation along Roudbar-Lorestan dam
LNEC which is manufactured by Interfels GMBH Com- axis are shown in figure 1.
pany. The modulus of deformation was computed based on The analyses of dilatometer test results described above
the solution for an isotropic elastic thick hollow cylinder as are commonly used in practice regarding the anisotropic,
follows: discontinuous and heterogeneous rock mass.
The tests carried out with the dilatometer are very quick
deformability can be determined continuously along the bore-
hole and can be considered as a quality index of the rock
formations.
where E = Modulus of deformation/elasticity, P = Pres- The directional character of the modulus of deformation
sure increment, D = Diameter displacement increment, of an anisotropic rock mass was measured along the walls
D0 = Diameter of drill hole and ν = Poisson’s ratio. of a borehole at the three sites intended for dam, most of the
The specifications of geotechnical drilled holes for anisotropy figures were found to be ellipses whose major axes
dilatometer in-situ tests are shown in Table 1. had the geological orientations.
A total of 29 borehole dilatometer tests were performed for A more meaningful result was achieved when the equiva-
the purposes of measuring the mean in-situ modulus of the lent directions in regard to the anisotropy of the studied dam
rock at discrete locations in Roudbar-Lorestan dam site. foundations and dam axes were nearly met by the minimum
The suite of 29 dilatometer tests were divided amongst 7 dilatometer moduli plotted in figures 1.
boreholes drilled in three parts of dam site. Totally, 5 tests
689

Figure 1. Geological map and anisotropy variation along Roudbar-Lorestan dam axis.
690

Application of rockfall risk assessment techniques in two aggregate quarries
L.R. Alejano, M. Veiga & I. Gómez-Márquez

Natural Resources & Environmental Engineering Department, University of Vigo, Spain
H. Dellero
Department of Geology, Faculty of Sciences, University Adbelmalek Essadi, Tetouan, Morocco
Keywords: risks and hazards, mining, case studies, rock slopes and foundations, site characterization
1 INTRODUCTION
ROFRAQ is a statistically-based empirical method for assess-

ing rockfall hazard in hard rock quarries. It is based on the
observation of rockfall-related accident quarries that typically
occurs as a result of 5 sequential events, as follows:
1. A detached block/rock mass exists on a slope.
2. The block/rock mass is close to equilibrium (under any
given instability mechanism).
3. A triggering phenomenon makes the block/rock mass
unstable.
4. The block/rock mass fall path is such that one or more
blocks reach the quarry bottom.
5. At least one block hits a worker or a machine.
The probability of an accident occurring, which is the prob-
ability of these 5 events taking place sequentially, can be Figure 1. Sketch of the quarry and two opposite views.
calculated by multiplying the individual factor probabilities.
Then a corrective parameter is added based on the history of Table 1. Joint set geometric and geomechanic features.
rock falls that occurred in the quarry.
The results of ROFRAQ ranges between 0 and 10000, and SETS J1 J2 J3
slopes can be classified in 6 categories with recommendations.
◦
Dip dir.* 258 (12) 360 (20) 079 (19)
◦
Dip* 66 (14) 89 (14) 45 (12)
Persist. m 15(10–20) 12 (10–20) 7 (3–10)
Spacing m 1,9(0,6->2) 2 (0,6->2) 2 (0,6->2)
2 ENLARGEMENT OF A GNEISS QUARRY DESIGN JRC* 7 (3) 6 (3) 9 (3)
JCS* MPa 138 (58) 183 (50) 138 (56)
An aggregate quarry (fig.1) was exploited for 25 years and was Width mm 0–1 0–1 Closed
about to achieve its final design stage. But, since the quarry Fill Oxides Oxides —
◦
was sitting on a significant amount of exploitable deposits, the Wheater. I–II I–II I–II
owners decided to extend the life of the quarry by deepening Water Dry-hum. Dry Dry
the bottom of the quarry.
The study of discontinuities identified 3 different join sets * Parameters where average and standard deviation values appear.
(Table 1); also samples were taken from the quarry and tested
in order to obtain the rock’s basic information. The val-
ues of cohesion and friction of the joints were calculated, With this in mind, we have designed a new design for the
and the instability mechanisms were identified by means of quarry. Rock was very hard so circular slope failure was not
hemispherical projections techniques. possible. For the case of 45◦ slopes, which is the present one
Based on the statistical data from the discontinuities study, no failure mechanism is cinematically possible, except for
safety factors was identified for every mechanism in dry, 50% block toppling in the central slope (which will not take place
saturated, fully saturated and seismic conditions. considering the persistence and spacing of joints). The gen-
The application of ROFRAQ to the present four slopes eral slope of the quarry was kept to 45◦ , and since according
of this quarry let us obtain values in the range 15 to 35, to ROFRAQ results of rockfall risk was low to average, it
corresponding to low and low to average risk. was considered that if the slope retains 90% of rock blocks,
691

failures were identified, confirmed in a similar slope of a
neighbor quarry. Also wedge failures were identified in the
northeastern slope (streak N-135◦ ) formed by two joint sets
and another one acting as a tensile crack. These wedges can
be very unstable under saturated conditions (fig.2).
The projection of joint traces derived from field data in the
slope face has been helpful to quantify continuity and include
geometry in the SF calculation in an indirect way. Even if
there are now advanced techniques to perform this task, it is
still difficult to obtain reliable persistence estimates in situ to
properly apply these techniques.
The ROFRAQ method has been also applied to find out that
most of the slope orientations can be rated as low risk ones,
except those two walls where problems were detected in the
basic geotechnical study that yield ratings of 127 (E-wall) and
55 (NW-wall). However, since this last slope, runs parallel to
Figure 2. Stability analysis of a wedge in a quarry. Different safety the boundary of the quarry, it was decided to keep it, and it was
factors obtained. Stereographic, 3D representation and picture of
recommended to create 12 meters wide berms, and restrict the
such a wedge in a quarry bench.
exploitation to the dry season.
4 CONCLUSIONS
The incorporation of rockfall quarry assessment and control

techniques (Alejano et al., 2007 & 2008) in the process of
quarry design and planning has proved to be a useful deci-
sion making tool. Therefore, ROFRAQ cannot only be used to
quantify rockfall hazard, but also to propose practical recom-
mendations regarding slope directions and catch bench width,
to achieve a better degree of convenience and security in mine
and quarry exploitations.
It is important to bear in mind that the quantification of
data for natural materials such as rock masses is invariably
subject to a certain degree of subjectivity and uncertainty.
Consequently, ROFRAQ should be used as a guideline. It is
relevant to note, that for design purposes, these approaches
should be used in combination with traditional general slope
stability studies (Wyllie & Mah, 2004).
Figure 3. Picture of a bench in the NW slope, with traces of the
joints observed. In the upper right part, a sketch of the quarry and ACKNOWLEDGEMENTS
the stereographic representation of the main joint sets are included.
The authors acknowledge the funding received from the
safety conditions were fulfilled. In this way, for the conditions Regional Government of Galicia (Consellería de Inno-
encountered the rockfall control method developed prescribed vación e Industria, Xunta de Galicia), for financial support
a catch-bench width of 7 m, increased in 1 m to account for of the research project under contract reference number
back-break problems. INCITE08PXIB304076PR, which has been of help in some
of the research topics presented in the paper.
3 STUDY OF A SERPENTINITE QUARRY

REFERENCES
In this case we study a serpentine quarry that opened four
years ago that produces aggregates for railway gravel foun- Alejano, LR., Pons, B., Bastante, F.G., Alonso, E., Stockhausen, H.W.
dation. The deposit extends to a depth of around 50 meters 2007. Slope geometry design as a means of rockfall control in
quarries. Int. J. Rock Mech. & Min. Sci. 44: 903–21.
therefore only three 10 m high benches were considered. The Alejano L.R., Stockhausen, H., Bastante, F.G., Alonso, E., &
owner wanted to find a suitable way to plan quarrying for the Ramírez-Oyanguren, P. 2008. ROFRAQ: A statistics-based empir-
following years and some guidelines to prepare a final design. ical method for assessing accident risk from rockfalls in quarries.
The same process of geotechnical survey, lab testing and Int. J. Rock Mech. & Min. Sci. 45: 1525–72.
statistical data analysis was performed. Results have shown Wyllie, D.C. & Mah, C. 2004. Rock Slope Engineering: Civil &
that in the eastern slopes (streak N175◦ ) some unstable planar Mining. 4th ed. Spon, London.
692

Seismic stability analysis of rock slopes using the numerical manifold method
X.M. An & Y.J. Ning

G.W. Ma
School of Civil and Resource Engineering, University of Western Australia, Crawley WA, Australia
Z.Y. Zhao

Keywords: Numerical modeling; Risks and hazards; Stability analysis
Seismic stability of rock slopes is numerically investigated φ (friction angle) > α (slope inclination angle), and unstable
using the numerical manifold method (NMM). A paral- for φ < α. This study investigates the horizontal ground accel-
lelepiped block resting on an inclined plane as shown in eration induced block instabilities based on these two cases.
Figure 1, which has been considered as a first approxima- Two different instability mechanisms are observed. The hor-
tion of the rock slope stability conditions, is employed in this izontal ground acceleration enlarges the destabilizing force,
study to find out the fundamentals of seismic instabilities of which makes the originally moving block (i.e., φ < α) move
rock slopes and to validate the NMM for seismic instability faster (e.g., Figure 4 for a block under a sinusoidal function)
analysis of rock slopes. and produces the originally stable block (i.e., φ > α) a finite
The block under gravitational loading only is first con- displacement (e.g., Figure 5 for a block under a sinusoidal
sidered to calibrate the NMM and to determine the control function & Figure 6 for a block under an arbitrary ground
parameters for the NMM calculation. A regularly-patterned acceleration record).
triangular mesh is used to generate the NMM models. The A vertical ground acceleration is lastly introduced and
proper mesh size is first determined by gradually refining its induced rock slope instability mechanism is discussed.
the mesh until the predicted displacement close to the ana- Consider a half-sine vertical ground pulse acceleration with
lytical solution. To this end, totally three different meshes are amplitude of a0 and duration of T. We assume φ = 30◦ and
investigated. Numerical results indicates that for such a sim- α = 25◦ . The block is initially stable since φ > α. However,
ple problem without high stress gradient, a very coarse mesh when a vertical pulse acceleration is introduced, the effec-
as shown in Figure 2 will produce an accurate result (with tive friction angle φ̃ vibrates around the static friction angle
an error of only 1.94%). The proper time step size t is then φ. For the case of a0 = 0.8 g, T = 0.8 ms, the φ̃ has a mini-
determined by being gradually reduced until the predicted dis- mum value of 6.71◦ (Figure 7), much smaller than φ, which
placement close to the analytical solution. To this end, totally is termed as ultra low friction phenomenon. The block moves
six different t, i.e., 0.01s, 0.008s, 0.005s, 0.002s, 0.001s when α > φ̃. Vertical pulse acceleration makes the originally
and 0.0001s, are considered. The errors to the analytical solu- stable block unstable and provides the block a finite displace-
tions of the displacement are 42.57%, 15.39%, 7.85%, 4.70%, ment. Its effect highly depends on the duration and amplitude
2.22%, 1.94%, respectively. Although when t = 0.002s, the of the pulse acceleration. Various vertical pulse accelerations
NMM predicted displacement is already acceptable within with the same amplitude but different pulse acceleration are
an engineering accuracy, we choose an even smaller t of
simulated. Figure 8 plots the variation of minimum φ̃ with
0.0001s throughout this study in order to achieve more accu-
the duration T. It is observed that there exists a critical dura-
rate results. Using the determined control parameters, five
tion, Tc . The minimum φ̃ decreases with the duration before
cases with different friction angles, i.e., 0◦ , 5◦ , 10◦ , 15◦ , 20◦ ,
are studied. The NMM predicted horizontal displacements
agree well with the analytical solutions for all the five cases,
as shown in Figure 3. Validity of the NMM for gravity induced
displacement is calibrated.
Horizontal ground acceleration is then introduced and its
induced rock slope instability mechanism is investigated.
When only under gravitational loading, the block is stable for
Figure 3. Block sliding under gravitation loading only: friction

Figure 1. A single block Figure 2. NMM model for the angle effect (solid line – analytical solution; discrete point – NMM
resting on an inclined plane. single block problem. results).
693

Figure 4. Block displacement under a sinusoidal horizontal ground
acceleration, φ = 20◦ , α = 25◦ .
Figure 6. Block response under an arbitrary horizontal ground
acceleration record, φ = 20◦ , α = 25◦ .
Figure 7. Effective friction angle under a vertical pulse accelera-

tion, a0 = 0.8 g, T = 0.8 ms.
Figure 8. Variation of minimum φ̃ with pulse duration.

Figure 5. Block displacement under a sinusoidal horizontal ground
acceleration, φ = 28◦ , α = 25◦ .
Tc and increases with the duration after Tc . Pulse amplitude

effect is also investigated. Pulse duration is kept as a constant
of 0.8 ms, while the amplitude varies from 0.1 g to 0.9 g. The
minimum φ̃ decreases almost linearly with the amplitude of
the pulse acceleration, as shown in Figure 9.
In summary, seismic instability of rock slopes is numeri-
cally investigated using the NMM. Instability mechanisms of
the rock slopes under horizontal and vertical ground accel- Figure 9. Variation of minimum φ̃ with pulse amplitude.
erations are revealed, respectively. The validity of the NMM the analytical solutions and Newmark-β numerical integra-
in predicting the ground acceleration induced permanent dis- tion solutions. Furthermore, the proper values of control
placement has been verified by comparing its results with parameters for NMM calculations are suggested.
694

Stabilization of a landslide in Valgrisenche, Italy
G. Barla, F. Antolini & M. Barla

Department of Structural and Geotechnical Engineering, Politecnico di Torino, Torino, Italy
Keywords: numerical modeling, rock slopes and foundations, risk and hazard, rock support, site characterization, stability
analysis
In April 2009, after a period of heavy rainfalls, a debris slide

with an approximate volume of 35,000 m3 occurred alongside
the SR25 road between Valgrisenche and Arvier villages in the
Aosta Valley (North Western Italian Alps).
The landslide overpassed the retaining wall built along-
side the road and a large amount of rock debris accumulated
over the road itself (Figure 1). After an initial period during
which the road was realigned on a provisional track reducing
the carriageway width to a permanent single-way road work-
ing with traffic lights, the continuous slope movements forced
the local authorities to close the road to traffic. The SR25
road represents the only connection between Valgrisenche
and the main Dora Baltea valley and its closure has led to
the isolation of several villages. Hence opening the road was
therefore required as a matter of urgency in order to restore
the road communication with the main valley. Urgent slope
remedial works started immediately after the landslide occur-
rence while slope movements were still taking place, posing
safety and logistic problems for both ground investigations
and remedial measures to be undertaken.
The geology of the site comprises heavily jointed parag-
neisses, micaschists and calcschists with intercalations of
metabasites overlain by a thick colluvial detritic cover where
the landslide has developed. The slope failure involves a pre-
dominantly rotational movement in the upper slope sector
which evolves in a superficial translational slide along the
slope. Further superficial rotational movements developed at
the bottom bench during slope reprofiling works.
Ground investigations included 4 seismic refraction lines
and 3 boreholes. In addition 4 inclinometer tubes and 3
piezometer boreholes were installed at the site to monitor the
slope movements and the groundwater level versus time.
The groundwater level tended to stabilize at a depth of about
25 m, along the contact between the bedrock and the colluvial
deposits. As a consequence of rainfalls and the snow melting Figure 1. (a) Geologic and geomorphologic map and (b) photo-
during the spring-summer period, groundwater level increases graph taken in May 2009 of the landslide area.
and saturates the colluvial deposits.
Back analysis with both the limit equilibrium method
(LEM) and the finite element method (FEM) allowed one to
evaluate the residual shear strength parameters along well- verified by LEM and FEM analyses, suggesting the necessity
defined slip surfaces detected by inclinometers. Removal of for integration of the preliminary design.
the failed rock material, reprofiling of the slope, drainage mea- The design of the stabilization works was constrained by
sures, and installation of pre-stressed and passive anchors, the priority to allow for quick opening of the the SR25 road.
together with the rebuilding of a preexisting damaged retain- As already mentioned, the SR25 road was partially opened at
ing wall were designed in order to improve the slope stability the end of May 2009 by providing a provisional single-way
conditions. The effectiveness of the stabilization measures was track working with traffic lights.
695

The safety factor along the sliding surfaces considered dur-
ing the back analysis was considerably increased (from 1.002
and 1.042 to 2.015 and 2.541 respectively). Some superfi-
cial slip surfaces, trending almost parallel to the slope, still
showed safety factors less than 1. In order to remove such
unstable surfaces, the following improvements were adopted:
• Installation of fully grouted dowels, 9 m long, 20 mm
diameter, 3 × 3 m spaced.
• Installation of a further row of passive anchor bars along
the row B tieback wall (20 m long, 40 mm diameter, 15◦
and 20◦ alternating inclination).
• Increasing length of all the other passive anchor bars up to
20 m.
Thanks to the installation of rock bolts, the superficial slope
displacements are considerably reduced (one order of mag-
nitude), the loads in the retaining/reinforcement structures
decrease and the overall slope stability improves. The pro-
cess described above was repeated by the corresponding FEM
analyses with the SSRM. The slope safety factors resulted to
be equal to 1.32 with the above mentioned improvement of
the original design (Figure 2a).
The remedial works and stabilization measures lasted till
the end of the summer, to be finally interrupted due to snow
falls. Figure 2b shows a photograph of the slope in Novem-
ber 2009, when most of the works were completed, with the
exception of the installation of the dowels and the passive
anchor bars. The pre-stressed anchor heads were connected
along each row by tieback walls in order to ensure a mechani-
Figure 2. (a) Results of the FEM analysis and (b) photograph of the
slope in November 2009, with the most important remedial works
cal interaction between single elements and provide continuity
completed. along the horizontal plane.
The particular slope topography and the presence of boul-
ders up to 10 m wide, forced to adopt discontinuous wall
stretches in spite of a single continuous wall for each pre-
stressed anchor row. The final lock-off loads for the pre-
The following stabilization measures were considered: stressed anchors were between 200 and 248 kN. Load cells
• Reprofiling of the slope by building three benches, removal were subsequently installed on 10 anchors in order to monitor
and stabilization of the unstable cover. their performance.
• Installation of a deep drainage system (sub-horizontal Load cells readings have been collected monthly after
drains). installation showing an increasing load in the anchors follow-
• Installation of three rows of pre-stressed anchors, 35 m long, ing a preliminary relaxation; only anchors T4 and T7, located
with inclination between 10◦ to 15◦ . on the upper north-western landslide sector, show a moderate
• Building of a new counterscarp wall founded on micropiles load reduction. Time interval is too short to draw any con-
and reinforced with a double row of passive anchor bars clusion, and landslide monitoring should continue in order
(15 m long, 40 mm diameter, 15◦ and 20◦ inclination). to identify any anomalous behaviour of the support measures
• Extension and reinforcement of the pre-existing damaged implemented and to check continuosly the overall behaviour
retaining wall with a single row of passive anchor bars (15 m of the slope.
long, 40 mm diameter, 40◦ inclination). Particular attention was posed on draining the groundwater
in the colluvial deposits. The final drainage system consists
The design of the stabilization measures was carried out by of 80 sub-horizontal drains installed along the tieback walls,
making effective use of both LEM and FEM methods. The 25–30 m long, 1 m spaced and 4◦ –5◦ inclination. The effect of
slope geometry was modified to take into account the stabi- the drainage system is particularly effective in the upper and
lization measures. Appropriate mechanical parameters were more critical sector of the landslide where the piezometric
adopted for the support elements. level showed a 6 m lowering since drains installation.
696

Development and characteristics of a deep-seated slope movement
Ya-Chu Chiu
Chia-Han Lee & Tai-Tien Wang

Tsan-Hwei Huang
Keywords: field measurements, site characterisation, rock slopes and foundations
1 INTRODUCTION
Generally describing with huge movement mass with slow

movement rate, a deep-seated slope movement or deformation
induced gravitationally is usually difficult for identification.
Potential hazards caused by deep-seated slope movements are
long-term disregarded and paid less attention for mitigation.
Therefore, characterizing a deep-seated slope movement and
thus evaluating the potential for slope instability are essential
for hazard mitigation.
Example of a deep-seated slope movement in southeast Tai-
wan, this study investigates the development of the movement
masses and accordingly characterizing the movement.
2 STUDY AREA AND METHODOLOGY Figure 1. Topographic map of the study area.
Figure 1 shows the topography of the study slope. The rock of modern geodetic total station to acquire 3D global coor-
nearby the slope is mica schist with schistosity dipping to dinates of the points on tunnel walls, with a precision of
northwest. A single-lane highway tunnel built in 1972 suffered 2–3 mm. Displacement of each profile is then obtained. Also,
from significant squeezing after completion of construction, the surface condition of tunnel lining walls is recorded by
and a double-lane tunnel was built as a substitute for the col- image-mosaic technology through photography and image
lapsed one and completed in 1991. The slope west of the tunnel processing. Finally, discussion about site characteristics and
is active. Lining cracks developed approximately two years associated tunnel displacement are made.
after tunnel completion. Repair measures were implemented
in 1998. Surface geological survey has been done since early
1970’s. However, the instability of the slope was recognized 3 INVESTIGATION RESULTS AND DISCUSSION
until 1992 after the completion of the existing tunnel. Since
that a series of field investigations and monitoring have been Six aerial photos taken through 1981-2006 are used to identify
implemented. Rock formation is than revealed from results the multitemporal ground variations on study area. Aerial pho-
of 17 boreholes and geophysics investigation. And monitor- tos interpreting results show that the instable ambits were local
ing records of inclinometers and ground displacements are and associated sliding surfaces were not deep initially. The
available for the past two decades. construction of a forest road and reclamation activity resulted
First, multitemporal aerial photos are utilized to interpret in the enlargement of instable ambits. Bank erosions leaded
the possible ambits of movement masses. Field investiga- to an erosion gully at a corner of the slope toe, which wors-
tions focusing on phenomena of slope movement are carried ened the slope stability by means of headward erosion. The
out to confirm the ambits. The deformation of the exist- headward erosion is sped up by the steep gradient the water
ing tunnel is monitored using a micro-displacement survey flowing down as well as the schistosity dipping down, and
technology which integrates Differential Global Positioning reduced the stability of the slope above the highway. Addi-
System (DGPS) and Automatic Target Recognition (ATR) tionally, the affecting area caused by the slope movement near
697

Figure 2. Ground surface movements and locations of discussed
profiles of the highway tunnel.
Figure 4. Horizontal ambits (a) and sliding surfaces (b) of move-

ment masses.
Figure 3. Movement of 4 typical tunnel cross-sectional profiles
(2009.02-2010.04).
33 and 99 m from the portal are mostly significant. The spa-
the erosion gully was magnified by the ubiquitous schistosity tial distributions of these two major cracks paralleled the slope
due as it has low strength and high deformability. The distribu- movement direction (Fig. 2). The sliding surface of movement
tive movement masses assembled gradually, meanwhile, the masses A and C are then identified.
sliding surfaces went deep; resulting in the three movement
masses A, B, and C (Fig. 2).
A series of instrumentations have been installed since 1999 4 CONCLUSION
to monitor slope stability of the study area. Figure 2 shows the
horizontal ground movement calculated from the inclinometer Three characteristics of the studied deep-seated slope move-
monitoring results. The displacement vectors measured inside ment can be identified: 1) Sliding surfaces are confirmed, at
movement masses point toward northwest, almost parallel the least for movement masses A and C with maximum depths
dip direction of rock schistosity, but not down-hilly to the of 75 m and 35 m (Fig. 4), by the inclinometers monitoring
northeast, which imply that the movements are affected by results and the tunnel deformation measurement. The ratios
the schistosity. Ground settlement and tunnel displacement of sliding depth to length are thus 0.17 and 0.16, respectively.
have been monitored every 6 months since Feb. 2009. The 2) The movements for distinct movement masses are inde-
ground settlements are limited during the first half year, and pendent. For a heavy rainfall induced movement, the upper
dramatically increase in the second half year with a maximum movement mass (C) moves quicker and greater than the lower
magnitude of approximate 200 mm at point Q indicated in one (A). 3) The movement rates vary with times. Once the
Fig. 2. movement is triggered, for instance, by a heavy rainfall, the
Tunnel cross sectional profiles are surveyed at an interval of movement rate is suddenly increased. The movement contin-
5 m using the micro-displacement technology. Two profiles, ues more than one year with a decreasing movement rate. And
① and ②, located in movement masses C and A translated to the movement mass may temporarily stop for a long time.
north and settled down significantly (Fig. 3). And the trans- Finally, schistosity plays a key role for the studied deep-
lational component of profiles ③ and ④ located outside the seated slope movement. The steep gradient of the slope inside
movement mass are limited. Additionally, there are numer- the movement slope is 35◦ , implying that the long-term appar-
ous cracks appeared in tunnel lining in the section 25–110 m ent friction angle for the schistosity should less than this
from the western portal. Among which the cracks located value.
698

3DEC modelling of p-y behaviour of laterally loaded piles in jointed rock
W.L. Chong, A. Haque & P.G. Ranjith

Department of Civil Engineering, Monash University, Victoria, Australia
A. Shahinuzzaman
Powerlink, Queensland, Australia

Keywords: numerical modelling, rock joints, rock slopes and foundations, weak rock
Pile foundations are often employed to provide support for

many civil engineering infrastructures to resist lateral loads.
For the analysis of laterally loaded piles, the p-y method,
also known as subgrade reaction approach, is widely used.
Extensive research has been conducted and well established
p-y methods based on continuum approach have been devel-
oped for soils. For laterally loaded rock socketed piles, the
existing p-y methods have extended the continuum theory by
adopting rock classification systems such as Rock Mass Rat-
ing (RMR) or Geological Strength Index (GSI) to simplify
the rock mass to a homogeneous condition. However, rock
in reality is a highly discontinuous medium where secondary
structures such as fractures and joints are common occurrence.
Therefore, the assumption of homogenous and continuum
rock condition may not be accurate in modelling the rock
mass behaviour. With such limited knowledge, current design Figure 1. Normalised p-y curves.
methods can either produce conservative or unsafe designs.
This implies that further research is required to develop bet-
ter understanding of the complex p-y behaviour of laterally
loaded piles socketed into jointed rock mass.
In this study, 3DEC was employed to investigate numer-
ically the complex interaction between the joint dip angle
and the p-y response, as well as the pile head load-deflection
behaviour of laterally loaded rock socketed pile. The initial
numerical model was first calibrated against two laboratory
model lateral pile tests socketed into jointed mudstone. The
calibrated model was subsequently extended to incorporate a
1-joint set model having a dip angle 45◦ and the p-y curve for
this model was then derived. The normalised p-y curve derived
for the laboratory model was in good agreement with the field
load test results (Figure 1). Subsequently, different joint dip
angles (30◦ and 60◦ ) were simulated and the p-y curves at dif-
ferent depths along the pile and the pile head load-deflection Figure 2. p-y curves for different joint dip angles.
responses were determined.
Figure 2 shows the p-y curves at different depths (25 mm,
50 mm and 75 mm) along the pile socket length for the three
joint dip angles of 30◦ , 45◦ and 60◦ . The p-y response becomes
stiffer with increasing depth especially in the non-linear sec-
tion of the curves which is due to higher overburden stress. It
is also to be noted that at a given depth, the p-y response was
found to be independent of joint dip angles (Figure 2).
The pile head loads and corresponding deflections were
obtained from the 3DEC simulation for all joint dip angles
(Figure 3). It can be seen that the linear response of the pile
which is crucial for practical engineering design is similar for
all the dip angles simulated. This study found that a lateral
displacement up to 10% of pile diameter where the load-
deflection response remains about the same while the ultimate
lateral capacity increases with steeper dip angles (Figure 3). Figure 3. Load-deflection for different joint dip angles.
699

Field test and interpretation of the side shear resistance of a drilled shaft in rock
D. del Olmo & C. Olalla

Department of Engineering and Morphology of the Terrain, Universidad Politécnica, Madrid, Spain
S. Melentijevic & L. Prieto

RODIO – KRONSA
Keywords: pile foundation, field measurements, case history, back analysis.
A static load test was performed on a drilled shaft in a flysch Considering those relations, their values should be as close
rock. The properties of the rock mass were extensively inves- as possible to a safety factor of 1.00, so it can be concluded that
tigated. The side shear resistance of the drilled shaft has been these theories are very conservative. This means that whenever
carefully studied during the load test. Two Osterberg cells were a pile is designed using these expressions, an additional Safety
embedded into the pile, along the 5 meter pile segment in the Factor (SF) is introduced in the analysis in addition to the one
good quality flysch rock, in order to introduce the load. Using considered for the pile. The synthesis of this comparison is:
laboratory test results, analyzing the field load test and mod-
eling the drilled shaft by the finite element method, the side
shear resistance transfer mechanism between the shaft and the Value
rock was reproduced. Author <MPa> SF
The test was carried out using two Osterberg cells embedded
in the pile, as have been said. The cells were separated from Rossenberg and Journeaux, 1976 2.45 1.43
each other. One was situated at half the tested height of the Horvath and Kenney, 1979 1.45 2.41
drilled rock shaft and another one near the base of the pile. Williams and Pells, 1980 1.77 1.97
The test was planned in two stages; during the first one only Rowe and Armitage, 1986 4.15 0.84
the bottom cell loaded the pile; while in the second stage only Carter and Kulhawy, 1988 1.38 2.53
the upper cell worked.
In addition to the results obtained from the load test by
analyzing settlements and mobilized shear forces, a review of A FEM analysis has been done, modeling the rock socket as
some of the existing empirical criteria to estimate the ultimate Horvath et al. (1983) proposed. This criteria causes the shaft-
side shear resistance of a pile has been presented. rock wall to be modeled as a toothed surface. According to
The following are conclusions from the load test results: the UCS of the rock and the excavation method, the shaft-
rock contact was modelled as a rough surface. This model has
– Neither the pile nor the rock failed during the test. allowed the load transfer mechanism to be represented from
– The analysis of the side shear resistance during the first the initial stages of loading to the failure. Analyzing FEM
stage of loading was complex due to the interaction between results it can be observed that the failure value obtained from
the base and the shaft of the pile. the analysis is very accurate to the one obtained in the load test.
– During the second stage of the test, only side shear resis-
tance was mobilized. Although the ultimate side shear
resistance could not be reached it was deduced that the
strength was near 3.5 MPa. The estimated value for ultimate
unit friction at failure obtained by the load test was com-
pared to the results obtained by the empirical relationships,
depending on the rock characteristics.
700

What modern rock mechanics owe to the Malpasset arch dam failure
Pierre Duffaut
French Committee on Rock Mechanics, Paris, France
Subject: Other: Dam failure case study
Keywords: case studies; weak rock; field measurements; rock joints; stability analysis; risks and hazards
1 INTRODUCTION
Since the most ancient construction works a rock foundation

was enough for safely building on it. On December 2, 1959,
the Malpasset arch dam, close to the French Côte d’Azur,
suddenly broke and released 50 hm3 , with huge destructions
and more than 400 fatalities. About this time, hundreds of arch
dams had already been built in the US, Japan, and Europe and
hundreds have been built since, higher and higher, without
any more accidents. The event initiated a lot of analyses and
studies, both in situ and at labs, in France and worldwide.
Clearly, ISRM was born from this failure.
2 THE FAILURE
Designed for water storage by famous Consulting Engineer Figure 1. Remains of Malpasset arch dam: a conspicuous crevice
A. Coyne, and built 1952–1954 in a narrow gneiss gorge, the appears along the concrete shell; the whole river discharge is flowing
60 m high dam has never been filled up before the occurrence below the dam (photo P. Duffaut, 1960).
of huge rains in November 1959. The failure was assigned to
water pressure through a unique combination of features:
3 MAIN LESSONS
i) high sensitivity of rock permeability to stress, creating a
deep impervious barrier below the dam; Two traps had mistaken the people in charge: the rock mass
ii) high deformability of rock, helping open a deep crevice “look”, as seen by many visitors, was treacherous, and the
behind the dam heel (Fig. 1); administrative status was rather low; nobody had realized this
iii) a fault dipping upstream ca.20 m downstream of the dam; dangerous object had been left without convenient control.
iv) lack of boreholes for foundation drainage. The main lesson from Malpasset failure was that, inside a
It must be added that the monitoring and supervision of the dam foundation, water forces may be the same order of mag-
dam were rather poor, as it was left unused by lack of any nitude that dead weight and dam thrust. All dams are only
water distribution network (nobody noticed a slight move of gravity dams under condition to include the weight of ground
the dam towards left side); lastly, the construction works of a abutments and engineers are the only ones, in the geotechni-
motorway bridge was glad no water could flow through. cal field, to be fully aware of water force, a true revolution
For having taken part himself in the investigations on many following the long time needed to understand uplift inside
dam sites in France at the same time (1948–1955), the author rock mass.
bears witness that the rules of art had been followed: As stated Before, only mining engineers had begun to formalize a
the technical literature, main dam site problems were bedrock meager corpus of rock mechanics. Since, civil engineering
depth below the valley floor, and bedrock imperviousness, Of and geological schools, labs and bureaus, as well public as
course all rules were quickly changed after the event: more private, developed an array of new concepts and methods,
site investigations became mandatory, how sound the rock which first appeared at the 8th ICOLD conference in 1964
may look; uplift drainage under gravity dams was extended to and later at ISRM ones, beginning in Lisbon, 1966. Civil engi-
arch dams; and the State control on all dams over 15 m height neers met petroleum and mining engineers to build together a
was reinforced. comprehensive Rock Mechanics.
701

Prediction of ultimate bearing capacity of discontinuous rock foundations

on the basis of limit analysis
A. Fahimifar & M. Imani

Department of Civil and Environmental Engineering, Amirkabir University of Technology, Tehran, Iran
Keywords: numerical modeling, rock failure, rock joints, rock mass, rock slopes and foundations
The ultimate bearing capacity is an important design consider- for determining the bearing pressure from the load-settlement
ation for heavy structures such as concrete dams, bridge piers, curves. The ultimate bearing capacities obtained in this man-
and multistory buildings in jointed rock masses. Because of ner are compared by the values obtained from an upper bound
existence of discontinuities within the matrix material, the limit analysis formulation developed in this paper. Results of
bearing capacity of rock masses is governed not only by the this comparison show that by increasing the degree of joint-
intact matrix material but also by the characteristics of the dis- ing, the ultimate bearing capacity of rock foundations will
continuities. According to the joint spacing in comparison to decrease. Also, the bearing pressure values obtained by the
the structure dimensions, two approaches called the direct and upper bound method are greater than the values obtained by
the homogenization approaches are applied for determining the distinct element using the Log-Log method.
ultimate bearing capacity. The effect of groundwater table on the ultimate bearing
In this paper, for the case of a non-homogeneous rock foun- capacity of rock foundations was also investigated. Using the
dation with large joint spacing, the ultimate bearing capacity distinct element method, different positions of water table were
is investigated using distinct element method. The distinct ele- analyzed and the reduction of ultimate bearing capacity was
ment modelling are performed for rock masses containing two investigated. It is found that in the cases analyzed, the maxi-
orthogonal joint sets in which, the orientation of the first joint mum reduction of bearing capacity caused by water pressure is
set was assumed to be 15◦ , 30◦ and 45◦ . Different spacing val- about 10 percent. By increasing the degree of jointing of rock
ues for joints were considered and load-settlement curves were foundation, the decrement of the ultimate bearing capacity
obtained in each case. Then, the Log-Log method was applied would increase.
702

Simulating the failure mechanism of rock slopes (due to kinked and secondary cracks
propagation) by a higher order displacement discontinuity method
M. Fatehi Marji & M. Eghbal

Faculty of Mining and Metallurgy Engineering, Yazd University, Yazd, Iran
Mine and Industrial Organization of Yazd, Yazd, Iran
Keywords: crack propagation mechanism, rock slope stability analysis, numerical methods, displacement discontinuity
method, LEFM
Numerical modeling of rock slopes is now used routinely in In this paper, the Fictitious Stress Method (FSM) is used to
the civil and mining engineering sectors as well as in academic discretize the boundary of the problem in two dimensions and
research of numerical applications available today. The stabil- the higher order Displacement Discontinuity Method (DDM)
ity of rock slope is mainly determined by its discontinuity and with kinked and special crack tip elements is used to dis-
Rock Bridge. However, the failure mechanism of discontinuity cretize the original cracks, the kinked cracks and the secondary
and Rock Bridge has not been studied comprehensively. In this cracks. A mixed mode fracture criterion known as maximum
paper, the stability analysis of jointed rock slope is carried out tensile stress criterion is used to predict the direction of crack
by kinked and secondary cracks propagation using the higher initiation and the crack propagation paths. This paper presents
order displacement discontinuity Method. This study presents a method for determining the effect of crack propagation in
a new approach for stability analysis of jointed rock slope. It rock slope stability analysis by the higher order Displace-
can also consider the complicated geological conditions and ment Discontinuity Method (DDM) which is a relatively newly
supported slopes. Fracture mechanics is the field of mechan- developed method for determining the effect of fracturing in
ics concerned with the study of the formation of cracks in the stability analysis of rock slopes. Here, some failure areas
materials. Previous researches have shown that Griffith’s brit- have been shown where the cracks can grow. The modified
tle fracture theory can be modified to account for the effects of DDM program can evaluate the effect of cracking of rock
crack closure in compression. They have also shown a useful slopes on the failure surfaces. Finally a real case study (the
basis for the study of the fracture of hard rocks based on the second tectonic block of Choghart iron mine) is solved numer-
modified Griffith theory. An analysis of the stress distribution ically and it is shown that there were some failure planes on
around a crack indicates the points of fracture initiation as which the joints may start to propagate. It has been concluded
well as the initial direction of crack propagation. As a result that the wing cracks can be produced in rocks under both
of the changes in stress distribution associated with fracture compressive and tensile loading conditions. Further more it
propagation it is, however, impossible to predict the final path may also be concluded that the secondary cracks which are
of the propagating crack. Consequently, a serious limitation produced due to shear stresses induced under compressive
of the Griffith theory lies in the fact that it can only be used loading may also be produced under tensile loadings (similar
to predict fracture initiation. Recently, several researches have to the wing cracks).
shown that the Mode II fracture toughness of rock material is
usually higher than the Mode I fracture toughness, especially
when the confining pressure increases.
703

The assessment of stability of grand Dallagkaus landslide in reservoir of

Zaramag hydro power development, Russia
E.G. Gaziev & V.V. Rechitski

Geodynamic Research Center, Moscow, Russia
Keywords: stability analysis, rock slopes and foundations, numerical modelling, case studies
1 DESCRIPTION OF GRAND DALLAGKAUS SLIDE mass movements which coincided with the contour of the
sliding mass.
The Grand Dallagkaus landslide is situated on the The Dallagkaus slide analyzed with the aid of a 3D model
Mamisondon River which is a left bank tributary of the Ardon (considering the width of the sliding body at all elevations),
River. The depth of the Dallagkaus landslide is up to 80 m and has shown that in the water saturated state (w = 28.5% at
its volume is about 13 mln m3 . The length of the slide body γR = 2.34 t/m3 ) it may pass in the limiting state of stability
is about 950 m, while its averaged width varies from 80 to at the angle of friction ϕ = 20.5◦ and cohesion C = 15 kPa.
400 m. The Grand Dallagkaus slide is devoid of wood cover According to the studies conducted on other slides situated
while the surrounding stable areas are covered with rather in this region, the following strength parameters for shear
a dense forest, which serves the evidence of the slide body ϕ = 24◦ and cohesion C = 25 kPa have been taken for the
movements. natural moisture conditions of the Grand Dallagkaus slide.
The landslide stays in the condition of limit equilibrium Analysis of the Quaternary deposits in the natural mois-
and its movements make 50 mm per year. And in spite of slow ture condition has yielded the safety factor of the slide mass
downward movement of the sliding mass these displacements, k = 1.20, while in the water saturated state, the safety factor of
in case of their re-activation, can constrict and dam the chan- the Quaternary deposits turned out to be k = 1.01. Thus, prac-
nel of the Mamisondon River. Especially this danger can be tically the same results have been obtained like in analyzing by
brought about by a strong earthquake shock effect. Consider- the limit equilibrium method the stability of the Dallagkaus
ing the volume of this slide body, the height of impoundment slide. These analyses confirm that excessive wetting of the
may be rather big and its potential breach may have catas- soil from intensive precipitation and snow melting may be the
trophic consequences for the dam and the downstream parts main triggering mechanisms of local slide processes. Besides,
of the Ardon valley. To preclude development of such a dan- the performed analyses permitted delineating of the least sta-
ger, the project provides for bypass spillway tunnel through ble zones of the mass where large movements were displayed
the opposite bank of the Mamisondon River which could be at low safety factors, hence stability.
capable of passing the river flow in case of the river channel The method of limit equilibrium was used to analyze the sta-
damming. bility of the sliding mass due to increase in its moisture content
(from precipitation and snow melt water) and due to seismic
shocks with intensity 7, 8 and 9, which has allowed the deci-
2 ANALYSIS OF STRESS-STRAIN STATE AND sion to be made on impounding of the reservoir with a fixed
STABILITY OF SLIDE BODY water stage. It has been established that in the water-saturated
state, the slide mass passes into the condition of incipient slip
A number of the following problems should have first been and any seismic shock effect will trigger its movement. At the
resolved to ascertain stability of the slide with due account for same time it should be stressed that the above movement will
the results of assessment of the slope stress-strain state and the not be of the avalanche nature as loose material of the slide
sliding mass contour: has been creeping (at the rate of about 50 mm/year) for a long
time and the displacements of the sliding mass most likely
1. First, to finalize the shear strength properties of the sliding
will be of the plastic nature.
mass understanding that they can not be defined exper-
To protect the landslide from excessive saturation with rain
imentally in the laboratory, considering the presence of
and snow melt water running down the slope, it would be
large size lumps in the slide mass,
practical to arrange gutters at upper elevations which would
2. Second, to define the cause of though-insignificant but on-
permit the rain and snow melt water running down the slope,
going movements of the sliding body,
to be channeled past the land slide.
3. Third, to define the conditions on implementation of which,
On displacement of the lower portion of the landslide
displacement of the sliding body will not present danger
for 10–15 m, the slide body becomes more stable and its
for functioning of the dam.
displacements are expected to cease.
The numerical model of the slope was used to analyze the In the natural moisture state, the loss of stability by the
stress-strain state and to determine the area of potential rock sliding mass will be triggered only by seismic shocks with
704

3 CONCLUSION
1. With Zaramag reservoir level at El. 1690.6 m and the natu-

ral moisture condition of the soil, the stability factor of
the Grand Dallagkaus slide makes k > 1.2. In this case
displacements of the slide may occur on saturation of the
slide mass with water from excessive precipitation or snow
melting.
2. In the water saturated condition, the slide mass will pass
in the limit equilibrium state and any seismic effect will
trigger its movement.
3. To protect the slide against excessive saturation with rain
and snow melt water running down the slope, it will be
practical to arrange rain water gutters at the upper levels
which permit the rain and snow melt water running down
Figure 1. Distribution of actuating (T) and resisting (R) forces along
the slope to be channeled past the slide.
the landslide body. 4. On 10–15 m downward displacement of the lower portion
of the slide body it becomes more stable and its movements
is to cease.
intensity 9 and more which may lead to its sudden downward 5. In case of an earthquake occurrence in the natural mois-
movement and damming of the Mamisondon River. ture conditions, the slide mass will lose its stability only at
It is of interest to analyze relation between the actuating and seismic shock with intensity 9 and higher which may lead
resisting forces lengthwise the slide (Fig. 1). to its sudden movement and damming of the Mamisondon
Examination of the diagram shows that shearing forces River channel. On filling the reservoir to NWL = 1730 m,
exceed the restraining ones in the top third portion of the slide. the slide may lose its stability at the seismic shock with
Therefore, reducing the risk of water saturation will decrease intensity 8.
the hazard of intensification of the slide mass movement. 6. To ensure safety in case of all potential hazard devel-
Concurrently the stability factor of the slide was being opments, behavior of the slides should be closely and
determined by the limit equilibrium method for the following continuously monitored: a geodetic monitoring service
design cases: must be provided with concurrent recording of river water
– natural moisture condition of mass; stages and precipitation.
– water saturated condition of mass;
– 10–15 m displacement of the slide body in the water
saturated condition;
– seismic shock with intensity 7, 8 and 9.
705

Slope stability analysis using a hybrid Finite-Discrete Element method code (FEMDEM)
G. Grasselli, A. Lisjak, O.K. Mahabadi & B.S.A. Tatone

Geomechanics Research Group, Lassonde Institute, Department of Civil Engineering, University of Toronto, ON, Canada
Keywords: numerical modelling, rock failure, rock slopes and foundations, stability analysis, rock joints, rock mass
1 INTRODUCTION
Traditionally, the stability of rock slopes is assessed using limit

equilibrium methods. Often these methods are limited to the
study of planar or wedge instabilities and other complex failure
kinematics, including toppling, are generally not addressed.
It follows that the failure mechanisms are often overlooked in
the assessment of the hazard of instability and, subsequently,
the risk to down-slope structures. Furthermore, quantitative
prediction of rock fall and runout trajectories is of extreme
importance in hazard assessment of natural and engineered
slopes and in the design of protective measures. To the authors’
knowledge, no commercial code is able to account for multiple
modes of instability and the subsequent fragmentation, runout,
and trajectories of failed material without assuming the mode
or volume of failure a priori. In the current study, the ability
Figure 1. Calibrated trajectory in the initial part of the motion
of FEMDEM to model the onset of failure and the subsequent (Lisjak et al. 2010).
movement of the unstable volume is discussed.
3 NUMERICAL EXAMPLES
2 THE 2D COMBINED FINITE-DISCRETE ELEMENT 3.1 Rock fall
METHOD (FEMDEM)
The capability of FEMDEM to quantitatively simulate rock
FEMDEM is a numerical tool pioneered by Munjiza et al. falls was verified by modelling a real scale in-situ test carried
(1995) for the dynamic simulation of multiple deformable and out by Giacomini et al. (2010) in eastern Australia.
fracturable bodies. Within the framework of FEMDEM, dis- As described by Lisjak et al. (2010), the block was mod-
crete element method (DEM) principles are used to model elled as an elastic body having Young’s modulus and Poisson’s
interaction between different solids, whose deformation is ratio equal to 40 GPa and 0.3, respectively. The block-slope
analyzed by discrete element analysis (FEM). A unique interface was characterized by a contact stiffness of 4 MPa
feature of such a numerical tool is its capacity to model and a friction angle of 40 deg. The contact parameter B was
the transition from continuous to discontinuous behaviour by calibrated by comparing the numerical travel path with the
explicitly considering fracture and fragmentation processes. experimental trajectory in the initial part of the block motion,
Since an explicit time-marching scheme is used to integrate determined through digital video cameras. In general, the
Newton’s equations of motion, fully dynamic simulations can higher the B value, the higher the energy dissipation intro-
be performed. A number of different dissipative mechanisms duced is and the lower the bounce heights are. As can be
(e.g., friction, contact damping and fracturing) are imple- observed in Figure 1 a very good agreement between exper-
mented into the code while no artificial numerical damping imental and numerical travel path was obtained for B equal
or other numerical parameters, often needed by other DEM to 2.
codes, are used. For further details the reader should refer to Since FEMDEM can account for internal stress distribu-
Mahabadi et al. (2011). tion, interaction between discrete bodies, and intact material
The capability of FEMDEM to quantitatively simulate fragmentation, it has the potential to largely overcome the
rock slope instabilities has been tested and validated against drawbacks associated with limit equilibrium analyses and sim-
experimental results obtained during controlled laboratory ulate the entire dynamic process. Four idealized examples of
and field tests. rock slope instability were chosen to better illustrate the use
706

Figure 3. Numerical modeling of the progressive failure process
for a rock slope. Results of a rock slope failure and consequent
fragmentation simulated using FEMDEM (Lisjak, unpubl.).
Figure 2. Simulated failure sequence for a homogeneous cliff model
(Mahabadi et al. 2011). fractures and fragmentation processes have a major role in the
rock-mass mechanical instability, are extremely costly (sev-
eral tens of millions in damages), create a high risk for human
of FEMDEM to model statically and dynamically triggered life, and cannot be fully described using conventional con-
slope failures. tinuum formulations as currently done by most practitioners.
Analysis of a structurally dominated slope using FEMDEM
3.2 Cliff recession modelling shows the initial development of a failure plane and sub-
vertical tension cracks, followed by the formation of a shear
Analysis of cliff stability for setback assessment has always zone along the main failure plane (Figure 3).
represented a challenging problem in rock mechanics. Main
difficulties have been associated with the inability of both
conventional slope stability analysis methods (e.g., limit 4 SUMMARY AND CONCLUSIONS
equilibrium methods) and continuum numerical models to
replicate the observed complex failure mechanism charac- In the present study, a new modelling approach, based on the
terized by a combination of intact material breakage, sliding combined finite-discrete element method (FEMDEM), has
along discontinuities, and rigid body instability. A 20 m high been used to simulate the behaviour of rock slopes subjected
cliff configuration presented by Carter et al. (2009) was cho- to static and seismic loading (the reader should refer to the
sen for this example. A classic wave cut cliff notch due to full paper for the dynamic simulations). The approach has
the marine erosion was created by excavating the base of the been tested against experimental results and then successfully
cliff face. Undercut depth was set to 9 m from the cliff face. extended to large-scale cases. Effectiveness of the approach
Rock mass properties from the case study published by Carter was presented using models that study how rock structure and
et al. (2009) were used. The dissipative contact algorithm with surface topography influence the instability phenomena.
a power damping coefficient B equal to 3 was applied in an Future development, together with the introduction of a
attempt to realistically simulate energy dissipation processes Discrete Fracture Network and hydro-mechanical coupling,
occurring during rock fallout. will expand the field of application of FEMDEM to a broader
As illustrated in Figure 2, failure initiates with a surface range of problems including open pits and gravity dams.
tension crack developing at approximately 18 m from the cliff Finally, the algorithms will be extended to a three dimensional
face. Consequently, a major fracture plane propagates towards version of the code which is currently under development.
the undercut tip, triggering a large scale toppling failure of the
cliff platform. Also, caving of material from the undercut roof
can be observed. Such a complex failure mode is in excellent ACKNOWLEDGEMENTS
agreement with field observations as reported by Bird et al.
(1979). The ability of the code to dissipate energy during rock This work has been supported by the Natural Science and
impact allows for a better match to field observations in terms Engineering Research Council of Canada in the form of
of fragmentation and run-out distances of failed material. Discovery Grant No. 341275 and RTI Grant No. 345516
held by G. Grasselli, an Alexander Graham Bell Canada
3.3 Rock wall failure Graduate Scholarship held by B.S.A. Tatone, an Ontario Grad-
uate Scholarship in Science and Technology held by O. K.
Phenomena such as the rock fall that occurred in July 2008 Mahabadi, and an Ontario Graduate Scholarship held by
on the Sea-to-Sky Highway in BC, where the presence of A. Lisjak.
707

Evaluation of the stability of a bullnose within Coedmore Quarry, Durban, South Africa
A.J. Greet, K. Naidoo & E.D.C. Hingston

School of Geological Sciences, University of KwaZulu-Natal, Durban, South Africa
Keywords: rock joints, rock mass classification, rock slopes and foundations, stability analysis
An assessment of the stability of a bullnose within Coedmore strength allowed for the back calculation of Joint Roughness
Quarry was conducted. The stability analysis was based on Coefficient (JRC). The rock mass classification concluded
detailed scanline surveys and rock strength testing. Kinematic that the slopes were stable. A slope stability analysis was
analysis of the discontinuity data was carried out and it can conducted using the deterministic approach, resulting in a
be concluded that a wedge failure was the most likely failure factor of safety >1. Analysis based on the worst case sce-
mechanism associated with the orientation of the discontinu- nario also rendered a stable slope but with a lower value of F.
ities. The shear strength of the discontinuities was analysed It must be noted that, in reality, the bullnose has been stable
and the estimated maximum normal stress acting beneath the for several decades apart from small localized failures, which
slope was determined. The empirical determination of shear is consistent with field observations.
708

Centrifuge model test on earthquake-induced failure behaviour of

slope in discontinuous rock mass
M. Ishimaru & T. Kawai

Civil Engineering Research Laboratory, Central Research Institute of Electric Power Industry, Chiba, Japan
Keywords: rock slopes and foundations, physical modelling, rock failure, dynamics, rock joints, stability analysis
The seismic stability of rock slopes has usually been evaluated As a result, the slope model collapsed when it was excited
analytically in terms of the sliding safety factors based on by a sine wave of 3.5 m/s2 , which was converted to a real field
limit equilibrium methods. However, because there are very scale. Figure 2a shows an aspect of the collapse of the slope
few model tests that focus on the dynamic failure behaviour model immediately after the collapse, and Figure 2b shows the
of rock slopes, quantitative studies on the applicability of the shape of the sliding surface, determined on removing the col-
seismic stability evaluation method based on these model tests lapsed block. From these figures, the artificial discontinuities
have also been few. considerably affected the collapse, and the type of collapse
We conducted a centrifuge model test on earthquake- was plane failure.
induced rock slope failure. Figure 1 shows the structure of a dip The test results were compared with safety factors related to
slope model for the centrifuge model test. The dip slope model the sliding surface, obtained by the limit equilibrium method;
was constructed from cement, sand, iron powder, and water, the stresses on the sliding surface were estimated by equiv-
and discontinuities were imitated by inserting Teflon sheets. alent linear analysis. Figure 3 shows the minimum sliding
In the test, the centrifugal acceleration was 30G, and the accel- safety factors at each excitation step. The trends of the com-
eration amplitudes of the input sine waves were increased puted safety factors showed that the threshold amplitude of the
gradually at every step. failure was lower than the actual amplitude that induced the
failure, implying that the seismic stability evaluation method
based on the safety factors was conservative.
Figure 2. Aspect of the collapse of the slope model.
Figure 1. Structure of the dip slope model for the centrifuge

model test. Figure 3. Minimum sliding safety factors at each excitation step.
709

Evaluation of slope hazards stability and failure probability in a

mountainous area along a highway
N. Jia, Y. Mitani, I. Djamaluddin & H. Ikemi

Department of Civil and Structural Engineering, Graduate School of Engineering, Kyushu University, Japan
Keywords: rock slopes and foundations, risks and hazards, case studies, back analysis
1 EXTENTED ABSTRACT
Landslides along highway routes always interrupt traffic and

endanger human life. In mountainous area, for the complexi-
ties in geology, vegetation cover and hydrological conditions,
it’s difficult to evaluate the stability of landslides along the
highway. It has been recognized that landslides will occur in
the same geological, geomorphologic conditions as in the past.
Based on the back calculation of mechanical parameters of a
past landslide site along a highway, this paper evaluates land-
slide stability and recurrence probability in this area to prevent
similar hazards from happening again. So, a GIS-based 3D
slope stability method is used to calculate the mechanical
parameters. Then, to locate possible landslides, the study area Figure 1. Safety factors of slope units.
is divided into slope units. Through Monte-Carlo simulation
and 3D limit equilibrium models, the possible landslides can
be located. corresponding safety factor. After calculating 100 times, the
For the mechanical parameters of sedimentary layer, after minimum safety factor is considered as the stability of this
sliding, the slip body is assumed to be in an equilibrium state, slope unit. After all the slope units undergo this process, the
which means that the safety factor equals 1. Three GIS-based distribution map of safety factor can be evaluated (Figure 1).
3D limit equilibrium models (Revised Hovland Model, 3D Safety factor alone is not enough to decide the stability of
Bishop Model and 3D Janbu Model) have been used to sim- slope unit. Failure probability analysis should be conducted as
ulate several combinations of mechanical parameters. After well. Assuming variables c, φ and the 3D safety factor to be in
comparison, c = 5 kN/m2 and φ = 26◦ are considered to be the normal distribution, the 3D probability of the landslide is
suitable values for cohesion and internal friction angle. For calculated by an approximate method which uses a probability
weathered granite layer, because the slip surface didn’t pass in the range of ±3σ (σ is the standard deviation), the 99.75%
this layer, a process for searching potential slip surface is need. of precision in this range is enough for the landslide hazard
In this study, the search is performed by Monte-Carlo simula- assessment. To each slope unit, the safety factor is calculated
tion. The initial slip surface is assumed to be the lower part of 100 times. The mean value µ and standard deviation σ can be
an ellipsoid and changes according to multiple layer strengths calculated based on 100 safety factors. The failure probability
and the conditions of the discontinuous surface. According to is the integral of normal distribution function when safety
the weathering degree and related specifications of Japan, the factor is less than 1.
internal friction angle of substratum layer is selected as 35◦ . Combining safety factor distribution and probability dis-
The cohesion is back calculated. The result shows that if the tribution, it can be seen that, for the slope units in part A, B
cohesion is 8.9 kN/m2 , the sliding mass is in a stable state. and C, both the safety factors and failure probabilities are low.
For the large scale landslide stability analysis, a key prob- These parts are not stable and it is very likely that the slope
lem is how to extract mapping unit as study range. Slope failure occur. But for part A and B, because the direction of
unit, that is, the portion of the land surface delimited by streams is not toward the highway, the landslides in these two
watershed divides and channels, has similar topographic and parts have little influence to the highway. In part D, because
geological characteristics. For each slope unit, stability evalua- most of unstable slope units are located within the valley, if
tion includes using Monte-Carlo simulation to locate possible landslides occur and slide into the valley, a debris flow is easy
slip surface and applying Revhovland model to calculating to be formed and.
710

Seismic instability assessment of rock slopes in a large area based on planar sliding mode
J.-C. Jiang & T. Yamagami

Q. Yang
The State Key Laboratory of Coastal and Offshore, Dalian University of Technology, Dalian, China
Keywords: rock slopes and foundations, risks and hazards, stability analysis
A GIS-based approach for regional assessment of seismic in the area is shown Fig. 2. An artificial acceleration time
instability of rock slopes is developed. The rock slope land- history is used for displacement calculations. The distribution
slides induced by an earthquake is assumed to occur in planar of the permanent displacements obtained using the proposed
failure mode. A rigorous Newmark’s analysis is integrated method is also shown in Fig. 2. It is of interest to note that
with a GIS to predict seismically induced deformations of rock the seismic displacements only take place in slopes in which
slopes. The proposed method includes the following steps: the slope inclination is close to the dip of the geological dis-
continuity planes. No displacements are distributed in slopes
1) A Digital Elevation Model (DEM), a digital geological map
whose dip is opposite to the inclination of the slope surfaces.
and possible ground motion (i.e. acceleration-time history)
are first prepared for a specified region.
2) Shear strengths of rock slopes are assigned to each grid cell
in the DEM coverage by overlaying the DEM and digital
geological maps.
3) For a given gird cell, 3 × 3 cells consisting of the cell under
consideration and eight adjacent cells around it are consid-
ered, as shown in Fig. 1(a), so that four vertical sections of
the slope surface, A-A , B- B , C-C and D-D are obtained.
As the slope profile for the cell, a section, which is the clos-
est to the direction perpendicular to the strike of geological
discontinuities, is selected from the four vertical sections,
as illustrated in Fig. 1(b).
4) Judge if plane failure will occur by considering the strike
and dip of the geological discontinuities. If the conditions
for plane failure are satisfied, then go to the next step.
5) The critical acceleration is computed for the selected slope
section and then Newmark seismic displacement is deter-
mined based on the given ground motion using a double
integration procedure.
6) The above described procedure is in turn repeated for
each grid cell to obtain the distribution of permanent Figure 1b. Generation of slope section for analysis.
displacements over the whole region.
The proposed method is applied to a mountain area in
Tokushima prefecture, Japan. A 10 meter DEM coverage
is built from 1:2500 contour line map. Homogeneous soil
with c = 19.6 kPa, φ = 35◦ and γ = 17.64 kN/m3 and dry con-
ditions are assumed for the whole study area. In addition, the
strike and dip of the main geotechnical discontinuity planes
Figure 1a. An area defined by 3 × 3 cells around the grid cell under
consideration. Figure 2. Distribution of the calculated displacements.
711

Deterministic stability analyses of a rehabilitated pit wall slope at

Coedmore Quarry, Durban
K. Naidoo, B.R. Jones & E.D.C. Hingston

School of Geological Sciences, University of KwaZulu-Natal, Durban, South Africa
Keywords: blasting, site characterization, stability analysis, deterministic approach, kinematic analysis
Coedmore Quarry’s on-going rehabilitation programme A pseudostatic analysis was conducted to account for the
involved the tipping of waste rock over final highwall’s. impact of blast induced acceleration. Results indicated that at
However, this was stopped as it was a concern that it might mean shear strength values with an angle of internal friction
prove to be a geotechnical hazard. The stability analysis was of 37◦ , the slope is stable. However the presence of a ground-
based on kinematic observation that failure would occur as water table, and occasional peak accelerations, of greater than
a translational slip of cohesionless waste rock material. This 0.060 g, reached during blasting could have an adverse impact
type of failure lends itself to the infinite slope stability model. on the stability of the slope.
712

Developing a slope stability curve of open pit coal mine by

using dimensional analysis method
S. Kramadibrata, S. Saptono, R.K. Wattimena, G.M. Simangunsong, & B. Sulistianto

Department of Mining Engineering, Institut Teknologi Bandung, Bandung, Indonesia
Keywords: stability analysis, case studies, coal mines
Slope stability is very critical in running open pit mining oper-

ation in the Adaro coal mine, east Kalimantan, Indonesia.
The reason being, very steep seam, coal bearing strata is dom-
inated by weak and friable to medium strong sandstone of
young formation, high rainfall, and high production plan. Hav-
ing learned the local environmental condition, it is apparent
that the most influence factor to the potential slope failure
is the strength deterioration of the coal bearing strata. Thus,
the geotechnical engineers who have been facing difficulty
of taking account the strength deterioration factor into the
calculation would then need more time to assess the slope
stability.
The potential slope failure is both controlled by the dimen-
sion and geometry of the slope, orientation of the discontinuity
and its shear strength. It is therefore appropriate to conduct a
geotechnical research which can link the rock mass character-
istics with the shear strength of discontinuity plane. Prior to
the rock mass characterisation, determining the most poten-
tial slope failure areas was done carefully, and it was decided Figure 1. The proposed slope stability curve at different slope
that investigation covering of not less than 20 cross sections angles.
stretches along 14 km long with the maximum height of slope
section was 190 m. Subsequently, geological structure map-
ping, determination of physical and mechanical properties of In order to develop the slope stability curve dimensional func-
intact rock, and hydrology evaluation in the investigation area tions associated with slope stability assessment were derived.
were carried out. Many believe that scale effect on cohesion Having done a series of mathematical exercise a propose slope
and internal friction is existence. Hence, different sizes of stability curve was obtained, as depicted in Figure 1.
rock sample were used for direct shear tests. Considering the It is however important to note that at this point in time the
strength deterioration, determination of intact rock strength validation process of this curve has not yet been done. It is
under various water contents was also observed. expected that this curve could enable geotechnical engineers
The rock mass characterisation revealed that the RMR of to assess slope stability in a better way and risk factor could
the rock mass investigated ranges from Class II to Class V. be set at minimum value.
713

Rock slope risk analysis based on non-linear failure criterion
A.J. Li
School of Engineering, Deakin University, Geelong, VIC, Australia
M.J. Cassidy
ARC CoE in Geotechnical Science and Engineering, The University of Western Australia, Perth, WA, Australia
A.V. Lyamin
ARC CoE in Geotechnical Science and Engineering, The University of Newcastle, Newcastle, NSW, Australia
Keywords: reliability, probability, factor of safety, limit analysis
1 INTRODUCTION
The probability of failure of a rock slope is generally estimated

by using the Limit Equilibrium Method (LEM) in conjunction
with a reliability analysis. Although the LEM is relatively sim-
ple and time efficient, recent studies have indicated that using
the LEM may overestimate the factor of safety by 21%, when
based on a non-linear failure criterion. Fortunately, the solu-
tions presented by Li et al. (2008, 2009) can provide more
accurate evaluations for rock slope stability as the numerical
upper and lower bound limit analysis methods (2002a, 2002b,
2005) were employed. The advantages of these methods are
used in this study to assess the rock slope probability of failure.
The motivation of this st5udy is that with more accurate meth-
ods to evaluate the factor of safety, more economic designs can
be performed.
In this study, the rock slope atability is estimated based on
the Hoek-Brown failure criterion (2002). A non-dimensional
stability number proposed by Li et al. (2008) is employed to
evaluate rock slope stability. It is defined in Equation 1. Figure 1. Distribution of safety factors based on different distur-
bance factors (D).
Table 1. Mean safety factor and probability of failure based on

different disturbance factors (D).
where N is the stability number, γ is the unit weight of the Disturbance factor Mean (F) Pf (%)
rock mass, and H and F are the height and the safety factor
of the slope respectively. D = 0.0 4.880 0.09
From the case study, the safety factor distribution and proba- D = 0.7 0.661 86.83
bility of failure (Pf ) obtained for different disturbance factors D = 1.0 0.126 100
(D) are shown in Figure 1 and Table 1. It can be observed
in Figure 1 that the distribution of the safety factor differs
significantly for the limit analysis solutions with different of safety was highly sensitive to the degree of disturbance D.
magnitudes of D. As expected, the Pf increases when the rock Therefore, the judgment of engineers in adjusting for rock
mass disturbance increases. mass disturbance is critical and required to perform more
This study showed that the rock slope risk can be estimated accurate and appropriate slope stability designs before and
by using probabilistic analysis in conjunction with numerical during operation. If there is not enough relevant information
upper and lower limit analysis solutions. However, the factor or data, conservative designs are recommended.
714

Control and evaluation for residual subgrade settlement of high speed railway
X.H. Li, Z. Chen, X.Y. Zhao, H.L. Sun, J.H. Guo & T. Zhao
China Railway SIYUAN Survey and Design Group CO., LTD., Wuhan, Hubei, China
Keywords: case studies, monitoring, rock failure, rock mass
At present, the total length of high-speed railway lines in oper- with bulk material pile (c such as rushed stones pile, sand
ation achieves about 7500 km in China, and about 8000 km pile, etc.) composite foundation, semi-rigid and rigid pile-
in Germany, Japanese, France, and other countries. The total net structure composite foundation;
length of high-speed railways under construction or to be con- • For the main line of high-speed railway with ballastless
structed all over the world is up to above 20,000 km. Subgrade track, the semi-rigid (such as CFG piles, etc.) composite
is an important part of high-speed railway, it’s therefore cru- foundation should be adopted, complemented with flexible
cial to control the subgrade settlement strictly so as to ensure pile (such as jet grouting pile, etc.), rigid pile-net structure
the long-term high-speed, smooth, comfortable and safe oper- composite foundation and sheet pile structure;
ation of railway. Base on a large number of research and tests • For stations, the combined treatment programs of rigid and
both in laboratory and field, China has formed a set of sophis- flexible piles composite foundation should be applied;
ticated theory, technology and standard for high-speed railway • For the transition of bridge-subgrade, culvert-subgrade,
subgrade foundation reinforcement and settlement control. and embankment-cutting and the uneven ground, combina-
In this paper, the concept of “settlement control” is put for- tion of different foundation reinforcement measures, pile
ward, and is required to be followed during the design and length & pile spacing transition, and pre-loading transition
construction of subgrade structures for high-speed railway etc. should be adopted, so as to achieve uniform subgrade
engineering; a brief introduction is made about the com- settlement and control the residual settlement.
prehensive survey methods for high-speed railway subgrade,
It’s indicated in the paper that the high-speed railway sub-
including exploratory boring, static Cone Penetration Test
grade settlement monitoring should mainly emphasizes the
(CPT), Vane Shear Test (VST), Stress Shovel Test (SST),
subgrade surface settlement and foundation settlement, with
Standard Penetration Test (SPT), Dynamic Penetration Test
adoption of settlement observation piles for subgrade surface
(DPT), large-scale loading tests, geophysical exploration,
settlement monitoring, settlement plate, single-point settle-
etc.; the concept, component and control conditions of post-
ment gauge & deep-layer settlement gauge for foundation
construction settlement on railway foundation are stated in
settlement monitoring, sectional settlement pipe for differ-
details; and the control standards of residual settlement and
ential settlement monitoring between subgrade in transverse
differential settlement with adjacent bridge, tunnel, and other
direction and transition segment, and inclinometer pipe for
structures for Chinese high speed (250 to 350 km/h) ballast
lateral deformation monitoring of subgrade; and specific
and Ballastless tracks are described.
requirements are specified for the setting, accuracy, obser-
It’s indicated in the paper that the high-speed railway
vation time, and frequency of settlement monitoring section.
subgrade settlement is closely related to the embankment
Acc. to the construction process, a curve showing relation-
filling height, ground soil properties, and the thickness of
ships between sectional settlement and embankment height &
compression layers; and then the calculation methods of sub-
time should be plotted, to carry out comprehensive analysis
grade settlement and post-construction settlement for Chinese
adopting the hyperbolic method, Logarithmic curve method,
high-speed railway is described in detail; base on field test,
settlement velocity method, Hoshino method and correction
the advice is given to the settlement calculation depth of
hyperbolic method, so as to predict the final settlement of
thick Quaternary stratum. In view of embankment height,
subgrade and the residual settlement after construction, for
ground conditions, construction schedule and engineering
which the specific assessment standards are illustrated. In
types, based on the systematic analysis of foundation soil layer
addition, acc. to the sectional measured settlement, predicted
distribution, thickness, depth, physical and mechanical and
overall settlement and post-construction settlement, longi-
deformation parameters, the assessment of post-construction
tudinal settlement curve incl. subgrade, bridge, tunnel and
settlement is made, and thereon the foundation treatment mea-
transition segment should be plotted, to analyze the unifor-
sures should be optimized in order of embankment setting,
mity of settlement along railway in longitudinal direction,
replacement, preloading, and foundation reinforcement.
so as to give comprehensive evaluation as weather the track
Foundation treatment measures should be adopted accord-
laying requirement for high speed railway is satisfied.
ing to project types:
In the course of high-speed railway construction in China,
• For the main line of high-speed railway with ballast track, to control the subgrade post-construction settlement, a series
the flexible pile (such as mixing pile, jet grouting pile, etc.) of theoretical studies and field tests have been carried out
composite foundation should be adopted, complemented on the reinforcement mechanism & effects, stress & strain,
715

foundation deformation of various composite foundation rein- thickness of 1.0∼3.0 m; after that is completely weathered
forcement measures. The settlement control effects of some muddy sandstone layer, thickness of 10.2∼15.3 m. Predicted
representative composite foundation projects of Chinese high- settlement after construction is 2.43∼11.77 mm, meeting the
speed railway are researched and introduced, including deep subgrade post-construction settlement control requirement for
soft soil pilot section within Fuzhou-Wenzhou high-speed 350 km/h ballastless high speed railway. Shijiazhuang-Wuhan
railway, thick soft soil foundation pilot section within Beijing- high-speed railway pilot section the surface layer is loose –
Shanghai high-speed railway, Wuhan comprehensive pilot slightly dense silty fine sand, thickness 3.4∼8.4 m; under it
section of soft foundation with Wuhan-Guangzhou Railway is slightly dense – medium dense silts, thickness 8.2∼21.0 m;
and Pilot section of deep soft foundation reinforcement within the layer under that is medium dense fine sand and silty clay of
Shijiazhuang-Wuhan high-speed railway etc. hard plasticity, thickness over 50 m. Predicted settlement after
Fuzhou-Wenzhou high-speed railway pilot section: the layer filling is 1.91∼6.27 mm, meeting the requirement of subgrade
is silty clay, thickness of 40∼45 m; the post-construction for laying ballastless track.
settlement 8.3∼30.8 mm, meeting the post-construction set- The foundation treatment measures should be optimized in
tlement control requirement for 250 km/h ballasted high view of settlement requirement, geological conditions, con-
speed railway. Wuhan comprehensive pilot section: the sur- struction schedule, application scope and reinforcement effect
face layer is clay of soft – flow plasticity with thickness of of foundation treatment measures, and technical and economic
about 4.0 m; under layer is cohesive soil of hard plasticity, benefits.
716

The role of rock mechanics investigations in Bakhtiary dam and

hydropower plant project
A. Mehinrad & Kh. Binazadeh

Project Management Team, Bakhtiary Joint Venture Consultants (BJVC), Tehran, Iran
A. Gheshmipour, H. Hamzehpour, H. Chehreh & M. Haftani

Geotechnical Engineer, Bakhtiary Joint Venture Consultants (BJVC), Tehran, Iran
Keywords: lab testing, rock mass, rock mass classification, rock properties, rock stress and site characterization.
Bakhtiary Dam & Hydropower Project is located on Bakhtiary from inside the exploratory galleries. The boreholes are sit-
River (the main branch of Dez River), about 70 km North-East uated at or around the arch dam, underground powerhouse
of the city of Andimeshk in South-West of Iran. The main dam complex, other main appurtenant structures and also at the
is a double-curvature concrete arch type with maximum height quarry areas. Geotechnical investigations performed in these
of 325 m and crest length of 509 m at El. 840 masl. It will be boreholes included RQD measurements, lithological survey,
the highest arch dam in the world now at the design stage. joint survey and analysis of permeability tests’ results.
The hydropower plant consists of an underground powerhouse Bakhtiary dam and hydropower plant project is unique for
complex with a total installed capacity of 1’500 MW which is its outstanding characteristics. One of them is the extensive
designed for annual energy output of 3’000 GWH. engineering geological and rock mechanics investigations car-
At present the dam site is only accessible by train through ried out during its design studies. These investigations were
Tehran-Ahwaz railroad. It is nearby a small railway station planned to fulfill the high technical requirements of Bakhtiary
called Tang-e-Panj Station. project. As a part of this extensive program, several laboratory
The updated feasibility studies of the project which was and in-situ rock mechanics tests were planned and executed
performed by Bakhtiary Joint Venture Consultants, BJVC during 2003 to 2010.
(2009a), confirmed that the higher the dam the more eco- A large number of laboratory rock mechanics tests were
nomical scheme can be planned for the project and the main performed on rock samples selected from the drill cores of
limitation is topography. Therefore, the optimum reservoir the exploratory boreholes including index tests, ultrasonic
normal water level was selected at 830 masl and the dam crest wave velocity measurements, unconfined compression tests,
elevation was designed at 840 masl, considering 10 m free triaxial tests and direct shear tests. The results of labora-
board for flood routing. Considering the narrow valley at the tory rock mechanics tests indicated that the rock material
dam location (width/height ratio of about 1.5) and suitable of siliceous limestone of Sarvak formation at Bakhtiary dam
condition of rock foundation, a concrete double curvature arch site is characterized by its very low porosity, very low water
dam was designed for Bakhtiary Project. absorption, relatively high compressive strength (UCSAve.
Bakhtiary dam site and its reservoir are located in the Folded ≥100 MPa) and high modulus of Elasticity (EAve. ≥60 GPa).
Zagross tectono–sedimentary province. The dam site is mostly Meanwhile, the average shear strength parameters along the
placed on siliceous limestone of Sarvak formation (Cretaceous bedding planes and discontinuities with planner-smooth sur-
period) which has been divided by BJVC (2009b) to seven faces (JRC <6) are C = 0–20 kPa and ϕ = 32◦ and for the
units from Sv1 to Sv7, based on the thickness of bedding discontinuities with planner-rough surfaces (JRC >6) the
planes and the state of siliceous combination outcrop. The average shear strength parameters are C = 80 kPa and ϕ = 36◦ .
most important geological structures at the dam site consist of In order to make a realistic estimate on the geomechani-
Griveh and Siah Kouh anticlines (dam axis will be laid almost cal parameters of the existing rock masses at the dam site, an
parallel to the axis of these structures), F1–F3 thrust and F2 extensive in-situ rock mechanics test program was performed
fault, kinkbands, chevron folds and a joint system (including in the exploratory galleries excavated in the dam foundation
bedding and 3 joints set) at the dam site. and underground powerhouse areas. This test program con-
Subsurface exploration at Bakhtiary dam site consisted sisted of deformability tests including Plate Load, Large Flat
of the core drilling of more than 178 boreholes, with the Jack and Dilatometer tests, in-situ stress measurement includ-
total length of about 12’500 m, in which water pressure ing Borehole Slotter and Hydraulic Fracturing tests and Direct
tests were also carried out. In addition, 11 exploratory gal- Shear test for determining the shear strength parameters of the
leries were excavated with the total length of 2’850 m for bedding planes as the most frequent and dominant discontinu-
engineering geological mapping, discontinuity survey and ity at the dam site. Since the Plate Load Tests provided the most
in-situ rock mechanics tests. Among these boreholes, 70 bore- reliable results for estimating the deformability characteris-
holes were drilled from the ground and 108 boreholes drilled tics of the rock masses therefore; a correlation was defined
717

between the overall modulus of deformation of the rock parameter. Both methods are based on the Hoek-Brown fail-
masses and their ratings based on GSI classification system. ure criterion and consider the rock mass as a homogeneous,
Results of Hydraulic Fracturing tests at Bakhtiary dam site isotropic and continuous medium. The basic theory of the first
revealed that the average orientation of the major horizontal method, Wyllie (1999), is the same as the one described by
principal stress at the dam site (N57◦ ±36◦ ENE) fitted well Goodman (1989) which considers a triaxial condition for the
with the general stress regime by which the Zagross mountain loaded rock foundation by taking into account the unconfined
ranges had been created, in the vicinity of the borderline of compressive strength of the adjacent rock masses as the minor
Arabian and Eurasian tectonic plates. principal stress. The second method, proposed by Serrano et al.
Engineering classification of different types of rock masses (2000) uses the theory of plasticity considering the effect of
at Bakhtiary dam site were performed based on the detail scale factor (between the width of foundation and the spacing
information gathered from engineering geological mapping of the major discontinuities) and the effect of stress level on the
and discontinuity survey data on the surface using the well- behavior of highly fractured rock masses in large foundations
known classification systems (RMR89, Q and GSI). Results which result in plastification of such rock mass even in rela-
of the engineering classification systems indicated that the two tively low stress levels. By comparing the average values of the
main groups of rock masses at the dam site could be classified ultimate bearing capacity of different types of rock masses at
as “fair” and “fair to good” rock masses, respectively. Bakhtiary dam site at zero and 15 m depths, calculated based
Correlating the results of several Plate Load Tests per- on the two above methods, it was concluded that the Serrano
formed inside the exploratory galleries with the GSI values et al. method provided about 2.5 times higher values than the
of the tested rock masses, using the Hoek & Diederichs for- Wyllie method. Considering the results of comparative study
mula (2006) resulted in definition of the best fit curve for the performed by Merifield et al. (2006), it was concluded to use
rock mass disturbance factor of D = 0.05. Using the best fit the more realistic values of the ultimate bearing capacity of
curve, the average modulus of deformation of Sv2, Sv3 and the rock masses calculated in accordance with the Serrano et
Sv4 rock masses (Group A) with GSIAve. = 52 in the dam al. method. However, in order to be conservative enough in
foundation was estimated as Em = 10 GPa and for the Sv5 estimating the allowable bearing capacity of Bakhtiary dam
and Sv6 rock masses (Group B) with GSIAve. = 60 in the rock foundation, a factor of safety of five was used, which was
underground powerhouse area it was as high as Em = 18 GPa. recommended by Goodman (1989).
Since the bearing capacity of the rock masses at Bakhtiary The estimated geomechanical parameters for the “Group
Dam site was one of the key parameters in arch dam design, A” rock masses confirm that a 325 m high arch dam can be
two well-known methods were used for estimation of this constructed at Bakhtiary dam site.
718

Design of spread foundations on rock masses according to Eurocode 7
T. Miranda, F. Martins & N. Araújo

Department of Civil Engineering, University of Minho, Guimarães, Portugal
Keywords: rock slopes and foundations
1 BEARING CAPACITY OF A ROCK FOUNDATION considering additional information related with the uncon-
ACCORDING TO THE EC7 fined compressive strength of the rock and the spacing of the
main joint set.
For the design of spread foundations on rock masses the EC7
provides some considerations. It points out to the necessity of 1.1 Exampe 1
taking into account the following features in the design:
This example is concerned with the computation of the allow-
• the deformability and strength of the rock mass and the able bearing capacity of a 2 m width square footing built in
permissible settlement of the supported structure; a jointed and weakly cemented sandstone. The weight den-
• the presence of any weak layers, for example solution sity (γ) of the rock was set to 22.5 kN/m3 and the spacing of
features or fault zones, beneath the foundations; the main joint set (s) of 0.2 mm. The following parameters
• the presence of bedding joints and other discontinuities and were also considered: Geological Strength Index (GSI) = 20;
their characteristics (for example filling, continuity, width Uniaxial compressive strength (σ c ) = 10 MPa and SF = 3. The
and spacing); calculation was carried out for the case of the footing built at
• the state of weathering, decomposition and fracturing of the
rock;
• disturbance of the natural state of the rock caused by
construction activities, such as, for example, underground
works or slope excavation, being near the concrete founda-
tion.
It states also that for intact igneous and gneissic rocks,
limestones and sandstones, the presumed bearing pressure is
limited by the compressive strength of the concrete founda-
tion. Finally, a remark for the necessity of assessment of the
foundation settlement by means of comparable experience.
The method presented in the annex G of the EC7 can be
used for the calculation of the bearing resistance of spread
foundations in weak and broken rocks with tight joints. For
weak and broken rocks with open or infilled joints, reduced
values of the bearing capacity should be used.
The application of the method starts with the choice of the
group of the rock considering the classification provided in
Table 1. After the definition of the group, the bearing resis-
tance of the foundation can be estimated using Figure 1,
Table 1. Grouping of weak and broken rocks.
Group Type of rock
1 Pure limestones and dolomites Carbonate sandstones of

low porosity Igneous
2 Oolitic and marly limestones Well cemented sanstones
Indurated carbonate mudstones Metamorphic rocks,
including slates and schist (flat cleavage/foliation)
3 Very marly limestones
Poorly cemented sandstones Slates and schists
(steep cleavage/foliation) Figure 1. Presumed bearing capacity for square foundations with
4 Uncemented mudstones and shales settlements not exceeding 0.5% of the foundation width for the four
groups.
719

Table 2. Results for example 1.
qa (MPa) qa (MPa)
Method depth = 0 m depth = 2 m
Wyllie (1992) 0.35 0.71

Serrano et al. (2000) 1.01 1.46
EC7 (CEN, 2004) 1.70 1.70
the surface and for a depth of 2 m using the three methods Figure 2. Variation of qa with σ c .
presented before.
The results of this calculation are presented in Table 2:
The results show considerable dispersion between the 2 CONCLUSIONS
results of the different methods. The method of Wyllie (1992)
provides the lowest values and the one suggested by the EC7 For the design of spread foundations in rock masses the EC7
the highest. One important issue is the significant increase of provide some aspects that should be taken into account and
qa with depth for two of the methods which were an expected proposes a method for the calculation of the bearing capacity.
fact due to the increasing vertical stress. For a 2 m depth of The comparison of three different methods based on two cases
the foundation base the method of Wyllie (1992) point out to allowed concluding that considerable different results can be
a double value of qa and the one by Serrano et al. (2000) to an found even for simple examples. The method from Wyllie
increase of almost 50%. The EC7 method (CEN, 2004) does (1992) seems to provide a lower boundary for the values of
not take into account with the increase of the bearing capacity the allowable bearing capacity.
due to this fact. The method proposed by the EC7 is rather simplistic and
does not take into account important aspects that influence the
bearing capacity like the depth of the foundation, its shape,
1.2 Example 2 the eccentricity of the loads, water, etc.
A pillar transmits to a square footing a load of 4 MN. The In spite of the suggested EC7 method has been based on
geotechnical survey identified a layer of soil up to a depth conservative assumptions, it gave the less conservative results
of about 3 m (γ = 20 kN/m3) and below a jointed weathered in example 1. However, the results obtained with this method
granite where the foundation will rest. The design parameters in example 2 are similar to the lower values obtained using
for this granite were the following: GSI = 40; γ = 24 kN/m3; Wyllie method.
σc = [20–50] MPa; s = 0.2 m and SF = 3. The objective is In conclusion the EC7 method for the design of spread foun-
to compute the allowable bearing capacity using the three dations should be used with caution and to provide only an
methods. Given the uncertainty related to the value of σc, approximate idea of the bearing capacity. The results should
it was decided to carry out a parametric study to analyse the be validated using other analytical procedures or even using
importance of this parameter. numerical methods.
The main results of this calculation can be observed in
Figure 2 where the relation between qa and σ c is presented.
In this example the influence of the variation of σ c was REFERENCES
investigated. For all the tested methods, qa increases almost
linearly with σ c with a small exception for the Wyllie (1992) CEN 2004. Eurocode 7 Geotechnical Design – Part 1: General Rules.
Hoek, E. and Brown, E. 1980. Underground excavations in rock. Inst.
method where this increase is slightly non-linear. This method
of Mining and Metallurgy, London, 627p.
and the one from the EC7 provide similar results in this case Wyllie, D. 1992. Foundations on rock. E & FN Spon.
pointing out for a value of qa approximately equal to 25% of Serrano, A., Olalla, C. and González, J. 2000. Ultimate bearing
σ c . On the other hand, the method of Serrano et al. (2000) capacity of rock masses based on the modified Hoek-Brown cri-
provides values of qa which are about 74% of σ c , almost three terion. Int. Journal of Rock Mechanics and Mining Sciences 37:
times higher than the previous ones. 1013–1018.
720

Unloading scheme to control sliding mass at Angouran open pit mine, Iran
P. Moarefvand
Department of Mining and Metallurgy, Amirkabir University of Technology, Iran
M. Ahmadi
Department of Mining Engineering, Tarbiat Modares University, Tehran, Iran
M. Afifipour
Department of Mining and Metallurgy, Amirkabir University of Technology, Iran
Subject: Rock slope stability.
Keywords: rock slope, site characterization, rock failure, numerical modeling, stability analysis.
Angouran zinc-lead open pit mine, located in the northwest right sides, without significant disturbance and overturning.
of Iran, has experienced a large-scale slope failure in the Finally the crushed zone dropped on lower benches with
order of 12 Mm3 volume. The whole failed mass moved about respect to the slip surface geometry and fragmented. A
100 m horizontally and 65 m vertically. The failure region was three-dimensional discrete element modeling was executed
mainly comprised of a jointed limestone rock mass as a hang- to evaluate the slip surface and predict responses of the slid-
ing wall at the northwestern wall of Angouran pit. The hanging ing mass. The slip surface geometry estimated through the
wall contact toward the thick schist bedrock might be consid- numerical simulation was in close agreement with the site
ered as the triggering zone. The interface between these two investigation results, and proposed the failure scenario. The
formations (hanging wall limestone and footwall schist) was slip surface and the sliding zone geometries implied that
estimated to be the slip surface, owing to the poor geomechan- unloading below this level would lead to lightening the toe
ical properties. The main challenges due to this giant failure of slope and consequently, causing instability in the whole
are subsequent responses of the sliding mass, geometry of volume. The gradient of the slip surface between level 2970
the slip surface, the regions of critical movements, and where and level 2960 approximately reached to zero and the surface
unloading procedures should be performed to stabilize the became almost horizontal. This fact was the main reason for
slope and restart mining operation. To find solutions for the the temporary stability of the sliding zone. The sheared zone
challenges, a comprehensive site investigation and monitor- does not need to be unloaded and disturbance in this zone
ing was conducted to compare the primary and secondary might decrease its strength and stability. Based on the findings
discontinuity patterns in order to identify the failure scenario from site investigation and numerical analysis and according
and determine the governing mechanism. Using the results of to the slip surface geometry and the relative position of the
field investigations and analyzing the primary and secondary mentioned main parts in the failure region, a decision was
discontinuity patterns, the failure area was divided into three made to stabilize the slide by unloading the active part of the
main parts: the sheared zone, the sliding zone, and the crushed complex unstable mass from the levels upper than 3050 to
zone. The sheared zone is the upper section that is restricted restart mining operation safely in the adjacent sections of the
by two lateral faults. These faults that dipped into the failure pit. It is estimated that over 4 Mm3 , i.e. approximately 35 per-
area were considered as two lateral slip surfaces. The slid- cent of the whole volume, should be unloaded to restart mining
ing zone moved and created the sheared zone in both left and operation.
721

Mechanism and dynamics of dip-slope failure revealed by LiDAR data and

Discrete Element Method
Chih-Hsiang Yeh & Ming-Lang Lin

Yu-Chang Chan
Academia Sinica, Institute of Earth Sciences, Taipei
Kuo-Jen Chang
Keywords: dip-slope failure, LiDAR DTM, discrete element method
Recent in Taiwan the dip-slope landslide failure frequently first obtained the basic engineering parameters, c, φ, and τ,
caused huge casualties. Except for rainfall or earthquake by numerical modeling method, 2-D discrete element method,
directly induced slope failure, the man-made interference such in one case study. At same time, we considered groundwater,
as cutting the toe of slope had been one of the crucial factors of retaining wall, and geological factors so that to simulate real
losing slope stability. One representative case of dip-slope fail- field condition. Finally, we are going to look for all of the
ure occurred on freeway locates Keelung river northern basin, potential slopes using LiDAR DTM, and evaluate the possi-
which consist of dip-slope strata. Noteworthy issue around bility of sliding of slope failure in those slopes through above
this area is that there is still a lot of potential slopes may slide case study.
because of influences of artificial building. In this study we
740

Deep foundations on weathered rocks
C.S. Oteo
University of Coruña. Spain
Subject: Analysis techniques and design methods. Others (Pile foundations)
Keywords: weak rock.
1 SUMMARY Olalla (2008) himself has grouped a number of real qp val-

ues (field trials), corresponding to different materials, with
It is very common in large bridge pilings for these to rest RMR’s ranging from 15 to 85 and σc between 0.5 and 90 MPa,
on more or less resistant rocks: gypsum marls and gypsum, having represented the relation qp − σc (Fig. 1). In the same
limestone, granite, basalt, etc. In these cases it is always a prob- figure there is a line that may be a good correlation between
lem to determine what the pile embedment is in this resistant the two variables, though with a large dispersion. This medium
material. line would imply that:
The top of the rock is always altered, with minimum thick-
nesses of 1 m that can sometimes be up to twenty metres. It is
always considered that altered zone should be penetrated for
the piles to be supported below, but it should be understood
that the intention of this idea is not to leave the tip of the piles
in crumple zones and not that it is necessary to reach areas
with a type I degree of weathering according to the Interna-
tional Society of Engineering Geology there are 6 degrees of
weathering. This means that the formula qp = 4, 5 σc for soils is still quite
In our view, this condition may be unnecessary, above all valid. Using (σc )1/2 can be quite useful in harder materials, the
because ground strength can often be very high and the average ratio to apply, (σc )1/2 , turning out to decrease with increasing
acceptable pressure can be well above the allowable structural σc .
pressure of the pile In Fig. 2 we have represented (using the Olalla data) the
It is necessary to determine tip strength, qp, and shaft bear- relationship between qp and the RMR index, obtaining:
ing capacity in the area of pile embedment, τr , which need not – A large dispersion (the rocks are different, as are the
be uniform, since it will reach different values in the altered fixities, etc.)
zone (alteration grades III to VI) than in the fresh rock zone – If the highest values are removed (and some fairly low ones,
(grades I–II). for their corresponding RMR value), you can see why it is
From the theoretical point of view there are a number of not a good idea to exceed qp of around 15–20 MPa.
formulas to estimate both τr and qp , developed for materials – For low and medium rock qualities (corresponding to soils,
of different consistency. Olalla (2008), has grouped several of i.e. degrees of weathering V–VI) the average values of qp
these formulas, whose specific references can be seen in that practically correspond to the expression:
work:
It can be applied within an acceptable degree of correlation

All these formulas correspond to the scheme: qp = Nc σc , – with actual data.
typical of “soils” – in which qp is in MPa, Nc is the load factor As for shaft friction or lateral skin friction, τr , there are also
and σc (in MPa) the compressive strength, in which only the many correlations:
term “cohesion” is considered.
In some cases, the value of Nc has been set according to the
RMR index of Bieniawski.
That is equivalent to establishing an influence of the degree
of quality of the rock or of its degree of alteration. Some of these expressions give enormous values, as we
All these expressions vary linearly with σc . For a quality first have to think what maximum value of τr can be used. For
rock, such as a granite a unconfined compressive strength of example, in limestone, a τr value of 0.45 MPa can be used (i.e.
100 MPa, qp could come to have a value of between 15 MPa 4,5 Kp/cm2 , which seems rather excessive). Other formulas
(reasonable value) and 1100 MPa, the latter being an unattain- give higher values (45 Kp/cm2 , for instance).
able value since, with a safety factor of 3, it would lead From all these expressions and the author’s personal experi-
to a working pressure of over 360 MPa far superior to any ence, we have drawn up Figure 3, which indicates the possible
structural allowable work pressure. shaft and tip strengths in granite materials, data with different
724

Figure 1. Field measurement data, σhp , in relationship with the rock
unconfined compressive strength, σc (Olalla, 2008).
Figure 3. Possible pile lateral and tip bearing capacity in granite

materials.
intubation the pile to ensure the continuity of the concrete. We

have done this in several jobs in gypsum karst. The problem
was solved by grouting the cavities through probes with spe-
cial mortar pumps (under pressure). We have also used this
injection solution when the karst area was under the tip of
large diameter piles, in order to ensure the karst areas were
properly treated.
We have used a similar solution, with drill holes under foot-
ings, in the foundations of several bridges on the motorway
and railway over gypsum. The existing gypsum could well
absorb the surface foundation loads, but the purpose of the
Figure 2. Relationship between, the RMR index and the fields pile drill holes was to check that there was no karst below. To find
tip bearing capacity (Olalla, 2008).
out whether or not this was the case, each hole was grouted
and where significant amounts were admitted, the groutings
weathered degree, based on engineering experience and some were repeated.
actual measurements taken from Figures 1 and 2. This figure At all events, in these hard grounds with cavities, shaft
can serve as a guide for many projects in northwest and central friction must be assessed realistically. In the text we have
Spain and other countries with granite rock formations, which represented an approach we recommend in a specific job
sometimes require excessive fixity, in our view. In this Figure, (piles in hard gypsum under a viaduct in the High Speed
τr has been limited to a maximum of 0.4 MPa Line Madrid-Valencia), where τr is determined according to
A problem that requires attention is the problem of the the speed of advance of the pile drilling. We have applied
piles excavated in chalk gypsum and limestone with possible this same approach (drilling speed) several times to establish
cavities. the pile embeddment in hard rock (as in the case of granite,
In both cases they are hard materials with good shaft and where we have left piles when the speed of advance was about
tip strengths, which can be affected by the presence of caves 15–20 cm/hour). The recommended τr value reached 4 MPa
and dissolution holes. If they are found in the shaft area, one (slow advance of 0.7 m/hour) and fell to 0.1 MPa when the
could use the solution of filling the cavity with mortar or of advance rose to 3 m/h.
725

Failure modes and countermeasures of large-scale rock slope composed of

Phyllite in Korea
Boo-Seong Park, Hyun Cho, Seung-Hun Cha & Dong-In Park

Ssangyong Engineering & Construction Co. Ltd., South Korea

Keywords: case studies, rock slopes, rock failure, site characterisation, stability analysis
Phyllite is a low grade metamorphosed rock. In Korea, phyl- are well developed along the lithologic boundary of phyllite
lite is mainly distributed in the Chungju and Boeun areas, and granite porphyry.
which belong to the Ogcheon Supergroup. Phyllite is mainly In the study site, slope failures occurred in four (4) cut
composed of platy minerals such as fine-grained micas and slopes as shown in the Fig. 2 and Table 1. The heights of the
chlorite, and it has foliation developed in metamorphic pro- cut slopes varied from 30 m to 70 m approximately. Plane and
cess. Phyllite is an anisotropic rock in terms of engineering wedge failures occurred by foliation and joint in the site A.
properties because strength and deformation modulus vary in In the site B, the 3.0 m thick intruding dyke of granite por-
different directions. It shows fissility developed along long- phyry, intruded into phyllite, is highly affected by shearing
persistent foliations, resulting in low shear strength. Also, and weathering, and exists under completely weathered soil
phyllite is easily weathered when exposed to the air since it condition, containing small rock fragments and fault breccia.
is basically formed with clay minerals. Those characteristics, Plane failures arose along foliation of phyllite in the site B. In
such as anisotropy, fissility and easiness for weathering are the site C, a large-scale wedge failure occurred by intersec-
considered as major causes for slope failures, and it is proven tion of joints. In the site D, the 25 ∼ 30 m thick dyke of granite
by the fact that many cases of small to large scale rock slope porphyry, intruded into phyllite, shows well developed fault
failure in phyllite have been reported in Korea, resulting in shear zones along the lithologic boundary, and plane failures
increase of construction cost and delayed construction sched- arose along the lithologic boundary and foliation in phyllite.
ule. Therefore, due considerations are required for design or
construction of rock slopes in phyllite. Even it becomes more
important when we consider that construction cases of national
roads and highways across inland areas in Korea have been
increasing rapidly, and as a consequence, many cut slopes
have been constructed within geologically poor areas in Korea
such as the Ogcheon Supergroup where shows complex geo-
logical structures within metasedimentary rocks. Considering
the fact, this case study aims to introduce failure modes of
rock slopes which consist of phyllite and granite porphyry,
intruding along foliations of phyllite. Figure 2. Location of the study sites.
The study site consists of low grade metamorphosed rock
Table 1. Failure modes.
(phyllite) and granite porphyry which belong to the age-
unknown Hwanggangri Formation of the Ogcheon Super-
group and Cretaceous dyke, respectively.And alluvial deposits
of Quaternary Period are unconformably overlaying them.
Most sites belong to the Hwanggangri Formation of the
Ogcheon System and major component of rocks is dark grey
or black pebble-bearing phyllite. On the slope surface, phyllite
is highly weathered and it has become platy rock fragments
as shown in the Fig. 1. The granite porphyry, which intruded
phyllite, shows differential weathering, and fractured zones
Figure 1. Typical geological condition of the study sites.
726

The major characteristics of the four (4) slopes is the fissility, image processing system were conducted to acquire infor-
which is well developed along long-persistent foliation, result- mation on the geological strata, types of rocks, geological
ing in low shear strength. Since the dominant dip angles and structures, dominant trends of discontinuities, etc. as shown
dip directions of the discontinuities are similar to those of the in the Fig. 3 and Fig. 4. And the results of them were used for
slopes, plane failures along foliation and sheared lithologic the analysis on the causes of the slope failure and preparation
boundary occurred frequently. Also, wedge failures occurred of the countermeasures against slope failures in the long-term
at the intersections between foliation and joint, and foliation view. By comparing and analyzing the results of the geotech-
and fault plane. nical survey, the causes of the slope failures were inferred as
During the original design, sufficient numbers of site sur- the easiness for weathering of the bedrock, well developed fis-
veys hadn’t been performed, causing lack of information for sility, discontinuities, foliation, etc. in phyllite, and the fault
design and resulting in instability of the slopes. Therefore, shear zones around the lithologic boundary between phyllite
additional geological survey such as the surface geological and granite porphyry.
survey, boring survey, electric resistivity survey and borehole For reinforcements of the slopes, as shown in the Table 2,
the soil nails were planned for securing the stability against
Table 2. Reinforcement method plane failures along foliations and wedge failures by intersec-
tions of joints for the site A. The slope gradient of the site
B is reduced from 1:0.8 ∼ 1.0 to 1:1.4 to minimize the slid-
ing force, and the weathered zone of the sheared dyke rock
was reinforced with the soil nails. For stabilizing the slope
and avoiding an infinite slope in the site C, the slope gradi-
ent was revised from 1:0.5 ∼ 1:1.5 in the original design to
1:0.8 ∼ 1:1.2, and the soil nails were installed simultaneously.
For the site D, the decrease of the slope gradient was applied
for the slope stabilization. The slope gradient is decreased
to 1:1.7 by considering the low shear strength of the discon-
tinuities. The PAP (Pre-stressed Anchor and Precast panel)
retaining wall in addition to the decrease of the slope gradient
was planned to enhance the stability of the key blocks as well
as to minimize the quantity of the excavation work. Also, the
PAP retaining wall suggested more uniformity with the wing
wall at the tunnel entrance. The surface of the slope was rein-
forced with the slope protection method since the stability at
the slope surface might be decreased due to the easiness for
weathering of phyllite. Therefore, the grid frame is installed
by using the fiber mold, and the grass seeds were sprayed onto
slope surface.
As discussed above, the slopes composed of phyllite gen-
erally have well developed discontinuities such as foliation,
joint, etc. In those slopes, the dominant trends of disconti-
nuities and the condition of discontinuities (filled with clayey
material) are major causes for slope failures. Therefore, selec-
tion of optimal methods for geotechnical investigation with
sufficient numbers shall be prepared when anisotropy rock
slopes are designed. Based on results of surveys, charac-
teristics of geology, geological structures and engineering
properties of bedrock shall be thoroughly understood and
applied for design and construction. We hope that this case
study would be a useful reference for stability analysis of rock
slopes whose site conditions are similar to those of this study
site such as geological structures and geotechnical properties.
Figure 3. Geotechnical investigation (Site D).
Figure 4. Results of boring survey and electric resistivity survey (Site D).
727

Foundations of gravity dams: Rock mechanics requirements
Manuel Romana
Technical University of Valencia, Spain
Subject: Dam foundations

Keywords: rock slopes and foundations
1 INTRODUCTION Table 1. Approximate values of celerity in terrain classes
The requirements for the foundation of concrete (either CVC C (Km/s) Terrain class
or RCC) dams traditionally have been expressed in qualitative
∼5 Very hard rock (good concrete)
terms: “sound rock”, “fresh rock”, “below the decompressed 5–3.5 Hard rock (concrete)
surface strata” . . . Fraser (2001) describes the “different 3.5–2 Medium/Soft rock
approaches for defining a (dam) foundation objective”: “1 2–1.5 Very soft rock/indurated soil
attain a specific geologic unit; 2 excavate to a grade based on <1.5 Soil
field testing results; 3 attain a specific rock quality; 4 achieve
a surface that meets a construction control test; 5 excavate to
a surface based on the ability of the excavation equipment;
6 excavate to a depth indicated by the design; 7 achieve a Table 2. Geotechnical zoning conditions used in Portugal.
material judged adequate based on visual observation”.Those
approaches are subjective. Fraser ends: “ultimate adequacy of Zone Weathering degree Celerity (Km/s)
the foundation must be confirmed . . . to insure the foundation
will perform” GZ 1 W1–W2 >42.5
The most usual requirement for rock foundation quality of GZ 2 W3 1.2–2.5
GZ 3 W4–W5 <41.2
gravity dams is “good quality, sound, fresh and unweathered
rock”. Sharma (1998): “the entire (foundation) area should
be stripped to firm rock”. Fell et al. (2005): “after normal
treatment . . . the rock foundation must be adequate to support
Table 3. Effect of Ec/Em on gravity dams behaviour
dam loads”.
Actually an increasing number of CVC and RCC dams have Ec/Em* Influence on dam Problems
to be founded in less sound rock, border-line terrains.
The problem of dams’ foundation in “low strength mate- <4 Negligible None
rials” was addressed in a quantitative way, by Rocha (1975, 4–8 Low importance Minor
1976). He had started to work on the problem of rock mass 8–16 Important Some
deformability for double arch dams (Rocha, 1964, 1974). >16 Very important Moderate to big
Sadly he died and his last papers for gravity dams, in
* Ec/Em concrete/foundation deformation modulus
Portuguese, have not been published in English.
Dam stability against horizontal sliding must be checked by
the designer taking account of the distribution of water pore
pressure across the dam foundation and of the mobilized shear Portuguese authors use the concept of “geotechnical zon-
strength of the governing joints. The calculation must be based ing”, with GZ 1 adequate fro dam foundations
in a good knowledge of geology below the dam. There are no According to Marulanda (2003) “foundation levels are
approximate solutions. acceptable when a minimum compressive seismic velocity
(celerity C) between 2.5 and 3.5 Km/s are recorded. Oth-
ers . . . define the rock foundation level on the basis of
2 ISRM SCALE OF WEATHERING GRADES
ripability . . . by D8 dozers”. The generally stated limit for
rippability is terrain celerity of 2 Km/s, (Caterpillar, 2008)
Weathered material must be striped away from the foundation.
which could be accepted as the lower limit for gravity dams’
In most cases the foundation is excavated to grade II rock in the
foundation, with 2.5 Km/s as desirable value
valley central part and to grade II–III rock in abutments. Spill-
ways should be founded on grade I rock. These requirements
could be minored for RCC dams. 4 ROCK MASS DEFORMABILITY MODULUS
3 LONGITUDINAL VELOCITY (CELERITY) Marulanda (2003) says that “a modulus of elasticity ratio
above 1/4 is commonly considered for compatibility between
A parameter frequently quantified to check the rock quality foundation rock and RCC”. Rocha’s work for arch dams was
has been C, celerity of P-waves, measured by geophysical endorsed by Serafim (1988) and Oliveira (1990) Rocha’s
methods (refraction). (1975, 1976) results for gravity dams are in table 3,
728

Zeballos (1992) analyzed 29 big dams with published Table 4. Minimum condition for foundation of gravity dams.
deformability data, and stated: “there are almost no concrete
gravity dams with Ec/Em > 20” which confirm Rocha results. Allowable with
There are two concrete (CVC) gravity big dams in Spain with Minimum condition Allowable special measures
Ec/Em ≥ 20. Puebla de Cazalla dam was designed with slopes ISRM weathering degree II/III III/IV–V
1:0.8, and 1:0.2 as a measure to cope with the deformability Celerity (Km/s) 2, 5 2, 0
problems (Bayan et al., 1993): the dam is in normal operation. RMRB 60/50 50/40
Rules dam was designed as a “normal” arch-gravity dam with
good quality concrete (Nevot el al, 1993); reservoir filling has
been careful, slowly increasing water levels over several years,
for foundation consolidation (Em ∼1 GPa, Soriano, 2010); the Table 5. Minimum terrain Em rock mass deformation modulus
dam has been displaced downstream t several centimetres. (GPa) for foundations of gravity dams
It can be deduced that the limit for terrain allowable min-
imum deformation modulus would be Ec/Em = 1/10 and if Minimum Em
Ec Dam type/concrete
Ec/Em = 1/20 deformability problems may appear and they
(GPa) Allowable With problems quality
would need special measures to cope with. (table 5).
30 3 1.5 CVC dams/ good
20 2 1 RCC dams/ normal
5 GEOMECHANICS CLASSIFICATIONS: RMR, DMR
10 1 0.5 Old dams/very poor
Use of RMR has been extremely frequent in underground
works, scarce in slopes and reduced in dam foundations. Many
authors have referred to the use of RMR as a useful tool for Table 6. Minimum conditions of the rock/soil for the foundation
the description of rock mass foundations: Pircher (1982) “the of hardfill dams
future seems to be in the development of quality index values
e.g. RMR by Bieniawski” and Serafim (1988) “appropriate Allowable with
rock mass classifications can . . . be used to obtain a good Minimum conditions Allowable special measures
estimate of (shear strength and deformability) parameters”, Compressive strength 2 MPa 1.5 MPa
in General Reports for ICOLD Congresses. Celerity (Km/s) 2.0 Km/s 1.5 Km/s
Difficulties of RMR for dam foundations derive from Em 1 GPa 0.5 GPa
several points: consideration of the water pressure is very
doubtful; there are no good rules for the adjusting factor for
joints orientation; properties of rock, rock mass and joints
change with watering changes; there are only guidelines for foundation conditions are less satisfactory . . . it may be more
general stability against horizontal sliding. economic . . . a larger volume of cheaper material…ultimate
Some authors propose a value of RMRB (basic RMR, solution . . . is the hardfill dam”. There are yet too few hardfill
obtained adding the five Bieniawski parameters without the dams built in the world to establish general criteria, but some
adjustment factor for joints orientation) of ∼60 (limit between guidelines (in Table 6) can be proposed from some built dams
Bieniawski classes II and III) as requirement for the concrete in operation: Cindere (Batmaz, 2003), Moncion (Capote et al.
dams foundation in the central part of the valley. This con- 2003), Koudiat-Acerdoune (Frossard 2011).
dition is roughly equivalent to asking for Em ∼ 20 GPa. In
the abutments, minimum RMRB would be 40 (lower limit of
Bieniawski class III) and Em ∼ 5–6 GPa. 7 CONCLUSIONS
Romana (2003a, 2003b, 2011) has developed a new geome-
chanics classification for use in dam foundations: DMR (Dam The minimum requirements are summarized below (but must
Mass Rating), based in the RMR scheme but with different be checked with the full text of the paper)
adjusting values for every problem and every dam class. Its
description falls out of the scope of this paper. ACKNOWLEDGEMENT
6 FOUNDATIONS OF HARDFILL DAMS The author will be very thankful for any comments which
could confirm, discuss, or deny, the limit numbers contained
A new class of gravity dams, built with hardfill, tries to avoid in this paper for the several requirements. The e-mail address
the problem of poor strength foundations. (Londe & Lino, of the author is mromana@stmr.es
1992; ICOLD, 2000). As Dunstan (1995) explained: “if the
729

Use of ground-based synthetic aperture radar to investigate the complex

3-D kinematics of a large open pit slope
J. Severin & E. Eberhardt

Geological Engineering-EOS, University of British Columbia, Vancouver, Canada
L. Leoni
IDS Ingeneria Dei Sistemi, Pisa, Toscana, Italy
S. Fortin
Highland Valley Copper Mine, Logan Lake, British Columbia, Canada
Subject: Site Investigation and Field Observations
Keywords: field measurements, monitoring, mining, rock slopes and foundations
1 INTRODUCTION
This paper reports a novel experiment involving the simultane-

ous deployment of two ground-based interferometric synthetic
aperture radar (InSAR) systems. These were used to collect
continuous, line of sight displacement data in “stereo” of a
large, moving open pit slope bisected by a large fault. The data
and results presented demonstrate that an improved under-
standing of the 3-D kinematics of a large rock slope and the
influence of large scale geological structure can be achieved
using advanced state-of-the-art monitoring techniques.
2 EMERGING SLOPE MONITORING TECHNIQUES

Figure 1. Location of IBIS-M radar units and direction of expected
New developments in slope monitoring include the use of principal movement.
remote sensing technologies like terrestrial radar, which pro-
vide high-resolution, full area spatial coverage as opposed to overlap with Site A at an oblique angle to the direction of slope
relying on geodetic point measurements. The commercially movement. The units were synchronized to run simultaneously
available radar sensor (IBIS-M) used in this study, manufac- under the same conditions for several days collecting data with
tured by IDS Ingegneria Dei Sistemi is based on: i) Stepped the same scan interval (6 minutes).
Frequency Continuous Wave (SF-CW), allowing range direc-
tion resolution, ii) Synthetic Aperture Radar (SAR), allowing
cross range resolution, and iii) differential interferometry, 4 DATA MANAGEMENT
which allows the measurement of displacements by comparing
phase information of the back scattered electromagnetic waves The monitoring data was collected and processed using the
collected at different times. The IBIS-M sensor is capable software IBIS Controller®and IBIS Guardian®to remove
of performing real-time, near-continuous (5 minute interval), the atmospheric artifacts (in both range and cross-range) from
line-of-sight (1-D) monitoring of large areas, day or night and the phase information with an advanced algorithm that is able
in all weather conditions (−50◦ C to 50◦ C). to automatically discern stable points from those that are mov-
ing thus avoiding the need to manually select ground control
points. Independent displacement maps were combined with a
3 FULL SPATIAL DETECTION OF A PIT SLOPE digital elevation model (DEM) of the mine site and converted
into global coordinates with displacement values. Approxi-
The west wall of the Lornex Pit at Highland Valley Copper mately 25,700 points were found to be common between the
(HVC) near Kamloops, British Columbia, Canada, involves two data sets.
a 400 m high slope with an overall angle of 29◦ . The rock is Based on the original location common to each raster
relatively competent but is altered in the vicinity of the fault set (XA1 ,YA1 ,ZA1 = XB1 ,YB1 ,ZB1 ), the corresponding line of
zones. The radar units (Sites A and B) were located near the sight displacement and the location of each radar instrument,
top of the east pit wall (Fig. 1). The distance of the monitored a displacement vector for each direction of movement was
points ranged between approximately 800 and 2500 m. Site A created (XA2 , YA2 , ZA2 and XB2 , YB2 , ZB2 ) using the formula:
was chosen as it was roughly in-line the expected direction of
slope movement. Site B was chosen to provide good spatial
730

Figure 2. Combined displacements from sites A and B.
where xo ,yo ,zo instrument location, and R is the ratio of the

measured displacement and distance from the instrument. For
each of the new vectors created, the equation of the plane per-
pendicular to that vector at the new point was determined.
Actual displacement of each raster cell must exist on both
planes; therefore, the combined displacement point exists on Figure 3. Combined vector map. Lower, enlarged (125 × 125 m)
the line created by the intersection of these planes. In the view of the upper slope with magnified displacements (10X) shows
absence of a third radar unit to allow for true 3-D triangu- direction of movement.
lation, one of the values of the new vector was assumed. In
this case, a weighted Z value based on displacement was used.
within the upper part of the slope which are likely controlled
by separate structures within the rock mass. A distinctly dif-
5 THREE-DIMENSIONAL DISPLACEMENT PATTERN ferent direction of movement exists below the Lornex Fault
suggesting that this region is experiencing a different failure
5.1 Displacement magnitudes mechanism than the toppling mechanism interpreted for the
The pattern of total cumulative displacements observed from upper slope.
Sites A and B during the monitoring period ranged between
+26 to −220 mm/day and +35 to −275 mm/day, respectively. 6 CONCLUSION
Displacement patterns created from each site were similar with
slightly different boundaries and magnitude. Simultaneous monitoring with two synthetic aperture radar
The combined displacements measured from the two instru- units has led to the creation of a high resolution, pseudo-3D
ments leads to the pattern observed in Figure 2. Displacement displacement map of an open pit slope. Review of the dis-
values ranged between 0 to 307 mm/day. These results show placement vector map allows for an increased understanding
maximum displacement values that are higher than the approx- of the overall slope kinematics, adding information to what
imately 100 mm/day reported for the geodetic prisms (Rose & would be attained by the less dense geodetic prism coverage.
Scholz, 2009). This is primarily due to the large number of Movement between prisms no longer needs to be interpreted
points being measured, the opportunity to measure near verti- and small areas of high displacement that may pose potential
cal rock faces and measurement of areas that would otherwise safety concerns, for example that along the west wall access
be moving too rapidly to install geodetic prisms. ramp, can be identified and monitored. Areas which cannot
By creating the pseudo-3D displacement map, a more accu- be monitored due to poor or dangerous access can be covered
rate representation of the location and pattern of displacement by the radar without the need to install prisms.
in the slope can be achieved.
5.2 Vector displacement REFERENCE

To further ascertain the general movement of the monitored Rose, N. & Scholz, M. 2009. Analysis of complex deformation
slope, displacement vectors were resolved and plotted (Fig. 3). behaviour in large open pit mine slopes using the Universal Dis-
The enlarged region shows an example of an area with high tinct Element Code (UDEC). In Slope Stability 2009: Proceedings
displacement rates. By plotting the displacement vectors, it is of the International Symposium on Rock Slope Stability in Open
observed that there are two independent lobes of movement Pit Mining and Civil Engineering, Santiago. CD-ROM, 11 pp.
731

Rock slope stability analysis using the discontinuous deformation analysis
J.P. Sun & Z.Y. Zhao

Keywords: numerical modeling, rock slopes and foundations, stability analysis
The stability of fractured rock masses is one of the classical method is higher than that by the DDA. Sarma’s method car-
problems in rock mechanics and still poses a major chal- ries out analysis based on the initial configuration of the rock
lenge to designers. The limit equilibrium method and the slope, which means that a bock slides only along its own base
continuum methods are best suitable to evaluate the factor segment. However, the DDA considers the initial configura-
of safety against failure for the rock slopes which are com- tion as well as the block kinematics. In the DDA analysis,
prised of massive intact rocks, weak rocks, or heavily fractured at the verge of the potential failure initiation, the small dis-
rock masses. However, when rock mass behavior is con- placement taking place to a block may carry it to contact with
trolled by the discontinuities, such as fractures and joints, the another base segment adjacent to its original base segment.
interaction between individual rock blocks defined by the dis- As the adjacent base segment may have different base angle,
continuities must be taken into account, the limit equilibrium friction angle and cohesion, the stability of the corresponding
method and the continuum methods may not be appropri- block and the whole slope will change accordingly. Figure 4
ate for the analysis of such block system. The Discontinuous
Deformation Analysis (DDA) is highly applicable to the mod-
eling of discontinuity-controlled rock block systems, allowing
two-dimensional analysis of slope failure mechanism and is
capable of simulating large displacements due to slip, or open-
ing along discontinuities. The advantage of the DDA is that
the mode of failure is one of the results of the analysis, rather
than one of the underlying assumptions. In addition, the DDA
reproduces the real displacements and interaction forces along
discontinuities within the sliding mass, as well as along the
failure surface.
This paper presents a detail study of the DDA for rock slope
stability analysis. The study concerns on the stability analysis
of a classic rock slope (Figure 1). A fictitious horizontal accel-
eration Kc is used in the DDA analysis as the safety measure
of the two-dimensional slope. For a given factor of safety, F, a
horizontal volume force, Kc Wi , will be acted on all the blocks
simultaneously. With the variety of Kc , the block displace- Figure 2. Displacement versus time at point A for a given factor of
ment will vary accordingly and a critical value of Kc can be safety.
obtained when the slope is at the verge of potential failure initi-
ation (Figure 2). The comparison of the F − Kc curves derived
by the DDA and Sarma’s methods (Figure 3) indicates that for
a given factor of safety, the value of Kc obtained by Sarma’s
Figure 1. Block partition of a rock slope. Figure 3. Factor of safety F versus horizontal acceleration Kc .
732

Figure 4. Distribution of normal thrust force between blocks.
Figure 6. The ratio of shear thrust forces to corresponding shear
strength along interfaces in DDA analysis.
relative to the right one in reality. The DDA can identify the
directions of the block movement and the shear thrust forces
automatically based on the block kinematics. In the DDA anal-
ysis as shown in Figure 6, the ratios of the shear thrust forces
to the corresponding resistant shear strength are all smaller
than 1, the limit equilibrium condition will not occur along
the interfaces between blocks at the initiation of slope failure
in this example. Meanwhile, since the ratio between the shear
thrust force and the shear strength varies greatly at different
interfaces, it can also be deduced that when the failure occurs
along the interfaces in the slope failure process, the failure
will not take place simultaneously among all the interfaces,
which is totally different from the corresponding assumption
in Sarma’s method. Sarma’s method is revised so that the rel-
Figure 5. Distribution of shear thrust force between blocks. ative movement directions of adjacent blocks are determined
based on DDA’s results and the failures don’t take place along
shows the distribution of normal thrust forces between blocks the interfaces. The results show that the factor of safety form
at the verge of the potential failure. Because different Kc is the revised Sarma’s method is closer to that from the DDA,
applied, the normal thrust forces obtained from the two meth- but there are still big differences (Figure 3). The reason is that
ods are different but with similar distribution patterns. Figure Sarma’s method carries out analysis based on the initial con-
5 shows that the distribution of shear thrust forces between figuration of the rock slope and the DDA considers the initial
blocks obtained from Sarma’s method is more regular than configuration as well as the block kinematics. The results illus-
that from the DDA. The distribution of shear thrust forces trate that the normal thrust forces and the shear thrust forces
from the DDA is very complex with smaller absolute values obtained from the two methods (DDA and revised Sarma’s
and shows that the direction of possible relative movement of method) are different but with similar distribution patterns
two adjacent blocks may not be the same as Sarma’s assump- (Figures 4 and 5). As a numerical modeling method, the DDA
tion. Sarma’s method only considers the condition that the is much more complicated both in theory and application;
left block moves upward relative to the right one. However, nevertheless, it gives a more realistic solution to a problem
it is sometimes possible that the left block moves downward and can be used to reproduce the whole rock failure process.
733

Mechanism of rock avalanche induced by earthquake – Insight from the discontinuous

numerical modeling approach
Xinbin Tu
Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, P. R. China
Gaofeng Zhao
School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW, Australia
Fuchu Dai
Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, P. R. China
Jian Zhao
Ecole Polytechnique Federale de Lausanne (EPFL), Laboratory for Rock Mechanics (LMR),
EPFL-ENAC-IIC-LMR, Lausanne, Switzerland
Keywords: numerical modeling; rock mass; rock slopes and foundations; stability analysis
1 INTRODUCTION
The Ms 8.0 Wenchuan earthquake triggered more than 56,000

landslides in the area along the seismogenic fault, including
shallow, disrupted landslides, rock falls, deep-seated land-
slides, and rock avalanches (Dai et al., 2010). According
to field investigation and remotely sensed imagery post-
earthquake mapping, we found that most of large rock
avalanches distribute along the surface rupture zone of seis-
mogenic fault. Based on the field observations, Dai et al. (in Figure 1. Result of DDA modeling: stable slope under static
review) proposed that oblique-thrust faulting should be a key condition.
factor in the initiation of some rock avalanches.
In this paper, the Discontinuous Deformation Analy- In this study, three types of slope stability analysis are con-
sis (DDA) is adopted to verify the mechanism of rock sidered, namely, the static analysis, the uplift analysis and the
avalanche induced by the Wenchuan earthquake. A typical earthquake loading analysis.
rock avalanche, the Shuimogou rock avalanche, is simulated
to verify the possible failure mechanism by oblique-thrusting 2.1 Case 1: The static analysis
faulting during the 2008 Wenchuan earthquake.
The static analysis is to verify the model and parameters
adopted. In this case, the only force acted on the slope is the
2 METHODOLOGY gravitation. Different shear strengths for the potential failure
surface are adopted in the numerical model to verify the static
For rock slopes with discontinuities which govern the behavior stability of the slope. Fig. 1 shows a stable state under static
of rock mass, DDA is a good method for simulating its defor- condition for the slope with cohesion of 20 kPa and friction
mation (Hatzor et al., 2004). DDA solves the stability analysis angle of 15◦ . The result suggests that the slope is stable under
for rock slope with large displacement and deformation, cal- normal condition of granite rock mass.
culating the forces and displacement among the blocks (Shi
and Goodman, 1985; Shi, 1988). DDA considers both stat-
2.2 Case 2: The uplift analysis
ics and dynamics using a time-step marching scheme and an
implicit algorithm formulation. In this case, in order to simulate the activity of oblique-thrust
The parameters raised in this paper were based on the fault, the hanging wall is uplifted by the fault slip along
observations during field work in the Wenchuan earthquake- the fault during the earthquake. The uplifting
√ distance along
affected regions between 2008 and 2010. The rock avalanche the fault surface is set to be 5 m (3.0 * 2 = 4.24 m), and the
debris originated from the granitic rock mass. At northeast- friction angle is 15◦ , the cohesion is 20 kPa for the slope fail-
ward of the scarp, NNW-side up thrusting and right-lateral slip ure surface. The results are shown in Fig. 2, which indicates
of the fractured rock mass are measured as 3.0 m and 3.0 m, that even at slow velocity of fault slip, the slope would undergo
respectively. serious failure.
734

3 DISCUSSIONS
Here we compare the different initiative mechanism for the

Shuimogou rock avalanche by DDA modeling. To reveal the
feasible initiative mechanism, two criteria are employed to
evaluate the modeling effect.
1) Modeling with the same material parameters. In this
case we can compare the slope stability under different
mechanisms.
2) Simulating the critical stable state with different parameters
for the slope. In this case we can compare the parame-
ters obtained under different mechanisms. The feasible one
should be with large shear strength for the potential failure
surface.
Due to variation of both friction angle and cohesion, it is dif-
ficult to obtain the suitable shear strength for the potential slip
surface. In this study, we adopt the first criterion to evaluate
the DDA modeling effect for rock avalanche mechanism.
As DDA can not directly give the Factor of Safety (FOS) of
Figure 2. The results of uplift analysis with DDA. slope, the slope stability under different loading cases can be
compared by the failure effect, i.e., rock mass displacement or
deformation. Comparing Figs.2 and 4, we can find that, under
the same condition, both the uplift loading of the hanging
wall for the fault and the earthquake loading can induce slope
failure, but the effect of the former is more prominent. This
indicates that the slope is more vulnerable during the fault
movement. The initiative mechanism for the rock avalanche
should be uplift loading by fault activity.
4 CONCLUSIONS
In this study, the Discontinuous Deformation Analysis (DDA)

is adopted to verify the mechanism of rock avalanche induced
by the Wenchuan earthquake. Two initial causative factors
are considered in the numerical model to compare and ver-
ify the effect on the formation of rock avalanche, namely,
ground motion and displacement along the seismogenic fault.
In spite of many landslides caused by ground motion during
earthquake, several large rock avalanches occurred during the
Figure 3. The earthquake loading adopted in dynamic DDA Ms 8.0 Wenchuan earthquake are proved to be mostly likely
modeling. caused by the uplift loading of the hanging wall during fault
movement. The oblique-thrust fault is an important aspect in
the analysis of the initiative mechanism of rock avalanches.
REFERENCES
Dai F.C., Tu X.B., Xu C., Gong Q.M., Yao X. In review. Rock

avalanches triggered by oblique-thrusting during the May 12, 2008
Ms 8.0 Wenchuan earthquake, China. Geomorphology.
Dai F.C., Xu C., Yao X., Xu L., Tu X.B., Gong Q.M. 2010. Spatial
distribution of landslides triggered by the 2008 Ms 8.0 Wenchuan
Figure 4. Result of dynamic DDA modeling under earthquake
earthquake, China. Journal of Asian Earth Sciences, 40, 883–895.
loads (friction angle = 15◦ , cohesion = 20 kPa).
Hatzor Y.H., Arzi A.A., Zaslavsky Y., Shapir A. 2004. Dynamic sta-
bility analysis of jointed rock slopes using the DDA method: King
2.3 Case 3: The earthquake loads Herod’s Palace, Masada, Israel. International Journal of Rock
Mechanics & Mining Sciences, 41, 813–832.
The Shuimogou rock avalanche locates about 80 km in the Shi G.H. 1988. Discontinuous DeformationAnalysis —A new numer-
northeast to the epicenter. The earthquake loads recorded by ical model for the statics and dynamics of block systems. Ph.D.
the Bajiao seismic station in Shifang city and the ground Thesis, University of California, Berkeley, USA, 378 pp.
acceleration monitored is shown in Fig. 3. The dynamic DDA Shi G.H. Goodman R.E. 1985. Two dimensional discontinuous
modeling result with given friction angle of 15◦, cohesion of deformation analysis. International Journal for Numerical and
Analytical Methods in Geomechanics, 9, 541–556.
20 kPa for the slope is shown in Fig. 4. It can be seen that the
slope fails under the earthquake loading.
735

Stability analysis of jointed rock slope evaluated by 3DEC
Alex Varughese
Central Soil and Materials Research Station, New Delhi, India
Ganesh W. Rathod & K.S. Rao

Department of Civil Engineering, Indian Institute of Technology Delhi, New Delhi, India
Subject: Modelling and Numerical Methods
Keywords: Rock Slopes and Foundations, Stability Analysis, Rock Joints
1 THE PROJECT
Two railway bridges is proposed to construct to cross River

Chenab and River Anjikhad, which is a tributary to the Chenab
River, near Reasi on the proposed railway line between Katra
and Qazigund of Northern Railways. The Chenab Bridge, once
it is constructed, would be the Highest Bridge in the World at a
height of 359m from the river bed level. The Anjikhad Bridge
will extend from Chainage 38.430 Km on left abutment on
Katra side to Chainage 39.087 Km on right abutment on Reasi
side. The Anjikhad Bridge will be at 190 m high above the river
bed level (with a deck level at c/l of bridge corresponding to
776.584 m) and 657 m long. This bridge consists of 16 piers
resting on ground. Among them 4 piers designated as A1 , P1a ,
P1b and P1 are on the left abutment and the other 12 piers
i.e. P10, P11, P12, P13, P14, P15, P16, P17, P18, P19, P20 and P21 Figure 1. Actual Topographical Model of Left Abutment.
are resting on the right abutment. Both P1 and P10 are the
arch foundations of the bridge on left and right abutments right side of the arch foundation P1 respectively and point 6 is
respectively. The loads exerted by these piers would affect the at toe. The fourteen Monitoring Points (MP) 1 2 3 4 5 6 7 8 9 10
stability of the slopes. The hill slopes are carved due to the per- 11 12 13 and 14 were installed in the model of right abutment
sistent geological processes with large nallahs and with steep at a depth of 10 m i.e. at foundation level. The point 2 and 3 are
natural slopes of 60–65◦ inclinations. The bridge alignment is on left and right side of the arch foundation P10 respectively
N120◦ E towards the left abutment to N300◦ towards the right and point 1 is at toe. Here displacements, shear stress and shear
abutment and the Anjikhad River flows in SW direction. velocities are recorded throughout the numerical simulation.
The stresses in X, Y, Z, directions, shear stresses in XY, YZ,
XZ planes, X displacement, Y displacement, Z displacement
2 NUMERICAL MODELLING contours , X velocity , Y velocity, Z velocity contours, maxi-
mum and minimum principal stresses at Z = −100 (along the
In order to determine the stability of both the abutments anal- proposed railway alignment), Z = −50 and Z = −150 (50 m
ysis was carried out using 3 Dimensional Distinct Element on the either side of the proposed railway alignment) were also
Code on an equivalent model of size with a width of 200 m recorded.
(100 m on both sides from the proposed railway alignment).
Since the actual topographical model (Fig. 1) is not accepted
by 3DEC so equivalent model (Fig. 2 depicting all the major 3 RESULTS AND DISCUSSION
topographic features were used in 3DEC. Roller boundary
conditions are assumed along the lateral sides of the model The numerical analysis was carried out for static loading case
such that no displacement is allowed in the x-direction. At the and pier loading case. In static loading case the prevailing
base of the numerical model, the boundary is fixed such that rock conditions at the site where the rock mass is observed to
no movement is allowed in the y and z direction. The depth be stable is stimulated considering rock properties and pres-
of foundation is assumed as 10 m. Out of twelve Monitoring ence of joint sets. In this case, pier loads were applied to the
Points (MP) six Monitoring Points (MP) 1 2 3 4 5 and 6 were rock slope to simulate the actual conditions. Each deformable
installed in the model of left abutment at a depth of 10 m i.e. block is then discretized into a mesh of tetrahedral elements
at foundation level and the other six were installed at ground with edge length of 10 m, where the motion of the block is
level as shown in Figure 5. The points 4 and 5 are on left and calculated at the grid point of the tetrahedral elements. The
736

3.2 Right abutment: Jointed rock with pier loads (case JR2)
The X displacements at all monitoring points varies between
−2.25 to 0.1 mm, Y displacements from −3 to 1–6.5 mm and
the Z displacements from −0.1 to 0.13 mm. The XX YY and
ZZ stresses at left of P10 varies from −1.18 × 106 Pa and right
of P1 varies from −0.77 × 106 Pa respectively. The XX, YY,
ZZ, XY,XZ andYZ stresses at Z = −100 m ( along the railway
alignment) varies from 0.3 × 106 Pa to −6.5 106 Pa respec-
tively. The XX,YY, ZZ, XY, XZ andYZ stresses at Z = −50 m
and Z = −150 m (50 m from the railway alignment) varies
from 0.4 × 106 Pa to −6.5 106 Pa respectively. The displace-
ments, stresses and velocities at Z = −50 m, Z = −100 and
Z = −150 m values confirm it to be a very stable slope.
However there are locations where nominal shear stresses
developed in the slope. The shear velocities observed are
approaching to zero confirms a stable state of the slope. The
shear stresses and displacement values are very less which
indicates that there is no movement of blocks. It is also very
common that movement of random blocks on the surface due
Figure 2. Jointed Slope Model of Chenab Left Abutment used in to delineation of joints and no surface cover. The results from
3DEC Analysis.
both methods indicate the stable nature of slope
case JL 12 of the left bank jointed rock with pier loads and
case JR2 of the right bank were considered to be the most 4 CONCLUSIONS
critical and are discussed here. Maximum Shear Stress, Dis-
placement, Minor and Major Principal Stress, and Velocity The vertical dimensions of left and right abutment considered
contours were observed in all the cases at Z = −50, −100 and in modelling are 404 m and 232 m respectively. And the cor-
−150. responding horizontal dimensions are 570 m and 670 m. The
effect of large difference in slope height of left and right abut-
3.1 Left abutment: Jointed rock with pier loads (case JL2) ments is observed from the results. The rock properties are
The X displacements at all monitoring points varies between relatively better at left abutment than the right. But the height
2.3 to 2.5 mm. The Y displacements at all monitoring points and base length found to be the governing parameters for the
varies between −5 to −8 mm. The displacements at all mon- slope stability. 3DEC analysis revealed left and right slopes
itoring points varies between −0.1 to 0.1 mm. The XX YY are stable under different loading conditions. A critical obser-
and ZZ stresses at left of P1 varies from −0.06 × 106 Pa to vation is attempted at P1 and P10 arch foundation. The shear
−0.6 × 106 Pa and right of P1 varies from −0.03 × 106 Pa displacements, shear velocities and shear stresses at all mon-
to −0.45 × 106 Pa respectively. The XX YY ZZ XY XZ itoring points for both left and right abutments are within the
and YZ stresses at Z = −100 m (along the railway alignment) acceptable limits and shows that the slopes are stable in static
varies from 0 Pa to −7.5 × 106 Pa respectively. The XX YY condition.
ZZ XY XZ and YZ stresses at Z = −50 m and Z = −150 m
(50 m from the railway alignment) varies from −0.5 × 106
Pa to 8 × 106 Pa respectively. The displacements, stresses and
velocities at Z = −50 m, Z = −100 and Z = −150 m values
confirm it to be a very stable slope.
737

Estreito HPP – Geotechnical and geomechanical characterization of the Sandstone

foundation rock mass
Erik Wunder
Estelar Engenheiros Associados Ltd.
Márcia Collares Meireles

Tractebel Engineering – Leme Engenharia
Nicole Borchardt & Alex Martins Calcina

Intertechne Consultores S.A.
Martin Afonso Camargo

CNEC WorleyParsons Engenharia S.A.
Keywords: weak rock, lab testing, site characterisation, rock mass, rock properties
1 INTRODUCTION Table 2.1. Average values from the physical characterization tests.
The Estreito HydroPower Plant is a concession of the consor- Absorption Porosity

tium formed by GDF Suez, Vale, Alcoa and Camargo Correa ρ (dry) ρ (wet) (S) (P)
g/cm3 g/cm3 % %
Energia. The total capacity is 1.087 MW with an average of
584.9 MW. Mean 1.89 ± 0.09 2.10 ± 0.05 11.63 ± 2.66 21.72 ± 4.07
The Estreito HPP is composed by a rockfill and a compacted Values
soil dam with 46.5 m maximum height, a spillway with 14
gates on the left bank and a water intake and power house
placed at the right bank. Table 3.1. Results from uniaxial compression tests and deforma-
The sandstone lithotypes of the Sambaíba Formation from bility modulus.
Parnaíba-Maranhão Sedimentary Basin correspond to the
foundation rock mass for the main civil structures. This strati- ρ σc E
graphic unit corresponds to an eolic sandstone sequence, with g/cm3 MPa GPa
large crossed stratifications (layer), which represents the final
Mean Value 1.86 ± 0.11 18.07 ± 7.85 16.89 ± 6.18
phase of desertification of the basin. It is tipically formed
by medium to course grained quartz, sub-angular and sub-
rounded and matte surface, with predominant red to orange Table 3.2. Results from Brazilian tests.
and yellow colors.
A serie of tests were done to characterize the sandstone ρaparente σt
rock as to its strength properties, deformability and hydraulic g/cm3 MPa
conductivity, beyond the physical characterization. It was also
Mean Value 1.73 ± 0.05 1.04 ± 0.32
developed 387 drill hole Lugeon tests were done in situ which
supplied interesting information to characterize the hydraulic
conductivity of the sandstone rock mass.
3 GEOMECHANICAL CHARACTERIZATION
3.1 Uniaxial compression strength and deformability

2 PHISICAL CHARACTERIZATION modulus
In this campaign 23 samples were tested for σc and E and 6
The samples tested were identified by the place they were samples for σt . For each test it was also obtained the density
collected, such as, 23 samples on the river bed, 18 samples of the sample. The average values are presented below:
on the right margin and 24 samples on the left margin. The It is shown on CD-ROM the graphics with the correlations
average values are presented below. between these parameters.
The graphics on CD-ROM show the relationships between It was also performed 40 punctual strength – Is(50) . The
dry specific weight, absorption and porosity. mean values were.
738

Table 3.3. Average values of tensile resistance. Table 4.2. Hydraulic conductivity associated to confinement
condition.
(1)
Is(50) σc
MPa MPa Number of tests k (cm/s)
Mean Value 0.79 ± 0.22 19.07 ± 5.31 confined(1) unconfined(2) confined unconfined
(1)–σc = 24.Is(50) River bed — 113 — 3.61 × 10−4

Right margin 59 92 9.37 × 10−5 3.44 × 10−4
Table 3.4. Average values of uniaxial strength tests. Left margin 2 121 2.88 × 10−5 2.45 × 10−4
Humidity condition σc
MPa Table 4.3. Influence of the lithostatic unload over permeability.
dry 8.09 ± 3.26 Local Condition Number k(cm/s)

saturated 9.52 ± 2.64
Left margin confined 2 2.88 × 10−5
Table 4.1. Mean values of hydraulic conductivity. Left margin unconfined 46 2.45 × 10−4
Left margin(1) post excavation 75 1.15 × 10−3
Number of tests k (cm/s)
(1)–20 to 30 m underneath the foundation rock surface
Mean Value 387 3.44 × 10−4
The results show an increase of 2 (two) orders of magni-
tude in the hydraulic conductivity condition “post excavation”
3.2 Uniaxial strength – complementary tests in relation to the condition “confined” and 1 (one) order of
Additional uniaxial compression tests were done on samples magnitude compared to the condition “unconfined”, which
collected in dry and saturated conditions. leaves clearly the conclusion of the direct influence of the
It should be considered that those tests were developed on lithostatic unloading over the hydraulic flow characteristics
site using hydraulic jack for concrete testing and with the of the sandstone rock mass.
inherent difficulty for getting parallel faces on rock sam-
ples, which could have influenced the results, conditioning 5 CONCLUSIONS
relatively lower values.
3.3 Direct shear tests The execution of laboratory tests campaigns had a essential
role in the geotechnical and geomechanical characterization
The direct shear tests were performed on 11 samples, 6 of of the Sambaíba sandstone.
them with the normal load parallel to bedding, and the other The results of the tests described the geomechanical Sam-
5 with the normal load applied perpendicular to the stratifi- baíba sandstone as a soft rock, with uniaxial strength of 18
cation. The linear approximations to the loading conditions MPa (average) and about 17 GPa for deformability modulus.
considered were: The porosity observed was 21.7% and the absorption was
the 11.6%.
−τ = 1.040σn + 1.081, (σn parallel)
The complementary campaign of uniaxial compression
−τ = 1.025σn + 1.320, (σn perpendicular)
tests showed a similarity between dry and saturated conditions
of the sandstone.
4 PERMEABILITY The results of direct shear tests showed the influence of the
anisotropy, given by the eolic stratification, especially on the
It was conduct 17 laboratory tests, where 8 were done with cohesion value.
variable load permeameter and 9 in the triaxial cell. The mean The results comparison for permeability tests performed
value was 5.78 × 10−5 cm/s. in the laboratory and in situ indicated that secondary perme-
The 387 drillcore hole Lugeon tests were executed along ability is about 1 order of magnitude larger than the primary
all the construction. The table below shows all the results permeability.
obtained. The results gathered by the series of drillcore hole Lugeon
It is clear the scale effect, since the permeability of the showed a direct influence of confinement conditions to the
sandstone rock mass (3.44 × 10−4 cm/s) is one order of sandstone rock mass on its hydraulic conductivity.
magnitude larger than the permeability of the sandstone rock All the results obtained by these campaigns of tests were
(5.78 × 10−5 cm/s). complementary to the geological model set to the implan-
The results of the field tests were also analyzed for the tation site of the plant, together, helped in making the right
condition of confinement, associated to the existence of decision for the development of the Executive Design and
Mosquito basalt covering and loading the Sambaíba sandstone Implementation of HPP Estreito.
where the tests were performed.
Note then that the condition of confinement has a direct REFERENCES
influence on permeability. The unconfined tests have a higher
conductivity than the confined condition in about 1 (one) order ES1-RT-C-USC/GG-001 Geomechanical tests on the rock
of magnitude. foundation.
Another interesting characteristic shown by the Lugeon ES1-RT-C-USC/GG-002 Geotechnical and geological model of the
tests results corresponds to the influence of the lithostatic load massive foundation.
relief created by the excavation over the remaining sandstone ES1-RT-C-USC/GG-003 Strength parameters estimation of the
rock mass. massive foundation of the structures.
739

Mechanism and dynamics of dip-slope failure revealed by LiDAR data and

Discrete Element Method
Chih-Hsiang Yeh & Ming-Lang Lin

Yu-Chang Chan
Academia Sinica, Institute of Earth Sciences, Taipei
Kuo-Jen Chang
Keywords: dip-slope failure, LiDAR DTM, discrete element method
Recent in Taiwan the dip-slope landslide failure frequently first obtained the basic engineering parameters, c, φ, and τ,
caused huge casualties. Except for rainfall or earthquake by numerical modeling method, 2-D discrete element method,
directly induced slope failure, the man-made interference such in one case study. At same time, we considered groundwater,
as cutting the toe of slope had been one of the crucial factors of retaining wall, and geological factors so that to simulate real
losing slope stability. One representative case of dip-slope fail- field condition. Finally, we are going to look for all of the
ure occurred on freeway locates Keelung river northern basin, potential slopes using LiDAR DTM, and evaluate the possi-
which consist of dip-slope strata. Noteworthy issue around bility of sliding of slope failure in those slopes through above
this area is that there is still a lot of potential slopes may slide case study.
because of influences of artificial building. In this study we
740

Interdisciplinary

A comprehensive approach in predicting excessive sand production
M. Bai, C. Santana & X. Shen

Halliburton, Houston, TX, US
Subject: Petroleum geomechanics
Keywords: rock failure, stability analysis, weak rock
In this paper, a comprehensive geomechanical approach is When the core samples from the field are secured, experimen-
proposed to predict the excessive sand production in an effec- tal testing is an effective way to simulate the sanding scenarios
tive way that includes: a) employing analytical method with under realistic, simulated down-hole conditions. The results
the combination of experience-based empirical rules and ana- can be used to validate the studies obtained from analytical
lytical expressions from the simplified sanding model (e.g., and empirical methods.
determining the correlation between critical drawdown rate The benefits of adopting the unified approach are partially
and formation rock strength, delineating the sand-free enve- explained in a case study of this paper. In conclusion, the
lope based on the reservoir depletion history and desired ultimate purposes of employing the unified approach can be
critical bottomhole flowing pressure as well as formation com- given as follows:
pressive strength, and assessing the suitability of each method
• To avoid applying one particular method that can be biased
under a particular case); b) performing a numerical method
and can lead to an insensible decision for sand control.
using the complex constitutive relationships and realistic com-
• To generate the self-calibrated results of sanding predic-
pletion configurations in three-dimensional field environment
tion that can result in making an effective and economical
(e.g., recording the transition of formation from brittle to
decision for sand control.
ductile states with increasing drawdown pressure until the
catastrophic failure occurs, i.e., the rock plastic deformation With respect to the application of the presented technology,
exceeds the critical value; employing the finite element tech- the presented method can be used to predict excessive sanding
niques to represent the behavior of heterogeneous rocks, such potential and assess completion-device stability near perfora-
as faults; and simulating the production scenario under desired tion tunnels to aid a good decision on the type of sand control
sand prevention devices); and c) conducting an experimental method to be employed. In a particular case study, the weak
method using laboratory rock mechanical testing techniques spots in the reservoir perforation intervals were identified.
(e.g., determining the correlation between wellbore strength The formation strength at the weak spots was determined from
and maximum differential pressure via thick-wall cylinder sonic measurement and rock mechanical testing data. The crit-
tests, calculating the maximum drawdown pressure using ical values of sanding rate, unconfined compressive strength,
simulated depletion tests, and determining the impact of critical plastic strain, critical bottomhole flowing rate, maxi-
stress path on the maximum depletion from pore-volume mum wellbore strength, and critical depletion pressure were
compressibility tests). assessed using the analytical, empirical, numerical, and exper-
Using one particular method (e.g., either analytical, or imental methods under the field conditions (e.g., reservoir
empirical, or numerical, or experimental) may be convenient, pressure, in-situ stress, stress path, rock strength, etc.).
but it can lead to an incomplete or a biased analysis of sanding Based on the conclusions drawn from the paper, it is noted
prediction. To avoid this shortcoming, the presented technique that the presented approach on the sand prediction differs
provides a unified approach that employs multiple methods from the conventional approach, where the latter is often
with complementary objectives. In the present study, atten- biased toward a particular method (e.g., empirical, analyt-
tion is focused on predicting the critical drawdown pressure ical, numerical, or experimental). The presented technique
(CDP) under tolerable sand production. provides a unified approach that uses multiple methods with
The primary purpose of the analytical and empirical meth- complementary objectives: a) analytical and empirical meth-
ods in this paper focuses on determining the critical drawdown ods initially assess the sand production risk under the simpli-
pressure constrained by the sand-free envelope. The sand-free fied conditions; b) the numerical method examines wellbore
envelope is defined by the border lines of: a) an equal line stability issues under the detailed configuration of sand pre-
between the critical bottom-hole flowing pressure and reser- vention devices; and c) the experimental method calibrates the
voir pressure, b) the critical drawdown pressure at the current sand prediction under the simulated downhole conditions.
reservoir condition, and c) the minimum critical bottom-hole There are number of significant results being derived from
flowing pressure. the paper. Noticeably, classical sand control techniques are
For complex perforation geometry, numerical methods, primarily based on installing the sand exclusion devices, such
such as the finite-element method, are more amenable. Using as gravel packing, expandable screens, etc., which dramati-
critical plastic strain to assess the sanding-induced perfora- cally increase the completion costs and detrimentally reduce
tion tunnel stability via the finite-element method, numerical the efficiency and potential for the hydrocarbon production.
methods can also evaluate the critical drawdown pressure The presented approach uses the self-calibrated predictions
under the specific production scenario. The process parallel on sand production via multiple methods, which can lead to
to the analytical and empirical methods can provide a sensible a sensible decision for selecting the effective and economical
validation for the determined critical drawdown pressure. sand control measures.
Numerous experimental methods involving laboratory-
rock mechanical testing are available for the sanding analysis.
743

The theory of underground structures design has gained strength
N.S. Bulychev, N.N. Fotieva & P.V. Deev

Tula State University, Tula, Russia
ABSTRACT: The theory of analytical methods for underground structures design was born in the middle of twentieth century
due to professor K.V. Ruppeneyt, who suggested to consider mine support and rock mass as elements of a united deformable
system. Small values of contact displacements make possible to use linearly deformable medium as a model of rock mass.
Today modern design methods based on rigorous analytical solutions of the elasticity theory plane problems are developed for
stress state determination of shaft supports and tunnels linings subjected to different loads.
Keywords: Mine design, back analysis
1 INTRODUCTION methods. The first analytical solution, which can be used for
lining design, had been obtained by Gabriel Lame in the mid-
The first attempts to obtain underground constructions stress dle of nineteenth century. The problem was quite simple and
state are dated by the end of nineteenth century. Early design consists in stress strain state determination of a ring sub-
methods are based on structural mechanic solutions of the jected to inner and outer pressures. Some developments of
problems consist in internal forces determination for a bent this problem led to appearance of an analytical method for
axis beam subjected to the vertical and horizontal pressures. shaft lining design.
Later these methods were developed by external load dividing Muir Wood (1975) and Curtis (1976) had obtained analyt-
into ‘active’and ‘passive’components. Passive part of the load ical solutions for a circular ring supporting an opening in the
consisted in the reaction of elastic foundation to the beam linearly deformable medium, which later has been used for
displacements. The exclusion of rock mass from the design the design of Channel Tunnel Rail Link in UK.
scheme turned internal load determination to a very difficult The analytical method for design of multilayer tunnel and
problem. shaft linings has been developed in the same time in Russia by
The conception of joint contact interaction proposed by Bulychev (1974). This method was used for design of some
Fenner and Ruppeneyt (1954) had led to the basis of a new tunnels with total length about 60 km along Baikal-Amur
theory of underground structures design. The new approach railway, and numerous shafts. Further development of this
does not demand rock pressure determination, because nor- method made possible to solve the inverse problem connected
mal and shear stresses along the contact line are obtained with initial stress field determination on the base of stress or
as intermediate values during calculations. The input data displacement measurements in some points of a tunnel lining.
for calculations are deformation properties of the rock mass Design of non-circular tunnel linings needs solution of
and a support material as well as parameters of the intact rock more complex problems of the elasticity theory. The general
mass stress field. approach to the wide class of elasticity theory problem was
Structural mechanics methods were not able taking into made by Muskhelishvily (1953). This approach was applied
account ‘rock-support’ interaction, so analytical design meth- for non-circular supports design (Fotieva, 1980). Analytical
ods based on solutions of elasticity theory problems have been design methods for non-circular tunnel linings subjected the
developed for underground construction design. The main action of rock weight, tectonic forces, underground water
features of underground structure analytical design theory are: pressure and internal pressure applied to the internal outline
of a lining cross-section have been developed.
– the rock mass and a support are considered as elements of
Another problem difficult to solve was seismic design of
a united deformable system subjected to external loads and
underground structures. The characteristic feature of seismic
actions;
loads is the fact, that seismic waves direction and combination
– rock mass model is linearly deformable medium;
are not known in advance. So, the design of an underground
– construction technology strongly influences upon under-
structure subjected to the action of prescribe directed seismic
ground structure stress state;
wave is not effective, because this direction hardly corre-
– mathematical apparatus of analytical design methods is
sponds to the real direction of incoming waves propagating
mathematical apparatus of the elasticity theory.
in the rock mass during the Earthquake.
Now there are various analytical design methods developed In the work (Fotieva, 1980) an original method of seis-
in the frame of general theory for stress state determination of mic design is proposed. The design consists in determining
shaft supports and tunnels linings subjected to different loads. the most unfavourable lining stress state at any combinations
and directions of long longitudinal (compressive-tensile) and
shear waves propagating in the plane of the tunnel cross-
2 MODERN ANALYTICAL DESIGN METHODS section during the Earthquake of certain magnitude. With the
aim of designing tunnel linings of an arbitrary cross-section
Analytical methods used for the design of circular tunnel shape two plane quasi-static contact problems of the elasticity
and shaft linings are the simplest among modern analytical theory are considered.
744

When underground structures are constructed close to each The general design theory being the base of modern ana-
other, the problem of their interaction arises. In the work lytical methods unites all types of construction methods and
by Fotieva & Kozlov (1992) a plane problem about stress all types of mine supports subjected vary types of loads and
state of the rings supporting circular openings in the linearly actions. It means that the theory has a large field for new
deformable medium is considered. The authors use special developments and achievements.
modification of approach proposed by Sherman (1951) for
stress state determination of multiply connected regions. The
sequence of tunnels construction was taken into account by REFERENCES
consequent consideration of a number of plane problems
corresponding to the all stages of tunnel complex construction. Aramanovich, I.G. 1955. Distribution of stresses in an elastic semi-
Now the design methods mentioned above are developed plane weakened by a supported circular opening. Reports of AN
USSR: Vol. 104, No. 3, 17–23.
for the case of a finite number of parallel underground struc- Bulychev, N., Amusin, B. & Olovyanny, A. Design of permanent
tures, which may have multilayer supports and may be con- support of mining working. Moscow: Nedra, 1974.
structed with the application of grouting (Fotieva et al., 2007, Curtis, D.J. 1976. Discussion on the Circular Tunnel in Elastic
2010). Ground, Geotechnique: 231–237, Vol. 26, No. 1.
The main difficulties in shallow underground structures Fotieva, N. 1980. Design of underground structures support in
design are the necessity taking into account the influence of seismic regions. Moscow: Nedra.
the surface and the initial stress field changing with the depth Fotieva, N. & Kozlov A. 1992. Design of the parallel openings
to the structures stress state. The first elasticity theory prob- support in seismic areas. Moscow: Nedra.
lem concerned underground structure stress state has been Fotieva, N., Bulychev, N., Sammal, A. & Deev, P. 2005. State
solved by Aramanovich (1955). The technique of complex and bearing capacity of shallow tunnel linings undergoing the
influence of nearly located buildings. Gornictwo I geoinzinieria.
potential analytical continuation been proposed in this article Kwartalnik Akademii Gorniczo-Hutniczey im. Stanislawa Staszica
later was applied to stress state determination of single shal- w Krakowie: 217–223.
low non-circular structures (Fotieva et al., 2005) and multiple Fotieva, N., Bulychev, N., Firsanov, E. & Deev, P. 2007. Tunnel linings
tunnel linings of arbitrary cross-section shape (Fotieva et al., design in a seismic area. Proc. of First Sri Lankan Geotechni-
2010). All analytical methods mentioned above are used for cal Society intern. conf. on soil and rock engeneering, Colombo,
underground structures design in Russia. August 6–11, 2007.
Fotieva, N., Bulychev, N., Deev, P. 2009. Multiple non-circular tunnel
linings design under seismic effects of Earthquakes. Performance-
based design in Earthquake geotechnical engineering. From case
3 CONCLUSIONS history to practice. Proc. intern. conf. Tokyo, June 15–18, 2009.
Fotieva, N., Bulychev, N. & Deev, P. 2010. Design of parallel
Today former design theories based on prescribe load principle non-circular tunnel lining constructed in urban areas with the
and structural mechanics solutions exhaust their facilities and application of grouting. Transport and City Tunnels. Proc. of
must not been recommended for practical use. the intern. conference Underground Construction, Prague, June
Numerical modelling of infinite and semi-infinite weighty 14–16, 2010.
areas gives results different from exact analytical solutions. Muir Wood, A.M. 1975. The circular tunnel in elastic ground.
Geotechnique, No 1.
In some cases non-controlled errors of modelling may lead to Muskhelishvili, N.I. 1953. Some basic problems of the mathematical
the inadequate results. There are some cases where numerical theory of elasticity. Netherlands: Noordhoof.
modelling data obtained on the same software by different Ruppeneyt, K. 1954. Some problems of rock mechanics, Moscow:
specialists have bad agreement with each other. So, mod- Ugletechizdat.
ern analytical methods have some advantages over numerical Sherman, D.I. 1951. About stresses in the plane weighty medium with
modelling and a good possibility of combined utilization and, two identical symmetrically located circular openings. Applied
in some cases, can replace them. mechanics and mathematics: 751–761. Vol. 15, No. 6.
745

Simulation of steel fiber reinforced shotcrete behavior using fracture mechanics
A. Fahimifar & B. Mehrzad

Faculty of civil and environmental engineering, Amirkabir University of technology, Tehran, Iran
Keywords: lab testing, numerical modeling, fracture mechanics
Many fracture mechanisms models have been suggested for particularly flexural strength and tensile strength of SFRS
concrete behavior, such as Hillerborg’s fictitious crack model specimens have increased significantly, the tensile strength
(Hilleborg et al. 1976), Bazant and Oh’s crack band theory corresponding to 4.5 percent steel fiber is two times of
(Bazant & Oh, 1983), Jenq and Shah’s two parameter fracture plain concrete. Also the post-peak behavior of this material
model (Jenq & Shah, 1985). improves with steel fiber content significantly.
In this paper, application of Hillerborg’s theory in Steel The experimental results obtained through the SFRS spec-
Fiber Reinforced Shotcrete (SFRS) is assessed, For this pur- imens were applied in the concrete damaged plasticity model
pose a series of laboratory tests were performed including, based on Hillerborg’s fracture theory using ABAQUS pro-
uniaxial compression, direct tension and flexural tests on gram, for different percentages of steel fiber.
SFRS specimens, then simulate the tests with concrete dam- This model named Concrete damaged plasticity uses two
aged plasticity based on Hillerborg’s theory for behavior of compressive and tensile mechanism of failure. Compressive
shotcrete and finally the results were compared. behavior, elastic modulus and poison’s ratio were extracted
The fracture energy required for extending the crack pro- from uniaxial compression test. Tensile behavior was pre-
posed by Hillerborg’s theory, for using this concept in the dicted in this simulation. The tensile behavior was estimated
present experimental investigation three parameters of uni- using Hillerborg’s fracture energy iteratively until the labo-
axial tension strength ( ft ), fracture energy (GF ) and crack ratory and simulating results were converged. In this way, a
cohesion stress curve (σ(ω)) are needed. tensile behavior for each steel fiber reinforced specimen was
The 28-days compressive strength of shotcrete was deter- suggested.
mined using 15 cm cubes according to ASTM C39 standard. Geometery of the model and loading is exactly similar
Direct tensile strength was performed in a particular man- to four-point flextural test, a three dimensional deformable
ner using cilynrical specimens with 5 cm in diameter and element and fine and middle-size meshes were applied in
10 cm length. In order to control the mode of failure in pure the model. The estimated curves suggest a two-line curve for
tension a weakend section was produced in the middle of tensile post-peak behavior.
the specimens (B. Mehrzad, 2010). Four-point flexural tests The results show a good agreement between laboratory and
on 10 * 10 * 35 cm specimens based on ASTM C1018 for concrete damaged plasticity based on Hillerborg’s theory. A
evaluating the flexural strength were used. two-line post-peak behavior for SFRS is appropriate. For lower
All the tests were carried out using a servo-controlled test- steel fiber contents the model curve lie below the laboratory
ing machine through strain control as the variable parameter curves and for higher steel fiber content it show higher results,
and the complete curves was obtained for each specimen. that means the curves stand above the laboratory results.
The results show that, there is a dramatic change in the
stress-strain curves of specimens containing steel fibers,
746

Rock mechanics aspects of drill bit rock interaction
S.A.B. da Fontoura, N. Inoue, I.M.R. Martinez & C. Cogollo

Computational Geomechanics Group ATHENA, Group of Technology in Petroleum Engineering GTEP,
Pontifical Catholic University of Rio de Janeiro, Brazil
D.A. Curry
Baker Hughes Incorporated, London, England
Keywords: numerical modelling, problem rocks, rock properties, rock stress
A clear understanding of the mechanics of the rock-cutting the interaction of a single PDC cutter with rock and hence
process is necessary to solve many problems of reduced to evaluate the MSE. In our modelling of the rock cutting
drilling efficiency. Assessing penetration rate performance process two type of analysis were carried out: 1) 2-D analy-
objectively is a difficult task. Numerical simulation can be sis of linear cutting and 2) 3-D analysis with curved cutting
a valuable instrument for analysing the particular conditions trajectory. For each model, the cutting simulation was done
of each borehole drilling operation without the often costly under atmospheric pressure (0 MPa) and at hydrostatic pres-
investment required for a full scale experimental investigation. sure (10 MPa), the latter simulating the drilling fluid pressure
In the petroleum industry, the Polycrystalline Diamond acting on the hole bottom when drilling at depth.
Compact (PDC) bit is a popular drilling tool that employs the The aim of this study was to evaluate the rock behaviour
cutting mechanism to shear a rock. It accounts for more than during cutting under pressure. In this condition, the rock
one third of the total footage drilled worldwide. It widespread behaviour is predominantly inelastic, so, the extended
use makes it important to understand the interaction of this Drucker & Prager model was used because it considers
tool with different rock environments and diverse conditions consider materials that exhibit pressure-dependent yield.
of drilling. The Drucker & Prager model provides three different yield
The single cutter test allows understanding the interac- criteria. The differences between these criteria relate to the
tion rock-cutting tool by determining the penetration and shape of the yield surface in the meridional plane, which can
drag forces acting on each cutter located on the bit face. be a linear form, a hyperbolic form, or a general exponent
Furthermore, this laboratory test enables to duplicate many form. The linear form assumes linear dependence of devia-
parameters that affect the forces including the rock type, cut- toric stress on hydrostatic pressure and the hyperbolic model
ter design and wear state, position on the bit, cutter interaction, makes a similar assumption at high confining pressure. There-
cutting speed, rock stress state, and fluid environment. In this fore, these forms of the model do not provide a particular
paper, a numerical model of single PDC cutter is developed good match to triaxial test data. Because the exponent form
to assess the forces of the cutting rock process. of the yield criterion provides the most flexibility in matching
The concept of Mechanical Specific Energy (MSE) has triaxial test data, it was selected for these simulations.
been used effectively to evaluate drilling efficiency, Pessier & For this study, we chose to model Carthage limestone, a rock
Fear (1992). The MSE is defined as the work done per unit whose deformation, fracture and drilling properties have been
volume of rock cut; the higher the MSE for a given drilling extensively studied and for which we had triaxial test data at
environment, the lower system efficiency. Currently, MSE several different confining pressure.
measurement provides the ability to detect changes in the First, we selected the point in each stress-strain curve
efficiency of the drilling system while drilling, Dupriest & corresponding to initial deviation from elastic behaviour for
Koederitz (2005). However, effective numerical simulation of calibrating the yield parameters. Thus, one stress data point
the drilling operation would allow an early estimate of the effi- from each stress-strain curve at a different level of confining
ciency of drilling, making it easier to propose and to improve pressure is plotted in the meridional stress plane. Finally, a
solutions for better drilling performance. least-square fit that minimizes the relative error in stress is
By definition, for a single cutter the MSE is the rock cutting used by ABAQUS to set the yield surfaces in the meridional
force integrated over the distance cut, divided by the volume plane and to determine the model parameters.
of rock cut. Hence, this paper focuses on the variation of the The model used the ductile damage initiation criterion.
cutting force (Fx), i.e, the component of the force parallel to According to this criterion the material starts to deform once
the direction of the movement of the cutter, with the position the shear strain reaches a certain critical plastic strain value.
of the cutter as that is directly related to the variation of the Furthermore, the Drucker & Prager compression type hard-
specific energy. ening option was used to define the hardening behaviour of
Our numerical simulation of cutting process used the com- the uniaxial compression yield stress, as a function of uniaxial
mercial Finite Element Analysis software ABAQUS to model compression plastic strain. The Displacement type option was
747

used to define the evolution of material damage as a function increases when the tool contacts the next element of the work-
of the plastic displacement from the initiation of damage up piece. So a finer mesh is more appropriate for representing of
to eventual failure. Once the damage reaches a critical level, the contact between the cutting tool and the rock as a continu-
the material fails and the element is removed from the mesh. ous medium. A more refined finite element mesh may better
The cutting process involves different problems such as the represent the rock as a continuum, but can also generate a
large deformations of the material at very high strain rates, higher cost of computational time, reducing the efficiency of
contact interactions rock-cutting tool, large plastic work and numerical simulation. Then the numerical simulation requires
high friction. All these problems make the numerical sim- a good choice of the number of elements in the mesh.
ulation of cutting process particularly complex. The results As expected from the triaxial test data, the models predicted
of the numerical simulation can be deteriorated because the that rock’s resistance to cutting increased with increasing
distortion of the elements resulting from large deformation. confining pressure. The simulations at atmospheric pressure
The finite element Lagrangian and the Eulerian formulations showed lower force required for cutting process than the
have been used in modeling of orthogonal cutting process of simulations at hydrostatic pressure (10 MPa).
metal and rock. In the Lagrangian formulation, the elements It was possible to observe the increasing forces with the
are attached to the material and are deformed with the move- cutting depth because. It can explain because when cutting
ment of the workpiece. This approach is appropriate when depth increases, the contact area between the rock and cutter
unconstrained flow of material is involved, then the FE mesh also increases, consequently the forces would be expected to
will correctly represent the material boundaries during the increase.
sequence of the analysis. The disadvantage of the Lagrangian The force behaviour shown in the numerical simulation by
formulation is the great distortion of the elements of the mate- Jaime et al. (2010) displayed a comparison of the cutting force
rial in front of the tool tip, such that, sometimes, the mesh has (horizontal) from FEM simulation and force in laboratory test.
to be regenerated. On the other hand, in the Eulerian approach The behaviour of the history force is similar with this study,
the elements are fixed in space and the material flows through where the element erosion was used in modelling fracture,
the mesh, consequently no mesh distortion occurs and no avoids the difficulty of no a priori knowledge of failure modes.
remeshing is required, Movahhedy (2000). Nevertheless, this The simulation results by Jaime et al. (2010) concluded that
approach is not suitable for modelling the unconstrained flow when using the element erosion, the simulated forces show
of material on free boundaries; the spatially fixed mesh means frequent drops to zero as the elements in front of the cutter
that an a priori assumption should be made about the shape of are deleted and there is a loss of contact between the damage
the chip and the contact conditions. material and the cutter.
Another type of formulation is the Arbitrary Lagrangian- In both 2D and 3D simulations, it was possible to obtain
Eulerian (ALE) method, which combines the advantages and good representation of the effect of confining pressure on the
avoids the shortcomings of both of previous methods. In this rock cutting process, using a material model in which stiffness
method, the FE mesh is neither attached to the material nor and strength increase with increasing confining pressure.
fixed in space, the mesh has a motion that independent of the Finally, this study allowed us to create models to repre-
material. sent reliably the mechanics of a single cutter experiment,
Jaime et al. (2010) observed that the ALE formulation of showing that the numerical simulation can be a valuable tool
the FEM presents some shortcomings similar to the Eulerian to evaluate various parameters without costly investments of
formulation. Due to the need of a mesh to encompass the mate- time and capital. Consequently the numerical simulation can
rial location at the current time, and a void region around it support understanding of the mechanics of the rock-cutting
to include the location of the material at a future instant in process and it can help to resolve many problems of reduced
time, the computational time is considerably larger than the drilling efficiency in different rock environments and diverse
time required of a Lagrangian solution. Therefore, our models conditions of drilling.
used only the Lagrangian method, avoiding the element diffu-
sion problem as that of Eulerian and ALE formulation when
chip were formed. REFERENCES
The results obtained with 2D and 3D models showed the
brittle failure behaviour when fine mesh was used. The cut- Dupriest, F.E. & Koederitz, W.L. 2005. Maximising drill rates with
ting force shows lower frequency variations in the coarse mesh Real-Time surveillance of Mechanical Specific Energy. Society
model than it does in the fine mesh model. The force increases of Petroleum Engineers Journal, paper No. 92194, present at
while the cracks develop and reduces when the element is SPE/IADC Drilling Conference, Amsterdam. The Netherlands.
removed from the mesh. Simulation of the rock cutting pro- Jaime, M.C., Gamwo, I.K., Lyons, D.K., & Lin, J.S. 2010. Finite
cess required the use of a model of contact with erosion. In element modelling of rock cutting. In 44th US Rock Mechanics
Symposium and 5th US-Canada Rock Mechanics Symposium, Salt
this type of contact model, the element is detached from the Lake City, Utah, USA.
mesh when it reaches a specified damage level. Following the Pessier, R.C. & Fear, M.J. 1992. Quantifying common drilling
removal of an element, the tool can remain without contact problems with Mechanical Specific Energy and a Bit-Specific
for a long time when a coarse mesh is used for the workpiece. coefficient of sliding friction. Society of Petroleum Engineers
Without contact, no force is required, and then in the force- Journal, No. 24584, presented at 67th SPE Annual Technical
displacement diagrams the value of force is zero. The force Conference and Exhibition. Washington, DC, USA.
748

The impact of structural-tectonic of the rock mass on the formation and

development of geo-deformation processes
E. Freiberg, E. Bellendir & V. Golitsyn

Vedeneev VNIIG, St. Petersburg, Russia
N. Ablyamitov
Odessa Port Plant, Odessa, Ukraine
E. Cherkez & E. Tchujko

Mechnikov Odessa National University, Odessa, Ukraine
G. Bich
Chernomorniiproject, Odessa, Ukraine
Keywords: rock slopes and foundations, rock mass, monitoring
1 INTRODUCTION and 3 at the first stage of landslide protection facilities at

Odessa shore.
In the construction of North Black sea coast slopes the Underground benchmark network was embedded into the
stratigraphic-genetic complexes of upper Sarmatian, Meiotic, lining of the tailrace tunnels 350–400 meters long. The adits
Pontiac, middle-upper Pliocene and Pleistocene periods have were worked in landslide and ended in the rock mass that
taken place. Genetically they are sea, lake and sea, alluvial- were not affected by sliding on the distance of 150–250 m
sea (deltaic), alluvial, old red sandstone and loess deposits. from plateau edge (fig. 2). The adits are connected in bed-
They are characterized by heterogeneity to variable degree of rock by drainage gallery, worked on the edge of Meiotic
lithological composition and by variability of physical and clays and Pontian limestones. The deformation of adits were
mechanical characteristics both of thickness and lateral. Dif- determined by measuring vertical and horizontal displacement
ferent weakness zones of structural-tectonic and lithogenous of benchmarks. A total of 18–20 measurement cycles were
nature developed in the rock masses composed with these performed in each adit for the period from April 1966 till
stratigraphic-genetic complexes. September 1992.
Lithogenous vertical heterogeneity and layering, tectonic As an example let’s examine vertical and horizontal dis-
discontinuity and modularity of rock mass structure determine placement of benchmarks in the adits No 2 and 3. One group of
space frame and dynamically complex structure of the strain benchmarks mainly raises, another part descends. So the adit
field. The impact of natural and anthropogenic factors results along the whole length can be divided into separate segments
in the fact that the most contrast changes of deformation sign (blocks of rock mass) with characteristic dimension about 60
and value happen on the border of structural-tectonic blocks meters that move up, tilt and make differentiated movements.
and within lithogenous weakness zones the deformations and The analysis of horizontal deformations distribution along
shifts are localized. adits shows that the distance between the benchmarks
Instrumental observations data of long-haul line struc- increases and the adits elongate. The biggest elongation of
tures behavior (e.g. rock mass with underground openings) the adits occurs within landslide build-up and on the junction
as well as long-term monitoring of drifts values and direc- of landslide body and bed-rock. The extension of the adits
tion within the thickness of rock mass with mine-surveying in bed-rocks amount to 230–300 mm. The average (over a
instrument are effective instruments for identification and long-standing observations) rate of benchmarks longitudinal
studying of rocks massif composed of tectonic blocks and displacements in the bed-rock amounts to 8–12 mm/year.
lateral variability of its structure. Such field observations Differentiated tectonic motions and alternation of modern
have been conducted in coastal mass of Odessa region and zones of tension and compression create the backgrounds for
North Black sea region including slopes on the territory of decrease of slope stability. It results in continuous landslide
Odessa city. The results of such observations help in con- displacements and slow deformations of Meiotic clay creep-
struction of landslide protection works along coastal section ing not only within bed-rocks but also within landslide slope.
Lonzheron – Arkadia. One can assume that thanks to differentiated displacements of
As an example let’s examine the nature of rock mass defor- small blocks not only in landslide slope but also in bed-rocks,
mation by the results of deformation of tailrace adits No 2 the zones of local plastic deformations are formed i.e. tectonic
749

processes form and maintain in constant active state structural meiotic deposits section are also caused, in our opinion, by
and geological basis of sliding processes. tectonic factor – compression and stretching of geospace in the
Another subject of examination is the territory of port impacted zones by tectonic breach system. It is worth noting
Yuzhny which has its facilities on both sides of Maliy Adzhalik that the dependence of engineering and geodynamic processes
estuary and of Odessa Port Plant which facilities with deep-sea on structural-tectonic peculiarities of rock mass is confirmed
terminals are situated on the right bank. Geological structures by the results of long-term instrumental observations of
of right bank sliding slope and of plant’s warehouse area where rocks deformation and displacement at the slopes of Odessa
observations have being conducted are similar to geological coast.
structure of landsliding slopes of Odessa coast.
Since 2000 Vedeneev National Research institute (Saint-
Petersburg, Russia) has been conducting observation of 2 CONCLUSIONS
hypogene deformations of slope rocks at western shore of
M. Adzhalik firth in four 42.0–43.0 m deep holes drilled at 1. The results of long-term geodesic monitoring of defor-
the +9,0 m bench. The holes have been driven up to primary mation of Odessa coast landslide protection works and
Meiotian deposits that were not affected by landslide dislo- underground conveyer gallery (port Yuzhniy) and of Port
cations and that are represented by alternate layers of clays, Yuzhny waterfront structures as well as the results of
sands and sandy loams. Several low-strength limestone inter- depth observations of the slope deformations of Odessa
layers have been discovered at the depths of 33.0 and 39.0 m. port plant warehouse area showed that current deforma-
The deformations were registered by mine-surveyor projection tions of Odessa coast slopes and of Adzhalik estuary coast
meter IIM-100 according to results of the survey of the bend of slopes (port Yuzhniy) are of the same nature defined by
longitudinal hole axes equipped with polyethylene pipe cas- combination and relationship of the following factors:
ings. From 2000 to 2009 authors of the project conducted a) rock masses include microblocks of structural-tectonic
eighteen series of instrumental monitoring. Their results were nature that form regularly oriented grid of geologic
used to define the displacement values of point projections environment nonuniformity with step from first hun-
of the holes longitudinal axes towards the firth and the sea. It dred meters to twenty – thirty meters that correlate with
has also been established that abrupt decrease of displacement buildings and structure sizes;
values is registered from the depth of 20.0–22.00 m, and this b) vertical differential displacement and tilting of the
depth can be considered the lower boundary of shear defor- microblocks, that happen with different frequency in
mations. Deformation accuracy for IIM-100 projection meter time and various range in magnitude;
is +0,0001 H, where “H” is measuring point depth. c) horizontal quasiperiodical compressions and stretch-
Data analysis of instrumental observations of in-depth rock ings within the boundaries of interblocks and lineary
deformations in holes revealed that the most substantial dis- localized zones weakened by the cracks;
placements of mark projections of the vertical axis of each d) rheological properties of the rocks of the main deformed
hole take place in Meiotian deposits at several levels: up to horizon;
depth 20–22 m (actual elevations −11 ∼−13 m) the displace- e) tectonic displacements of microblocks form the areas of
ment rate is 5–7 mm/year and at the depth 30–34 m (actual local plastic deformations i.e. tectonic movements gen-
elevations −21 ∼−25 M) the displacement rate is not more erate and continuously keep up rocks sliding in active
than 2–3 mm/year. phase;
Values of the point projections displacement of the longitu- f) the directions of shear deformations and shifts com-
dinal axis drillhole towards the firth and sea side was used to ply with main directions of weakness zone system of
form displacement trace.As an example let’s analyse the obser- tectonic nature (according to borehole monitoring).
vation results of well 1. With general South-East direction
some of the traces have step view as a result of displacement 2. Results analysis of instrumental observations of Malyi
direction change for 30◦–60◦–90◦. During some observation Adzhalik right bank slopes deformations allow us to mark
cycles the displacements have reciprocating character. out within the limit of rock mass separate blocks of
Comparison of the main directions of erosion cuts linea- rock which spatially-kinematic displacement characteris-
ments with characteristic directions of displacements of hole tics indicate their common, probably sliding, nature.
axis marks projections makes it possible to assume the impact 3. On the cross-section of Meiotic deposits of Malyi Adzhalik
of tectonically weakened zones on the displacements direc- right bank slopes it is possible to distinguish two struc-
tion. Deformation activity of different tectonic weakened zone tural levels of shear deformations at absolute elevations
systems and their regular “activation” or “occurrence”, change of −11 ∼ −13 m and −21 ∼ −25 m, on which ancient
of slide direction field in upper and lower parts range of landslides confined to other erosion levels could form.
750

Simulation of arenaceous weak rocks by means of cemented sands
V.R. Galván Liévano

Intertechne Consultores SA
M.A. Kanji
Keywords: lab testing, rock properties, weak rock
1 INTRODUCTION
The study of weak rocks is of increasing interest. In many

cases, due to lack of sufficient knowledge of the proper-
ties and behavior of weak rocks, their parameters can be
underestimated, as conservative assumptions are adopted.
Another indication of the importance of this subject is the
former existence of technical committees on weak rocks in
the geotechnical societies (ISRM, IAEG and ISSMGE).
The present research started with a thorough data collection
of published data along about three decades on the properties Figure 1. Plot of sand/cement mixtures in the γ d vs. n graph.
of weak rocks, which allowed the establishment of correla-
tions between the several properties and parameters (Kanji &
Galván Liévano, 1998; Galván Liévano, 2000). In parallel,
an overview of Southamerican weak rocks and correlation
between their properties were presented by Bosio & Kanji
(1998).
In the case of the research related to this paper, the main
interest was to investigate the possibilities of studying the
properties of arenaceous weak rocks by means of artifi-
cially manufactured materials built with mixtures of sand and
Portland cement in various proportions. If the properties of
such materials would match those of similar natural arena-
ceous rocks then their characteristics could be reproduced and Figure 2. Range of values of sand/cement mixtures of Porosity and
Water Absorption.
could be investigated under controlled conditions.
For this purpose, specimens of varied mixtures of sand and properties of natural weak rocks (Galván Liévano, 2000), as
cement were prepared and tested to determine their physical their results acquire more significance this way.
and mechanical properties Their preparation, testing, results
and comparison with natural rocks are presented in this paper.
2.1 Physical properties
The content of common Portlant cement type I varied in
10%, 20%, 50% and 80% of the mixture. The sand had a As shown in Figure 1, the plot of testings results of
minimum diameter of 0,3 mm. The mixtures were compacted sand/cement mixtures presented in the same Figure 1 range
similarly to the Proctor test (on cylinders of 5.4 cm in diame- from 40% to 50% in porosity (similar to tuffs) for the less
ter and 15 cm high) to result in the loose, medium and dense compact mixtures to about 20% for the more compact ones,
states (except for the 10% cement which did not had samples falling within the normal range of natural rocks but slightly
with medium state), in the attempt to simulate sandstones with lower. The Water Absorption (A) also has a theoretical rela-
different degrees of cementation. The specimens, were sub- tionship with the Porosity, shown in Figure 2, with a small
merged during at least 48 hs for setting and then removed from spread of results for natural weak rocks, also due to differ-
the moulds and maintained for 10 days in the humid chamber, ent procedures of saturation and measurement of the porosity.
after which they were oven dried to reach constant weight. The results for sand/cement mixtures also follow the theoret-
After trimmed and saw cut the specimens were tested follow- ical line very closely. Specimens subjected to vacuum show
ing the ISRM Suggested Methods referring to each type of test. higher absorption that those simply immersed in water for
short (1 hr) or longer periods (1 day or more).
2 RESULTS
2.2 Mechanical properties
The results of the different testings are presented in plots The testing results of Sonic Velocity determinations on the
superimposed to data and various previous correlations of the specimens showed very good correlation with the UCS, as
751

Figure 6. E50 and UCS for the mixtures of sand/cement and for
Figure 3. Relationship Vp × UCS for sand/cement mixtures. other natural weak rocks.
Figure 4. Schmidt Hardness with UCS for the sand/cement

mixtures. Figure 7. Static and Dynamic Moduli for sand/cement mixtures
compared to some correlations for natural rocks.
Figure 5. Dry Unit Weight × UCS for sand/cement mixtures.

Figure 8. Brazilian tests for sand/cement mixtures.
seen in Figure 3, while the data for natural weak rocks show
very wide dispersion. Schmidt Hardness index determinations the specimens according to the percentage of cement, shown
(H) for the sand/cement mixtures follow very well the trend in Figure 8. Several other correlations obtained with different
of natural weak rocks, but plot slightly above, as presented in determinations are not here presented but show the same type
Figure 4. of coincidence with natural weak rocks.
The relationship between Unconfined Compression The several tests on sand/cement mixtures with various
Strength UCS and Dry Unit Weight is not unique for the proportions showed results that compare very well to those
diverse types of rocks, showing a great variation according properties and characteristics of natural weak rocks. There-
to rock type and inherent strength, as shown in Figure 5, in fore, artificial materials as those tested in this study can
which each curve correspond to the same type of rock but reproduce and simulate the properties of natural arenaceous
under different condition of unit weight due to cementation rocks.
or weathering degrees. Typically the curves have an S shape
in the semi-log graph. The results of sand/cement mixtures
followed the same trend, in the lower range. REFERENCES
The E50 Deformation Modulus of the sand/cement mixtures
yield results of the same order as of the natural weak rocks, Bosio, J. J. & Kanji, M. A. (1998): Soft Rocks of the Rio de la Plata –
as shown in the Figure 6 relating E50 to UCS. S. Am., Proc. 2nd. Int. Symp. Hard Rocks and Soft Rocks, Naples,
Figure 7 presents the curve of the Static Modulus of Defor- V.1: 55–61.
Galván Liévano, V. R. (2000): Simulation of the geotechnical proper-
mation (Est) against the Dynamic Modulus (Edyn) calculated ties of the weak arenaceous rocks by means of artificial materials,
from the Vp sonic velocity and the Poisson’s ratio. The results Doctoral Thesis, Dept. Struct. & Geotech. Eng., Polytechnical
are mostly in the range of low values of Edyn, almost coin- Sch., Univ. S. Paulo, 302 pp.
cident with the existing published curves for natural weak Kanji M. A. & Galván Liévano, V. R. (1998): Correlation of Prop-
rocks. Finally, the curves of the Tension Strength as measured erties of Soft Rocks, Proc. 2nd. Int. Symp. Hard Soils and Soft
by the Brazilian test is correlated to the Dry Unit Weight of Rocks, Naples, V.1: 239–244.
752

3D geological documentation system based on digital photogrammetry
L.Q. Guo
College of Manufacturing Science and Engineering, Sichuan University, Chengdu, China
Northwest Institute of Nuclear Technology, Xi’an, China
J.B. Liao
College of Manufacturing Science and Engineering, Sichuan University, Chengdu, China
L.Q. Guo, J.J. Chen, X. Yu & J. Tao

Northwest Institute of Nuclear Technology, Xi’an, China
Keywords: rock mass; rock joints; numerical modeling; field measurements
1 INTRODUCTION The technique framework of this system consists of the

self-calibration techniques of binocular CCD cameras, the
The stability and permeability of underground rock engi- digital photogrammetry with the binocular system, the image
neering is related tightly to the distribution of rock mass encoding and pre-processing, the image feature’s detection
fractures. Using compass-clinometer and scratchpad to mea- and stereo matching, image mosaic, computation of 3D point
sure fracture orientation is a traditional method. First, it is cloud, 3D image model reconstruction based on OpenGL,
slow, difficult and dangerous. Second, time restriction leads to discontinuity orientation measurement from 3D metric image,
incomplete data as excavation proceeds or support is applied. lithology, physical and mechanical parameter management,
Third, the documentation data can not be verified after the spatial database management, and so on.
tunnel or rock cavity is covered by sprayed cement. Then,
how to acquire high precision rock mass discontinuity ori-
entation data rapidly and comprehensively is significant in 3 SOME KEY TECHNIQUES AND METHODS
rock engineering design, construction and stability appraise-
ment. Some scholars researched the indirect measurement of 3.1 Self-calibration of binocular CCD cameras
rock mass discontinuity surfaces by digital photogrammetry. Pollefeys et al. (1998) firstly proved that self-calibration and
The JointMetriX3D system is a rotating line-scanner imaging metric reconstruction is possible in the presence of varying
system, which is developed by 3G Software & Measurement and unknown intrinsic camera parameters by a certain num-
GmbH Austria. The rock mass discontinuity network could ber of image sequence with zero camera skew (s = 0).Agapito,
be measured from its metric 3D images. In this paper, the 3D Hartley and Hayman investigated the theoretical and practical
geological documentation system based on digital photogram- feasibility of self-calibration in the presence of rotating and
metry is a rotating CCD area scanner imaging system with two zooming cameras. In this paper, the self-calibration of binoc-
area scanning camera. This system can be freely placed in the ular CCD cameras is conducted by improving the Agapito’s
tunnel to finish digital photogrammetry with a binocular sys- method. First, the primary value of two CCD cameras’ intrin-
tem by the online self-calibration techniques. The acquired sic parameters and exterior parameters could be calibrated
images record the real fracture orientation of the tunnel or by Zhang’s method. Second, the stereo rig is rotated purely
rock cavity. The feature of rock crevice can be recognized and with fixed parameters except the camera focal length. Images
extracted by the method of digital image processing. 3D coor- are overlapped 60% between two adjacent rotating shootings.
dinates of spatial points can be calculated by the homonymous The calibration algorithm is based on the use of the infinite
feature points located in the left and right image pairs. 3D met- homography constraint which describes the mapping of the
ric image model can be reconstructed by the cloud of points image of the absolute conic between views. By this algorithm,
with 3D coordinates. By using the semi-automatic method, the self-calibration of rotating binocular CCD cameras with
the orientation of rock mass discontinuity can be measured varying focal lengths can be realized.
interactively from the 3D metric image model.
In this study, the system components and technique frame- 3.2 Image pre-processing
work are introduced first, and then some key techniques and
methods are discussed separately. Image pre-processing includes image encoding, distort cor-
rection and image strengthening. High pass filter, fractional
differential template or the gray level expansion can be used
2 SYSTEM COMPONENTS AND TECHNIQUE to strengthen the edge feature of the rock crevice.
FRAMEWORK
3.3 Feature detection of the rock crevice
The 3D geological documentation system consists of binoc-
ular CCD cameras, theodolite, auxiliary light source, tripod, The 3D geological documentation system requires only a small
battery and laptop. amount of spatial control points with known coordinates in
753

the tunnel to get a relationship between the images and the By the methods of vector data overlay display and spatial
tunnel. The edge and corner feature points of the rock mass entity multi-layer overlay modeling based on OpenGL, the sur-
discontinuity are applied in the self-calibration of rotating and face geological documentation data and internal 3D data could
zooming cameras, image stereo matching, 3D model recon- be displayed on the 3D reconstructed image model together. It
struction and discontinuity orientation measurement. There- also relates spatial objects to the attribute parameters and can
fore, to achieve rapid and accurate identification and location realize visual query analysis. The 3D geological documenta-
of rock crevice is the basis of various crevice measurements. tion system would be of high resolution, true color, with an
In laboratory, Xu’s method and Wang’s method were objective record of the tunnel surface geological information,
applied in rock crevice feature detection. The adaptive and provide users with a virtual documentation environment,
weighted median filter and the correlation test of adjacent and provide data support for the subsequent rock mechanics
pixels along the gray gradient direction were researched. Com- analysis.
bined with Sobel operator, this algorithm is able to improve
the accuracy of crevice edge detection. Not considering the
problem of the excavated pseudo-crevice edge and shadow 4 FIELD PHOTOGRAMMETRY SCHEME
noises, this algorithm needs further study and improvement.
Experimental result shows that in order to get higher corre- In order to improve image quality, jitter, light intensity and
sponding feature point matching ratio, it is necessary to select exposure time must be considered. The 3D geological docu-
right feature detection algorithm and detection zone by using mentation system can be mounted on a tripod to prevent from
the manual method. camera shake. By using the auxiliary light source installed on
the system, the working environment lighting conditions are
3.4 3D image model reconstruction improved. The relationship between the camera exposure time
and light intensity can be acquired through the field shooting
Firstly, the actual scanner poses are determinate right from the experiment.
observation of the ground control points by theodolite. Two The 3D geological documentation system’s field pho-
cameras’ intrinsic and exterior parameters could be calculated togrammetry scheme is as follows. Firstly, this system is
from the self-calibration of rotating and zooming cameras. placed in nearby the tunnel axis, and use theodolite to measure
Secondly, stereo matching could be calculated between the the tunnel control points to obtain the absolute spatial coordi-
left and right images. Mismatching points can be filtered by nates of the system cameras. Secondly, the primary intrinsic
the parallax mean filter algorithm. Thirdly, spatial points’ 3D parameters of two CCD cameras can be calibrated by Zhang’s
coordinates could be calculated from the corresponding fea- method when system aims at the tunnel wall. Thirdly, the other
ture points between views. Finally, the 3D image model could intrinsic parameters remain unchanged except the focal length,
be reconstructed by triangulation, mesh building and texture and the binocular CCD camera takes stereo image pairs after
mapping of the point cloud data. the two CCD cameras rotate the same angle. In accordance
with the order of the wall – the top – the other side wall,
3.5 Geological documentation element measurement the continuous scanning images overlap during rotating and
By using the semi-automatic method, the geological documen- zooming shooting. Then the system moves to the next posi-
tation elements of rock mass discontinuity can be measured tion to finish photogrammetry segment by segment. Finally,
from or be marked on the 3D metric reconstructed image the binocular CCD camera self-calibration, spatial coordinate
model or the cylinder expanded view images. calculation, 3D metric image model reconstruction and data
analysis can be carried out in the laboratory.
3.6 Geological documentation data management and
analysis
5 CONCLUSION
The system use ORACLE 9i to establish the spatial database.
During geological recording, the digital photographic image Without placing artificial marks on the target rock surface, 3D
pairs, the orthophoto maps, the mosaic maps, the rock geological documentation system based on digital photogram-
joints distribution images are stored into the image-graphic metry could be freely placed in the tunnel to realize online
database, and the camera calibration parameters of the intrin- self-calibration of rotating and zooming cameras and non-
sic and exterior orientations, the spatial coordinates of the contact measurement of rock mass discontinuity orientation.
control points and feature points, the interpretation infor- The geological documentation elements can also be measured
mation of the discontinuity orientation, the parameters of from the 3D metric image model. This system can improve the
the physical and mechanical properties are stored into the automation level of geological documentation data collection
properties database. and processing.
754

Uncertainty and sources of error in rock engineering
J. Hadjigeorgiou & J.P. Harrison

Lassonde Institute for Mining, University of Toronto, Toronto, Canada
Keywords: site characterization
1 INTRODUCTION and equipment then it can be relatively easy to quantify the

magnitude of error.
Since the first ISRM Congress in 1966 there have been sev- In well designed equipment, some of the easiest errors to
eral substantial developments in rock engineering focusing quantify are those stemming from the accuracy and precision
on new engineering tools for the analysis and design of struc- of the testing equipment. By calibrating the equipment, and
tures in rock. One early development was the introduction of noting the applicable ranges of the various sensors, it is possi-
rock mass classification systems which, when combined with ble to both quantify such errors and reduce them to acceptable
empirical rules, provides a framework that is still widely used levels.
for design. Another development, partially fuelled by increase A different type of error that seems particularly prevalent
in computer power, is the proliferation of numerical techniques is that arising from the use of inappropriate equipment and
that are routinely applied to the analysis, and visualization of methods. This is applicable to field investigations, laboratory
the results, of complex 2D and 3D problems. Unfortunately, testing and engineering analysis.
our understanding of uncertainty and sources of errors in rock
engineering does not seem to have been developed to the same
degree as these design procedures. Here, attention is drawn 1.3 Uncertainty
to the sources of error and uncertainty within these proce-
dures, and how these might detrimentally affect our designs Uncertainty is a term that tends to be employed quite loosely
and management of rock engineering projects. We conclude by in rock engineering. Baecher and Christian (2003) distin-
suggesting that rock engineering, particularly given the tools guish two types of uncertainty: that due to naturally vari-
and procedures currently available and under development, able phenomena in time or space; and that due to lack of
would benefit from a substantial effort to properly quantify knowledge or understanding. Although there seems to be no
and manage uncertainty and sources of error. consensus in rock engineering, elsewhere these are usually
referred to as aleatory and epistemic uncertainty, respectively
1.1 Sources of error and uncertainty (Kiureghian & Ditlevsen, 2009).
Uncertainty often results from the inherent variability of
In this paper the case is made that uncertainty and sources of rock masses, which in turn is a function of both geological
error are two different components of engineering analysis and materials and setting. In some cases, we implicitly assume
design, but understanding both is essential for rock engineer- that the uncertainty is aleatory and apply statistical methods
ing design. Furthermore, we show that the two components to sampling and analysis strategies.
are linked in certain circumstances, such that not accounting
for uncertainty can lead to substantial errors.
We introduce sources of error and uncertainty using exam- 2 IMPACT OF ANISOTROPY
ples from rock engineering, and use these to postulate that
there are areas of rock engineering analysis and design in The prevalence of rock mass classification systems in rock
which uncertainty and sources of error overlap. The practical engineering overshadows the fact that they are limited in their
implications of gaining an improved understanding of uncer- ability to fully capture the anisotropy that exists in fractured
tainty and sources of engineering error are the development rocks. It is unfortunate that the effects of this in terms of error
of appropriate strategies for resource allocation and require- and uncertainty are often overlooked.
ments. Thus, when considering limited human and financial A further limitation is the tendency to develop links between
resources, the impact of reducing uncertainty or sources of the different classification systems. This is driven by the
error needs to be quantified with respect to generating an need to provide quantitative input to specific empirical design
improved design. methods.
Finally, and separate from issues of linking two schemes,
the necessity to arrive at a unique index implies a simplifica-
1.2 Sources of error
tion. Consequently, all classification systems suffer from the
Sources of error for engineering purposes are often associated problem of false similarity (i.e., rock masses with different
with data collection and material testing methods. In cases engineering characteristics can have the same classification
where data collection uses specific standardized techniques value (Pine & Harrison, 2003)).
755

3 DATA COLLECTION METHODS objective. Clearly, the objective has to be well defined and
communicated: without these, there is the potential for error
3.1 Data collection methods in that an incorrect characterization may be specified.
Errors of convenience seem to stem from a lack of resources
Field sampling is liable to biased, particularly when the sam-
(e.g., time, funds, personnel), and include such practices as
pling of discontinuities is involved (Priest, 1993). Bias in this
combining limited data sets from geomechanically different
respect will include orientation, location and extent of dis-
domains into one data set of required size.
continuities. Much of these are a function of accessibility, as
Finally, we will have errors due to ignoring variability and
a human operator is required to physically obtain these data
uncertainty. The first case is when we know that properties
from a rock face.
vary with location but adopt only one value. In the second
Currently, much effort is being expended on the use of
case, we are simply uncertain about the value to assign to a
improved remote sampling technology in the form of dig-
property (e.g., an assessment of discontinuity surface rough-
ital photogrammetry, laser scanning, LiDAR and the like.
ness). These are distinctly different errors, but tend to have the
Although these technologies permit the collection of much
same effect in that only a single value, rather than a range or
greater amounts of data than manual methods, and often lead
distribution, is used. The consequence is that the conditions
to much higher spatial resolution, they nevertheless are sub-
are over simplified, as discussed above.
ject to similar problems of bias – particularly with respect to
Despite these limitations and potential errors, it is not
the orientation of the face being sampled.
unusual to proceed with an engineering design based on these
empirical methods. It is unfortunate that, despite having vari-
3.2 Measurement errors ous appropriate analytical tools, we do not attempt to quantify
These are arguably the best defined and easier to quantify. these errors. In fact, the continued application of these meth-
Use of both experimental protocols and traceability ensures ods can perhaps be attributed to the perception that ‘they work’
that measurement errors are minimized and quantified. more than the fact that any errors are of acceptably small
magnitude. In reality, this is further indication that empirical
3.3 Sampling bias errors techniques have an inherent, and often unknown, conservatism
that stifles improved and innovative design.
Human bias is demonstrated by the need to collect ‘good The results of a characterization campaign are often used to
quality’ samples from which to prepare laboratory specimens. provide the input for extensive numerical analysis in support
of design. It is interesting that there is no theoretical rela-
3.4 Inappropriate numerical tools tion between the results of any classification system and rock
Detailed models provide less uncertain (i.e., ‘better’) represen- mass constitutive and strength relations. Instead, the parame-
tation of the subject under investigation. However, increasing ters required for numerical analysis are obtained by empirical
the detail in a model requires more information, and this is correlation. Such correlations have the potential to be in error,
subject to the previously discussed errors and uncertainties. and these will compound the errors previously resulting from
Thus, what one gains by reducing model uncertainty, one may the characterization.
in turn lose by increased information error and uncertainty.
5 CONCLUSIONS
4 ENGINEERING DESIGN
Our understanding of uncertainty and sources of errors in rock
An example on how sources of error and uncertainty overlap engineering does not seem to have been developed to the same
can be observed in developing a strategy for design associ- degree as instrumentation, analysis and design procedures.
ated with an underground excavation. The first step would Nevertheless, we are still faced with the need to develop solu-
arguably be to identify the key features of the structure, such tions and provide results to our rock engineering problems.
as excavation shape, principal dimensions and location. These A major step forward would be to “. . . recognize the extent
parameters can often be determined early in the design process of our ignorance and to understand whence it arises. We can
with limited uncertainty or error. reduce uncertainty by obtaining more information, especially
The second step involves characterization of the rock mass when the search of more information is guided by rational
(i.e., determining the engineering characteristics of the mate- understanding of the nature of uncertainty and its impact on
rial), and this has to be defined in the context of the engineering our decisions” (Christian 2004).
756

Analysis of the infrastructure system and core factors for environment-friendly

urban regeneration
K.C. Han, D.W. Ryu & H.M. Kim

Korea Institute of Geosciences and Mineral Resources, Daejeon, South Korea
T.H. Kim
Posco Engineering & Construction Co. Ltd., Seoul, South Korea
Keywords: tunneling, stability analysis
These days, cities have dramatically expanded due to over- weighted values by core element technology for construction
population, the concentration and extension of urban func- of the infrastructure.
tions. As a result, multiple urban renewal projects have been Lately, Gyeonggi-do announced the Great Train eXpress
planned and executed to strengthen the efficiency of urban (GTX) plan which connects Gyeonggi-do with Seoul while
functions and improve living environment. Because of the the City of Seoul unveiled the U-smart Way project. In con-
constraints of urban ground space, there have been limi- sideration of current social and economic situations, these
tations in the improvement of urban functions through the urban underground space development projects are targeted
horizontal expansion of urban spaces. Therefore, it is nec- to overcome the following problems; increase in the ratio
essary to set underground express ring roads which can be of intercity buses and traffic congestion due to large hous-
highly applicable to urban development and improvement ing development in the outskirts of capital region, difficulties
as target infrastructure, examine the location conditions and in securing spaces and decent ground traffic infrastructure,
ripple effects of the development, collect necessary informa- huge land compensation cost and civil petitions by civil and
tion for conceptual design and estimate weighted values by environmental groups.
core element technology for construction of the infrastructure. The development of underground transport infrastructure
When underground transport infrastructure is constructed in is one solution to the problem of limited ground space. It
the downtown, there are many technical problems to be solved appears that the development project would generate various
such as interaction between underground facilities and the positive effects such as decrease in ground traffic congestion
existing facilities on the ground, tunnel stability, ventilation and preservation of current urban living environment. When
and environmental problems. Hence, this paper has inves- underground transport infrastructure is constructed down-
tigated the matters and technical problems that should be town, there are many technical problems to be solved such
considered during the construction of underground space and as interaction between underground facilities and the exist-
analyzed the technologies that are necessary in developing ing facilities on the ground, tunnel stability, ventilation and
urban spaces based on patent trend analysis in construction of environmental problems. Hence, this paper has investigated
underground spaces. the matters and technical problems that should be considered
There have been limitations in urban regeneration and during the construction of underground space.
expansion of urban functions with the horizontal expansion The prerequisite technologies for urban underground space
of urban spaces only. Therefore, it has been suggested to pro- development have targeted to investigate the current status of
mote urban development and improvement in 3-dimensional conventional technologies based on patent trend analysis, ana-
underground structure. In Korea, there have been efforts to lyze the technical directions of patent applicants and inventors
achieve qualitative urban growth and develop a pleasant urban and figure out the current status of core patents. On the con-
space by overcoming the limitations of ground spaces since trary, the patent trend analysis in this paper has dealt with
early 1990s. technologies associated with the construction of main tun-
Unlike ground facilities, it is difficult to expand or improve nel and shaft, tunnel ventilation and disaster prevention and
underground facilities once they are built. Hence, it is impor- measurement of groundwater for maintenance.
tant to come up with a reasonable and detailed development Figure 1 reveals the possible three GTX lines; KINTEX,
plan based on the precise forecast on the overall conditions of a Goyang-Dongtan New Town (74.8 km), Cheongyangri-
future city. For this, it is necessary to set underground express Songdo, Incheon (49.9 km) and Uijeongbu-Geumjeong,
ring roads which can be highly applicable to urban develop- Gunpo (49.3 km). Gyeonggi-do estimates that the GTX
ment and improvement as target infrastructure, examine the Project may need KRW 13.9 trillion if promoted with gov-
location conditions and ripple effects of the development, col- ernment money and KRW 11.1 trillion if promoted with
lect necessary information for conceptual design and estimate private funds.
757

Figure 2. Seoul’s U-smart way lines.
Figure 1. Gyeonggi-do’s GTX Lines.

– Early detection and reinforcement of composition ground
during excavation of underground structure.
The provincial government anticipates that GTX would save – Technology to secure stability in weak zone to prevent
energy by 450,000 TOE/year (KRW 584.6 billion) and reduce ground subsidence and minimize inflow of groundwater
CO2 emissions by 1,490,000 tons/year and CO2 treatment during excavation (cutoff technology).
costs by KRW 59.5 billion/year (Gyeonggi Urban Innovation – Technology to minimize blast vibration during tunneling in
Corporation, 2009). In addition, it’s forecasted that the new the central business district.
means of transportation would shift commuting patterns from – Maintenance system which can comprehensively evaluate
car to train and bring fundamental change in urban space. and manage the stability of current structures and new
Figure 2 shows Seoul’s U-smart way lines which include underground structures.
a total of 6 lines (149 km, 3 south-north lines, 3 east-west – Disaster prevention system which makes it possible to
lines). The three south-north lines are Siheung-Central City- detect fire and accidents early and handle them properly.
Eunpyeong (24.5 km), Yangjae-Hannam-Dobong (26.3 km) – Fire-resistant technology to keep underground structure
and Segok-Seongsu-Sanggye (22.8) while the three east-west safe from fire.
lines are Sangam-Central City-Jungrang (22.3 km), Sinwol- – Treatment of polluted air from portals and ventilation
Central City-Gangdong (22.3 km) and Gangseo-Seocho- shafts.
Bangi (30.5 km).
Seoul City expects that once the U-smart way is built, citi- In terms of the number of patent applications for under-
zens would be able to go to any place in the city within about 30 ground space development, the U.S. was the highest until
minutes. In addition, traffic flow would be greatly increased by 1986. Since 1987, Japan has taken the lead. In terms of the
reducing traffic volume by 21%. A total of KRW 11.2 trillion share of the patent to total patents, public patents in Japan were
(mostly private funds) will be invested in this project. the most dominant (83%, 6,964 cases), followed by published
An underground space is an important social infrastruc- patents in the U.S. (7%, 580 cases), public patents in Korea
ture which offers a new opportunity for environment-friendly (6%, 500 cases) and public patents in Europe (4%, 330 cases).
urban regeneration. Various problems (ex: traffic, environ- Since 2000, the number of patent applications has declined in
ment, culture, etc.) caused by the excessive expansion of a Japan. In Korea, it has been increasing since 1997. It appears
city have reduced quality of life. In addition, climate change that it is urgent to achieve customized technology develop-
has caused disasters in the city. As a result, social costs have ment and apply for related patents after analyzing major patent
increased every year. applications by technology in Korea to introduce the latest
The core technologies which are necessary for GTX technologies.
and U-smart way projects which are targeted to build In terms of patent share by technology, the number of patent
3-dimensional underground structure for urban regeneration applications for shaft downward excavation, prevention of
are as follows: environmental disasters and groundwater in maintenance was
high in Japan. In the U.S. and Europe, on the contrary, tun-
– Statistical techniques for qualitative evaluation of uncer- nel reinforcement was strong. Lastly, Korea has focused on
tainty of ground properties and technologies to evaluate groundwater in maintenance, measurement in maintenance
the stability of underground structure based on reliability. and tunnel reinforcement.
– Analysis on interactions with adjacent tunnels and ground
structures and behavior of underground structures in
composition ground.
758

Microwave assisted mechanical rock breaking
F. Hassani & P.M. Nekoovaght

Department of Mining and Materials Engineering, McGill University, Montreal, Canada
P. Radziszewski
Department of Mechanical Engineering, McGill University, Montreal, Canada
K.E. Waters
Department of Mining and Materials Engineering, McGill University, Montreal, Canada
Mining Engineering, Montreal, QC, Canada
Keywords: microwave, Lab testing, Mechanical excavation, rock breakage
1 INTRODUCTION
The wear on mechanical rock breaking equipment due to going

through hard rock is a significant issue in mining operations.
Therefore investigation of methods of reducing the energy
requirements when breaking the rocks is of vital importance.
The work presented here details the effect of microwave radi-
ation on the tensile strength of a number of rocks from around
the world after exposure to microwave radiation at a power
level of 2.45 GHz.
Microwave radiation has been investigated for a number
of years as a method of reducing the grinding requirements
in the comminution process. The mineral that the microwave
encounters is either transparent to the radiation, it reflects it or
it absorbs the energy, thereby heating up and expanding. This
induces fractures along the grain boundaries, ensuring that it
is easier to break the host ore body. Significant reductions
in the Bond Work Index of mineral ores have been reported. Figure 1. Norite disc shape specimen cracked in 30 s at 5 kW.
As yet this is not economically viable in a mineral processing
plant, due to the high energy costs. However, it may well prove
viable not only in the comminution process, but also directly
at the rock face, thereby reducing the wear on mechanical rock certain extent, being absorbed by components of the rocks.
breaking equipment. This increase was linear in nature. The disc shaped specimens
During this study, four types of rock were investigated: heated at a greater rate than the cylindrical ones, and melting
of the rock was observed.
• Basalt samples from California region (2.78 g/cm3 )
Throughout the investigation, the samples showed a
• Mafic norite from Sudbury complex basin (2.8 g/cm3 )
decrease in the tensile strength of the rocks with exposure to
• Basalt from China (2.89 g/cm3 )
microwave radiation, with the norite from Sudbury showing
• Granite from Vermont region (2.65 g/cm3 )
the greatest affinity for the microwave radiation. This is most
All rock samples were cored and their ends were cut and likely due to the high degree of sulphidic minerals that are in
polished according to the ISRM suggested techniques. Cylin- the norite matrix. It has been shown previously that many sul-
drical rock samples were 50 mm in diameter with a height phide minerals absorb microwave radiation to a high degree.
of 100 mm for the uni-axial tests and 25 mm for the tensile The effect of the microwave radiation is shown in Figure 1,
tests. In order to have a good comparative study all dry spec- Figure 2 and Figure 3.
imens were randomly chosen to be treated in three different Figure 1 shows the cracks formed in the disc shaped speci-
microwave power levels (1.2, 3 and 5 kW) for different expo- men after 30 s of exposure to microwave radiation at a power
sure times, ranging from zero to 120 s in a multi-mode cavity. level of 5 kW. It can be observed that a crack is formed along
All samples investigated observed linear increases in tem- the height of the disc. Figure 2 shows a crack propagated
perature, indicating that the microwave radiation was, to a along the entire length of the cylindrical section, after 120 s at
759

Figure 4. Power input of microwave vs. tensile value for basalt 2.
Figure 2. Norite being cracked at 120 sec in 5 kW. 5 kW, and Figure 3 shows the reduction in tensile strength as
a function of power level and exposure time.
By comparing the power level and the tensile strength value
in 10 and 20 s of exposure of Basalt samples, it can be observed
that in 10 seconds of exposure time, the tensile strength value
is reduced very little. At 20 s of exposure, the tensile strength
was reduced significantly as the power level increased (Fig. 4).
This work shows that the exposure to microwave radiation
can significantly reduce the tensile strength of rocks, which
would lead to reduced wear on breakage equipment. Sig-
nificantly, some samples melted after long exposure times,
therefore optimal exposure times and power levels must be
found for each rock that is being investigated.
Figure 3. Power input of microwave energy vs. tensile strength value

of norite.
760

Continuously variable frequency electrochemical characteristics of rock
J.C. Hu
State Key Laboratory for GeoMechanics and Deep Underground Engineering,
School of Civil & Architecture Engineering, Zhongyuan University of Technology, Zhengzhou, China
F. Wang
School of Materials & Chemical Engineering, Zhongyuan University of Technology, Zhengzhou, China
J.J. Cui
School of Civil & Architecture Engineering, Zhongyuan University of Technology, Zhengzhou, China
M.C. He
Keywords: rock properties; rock failure; lab testing; physical modelling
1 INTRODUCTION 3 EIS EQUIVALENT CIRCUIT OF ROCK
Electromagnetic properties of rock and mineral are important Based on the microstructure analysis of electrical signal pass-
in resources engineering, civil engineering and environmental ing rock, we know that when alternating current acts upon rock
engineering. The paper includes three parts as follows. specimen, there are different conductive paths, and called as
SCP, CCP, DCP, ICP.
The simplified microstructure model of conductive path of
2 EIS PHYSICAL MODEL OF ROCK rock is established.
Electrical model is drawn out base on the basis of physical
Based on the experimental device, the microstructure of model, such as figure 2.
electrical signal passing rock is established as figure 1. The total impedance of rock can be drawn from figure 2:
In which where the meaning of parameters is the same as

figure 2.
Then, the circuit parameters of rock are analyzed as
following.
1) As for rock surface, the attachment of conductive liquid and
migration path of conductive ion are affected by surface
Figure 1. Microstructure of electrical signal passing rock. Figure 2. Electric conduction model of rock.
761

roughness and connected pattern, etc. The impedance is
including of two parts.
where ρ1 is resistivity of rock surface attaching solution

in SCP; L1 is the shortest path migrated by ion on the
Figure 3. Electrical equivalent circuit of rock microstructure.
rock surface; S1 is vertical average cross sectional area in
the electrical field; ξ 1 is torsional coefficient of SCP; λ1
is percentage of the total concaves volume of rock speci- 4 AC IMPEDANCE SPECTROSCOPY EXPERIMENT
men surface that SCP concaves volume accounted for; ϕ1 OF ROCK
is concaves volume of rock specimen surface attached by
conductive liquid. According to the principle of AC impedance, testing appara-
2) Impedance statistical value of the entire continuous con- tus was designed for measuring rock impedance, carried out
ductive path in the rock specimen is as follows: impedance measurement of sandstone after different loading.
Electrochemical workstation of type CHI600A for
the experimental using, can provide electrical signal of
where ρ2 is resistivity of pore solution in CCP; L2 is 1 × 10−3 ∼1 × 108 Hz. Sine-wave electrical signal was taken
thickness of rock specimen in parallel with electric field as disturbance signal.
direction; S2 is vertical cross-section of electric field; ξ 2 is AC impedance spectroscopy of sandstone displays four
CCP torsional coefficient; λ2 is percentage of the total pore sections of arc, to illustrate the four relaxation processes
volume of rock specimen surface that CCP pore volume of rock electrochemical process which correspond to four
accounted for; ϕ2 is the porosity of rock specimen. time constants. They denote respectively characteristics of
3) Impedance statistical value of the discontinuous conductive inductive reactance, capacitive reactance and impedance of
path in the rock specimen is as follows: rock under the action of continuous frequency current, where
high frequency range is inductive reactance arc, represented
surface characteristics of rock, capacitive reactance arc of
medium and low frequency range represented internal surface
where ZCP is impedance of connected part on DCP; ZDP is characteristics of rock.
parallel plate capacitive formed by unconnected part. The results show that radius of capacitive reactance
arc of medium frequency range is in sequence of
S3D < S20 < SL < SW, which has close relationship with
where ω is AC frequency; CDP is parallel plate capacitance rock porosity. Under above-mentioned load, all the internal
of discontinuous pores. cracks of sandstone began to extend, it is showed on the figure
of Nyquist that radius of capacitive reactance arc of medium
frequency range are augmented. Within the same frequency
range, the variation of impedance real part of high frequency
where d is equivalent cumulative thickness of all the dis- arc on the start-stop segment represents connectivity degree
continuous pores in parallel with electric field direction; of rock interstices (connectivity degree of conductive liquids
ε0 is vacuum dielectric constant, 8.85 × 10−14 F/cm; εr is in the interstices). The values of R2 -R1 of S202, S3D2, SL2
relative dielectric constant of rock specimen. Others are the and SW2 are in sequence of 0.38, 0.65, 0.86 and 0.57 at
same as above. water-saturated state, which reflects changes of connectivity
degree of sandstone interstice after loading. (The S202 curve
was AC impedance spectroscopy of water-saturated sandstone
All the above results give us as following: under loading of 20 h at pressure of 40.8 MPa, the SL2 curve
was AC impedance spectroscopy of water-saturated sand-
stone under loading and unloading condition at pressure of
4) Impedance of insulated part 40.8 MPa, the S3D2 curve was AC impedance spectroscopy
of water-saturated sandstone with quick damage at pressure of
60 MPa, the SW2 curve was AC impedance spectroscopy of
where Cmat = Sε0 εr /L. water-saturated sandstone with no damage.)
When RCP and RCCP have a big value, that is to say, when rock
pores is empty, the resistance of solid parts of rock should be ACKNOWLEDGEMENTS
calculated.
According to the above discussions, we should consider The authors would like to acknowledge the financial sup-
diffusion reaction caused by the changes of double-layer port of the National Natural Science Foundation of China
capacitance and interstitial solution concentration of rock (51074196), and the State Key Laboratory for GeoMechan-
internal interface. The equivalent circuit model of whole rock ics and Deep Underground Engineering, China University of
is established as showing figure 3. Mining & Technology (SKLGDUEK0917).
762

Workflow design and software customization to enable integration of lidar data with active
drill-and-blast tunnel construction projects
Matthew J. Lato, Malte Vöge, Erling Dale & Elin Morgan

Norwegian Geotechnical Institute, Oslo, Norway
Subject: Information systems, artificial intelligence and other advanced techniques
Keywords: Tunnelling, rock caverns, rock mass classification, field measurements
Lidar imaging is a powerful technology that has recently been resources over the past three years to develop an operational
adopted by geotechnical engineers as an operational tool. The workflow, as illustrated in Figure 1. The workflow is designed
remotely collected data has the ability to enable geotechni- to enable accurate and rapid processing, including feature
cal engineers to perform advanced geometric, mathematic, extraction and interpretation of the lidar data. Through the
and spatial analyses without direct exposure to the rockmass. development of in-house macros and processing standards, the
Often working environments, such as drill and blast tunneling, use of lidar data has become efficient, reliable, and extremely
require immediate results and interpretation while the time robust. This paper outlines the approach taken by NGI to
allotted for on-site data collection is limited by a demanding develop the workflow as well as demonstrate numerous tools
production/advancement schedule. The Norwegian Geotech- currently used on-site in active drill and blast environments.
nical Institute has invested a significant amount of time and NGI has adopted the software suite PolyWorks as the prin-
ciple data processing option for all lidar data. The selection
of PolyWorks was made for two distinct reasons. The first is
its ability to manage and manipulate raw lidar data through
advanced computational algorithms. Secondly, and the focus
of this paper, is the ability to customize the software, create
macros, and expand the capability of the modules within the
PolyWorks framework.
The workflow for processing lidar data, in place at NGI,
is optimized for efficient and reliable results. The steps taken
to process the data (Figure 1) involve the user to combine
datasets, align the data with scan images collected from the
previous blast round, convert the lidar points to surface mod-
els, and extract geotechnically relevant information from the
surface models. The development of the surface model from
the lidar points is a critical process that is computationally
demanding. Through extensive sensitivity testing of different
surface input parameters, optimal values have been deter-
mined and set as user defaults. The setting of input and surface
meshing variables reduces the potential for user error and
variability between individuals processing lidar data.
Figure 1. Lidar data processing workflow developed at, and in This workflow, and the use of internally established param-
place, at NGI. eters, prepares the data for detailed and accurate geotechnical
Figure 2. Process of identification through digitization of a rock joint.
763

Figure 3. Shotcrete thickness evaluation by combining data from sequential blast rounds.
analysis. Upon completion of the conversion from raw lidar blasting, and prior to the application of shotcrete. This struc-
points to a true 3-dimensional surface model, the IMInspect tural evaluation (Figure 2), as well as assessment of shotcrete
module within PolyWorks is used to analyze the model and thickness (Figure 3), shotcrete tolerances, and volume of
extract pertinent features. blasted rock is calculated by the user through semi-automated
Construction of a drill-and-blast tunnelling requires a geo- processes.
logical investigation of the face and back of the tunnel post
764

Parametric sensitivity investigation: Analysis of wellbore stability
G. Li & M. Bai
Halliburton
Keywords: stability analysis, rock stress, numerical modelling
Wellbore-stability problems are frequently encountered when were performed to determine the relative effects of various
drilling deviated wellbores near salt or in tectonically active factors in maintaining wellbore stability. This study focuses
regions. When these unplanned geomechanical issues go on the most-sensitive parameters to develop a better under-
unresolved, it can result in wellbore collapse, lost circulation, standing of the controlling parameters in determination of the
stuck pipe, and casing collapse. The quality of wellbore- wellbore stability. Consequently, the less-sensitive parame-
stability predictions in drilling operations strongly depends ters were filtered out to minimize both the requirement for
on the quality of input parameters. The Quality Control (QC) necessary data entry and computational time.
of input parameters described in this paper includes perform- Case studies using the proposed methods are provided to
ing a comprehensive sensitivity study to identify the key demonstrate the workflow of the sensitivity investigation.
parameters that dictate wellbore stability and eliminating the After providing each input parameter with the designated func-
less-sensitive parameters that are insignificant to wellbore sta- tional distribution (e.g., normal or log normal distributions),
bility. Case studies using the proposed methods are provided the Monte Carlo simulation was launched. The probabilities
to demonstrate the workflow of the sensitivity investigation. of a stable drilling process were computed from the model-
This study explores many factors that affect the stability ing, using the shear-failure gradient as a lower bound and the
of a wellbore, such as the in-situ state of stress, rock proper- fracture gradient as an upper bound to guide the selection
ties from core and logs, and the failure criteria selected. In this of the desired mud-weight windows. The advantage of using
study, methods for determining various input factors using the the Monte Carlo simulation rests on the freedom to select
Wellbore Risk-Analysis (WRA) method with the wellbore- the uncertainty range for each parameter and to indicate the
stability probability assessment by means of Monte Carlo confidence level of each input parameter. Consequently, the
simulations to understand the influence of various wellbore- determined results can be more sensitive and realistic than
stability factors were analyzed first. The sensitivity analyses those calculated from a deterministic model.
765

Study and application of visualized model test technologies

for gas storage cluster in salt rock
L.J. Cheng, Z.K. Li & B. Xu

State Key Laboratory of Hydro-science and Engineering, Tsinghua University, Beijing, China
Keywords: lab testing; monitoring; physical modeling; rock caverns
1 INTRODUCTION Concealed excavation is difficult and uneconomical. Dissolv-

ing the model material to form the cavities is practicable no
The main purposes of modeling test for gas storage cluster other than the model material can be dissolved by some sol-
in salt rock are to study the interaction of the caverns and vent. While rosin can be dissolved by alcohol, the shape of the
research the chain destruction mechanism. Comparing with cavity is not easy to control. The difficulty of pre-embedding
the ordinary geo-mechanic physical model test, this test has the cavity mold is how to take out the mold without exerting
the following difficulties: the cavern must be sealed enough influence on the air impermeability of the cavities. Fortunately,
for gas storage, the destruction process of the caverns must melting rosin has viscosity which makes pre-embedding the
be visual and as the prototype material, rock salt can self-heal gasbag mold easier.
and has rheology property. The mold for physical model with single cavity is showed as
The way of modeling gas storage cluster in ordinary gran- Figure 2. The physical model with single cavity is the elements
ular model material, is usually putting an inner gasbag inside of the model with several cavities; they are showed in Figure 3.
the material. In this case, the process of caverns destruction is
not visible from outside, and one must cut the model separated 3 LOAD CONTROL SYSTEM
after the test for investigation. But this is normally not suc-
cessful. The assumption is, if there was an air impermeable The simulation of field stress is very important in a geo-
material which is transparent, and it is easy for gas storage mechanic physical model test. Besides, the gas pressure must
modeling and destruction process is easy to be observed. We be simulated inside the storage physical model too. The main
find that rosin is transparent solid at normal temperature while component of field stress is gravity stress in salt rock, the
it is able to self-heal under the influence of alcohol.The authors distribution of field stress is similar to lenitic stress as the
have done a mass of experiments with rosin and confected a gas storage is deep embedded. The crustal stress can be sim-
transparent model material mixed by rosin, paraffin and some ulated by gas stress and lenitic stress. Gas stress is provided
other additives, which is air impermeability, self-healing and by external gas resource.
rheologic. The details about how to make this material and The load control system consists of three parts: sealed
the mechanic properties of this material will be described in cuboids container, gas stress providing and controlling
another paper. In this paper, the main techniques about this facilities and water. The sealed cuboids container is made of
visual destruction process test will be presented detailed. The
illustration of the whole test is presented as Figure 1.
2 TECHNIQUES TO CONSTRUCT THE MODEL
As was stated above, rosin is transparent, yellowish or yellow-

ish red solid at normal temperature, but it changes into liquid
when the temperature exceeds its softening point of 70–90◦ .
This property makes easy for constructing the physical model.
The hardening of the material from liquid to solid is somewhat
slow; therefore, time is enough to construct the solid model.
There are three methods to form the cavities: concealed Figure 2. Mold for the physical model with single cavity.
excavation, dissolving or pre-embedding the cavity mold.
Figure 1. Illustration of the concept of visual physical model test. Figure 3. Models with single cavity and several cavities.
766

Figure 6. Illustration of destruction process monitoring system.
Figure 4. Illustration of the load control system.
Figure 7. Actual image of the visible model test.
In this experiment proposal, not only the measurement of

Figure 5. Illustration of displacement measurement principle. model internal deformation is to achieved, but also the moni-
toring of the destruction process of model cavity is required.
Each of the three high speed video cameras which fixed in
steel and with glass window on each side. Angle iron is used
3 orthogonal directions in this experiment can reach a speed
to reinforce the side boards. The top board connects with the
of 200 figures per second, which can fulfill the requirement
side boards by bolds and flange. In this way, the top board can
of destruction process recording. By adopting high speed
be taken off when it is necessary. Rubber mat is used as water-
camera, both the recording of destruction process and acqui-
proofer of flange. The illustration of load control system is
sition of deformation process of monitoring points via images
presented as Figure 4.
contrastive analysis can be achieved. In order to attain high-
definition images, a LED accent light are placed opposite each
4 THE MEASUREMENT AND OBSERVATION SYSTEM set of high speed camera, shown as in Figure 6.
Digital Speckle Correlation Method (DSCM) is a deforma-

tion measurement and analysis method, in which target digital 5 CONCLUSIONS
images are acquired by utilizing digital image acquisition sys-
tem and processed by image processing technology. At present In this article, the main techniques and theories of visualized
the application of this method commonly concentrates in the model test for gas storage cluster in rock salt is introduced
field of surface deformation and damage measurement, how- in detail. Techniques for model construction and load control
ever measurement of internal deformation and damage of and destruction observation system have been demonstrated
3-D model via this technique is rarely reported. Theoretically, by several primary tests. The whole appearance of the model
measurement method of 3-D model’s internal deformation and frame is showed in Figure 7. The work of the test mentioned in
destruction is similar with surface’s 3-D deformation measure- this article is under progressing and improving in the moment,
ment methods.The principle of 3-D deformation measurement and the method of deformation measurement has only been
by two collectors (such as cameras) is showed in Figure 5. demonstrated theoretically here, the next target is to utilize the
Speckle images utilized in DSCM are digital speckle fields. deformation measurement system in practice and to make the
To monitor the surface displacement, artificial speckles can test completed.
be created. For internal deformation monitoring, it is diffi-
cult to create 3-D continuous artificially speckle field, and ACKNOWLEDGEMENTS
for 3-D imaging, all points are projected onto the same plane,
there will be mutual interference among points in the depth The authors acknowledge the support of the National key
direction and increase of difficulty of image recognition. In basic research effort program (973) of China (Grant No.
this article, DSCM and the design thought of multipoint dis- 2009CB724604).
placement meter method are combined. In model manufacture
process, small size marker points are embedded in the internal
key monitoring area of the model. The displacement process REFERENCES
of each marker point can be obtained by two collectors, and
the arrangement of the marker points is similar to that of L.J. Cheng, PhD Candidate, chenglj068@gmail.com, 086-010-
multipoint meter. 62791425, 086+13699231535.
767

Real-time microseismic navigation for CBM directional drilling
X. Luo
CSIRO Earth Science and Resource Engineering, Australia
Subject: Others – Directional drilling
Keywords: monitoring, directional drilling, navigation, passive seismic
Directional drilling of a CBM hole can be difficult, risky and An increase of seismic amplitude with the approach of the
costly. One of the key issues in guaranteeing the success of drill bit to the borehole geophones was evident. Good seismic
the drilling is to ensure the drill bit is steered in the correct signal to noise ratio was observed when the drill bit was within
direction. Using drilling noise to locate the seismic source 150 m from the geophones (Figure 2).
(drill bit) is proposed (Figure 1) and trialed in this study. The results have shown that the seismic signals generated
This field trial was carried out at a CBM site in Bowen from drilling inside a coal seam can be efficiently observed
Basin, Australia, where directional drilling in the 5 m thick using triaxial geophones when the drill bit was within 150 m
Pleiades seam (P-seam) at 340 m deep was conducted to from the geophones. Consistent azimuth values of the drill
extract coal gas for commercial use. The distance from the bit were determined when the bit was within 120 m from the
drill rig to the target well is 1,344 m. geophones. Through a combination of the drill bit azimuth
and the drill bit distance from the drill rig, the trajectory of
the drill bit can be determined in real-time.
Figure 1. Passive seismics for CBM drill bit navigation. (A): A

vertical cross section showing drill bit in a coal seam and the locations
of geophones in the production well that is to be intersected by the
drill bit. (B): Planview of the drill bit location in reference to the
geophones. The bit location can be determined in real-time through Figure 2. The changes of the seismic amplitudes of S1 and S2
analysis of the azimuth. associated with the drilling progress.
768

On the minimum size of Representative Volume Element:

An n-point probability approach
D. Łydżba & A. Różański

Institute of Geotechnics and Hydrotechnics, Wrocław University of Technology, Wrocław, Poland
Keywords: rock properties, physical modeling
1 INTRODUCTION
A random heterogeneous material is a realization of a specific

random or stochastic process. More precisely, a realization
is an event, ω, that belongs to a sample space, . Further-
more, a collection of all the possible realization of a random
medium generated by a specific random/stochastic process is
an ensemble. The overall property/effective property of ran- Figure 1. Graphical illustration of the notion of a sample.
dom composite medium is defined as an ensemble average of
being a number of pixels in a row of RVE digital image. The
local property, i.e.:
ensemble average is then estimated as:
where η(x; ω) is a local random medium property, being a where ηj is a structural function corresponding to the
function of a space position and a realization ω, p(ω) is a j-realization of RVE and n is the size of the sample.
probability density function. The crucial role in this approach plays a proper determina-
The definition (1) implies a necessity of generation of tion of the RVE size, NRVE , as well as the size, n, of the sample.
all realizations forming ensemble in order to determine the The both values are related to each other, i.e. larger size of RVE
ensemble average. This exceptionally complex procedure is then a smaller sample size n can be used or inversely smaller
usually overcome using an ergodic hypothesis which allows size of RVE requires necessity of larger size of the sample use.
replacing the ensemble average with a volume average, provid-
ing that a volume of the medium considered tends to infinity,
i.e.: 2 MICROSTRUCTURE DESCRIPTORS
Consider an M -phase random medium. The total volume of

V is partitioned into M -disjoint random sets or phases. Let
Vi (ω) denotes region occupied by a phase i. Introduce now a
structure function η(x; ω), for phase i, such that:
The relation (2) gives the possibility of considering only
one arbitrary realization providing that the sample volume
is infinite. The term a sample with infinite volume means, in
an engineering application, a sufficiently large sample. This
sufficiently large sample is called as Representative Volume The expectation or probability of finding phase i at a chosen
Element (RVE). point, x, is then evaluated as:
A lot of attempt has been made in order to quantify the
RVE size on the basis of statistical and numerical analysis.
Within all the methods proposed the RVE size is usually deter-
mined by investigating the convergence of apparent property,
with increasing the size of RVE. It causes that the process of The above function is referred to as a 1-point probability func-
RVE size evaluation requires a very large number of numerical tion for phase, i, since it gives the probability to find phase i at
calculations, e.g. the finite element method analysis. position x. Similarly, the n-point probability function denotes
In this paper a methodology developed and presented in the probability that n points at positions x1 , x2 , . . . , xn are
a series of papers (Różański et.al. 2009, Różański 2010) is simultaneously found in phase i. According to its definition it
adopted for an estimation of the ensemble average (1) or (2). is defined as:
The sample is here considered as a set of a finite number n of
RVE elements (Fig. 1), each having the same finite size, NRVE ,
769

Hereafter, the primary attention is limited to the 2-point
probability function, since a determination of the minimum
RVE size used in this paper is based only on this microstruc-
ture descriptor. One can interpret the 2-point probability
function as a measure of morphology of phase distribution
within composite medium, i.e. the information of how the end
points of line segment of length r are correlated within the
microstructure.
3 MINIMUM RVE AND SAMPLE SIZE
The sample is said to be representative with respect to

microstructure geometry of composite medium if it assures
a satisfactory replica of the 2-point probability function of
composite.
The condition for the minimum RVE size developed in the
paper can be presented as: Figure 2. Original 2-point probability function and its replica:
Fontainebleau sandstone.
for Fontainebleau sandstone. It has been estimated that

Nmin = 133 pixels. The sample size has been determined to
where ε is an assumed error tolerance, lc is so-called correla- be n = 512.
tion length and the numbers N1 (ε) and N2 (ε) are such that the The parameters determined, i.e. the minimum RVE size and
following inequalities hold true, i.e.: the sample size has been then used to create the sample. This
sample has been then used to evaluate the Monte Carlo esti-
mator of the 2-point probability function of the Fontainebleau
sandstone. The estimator obtained is almost perfect replica of
the original 2-point probability function
REFERENCES
Różański A., Łydżba D., Shao J.F. 2009: Numerical determina-

The variables ξ and ψ denote volume fractions of composite tion of minimum size of RVE for random composite materials:
constituents contained within RVE. two-point probability approach, Proceedings of the First Interna-
The condition for the sample size is formulated by applica- tional Symposium on Computational Geomechanics, COMGEO
tion of the Central Limit Theorem. The final form is presented I, Juan les Pins, France, 29 April – 1 May, 2009/eds S. Pietruszczak
below: et al., Rhodes: IC2E International Centre for Computational
Engineering, 2009, pp. 26–37.
Różański A. 2010, Random composites: representativity, mini-
mum RVE size, effective transport properties, PhD dissertation,
Universite Lille 1, LML (UMR CNRS 8107), No.: 40444.
where (∗) is the cumulative distribution function of standard

normal random variable and the error tolerance ε represents
the relative error.
4 THE SIZE OF RVE FOR THE FONTAINEBLEAU

SANDSTONE
The methodology developed has been used for evaluation of

RVE size, based only on a digital image of microstructure,
770

Estimation of rock mass deformation modulus based on GSI system
A. Majdi, A. Bashari & M. Beiki

School of Mining Engineering, University College of Engineering, University of Tehran, Tehran, Iran
Keywords: deformation modulus, rock mass properties, plate loading test; empirical equation. GSI-System
Rapid growth of engineering rock mechanics and rock engi- The source of data used in this paper were obtained from
neering makes the determination of rock mass properties as six different Iranian dam sites and powerhouse locations,
one of the most important requirements in design and con- namely, Bakhteyari, Gotvand, Khersan-3, Chamshir, Parsian
struction of rock engineering projects such as underground and Siahbishe.
excavations, tunnels, dams, etc. Among the rock mass prop- The data used in this study include; the deformation modu-
erties, modulus of deformation is of paramount practical and lus of rock masses measured by plate loading tests, laboratory
economical importance. measured modulus of Elasticity (Ei), Uniaxial Compressive
Though, the in-situ determination of rock mass deforma- Strength (UCS), along with the information obtained during
tion modulus is of vital importance, however, the complex rock core recovery such as; Rock Quality Designation (RQD),
and time consuming procedures makes the method uneco- ground water condition, and all the other parameters required
nomic and unfavorable for some projects. Hence, in the recent to determine GSI and RMR indices. A summary of the
years, predictive models have been employed for the indirect employed data is given in Table 1. Finally, a database includ-
estimation of the deformation modulus. Nonetheless, the pre- ing a total of 150 cases was acquired for further statistical
dictive models results may not represent the realistic values analysis.
of deformation modulus as expected. The frequency histograms of GSI and Em to determine the
The empirical equations for indirect estimation of the defor- number of data are given in Figure 1. The plots help us to
mation modulus are simple; hence, they are unable to include signify density of samples in each parameter. As it can be
some uncertainties with regards to the limited data availability, seen, distributions of the data include a wide range of the
variability of rock type and complex heterogeneous nature of parameters. In other words, the GSI parameter ranged between
the rock masses. Hence, comprehensive investigations on the 34 and 76 (Figure 1a) and modulus of deformation is ranged
integrity of the results obtained by these empirical equations from 1.0 GPa to 40 GPa in which samples less than 20 GPa are
are crucial for rock engineering projects. more grouped than other intervals.
Consequently, several researchers have proposed some A plot of deformation modulus, measured by plate loading
empirical equations to estimate deformation modulus of test, against GSI is shown in Figure 2. As it can be seen,
isotropic rock mass on the basis of classification schemes there is a good and reasonable agreement between these two
such as the Rock Mass Rating, RMR, the Tunneling Quality parameters. The Root Mean Square Error (RMSE) index was
Index, Q, and the Geological Strength Index, GS. also calculated in order to check and compare the performance
In this paper, at first some statistical evaluations of the capability of predictive models developed in this study.
gathered information including plate-loading test data and The Comparative results obtained based on prediction per-
the corresponding rock mass properties have been presented. formances of the proposed models with the existing empirical
Then, two new empirical equations, based on simple and equations leads to the following concluding remarks;
multiple regression analyses, are proposed. Finally, a compar- Considering the RMSE values, the two equations proposed
ison of prediction performance between the existing empirical in this paper can be used as highly reliable predictive models
equations and the new proposed equations is made. as compared with the existing empirical equations.
Table 1. Results of basic statistical evaluations of some parameters.
Parameters RQD WD1 GSI Em (GPa) UCS (MPa) Ei (GPa) RMR89
Statistics
No. of data Valid 130 95 139 150 150 150 101
(PLT2 ) Missing 20 55 11 0 0 0 49
Mean 58.2 2.1 53.3 9.9 89.5 46.2 58.9
Median 60.0 2.0 51.0 7.9 100.0 60.0 56.0
Std. Deviation 21.5 0.7 11.2 8.0 48.6 27.2 11.6
Variance 463.9 0.5 125.0 64.3 2360.6 741.1 135.5
Minimum 0.0 1.0 34.0 1.0 10.0 2.5 39.0
Maximum 98.0 4.0 76.0 39.0 180.0 85.0 81.0
1
Weathering Degree
2
Plate Load Test
771

Figure 1. Histograms of the (a) GSI parameter and (b) rock mass deformation modulus obtained by in-situ plate loading tests.
Figure 2. Measured rock mass deformation modulus by plate

loading test versus GSI parameter.
772

New physical modeling of flexural toppling failure with titling machine
A. Majdi, Ö. Aydan & M. Amini

School of Mining Engineering, University College of Engineering, University of Tehran, Tehran, Iran
School of Marine Civil Engineering, Tokai University, Shimizu, Japan
Department of Mining Engineering, Faculty of Engineering, University of Kashan, Kashan, Iran
Keywords: rock slope, flexural toppling failure, tilting method, theoretical approaches
Toppling failure is one of the most common hazardous insta-

bility of rock slopes that engineers and researchers are facing
with. Flexural toppling failure is one of the specific modes
of toppling failure which occurs due to bending stresses. In
the research, some experimental models were carried out on
several single rock columns and rock slopes with potential of
flexural toppling failure by tilting machine. Columns used in
preparing physical models of layered rock slopes were created
through the consolidation of a special powder mixture under a
given consolidation pressure. Frictional angles, unit weights,
tensile and compressive strengths of these samples have been
measured in the laboratory and the results are summarized in
Table 1.
Modeling of single columns and controlling, measuring
and predicting the corresponding parameters before and dur-
ing the tests are very common. Hence, before the modeling
of the rock mass, single columns with potential of flexural
toppling failure were modeled. In order to construct single
column models, some samples with effective length of 5 to 30
centimeters were built at 600 KPa consolidation pressure. The
ends of each sample were fixed on a horizontal plate of tilting
machine by a small screw jack (Figure 1).
Then the monitoring equipments such as laser transducers,
acoustic emission sensors and accelerometers were setup both
on the machine and on the samples. Next, the inclination of
tilting machine plate was gradually and slowly increased till Figure 1. Modeling of single column with potential of flexural
failure occurs. During the test, tilting angle of the machine, toppling failure with tilting machine before and after failure.
displacement of the sample and crack initiation were mea-
sured by digital accelerometer, laser transducer and acoustic
emission sensors, respectively. In Figure 2, the results of these
tests are compared with the theoretical ones. As it can be seen
from the figure, the results of experimental models are in a
good accord with the theoretical ones.
To construct the required representative physical models of
the flexural toppling failure, some continuous columns with
effective length of 5 to 42 centimeters were built at 600 KPa
consolidation pressure. Then approximately isosceles trian-
gle shape models were built with the samples in a special
frame. The ends of the models were fixed firmly on the frame
Table 1. Engineering properties of laboratorial solid materials

with 600 KPa consolidation pressure.
Consolidation Tensile Fictional

pressure UCS strength Density angle
Parameters (KPa) (KPa) (KPa) KN/m3 Degree
Values 600 50–110 15–35 17–22.3 35–40 Figure 2. Comparison between theoretical and experimental results
for a single column.
773

Figure 3. Physical modeling of flexural toppling failure in rock slopes with tilting machine before and after failure.
by a small screw jack. Next the frame was fixed on the tilt-
ing machine and then the monitoring equipments were setup.
Finally, the angle of main plate of the tilting machine was
increased gently and slowly till failure occurs or the angle
of tilting reaches at 45 degrees (Figure 2). In the process of
this modeling the dimensions of the models are of paramount
importance. If the dimensions are too small, then the failure
will not occur. On the other hand, as was mentioned earlier
in this paper, the samples are so brittle, hence, construction
and handling of large samples are so difficult. Thus, the max-
imum height of slope models, in this study, was determined
to be as of 42 centimeters. For this limitation, most of our
models were stable during the test and only one model was
unsuccessful (Figure 3).
To assess the results of the existing theoretical approaches,
the stabilities of these models have been analyzed with Majdi
and Amini, and Amini et al methods. The comparative results
are presented in Figure 4. As it can be seen from the figure,
the results of the experimental and the theoretical models are Figure 4. Comparison between the theoretical and experimental
fairly close to and are in a good agreement with each other. results on the basis of Majdi and Amini, and Amini et al. method.
774

Fusion of drill monitoring data with geological borehole assays
Arman Melkumyan, Peter Hatherly & Hang Zhou

Australian Centre for Field Robotics, The University of Sydney, NSW, Australia
Keywords: numerical modelling, mining, site characterisation
The integration of the various forms of geological information

obtained by drilling is an important goal in mining and site
investigations. In mining, drill hole geochemical assays often
provide detailed information on rock types and ore grades.
Other potential sources of information can come from geo-
physical logging and drill performance monitoring usually
known as Measurement-While-Drilling (MWD). In this paper,
we investigate the fusion of assay data and MWD data.
To bring these disparate data together, we employ super-
vised statistical machine learning techniques using Gaussian
Processes (GPs) to build separate 3D models of grade and the
physical properties of the rocks. We then conduct fusion of the Figure 1. Plan view of the drilling pattern. Exploration holes are
rock property and grade models using Multiple Task Gaussian marked by crosses and the blast holes which had drill monitoring
Processes (MTGPs). MTGPs are a popular statistical super- data available are marked by the dots.
vised learning and fusion technique in the machine learning
community. The proposed algorithm autonomously learns the
intrinsic interconnections between the rock parameters and
geochemistry and uses these interconnections to improve the
quality of the geological model.
Supervised statistical machine learning represents a method Figure 2. Iron grade from exploration holes before fusion.
for automated generation of algorithms that reason from exter-
nally supplied training dataset to produce general hypotheses,
which then can be applied to make inferences for new datasets.
In other words, the goal of supervised machine learning is
to build an automated system that can accumulate and fuse Figure 3. Adjusted Penetration Rate (APR) from blast holes.
information from different sources and build concise models
of the underlying properties of functions and provide esti-
mations with corresponding uncertainties for the values of
these functions in new situations. Within a geological con-
text, this allows us to use measured discrete information about
grades and mechanical properties to automatically construct Figure 4. Iron grade after fusion with APR.
a continuous 3D predictive model.
We demonstrate the principles of our approach by fus- Figures 2 and 3 illustrate the geological model (iron grade)
ing MWD data from closely spaced blast hole drilling at an and the model of APR for the mining bench which utilized
open pit iron ore mine and assay results from more widely the blast hole pattern involving the blast holes shown in Fig-
spaced exploration drill holes. For the geological model, ure 1. Figure 2 is smooth and shows general trends. Figure
chemical assays of samples from consecutive 2 m intervals 3, on the other hand, is based on closer spaced drilling and
taken down exploration drill holes which are at spacings of shows more detail. However both images show similar fea-
25–50 m are used. From the blast holes, rock strength is rep- tures which allows statistical propagation of information from
resented by a parameter we call the Adjusted Penetration one model to another. In the resulting fused iron grade model,
Rate (APR). APR is calculated from the relevant rotary drill Figure 4, it is clearly seen that detail from the APR model has
monitoring measurements concerning Penetration Rate (PR), been incorporated. A more detailed geological model is thus
weight on bit (also called Pulldown Pressure, PP) and Rotation obtained.
Pressure (RP).
775

Parametric study of formation stability using a hollow cylinder model
P.A. Nawrocki, Z. Qi & D. Wang

Department of Petroleum Engineering, The Petroleum Institute, Abu Dhabi, UAE
Keywords: numerical modelling, rock failure, rock properties, rock stress, stability analysis
1 INTRODUCTION General distribution of stresses in HC has been analyzed

first and the concept of “operational mud weight window”
Hollow Cylinder (HC) modelling has been a popular method has been introduced. It was shown that low Pi values favor
for borehole stability analysis as geometry of HCs makes them shear failure; tensile failure at the wellbore wall can be
an ideal tool to simulate wellbore situations. In this paper, the expected for high hole (well) pressures; in both cases failure
linear elastic theory has been used in the parametric analysis is always at the inner wall.
of borehole stresses in the HC setting. Different hole sizes Then failure criteria mentioned above have been used in
and loading conditions have been considered and the influ- assessment of critical pressures, fracturing and collapse, that
ence of the internal and external pressure on stability has been define safe mud weight window. To assess the influence of
analyzed. pore pressure on stability, analysis has been conducted both
for saturated and dry conditions. The set of rock properties
and in-situ stress data from one local oilfield has been used
2 ROCK FAILURE CRITERIA and the results presented in terms of critical mud weights. The
critical values of pressure ratio β have been obtained for three
The three popular failure criteria, the conventional “triax- failure criteria considered and the safe mud weight window
ial” Mohr Coulomb (MC), the Drucker-Prager (DP) and the has been defined in each case.
Modified Lade (ML) criterion have been used to explore the Finally, predictions given by the three failure criteria
variation trends of critical wellbore pressures for both dry and considered have been compared.
saturated conditions.
The differences between these criteria have been high-
lighted by comparing their predictions in σ1 –σ2 space for 4 CONCLUSIONS
different values of confining stress σ3 . It has been shown that
for fixed values of σ2 and σ3 , the inscribed DP predicts failure The HC model has been confirmed as an effective tool to
at lower stresses than the circumscribed DP, which behaves study wellbore stresses and stability. Its geometry and loading
significantly different than the other criteria. The ML first adaptability makes it ideal for reproducing stress states around
predicts strengthening effect with increasing σ2 followed by wellbores and simulating a much wider variety of loading
a slight reduction in strength once σ2 becomes “too high”. conditions than other available tests.
Thus, the ML criterion seems to provide a good alternative to Pore pressure plays an important role in borehole stability
the MC criterion. analysis and mud weight design. It can significantly weaken
the rock narrowing down the safe mud weight window to a
great degree. Hence, neglecting pore pressure (or inaccurate
3 ANALYSIS pore pressure estimation) will mislead the safe mud weight
choices and bring about wellbore instability problems.
HC with an inner radius Ri and an outer radius Ro has been The MC criterion is apparently conservative. It is a two-
considered. The external pressure is Po , the internal pressure dimensional failure criterion that does not account for the
Pi and the axial pressure is F. Utilizing linear elastic theory, intermediate principal stress strengthening effect. This anal-
the HC effective stresses, radial (σr ), hoop (σθ ), and axial (σz ), ysis has also re-confirmed that the DP criterion is apparently
have been calculated and parametric analysis of HC stresses non-conservative and overpredicts the σ2 strengthening effect.
and stability has been conducted in terms of two major param- The ML criterion is a moderate one, between the extremes of
eters, α and β, that introduce different HC geometries and the other two criteria. Thus, it can provide reasonable mud
loading conditions into the analysis. Different values of these weight predictions.
parameters give rise to different stress distributions in the HC
and, subsequently, affect the critical internal pressures.
776

Rock mass characterization using attenuation of coda waves
R.A.O. Pedroza
Geotechnical and Mining Department, Faculty of Engineering, Francisco de Paula Santander University,
Cucuta, Colombia
J.C.A. Vargas
Geosciences Department, National University of Colombia, Bogotá, Colombia
P.C.E. Rodríguez
Faculty of Engineering, Javeriana University, Bogotá, Colombia
Keywords: field measruments, lab testing, rock properties, site characterisation
1 INTRODUCTION territory is a zone with high seismic activity and a com-

plex geological and tectonic configuration as result mainly of
To evaluate the seismic site response is of great significance convergence of three tectonic plates acting against the South
in geotechnical engineering research and applications. The American Plate, they are: Nazca, the Caribbean and Coccos
amplitudes of seismic waves at observing sites are influenced plate. The rock mass evaluated on this research belongs to the
by the source characteristics, travel path and local site con- Eastern Mountain Range of the Colombian Andean region.
ditions. Seismic wave attenuation in geological materials is a At site, the hard sandstone formations of the Guadalupe
process resulting of the interaction of several mechanisms that Group, consists of medium-sized banks to thick quartz sand-
contribute to the energy reduction during dynamic excitation. stone, with some calcareous fossils, yellowish-white, very fine
Anelastic attenuation (expressed as the inverse of the q fac- grained to medium, subrounded, with medium to high spheric-
tor) is the energy loss that seismic waves experiment as they ity, with stratification very thin to very thick, hard, compact,
propagate through a medium. In the seismogram the atten- generally well cemented by silica or carbonates, and with spo-
uation is represented by the amplitude reduction, caused by radic intercalations of light gray siliceous mudstone layers 3 to
geometric spreading, intrinsic attenuation due to the medium 10 cm (Fierro J. & Pedroza R, 2010).
anelasticity, and scattering due to inhomogeneities in the
media, inelasticity and multipathing, among other factors.
Vargas & Pedroza (2009). 3 METHODOLOGY
In the present study, the seismic quality factor and attenua-
tion coefficient of the seismic waves of coda in the Mondoñedo The seismic survey was made by using a Geometrics Geode,
region of the Cundinamarca state, Colombia has been inves- 24 channel seismograph and the seismic configuration design
tigated. Both parameters are physically diagnostic of the was carried out by covering all the study area by using six seis-
different types of geological materials [soils and rocks], mic line arrays of 24 sensors Dx-20x, with distance among
because they reflect the differences in their anelastic prop- them of 2.0 m and connected to the multichannel recorder
erties. A number of physical and mechanical properties have (Geode); two of them oriented E-W; two lines in S-N direction
been inferred from the velocities of propagation of the differ- and two diagonals NW-SE and NE-SW. In this experiment, a
ent types of seismic waves as well as their bulk densities to mass of 16 pound (7.25 kg) was employed to generate excita-
provide direct information about the rock material quality. In tions by hammer impacts on a steel plate to create energy for
the same way, the attenuation of seismic energy was also used acquisition of seismic data. Five hammer blows were applied
as a measure of the rock material quality and to estimates de on steel plate near placed each sensor. The total length of the
dynamic quality of the rock mass. geophone array was 2.2 km. More than 132,000 signals were
captured, of which 26,400 were selected by their quality, so
the Qc calculation was made using the model of Sato (1977)
and with a window time of 2 s.
2 GEOLOGICAL OUTLINE
To determine the Qc quality factor using the single back
scattering method (Aki & Chouet 1975), it has implemented
Colombian, located at the northwestern corner of the South
the following procedure:
American plate, presents following structural systems, each
one with their own seismotectonic characteristics (e.g. i. Filtering of the seismogram. It was made a bandpass-
Taboada et al. 2000): a) the Caribbean zone, b) the Andean filtered process for each seismogram over the frequency
region and c) The basin named “Llanos Orientales”. This bands 20–40 Hz in order to detect heterogeneities.
777

ii. Calculation of the envelope of energy from of quadratic
average amplitudes, for t ± 0.05 s with sliding windows
of 0.05 s.
iii. Choosing a time window defined by the time between
2 * (r / v) to 2 * (r / v) + 1 s, from the excitation’s initial
time.
iv. Adjusting the root mean square (rms) of the amplitudes
using linear regression by Least Squares. The value of
Qc-1 was obtained from the slope of the regression. Only
amplitudes greater than two times the background noise
level have been considered.
v. Only considered the Qc-1 values for which the adjust-
ment by using least square shows a correlation coefficient
greater than 0,7. Finally, the frequency dependence law of
the coda Q [Qc-1 = Qo-1f-n] was adjusted by using the
set of values of Qc-1 found at different frequencies.
Borehole seismic measurements were carried out using the Figure 1. P- and S-wave profiles from downhole test.
downhole method.
The laboratory testing program evaluates the general phys-
for a Poisson’s solid Lame’s constants equal to each other ( ).
ical and mechanical properties of rock. The natural water
In this research, the ratio Vp/Vs (compressional/shear wave
content, apparent porosity, weight unit, permeability, triaxial
velocity) for the hardstone had an average value of 1.93 taken
and unconfined compression strength, petrographic exami-
as base the downhole test data (Figure 1).
nation, ultrasonic elastic constant, among other properties
were evaluated. The specimen of sandstone were tested both
unsaturated and saturated conditions. 5 CONCLUSIONS
It has been presented a useful tool for geotechnical purposes;

4 SOME RESEARCH RESULTS
it consists in a procedure that extrapolates the use of the coda
wave in evaluating the rock mass characteristics. in such a way
Coda Q values were obtained for frequency bands centred
that it allows determining of the lateral variation of attenuation
at 5, 10, 15, 20 and 25 Hz. Table 1 show a sample of the
anomalies.
dimensionless quality factor Q and its components, scattering
Three main subjects involved in the present paper are:
and intrinsic values.
(1) Description of the geological conditions of the evaluated
As it can been observed the seismic wave amplitudes reduce
rock mass. (2) Outline of the field and in laboratory testing
as waves propagate through the rock mass. This reduction
program. (3) Determination of the P and S velocities and both.
is frequency dependent in such way that Q increases with
the seismic quality factor and attenuation coefficient for coda
increasing frequency. From the separation of Qs and Qi values,
waves. From such parameters an analysis about properties of
it is observed that the study area can be characterized by a high
the rock mass were inferred.
scattering attenuation (high scattering Q inverse, Qs-1) and by
a relatively low intrinsic attenuation (low intrinsic Q inverse,
Qi-1). REFERENCES
On the other hand, by using downhole test information
was obtained the dynamic elastic properties of the sandstone. Aki K. & Chouet B. 1975. Origin of coda waves: source. attenuation
Poisson ratio for the analyzed hard sandstone ranges between and Scattering effects. J. Geoph. Res. 80. 3322–3342.
0.01 and 0.45, with average value of 0.26. Usually loose mate- Fierro M. J. & Pedroza R.A. 2010. Some aspects of the Herrera
rials posses higher values µ = 0.45 and very hard rocks have Lagoon geology. (in press).
lower values µ = 0.05. Sato. H. 1977. Energy propagation including scattering effects.
The shear modulus ranges from 0.25 MPa to 5.82 MPa and Single Isotropic scattering approximation. J. Phys. Earth. vol 25.
theYoung’s modulus (E) ranging from 3.83 MPa to 16.72 MPa. p. 27–41.
Taboada. A. et al. 2000. Geodynamics of the northern Andes:
The previous data show a natural material with high variation
Subductions and intracontinental deformation. Colombia).
in its elastic properties. Tectonics. Vol. 19. No. 5.
Fracturing and/or water saturation of the studied geological Tatham. R. H. 1982. Vp/Vs and lithology: Geophysics. v. 47.
materials have stronger influences on observed Vp/Vs ratio p. 336–344
than elastic parameters of the mineral comprising the matrix Vargas J. C. A. et al. 2010. Anelastic tomography by means of coda
(Tatham 1982). The ratio Vp/Vs taken as reference is 1.73 waves for geotechnical purposes. (in press).
778

Rock mass classification based on seismic measurements using Support Vector Machines
A.J. Rechlin
Scientific Drilling, GFZ German Research Centre for Geosciences, Germany
S. Lüth
Centre for CO2 Storage, GFZ German Research Centre for Geosciences, Germany
R. Giese
Scientific Drilling, GFZ German Research Centre for Geosciences, Germany
Keywords: rock mass classification, tunneling
1 INTRODUCTION A new approach of rock mass classification with seismic

velocities using support vector machines is presented. Over
The success of tunnelling is strongly influenced by an accurate the past few decades, an increasing number of self-learning
knowledge of the host rocks’ geotechnical behavior. Seis- algorithms for pattern recognition have been developed. One
mic measurements during tunnel construction can provide of the newest and most promising methods for the application
detailed, continuous and cost-effective prior information on to real datasets are the Support Vector Machines (SVMs) that
the rock mass characteristics along the projected roadway. In are based on the Statistical Learning Theory (Vapnik, 1998).
this framework the Integrated Seismic Interpretation System SVMs belong to the Kernel Methods (KMs) and are a rela-
(ISIS) was developed at the GFZ (Borm et al., 2003) for the tively new family of algorithms that present a series of useful
seismic exploration of the rock ahead of the tunnel as much features for pattern analysis in datasets. KMs combine the sim-
as in the tunnel near field. ISIS is applicable to construction plicity and computational efficiency of linear algorithms, such
sites in highly complex geological settings due to its modular as the perceptron algorithm, with the flexibility of non-linear
design and continuous development of new hardware (Lüth, systems- such as neural networks, and the rigor of statisti-
2008). Moreover algorithms for data processing are adjusted cal approaches such as regularization methods in multivariate
and developed. The seismic results obtained need to be inter- statistics.
preted within a small timeframe, enabling engineers on-site An important property of SVMs is that the determination
to make adjustments to the excavation method and the tunnel of the model parameters or the “learning step” corresponds
design as the excavation proceeds, thereby making optimal to a convex optimization problem, so that any local solution
use of time and resources. To this end a rock mass classi- becomes a global optimum. This is a major advantage over
fication, e.g. the Rock Mass Rating (RMR, e.g. Bieniawski, neural networks. Moreover, SVMs are very resistant to over-
1989), based on seismic velocities is done using Support Vec- fitting, especially in regimes where other methods are affected
tor Machines. We will discuss the approach applied using an by the “curse of dimensionality”.
example from one test site in Switzerland. In rock mechanics, the system (rock mass) is usually
complex, fuzzy and nonlinear. That means the relationship
between the known data distribution and the variables cannot
2 METHOD be obtained accurately, therefore the usage of conventional
methods like regression and statistics might be inefficient.
Rock mass classification is an important tool in rock engi- The basic idea behind an SVM is that a non-linear and non-
neering decision making. However, a direct and detailed separable dataset becomes separable in a higher dimension.
assessment of the features, like the RQD, the joint attributes Thus, the data is mapped to a higher dimension by a mapping
or the UCS is time consuming. Seismic measurements on the function. In this higher dimension, the SVM searches for the
other hand are cost-effective and can provide high-resolution optimal decision surface (hyperplane) that acts as the linear
data along the projected roadway. El-Naqa (2001) stated that it decision function with a maximum distance (margin) to the
is useful to correlate the seismic velocity to a geomechanical nearest example vectors of the respective classes.
classification system (e.g., the RMR) that considers several
features that affect the geomechanical behavior of the rock
mass, such as the discontinuity characteristics (alteration, 3 FIELD INVESTIGATIONS
roughness etc.), in addition to the RQD. To determine how
the RMR and RQD are predictable via seismic velocities both The test-site was situated in the Faido adit that belongs to the
classification approaches are used in this work. One of the few Gotthard Base Tunnel construction site (Switzerland).
applications of SVM to rock engineering purposes has been Seven seismic profiles were measured between September
carried out by Hong-Bo et al. (2002). 2000 and June 2001 in the Faido adit in southern Switzerland
779

by the GFZ German Research Center for Geosciences. The Both classification approaches using SVMs showed good
adit, excavated by drill and blast method, belongs to the training and testing accuracies, though the RQD was not as
Gotthard base tunnel under construction, which will constitute sensitive to the seismic velocities as had been expected.
an important link in the Trans-European Transport Network.
5 CONCLUSION
3.1 Geological setting
The lithology along the Faido adit is separated into two The nonlinear mapping between the rock mass classes and
units: the Leventina Gneiss (LeG) and the Lucomagno (LuG) the seismic velocities are learned from a finite number of
Gneiss. These varieties separate the Penninic Gneiss zone into samplesalong the tunnel alignment. The rock mass conditions
two parts – the southern (LeG) and the northern (LuG) sec- realized confine the applicability of the trained SVM. In the
tions. Both varieties are composed of multiple deformed and presented example only “high” and “low” values in the RMR
metamorphically overprinted rocks of mainly granitic origin. classification of two gneiss varieties could be distinguished.
The boundary between the gneiss varieties is complex and Thus, more training datasets, detailing the 5 possible RMR
overprinted by isoclinal folds. Four major faulted and fractured classes and different rock types should be regarded next.
zones are encountered along the surveyed sections.
ACKNOWLEDGMENT
3.2 Seismic measurements
Seven seismic profiles of approximately 70 m length were The collaborative research project OnSITE was funded by
acquired between tunnel meter 880 and 2440 during the exca- the German Federal Ministry of Education and Research
vation of the 2600 m-long, NE-SW oriented Faido adit. The within the GEOTECHNOLOGIEN programme. The project
measurement campaigns were carried out at intervals of about was coordinated by the GFZ Potsdam. Gratefully acknowl-
200 m. Each seismic profile began less than 10 m behind the edged are the operational and logistical support provided by
advancing tunnel face. As the tunnel was excavated with the Herrenknecht AG, Hochtief Construction AG, Ed. Züblin AG,
drill and blast method, the seismic source was moved inde- and Amberg Technologies AG.
pendently. Eight to ten 3-component receivers, installed on
rock anchors drilled two meters deep into the left tunnel wall REFERENCES
(referring to the tunnel advance direction), were recording
simultaneously along each profile. One rock anchor was posi- Bieniawski, Z. T. (1989) Engineering rock mass classifications – a
tioned on the opposite wall of the tunnel to control several complete manual for engineers and geologists in mining, civil, and
guided waves along and around the tunnel surface. A tomo- petroleum engineering, John Wiley and Sons
graphic model of grid size 0.5 m built the basis for further Borm, G., Giese, R., Otto, P., Amberg, F. & Dickmann, Th. (2003)
interpretations of the seismic results (Giese et al., 2005). The Integrated Seismic Imaging System for Geological Prediction
seismic velocities were acquired from the tomograms along During Tunnel Construction, In Proc. Intern. Symposium Series
scanlines 5 m deep into the tunnel wall to avoid the influence S33 ISRM-2003 Technology roadmap for rock mechanics, South
of the Excavation Damage Zone (EDZ). African Institute of Mining and Metallurgy, 137–142
El-Naqa, A. (2001) Assessment of geomechanical characterization of
a rock mass using a seismic geophysical technique, Geotechnical
and Geological Engineering, 14, 291–305
4 RESULTS Giese, R., Klose, C. & Borm, G. (2005) In situ seismic investiga-
tions of fault zones in the Levantina Gneiss Complex of the Swiss
The validity of the approach was tested on two popular rock Central Alps, In Petrophysical Properties of Crystalline Rocks,
mass classification systems: the Rock mass Quality Index The Geological Society, 240, 15–24
(RQD) and the rock mass rating (RMR). Based on the RMR, Hong-Bo, Z. & Xia-Ting, F. (2002) Classification of engineering
two rock mass classes were determined along the seven seis- rock based on support vector machine, Rock and Soil Mechanics,
mic profiles. The boundary between those classes was at 60 23 (6), 398–701
Lüth, S., Rechlin, A.J., Giese, R., Tzavaras, J., Gross, C., Buske, S.,
RMR which separates “fair” from “good” rock in the classifi- Jetschny, S., DeNil, D., Bohlen, T. (2008) Seismic Prediction
cation scheme. The RQD was separated into 3 classes, based ahead of a tunnel face – Modeling, field surveys, geotechnical
on the number of occurrences, with either values in the range interpretation, International Journal of the JCRM, 4 (2), 47–51
“excellent” (RQD > 90), “good” (RQD 75 to 90) or “lower” Vapnik, V. N. (1998) Statistical Learning Theory, John Wiley and
(RQD < 75). Sons, 736
780

A rock mass classification to estimate excavation rate, support and rock properties
in a borehole survey
M. Sawada, K. Shin & Y. Inohara

Civil Engineering Laboratory, Central Research Institute of Electric Power Industry, Abiko, Japan
T. Shidahara
Newjec Inc., Tokyo, Japan
T. Hatano
Electric Power Development Co. Ltd., Tokyo, Japan
T. Miwa
Tohoku Electric Power Co. Inc., Sendai, Japan (formerly at Nuclear Waste Management Organization of Japan)
Keywords: nuclear repository, site characterization, rock mass classification, tunnelling, rock properties
A method of evaluating engineering rock mass characteristics level. The distribution of excavation rate in the qu-RQD-Jc
has been proposed for the purpose to help the adequate selec- space has been also analyzed in the same way.
tion of the detailed investigation areas, which is an important Relation between 3 parameter-rock mass description and
decision making for the selection of the final disposal site of rock mass properties of strength and Young’s modulus has
high level radioactive waste in Japan. been analyzed based on published in situ rock shear tests and
The method requires the estimation of excavation rate, plate jack tests in Japan. Figure 2 shows the contour lines
support and mechanical properties of rock mass using the (red) for cohesion on qu-RQD graph at Jc = 3. The values in
information from the rock core, which is qu: uniaxial strength the small squares are the cohesion from the in-situ rock shear
of rock core, RQD and Jc: joint condition. tests at dam sites in Japan. The contour lines are drawn based
Relation between the 3 parameters and advance rate of on the values in the squares. It is obvious that higher RQD and
excavation and support has been analyzed based on published qu gives higher cohesion. This qu-RQD graph has also been
records of tunnel excavation. Each of qu, RQD and Jc is cat- made for other values of Jc. Lower Jc, which means better
egorized into 5 groups. Jc is categorized in the same way joint condition, gives higher cohesion. The contour lines for
as in Rock Mass Rating system, RMR. Figure 1 shows the Young’s modulus on qu-RQD graph have also been obtained
distribution of support classes for tunnels excavated by drill in the same way.
and blast in the qu-RQD graph at Jc = 3. The average numbers We can roughly estimate excavation rate, support, shear
of rockbolt are written at corresponding positions in the field strength and Young’s modulus by the contour lines obtained
with colored ellipses. The degree of support has been catego- in this study like Figure 1 and 2, if qu, RQD and Jc are deter-
rized into 3 classes of A, B and C. A is light support level, B mined. All of these 3 parameters are able to be obtained in a
is medium support level, and C is heavy support level. This borehole survey. The feature of this method is that it uses the
qu-RQD graph has been made for each Jc value. It is found 3 dimensional information of qu, RQD and Jc without con-
that higher RQD and Jc, and lower Jc require lighter support verting the information to one dimensional value as in RMR.
Figure 1. Numbers of rockbolt for drill and blast excavation in the

qu-RQD graph at Jc = 3. Figure 2. Cohesion (unit: MPa) in the qu-RQD graph at Jc = 3.
781

Charts for estimating active earth pressures on retaining walls assuming

Hoek & Brown non-lineal strength criterion
A. Serrano
Universidad Politécnica deMadrid, Spain
A. Perucho & M. Conde

Laboratorio de Geotecnia del CEDEX, Ministry of Public Works, Spain
Keywords: physical modeling, stability analysis
1 INTRODUCTION
Coulomb’s classic method for calculating earth thrust on walls

was developed for incoherent materials with a lineal strength
criterion assuming a plane failure surface. With the existing
methods, based on Coulomb’s, the calculation of thrusts with
materials that follow non-linear strength criteria requires a
previous linearization of the failure criterion choosing a range
of stresses.
The new developed method is an extension of Coulomb’s
classic method, but it is valid for materials with linear and non
linear strength criteria, and not plane failure surfaces. The Figure 1. Wall geometry.
charts presented in this paper have been obtained assuming
Hoek & Brown strength criteria. They allow for determination
of the thrusts on walls due to materials whose mechanical
behaviour is clearly non linear (armourstone, highly fractured
rock masses, pyroclasts, etc.)
2 BASIC HYPOTHESES
1. The wall is assumed to be infinite so that it is a two-

dimensional problem in plane deformation condition.
2. The earth surface is plane, forming an angle, α, with the
horizontal.
3. The wall has a vertical backfilling.
4. A wedge of earth limited by a surface passing through the Figure 2. Forces in a slice (a); Oblique axes (X, Y) (b).
foot of the wall thrusts against the wall.
5. The earth is dry, i.e., pore pressures are not taken into the wall has to be moved towards it. In these situations the
account. Calculations are done on effective pressures. wedge attempts to lower or rise itself and thus there are
6. The earth has a specific weight of γ. established the directions of the reactions on the failure
7. The normal (σ) and tangential (τ) stresses acting on the surface of the wedge.
surface of the thrust wedge verify a failure criterion.
8. The resultant force on any vertical plane of the wedge forms
a constant angle with the horizontal (i.e. t/n = constant).
The value adopted for this angle is the value of the 3 EQUILIBRIUM EQUATIONS AND CHARTS
earth-wall friction angle, δ, defined by its tangent:
µ = tan δ = t/n. The horizontal and vertical balance of forces and the bal-
9. The failure surface produces the external force on the wall ance of momentum let to get the following differential system
(maximum and minimum in the active and passive case (Fig. 2a):
respectively).
10. For the maximum force on the wall to be produced the wall
has to be moved away from the earth while for the minimum
782

Table 1. Dimensional measuring units.
Parameter Unit
Specific weights: γ kN/m3

Pressures: β kN/m2
Lengths: L = β/γ m
Forces: L β = β2 /γ kN/m
Moments: L2 β = β3 /γ 2 kN
Figure 4. Values for B, position of the resultant of the force

(H = H ∗ /L; H *: real wall height).
4 CONCLUSIONS
A new method for estimating earth pressures and total thrust

on retaining walls has been developed. It is an extension of
Coulomb’s earth pressure theory for non cohesive materials,
valid in more general cases like:
• Materials following either a lineal or a non-linear strength
Figure 3. Values for λ (H = H∗ /L; H*: real wall height). criterion (any parabolic or H & B criteria).
• Non-horizontal earth surface.
• Earth-wall friction angle.
where: t = µn (µ is a constant); m = nB; h = x − y tan α being • The material dilatancy is considered.
µ a constant, t and n the tangencial and normal forces in the • The failure surface does not need to be plane, as in pre-
slices respectively, m the momentum in the slice base (point viously developed methods, but its shape is obtained as a
M , fig. 2a), B the distance of force n to point M . result of the calculus, being this point the most remarkable
Expressing the equations in oblique axes (Fig. 2b), adimen- achievement of this method.
sionalizing and grouping terms, the system has been solved.
The final aim is to determine n (horizontal component of the The method can be used to calculate active and passive
force on the wall) and m (moment acting on the foot on the thrusts.
wall) for X = H , height of the wall. Y (X ) is also found, and A number of charts calculated with this method are included
it defines the wedge surface making the integral maximum in the paper, useful for estimating horizontal thrust and
(active case) or minimum (passive case). momentum at the foot of the wall, covering a wide range of
All the developments have been done adimensionally, cases.
adopting the dimensional measuring units in Table 1.
Where β is a strength parameter, as defined by Serrano &
REFERENCES
Olalla (1994) and L is an intrinsic length (L = β/γ).
Having solved any problem the real results will be obtained
Serrano, A. & Olalla, C. 1994. Ultimate bearing capacity of rock
by multiplying them by the respective physical units. masses. Int. J. Rock Mec. Min. Sci. Vol. 31.
The results may be expressed as adimensional charts useful Serrano, A., Olalla, C. & Perucho A. 2007. Active and passive earth
for obtaining the thrust and the moment acting on the wall. A pressures on retaining wall assuming a non-linear strength crite-
group of charts, obtained assuming Hoek & Brown strength rion and constant dilatancy. ISRM 11th Int. Cong. on the Sec. Half
criterion have been obtained for the active case and for a range Cent. of Rock Mech. Lisbon, 2007.
of parameters. They are included in the paper. Figs. 3 and 4 Serrano, A., Perucho A. & Conde, M. 2010. General method for
show some of them. estimating the active and passive earth pressures on retaining
Adimensional horizontal thrust, Eh , on the wall and momen- walls assuming different strength criteria. Eurock 2010. Lausanne
tum at the foot of the wall, M , are given by: (Switzerland).
being H is an adimensional height = H ∗ /L (H *: real wall

height). To obtain the real horizontal thrust and momentum
they will have to be multiplied by Lβ and L, respectively:
Some examples of use have been developed and included
in the paper.
783

A new classification system for wet rocks (MiC System)
H. Soroush
Well Engineering Centre of Excellence (WECOE), Weatherford Oil Tool M.E., Dubai, UAE
A. Fahimifar
Department of Civil and Environmental Engineering, Amirkabir University of Technology, Tehran, Iran
Subject: Rock material and rock mass property testing (laboratory and in-situ)
Keywords: rock mass classification, rock properties, lab testing, rock failure
1 INTRODUCTION microfissures index, swelling index, and grain size index).

To calculate each index, the magnitude of the parameter is
Rock masses are usually in wet or saturated condition in the divided by its maximum possible magnitude multiplied by the
nature. Interaction of rocks with water leads to a reduction corresponding score.
in rock mechanical properties, and quantifying this effect has
been always a problem in rock engineering projects. Knowl-
edge of the mechanisms of the interaction between water and 5 MOISTURE INDEX CLASSIFICATION SYSTEM
earth materials and its effects rock properties is of paramount
importance in geotechnical engineering. Despite the fact that In the MiC system, rocks are classified into 5 classes according
a considerable number of research has been carried out on this to their final score:
topic, they are not thorough and general.
1. Very strong rocks do not encounter any weakness in wet
In this paper, a new rock classification system is presented
environments. These rocks will behave as in dry conditions.
in order to evaluate wet rock properties. The proposed system
2. Strong rocks encounter a little weakness in a wet environ-
includes the parameters with major effect on the strength of
ment; however, this effect may be more significant in a
wet rock. These parameters include: water absorption, poros-
particular aspect depending on the classification factors.
ity, weathering degree, swelling index, microfissures intensity,
3. Medium rocks in a wet environment encounter consider-
and grain size.
able weakness. It means that rock physical, mechanical or
physico-chemical properties may be affected considerably.
2 EXPERIMENTAL STUDY 4. In the case of weak rock, a high degree of weakness occurs
and UCS, cohesion and internal friction angle reduce sig-
A comprehensive laboratory study was carried out on a variety nificantly. In fact, nearly all rock properties are affected
of rock types to measure the required physical and mechan- notably.
ical rock properties. Each of the six chosen parameters was 5. Very weak rocks show remarkable weakness in wet envi-
plotted versus UCS in saturated condition to understand their ronments and all rock parameters are affected extensively.
relationship.
6 CONCLUSION
3 SENSITIVITY ANALYSIS OF PARAMETERS
A new classification system was proposed for evaluation of
To determine the sensitivity of the saturated UCS to each moisture effect on rocks utilizing some parameters which con-
parameter, sensitivity analysis was performed using the cross- trol mechanical, physical and physico-chemical properties of
plots. The distribution of the lab data shows that they can be rock, predominantly at laboratory scale. Using a rating sys-
divided into two separate groups: rocks with UCS smaller and tem, an index called Moisture index (Mi) is allocated to each
higher than 50 MPa (Group I & II respectively). These slopes rock. This classification system (MiC) can be carried out in
of the fitted lines to the lab data indicate the importance of four steps:
each parameter and correspondingly their score.
1. Petrographic description including determination of rock
type, grain size, number of microfissures and weathering
4 DETERMINATION OF THE INDICES degree.
2. Laboratory testing for determination of effective porosity,
In MiC classification system, six indices are defined to reflect water absorption and swelling index.
the influence of each parameter on rock properties (i.e. water 3. Calculation of required indices, factors and finally Mi.
absorption index, effective porosity index, weathering index, 4. Determination of the rock class.
784

The life-cycle information management practice of Shanghai Yangtze River Tunnel and
Bridge Project
H.Y. Tian & Q.W. Liu

Shanghai Changjiang Tunnel & Bridge Development Co., Ltd, Shanghai, China
Keywords: tunneling, Shanghai Yangtze River Tunnel, life-cycle information management, project information management,
construction information systems, operation information management
Shanghai Yangtze River Tunnel & Bridge Project is an extra- introduced. The difficulties of the overall life-cycle informa-
large transportation infrastructure project that is an important tion management for ShanghaiYangtze River Tunnel & Bridge
part of the express way from Shanghai to Xi’an. The com- Project such as large site location, multi-parties involved, huge
pletion of this project will improve the structure and layout amount of information data and new technologies, equipments
of transportation system in Shanghai, speed up the regional and materials adopted were illustrated.
economic integration of the Yangtze River Delta, better pro- Second, in order to better carry out information manage-
mote the economic developments of the Yangtze River area ment for such an extra-large river-crossing project and to
and even China overall, and enhance overall economic com- improve the efficiencies for management, the engineering
petitiveness of Shanghai within China. The project starts from information management system was established to achieve
wuhaogou in Pudong, crossing Yangtze River South Channel the project progress control, investment control, contract man-
waterway by tunnel to Changxing Island and spanningYangtze agement, quality control, construction management, design
River North Channel waterway by bridge to the Chongming management, document management, security management,
Island. The total length of the project is 25.5 km long, within materials management and equipment management.
which the length of the tunnel is 8.95 km (the length of river- Third, for such a project, all kinds of information and
crossing is 7.5 km), and the length of the bridge is 16.63 km data from survey, design and construction should be received,
(the length of river-crossing is 9.97 km). There are many
difficulties of a huge river-crossing project, such as large
scale, large investment, long construction period, complex
technologies, multi-parties involved, high risk and long site
locations.,etc. Then information management has become the
effective method for construction and operation management
for such a project. For this, overall life-cycle information
management was set as the target, the project information
management system, construction information management
system, operation information management system are estab-
lished to effectively manage all the in formations from project
management, design, construction and operation of such an
extra-large project, that have effectively improved the effi-
ciency and quality of the construction and operation of the Figure 2. Universal fitting shield tunnel segment arrangement.
Shanghai Yangtze River Tunnel & Bridge Project.
This article first described the trend of information sys-
tem applications for civil engineering project. The details of
the Shanghai Yangtze River Tunnel and Bridge Project were
Figure 3. The structure of integrated structural health monitoring

Figure 1. Structure of project information management. system.
785

Figure 4. The complex operational safety monitoring system.
Figure 6. The operation information system.
Fourth, on the basis of project management system and

construction information system, structural health monitoring
system and integrated monitoring system were set to analyze
the data from structures and equipments operation. Then oper-
ation system was later set to integrate all the information’s
during different period of the project which is helpful for the
operation and maintenance for the project.
Through this integration among systems illustrated above,
the information management of Shanghai Yangtze River
Tunnel & Bridge Project has realized the basic contributions
to the overall life-cycle informationization. It has played an
important role to upgrade the level of the construction and
Figure 5. Research of information system. operation management for such an extra-large project. Mean-
while, during the follow-up operation and maintenance period,
more technologies such as digital imaging detection and analy-
analyzed and fed back to optimize the design and construction
sis, mobile visualization and the integration of existing digital
works. Then on one hand the specialized information system
systems should be further researched to improve the manage-
was built including the information’s from TBM attitude con-
ment level of construction and operation for tunnel & bridge
trolling, construction monitoring, risk analyses, freezing data
integrated project.
for cross passages, and on the other hand, the digital platform
of tunnel and bridge construction period was established for
visual and integrated management during survey, design and
construction period which is the bases for effective applica-
tion of all the data and for establishing intelligent analysis and
decision-making system for the project.
786

In-situ experimental studies on improvement of deformability

of rock masses by grout treatment
S. Utsuki
Hazama Corporation, Japan
T. Asakura & K. Kikuchi

University of Kyoto, Japan
Keywords: case studies, rock mass, rock properties, rock mass classification
1 INTRODUCTION
The grout treatment is to inject cement milk from the bore-

hole at the dam foundation in order to improve the mechanical
properties and permeability of the rock masses. The effect of
improvement about the mechanical properties are gained by
cement’s being fulfilled to the joints, and as a result, it can
be made rock masses “hard”, “homogenized”, and “unified”.
However, at the dam sites in Japan, the effect by grouting has
been verified only from the permeability tests outcome, and
there are not enough examinations about the mechanical prop-
erties. Therefore, regardless of many construction results such
as the consolidation grouting, the rise of mechanical proper-
ties by grouting is not considered in the dam foundation design
in Japan.
On the other hand, authors executed in-situ experiments at
the dam sites in 14 places and 5 kinds of rock types before and
after grouting. In this paper, the effect of improvement about
the deformability of rock masses is examined quantitatively Figure 2. Relation between the modulus of deformation before and
after grouting.
in detail from experimental results.
2 CONTENT OF IN-SITU EXPERIMENTS is improved to CM class at all the measurement points. More-
over, about the comparison of rock types, it is clear that the
The borehole load test is comparatively simple and used to igneous rocks are made harder than the sedimentary rocks.
measure the deformability of rock masses in the borehole.
So, in order to confirm the effects of improvement about the 3.2 Examinations of making rock masses homogenized
mechanical properties by grouting quantitatively, the borehole To make rock masses homogenized is to make variety of the
load tests were executed before and after grouting at the same mechanical properties small after grouting. It is shown in
position of boreholes. Figure 5 that the ratio of the modulus of deformation before
and after grouting is small as the modulus before grouting is
3 EXAMINATIONS OF EFFECT OF IMPROVEMENT large originally, on the other hand, it is large as the modulus
ABOUT DEFORMABILITY OF ROCK MASSES before grouting is small. As a result, it is clear that rock masses
are made homogenized by grouting.
3.1 Examinations of making rock masses hard
3.3 Examinations of making rock masses unified
Here, the result of tests is shown in Figure 2 that indicates
the relation between the modulus of deformation obtained at To make rock masses unified is to make joints fulfilled with
the same point in the boreholes before and after grouting. This cement, and the difference between the modulus of deforma-
figure shows that the rise of the modulus is admitted in almost tion and the modulus of elasticity is made small after grouting.
all the points, so it is clear that the rock masses are made hard In Figure 7, at many measuring points, D/E ratio after grouting
by grouting. And it is also seen the tendency roughly that is larger than before grouting, so it is thought that, as the joints
the effect of improvement is small when the modulus before are fulfilled with cement, the difference between the modulus
the grouting is large originally, on the other hand, it is large of deformation (D) and the modulus of elasticity (E) is made
when the modulus before grouting is small originally. And it is decreased and as a result, it is clear that rock masses are made
remarkable in CL-1 class so that the modulus of deformation unified after grouting.
787

Intelligent recognition of rock mechanical models
C.X. Yang & X.T. Feng

School of Resources & Civil Engineering, Northeastern University, Shenyang, China
B.R. Chen
Institute of Rock and Soil Mechanics, the Chinese Academy of Sciences, Wuhan, China
Keywords: neural network; numerical modeling; rock properties; back analysis; rock failure
During recent several decades, different theories and tech- This kind of methods has been successfully used in mod-
niques from the fields of artificial intelligence, system science eling the creep behavior of some engineering softrock and
and nonlinear science have been systematically investigated the mechanical behavior of limestone under different water-
for different purpose of modeling tasks in rock mechanics chemical environment (Yang et al. 2008; Chen et al. 2010).
and rock engineering. In these methods, the rock mechanics Modeling based on global optimal search of nonlinear
models are established through nonlinear relationship map- parameters. By adding some nonlinear items to simple mod-
ping, global optimal search, self-learning, self-organization els (e.g., linear elastic), we can also find another way to build
and data mining in a data-based modeling manner. The results up nonlinear models. In these cases, the nonlinear items should
are encouraging and present us another alternative method- be carefully derived to obtain a reasonable level of nonlinear-
ology for rock mechanics modeling (Feng et al. 1997; Feng ity. Usually, the nonlinear levels are dependent on a series of
2000). This paper gives some brief review of these works and parameters, for example, the coefficients of polynomial, the
makes some summery issues concerning with the application exponential parameters of the time dependent item, et al. The
aspects. purpose of the optimization problem is to find these struc-
Modeling based on nonparametric regression. Because ture parameters of rocks that will result in good agreement of
the mechanisms of most of the constitutive factors are farm simulating results with lab or field tests. For nonlinear rock
form well-known. Sometimes, it is impossible to derive a materials, the objective function is a highly nonlinear function
mathematic expression incorporating all those parameters to of parameters. It is very difficult to obtain an analytical evalua-
establish a reasonable model. For these cases, nonparame- tion of the function gradients. Those optimization procedures,
ter modeling techniques can provide potential alternative. Of such as Genetic Algorithms (GAs), Particle Swarm Optimiza-
these methods, Neural Networks (NN) provided rich, powerful tion (PSO), requiring no knowledge of gradient information
and robust nonparametric modeling framework with proven of the objective function, provide powerful solving tools. We
and potential applications across sciences. By selecting differ- have made systematically investigation of these methods and
ent input and output variables, we can build up different styles have proposed a series of algorithms for the optimization prob-
of models. It is worth to note that the new-found constitutive lem. Here, a GA procedure is employed to explain the solving
affects, such as the time-dependent (t), hydraulic, chemical process.
parameters, et al., can be conveniently incorporated into the The solution procedure starts from a population of candi-
model. If provided relative testing data, the NN model can date solutions. Each possible solution is used as real structure
learn the unknown mechanism of these factors. parameters of rock mechanics models for output analysis
The underlying relationship can be learned directly from to predict the response of ui at the observation points. The
testing data by train process of the NN model. Of these kinds response obtained from the output computation is then com-
of method, efficient learning algorithms and sufficient test- pared with those obtained from the lab or field tests to evaluate
ing data containing enough response information are of great the fitness of the population of the candidate solutions. The
importance. For the former, we have evolved a series of opti- estimated fitness values are then used by the GA operations to
mal algorithm, e.g., Genetic Algorithm (GA) optimized NN, obtain a new population of the parameter sets and thus evolve
to approve the conventional gradient search procedures and into a new generation. The population of candidate parameter
have been proved with powerful self-learning capacity. As to sets is updated until the prescribed stopping criterion is met.
the sufficiency of the testing data, different lab experiments For practical application, the output analysis procedure can
have to be designed to investigate each constitutive factor be an analytical or numerical computation process, depend-
under consideration. On the other hand, the macro response of ing on the testing data. Usually, if global load-deformation
rock structure, such as the load-deformation and excavation- data is used, a numerical procedure should be employed to
deformation responses, may contain abundant constitutive link the testing data to the stress-strain response.
information. This information can also be extracted through Using these kinds of modeling methods, we established the
numerical computation and be used for NN learning. nonlinear elastic constitutive model of laminated composites
788

and the stress-strain-time relationship of diatom softrock to certain mechanisms or experiences and the modeling pro-
(Feng & Yang 2001, Feng et al. 2002). cess reduced to a parameter optimization problem, a global
Modeling based on coupling recognition of model struc- optimal procedure is suitable. Commonly, the nonlinear mod-
ture and its parameters. Mathematically, rock mechanics els are partially known, e.g. the input variables are know but
models consist of a set of input variables, such as the stress, the nonlinear structure is difficult to establish, in this case, the
strain, stress history, strain history et al., or a set of constitutive coupling recognition of model structure and its parameters
units, such as the elastic unit, viscous unit and plastic unit. provide powerful alternative.
Although all these building blocks can be measured, tested
and analyzed based lab or field experiments, how to derive
a reasonable model incorporating all these relatively simple ACKNOWLEDGEMENTS
constitutive behavior to simulate complex nonlinear behavior
is still a difficult problem. Because the non-linear mechani- Financial supports from the National Basic Research Project
cal mechanism is not clearly understood in most cases, it is under Grant no. 2010CB732006, the Program for New
very difficult to assume the structure of the model in advance. Century ExcellentTalents in University under Grant no NCET-
Instead of the usual method that select a model by some test- 10-0275, the Special Fund for Basic Research of Central
ing analysis among several candidate models, which has great Colleges under Grant No. N090401002 and the SRF for
opportunity to lead to local minimal results, a recognition ROCS, SEM under Grant No. 20071108-4 are all gratefully
procedure based on global search of the model structure and acknowledged.
its parameters may obtain reliable models. This recognition
is a dynamic search problem being highly multimodal, multi-
variable with high order, and needing a large parameter space. REFERENCES
How to obtain a global optimum solution is a key to this
problem. Chen B.R., Feng X.T., Yao H.Y. et al. 2010. Study on mechani-
Genetic Programming (GP) is new structure description cal behavior of limestone and simulation using neural network
method and its nature is a generalized layering computer pro- model under different water-chemical environment. Rock and Soil
gram description. It does not need to determine in advance or Mechanics, 31(4): 1173–1180 (in Chinese).
restrict the form of the structure and its size. The description Feng X.T. & J.A. Hudson. 2010. Specifying the information required
can automatically modify the size of the structure according for rock mechanics modelling and rock engineering design.
to the change of its environment. Therefore, it is attractive for International Journal of Rock Mechanics and Mining Sciences,
47(2): 179–194.
recognition of the structure of a non-linear constitutive mate- Feng X.T. &Yang C.X. 2001. Genetic evolution of nonlinear material
rial model. By coupling the GP and GAs, we have proposed constitutive models. Computer Methods in Applied Mechanics and
a hybrid evolutionary algorithm simultaneously recognition Engineering. 190(45): 5957∼5973.
of the structure of the non-linear constitutive material model Feng X.T. (ed.) 2000. An Introduction to Intelligent Rock Mechanics.
and its coefficients. The GP is used to recognize the struc- Beijing: Science Press. (in Chinese).
ture of the non-linear stress–strain relationship without any Feng X.T., Chen B.R., Yang C.X., et al. 2006. Identification of visco-
assumption in advance and a GA procedure is then employed elastic constitutive models of rock materials using coupling genetic
to recognize its coefficients. Thus, the models are recognized programming with particle swarm optimization algorithm. Int J.
in a self-organization manner. The obtained non-linear stress– Rock Mech. And Min. Sci. 43(5): 789∼801.
strain relationship can not only satisfy the dynamic change in Feng X.T., Chen B.R., Yang C.X., et al. 2006. Intelligent analysis
of rheological characteristic of rock materials. In Multiphysics
its structure but also its variables and coefficients. coupling and long term behaviour in rock mechanics, 275–280,
The proposed recognition methods have been used in Liège: Belgium.
macro-mechanical modeling of the non-linear behavior of Feng X.T., Katsuyama K., Wang Y.J. et al. 1997. A new direction –
composite materials and visco-elastic rock materials (Feng & Intelligent Rock Mechanics and Rock Engineering, International
Yang 2004, Feng et al. 2006). Journal of Rock Mechanics and Mining Sciences, 34(1): 135–141.
For nonlinear rock materials, the structure of the non-linear Feng X.T., Li S.J., Liao H.J. et al. 2002. Identification of nonlinear
constitutive models is a key to control the non-linear behav- rock-like material constitutive models using genetic algorithm.
iors. There are different ways to representation the nonlinear International Journal for Numerical and Analytical Methods in
models implicitly of explicitly. The selection of the model- Geomechanics, 26(8): 815∼830.
ing methods may depend on the available knowledge about Feng, X.T. & Yang, C.X. 2004. Coupling recognition of the structure
and parameters of nonlinear constitutive material models using
the studied objects (Feng & Hudson 2010). For those newly hybrid evolutionary algorithms. Int J. for Numerical Methods in
observed constitutive factors, without enough support of cer- Engineering. 59: 1227∼1250.
tain theories, the NN-based nonparametric modeling method Yang, C.X., Feng, X.T. & Chen, B.R. 2008. Modeling creep behavior
may be more useful with no need to derive the mathematic of soft rock using an evolutionary neural computing method. In
expression based on available constitutive laws. While the J.H. Fan & H.B. Chen (eds.), Advances in Heterogeneous Material
structure of the models can be determined firstly according Mechanics 2008, 456–459, Lancaster: DEStech Publications.
789

Verification of reinforcing effects of a tunnel face reinforcement method

by centrifuge model tests and numerical analysis
Y. Yokota, K. Date & T. Yamamoto

Department of Geotechnical Engineering and Engineering Geology, Kajima Technical Research Institute, Tokyo, Japan
Keywords: bolts and anchors, lab testing, numerical modeling, rock support, tunneling, weak rock
1 EXTEND ABSTRACT their excavation performance. In recent years, small-diameter

steel pipe bolts have been employed most frequently. However,
In the past, the conventional method for excavating special in the authors’ view, the steel pipe bolts that are currently used
ground having a low competence ratio and high expandability, have a smooth surface, so they cannot provide the bonding
was to make excavation areas small while stabilizing faces and strength around steel pipes, and they do not have the same
crowns. However, that method was unable to prevent excessive effects as GFRP bolts.
displacement, and timbering frequently had to be replaced. In To address this problem, the authors carried out a tunnel
recent years, full-face excavation has been increasingly carried excavation test in a centrifugal force field and parametric
out on such unstable ground so as to restrain the final displace- studies with 3D-numerical analysis method. According to our
ment by early closure. In full-face excavation, it is important studies, it demonstrated that bonding strength was influential
to employ a forepoling method in combination with a face on failure pattern and ground movement. We subsequently
reinforcement method that utilizes long facebolts in order to developed new facebolts with checkered steel surface which
stabilize the face. can present much larger bonding strength. Furthermore, this
As reinforcing materials, GFRP (Glass Fiber Reinforced paper describes an actual employment of these new bolts to a
Plastic) bolts have been widely used in a face reinforcement tunnel with strongly weathered breccia.
method that utilizes long facebolts that are selected based on
Figure 2. Newly-developed checkered

Figure 1. Results of centrifuge model tests (failure patterns). steel pipe.
Figure 4. Tension-displacement curve obtained

Figure 3. Results of numerical analysis (Distribution of Maximum shear strain). from in-situ pull-out tests.
790

Author index
Abbasi, B. 224 Barton, N.R. 3 Chan, P.-C. 406

Abdulin, I.L. 285 Bashari, A. 771 Chan, Y.-C. 740
Abdulin, I.M. 167 Batalha, C.R.V. 297 Chang, K.-J. 740
Ablyamitov, N. 749 Bauer, C. 549 Chang, Y. 165
Abrah, B. 689 Bayerl, M. 259 Chanyshev, A.I. 167, 285
Adikaram, N.M. 356 Beck, D.A. 223 Chau, K.T. 391
Afifipour, M. 721 Beiki, M. 771 Chehreh, H. 717
Ahangari, K. 446 Bell, W. 580 Chen, B.R. 788
Ahmadi, M. 350, 721 Bellendir, E. 749 Chen, C.-H. 476
Ahn, C. 542 Belousova, O.E. 167 Chen, C.-N. 625
Ahn, J.R. 478 Bennani, M. 404 Chen, C.H. 587
Airey, D.W. 338 Béraud, J.F. 622 Chen, J. 386
Ajamzadeh, H. 177 Bérest, P. 622 Chen, J.J. 753
Akiyama, M. 569 Berger, J. 338 Chen, L. 556
Aksoy, C.O. 664 Bergmann, P. 656 Chen, L.-H. 625, 626
Akutagawa, S. 403 Bernardo, F. 413 Chen, S.H. 543
Alejano Monge, L. 593 Bewick, R.P. 376 Chen, Y. 292
Alejano, L.R. 570, 691 Beyhan, S. 511 Chen, Y.-C. 625, 626
Alonso, E.E. 672 Bhardwaj, V. 176 Chen, Z. 715
Alramahi, B. 295 Bhasin, R. 609 Cheng, C. 307
Alvarez-Fernandez, M.I. 471 Bian, Y. 362 Cheng, L.J. 766
Alvarez-Vigil, A.E. 471 Bich, G. 749 Cherkez, E. 749
Alves, C.A.S. 378, 380 Bigarré, P. 404 Chikahisa, H. 408
Amaral, C.S. 629 Binazadeh, Kh. 717 Chiu, Y.-C. 697
Amberg, F. 621 Blaisonneau, A. 524 Cho, H. 726
Amini, M. 773 Blankenship, D. 522 Cho, J.-W. 293
An, X.M. 497, 693 Borchardt, N. 738 Cho, S.H. 478
Anagnostou, G. 163, 623 Bost, M. 473 Cho, W.J. 323
Andrade Pais, L.J. 297 Bouazza, A. 209 Choi, S.K. 537
Antolini, F. 695 Boulamatsi, S.A. 573 Choi, S.O. 578
Aoki, K. 345, 638 Boulon, M. 291 Chong, W.L. 699
Aoki, T. 663 Bräuer, V. 554 Choudhury, P.B. 162
Arai, S. 458 Brouard, B. 622 Christiansson, R. 419, 423
Araújo, N. 719 Brown, E.T. 29 Chugh, Y.P. 224, 580
Armand, G. 157, 291 Brown, J. 254 Chung, S.-K. 535, 627
Arpaz, E. 472, 511 Brzovic, A. 313 Cogollo, C. 747
Arsyad, A. 521 Bulychev, N.S. 744 Conde, M. 782
Asakura, T. 490, 787 Bussard, T. 248 Conil, N. 157, 291
Ashraf, M. 260, 675 Contrucci, I. 404
Athanasopoulos, S.D. 159 Cabrera, J. 309 Cosenza, P. 309
Auvray, C. 585 Cabrera, R.L. 198 Cosgrove, J.W. 169
Aydan, Ö. 161, 463, 773 Cai, M. 575 Crawford, B. 295
Aziz, N. 572, 577 Calcina, A.M. 738 Cruchaudet, M. 157
Calçada, R. 501 Cui, J.J. 761
Baafi, E. 601 Camargo, M.A. 738 Cui, Y.J. 373
Babendererde, L. 679 Canelli, L. 474 Curry, D.A. 747
Bae, S.-H. 535 Cantieni, L. 163, 623
Bagde, M.N. 162 Cao, C. 577 da Costa, A.M. 629
Bahrani, N. 376 Cao, S.F. 556 da Fontoura, S.A.B. 747
Bai, E.L. 213 Carnero, F. 593 Dai, F. 480, 734
Bai, M. 743, 765 Carranza-Torres, C. 586 Dale, E. 763
Bandis, S.C. 668 Carter, T.G. 679 Danjo, T. 409
Barla, G. 695 Cassidy, M.J. 714 Date, K. 790
Barla, M. 695 Cavus, M. 511 Dayal, R.K. 176
Barnichon, J.D. 549 Çavuş, M. 472 de Greef, V. 622
Baroudi, H. 404 Cha, S.-H. 726 de los Santos, C. 313
791

DeDontney, N. 295 Fujii, Y. 415 Ha, T. 186
Deev, P.V. 744 Fukuda, T. 173 Ha, T.W. 235
Dehkordi, M.S. 370 Fukui, K. 306 Haba, T. 403
del Olmo, D. 700 Fukushima, Y. 447 Hadadou, R. 404
Deleruyelle, F. 550 Funahashi, T. 490 Hadei, M.R. 299
Delisio, A. 631 Funatsu, T. 231 Hadj-Hassen, F. 232
Dellero, H. 691 Furukawa, Y. 633 Hadjigeorgiou, J. 755
Deng, J.Q. 214 Haftani, M. 717
Deramore Denver, L. 305 Gabriel de Almeida, P. 297 Hagan, P. 305
Descamps, F. 296 Gaillot, P. 295 Haghnejad, A. 446
Dick, P. 549 Gainey, A. 338 Hakala, M. 434
Dickmann, T. 410 Galván Liévano, V.R. 751 Hakami, E. 423
Didier, C. 611 Gamwo, I.K. 254 Halakatevakis, N. 392
Diek, A. 522 Gao, E.X. 516 Hamzehpour, H. 717
Djamaluddin, I. 710 Gao, R. 225 Han, J.L. 245
Doghozlo, H.M. 441 García-Bastante, F. 570 Han, K.C. 757
Dong, J.-J. 211 Gardner, L.J. 583 Haque, A. 209, 699
du Plessis, M. 581 Garitte, B. 634 Harrison, J.P. 755
Du, X. 225 Gasc-Barbier, M. 268 Harte, N.D. 641
Duan, Q.W. 452 Gatmiri, B. 585 Hashiba, K. 306
Dubois, L. 473 Gattass, M. 629 Hassani, F. 759
Dudt, J.P. 388 Gautama, R.S. 592 Hatano, T. 458, 781
Dudoignon, P. 309 Gaziev, E.G. 704 Hatherly, P. 775
Duffaut, P. 701 Ge, X.R. 174 Hatsuyama, Y. 204
Dufour, N. 550 Geni˛s, M. 463 Hatzor, Y.H. 608
Dwivedi, R.D. 632 Gens, A. 634 Hayashi, H. 638
Dyskin, A.V. 171 Gentier, S. 524 He, L. 237
Geršak, A. 448 He, M.C. 307, 486, 761
Eberhardt, E. 730 Gharouni Nik, M. 416, 428 Hebblewhite, B. 305
Eff-Darwich, A. 310, 425 Ghavami, N. 441 Hedan, S. 309
Eghbal, M. 703 Ghazal, R. 232 Heissenberger, R. 442
Eickemeier, R. 227, 239 Ghazvinian, A. 115, 299, 350 Hernández, L.E. 310, 425
Eitzenberger, A. 482 Gheshmipour, A. 717 Hertz, E. 622
Elias, E. 305 Ghosh, C.N. 584 Hesami, M. 354
Eltsov, I.N. 194 Giese, R. 779 Hesser, J. 554
Epov, M.I. 194 Gineys, J. 473 Heusermann, St. 227, 239
Eslami, J. 303 Giot, R. 585 Hingston, E.D.C. 708, 712
Esteves, C. 193 Giraud, A. 265, 585 Hiroshima, T. 403
Giwelli, A.A. 301 Hiroto, K. 592
Fahimifar, A. 702, 746, 784 Glamheden, R. 417, 419 Hiroyoshi, N. 348
Fahland, S. 227 Goel, R.K. 176, 632 Hökmark, H. 241, 558
Fairhurst, C. 81, 109 Goh, A.T.C. 684 Hong, E.-J. 184
Fälth, B. 419 Golitsyn, V. 749 Hormazabal, E. 586
Fan, L. 411 Gomes, A. 552 Hosoda, T. 530
Fan, L.F. 484 Gómez-Márquez, I. 691 Hou, M.Z. 564
Faramarzi, L. 689 Gong, W.L. 486 Hou, Z. 526, 528
Fatehi Marji, M. 703 Gonzalez-Garcia, A.J. 81 Hoxha, D. 265, 303
Fauchille, A.L. 309 Gonzalez-Nicieza, C. 471 Hu, C. 650
Feng, Q. 412 González del Álamo, J.A. 297, 636 Hu, J.C. 761
Feng, X.-T. 81, 499, 681 Goodarzi, M. 441 Hu, Y. 515
Feng, X.T. 788 Görgülü, K. 472 Huang, T.-H. 476, 697
Fernandes, R. 265 Gorjian, M. 177 Huang, T.H. 179
Ferreira Gomes, L.M. 297 Götz, J. 656 Huang, X. 237
Ferrero, A.M. 229, 474 Grasselli, G. 706 Hudson, J.A. 47, 169, 434, 681
Figueiredo, B. 413 Greet, A.J. 708 Hug, R. 163, 623
Fitzenz, D. 532 Grgic, D. 303 Huh, D.-H. 184
Flansbjer, M. 313 Grossmann, N.F. 421 Hyun, C.-U. 357
Forlani, G.F. 229 Guan, F.H. 654
Fortin, S. 730 Guo, J. 283 Ichihara, Y. 415
Fotieva, N.N. 744 Guo, J.H. 715 Ijiri, Y. 204
Fredriksson, A. 417 Guo, L. 250 Ikemi, H. 521, 710
Freiberg, E. 749 Guo, L.Q. 753 Illankoon, T.N. 356
Fu, B.J. 562 Gurley, H. 580 Imani, M. 702
Fuenkajorn, K. 509 Gutierrez, M. 233 Inagaki, D. 204
792

Inohara, Y. 458, 781 Kawahara, T. 408 Labiouse, V. 561
Inoue, N. 747 Kawai, T. 709 Laigle, F. 265
Ishijima, Y. 415 Kawasaki, H. 445 Lamas, L. 413
Ishimaru, M. 709 Kawasaki, S. 409, 569 Lamas, L.N. 501
Ito, T. 312 Kazerani, T. 489 Lan, H. 257
Iura, T. 490 Ke, D. 556 Latham, J.-P. 250
Ivanova, A. 656 Kemppainen, K. 276, 434 Lato, M.J. 763
Iwano, M. 663 Kerris, S. 602 Laughton, C. 649
Iwasaki, T. 457 Keshavarz, M. 291 Lazemi, H.A. 370
Iwata, N. 181 Keyter, G.J. 641 Laín Huerta, C. 593
Khademian, Z. 428 Laín Huerta, R. 593
Jacobsson, L. 313 Khalili, N. 281 Le Bron, K.B. 188, 595
Jafari, A. 266 Kham-ai, D. 427 Leal e Sousa, R. 499, 552
Jaime, M.C. 254 Khoshrou, H. 503 Lee, C.-H. 697
Jalalifar, H. 572 Kikuchi, K. 369, 787 Lee, C.S. 323
Jang, H.-D. 615 Kim, B.-H. 376 Lee, H.-H. 382
Jang, M.-H. 615 Kim, H. 293, 437 Lee, H.G. 235
Javadi, M. 539 Kim, H.M. 757 Lee, S.-C. 184
Jeng, F.-S. 382, 406 Kim, J.-G. 615 Lee, Y.-K. 261, 325
Jeon, S. 293, 437 Kim, J.S. 323 Lee, Y.K. 676
Jeong, W.-C. 535, 627 Kim, K.-S. 460 Lemos, J.V. 91, 501
Jia, N. 710 Kim, S.-J. 615 Lenti, L. 473
Jia, P. 182 Kim, T.-K. 535, 627 Leoni, L. 730
Jia, X.N. 486 Kim, T.H. 757 Lheur, C. 622
Jia, Z.X. 452 Kim, W.-B. 186, 615 Li, A.J. 190, 714
Jiang, J.-C. 711 Kimura, S. 447 Li, B. 327, 488, 654
Jiang, J.C. 396 Kinoshita, N. 545 Li, C.C. 492, 605
Jiang, W. 241, 243 Kishida, K. 530, 545 Li, D.Y. 191
Jiang, W.J. 556 Kitaoka, T. 247 Li, G. 765
Jiang, Y. 173, 327, 488 Kitayama, N. 408 Li, H.B. 494
Jiang, Y.D. 245, 343, 398 Kitou, N. 638 Li, H.Q. 384
Jiang, Y.Z. 315 Kiyama, T. 415 Li, J.C. 494
Jiménez, R. 636, 639 Kizaki, A. 301, 339, Li, L. 252, 516
Jing, F. 587 453, 648 Li, L.P. 650
Johansson, E. 434 Klein, E. 404 Li, M.T. 616
John Loui, P. 584 Kobayashi, A. 249 Li, P.C. 531
Johnson, R.A. 188, 595 Kodama, J. 415 Li, Q. 328
Jones, B.R. 712 Kodikara, J. 209 Li, S. 488
Joo, K.-S. 184 Kohno, M. 320 Li, S.C. 616, 650, 652, 652
Josien, J.P. 404 Koike, M. 204 Li, W.S. 587
Judeel, G. du T. 641 Kojima, Y. 490, 659 Li, X.B. 191
Juhlin, C. 656 Koliji, A. 248 Li, X.H. 715
Juhojuntti, N. 656 Konecny, P. 491 Li, Y.P. 682
Jung, J. 317 Konečný, P. 591 Li, Z.K. 562, 766
Konicek, P. 491 Liang, W. 496
Kadiri, K. 643 Konstantas, I. 263 Liao, J.B. 753
Kaewkongkaew, K. 427 Koyama, T. 249 Liao, J.J. 513
Kaiser, P.K. 376, 575 Kramadibrata, S. 465, 713 Lima, C. 193
Kaneko, K. 319, 348, 478 Kubota, S. 445 Lin, D.M. 330
Kaneko, T. 663 Kumagai, T. 280 Lin, G.-L. 626
Kang, H.P. 589 Kumakura, S. 415 Lin, J. 589
Kang, M.S. 478 Kumar, Rakesh 187 Lin, J.-S. 254
Kang, S.-S. 360 Kunz, H. 554 Lin, M.-L. 740
Kanji, M.A. 430, 644, 751 Kuroda, Y. 546 Lin, S.-S. 211
Kannan, G. 646 Kuruppu, M. 318 Lin, W. 439
Karami, M. 689 Kusaka, E. 292 Lisjak, A. 706
Karino, Y. 339 Kusuda, H. 292 Liu, B. 652
Kashubin, A. 656 Kusuma, G.J. 592 Liu, G.J. 486
Käsling, H. 259 Kusumi, H. 247, 444 Liu, H.Y. 496
Kataoka, M. 318 Kuula, H. 434 Liu, J.Z. 213
Katayama, T. 249 Kuze, K. 521 Liu, M.J. 343
Kato, J. 317 Kwaśniewski, M. 321 Liu, Q.W. 785
Kato, M. 319, 542 Kwon, S. 323 Liu, Y.K. 587
Kavur, B. 432 Kyoya, T. 317 Liu, Y.M. 556
793

Liu, Y.R. 214, 654 Merifield, R.S. 190 Ogata, Y. 478
Liu, Z.X. 191 Mese, A.I. 534 Ohnishi, Y. 249
Lönnqvist, M. 241, 558 Migliazza, M. 474 Ohta, H. 660
Loock, S. 473 Migliazza, R. 229, 570 Ohta, R. 408
Lourenço, P.B. 380 Min, K.-B. 293 Ohta, Y. 161
Lu, A.Z. 616 Minami, Y. 403 Okamoto, Y. 408
Lu, D.T. 531 Miranda, T. 378, 380, 499, 719 Okano, N. 659
Lu, Q.C. 214 Miskovsky, K. 368 Okatsu, K. 301
Lu, T. 332 Missal, C. 678 Okubo, S. 306, 345
Lu, Y. 257, 332 Mitani, Y. 521, 633, 710 Okui, Y. 660
Lunn, J. 658 Mito, Y. 260, 546, 675 Olalla, C. 425, 700
Luo, X. 526, 768 Mitri, H.S. 394 Olofsson, I. 417
Lüth, S. 656, 779 Miwa, T. 458, 781 Olsson, R. 609
Lv, X.C.C. 213 Miyabayashi, H. 490 Onuma, T. 280
Lyamin, A.V. 714 Miyazaki, K. 345 Ootsuka, Y. 440
Łydżba, D. 769 Moarefvand, P. 721 Oparin, V. 196
Lyu, S.-J. 406 Mohlfeld, M. 678 Osada, M. 356
Molladavoodi, H. 346 Oteo, C.S. 661, 724
Moon, S.-J. 184 Otsuka, I. 663
Ma, C.D. 191
Moosavi, F. 441 Oura, S. 521
Ma, F.S. 283
Moosavi, M. 177, 441 Ozacar, V. 664
Ma, G.C. 440
Morgan, E. 763 Ozer, S.C. 664
Ma, G.W. 99, 237, 484, 494,
Morimoto, K. 348
497, 693
Moritz, B. 442
Ma, L.K. 556 Pamplona, J. 380
Morris, T. 338
Ma, X.D. 255 Panthi, K.K. 666
Mortazavi, A. 346
Mabuchi, M. 292 Park, B.-S. 726
Mostafavi, S.S. 658
Maeda, H. 320 Park, C.-S. 357
Mulder, A. 349
Mah, J. 334 Park, Chan 460
Muralha, J. 413
Mahabadi, O.K. 706 Park, Chulwhan 460
Murata, O. 440
Majdi, A. 771, 773 Park, D. 17
Maji, V.B. 336 Naidoo, K. 708, 712 Park, D.-I. 726
Malan, D.F. 581, 597, 599 Nakamura, M. 247, 444 Park, E.-S. 535, 627
Malmgren, L. 507 Nakamura, Y. 478 Park, H. 319, 359
Manchao, H. 552 Nakano, T. 445 Park, H.-D. 357
Mandal, P.K. 584 Nakashima, S. 445, 530, 545 Park, J.-W. 261, 325
Maritz, J.A. 599 Nakata, C. 660 Park, K.-H. 184
Martha, L.F. 629 Nakata, E. 415 Park, Y.-H. 360
Martin, C.D. 257 Nanda, A. 646 Paschalidis, K.P. 668
Martinez, I.M.R. 747 Nara, Y. 348 Pasternak, E. 171
Martins, F. 499, 719 Narita, T. 415 Pedroza, R.A.O. 777
Masoudian-Saadabad, M. 338 Nawrocki, P.A. 776 Pehovaz, H.I. 603
Masunari, T. 457 Nazarov, L.A. 194 Peixoto, A. 499
Mathier, J.F. 388 Nazarova, L.A. 194 Pellet, F.L. 291
Matsui, K. 216, 592 Nejati, H.R. 299 Pereira, A.M.B. 629
Matsuki, K. 301, 339, 453, 648 Nekoovaght, P.M. 759 Perera, M.S.A. 537
Matsuoka, T. 546 Nemcik, J. 572, 577, 601 Perucho, A. 782
Matt, R. 442 Neves, J. 193 Peter-Borie, M. 524
Mauke, R. 678 Ngan-Tillard, D.J.M. 349 Petkovšek, B. 448
Maupin, V. 609 Nie, L.C. 652 Peura, J.J. 263
Maury, V. 532 Nie, W. 307 Phien-wej, N. 427
Maya, J.M. 661 Nikkhah, M. 350 Piau, J.-M. 532
McKinnon, S.D. 334 Ning, Y.J. 497, 693 Pierce, M.E. 109
Mehinrad, A. 717 Nipp, H.-K. 239 Pilgerstorfer, T. 450
Mehrzad, B. 746 Nishimura, T. 560 Pirulli, M. 722
Meireles, M.C. 738 Noiret, A. 157, 585 Pittard, F. 410
Melentijevic, S. 700 Nomikos, P.P. 159, 352, 573 Pittino, G. 362
Melkumyan, A. 775 Noorani, R. 354 Plassart, R. 265
Melmeth, C. 658 Noorzad, A. 446 Poiate, E. 629
Memarian, H. 177 Nordlund, E. 482, 507 Pöllä, J.P. 263
Mendes, E. 297 Nozawa, T. 403 Pomares, M.J. 425
Mendieta, H.J. 341 Nyberg, U. 507 Porter, I. 601
Mendoza, J.A. 254 Poulard, F. 611
Meng, L. 343 Obara, Y. 318, 360, 447 Prashant, 584
Menschik, F. 259 Occhiena, C. 722 Prestor, J. 448
794

Prieto, L. 700 Samaniego, A.A. 198 Sofianos, A.I. 159, 352, 392, 573
Ptacek, J. 491 Samson, C. 334 Sommer, R. 202
Putzar, G. 223 Samsri, P. 509 Song, J.-J. 261, 325
Sandrone, F. 454 Song, L. 328
Qi, Z. 776 Santana, C. 593, 743 Song, S.G. 650
Qian, Q.H. 57, 685 Sanz, P. 295 Song, S.W. 562
Qu, D.A. 682 Saptono, S. 713 Song, W.K. 676
Rachez, X. 524 Sarfarazi, V. 299 Soroush, H. 541, 784
Radončić, N. 670 Sarmadivaleh, M. 455 Soto, C. 586
Radziszewski, P. 759 Sasa, K. 369, 538 Sriapai, T. 509
Rafiai, H. 266 Sasaki, T. 181 Stacey, T.R. 137
Rafiee, A. 268 Sasaoka, T. 216, 592 Stahlmann, J. 678
Rahimi Dizadji, M. 271 Sato, A. 369, 538 Stambuk Cvitanovic, N. 432
Raina, A.K. 162 Sato, H. 530 Stephansson, O. 269
Rakesh Kumar, 187 Satoh, A. 369 Suarez-Burgoa, L.O. 372
Ramamurthy, T. 115 Sawada, A. 530 Sueoka, M. 660
Ramon, A. 672 Sawada, K. 440 Sugawara, T. 415
Ramos da Silva, M. 296 Sawada, M. 415, 458, 781 Suh, Y.-H. 184
Ramses, W. 455 Saydam, S. 305 Sui, B. 650
Ramírez Oyanguren, P. 593 Sazid, Md. 505 Sulistianto, B. 592, 713
Ranjith, P.G. 209, 537, 699 Scavia, C. 722 Sun, C. 606
Rao, K.S. 371, 736 Schroeder, C. 296, 643 Sun, H.L. 715
Rasouli, V. 455, 541 Schubert, W. 127, 450, 670 Sun, J.P. 732
Rath, R. 646 Segalini, A. 474 Sun, L. 278
Rathod, G.W. 371, 736 Seiki, T. 369 Sun, X. 207
Read, S.A.L. 364 Sekine, K. 312 Suorineni, F.T. 376
Rechitski, V.V. 704 Sekino, H. 301 Suzuki, M. 633
Rechlin, A.J. 779 Semyonov, V.N. 285 Swarup, A. 176
Recio, D. 639 Serrano, A. 782 Swift, G. 386
Ren, A.W. 452 Severin, J. 730 Synn, J.-H. 460, 535, 627
Ren, F. 484 Shabanimashcool, M. 605
Ren, G. 602 Shahinuzzaman, A. 699 Taboada, J. 570
Ren, X. 515 Shahriar, K. 370, 428 Tada, M. 660
Resende, E. 193 Shahrour, I. 543 Takada, M. 415
Resende, R. 501 Shang, Y.J. 200, 330, 562 Takahashi, K. 249
Ribeiro e Sousa, L. 499, 552 Shao, H. 554 Takahashi, M. 319, 359, 542
Richards, L. 364 Sharifzadeh, M. 539 Takahashi, T. 173
Rinne, M. 269, 276 Sharma, K.G. 187 Taki, H. 663
Rocher-Lacoste, F. 473 Shen, B. 269, 276 Tal, Y. 608
Rodríguez, P.C.E. 777 Shen, X. 743 Tanabashi, Y. 488
Rodriguez, R.J. 255 Shi, G.H. 273 Tang, A.M. 373
Rodríguez-Losada, J.A. 310, 425 Shi, S.H. 650 Tang, C.A. 182
Rodríguez-Oitabén, P. 321 Shi, Y.Y. 200 Tano, H. 161, 463
Roegiers, J.-C. 522 Shidahara, T. 458, 781 Tao, J. 753
Roehl, D. 629 Shimada, H. 216, 592 Tarasov, B.G. 374
Romana, M. 366, 728 Shimizu, N. 231, 445, 457 Tatone, B.S.A. 706
Roncella, R. 229 Shimo, M. 663 Tchujko, E. 749
Röshoff, K. 412 Shin, K. 458, 781 Tenma, N. 345
Różański, A. 769 Shin, T. 447 Terada, K. 317
Russo, A. 586 Shin, Y.W. 235 Thuro, K. 259
Rutqvist, J. 280, 674 Shiote, T. 538 Tian, H.Y. 785
Ryu, D.W. 676, 757 Shirasagi, S. 675 Tian, S.B. 396
Shirzadegan, S. 507 Tijani, M. 232
Sadeghee, A. 503 Shono, T. 312 Tola, Y. 472, 511
Saharan, M.R. 505 Shou, K.-J. 626 Tsai, M.C. 179
Saiang, C. 368 Shou, K.J. 275 Tsesarsky, M. 462
Saito, H. 440 Shrivastava, A.K. 371 Tshering, T. 609
Sakaguchi, K. 301, 339, 453, 648 Simangunsong, G.M. 713 Tshibangu, J.-P. 296
Sakamoto, Y. 345 Singh, R. 459 Tsoi, P.A. 285
Sakkas, K.M. 352 Singh, T.N. 505 Tsuji, T. 444
Sakuma, S. 660 Sinha, A. 162, 584 Tsuno, K. 659
Sakurai, S. 663 Siren, T. 276 Tsuruhara, T. 660
Salari-Rad, H. 271 Smeallie, P.H. 681 Tsusaka, K. 204
Samaniego, A. 603 Šňupárek, R. 591 Tsutsumi, T. 375
795

Tu, X. 734 Wei, A.H. 283 Yiouta-Mitra, P. 392
Tutuncu, A.N. 534 Weng, M.-C. 382 Yokoi, K. 648
White, L. 522 Yokota, Y. 790
Ueda, A. 546 Whiteley, R.J. 466 Yokoyama, T. 447
Uehan, F. 440 Widodo, N.P. 465 Yoneda, T. 348
Ueng, T.-H. 406 Williams, D.J. 496 Yoo, S.Y. 546
Ujihara, M. 447 Wissler, R. 454 Yoshinaga, T. 447
Ulusay, R. 463 Wittke, W. 202 Yoshinakar, R. 181
Uripto, R.K. 465 Wohnlich, A. 248 Yoshioka, H. 638
Usoltseva, O.M. 285 Wong, H. 550 Yoshitake, H. 538
Utsuki, S. 787 Wong, J.Y.K. 684 You, S. 561
Uysal, Ö. 472, 511 Wong, L.N.Y. 191, 217, 384 Yu, H.Y. 275
Wong, R.H.C. 391 Yu, Q. 216
Valle, V. 309 Wu, D.S. 513 Yu, X. 753
Valley, B. 376 Wu, G. 386 Yun, X. 394
van der Merwe, J.N. 611 Wu, J.-H. 211 Yuvka, S. 511
Varadarajan, A. 187 Wu, W. 219, 388
Vardakos, S. 233 Wu, W.-J. 211
Vargas Jr., E. 552 Zeng, Y.W. 225
Wu, Z. 516 Zhang, H.M. 389
Vargas, J.C.A. 777 Wunder, E. 738
Varughese, A. 736 Zhang, J.J. 562
Wust-Bloch, G.H. 462 Zhang, J.T. 486
Vasconcelos, G. 378, 380
Vaskou, P. 646 Zhang, K. 396
Veiga, M. 691 Xia, C. 233 Zhang, L. 245
Verbrugge, J.-C. 296 Xia, K. 480 Zhang, L.C. 658
Verbrugge, J.C. 643 Xiang, J. 250 Zhang, L.L. 516
Verman, M. 679 Xie, H. 613 Zhang, N. 616
Vibert, C. 205 Xie, J.L. 556 Zhang, P. 482, 507
Vicente, J. 661 Xiong, J.H. 328 Zhang, Q.B. 388
Viete, D.R. 209 Xiong, Y. 515 Zhang, Q.S. 652
Vinches, M. 268 Xu, B. 766 Zhang, S.-S. 382
Viñas, R. 310 Xu, J.Y. 213 Zhang, W. 254, 330
Vöge, M. 763 Xu, Z.H. 650 Zhang, W.G. 684
Vogel, P. 239 Xue, L.L. 543 Zhang, X.P. 217
Vostrikov, V. 196 Zhang, Y.B. 182
Vrkljan, I. 432 Yamachi, H. 403 Zhang, Y.H. 411
Vuidart, I. 404 Yamada, H. 173 Zhang, Y.M. 283
Yamagami, T. 711 Zhang, Y.X. 245, 287
Wang, C.D. 513 Yamaguchi, T. 345 Zhang Na, 552
Wang, D. 776 Yamamoto, H. 280 Zhao, G. 734
Wang, D.G. 328 Yamamoto, K. 312 Zhao, G.F. 278, 281, 561
Wang, F. 761 Yamamoto, T. 260, 444, 675, 790 Zhao, H.-G. 554
Wang, G. 327, 412 Yamazaki, A. 292 Zhao, H.J. 283
Wang, J. 554, 556 Yan, J. 225 Zhao, J. 147, 219, 248, 278, 281, 489,
Wang, J.H. 589 Yanagawa, K. 638 494, 631, 734
Wang, J.S.Y. 681 Yang, C.H. 682 Zhao, M.J. 207
Wang, K. 207 Yang, C.X. 788 Zhao, T. 715
Wang, R.H. 315 Yang, G.S. 389 Zhao, W.F. 398
Wang, S.J. 217 Yang, H. 186 Zhao, X.G. 556
Wang, T. 398 Yang, H.-S. 615 Zhao, X.Y. 715
Wang, T.-T. 406, 476, 697 Yang, H.Q. 287 Zhao, Y.X. 179, 245, 343, 398
Wang, W. 515 Yang, H.S. 235 Zhao, Z.Y. 307, 497, 693, 732
Wang, W.J. 245 Yang, L. 488, 616 Zheng, H. 243
Wang, X. 250 Yang, Q. 214, 396, 654, 711 Zhigalkin, V.M. 285
Wang, X.G. 452, 516 Yang, X. 394, 496 Zhong, Z.W. 411
Wang, Y. 386 Yang, Z.-X. 625 Zhou, H. 613, 775
Wang, Y.G. 343 Yang, Z.Y. 179 Zhou, H.M. 411
Wang, Y.J. 452 Yao, Q. 658 Zhou, L. 528, 564
Wang, Z. 613 Yashima, A. 440 Zhou, X.P. 287, 685
Wang, Z.D. 213 Yashiro, K. 490 Zhou, Y. 254, 650
Wasantha, P.L.P. 209 Yasuhara, H. 530, 545 Zhou, Y.X. 147
Waters, K.E. 759 Yeh, C.-H. 740 Zhou, Z.Q. 650
Wattimena, R.K. 465, 713 Yin, P. 391 Zhu, J.B. 219, 388
796

Harmonising Rock Engineering and The Environment - Qihu - Qian

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Harmonising Rock Engineering and The Environment - Qihu - Qian

Uploaded by

Copyright:

Available Formats

© 2011 by Taylor & Francis Group, LLC