You are on page 1of 213

NUMERICAL ANALYSTS OF LARGE S@E

HORIZONTAL STRIP ANCHORS

A Thesis
Submitted for the Degree of

&lKasfer of 3 r i e n t e
L

in the Faculty of Engineering

BY
Y. S . R. KRISHNA

Department of Civil Engineering


INDIAN INSTITUTE OF SCIENCE
Bangalore-560 012, INDIA
July 2000
Dedicated
To
Lord Venkateswara
Acknowledgements

I express my deep sense of gratitude and sincere thanks to Pro$ B. R. Srinivasa Murthy,
Professor, Department of Civil Engineering and Dr. (Mrs.) A. Vatsala, Senior ~cient$c
Ojficer, Department of Civil Engineering, Indian Institute of Science, for their valuable
guidance, encouragement and support throughout the course of this study. Mere words
cannot describe the persona they are. Always kind, warm and affectionate, it is very pleasant
to work with them and I feel honored being associated with them. In short they are "epitome
of learning by example". I sincerely appreciate their deep understanding and constant care
they took throughout my stay at institute.

I sincerely thank Pro$ A. Sridharan, Prof T. S. Nagaraj, Pro$ K. S. Subba Rao,


Prof. N. S. Pandian Dr. Sivapullaiah, Dr. T. G. Sitharam, Dr. G. 1;. Sivakumarbabu,
Dr. SudhakarM Rao, Dr. Jyant Kumar, Dr. M. M. Allam, Dr. C. S. Manohau,
Dr. J. M. Chandrakishan and all other faculty members, Department of Civil Engineering,
Indian Institute of Science, who helped directly or indirectly during the present research
program.

I sincerely thank the authorities of Indian Institute of science, previous and present Chairman
of the Department of Civil Engineering, the past and present Convenors, non-teaching staf of
Soil Mechanics Laboratory and Civil Engineering ofice staff for providing the facilities for
the study and stay on campus.

I am very much thankjiul to Mr. P. Raghuveer Rao, Scientific Assistant, Department of Civil
Engineering, for his help in computational problems. X thank to my friends Giri, Mohen und
Parthasarathi, without their help the thesis would not have been in this form.

I wish to express thanks to Itasca Consulting Group, Inc. for providing FLAC software and
consistent support.

I acknowledge gratefully the cooperation extended by my friends and colleagues K.S.R


Murthy, C. Madhava Rao, Giri, Nandu, C. R. Parthasarathi, A. Srinivas, S. Minhaj,
K. C. Krishna, Mohan, Sitaram Naik, V. Sujatha, V.M Banahatti, Raghavendra Rao, Dinesh,
Venkataswamy, Tyagaraj, and Deepankar in all phases of this work and for making life at the
Institute memorable.

Mere words cannot express my sentiments and affectionate feelings towards my parents,
brothers, sisters-in-law, sisters, brothers-in-law for their encouragement, support and
cheerful cooperation.

Krishna. YSR
ABSTRACT

Structures like transmission towers, tele-communication masts, dry-docks, tall


chimneys, tunnels and burried pipelines under water etc are subjected to considerable uplift
forces. The net effect of external loading on the foundations of these structures results in
forces that try to pull the foundations out of the ground. Anchors are usually provided to
resist such uplift forces.

Earlier theoretical research of anchor behavior has focussed on elastic response and ultimate
pullout capacity. Many investigators have proposed techniques for determining the collapse
load of anchors. Essentially the approaches involve the use of limit equilibrium concepts,
with some assumptions regarding the shape of the failure surface andor the influence of the
soil above the anchor. The possible effect of dilatency and initial stress state are not
considered in these methods. A number of investigators have used the results of small size
model anchors to understand the behavior and extrapolated the results for predicting the
behavior of large sized anchors. This has lead to unsatisfactory results. It has been clearly
shown by Dickin (1989) that the failure displacements and load displacement curve patterns
are very different for small and large sized anchors, i.e. they are not just proportional to the
size of the anchor. Critical pullout load and the load displacement behavior are required for
the complete analysis of anchor foundations. Though, many theories have been proposed to
predict the uplift capacity within the limits of accuracy required at engineering level, at
present no simple rational method is available for computing deformations.

In the present investigation attempts have been made to analyze the load deformation
behavior of large size strip anchors in sands, clays and layered soils using two-dimensional
explicit finite difference program FLAC (Fast Lagrangian Analysis of Continua), well suited
for geomaterials, by assuming soil to be a Mohr-Coulomb material in the case of sands and
modified Cam-clay material in the case of clays.
It is now well understood that the shearing resistance of a granular soil mass is derived from
two factors frictional resistance and the dilatency of the soil. So the peak friction angle can be
divided in to two components critical friction angle (I$,,) and dilation angle (y). Critical
friction angle is the true friction angle as a result of frictional resistance at interparticle level
when the soil is shearing at constant volume. If $, for a given soil remains constant, the
value of y~has to increase with the increase in initial density of soil packing. The dilatency of
a soil mass gradually decreases with continued shearing from its initial high value to zero
after very large shear strains, when the soil finally reaches a constant, steady volume at
critical states. Correspondingly the observed friction angle (I reduces from its peak value to
(I, at a very large strains

In earlier days, clays used to be characterized by the strength parameters c and 4. Often,
under undrained conditions, 4 would be even considered zero. But in the recent
developments, it is understood that all the strength of clays is frictional. There is nothing like
cohesion. The part of shear strength, which appears to be independent of normal stress, is
shown to be the effect of over-consolidation and the resulting dilation. Thus although Cam-
clay model uses zero cohesion for all clays, it reflects this component of strength through
over-consolidation and in a more realistic way. Hence, it is appropriate to consider the pre-
consolidation pressure as parameter in the analysis.
More specifically, the various aspects covered in this investigation are as follows.

Chapter 1 provides the general introduction. In chapter 2, the existing literature for the
analysis of anchors for both experimental and analytical investigations on the pullout capacity
of anchors in homogeneous and layered soils and the load deformation behavior of anchors
under pullout are briefly reviewed.

Chapter 3, deals with the features and the implementation of the two dimensional explicit
finite difference program, Fast Lagrangian Analysis of Continua (FLAC) and the constitutive
modeling of soils. It discusses the background and implementation of Strain softening /
hardening model. This model is based on the Mohr- Coulomb model with non-associated
shear and associated tension flow rules. In this model the cohesion, friction, dilation and
tensile strength may harden or soften after the onset of the plastic yield. Further the critical
state concepts and implementation of the modified Cam-clay model have been discussed.
Cam-clay model originally developed for clays reflects the hydrostatic pressure or density
dependent hardening material response.
Chapter 4, focuses on the analysis of load deformation behavior of large size anchors in
granular soils. Two-dimensional explicit finite difference program (FLAC) is used for the
simulations and the soil is modeled as a Mohr-Coulomb strain softeninghardening material.
In this chapter a series of simulations have been carried out on large size anchor plates, with
parametric variation. By analyzing these results, a generalized load defoniiation relationship
for different sizes of anchors and different types of soil have been proposed. The results are
presented in the form of influence/design charts which can be used in hand calculations to
obtain an estimate of anchor capacity and deformation for a wide range of soil types and size
of anchors.

Chapter 5, deals with the analysis of the drained and undrained behavior of large size
horizontal strip anchors in clays using modified Cam-clay model. Earlier investigators have
studied the undrained behavior of anchor plates in clays, but no studies are reported in
literature for the drained behavior of anchors in clays. Further it is not clear whether, drained
or undrained condition will be critical for an anchor. In this chapter the drained and
undrained behavior of large size anchor plates in both normally consolidated and over-
consolidated states have been made. It has been found that the undrained pullout capacity of
an anchor in a soil of normally consolidated state will always be more than the drained
capacity. This is contrast to the usual understanding that undrained behavior is more critical
than the drained behavior.

In Chapter 6, an attempt has been made to analyze the behavior of large size anchors in two
layered sands and in conditions where backfill material has a higher or lower strength than the
native soil, for different shape of excavations. Soil is assumed to be a Mohr-coulomb strain
softeninghardening material.

In Chapter 7, the entire investigation covered in earlier chapters has been synthesized and
some specific conclusions have been highlighted.
CONTENTS
rNDEX Page No.

CHAPTER 1
INTRODUCTION

CHAPTER 2
LITERATURE REVIEW
2.1 Introduction
2.2 Pullout capacity of anchors in homogenous soils
2.2.1 Experimental Investigations on Anchors
2.2.2 Theoretical Analyses of Anchors
2.3 Anchors in Layered Soil
2.4 Load-Displacement Behavior
2.5 Summary and Scope of work

CHAPTER 3
NUMERICAL ANALYSIS USING FLAC
3.1 General
3.2 Fast Legrangian Analysis of Continua
3.2.1 Overview
3.2.2 Basic Calculation cycle of FLAC
3.2.3 Optional Features of FLAC
3.2.4 Numerical Techniques Used
3.2.5 Explicit Time-Marching Scheme
3.2.5.1 Comparison of explicit and implicit solutions
methods
3.2.6 Lagrangian Analysis
3.2.7 Field Equations
3.2.7.1 Motion and equilibrium
3.2.7.2 Constitutive Relation
3.2.7.3 Boundary Conditions
3.2.7.4 Finite difference Equations
3.2.8 Ground Water and Consolidation
3.2.8.1 Constitutive Law
3.2.8.2 Continuity equation
3.2.9 Numerical stability
3.2.9.1 Mechanical timestep
3 -2.9.2 Fluid timestep
3.2.10 Structural Elements
3.3 Constitutive modeling of soils
3.3.1 Constitutive Models Provided in FLAC
3.3.2 Mohr-Coulomb Model
3.3.3 Strain-SofteningkIardeningModel
3.3.4 Critical State Concepts
3.3.5 Modified Cam-Clay Model
3.4 Summary

CHAPTER 4
ANALYSES OF HORIZONTAL S T R P ANCHORS IN SAND
4.1 Introduction
4.2 Friction angle (4) and dilation angle ( y )
4.3 Formulation of the Problem
4.4 Validation of Numerical program FLAC
4.4.1 Properties of soil reported by Ramesh Babu (1998)
4.4.2 Comparison of results
4.4.3 Properties Reported by Rowe
4.4.4 Comparison of results
4.5 Analysis of large size anchors
4.6 Definition of Failure : The K2 failure concept
4.7 Load - Displacement behavior of Anchors
4.8 Generalization and Prediction of Load-Displacement behavior
4.9 Summary

CHAPTER 5
ANALYSIS OF HORIZONTAL S T R P ANCHORS IN CLAY
5.1 Introduction
5.2 Formulation of the problem
5.3 Analysis of large size anchors
5.4 Discussion on Simulation Test Results
5.4.1 Undrained Tests on Norrnally consolidated soils
5.4.2 Drained Tests on Normally consolidated soils
5.4.3 Drained Tests for over-consolidated clays
5.5 Summary

CHAPTER 6
ANALYSES OF HORIZONTAL STRIP ANCHORS ICN LAYERED SOILS
6.1 Introduction
6.2 Analyses of anchor behavior in two layered sand
6.3 Analysis of anchor plates with backfill having higher or lower
6.4 Summary

CHAPTER 7
RESULTS AND DISCUSSIONS

REFERENCES
LIST OF FIGURES

Figure No INDEX Page


No.

Fig 2.1 Anchor under pullout load


Fig 2.2 Failure surface assumed by Mors (1959)
Fig 2.3 Failure surface for Friction Cylinder Method
Fig 2.4 Failure Surface assumed by Veesaert and Clemence (1977)
Fig 2.5a Failure Surface assumed by Balla (1961)
Fig 2.5b Nature of variation of Nu with D/B (Balla; 1961)
Fig 2.6a Failure surface assumed by Mariupol'skii's
for shallow circular plate anchor
Fig 2.6b Failure surface assumed by Mariupol'skii's
for deep circular plate anchor
Fig 2.7 Failure surface assumed by Meyerhof and Adams (1968)
Fig 2.8 Failure surface assumed by Saeedy (1987)
Fig 2.9 Failure surface assumed by Subba Rao and Jyant Kumar (1994)
Fig 2.10a Layered soil system used in Stewart's (1985) experimental
investigation
Fig 2.1Ob Load-displacement curves of anchors in Layered soil system
Fig 2.1 l a Layered soil system used in Bouazza & Finlay, (1990) experimental
investigations
Fig 2 . l l b Ultimate pullout capacity against D,/B ratio
(Bouazza and Finlay; 1990)
Fig 2.12 Failure mechanism in two layered soil system by Manjunath (1998)
Fig 2.13 Load versus Displacement curves

Fig 3.1 Basic Explicit Calculation cycle


Fig 3.2 General Solution Procedure of FLAC
Fig 3.3a Mohr-Coulomb failure Envelop
Fig 3.3b Mohr-Coulomb and Tresca yield surfaces in principal stress space
Fig 3 . 3 ~ Intersection with n-plane
Fig 3.4 Mohr-Coulomb failure criterion in FLAC
Fig 3.5 Typical Variation of Cohesion, friction, dilation and tensile strength
with plastic strain
Fig 3.6 Normal consolidation line and unloading-reloading (swelling)
Fig 3.7 Yield surface of modified Cam-Clay Model in q-p space

Fig 4.1 Problem Analyzed


Fig 4.2 Finite difference grid used for analysis
Fig 4.3 Comparison of load-displacement curves with experimental results
Fig 4.4 Comparison of load-displacement curves with experimental results
Fig 4.5 Comparison of pullout capacity factor with Rowe's experimental
results
Fig 4.6 Comparison of load-displacement curves with Rowe's FEM analysis
Fig 4.7 Comparison of pullout capacity factors for strip anchors by various
theories
Fig 4.8 Comparison of pullout capacity factors for strip anchors by various
theories
Fig 4.9a Displacement vectors at failure in model anchors
Fig 4.9b Plastic regions of failure in model anchors
Fig 4.10 Definition of failure
Fig 4.1 1 Load-displacement curves for different D/B ratios, B = 1.00m, y = 0"
Fig 4.12 Load-displacement curves for different D/B ratios, B = 0.75m, y = 0"
Fig 4.13 Load-displacement curves for different D/B ratios, B = OSOm, y = 0"
Fig 4.14 Load-displacement curves for different D/B ratios, B = 0.50m, 11, = 5"
Fig 4.15 Load-displacement curves for different D/B ratios, B = 0.50m, y = 10"
Fig 4.16 Load-displacement curves for different D/B ratios, B = OSOm, y = 15"
Fig 4.17 Load-displacement curves for different dilation angles, B = 1 m, D/B=l
Fig 4.18 Load-displacement curves for different dilation angles, B = 1 In, D/B=2
Fig 4.19 Load-displacement curves for different dilation angles, B = 1 m, D/B=3
Fig 4.20 Load-displacement curves for different dilation angles, B = lm, D/B=4
Fig 4.2 1 Load-displacement curves for different dilation angles, B = lm, D/B=6
Fig 4.22 Load-displacement curves for different dilation angles, B = lm, D/B=8
Fig 4.23 Load-displacement curves for different dilation angles,
B = 0.75m, D/B = 1
Fig 4.24 Load-displacement curves for different dilation angles,
B = 0.75111, D/B = 2
Fig 4.25 Load-displacement curves for different dilation angles,
B = 0.75m, D/B = 3
Fig 4.26 Load-displacement curves for different dilation angles,
B = 0.75m, D/B = 4
Fig 4.27 Load-displacement curves for different dilation angles,
B = 0.75m, D/B = 6
Fig 4.28 Load-displacement curves for different dilation angles,
B = 0.75n-1,D/B = 8
Fig 4.29 Load-displacement curves for different dilation angles,
B = 0.50m, D/B = 1
Fig 4.30 Load-displacement curves for different dilation angles,
B = 0.50m, D/B = 2
Fig 4.3 1 Load-displacement curves for different dilation angles,
B = 0.50m, D/B = 3
Fig 4.32 Load-displacement curves for different dilation angles,
B = OSOm, D/B = 4
Fig 4.33 Load-displacement curves for different dilation angles,
B = 0.50m, D/B = 6
Fig 4.34 Load-displacement curves for different dilation angles,
B = OSOm, D/B = 8
Fig 4.35 Displacement vectors at failures in large size anchors, B = lm, y~ = 5"
Fig 4.36a Load-displacement curves
Fig 4.36b Displacement vectors and plastic regions at different load levels
Fig 4 . 3 6 ~ Displacement vectors and plastic regions at different load levels
Fig 4.36d Displacement vectors and plastic regions at different load levels
Fig 4.37a Displacement vectors and plastic regions at different dilation angles
Fig 4.37b Plastic regions at failure for different dilation angles
Fig 4.38 Displacement vectors and plastic regions at failure
for different dilation angles
Fig 4.39a Displacement vectors at failure for different dilation angles
Fig 4.39b Plastic regions at failure for different dilation angles
Fig 4.40 Variation of P, with D, y = 0"
Fig 4.4 1 Variation of P, with D, y = 5"
Fig 4.42 Variation of P, with D, y = 10"
Fig 4.43 Variation of P, with D, y = 15"
Fig 4.44 Variation of Z, with D, y = 0"
Fig 4.45 Variation of Z, with D, y = 5'
Fig 4.46 Variation of Z, with D, y = 10"
Fig 4.47 Variation of Z, with D, y~ = 15"
Fig 4.48 Normalized load-displacement curves for sands
Fig 4.49 Normalized plot of pullout capacity factors vs. dilation angle
Fig 4.50 Normalization of Z,,B % vs. D/B for different size of anchors and
dilation angles
Fig 4.5 1a Variation of pullout capacity factor with critical friction angle,
B = 0.50m
Fig 4.51b Variation of pullout capacity factor with critical friction angle,
B = 0.75m
Fig 4.5 1c Variation of Pullout capacity factor with critical friction angle,
B = 1.00m

Fig 5.1 Load-displacement curves in normally consolidated soil


for undrained case, B = 1.00m
Fig 5.2 Load-displacement curves in normally consolidated soil
for undrained case, B = 0.75m
Fig 5.3 Load-displacement curves in normally consolidated soil
for undrained case, B = 0.50111
Fig 5.4 Displacement vectors at failures for undrained conditions
Fig 5.5 Plastic regions at failure in normally consolidated soil
for undrained case
Fig 5.6 Load-displacement curves for normally consolidated soils
for drained case, B = 0.50m
Fig 5.7 Load-displacement curves for normally consolidated soils
for drained case, B = 0.75m
Fig 5.8 Load-displacement curves for normally consolidated soils
for drained case, B = 1.00m
Fig 5.9 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 1.00, D/B = 2
Fig 5.10 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 1.00, D/B = 4
Fig 5.1 1 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 1.00, DIE3 = 6
Fig 5.12 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 1.00, D/B = 8
Fig 5.13 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.75, D/B = 2
Fig 5.14 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.75, D/B = 4
Fig 5.15 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.75, D/B = 6
Fig 5. I6 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.75, D/B = 8
Fig 5.17 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.50, D/B = 2
Fig 5.18 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.50, D/B = 4
Fig 5.19 Load-displacement curves at different pre-consolidation pressure for
drained case, B = 0.50, D/B = 6
Fig 5.20a Variation of P, with D/B
Fig 5.20b Variation of P, with DLB
Fig 5 . 2 0 ~ Variation of P, with DLB
Fig 5.21a Variation of P, with B for normally consolidated soil
Fig 5.2 1b Variation of P, with B for PC= 25kPa
Fig 5 . 2 1 ~ Variation of P, with B for PC= 5OkPa
Fig 5.21 d Variation of P, with B for PC= lOOkPa
Fig 5.21e Variation of P, with B for PC= 200kPa
Fig 5.22a Variation of pullout capacity with D 5 for B = 1.00m
.Fig 5.22b Variation of pullout capacity with D/B for B = 0.75m
Fig 5 . 2 2 ~ Variation of pullout capacity with D 5 for B = 0.50m
Fig 5.23 Generalized load-displacement curve for drained case
Fig 5.24a Displacement vectors at failure for drained case at different pre-
consolidation
Fig 5.24b Plastic regions at failures for different pre-consolidation pressure
Fig 5.25a Displacement vectors at failures for drained case, B=l m, PC= 25kPa
Fig 5.25b Displacement vectors at failures for drained case, B=l m,PC= 25kPa
Fig 5.26 Plastic regions at failures for drained case, B=l m, PC= 25kPa

Fig 6.1 Problem analyzed


Fig 6.2 Load-displacement curves for layered soil, D/B = 4
Fig 6.3 Load-displacement curves for layered soil, D/B = 6
Fig 6.4 Load-displacement curves for layered soil, D/B = 8
Fig 6.5 Variation of P, with D/B
Fig 6.6 Variation of Z, with DIE3
Fig 6.7 Load-displacement curves for layered soil, D/B = 4
Fig 6.8 Load-displacement curves for layered soil, D/B = 6
Fig 6.9 Load-displacement curves for layered soil, DIB = 8
Fig 6.10 Variation of P, with D/B
Fig 6.1 1 Variation of Z, with D/B
Fig 6.12 Normalized load-displacement curves for layered soil
Fig 6.13a Displacement vectors and plastic regions at failure in layered soils
Fig 6.13b Displacement vectors and plastic regions at failure in layered soils
Fig 6.14a Displacement vectors and plastic regions at failure in layered soils
Fig 6.14b Displacement vectors and plastic regions at failure in layered soils
Fig 6.15 Shape of Excavation walls considered
Fig 6.16 Load-displacement curves for backfill with high or low strength than
the native soil, DO3 = 2
Fig 6.17 Load-displacement curves for backfill with high or low strength than
the native soil, DA3 = 4
Fig 6.18 Load-displacement curves for backfill with high or low strength than
the native soil, D/B = 6
Fig 6.19 Load-displacement curves for backfill with high or low strength than
the native soil, D/B = 8
LIST OF TABLES

Table No. INDEX Page No.

Table 4.1 Soil properties reported by Ramesh Babu (1998)


Table 4.2 Soil properties reported by Rowe (1982b)
Table 4.3 Comparison of Numerical results with experimental results
Table 4.4 Soil properties used in the analysis

Table 5.1 Model Properties of London Clay

Table 6.1 Soil properties used in the analyses


Table 6.2 Dense native sand with Loose sand backfill
Table 6.3 Loose native sand with Dense sand backfill

XIV
LIST OF SYMBOLS

Symbol Description

Area of anchor plate


Width of strip anchor or diameter of circular anchor
Soil cohesion
Fourth order tensor of compliance matrix
Un-drained cohesion
Diameter of shaft
Total embedment depth
Fourth order tensor of stiffness matrix
Relative density of compaction
Depth of the top layer
Modulus of elasticity
Strain rate components
Force
Mohr-Coulomb shear yield function
Factor of safety
Tension yield function
Shear modulus
Shear potential function
Tension potential function
Vertical extent of failure surface
Bulk ~nodulus
Earth pressure coefficient at rest
Nominal pullout coefficient of earth pressure on a convex cylindrical wall
Length of anchor
Frictional constant
Porosity
Number of steps
Pullout load
Mean principal stress
Initial pressure
Reference pressure
Pre-consolidation pressure
Critical pullout load
Pullout factor
Shearing resistance developed along the failure surface
Suction force below the anchor
Ultimate pullout load
Ultimate load capacity by two tangent method
Net average ultimate pullout capacity
Deviator stress
Time
Specific volume
Volume of the soil
Value of the specific volume at the reference pressure
Initial specific volume
Specific discharge vector
Volume of the solid particles
Effective weight of soil located in the failure zone
Weight of the anchor
Weight of the soil inside the cylinder
Anchor displacement at net pullout load P
Ultimate displacement
Unit weight of soil
Angle of internal friction
Apex angle
Angle of the failure surface with vertical at the edge of the anchor
Mass density
Shear stress
Dilation angle
Slope of the normal consolidation line
Slope of the elastic rebound line
Poisson's ratio

xvi
Major principal stress at failure
Intermediate principal stress at failure
Minor principal stress at failure
Critical frictional angle
Critical embedment ratio
Components of stress tensor
Normal stress
Pullout capacity factor for rectangular anchors
Plane strain friction angle
Peak frictional angle
Time step
Tensile strength
Triaxial friction angle
Recoverable change in specific volume
CHAPTER 1

INTRODUCTION

Structures like transmission towers, tele-communication masts, dry-docks, tall


chimneys offshore structures, suspension bridges, radar towers, aircraft's moorings, tunnels,
buried pipe lines under water etc., are subjected to considerable pullout forces. The net effect
of external loading on these foundations is a force that tries to pull them out of the ground.
Anchors are usually provided to resist such pullout forces. Anchors are also used as a
measure to increase the stability of the slopes. Anchors also used for tieback resistance of
earth-retaining structures, waterfront structures, at bends in pressure pipelines and when it is
necessary to control the stress. Different forms of anchors like plate anchors, screw anchors,
grouted anchors, anchor piles, drilled shafts, etc., are in common uses. Studies reported in
this thesis concern with the load deformation behavior of horizontal plate anchors.

Depending upon the relative depth of the embedment (DIB) of the plate in the soil the plate
anchors are classified as shallow anchors and Deep anchors. In the case of the shallow
anchors the anchor is installed close to the surface of the soil and the failure surface reaches
the ground surface with significant ground movements, and a small increase in embedment
depth results in a considerable increase in pullout load. Where in the case of deep anchors,
due to limiting displacement considerations the failure surface does not reach the ground
surface, but instead forms locally around the anchor. Besides any increase in depth of
embedment will not result in commensurate increase in the pullout load as in the case of
shallow anchors.

Many investigators have proposed theoretical techniques to estimate the pullout capacity of
anchors. Essentially the approaches involve use of limit equilibrium concepts, have generally
been based on some assumed geometry of failure surface. Further the possible effect of
dilatency, initial stress state and the effect of backfill having higher or lower strength than the
native soil is not been considered. A number of investigators used small size model anchors
in laboratory experiments and extrapolated the results for predicting the behavior of large-
scale anchors, which may lead to unsatisfactory results. It has been clearly shown (Dickin;
1989) that the failure displacements and load displacement curve patterns are very different
for small and large sized anchors, i.e. they are not just proportional to the size of the anchor.
Generally the proposed approaches in literature to estimate the ultimate pullout capacity of
anchors in homogeneous and isotropic soil conditions. Critical pullout load and the load
displacement behavior are required for the complete analysis of anchor foundations. Though,
many theories have been proposed to predict the pullout capacity with in the limits of
accuracy required at engineering level, at present no simple rational method is available for
computing deformations. Attempts have been made to study the deformation of anchors using
finite element technique (Rowe and Davis; 1982a&b). Swami Saran et a1 (1986) and Das and
Puri (1989) suggested a method based on hyperbolic load displacement response model.
Trautmann and Kulhawy (1988) and Ramesh Babu (1998) have proposed a statistical best-fit
approach.

Not much work has been done to study the behavior of anchor plates in Layered soils, the
problem often encountered by the practicing engineer in the field. Stewart (1985) and
Bouazza and Finlay (1990) investigated experimentally the behavior of circular anchor
embedded in two layered soils. Manjunatha (1998) and Ramesh Babu (1998) proposed
expressions to estimate the pullout capacity of strip and circular anchors in layered soils.

With this introduction, it is evident that the earlier studies on anchors are far from adequate.
It is attempted in this investigation to study some aspects to remove the inadequacies.

The organization of the thesis and the main features in each chapter are outlined below.

In Chapter 2, experimental and theoretical investigations on pullout capacity and load


deformation behavior of homogeneous and layered soils are briefly reviewed, and the
objective of the investigation are outlined.

Chapter 3, deals with the features and implementation of explicit two-dimensional numerical
program, Fast Lagrangian Analysis of Continua (FLAC). It also discusses the constitutive
models for soils and the implementation of Mohr-Coulomb model, Strain
SofteningIHardening model and modified Cam-Clay model. It also discusses the critical state
concepts.
In chapter 4, the vertical pullout capacity of a horizontal strip anchor has been modeled as a
plane strain problem, for theoretical analysis of load deformation behavior of large size
anchors. It deals with the formulation of the problem and validation of the numerical program
FLAC with experimental results of model anchors and some analytical solutions. It also deals
with the analysis of behavior of large size anchor plate in sands. At present the design of
anchor plates for pullout loads is based on ultimate pullout capacity considerations. There has
been only a limited effort to characterize the load-deformation response of these anchor
foundations. In this chapter a series of simulations have been carried out on large size anchor
plates, with parametric variation. By analyzing these results a generalized load deformation
relationship for different sizes of anchors and different types of soil have been proposed.

Chapter 5, deals with the analysis of behavior of large size anchors in clays using the
modified Cam-Clay model. It discusses the behavior of anchors in normally consolidated and
overconsolidated clays in both drained and undrained conditions. In earlier days, clays used to
be characterized by the strength parameters c and @. Often, under undrained conditions, @
would be even considered zero. But in the recent developments, it is understood that all the
strength of clays is frictional. There is nothing like cohesion. The part of shear strength,
which appears to be independent of normal stress, is shown to be the effect of over-
consolidation and the resulting dilation. Thus although Cam-clay model uses zero cohesion
for all clays, it reflects this component of strength through over-consolidation and in a more
realistic way. Hence, it is appropriate to consider the pre-consolidation pressure as parameter
in the analysis

Chapter 6, deals with analysis of large size strip anchors in two layered soils. A series of
simulations are made for bottom layer dense sand and top layer loose sand and vice versa.
Normalized load-displacement curves are presented for the above two conditions. Anchors in
backfill are very sensitive to construction techniques because they rely on the manner in
which the trench is excavated and filled up. Available methods in literature for the pullout
capacity of anchors are all based on backfilling material being identical with the native soil.
In this chapter an attempt has been made to analyze the behavior of anchors plates with
refilling with stronger or poorer material than the native soil.

Chapter 7 deals with the conclusions and remarks drawn from the analysis presented in
earlier chapters.
CHAPTER 2

LITERATURE REVIEW

2.1 Introduction

Anchors in soil are provided to re sist any type of outwardly direc:ted forces.
Foundations of tall skeletal structures like transmission towers, offshore structures, antenna
masts and other under ground structures like, dry-docks, tunnels, buried pipelines under water
etc., are subjected to tensile forces. Depending upon the requirements of the given structure,
different types of anchors such as plate anchors, direct ernbedmnet anchors, helical anchors,
grouted anchors, drilled shafts and anchor piles etc., can be adopted. Based on the orientation
of anchor plate, anchors can be classified as horizontal, vertical and inclined. Critical pullout
load and deformation under different loads are required for the complete analysis of anchor
foundations. Many theories have been proposed to predict the pullout capacity of these anchor
foundations in homogenous soils based on the results of the laboratory investigations and field
tests. A number of investigators have used small size anchors in their experimental
investigations and extrapolated the results to analyse and predict the .behavior of large size
anchors., Some researchers have developed expressions empirically to obtain the values of
shape factors for determining the pullout capacity of the anchors of all shapes from the results
of strip anchors. It is necessary to discuss all their works before formulating the scope of the
present work.

In the following paragraphs a review of the available literature on pullout capacity and load
deformation behavior of horizontal anchor foundations in homogeneous soils and layered
soils is presented.
2.2 Pullout capacity of anchors in homogenous soils

The general system of forces acting on a simple horizontal anchor is shown in Fig 2.1.
The pullout capacity is given by the general equation

Where
P, = ultimate pullout capacity
W = effective weight of soil located in the failure zone
P, = suction force below the anchor
P, = shearing resistance developed along the failure surface
In the case of sands P, is equal to zero.

Fig 2.1 Anchor under pullout load

2.2.1 Experimental Investigations on Anchors

Davie & Sutherland (1977) studied the effect of large soil strains and tensile stresses
in soil on vertical pullout forces of circular anchors in purely cohesive soils. Dimensional
analysis has been used to establish similarity conditions between prototypes and models, and
the relative contribution of body forces and soil strength to total pullout resistance are
examined. A series of 65 model pullout tests were conducted, comprising both pullout and
push out tests. Push out tests were conducted because greater control could be exercised over
the location of the anchor plate, and the soil above the plate could be better placed due to the
absence of the anchor shaft. The available theories, which do not consider the effects of large
soil strains and tensile stresses over estimate the actual ultimate pullout resistance.

Das (1978) conducted a number of laboratory model tests on circular, square and rectangular
anchors embeded in saturated clay with the undrained cohesion (c,) varying from 5.18 kPa to
172.5 kPa. It was found that the critical embedment ratio (D/B),, was a function of c,, . The
following empirical equations were proposed for the critical embedment ratio.

For square and circular anchors


(D/B),,., = (0.107 c,, + 2.5 ) _< 7

For rectangular anchors


(D/B)c,.R=1.6(D/B),,.,for U B > 4

Where
(D/B),,, = critical embedment ratio of square or circular anchors.
c, = undrained cohesion in kPa
L = length of rectangular anchor

Based on further model test results of Das (1980) modified the equation for critical
embedment ratio (D/B),,.R as
(D/B)cr.R= (D/B),,., (0.73 + 0.2 7 ( U B ) )_< 15.5 (D/B),,., (2.4)

Dickin (1988) studied the influence of anchor geometry, embedment depth and the soil
density on the pullout capacity of one-meter prototype anchor by subjecting 25mm models to
an acceleration of 40g in the Liverpool centrifuge. It was found that for strip anchors pullout
resistance expressed as dimensionless breakout factor increases significantly with anchor
embedment depth and soil density but reduce with increase in aspect ratio which is the ratio of
length to width of the strip anchor. Failure displacements also increase with embedment
depth but reduce with soil density and aspect ratio.
Frydman and Shaham (1989) performed a series of pullout tests on prototype slabs placed at
various inclinations and depths in dense sand. A simple semi empirical expression is found to
reasonably predict the pullout capacity of continuous, horizontal slab as a function of depth-
to-width ratio. Factors to account for shape and inclination are then established leading to
expressions, for the estimation of pullout capacity of any slab anchor.

The following expressions have been proposed for the pullout capacity of horizontal,
rectangular slab anchor in sand for dense sand

for loose sand, D/B22

Where (N4)ris the pullout capacity factor P,Jy D f x rectangular anchor

Krishnaswamy N.R and Suresh P.P (1992) conducted the model tests to investiga~ethe
various aspects of the pullout behavior of plate anchors with and without geo-synthetic
inclusions under different test conditions. Three different types of geo-synthetics
manufactured in India were used in their investigation. Tests were conducted in cohesive as
well as cohesionless soil medium under dry and submerged conditions. Results are presented
in the form of load-displacement curves. Based on the experimental results, similar
expression as proposed by Mayerhof and Adams (1968) is suggested for predicting the
ultimate pullout capacity of plate anchors with geo-synthetic inclusions. A simple laboratory
test is also suggested for determining parameter k, required in the proposed theoretical
method.

Rameshbabu (1998) investigated the pullout capacity and the load deformation behavior of
shallow horizontal anchors. Laboratory experiments have been conducted on anchors of
different shapes (square, circular and strip) and embedded in medium dense and dense sands.
In addition, the effect of submergence of soil above anchor has been investigated. Ashraf
Ghaly (1977) had recommended a general expression for pullout capacity of vertical anchors
based on statistical analysis of the experimental test results from published literature. On
7
similar lines and incorporating appropriate correction, Rameshbabu (1998) proposed a general
expression for horizontal anchors in sand by analyzing the results of published experimental
data and his own pullout tests data.
For horizontal strip anchor

For square and circular anchors

Where P, is pullout capacity factor and ( D ~ / Ais) geometry factor


ymtan@
Many other investigators Baker and Kondner (1966) etc. conducted several laboratory
model anchor tests.

2.2.2 Theoretical Analyses of Anchors

In literature several theories have been proposed to calculate the pullout capacity of
anchors, the difference between each one of them being mainly in the shape of the assumed
failure surface.

Mors (1959) proposed a failure surface in the soil at ultimate load which' may be
approximated as a truncated cone having an apex angle a equal to (90°+@ /?as shown in the
Fig 2.2. The net ultimate pullout capacity was assumed to be equal to the weight of the soil
mass bounded by the sides of the cone and the shearing resistance over the failure surface was
ignored.
P,=3.v (2.9)

Where
V = Volume of the soil in the truncated cone
y=Unit weight of soil
Fig. 2.2 Failure surface assumed by Mors (1 959)

Downs and Chieurzzi (1966) based on similar theoretical work, suggested that the apex angle
a shown in Fig 2.2 is always equal to 60°, irrespective of friction angle of the soil. But Teng
(1962) and Sutherland (1988) found that this assumption might lead to unsafe design in many
cases particularly with increase in depth.

Ireland (1963) proposed that the pullout capacity can be taken as a sum of the weight of the
soil mass vertically above the anchor plate and the shearing resistance along the sides of the
vertical surface as shown in Fig 2.3. In evaluating the shearing resistance the state of stress
can be assumed to be at rest. Following values of earth pressure coefficient at rest (KO)were
suggested,
KO= 0.5 for granular soils
KO= 0.4 for silts and clays
Failure surface I

- Anchor plate

Fig. 2.3 Failure surface for Friction Cylinder Method

Veesaert & Clemence (1977) presented a formulation for shallow circular anchors in sand
assuming a linear failure surface making an angle of P= (b /2urith the vertical through the edge
of the anchor plate as shown in Fig 2.4. The contribution of shearing resistance along the
length of failure surface was approximately taken into consideration by selecting a suitable
value of earth pressure coefficient from laboratory model tests. The net ultimate capacity can
be given as

Where

V is the volume of the truncated cone above the anchor


KO is the coefficient of lateral earth pressure, they suggested that the magnitude of Ko may
vary between 0.6 to 1.5 with an average value of about 1.
-G L A

Planar failure surface

Anchor plate

Fig. 2.4 Failure Surface assumed by Veesaert and Clemence (1 977)

Vermeer and Sujiadi (1985) observed that the angle P of the inverted cone slip surface with
the vertical (Fig. 2.4) is equal to the dilatancy angle y~ of the soil. Dilatancy may be described
as the change in volume that is associated with the shear distortion of an element of granular
material. For sand dilatancy angle is typically in the range between 0" and 20". For loose sand
it tends to zero but values beyond 15' may be found for dense sands. The following
expression is suggested for estimating the pullout capacity.

Where
@ ,= Critical state friction angle.

Murray and Geddes (1987) used limit equilibrium as well as limit analysis approach to deal
with the problem of horizontal anchors in sands. Model tests were also conducted in sands for
determining the vertical pullout capacity of anchor plates. While making use of limit
equilibrium approach, it was assumed that the failure surface is a straight line making an
angle of P=$ R with the vertical as shown in the Fig 2.4. Also in order to arrive at the
contribution of stresses over the failure surface it was assumed that the ratio of frictional shear
to the normal stress at all the points over the straight failure surface is approximately equal to
tan ($ 14).
II
In limit analysis approach, it was again assumed that the failure surface is a straight line the
inclination of which was suitably established and was found'to be making an angle of P=$
with the vertical for the upper bound solution as shown in Fig 2.4. Their lower bound
solution, results in practically unrealistic values.

Balla (1961) proposed a theory based on laboratory model tests and field load test data on
pullout capacity of shallow circular anchors in dense sand. It was shown that the shape of the
failure surface can be approximated by an arc of a circle making an angle of (45 - (b /w
the horizontal at the ground surface, as presented in Fig 2.5a. The ultimate pullout capacity of
the anchor is the sum of the weight of the soil in the failure zone and the shearing resistance
developed along the failure surface.

Ballas's theory indicated good agreement with the measured pullout capacity of anchors in
dense sand for an embedment ratio of D/B I 5. However for anchors in loose and medium
dense sands, for all D/B values and for anchors in dense sands with D/B > 5, the theory
overestimates the ultimate pullout capacity. The reason for the later observation is that it is
essentially a deep anchor condition and the failure surface does not extend to the ground
surface.

It was found that non-dimensional breakout factor Nu (i.e., Nu = -) P, increases with DIE3
YAD
ratio up to a maximum of Nu = N.* at D/B = (D/B), which was called the critical embedment
ratio (Fig 2.5b). For D/B > (D/B),, the breakout factor remains practically constant. Anchors
located at an embedment ratio D/B I (DB),, are defined or considered as shallow anchors and
those located at D/B > (D/B),, as deep anchors. Similar findings have been made by
Chattopadhaya and Pise (1 986).
\ Circular anchor plate

Fig 2.5a Failure Surface assumed by Balla (1961)

DIB

Fig. 2.5b Nature of variation of Nu with D B (Balla; 1961)

Mariupol'skii (1965), proposed separate mathematical formulae for the estimation of the
ultimate pullout capacity of shallow and deep anchors. He has assumed that for shallow
anchors the progressive failure mechanism commences with compression of the soil located
above the anchor plate Fig 2.6a. This compression occurs within a column of soil with the
same diameter as that of the anchor plate. As pullout progresses there is a continued
compaction of soil and this leads to an increase in the vertical compressive stress. Thus there
is a continued increase in frictional resistance along the surface of the soil colurnn. The
13
increase of the frictional resistance entrains adjacent rings of soil. Ultimately sufficient
tensile stress is developed so that failure occurs with the separation of soil in the form of a
cone with a curvilinear geneatrix. The net ultimate pullout capacity thus calculated by this
theory can be given as

Where KO= lateral earth pressure coefficient


c = cohesion
n = an empirical coefficient = 0.025@(degrees)
d = diameter of the shaft
For sand c=O, so that the last term in the numerator will be zero.

-~rdil~n~tW

Fig 2.6a Failure surface assumed by Mariupol'skii's for shallow circular plate anchor
Fig 2.6b Failure surface assumed by Mariupol'skii's for deep circular plate anchor

For deep anchor, it was assumed that under the applied load the anchor will reach a limiting
condition, after which additional work is required to raise the anchor through a distance L
which is equivalent to the work required to expand a cylindrical cavity of height L and
diameter d to a diameter B as shown in Fig 2.6b. Based on this concept the net ultimate
pullout capacity can be expressed as

4 =[?I (B' - d ' )


2- tan@] + f ( m i ) [ D - ( B - d ) ]

Where qo = radial pressure under which the cavity is expanded


f = unit skin resistance along the stem of the anchor
It was recommended that the lower of the two values be adopted for design. This was
primarily because the limit of D/B=(D/B),. for deep anchor condition was not clearly
established.
Meyerhof and Adams (1968) proposed a semi-empirical relationship for estimation of the
ultimate pullout capacity for a continuous or strip anchor foundations subjected to vertical
load and then modified it to the rectangular and circular anchors. Based on experimental
model test results they concluded that for shallow anchors the pullout capacity increases with
increase in depth, further distinct slip surface occurs in dense sands which extends in shallow
arc from the edge of the anchor to the ground surface as shown in Fig 2.7. In clays a complex
system of tension cracks was observed along with significant negative pore pressure above
and below the anchors. For deep anchors the failure surface is less distinct for both sand and
clay and the pullout capacity reaches the limiting value with increasing depth.

For shallow circular anchors the ultimate pullout capacity P, may be expressed as the sum of
the cohesion and the frictional force due to passive earth pressure developed on the cylindrical
surface extending vertically above the anchor foundation base, the weight of the anchor, Wf
and the weight of the soil, W, inside the cylinder. The ultimate pullout capacity is given by

P= = W,+w,+ i r ~ c S~, ,+( ; ) B ~ P ~ , , tan)

Where
c = Soil cohesion

S, = Shape factor governing passive earth pressure on a convex cylindrical wall


k, = Nominal pullout coefficient of earth pressure on the vertical surface
W,= Weight of lifted soil mass
Wf= Weight of footing
For strip anchors in shallow depth the ultimate capacity is given by

P, = 2 c ~ + @ ~ ktan$+^
,,

Where k, varies from 0.75 to 1.

+ W,)
W=weight of lifted soil mass and weight of footing, (Wf
P, Total passive eailh pressure GL
, 6 = 912 to 3414
W 4

q = Surcharge
H,, Critical depth
.
I
9

GL
w

e ":
4

"P

- -
c:
t
:

I II \ Anchor plate
Dj; - Hc,

BIZ ! 812
i

(a) At shallow depth


I (b) At great depth

Fig 2.7 Failure surface assumed by Meyerhof and Adams (1968)

For strip anchors in greater depth

P, =2cH+3(2D-H)Hk,tan$+W

Where H is the vertical extent of failure surface

The theoretical values of the pullout capacity appear to somewhat underestimate the actual
pullout resistance in dense sand and slightly overestimate the phllout resistance in loose sand.

For rectangular anchors in sand, the ultimate pullout capacity of shallow anchors may be
expressed as

P, = Wf+W, + 2cD(B + L) + @*(zs,B + L - B)k, tan@


Where B = width of anchor
L = Length of the anchor
Chattopadhyay and Pise (1986) proposed a theoretical model assuming a curved surface of
failure through the surrounding soil, to evaluate the ultimate breakout resistance of horizontal
plate anchors. During the axial vertical pullout of a horizontal plate anchor in sand, an axi-
symmetric body of revolution of soil is assumed to be initiated to move up along the resulting
failure surface. The mobilized shear strength of the soil along the failure surface and weight
of the body of the soil and plate anchor resist the movement. In the limiting equilibrium
condition, ultimate breakout capacity of the plate anchor is attained.

Saeedy (1987) proposed a theory to estimate the ultimate pullout capacity of circular plate
anchors embeded in sand, assuming the failure surface to be an arc of a logarithmic spiral as
shown in Fig 2.8. The author has converted the solution into a plot of breakup factor
F, = P& AD (A = Area of the anchor plate) vs. the soil friction angle $.

According to Saeedy, during the anchor pullout, the soil located above the anchor gradually
becomes compacted, in turn increasing the shear strength of the soil, and hence the net
ultimate pullout capacity. For that reason he introduced an empirical compaction factor which
is given in the form
p = 1.044 D,+ 0.44 (2.18)

Where
p =The compaction Factor
D , = Relative density of cornpaction
They have not considered the effect of surcharge (q=O).
The actual net ultimate capacity is given as P, (actual) = (F,y AD) p
Vertical t a n g e n t

(Focus) 1
D

Fig 2.8 Failure surface assumed by Saeedy (1987)

Subbarao and Jayantkumar (1994) proposed a theory for strip anchors in a general
homogenous c-$ soils with surcharge (q) at the horizontal ground surface by using the method
of characteristics coupled with a log-spiral failure surface in the lower region for determining
the pullout capacity in-terms of pullout capacity factors for cohesion, surcharge and unit
weight, similar to the bearing capacity factors. The failure surface was assumed to be a log-
spiral from the edge of the anchor in the curved rupture zone and straight line there after in
the Rankine passive zone as shown in the Fig 2.9.

q (Surcharge)

Fig 2.9 Failure surface assumed by Subba Rao and Jayant Kumar (1994)
19
The focus F of log-spiral lies along a line inclined at ( 4 5 4 /2) with the horizontal. The critical
failure surface was obtained by satisfying the vertical equilibrium of soil mass bounded by the
failure surface. The horizontal extent of failure surface at the ground level predicted using this
theory is however larger than the observed experimental values due to the assumption of the
Rankine passive zone near the ground surface. In addition to this, for dense to very dense
sands, the predicted pullout loads from the theory are lower than the experimental values.

Manjunath (1998) has modified the earlier theory by Subbarao and Jayantkumar (1994) by
using the complete log spiral failure surface without the Rankine passive zone near the ground
surface Fig 2.8. Pullout capacity factors Fc, Fq, and Fy separately for cohesion, surcharge and
unit weight have been computed for both strip and circular anchors by satisfying the vertical
equilibrium of the wedge bounded by the critical failure surface and are presented as
functions of embedment ratio and friction angle of soil.

Results from the theory have been compared with the model tests both conventional and
centrifugal and also with various available theories. The net average ultimate pullout capacity
(P,.,,J for strip anchor has been proposed as given below.
P,-,,t = cF, + qF, + 0.5 y 4 (2.19)

and the average pullout capacity P, as


Pu = Pu-net + y D

Vasic (1971) developed an analysis for the vertical pullout capacity of shallow circular
anchors in a general c-$ soil based upon the theory of expansion of cavity close to the surface
of a semi-infinite, homogeneous, isotropic solid (Vasic 1965) as shown in Fig 2.10 (Das;
1990). This method assumes an isotropic stress state and isotropic soil behavior. These
assumptions lead to an overestimation in dense sands and heavily over consolidated soil.
However the assumptions are reasonable for intermediate conditions,
which approximate isotropic behavior.
Swami Saran et a1 (1986) proposed an analytical procedure to predict the load-displacement
characteristic of shallow anchors in c-r) soils using a non-linear constitutive relationship.
Expressions have been developed to obtain the critical load and breakout load for strip, square
and circular anchors. A linear surface of sliding was assumed in this approach. The anchor is
buried at shallow depth in a homogeneous, isotropic medium of semi-infinite extent. At any
stage of loading pullout load P is resisted by a linear soil wedge of height D which makes an
angle a with the vertical, starting from the edges of the plate. The load at which the height of
D becomes equal to the depth of embedment is the critical pullout load PC,.For load Pd),,.,
the strength of soil (i.e. c and $) is considered to mobilize fully at the anchor level and to be
zero at the top level of the wedge. In between the strength is assumed to mobilize linearly.
For loads PC,IP IP,, the strength of soil is considered to mobilize fully at the level and 'm'
(mobilization factor) times the strength of the soil at the ground level, 'm' being less than
unity. P, is the pullout capacity and will be the load P when m equals unity.

Davie & Sutherland (1977) also analysed the behavior of circular anchors by using finite
element analysis, with an axi-symmetric elastic-plastic iterative procedure employing Von-
Mises failure criteria. The procedure could predict general type of failure regardless of the
depth of the anchor. The model was accurate only at small element strains because linear
elastic, non-strain hardening plastic stress-strain curve was used. The finite element program
developed can estimate magnitude and direction of stress occurring in each element of the
mesh at any stage of pullout resistance.

Rowe and Davis (1982 a & b) by using elasto- plastic Finite Element Analysis proposed a
general theory for horizontal as well as vertical anchors. The soil was modeled as a Mohr-
Columb elasto-plastic material with either associative or non-associative flow rule. A number
of factors such a roughness of the plate, initial stress condition, dilatancy effect and adhesion
or suction belowhehind the plate were considered. The assumption made was that the
practical failure load is reached when the displacement is a selected multiple of the
displacement that would have occurred if the condition had remained entirely elastic. They
choose 4 as the multiple and their practical failure load was described as K4 failure load.
Model tests on anchors in both clays and sands were also conducted for an anchor of LIB ratio
of 1 to 9.75 and D/B ratio of 1-8. Pull out capacity factors for cohesion and for unit weight
were developed for the basic case of smooth anchors placed in a givcn KO condition for a non-
dilatant soil without any suction or adhesion effect below the anchor. It was shown that the
roughness as well as initial stress state have negligible effect on the pullout capacity of
horizontal anchors. Dilatancy plays an important role in ultimate failure load in the case of
sandy soils. In the case of horizontal anchors, anchor pullout capacity increases with increase
in dilatancy angle of the soil. The increase is more predominant, as expected, in the case of
higher values of $. It was also shown that the presence of suction belowhehind the anchor
plate in saturated cohesive soils might lead to a significant increase in pullout capacity. The
theoretical-results were presented in the form of influence or design charts to be used in hand
calculations to obtain an estimate of anchor capacity for wide range of size of anchors and soil
types.

Kozo Tagaya et a1 (1983) analysed the pullout capacity of a buried anchor from finite
element program based on the Lade's elasto-plastic constitutive model for sands to clarify
ground stresses, maximum pullout resistance etc. They investigated the effect of finite
element mesh, load increment, soil constants, initial stresses and boundary condition on the
pullout resistance and displacement of the anchor and the stress distribution in the ground.

Basudhar and Singh (1994) studied the lower bound breakout factors for horizontal and
vertical strip anchors embedded in sand. A generalized procedure based on finite element and
non-linear programming has been developed.
2.3 Anchors in Layered Soil

Based on a large number of laboratory model and large-scale test results many
investigators reported the pullout resistance of anchors embedded in homogeneous soils, A
review of related literature shows that not much work has been done to determine the ultimate
pullout capacity in a two-layered soil, a problem that is often encountered in field.

Stewart (1985) studied the behavior of anchor plate embeded in a saturated clay layer
overlain by a compacted sand deposit, the experimental consisted of laboratory
model tests on circular anchor of 50mm diameter and 5mm thick as shown in Fig. 2.10a.
From this study it was found that the cohesionless soil overlay significantly increased the
ultimate pullout capacity of the plate anchor compared with its value when embeded in clay
alone (Fig. 2.10b). The increase in pullout capacity is due to two main factors. The first is
the additional over burden pressure, which converts the original shallow anchor into a deep
anchor, the second is the mobilization of the frictional resistance of the overlays. He showed
that large displacements were required to mobilize the frictional resistance of the overlying
sandy soil.

Loose I Dense Sand

Clay . ::
"t

Anchor plate
1'.
Fig 2110a Layered soil system used in Stewart's (1985) experimental investigation
DIB Tests
9.0

Test 6
3.0 2

1.5

( ) sand overburden pressure in kPa

Tests 2,4,6 - Dense sand


~ e s 7t - Loose sand

0 20 40 60 80
Displacement (mm)

Fig 2. lob Load-displacement curves of anchors in Layered soil system

Sutherland (1988) pointed out that in practice little real benefit to pullout capacity could be
achieved by placing a sandy soil over a plate anchor ernbeded in clay since a large
displacement is required to mobilize the shear strength of the overlays. It was suggested that
if a sand overburden were to be used, a more sensible solution would be to place the anchor
on the surface of the clay layer and then place sandy soil on top.
Bouazza and Finlay (1990) reported the behavior of a plate anchor burried in a two layered
sandy soil. The testing program consisted of a 37.5mm diameter circular plate anchor buried
in dense sandy soil overlain by loose or medium dense sandy soil as shown in Fig 2.1 la. The
pullout tests were carried out on an anchor embeded at a depth D in a combination of layers of
sand. The thickness of each layer was increased to a certain proportion of the anchor diameter
and it was increased from 1 to 4 times the anchor diameter. It was reported that for upper
layer thickness ratio of Iess than one and for a given D B (embedment ratio) there was no
difference between the pulling a plate anchor from a dense-medium bed or a dense-loose bed.
For a given D/B ratio and the upper layer thickness ratio of 1 to 4 a dense-medium bed gives a
greater pullout than a dense-loose bed as shown in Fig. 2. I lb. It is observed that the ultimate
pullout capacity is dependent on the relative strength of the two layers, the depth ratio of
embedment and the upper layer thickness ratio.

Loose / Medium Sand

Dense Sand
.": 1 I
Dt

Anchor plate

Fig 2.1 l a Layered soil system used in Bouazza & Finlay, (1990) experimental investigations
150
Bottom layer dense sand
- -
ULMS Upper layer medium sand
ULLS - Upper layer loose sand
D/B = Embkdrnent ratio

"

1
0 1. 2 3 4 5
Upper layer thickness ratio (D,/B)
Fig 2.1 l b Ultimate pullout capacity against D,/B ratio (Bouazza and Finlay; 1990)

Manjunath (1998) proposed a theory for the determination of vertical pullout capacity of a
shallow horizontal strip anchors in two layered c-4 soil system. The effect of surcharge has
also been considered. The theory has been developed by using method of characteristics
coupled with log spiral failure surface having different foci for different layers shown in Fig.
2.12. Pullout capacity factors, separately for cohesion (F,), surcharge (Fq) and unit weight
(F,) have been presented as a function of embedment ratio, friction angle of each layer and
ratio of top layer to the total embedment depth. Hence using these factors for Lhc chosen
values of friction angle of bottom layer ($t,), top layer ($,) and embedment ratio the vertical
pullout capacity of shallow horizontal strip and circular anchors can be determined in two
layered soil system with surcharge for any value of Df/D ratio. The net average ultimate
pullout capacity (P, J,. has been written in the form

And the average ultimate pullout capacity (Pu) as


?:,= c-,,,I+ (DJ,+ ~ b ? b

Fig 2.12 Failure mechanism in two layered soil system by Manjunath (1998)

Where rot= initial length of radial line of log-spiral failure surface in the top layer
rob = initial length of radial line of log-spiral failure surface in the bottom layer
rlt = final length of radial line of log-spiral failure surface in the top layer
rlt = final length of radial line of log-spiral failure surface in the bottom layer
pt = angle between the final radial line (r,,) of log-spiral failure surface in the top
layer and the horizontal at the interface .
Pb = angle between the final radial line (rib )of log-spiral failure surface in the bottom
layer and the horizontal at the interface ,
2.4 Load-Displacement Behavior

Along with critical pullout load it is also important to consider the force-deformation
relationship under different loads, which are required for the complete analysis of an anchor
foundation and to limit the anchor displacement to a serviceability limit. In order to
determine the allowable net pullout capacity of plate anchors, two different procedures can be
adopted.

1. Use of a tentative factor of safety, F.r,for the ultimate pullout capacity, based on the
uncertainties of determination of soil shear strength parameters and other associated
factors. For this type of analysis

2. Use of load - displacement relationship. In this method the allowable net pullout capacity
is calculated which corresponds to a predetermined allowable verticd displacement of the
anchor.

Though many theories have been proposed to predict the ultimate pullout capacity, at present
no satisfactory method is available for computing load-deformation behavior of anchors.
However a few attempts have been made to study the deformation behavior of anchors under
pullout.

Swami Saran et a1 (1986) proposed an a~alyticalmethod to predict the load-deformation


characteristics of shallow anchors in c-4 soils using non-linear hyperbolic stress-strain curves
of soils as the constitutive law.

Das and Puri (1989) investigated the load-displacement relationship of shallow horizontal
square and rectangular plate anchors embeded in medium and dense sand. For these
laboratory tests the width of the anchor plate (B) was kept at 50.8mm. The length-to-width
ratio of the anchor (WB) varied from one to five. Based on their laboratory observations, the
plots of net load P verses vertical deformations Z, can be of two types as shown in Fig.2.13.
In type I, the net load increases with displacement up to a maximum value, at which sudden
pullout occurs. The maxirnurn load i n this case is the nct ultimate pullout capacity P,,.
28
In type I1 the net load increases with the vertical displacement fairly rapidly up to a certain
point beyond which the load-displacement relationship becomes practically linear. For this
case the net ultimate pullout capacity is defined as the point where the slope of the P verses Z
plot becomes minimum. The vertical displacement corresponds to load P, is defined as 2, in
Fig 2.13

The magnitude of Au for anchors with various lengths to width (UB) ratios placed at varying
embedment ratio (DfB) was obtained. For tests conducted on medium sand, the relative
density of compaction Dr was about 48% and for tests conducted on dense sand the average
value of D,was about 73%. From their experimental results, Das and Puri proposed a non-
dimensional empirical load displacement relationship for shallow plate anchor, which is of the
form

Where P* =P/P,
z"=z/z;
Z= anchor displacement at net pullout load P
a and b = constants.

The constants a and b are approximately equal to 0.175 and 0.825 respectively and they are
not functions of the relative density of compaction.
It follows that

The preceding equation implies that the plot of z*/P* verses Z*will be approximately linear.
Fig 2.13 Load versus Displacement curves

Das and Seely (1975), followed a similar approach to study the load displacement response of
.
isolated shallow vertical anchors with limited width-to-height ratio.

Trautmann and Kulhawy (1988) obtained the basic load displacement data from both the
general literature and unpublished utility reports and has been summarized by Kulhawy and et
a1 (1983b), summarized the same. This includes load displacement response and various
supporting data on soil characteristics, construction procedures, and test conditions. The data
represents a total of 75 test results at 17 sites around the world and include various anchor
foundation types, sizes and depths; construction practices; soil conditions and loading
sequences. After compilation the data were evaluated to determine the failure load and
displacement. For this "slope tangent" method, i n which the failure load is taken as the
intersection of the tangents to the initial and final portion of the load displacement curve, was
adopted. The failure displacetnent was taken as that corresponding to the failure load.
Further the displacement data were put in dimensionless form by dividing by the depth of the
anchor and grouped according to soil and anchor type. These data were then plotted as a
normal probability plot to evaluate the statistical distribution. A logarithmic transformation
of the data provided a better straight line fit than an arithmetic plot. The results of the data
indicate the geometric mean displacements for both anchor foundation types and in situ soil
conditions. Also included are the geometric standard deviations, students t-values and the one
sided 95% confidence limits. Based on the limiting displacements obtained from the
statistical analysis, the load displacement data were then fitted with a hyperbolic equation and
the pullout load-displacement response was characterized at 95% confidence level as given
below.

In which Z is the displacement at load P.


D is the depth of embedment and
P, is the ultimate load.

Rarneshbabu (1998) studied the load-displacement behavior of the vertically loaded


horizontal plate anchors from both field and laboratory investigations. Laboratory and field
results documented in the literature have been collected for shallow and deep anchor plates
with variety of configurations and embedment in different types of sands. Trautmann and
Kulhawy's approach in providing statistical expression for the load displacement behavior has
been modified and the following two separate sets of expressions have been developed, one
for cohesive soils and the other for granular soils.

For square anchors in sand

Where PlI2is the ultimate capacity by two tangent rnethod


31
For circular anchor in sand

For strip anchors in sand

General load displacement relation for square anchors in cohesive soils

With the help of the above expressions for prediction of load-displacement response he also
proposed a deflection-dependent shape factor. Shape factor for any required deflection level
is the ratio of pullout factor for isolated anchor to that of strip anchor at the same deflection
level. Pullout factor (PF) is defined as the ratio of the load corresponding to the specified
deflection level to the weight of the soil colurnn above the anchor.

Where Ps is the pullout load corresponding to deflection 6

2.5 Summary and Scope of work

From literature presented on pullout capacity and deformation behavior of horizontal


plate anchor foundations in homogeneous and layered soils the following important points
need to be mentioned.
A number of investigators used small size model anchors in all the laboratory experiments
and extrapolated the results for predicting the behavior of large-scale anchors, which may lead
to unsatisfactory results.

Many investigators proposed theoretical techniques to estimate the pullout capacity of anchor
plates. Essentially the approaches involve use of limit equilibrium concepts. The methods
involve some assumptions regarding the shape of failure surface andlor the influence of the
soil above the anchor. Further the possible effects of dilatency, initial stress state and the
effect of backfill having higher or poorer strength than the native soil have not been
considered.

Critical pullout load and the deformation under the loads are required for the complete
analysis of anchor foundations. Though, many theories have been proposed to predict the
pullout capacity, at present no simple rational method is available for computing
deformations.

Not much work has been reported to study the behavior of anchor plates in layered soils, the
problem often encountered by the practicing engineers in the field. A review of literature
shows that no numerical work has been carried out to determine the pullout capacity and
deformation behavior in any type of layered soils.

By considering the above points and by using a numerical technique with an appropriate
constitutive model one can simulate field conditions more realistically. With this in view, the
present work has been undertaken. Numerical solutions for obtaining the ultimate pullout
capacity and deformation behavior of large size plate anchors in Sands, Clays and Layered
soils have been developed. In the analyses considerations have been given to the factors like
friction angle, dilatency, initial stress state, embedment ratio and the backfill having higher or
poorer strength than the native soil.

An attempt has been made to study the load-deformation behavior of the horizontal anchor
plate. Based on Numerical results of large size anchors with realistic soil parameters the
behavior has been generalized. Results are presented in the form of design charts, which may
be used in hand calculations to obtain an estimate of anchor capacity and displacements for a
wide range of soil types and size of anchors.
33
CHAPTER 3

NUMERICAL ANALYSIS USING FLAC

3.1 General.

Till recent years, all problems in geotechnical engineering such as bearing capacity,
slope stability, earth pressure and retaining walls etc, were being solved by limit state
methods, and no computations of deformations were possible in these methods. Structures
were designed with these failure loads together with a suitable factor of safety. But, it is now
well recognized that most problems are deformation controlled, and a mere factor of safety on
the ultimate load does not ensure acceptable levels of deformation under service loads
because of the highly nonlinear response. A complete knowledge of load-deformation pattern
is necessary for more realistic designs.

Problems involving complex materials such as soils and rocks and complex boundary
conditions such as in geotechnical engineering problems, do not have closed form solutions
and the engineer has to resort to computational methods that provide approximate but
acceptable solutions. With out numerical techniques it would be difficult to solve practical
geotechnical engineering problems analytically with a reasonable degree of accuracy. High-
speed digital computers have enabled engineers to employ various numerical discretization
techniques for approximate solutions of complex problems. As a result of broad applicability
and the systematic generality of different computational methods such as finite difference,
finite element, discrete element and boundary element techniques, the numerical techniques
have gained wide acceptance by designers and research engineers in engineering analysis.

In simple finite element formulations, we generally assume that the displacements of the finite
element assemblage are infinitesimally small and that the material is linearly elastic. In
addition, we also assume that the nature of boundary conditions remains unchanged during
the application of the loads on the finite element assemblage. With these assumptions finite
element equilibrium equations were derived for static analysis. These epations correspond
to a linear analysis of a structural problem because the displacement response is a linear
function of the applied load vector. When this is not the case we need to perform a nonlinear
analysis.

The nature of soil and rock is complex and analysis of such material requires nonlinear
considerations. The nonlinearity occurs in two different forms. The first is physical or
material non-linearity, which results from nonlinear constitutive laws. The second is
geometric non-linearity, which derives from finite changes in the geometry of the deforming
body (large strains). Depending on the sources of non-linearities, nonlinear problems can be
divided into three categories; problems involving material non-linearity, problems involving
geometric non-linearity and the problems involving both material and geometric non-
linearity.

In the case of material non-linearity, stresses are not linearly proportional to the strains, but
small displacements and small strains are considered, linear strain displacement relations are
used. Many significant engineering problems fall under this category. For problems
involving geometric non-linearity, non-linearity arises from finite changes in the geometry.
Linear stress strain relations are assumed in this case where as it encompasses large strain and
large deformation (due to rigid body movements). The most general category of nonlinear
problems is the combination of the first two categories. It involves non-linear constitutive
behavior as well as large strains and finite displacements. Thus to solve such problems, it is
necessary to employ an incremental formulation.

The numerical techniques that are well accepted for solving the structural problems may not
be recommended for the analysis of geotechnical problems due to the complexity of the soils.
The commercially available finite element packages like NISA, ANSYS, NASTRAN etc., are
useful to analyze structural problems but they are not very satisfactory to analyze the highly
non-linear geotechnical engineering problems as they do not have the right constitutive
models for the soils. Fast Lagraningan Analysis of Continua (FLAC) is an Explicit two-
dimensional analysis scheme, specially developed for geotechnical purpose, which simulates
the behavior of structures built of or on rocks, soils or other materials that may undergo
plastic flow when their yield limits are reached. This program contains several built-in
constitutive models, which sirnulate the highly nonlinear, irreversible response representative
of geologic or similar material and also allows to building and adding new constitutive
models in to the program.

3.2 Fast Legrangian Analysis of Continua

3.2.1 Overview

Fast Lagrangian Analysis of Continua (FLAC) is a two-dimensional explicit finite


difference program for engineering mechanics computation. It was originally developed for
geotechnical and mining engineers. Several built -in constitutive models are available that
permit the simulation of highly non-linear, irreversible response representative of geologic, or
similar materials. This program simulates the behavior of structures built oflon rocks, soils or
other materials that may undergo plastic flow when their yield limits are reached. FLAC has
the following special features, some of which are particularly suitable for geotehnical
problems.

P Interface elements to simulate distinct planes along which slip and/or separation can
occur.
P Plane strain, plane stress and axisymmetric geometry models,
> Groundwater and consolidation (fully coupled) models,
P Structural element models to simulate structural support of beam, pile or cable elements
(e.g., tunnel liners or rock bolts)
P Optional dynamic analysis capability,
F Optional visco-elastic (creep) models,
P Optional thermal (and thermal coupling to mechanical stress and pore pressure) modeling
capability,
P Extensive facility for generating plots of virtually any problem variable.

It also contains the powerful built-in programming language FISH. With this, one can write
his own functions to extend FLAC's usefulness and even implement his own constitutive
models if so desired. Materials are represented by elements, or zones, which form a grid that
is adjusted by the user to fit the shape of the object to be modeled. Each element behaves
according to a prescribed linear or non-linear stressktrain law in response to the applied
forces or boundary restrains. The material can yield and flow, and the grid can deform and
move with the material that is represented (in large-strain mode). FLAC ensures infinite
difference technique, that plastic collapse and flow are modeled very accurately.

3.2.2 Basic Calculation cycle of FLAC

The code embodies special numerical representations for the mechanical response of
geological materials. It has nine built-in material models: the "null" model, which represents
holes (excavations) in the grid, the isotropic elastic material; the transversely isotropic elastic
model; and six plasticity models (Drcker-prager, Mohr-Coulomb, Ubiquitous joint, Strain-
hardeninglsoftening, Double yield, and Cam-clay). Each zone in a grid may have a different
material model or property, and a continuous gradient or statistical distribution of any
property may be specified. Additionally, an interface or slip-plane model is available to
represent distinct interfaces between two or more portions of the grid. The interfaces are
planes upon which slip andlor separation are allowed, thereby simulating the presence of
faults, joints or frictional boundaries.

The basic formulation for FLAC that are being used here assumes a two-dimensional plane
strain state (3-D version is also available). This condition is associated with long structures or
excavations with constant cross section and, acted upon by loads in the plane of the cross-
section. In addition, it offers a plane stress option for elastic and Mohr-Coulomb plasticity
analysis. This is encountered, in thin plates loaded only in their plane. Finally, an option also
exists to model axisyrnmetric geometry. This geometry applies, to problems involving
cylindrical test specimens or cylindrical and spherical holes in a continuum.

Either velocity boundary conditions or stress boundary conditions may be specified at any
boundary orientation. Initial stress conditions, including gravitational loading, may be given,
and a water table may be defined for effective stress calculations. It incorporates the facility to
model ground water flow and pore pressure dissipation, and the full coupling between a
deformable porous solid and viscous fluid flowing with in the pore space. Structures, such as
tunnel liners, piles, sheet piles, cables, rock bolts or yielding props, that interact with the
surrounding rock or soil may be modeled with the structural element logic in FLAC.
3.2.3 Optional Features of FLAC

Three optional features for dynamic analysis, thermal analysis and modeling creep
material behavior are available as separate modules that can be included in FLAC. There are
three optional material models available that simulate visco-elastic (creep) behavior, the
classical visco-elastic (Maxwell) model, a two-component power law, and a reference creep
formulation implemented for nuclear waste isolation studies. All three models are available in
the creep module.

3.2.4 Numerical Techniques Used

FLAC is a two-dimensional explicit finite difference program, which uses an


"explicit", time-marching method to solve the algebraic equations. However, the finite
difference method is the oldest numerical technique used for the solution of sets of differential
e.quations, given initial values andlor boundary values. In the finite difference method, every
derivative in the set of governing equations is replaced directly by an algebraic expression
written in terms of the field variables (e.g., stress or displacement) at discrete points in space.
Hence no matrices are formed, large two-dimensional calculations can be made without
excessive memory requirements. In contrast, the finite element method has a central
requirement that the field quantities vary through out each element in a prescribed fashion,
using specific functions controlled by parameters. The formulation consists in adjusting these
parameters to minimized error terms or energy terms. Finite element programs often combine
the element matrices into a large global stiffness matrix, where as this is not normally done
with finite difference method because it is relatively efficient to generate the finite difi'erence
equations at each step. Wilkins (1964) presented a method of deriving difference equations
for elements of any shape and this method is used in FLAC.

3.2.5 Explicit Time-Marching Scheme

Even though FLAC is used to find a static solution to a problem, the dynamic
equations of motion are adopted in the formulation, One reason for doing this is to ensure
that the numerical scheme is stable even when the physical system being modeled is unstable.
With nonlinear materials, there is always the possibility of physical instability e.g., the sudden
collapse of a pillar. In real life, some of the strain energy in the system is converted in to
kinetic energy, which then radiates away from the source and dissipates. FLAC models this
38
process directly, because inertial terms are included i.e., kinetic energy is generated and
dissipated.

The general calculation sequence embodied in FLAC is illustrated in Fig. 3.1.

Equilibrium Equation
(Equation of Motion)

New Velocities
And Displacements
,,- \New Stresses
or Forces

StressIStrain Relations
(Constitutive Equations)

Figure. 3.1 Basic Explicit Calculation cycle

This procedure first invokes the equations of motion to derive new velocities and
displacements from stresses and forces. The strain rates are derived from velocities, and new
stresses from strain rates. The important thing to realize is that each box in Fig. 3.1 above,
updates all of its grid variables from known values that remain fixed while control is within
the box. For example, the lower box takes the set of velocities already calculated and, for
each element, computes new stresses. The velocities are assumed to be frozen for the
operation of the box i.e., the newly calculated stresses do not affect the velocities. However
we choose a time step so small that information cannot physically pass from one element to
another in that interval. The central concept is that the calculation wave speed always keeps
ahead of the physical wave speed, so that the equations always operate on known values that
are fixed for the duration of the calculation. No iteration process is necessary when computing
stresses from strains in an element, even if the constitutive law is highly nonlinear. In an
Implicit method (which is commonly used in finite element programs), every element
communicates with every other element during one solution step: several cycles of iterations
are necessary before compatibility and equilibrium is obtained. The disadvantage of explicit
method is seen to be the small timestep, which means that large number of steps must be
taken. Overall, explicit methods are best for ill-behaved systems - e.g., nonlinear, large strain
and physical instability systems and they are not efficient for modeling linear and small-strain
problems.

3.2.5.1 Comparison of explicit and implicit solutions methods

Explicit Implicit

Timestep must be smaller than a critical Timestep can be arbitrarily large, with
value for stability unconditionally stable schemes

Small amount of computational effort per Large amount of computational effort per
timestep timestep

No significant numerical damping Numerical damping depend on timestep


introduced for dynamic solution present with unconditionally stable
schemes.

No iterations necessary to follow nonlinear Iterative procedures necessary to follow


constitutive law nonlinear constitutive law

Provided that the timestep criterion is Always necessary to demonstrate that the
always satisfied, nonlinear laws are always above mentioned procedure is: (a) stable;
followed in a valid physical way and (b) follows the physically correct path
(for path-sensitive problems)

Matrices are never formed. Memory Stiffness matrices must be stored. Ways
requirements are always at a minimum. No must be found to overcome associated
bandwidth limitations. problems such as bandwidth. Memory
requirements tend to be large

Since matrices are never formed, large Additional computing effort needed to
displacements and strains are follow large displacements and strains.
accommodated without additional effort

3.2.6 Lagrangian Analysis

The Lagrangian strain component is defined as the increase in length of a line in terms
of the initial length and along a line initially parallel to x direction. The Eulerian strain
component is the increase in length of a line finally parallel to the x-axis, referred to its final
length. In evaluating derivatives, each of these two directions must be considered. If the
Lagrangian description is used, the displacement must be differentiated along a line initially
parallel to a coordinate direction (on the undeformed body). If the Eulerian description is
used, the displacement values must be differentiated along a line finally parallel to a
coordinate direction (on the deformed body). An assumption that is often made in developing
the relations is that the partial derivatives of the displacements be small, so that products and
powers of the derivatives can be neglected with respect to the derivatives and that the
derivatives can be neglected with respect to unity. This assumption often made in theory of
elasticity and restricts the analysis to small strains and to small rotations.

Since in FLAC there is no need to form a stiffness matrix, it is a trivial matter to update
coordinates at each timestep in large-strain mode. The incremental displacements are added
to the coordinates so that the, grid moves and deforms with the material it represents. This is
termed as "Lagrangian" formulation, in contrast to an "Eulerian" formulation, in which the
material moves and deforms relative to a fixed grid. The constitutive formulation at each step
is a small strain one, but is equivalent to large-strain formulation over many steps.

The general solution procedure is illustrated in Fig 3.2. It represents the sequence of processes
that occurs in the physical environment. In order to set up to run a simulation with FLAC, the
following three fundamental components of a problem must be specified:
1, Finite difference grid;
2. Constitutive behavior and material properties; and
3. Boundary and initial conditions.

The finite difference grid spans the physical domain being analyzed. Many people believe
that finite differences are restricted to rectangular grids. This is not true. Wilkins (1964)
presented a method of deriving difference equations for elements of any shape. This method
is used in FLAC. The constitutive behavior and associated material properties dictate the type
of response the model will display upon disturbance (e.g., deformation response due to
excavation). Boundary and initial conditions define the in-situ state (i.e., the condition before
a change or disturbance in problem state is introduced).
C)
*
Start
I

Model S e t u ~
>
---------,---,
I
Generate Grid, Deform to desired shape
I
I
9 Define constitutive behavior and material property
I
II
9 Specify boundary and initial condition

Step to Equilibrium State

I
I
I
I
I
I
I
I
I
I
Results Examine the
I
I Unsatisfactory Model Res~onse

Model makes Sense


I 1

I
b-
Perform Alterations
I
Example:
1. Excavate the Material
2. Change Boundary Conditions

Step to Solution

s
Needed Model Res~onse

4 Acceptable Results

Fig 3.2 General Solution Procedure of FLAC


3.2.7 Field Equations
The solution of solid-body, heat-transfer or fluid-flow problems in FLAC invokes the
equations of motion and constitutive relations, Fourier's law for conductive heat transfer, and
Darcy's law for fluid flow in a porous solid, as well as boundary conditions.

3.2.7.1 Motion and equilibrium: The equation of motion relates the acceleration, &/dt of
a mass, m, to the applied force, F, which may vary with time.
Newton's law of motion for the mass-spring system is

When several forces act on the mass, the above equation also expresses the static
equilibrium condition when the acceleration tends to zero, i.e. F=0 , where the
summation is over all forces acting. This property of the law of motion is exploited in FLAC
when solving static problems.

In a continuous body, equation 3.1 is generalized as follows.

p = massderzsity
t =time
x = coinportentsof coordinnrevector
j
gi = c~mponerztsofgravitntiaznl (body forces) and

0. - = coinpoizentsof
'J
stress terlsor

3.2.7.2 Constitutive Relation: The other set of equations that apply to a solid, deformable
body is known as the constitutive relation, or stresslstrain law. First, strain rate is derived
from velocity gradient as follows:
Where

e.. = stmi11- rate components, and


rl
u = velocity components.
j

Mechanical constitutive law is of the form

Where M( ) is the functional form of constitutive law,


K is a history parameter(s) which may or may not be present, depending on the
particular law, and := means "replaced by"
In general, the constitutive laws are written in incremental form because there is no unique
relation between stress and strain. The simplest example of a constitutive law is that of
elasticity:

Where
dB is the Krortecker delta,
At = time step
G, K = shear and bulk modulus, respectively

3.2.7.3 Boundary Conditions: Either stress or displacement may be applied at the boundary
of a solid body in FLAC. Displacements are specified in terms of prescribed velocities at
given grid points: equation 3.2 is not invoked at those grid points. At a stress boundary, forces
are derived as follows:

Where nj is the unit normal vector of the boundary segment, and As is the length of the
boundary segment over which the stress CTijb acts. The force Fiis added to the force sum for
the appropriate grid point.
3.2.7.4 Finite difference Equations: The difference equations for a triangle are derived from

the generalized form of Guass's divergence theorem (e.g., Malvern, 1969)

Ini fds = I-dA


af
s ax;
Where

j = is the integral around the boundary of a closed surface.


S

izi = the unit normal to the surface, s.


f = is a scalar, vector or tensor,
Xi = are position vectors,
ds = is an incremental arc length, and

I = is the integral over the surface area, A.


A

Defining the average value of the gradient of ' f ' over the area A as

one obtains, by substitution into Eq.(3.7)

For a triangular sub-element, the finite difference form of Eq. (3:9) becomes

where As is the length of a side of the triangle, and the summation occurs over the three sides
of the triangle. The values of <f > is taken to be the average over the side.
3.2.8 Ground Water and Consolidation

FLAC models the flow of ground water through a permeable solid, such as soil. The
flow modeling may be done by itself, independent of the usual, mechanical calculation of
FLAC, or it may do in parallel with the mechanical modeling, so as to capture the effects of
fluid/solid interaction. One type of fluidholid interaction is consolidation, in which the slow
dissipation of pore pressure causes displacements to occur in the soil. This type of behavior
involves two mechanical effects. First, changes in pore pressure cause changes in effective
stress, which affect the response of the solid - for example, a reduction in effective stress may
induce plastic yield. Second, the fluid in a zone reacts to mechanical volume changes by a
change in pore pressure, since the fluid is quite stiff. The code handles both fully saturated
flow as well as flow in which a phreatic surface develops: in the later case, pore pressures are
zero above the phreatic surface.

3.2.8.1 Constitutive Law


Darcy's law for an anisotropic porous medium is

Where Vi is the specific discharge vector ( units of velocity)


, P is the pressure, and
ISijis a "permeability" tensor
Each quadrilateral element is divided into triangles in two different ways. The specific
discharge vector can be derived for generic triangles, using the formula

Where C is the summation over the three sides of the triangle.


,The specific discharge vector is then converted to scalar volumetric flow rates at the nodes by
making dot products with the normal to the three sides of the triangle. The general expression
is
(3.13)
The stiffness matrix [MI of the whole quadrilateral element is defined in terms of the relation
between the pressure at the four nodes and the four nodal flow rates, as derived above
{el= [MI {PI (3.14)

The effect of gravity is incorporated as follows. If the grid point pressures around a zone
conform to the gradient aP/axi = gipw where gi is the vector of gravitational acceleration, then
the nodal flow rates {Q}should be zero. Hence the above equation is modified as follows
{Q/ = [MI {P - (xi - xi'" )gi p w/ (3.15)

Where xi'') is the x- coordinate of one of the corners.

3.2.8.2 Continuity equation


The flow imbalance, CQ, at a node causes a change in pore pressure at a saturated node as
follows

Where nV is the pore volume associated with the node.( n is the porosity and V is the total
volume). The term CQ includes contribution from the four surrounding zones and the source
's' that are specified.

In finite difference form the above equation becomes

WhereA V;,, is the equivalent nodal volume increase arising from mechanical deformations
of the grid.

3.2.9 Numerical stability


3.2.9.1 Mechanical Timestep

As described previously, the explicit-solution procedure is not unconditionally stable:


the speed of the "calculation front" must be greater than the maximum speed at which
information propagates. A time step must be chosen such that it is smaller than some critical
timestep.
The stability condition of elastic solid discretized into elements of size Ax is

Where C is the maximum speed at which information can propagate - typically, the p-wave
speed, C , where

For a single mass-spring element, the stability condition is

(3.20)

Where m is the mass, and k is the stiffness. In a general system, consisting of solid material
and arbitrary networks of interconnected masses and springs, the critical timestep is related to
the smallest natural period of the system, TInin:

3.2.9.2 Fluid Timestep

There are two aspects of numerical stability associated with the pore-fluid scheme.
First, an explicit solution of the fluid flow equations requires that the timestep be less than a
critical value; second, the bulk modulus of the fluid increases the mechanical stiffness of a
saturated zone. The effect of increased mechanical stiffness is incorporated into the density-
scaling scheme already in FLAC; the apparent mechanical bulk modulus of a zone is modified
by the presence of fluid as follows:

Where n is the porosity of the zone.


The explicit fluid timestep can be derived by imagining that one node at the center of four
zones is given a pressure of Po. The resulting nodal flow is then given by Q=POCMkk,where
CMkkis the sum over the 4 zones of the diagonal terms corresponding to the selected node.
The excess nodal flow gives rise to an increment in pressure AP, according to Eq.3.12

This relation is stable and monotonic if

The value of At used in FLAC is that given in above equation is multiplied by the safety
factor of 0.8.

3.2.10 Structural Elements

An important aspect of geomechanical analysis and design is the use of structural


support to stabilize a rock or soil mass. Structures of arbitrary geometry and properties, and
their interaction with the rock or a soil mass, may be modeled with FLAC.

Types of Structural Elements


Four forms of structural supports may be specified.
1. Beam Elements: Beam elements are two-dimensional elements with three degree-of-
freedom (x-translation, y-translation and rotation) at each end node. Beam elements can
be joined together with one another and/or the grid. Beam elements are used to represent
structural members in which bending resistance and limited bending moments are
important. They may be used to model a wide variety of supports, including support
struts in an open-cut excavation and concrete or shotcrete lining in a tunnel.
2. Cable Elements: Cable elements are one-dimensional axial elements that may be anchored
at a specific point on the grid (point-anchored) or grouted so that the cable element
develops forces along its length as the grid deforms. Cable elements can yield in tension
and compression, but they cannot sustain a bending moment.
3. Pile Elements: Pile elements are two-dimensional elements that can transfer normal and
shear forces and bending moments to the grid. Piles offer the combined features of the
beams and cables.
4. Support Members: Support members are intended to model hydraulic props, wooden
props or wooden packs. In its simplest form, a support member is a spring connected
between two boundaries.
Beam Elements
The beam elements in FLAC are standard two-dimensional beam elements with 3
degree-of-freedom (two displacements and one rotation) at each end node. Typical beam
element is defined by material and geometric properties, which are assumed to be constant for
each element. In general, the beam is assumed to behave as a linearly elastic material with in
failure limit. However if desired, a maximum moment (plastic moment) may be specified.
The beam is considered to have a symmetric cross-section, with area, A length L and second
moment of area, I
The orientation of the beam in two-dimensional space is defined by its direction cosines, ni, ti,

The force vector, Fi, at each node can be resolved into tangential (axial) and normal (shear)
component vectors:
Fi= F~'+ Fi"
= (Fjtj)ti+ (Fjnj)ili
= Ftti + F1izi

where
F,tj = IF;' I = F', and
Fjptj = IF: I = F"

Thus

Where the subscripts [a] and [b] identify the ends of the beam.
The components of axial and shear forces and moments at each node are given by the stiffness
matrix for a flexural element similar to the FEM:

Where utfal= tangential (axial) displacement at a,


unial= normal (shear) displacement at a,
u ' [ ~=
' tangential (axial) displacement at b,
unLbl= normal (shear) displacement at b,
' ( p l = rotation at a,

orb' = rotation at b, and


K = stiffness matrix
Various moment-release conditions (i.e., pinned joint) may be applied at each end node.
These conditions are:
3.3 Constitutive modeling of soils

Constitutive relations are mathematical expressions relating the response of a material


to the external changes, forces or environment etc. They are relations between cause and
effect. For soils in particular, we are talking about stresses and strains, or forces and
deformations. Constitutive relations are frameworks based on established principles of
continuum mechanics which describe the behavior of the material under any type of loading
in a global way, knowing some basic properties of the material. Different types of material
responses can described; elastic, plastic, elasto-plastic and visco-elastic and visco-plastic.
Although it is questionable whether the principles of continuum mechanics hold good for the
particulate material, soil, it is assumed that, for the scale of problem that we handle, soil mass
i
can be treated as a continuum.

A general stress-strain relation can be written as

Where D & C are 4th order tensors called the stiffness and compliance matrix respectively.

Elastic material models based on the theory of continnum mechanics can be generally
classified as linear elastic, Cauchy elastic, hyperelastic and hypoelastic models.

The linear elastic model gives a unique and linear relation between the state of stress and
strain, and it can be classified further as isotropic, trans~erst&~
isotropic, orthotropic or
anisotropic depending on the material response. The most general form of linear stress-strain
relations for an elastic material can be represented by the generalized Hooke's law as:

Where Bu are components of initial stress tensor


CUklis a fourth order tensor of elastic material constants
For other non-linear models, the various stiffness coefficients will not be constants, but will
be functions of either the current state of stress or strain. For Cauchy elastic material, the
current state of stress, Gij depends only on the current state of strain, ~ i j ,that is, stress is a
function of strain (or vice versa) i-e., the stiffness coefficients corsespond to secant modulus
values. The constitutive relation of this material has the general from
a ~Fu
=( E N) (3.34)

where Fij is the elastic response function of the material.

In hypoelastic formulation the stress rate can in general be represented by the material
response function that is a function of the current stress or strain state and strain rate. Here
the stiffness coefficient corresponds to tangent modulus values. The general form of
constitutive equation for this type of material is mathematically expressed as

where the dot indicates the rate of stress or strain.

Above type of non-linear-elastic models are often used for modeling the inelastic or plastic
behavior of soils, but they are not found to be satisfactory. Basically, in these models, the
various stiffness coefficients are arbitrarily chosen from experimental data. It is found that
these models violate the basic principles of thermodynamics, and hence may lead to some
problems under certain loading conditions. There are other class of no-linear elastic models,
the hyper elastic models, in which the stiffness coefficients are not arbitrarily chosen, but are
derived from a strain energy density function. These models are capable of reflecting a11 the
features of plastic behavior, most importantly the coupling between hydrostatic and shear
components of stresses and strains, in addition to non linearity, stress path dependency,
anisotropy, hardening, softening etc. (except irreversibility and hence suitable for only
monotonic loadings). However the difficulty with these models is to find a strain energy
density function which would be suitable for all conditions of loading. The model may end up
with too many material parameters, which often have no material meaning. All these indicate
that, we need plasticity based models are required to model the behavior of soils. Thus,
plasticity based models are required to model the behavior of soils.
Plasticity theory was initially developed for metals. In the initial stages, the models describe a
perfectly plastic response, i.e., the material will be either elastic or rigid until it reaches some
limiting stress and once that stress is reached, the material fails exhibiting infinite strains.
Subsequently the models were extended to include a hardening plastic response, i.e. the
material yields or experiences plastic strains once it reaches a particular limiting stresses, but
it does not fail, and the stress carrying capacity or the yield stress of the material increases
with plastic strains. Soils exhibit many special features when compared to metals. They
exhibit plastic volume changes, which no metal does. They exhibit contractancy and
dilatency during shear. Their hardening (and softening) is associated with plastic volume
changes where as in metals, the, hardening is associated with plastic shear strains. The
softening itself is a special feature not observed in metals. Being multiphase systems, their
interactions with pore fluid necessitates consideration of drainage conditions. Added to these,
there are problems of anisotropy, stress path dependency, cumulative plastic response under
repeated cyclic loading, etc.

The basic features of plasticity theory are yield function, failure criteria, hardening 1;lw and
flow rule.

3.3.1 Constitutive Models Provided in FLAC


There are nine basic constitutive models provided in FLAC version '3.3 arranged into null,
elastic and plastic model groups.
Null Model
A null model is used to represent material that is removed or excavated. The stresses
within a null zone are set to zero, no body forces (e.g., gravity) act on these zones. The null
material may be changed to a different material model at a later stage of the simulation. In
this way, backfilling an excavation, for example, can be simulated.

Elastic model group


The models in the group are characterized by reversible deformations upon unloading. The
stress-stain laws are linear and path independent. There are two models of this kind
1. Elastic, isotropic model
2. Elastic, transversely isotropic model
One can extend these elastic models to nonlinear hyperbolic models like Duncan-Chang
model by using the FISH programming.

Plastic Model Group


All plastic models potentially involve some degree of permanent, path-dependent
deformations, a consequence of the non-linearity of the stress-strain laws. The plastic flow
formulation in FLAC rests on basic assumption from plasticity theory that the total strain
increment may be decomposed into elastic and plastic parts, with only the elastic part
contributing to the stress increment by means of an elastic law. The following are the plastic
models available in FLAC
I . Druker - Prager model
2. Mohr - Coulomb model
3. Ubiquitous joint model
4. Strain softening/Hardening model
5. Double yield model
6. Modified Cam-Clay model

For the Druker -Prager, Mohr-Coulomb, Ubiquitous joint and strain softening models, a shear
yield function and non-associated shear flow rule is used. For Double yield model, shear and
volumetric yield functions, non-associated shear flow rule and associated volumetric flow
rules are included. In addition, the failure envelop for each of these models is characterized
by a tensile yield function with associated flow rule. The modified Cam-clay model
formulation rests on a combined shear and volumetric yield function and associated flow rule.

In FLAC, the out-of -plane stress is taken into consideration in the formulation that is
expressed in three-dimensional terms. All the models are based on plane strain conditions,
with exception of the strain softening model, which is also available in plane stress option. All
plasticity models are formulated in terms of effective stresses, not on total stresses.

In the present analysis, Mohr-Coulomb strain softeninghardening model is used for sands and
the modified Cam-clay model is used for clayey soils. These models are described in detail in
the next section.
3.3.2 Mohr-Coulomb Model

A well-known failure criterion used in soil ~nechanicsmuch before plasticity theory


was developed for metals is the Mohr-Coulomb criteria. Mohr's theory of failure involves the
construction of an envelope to all possible circles of stress that can be drawn for a particular
problem. These envelopes are generally curved but are usually replaced by a straight line.
This is equivalent to assuming that the soil conforms to the Coulomb failure criterion which
states that there is a linear relationship between the shear stress a at failure an the normal
stress, o,,
z = c + o , tan q5
where c = apparent cohesion
4 = angle of internal friction.

This defines a pair of straight lines in the o' : T stress plane. If Mohr's circle of effective stress
touches these lines, then failure of the soil will occur. If the state of stress Iies below this line
then it is in elastic state.
From Fig 3.3 (a). It may be deduced that

Wherea I a n d a jare the major and minor principal stresses at failure.

This yield criterion is independent of the intermediate principal stress 0 2 and is therefore not a
complete generality of the true behavior. For the case of cohesion c=O, this is quite similar to
extended Tresca criteria. In principal stress space it represents an irregular right hexagonal
pyramid, the axis of which lies along the space diagonal Fig 3.3 (b) and (c). A hexagon with
unsymmetrical edges reflects higher strength in compression than in extension.

Mohr coulomb model is an elastic perfectly plastic model. The failure envelop (which is also
the yield function) for this model corresponds to a Mohr-Coulomb criterion (shear yield
function) with tension cutoff (tensile yield function). The shear flow rule is non-associated
and the tensile flow rule is associated.
A
b
V1
V)
al
L

.-
C
L
m

.c
'(33, Normal s t r e s s O;,

- a,---------+
Fig 3.3 (a) Mohr-Coulomb failure Envelop
&
4

(b) (c)
& Fig 3.3 (b) Mohr-Coulomb and Tresca yield surfaces in principal stress space;
(c) Intersection with x-plane

Itrcr-etwrztal Elastic Law


In the FLAC implementation of this model, principal stresses 01, 02,03 are used, the out-of-
plane stress, o,,, being recognized as one of these. The principal stresses and principal
directions are evaluated from the stress tensor components and ordered so that (compressive
stresses are negative)
c ~ / C0 21 0 3 (3.38)
The corresponding principal strain increments Ae 1, Ae2, Ae3 are decomposed as follows

where the superscript e and p refer to elastic and plastic parts, respectively, and the plastic
components are non-zero only during plastic flow. The incremental expression of Hooke's
law in terms of principal stresses and strains has the form

wherea [=K +4G/3 and a 2=K - 2G/3.

Yield and potential functions


With the ordering convention of Eq. 3.38 the failure criterion may be represented in the plane
as
((rl,03) illustrated in Fig 3.4.

Fig 3.4 Mohr-Coulomb failure criterion in FLAC


The Mohr-coulomb shear yield function
S=aI-03&
+2cJ%
Tension yield function of the form
f t = aI -03
where$ is the friction angle, c, the cohesion, a 'fie tensile strength and

It may be noted that only the major and minor principal stresses are active in the shear yield
formulation; the intermediate principal stress has no effect. For a material with friction, $+O ,
the tensile strength of the material cannot exceed the value ot,,, given by

=-
tan (b

The shear potential function gScorresponds to a non-associated flow rule and has the form

where y is the dilation angle and

The associated flow rule for tensile failure is derived from the potential function gt, with
1
g =-aj (3.47)

Plastic correcriorzs
First consider shear failure. The flow rule has the form

where hS is a parameter of magnitude as yet unknown. Using the expression (3.45) for gs,
these equations become, after partial differentiation:

59
AeP1= hS
AeP2= 0
AePj= -hSN,,, (3.49)
The elastic strain increments may be expressed from (3.39) as total minus plastic increments.
In further using the flow rule (3.49), the elastic laws in (3.40) become

Let the new and old stress states be referred to by the superscripts N and 0, respectively.
Then, by definition,

Substituting these equations the expressions (3.50) for A a , d , 3 we may write

G I
N
= Dli - A S G l - a 2 N v )
N I
a2 =02 -2a2(1-~J

Where the superscript I is used to represent the elastic guess obtained by adding to the old
stresses, elastic increments computed using the total strain increments, i.e.,

The parameter ISmay now be defined by requiring that the new stress point be located on the
shear yield surface. Substitution of oINand 0 - 3 ~ for 01 and 03 in f=O gives, after some
manipulations (see 3.52 and 3.41)
In the case of tensile failure, the flow rule has the form

where the magnitude of the parameters ht is not yet defined. Using the expression (3.47) for
!gt. this expression gives, after partial differentiation:

Repeating a reasoning similar to that described above, we obtain

and

3.3.3 Strain-SofteningMardening Model

This model is based on the Mohr-coulomb model with non-associated shear and
associated tension flow rules, as described earlier. The difference however, lies in the
possibility that the cohesion, friction, dilation and tensile strength may harden or soften after
the onset of plastic yield. In the Mohr-Coulomb model these properties are assumed to
remain constant. Here one can define the cohesion, friction and dilation as piecewise-linear
functions of a hardening parameter measuring the plastic shear strain. A piecewise-linear
softening law for the tensile strength can also be prescribed in terms of another hardening
parameter measuring the plastic tensile strain. The yield and potential functions, plastic flow
61
rules and stress corrections are identical to those of the Mohr-coulomb model, as discussed
above.

Hanielling/Softenirzg Parameters
The plastic hardening parameters eP"s measured in by the accumulated plastic shear strain
whose incremental form is defined as

where
1
Aer = - (Ae? + Aef"
3

andA qs,j=1,3 are the principal plastic shear strain increments.


The tensile hardening parameter eP' measures the accumulated tensile plastic strain; its
increment is defined as

where ~e~~~is the increment of tensile plastic strain in the direction of the major principal
stress (tensile stresses are positive).
In the softeninghardening model, the user defines the cohesion, friction, dilation and tensile
strength variance as a function of the plastic portion ePof the total strain shown in Fig 3.5

Fig 3.5 Typical Variation of Cohesion, friction,


dilation and tensile strength with plastic strain

3.3.4 Critical State Concepts

Soils, when sheared to very large strains, finalIy reach a state where there will be no
further changes in stresses or volume and the soil just undergoes continuous deformation.
Such a state is called the critical state. Samples with different initial void ratios reach
different critical states. Or, there is a unique state of stress at critical state for each given void
ratio. The line joining ail these critical states for different initial states is found to be parallel
to the isotropic virgin compression path in v-p plane, and to be a straight line within a slope of
M, passing through the origin in q-p plane. It is well accepted that the compression path a
[e-logp] can be approximated by straight line in a semi-logarithmic plot.
I ,/critical state line

Fig. 3.6 Normal consolidation line and unloading-reloading (swelling)


lines for an isotropic compression test

The slope of the normal consolidation line is defined as h and the slope of the rebound line as
K. These known soil responses are used in the development of the Cam-Clay models
discussed in next section.

Equation of the normal consolidation line is defined as

where h and q are two material parameters and p, is a reference pressure. (note that VA is the
value of the specific volume at the reference pressure.)
The equation of the swelling line is define as
The recoverable change in specific volume Ave may be expressed in incremental form after
differentiation of Eq. (3.63)

After division of both members by v and using Eq. (3.65), we may write

Tangential Bulk modulus K can be derived to be

where^ is the slope of the elastic rebound path as defined in Fig 3.6

3.3.5 Modified Cam-Clay Model

The Cambridge group developed a systematic and comprehensive elasto-plastic model


in 1960. The model developed for clays was called Cam-Clay model. The modified Cam-
Clay model is an incremental hardeninglsoftening elasto-plastic model. Its features include a
particular form of non-linear elasticity and a hardeninglsoftening behavior governed by
volumetric plastic strain ("density" driven). Its yield surfaces, is an ellipse in the q-p plane
shown in Fig 3.7. The yield surfaces, after different levels of hardening are all similar in
shape and correspond to ellipsoids of rotation about the mean stress axis in the principal stress
space.

As the ellipse changes in size, the locus of the critical state point is a pyramid with its apex at
the origin shown by the critical state line in the qp-plane of Fig 3.7. It follows an extended
Von-Mises failure criterion (Fig. 3.3(b)). The flow rule is associative Cf=g) and the principle
of normality therefore applies to the yield surface. Since the surface is smooth, the direction
of plastic straining is uniquely defined for every point of the surface. At the intersection of
the critical state line and the ellipse, the normal to the yield surFace is vertical, Hence at this
point no component of plastic volun~etricstrain exists and all the plastic strain is distortional:
the soil can deform at a constant volume.

The yield surface is therefore strain dependent and expands or contracts as the soil hardens or
softens. Strain hardening is associated with compaction and strain softening is with dilation.
The initial size of the ellipse is governed by the maximum pre-consolidation pressure 2pCoto
which the soil has previously been subjected during its past history. If the soil has been over
consolidated at some time in its history, the p,o may be quite large and the soil can sustain
substantial load before any yielding occurred. For a stress path 1-2 in Fig 3.7, the plastic
strain vector normal to the ellipse produces a plastic volumetric decrease, which causes the
soil to harden. The ellipse expands until eventually position 2 is reached, at which point no
further volumetric strain occurs. The soil flows as a frictional fluid with constant volume.

The stress path 3-4 shows a strain-softening behavior due to the expansion of the material.
Consequently the ellipse decreases in size and eventually at point 4 the no volume change
limit is reached and coIlapse occurs at constant volume.

(Eq)
Strain hardening
S t r a i n softening

I
*
(6,)~
C- -1

Fig. 3.7 Yield surface of modified Cam-Clay Model in q-p space

I~tcreitzentalElastic Law
The generalized stress components involved in the model definition are the mean effective
pressure p and deviatoric stress q, defined as
where the Einstein summation convention applies and J2 is the second invariant of the
effective deviatoric stress tensor [s], i.e.

The incremental strain variables associated with p and q are the volumetric and deviatoric
strain components defined as

where AJ2 stands for the second invariant of the incremental deviatoric-strain tensor A[e], i.e.

The generalized strain increments.are decomposed into elastic and plastic parts so that:

And the new specific volume vN for the step may be calculated as

v N = ~ ( 1 -AE,,)

The incremental expression of Hooke's law in terms of generalized stress and strain is as
follows

The specific volume and mean pressure at the intersectidn of swelling line and normal
consolidation line are referred to as (normal) consolidation (specific) volume and (normal)
consolidation pressure: vCAand in the case of point A in Fig 3.6. Consider an incremental
change in stress bringing the point from state A' to state B' passing through B by loading and
At B it corresponds to consolidation volume vcB and consolidation pressure
The increment of plastic volume change AvP is measured on the figure by the vertical distance
between swelling lines (associated with points A and B) and we may write, using incremental
notation:

After divison of the left and right hand terms by v, we obtain, comparing with Eq.(3.65)

Hence, whereas elastic volume changes take place whenever the mean pressure changes,
plastic volume changes occur only when the primary consolidation pressure changes.

Yield and Potential Functiom


The yield function corresponding to a particular value pc has the form

The yield condition f=O is represented by an ellipse in the (q,p) plane, Fig 3.7. Note that the
ellipse passes through the origin; hence, the material in this model is not able to support an
all-round tensile stress.
The potential function g corresponds to an associated flow rule and we have

Plastic correctioizs
The flow rule is used to describe plastic flow has the form
where hv is the plastic (volumetric) multiplier whose magnitude remains to be defined.
Using the expression Eq.(3.79) for g, these expressions give, after partial differentiation:

The elastic strain increments may be expressed from Eq. (3.73) as total minus plastic
incremental. In further using Eq. (3.8 1)' the elastic laws in (3.75) become:

Ap = K ( A E , - XC,)
Aq = 3G(A&, - Xc,) (3.83)
Let the new and old stress states be referred to by the superscript N and 0, respectively.
Then, by definition, for one step,
PN = +A p
qN = q0 +A q
Substitution of the expression Eq. (3.83) gives
PN = -A 'Kcll
qN = q' -A. "3Gcb (3.85)
where the subscript 'I' is used to represent the elastic guess obtained by adding to the old
stresses, elastic increments computed using the total strain increments, i.e.,
0
Pi=P +ME,,
q ' = q o + 3 ~ ~ y (3.86)
The parameter hv may now be defined by requiring that the new stress point be located o n the
yield surface. Substitution of pN and qNas given by Eq. (3.85) for p and q in f(q,p)=O gives,
after some manipulations Eq. (3.78):

where
(3.88)
Of the two roots of this equation, the one with the smallest magnitude must be retained.

Note that the critical point corresponds top,, = p a , q,, = Mp&? in Fig 3.7, the normal to the
yield curve f=O is parallel to the q axis. Since the flow rule is associated, the plastic
volumetric strain rate component vanishes there. As a result of the hardening rule Eq. (3.77),
the consolidation pressure p, will not change. The corresponding material point has reached
the critical state in which unlimited shear strains occur with no accompanying change in
specific volume or stress level.
The new stress components qjNare expressed in terms of old and new generalized stress
values using the expressions

where

and [s] is the deviatoric stress tensor.

Harderzirzg/Softening Rule
The size of the yield curve is dependent on the value of the consolidation pressure p,,
Eq.(3.78). This pressure is a function of the plastic volume change and caries with the
specific volume as indicated in Eq. (3.77). The consolidation pressure is updated for the step
using the formula

where E~~ is the plastic volumetric strain increment for the step, v is the current specific
volume and h and K are material parameters as given previously in Eq.(3.62)

Over-corzsolidationRatio
The over-consolidation ratio R, is defined as the ratio between the zone initial
preconsolidation pressure ( a material property) and initial pressure po, i.e.,
3.4 Summary
The features and implementation of the Fast Lagrangian Analysis of Continua (FLAC)
has been discussed. FLAC is originally developed for geotechnical engineering. It has
several inbuilt constitutive models to m nod el the soil mass and also has the facility to
incorporate any user-defined constitutive law by using the inbuilt programming language
FISH. It is based on the explicit two-dimensional finite difference program. A discussion on
background and implementation of Mohr-Coulomb and Strain softeninglhardening model has
been presented. The strain softeninglhardening model is based on the Mohr-Coulomb model
with non-associated shear and associated tension flow rules. In this model the cohesion,
friction, dilation and tensile strength may harden or soften after the onset of plastic yield.
This chapter also discusses the critical state concepts and implementation of the modified
Cam-clay model. Cam-clay model was originally developed for clays. It reflects the
hydrostatic pressure or density dependent hardening material response.
CHAPTER 4

ANALYSES OF
HORIZONTAL STRIP ANCHORS IN SAND

4.1 Introduction

This chapter deals with the analysis of the behavior of large size horizontal strip
anchors in sands using a two-dimensional explicit finite difference program FLAC. The soil
has been modeled with a Mohr-Coulomb strain softeninglhardening model. Earlier
investigators have generally used small size anchors in their experimental investigations and
extrapolated the results to analyse and predict the behavior of large size anchors. This may
lead to unsatisfactory results. It has been clearly shown (Dickin; 1989) that the failure
displacements and load displacement curve patterns are very different for small and large
sized anchors, i.e. they are not just proportional to the size of the anchor. Load deformation
behavior together with the critical pullout load is required for the complete analysis of anchor
foundations. Though, many theories have been proposed to predict the ultimate pullout load,
at present no simple rational method is available for computing deformation at a given pullout
load. It is attempted in this chapter to analyze the behavior of large size anchors in granular
soil media.

The applicability of the numerical program FLAC to anchor pullout problem is validated by a
series of simulations and comparisons with the published data of model test results
(Rameshbabu; 1998 and Rowe; 1982b). A series of numerical simulations were made on the
large size horizontal strip anchors in different types of soils to analyze the load-deformation
behavior under vertical pullout load. Parametric study is made by varying width of anchor
(OSm, 0.75m and lm) depth of anchor (Dm ratio of 1 to 8) for loose and dense sand
conditions considering variation in dilation angle and critical friction angle of the soil. An
attempt has been made to present the results in the form of influenceldesign charts which can
be used in hand calculations to obtain an estimate of anchor capacity and deformation for a
wide range of soil types and size of anchors.
4.2 Friction angle ($) and dilation angle (yr)

It is now well understood that the shearing resistance of a granular soil mass is derived
from two factors. The external energy applied is used to overcome both the actual frictional
resistance at inter-particle contacts and also to expand the soil against the confining pressure.
The soil has to expand during shearing because of the highly irregular shape of the soil grains
causing interlocking and the grains have to be lifted up sometimes in order to cause shearing.
This behavior is called dilatency (a detailed study of stress dilatency theory was presented by
Rowe (1962)). Hence the observed friction angle Q, will be made up of two co~nponents,

@ =#Ja + w (4.1)
It is reasonable to assume $, to be constant for a given soil.
where
@, is the true friction angle as a result of frictional resistance at interparticle level when the
soil is shearing at constant volume, called the critical friction angle or frictional angle at
constant volume.
t
y is the dilatency angle which represents the energy required to expand against confining
pressure

The observed friction angle @ of granular soil varies with the nature of the packing of the soil:
the denser the packing, the higher the value of @.If @,, for a given soil remains constant, the
value of y has to increase with the increase in initial density of soil packing for a given
confining pressure. Further, the dilation decreases with increase in confining pressure for a
given density. The dilatency of a soil mass gradually decreases with continued shearing from
its initial high value to zero after very large shear strains, when the soil finally reaches a
constant, steady volume at critical states. Correspondingly the observed friction angle Q,
reduces from its peak value to $, at a very large strains.

In plasticity theory based models, such as Mohr-Coulomb model, proper values of friction
angle and dilatency angle are required to be specified. But most of the experimental data
reported in literature provide only the observed peak friction angle, and do not refer to the
dilatency component. For the present investigation we have chosen Chattachoochee River
Sand. The triaxial friction angle in it loose condition is 30". For plane strain case we are
using 32.5" (1.08+,). Observed peak friction angles are separated into dilation and the critical
73
friction angle of 32.5". For example, for a soil of frictional angle 4 5 O , it is separated in to a
friction angle of 32.5'and a dilation angle of 12.5"

In the present investigation, for Mohr-Coulomb strain softeninglhardening model, the friction
angle is varied from its peak value ope& to $cvand corresponding dilation from to
vpeak zero
over a 20% plastic shear strain.

Data of several sands reported in literature also indicates that the critical friction angle for
different sands lies in a narrow range of 30" - 35" (Stroud; 1971, Rowe; 1962). Hence, we
have carried out most of the analysis for this standard soil with $, as 32.5'. Although we
have used a representative value of 32.5" as $,,, further analysis of anchor behavior is carried
out for possible variation of $, of the sand from 25" to 35". The basic friction angle @,, of a
soil may vary to some extent depending on angularity of particles.

4.3 Formulation of the Problem

The boundary conditions used for the analysis of behavior of anchor are shown in Fig.
4.1. Boundaries of the soil domain are located at roughly 10 times the width of the anchor
plate from the center of the anchor plate. This is sufficiently large distance to ensure
minimum boundary influence on the results. A displacement boundary fixed in x direction is
provided on both sides of the grid and a rigid displacement boundary (fixed in x and y
direction) is provided at bottom of the grid. The finite difference grid used for the analysis is
shown in the Fig. 4.2. Finer mesh is used in the region around the anchor to get more
accurate results. The soil mass is modeled as a strain-softening Mohr-Coulomb material, as
explained in the earlier chapter. The anchor plate is treated as a beam element. The tie rod is
rigidly connected to the anchor plate, but the frictional interaction of the tie rod with the
surrounding soil mass is neglected. Thus the load applied on the top of the tie rod is directly
transferred to anchor point.

In all geotechnical-engineering problems, there is an in-situ state of stress in the ground,


before any excavation or construction is started. By setting initial conditions in the FLAC
grid, an attempt is made to reproduce this in-situ state, because it can influence the subsequent
behavior of the problem. In a uniform layer of soil or rock with free surface, the vertical
Fig 4.1 Problem Analyzed
h

Fig 4.2 Finite difference grid used for Analysis


stresses are usually equal to gpz, where g is the gravitational acceleration, p is the mass
density of the material, and z is the depth below surface. However the in-situ horizontal
stresses are more difficult to estimate. These are usually estimated in practice by the
empirical relation with the friction angle of the soil. The soil domain is initially equilibrated to
gravity stresses. An elastic model with a poison's ratio of 0.3 is used during this stage, which
is quite a reasonable value. Then the soil model is changed over to Mohr-Coulomb strain
softeninglhardening model. This constitutes the initial state for in the soil mass.

The simulations are made under strain controlled test by applying a displacement at the top
node of the tie rod. The displacement is controlled interns of velocity in FLAC. In order to
apply a given displacement to a boundary, it is necessary to prescribe the boundary's velocity
for a given number of steps. If the desired displacement is D, a velocity V over N steps
(where N=DN) may be applied. V should be kept small and N large in order to minimize
shocks to the system being modeled.

The convergence criterion for FLAC is the nodal unbalanced force, i.e., the sum of forces
acting on a node from its neighboring elements. If a node is in equilibrium, these forces
should sum to zero. For this study, the unbalanced force of each node was kept less than the
0.1 percent of the gravitational force.

4.4 Validation of Numerical program FLAC

Before the actual analysis for field anchors, it is necessary to verify that the FLAC
program can model the behavior of anchors under uplift load. For these problems, neither
analytical solutions nor field experimental results are available for comparisons. Hence it is
attempted'to compare the FLAC solutions with the available experimental results for model
anchors published in literature. For comparisons we have chosen two sets of data published
by Rowe (1982 b) and Rameshbabu (1998) for strip anchors.

One of the soil properties, which significantly affect the load-deformation curve, is the
Modulus of elasticity (E). It is an important input property required for the numerical
analysis. But most often, this property of the soil will not be reported in the experimental
investigations of anchors in literature. In such cases, a suitable value will have to be assumed.
The modulus value for soils is usually a variable with density of the soil. There are certain
approximate guidelines for possible values of E for different types of sands (Ref; Bowles;
1988).

4.4.1 Properties of soil reported by Ramesh Babu (1998)

Soil properties and experimental details reported by Ramesh Babu are shown in Table 4.1.
Table 4.1 Soil properties reported by Ramesh Babu (1998)

Sand @cv W Range of


y (kN/m3) 4'1 @P (assumed) (assumed) D/B
s1 15 38 41 32.5 8.5 4,6
s2 15.7 42.9 47 32.5 14.5 4,6

Where qt = triaxial friction angle


@, = plane strain friction angle
Width of the anchor plate = 60mm and thickness of the anchor plate = 6mm

The reported @ values are experimentally determined from triaxial tests. Based on that, the $
values for plain strain conditions were assumed by Rameshbabu as qp=l.l+t as it is generally
observed that a friction angle under plane strain condition is nearly 10% higher than that
under axi-symmetric condition (Hansen; 1961). For the present investigation we have
separated this plane strain friction angle in to critical friction angle (&=32.5) and the dilation
component as shown in table 4.1.

4.4.2 Comparison of results

Ramesh Babu (1998) has not reported the value of E for the material. Hence a value
of 35Mpa has been assumed. At a value of E equal to 35Mpa, the predicted load-
deformation curves show good agreement with the experimental results published by -Ramesh
Babu (1998) shown in Fig 4.3 and 4.4. E of 35Mpa is a reasonable value for sand of density
1500kg/m3to 1570kg/m3 (Ref. Bowles; 1988, Ortiz et al; 1986).
O.OE+OO 20E-04 4.OE-04 6.OE-04 8.OE-04 1.OE-03 1.2E-03 1.4E-W 1.6E-03 1BE-03 2.OE-03
Displacement(m)

Fig 4.3 Corrparision of load-dispimnt curves d h Exp. results

----- Ramesh Babu(Exp)


O.OE+OO I I I 4 I I I I I

O.OE+OO 2.OE-04 4.OE-04 6.OE-04 8.OE-04 1.OE-03 1.2E-03 1.4E-03 1.6E-03 1.8E-03 2.OE-03
Displacement (rn)

Fig 4.4 Cornparision of load displacement curves with Exp. data


4.4.3 Properties Reported by Rowe

The soil properties and experimental details reported by Rowe (1982b) are shown in Table 4.2
Table 4.2 Soil properties reported by Rowe (1982b)
y (w/m3' @ \V Range of D/B
1

Width of the Anchor plate = 5 1mm, thickness of the anchor plate = 8mm

Rowe (1982b) has also not reported the value of E for the material. For Medium dense sand
the range of E value is lOMpa to 5OMpa (Ref. Bowels; 1988, Ortiz et al; 1986). Hence for
the present comparisons we assumed a E value of 35Mpa.

4.4.4 Comparison of results

The definition of ultimate load will be a difficult thing in cases where the load-
displacement curves continue to harden even after very large strains. The practical ultimate
load in that case may have to be obtained by judgement or indirect methods. Many
investigators have used different definitions or techniques to fix the practical ultimate load.
The commonly used methods are single tangent and double tangent methods. Rowe (1982b)
used Kq failure theory i.e. the failure load is considered to have been reached when the
displacement is a selected multiple of that which would have been reached had conditions
remained entirely elastic. He used 4 as a multiplying factor and the failure load is denoted as
Kq failure load. For the comparison purpose, the same definition of failure load has been used
in the present analysis to estimate the ultimate failure load. The experimental values of
pullout capacity factor (P,/yBD) for various D/B ratios as obtained by Rowe (1982a) have
been compared with the present analysis in Table 4.3. The same comparison is showri in Fig
4.5. Present analysis is giving good agreement with the Rowe's (1982b) experimental results.
Table 4.3 Comparison of Numerical results with experimental results
Friction
angle @ in "

Rowe (1982b) has not presented the individual load displacement curves for his experimental
tests. However he has presented load displacement curves in the normalized form of P,/yBD
vs GEIyBD from the Finite element analysis of anchors, but again, the value is not reported.
As mentioned above value of input parameter E is required. A possible range of values of E
has been selected based on density (10Mpa to 50 Mpa for loose to Medium dense sands;
Bowles 1988 and, Ortiz et al; 1986). The load displacement curves for a E value of 35Mpa
obtained from the present analysis for the problem considered by Rowe (1982b) in the FEM
analysis have been presented in Fig 4.6 together with the load displacement curves by Rowe.
Rowe assumed an initial state of stress corresponding to KOequal to one. This state can be
achieved only with v equal to 0.5. So in the present analysis we have used v equal to 0.499.

Further, for the purpose of comparisons, theories of Meyerhof & Adams (1968), Murray &
Gedds (1987), Dickin (1989), Manjunath (1998), Subbarao & Jayanthkumar (19911) and
Vermeer & Sutjiadi (1985) have been used and the ultimate pullout capacity factor predicted
by them are compared with the present results in Fig. 4.7 and Fig. 4.8. It can be seen that the
values predicted by the present investigation are showing a close agreement with the
experimental and analytical results, except for the experimental results of Dickin (1989).
Infact Dickin's results (centrifuge tests) are showing unusual low pullout loads at higher D/B
ratios, when compared to experimental results of other investigators.
9 = 32'
q f = 40
B = 51mm

-0- Present (Theo)


-O- Rowe(Exp.)

0.0 1.O 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
DIB

Fig 4.5 Comparison of pullout capacity factor with Rowe Exp. results

O.OE+OO 1.OE-04 2.OE-04 3.OE-04 4.OE-04 5.OE-04 6.OE-04 7.OE-04


Displacement (m)

Fig 4.6 Comparison of load displacement curves with Rowe's FEM analysis
+Present
R Manjunath
A Meyerhof & Adams $=36O
A Subbarao & Jyant Kumar B=5lrnrn
X Dinkin
0 Murray & Geddes

8
0

0.0 1.O 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
D/B
Fig. 4.7 Comparison of pullout capacity factor of strip anchors by various theories

-c Present
Subba Rao & Jayant Kumar +=30
A Manjunath B=0.051m
A Rowe and Davis
X Vermeer asutijiadi
0 Meyerhof and Adams

0.0 1 .O 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0


DIE
Fig 4.8 Comparison of pullout capacity factor by various theories
FLAC also has an extensive facility of generating plots of various quantities. Fig 4.9a&b
show the displacement vectors and plastic regions at failure for different embedment depth
(D/B). It has been observed that there is significant surface displacement for DIB values up to
4, No significant surface displacements are seen for D/B ratios of 4 to 6, and practically zero
surface displacements for D/B of 8. These findings are in broad agreement with Rowe
(1982b).

The above FLAC simulations and comparisons with experimental and analytical results show
that FLAC can simulate the pullout behavior of anchors quite well. We will now proceed to
analyze the behavior of large size anchors for various conditions.

4.5 Analysis of large size anchors

FLAC simulations of pullout response of large size anchors embedded in


homogeneous sands are carried out with different parametric variations. The geometric
parameters considered are the width of the anchor plate B, and the depth of the anchor, D
usually specified as a ratio of width of anchor i.e. Dm. Three sizes of strip anchors, B=0.5m,
0.75m and l m are considered. Embedment ratios (D/B) of 1 to 8 are considered. This may
cover the practical range of sizes of anchors in the field.

The soil properties that affect the pullout response are soil properties cohesion (c), friction
angle (@),dilation angle (v), unit weight of soil (y) and Elastic constants of the soil E and G
or V. Usually while doing parametric studies, above properties are arbitrarily varied (in the
usual range), say Q, values from 0" to 50°, \y values from 0" to 40°, c values from 0 to 100 kpa
etc. Such variations may be meaningless because these properties are not independent and
mainly depend on density of the soil. Hence in the present analysis we have chosen published
properties of Chattahoochee River sand both at loose and dense conditions (Vesic and
Clough; 1968). The reported values were $,=32S0, v=O, E=6 Mpa and F1317kglm3 for
loose state, Q,,,=32.5, y=lOO,E=19.6 Mpa and ~ 1 5 4 kg/m3
3 for the dense state. We have
suitably extrapolated and considered four different states with y ranging from 0" to 15" to
cover the possible range. The corresponding assumed properties of the soil used for the
analysis q-e given in Table 4.4.
Fig 4.9a Displacement vectors at failure in model anchors
B=Slrnm, ( 9 = 3 2 O , Y=5"
Plasticity indicator
'at yield in shear or vol.
X elastic, at yield in past
o at yield in tension

Fig 4.9b Plastic regions at failure in model anchors


B = 5 1 m m $ = 3 2 " y=5"
Table 4.4 Soil properties used in the analysis
S.No. Y (kg/m3) 9 'Y E (MP) Range of
Dm
1 1317 32.5 0 6 1,2,3,4,6,8

Cohesion c is taken to be zero since the analysis here is for anchors in pure sands. Anchors in
cohesive soils are treated separately in the next chapter, where totally different constitutive
model, i.e., Modified cam-clay model is used, being a more appropriate model for clayey
soils. Although Capped models such as modified Cam-clay models, with density dependent
hardening and cap-like yield curves are better suited for soils. Mohr-Coulomb model may be
good enough for sandy soils, which generally exist in relatively dense states.

The material properties of the anchor plate are kept constant. It is assumed that the plate is
sufficiently stiff as not to affect the pullout response. Though for the purpose of analysis we
have used steel as the material of the anchor, the stiffness value used is so high, that the
results can be used for concrete anchor foundations

The same grid and boundary conditions used for model anchors shown in Fig 4.2 is used for
the analysis of large size anchors also. The strain-controlled simulations are made by
applying a displacement at the top node of the tie rod and corresponding load was plotted.

4.6 Definition of Failure: The K2 failure concept

As discussed earlier, the definition of failure for a continuously hardening'load -


displacement curve is a matter of judgement.

Trautmann and Kulhawy (1988) considered double tangent method to obtain the ultimate
pullout capacity P,, from the load displacement curve, where as many analytical methods
proposed in the literature suggest single tangent method. The results of model and field tests
on anchor plates suggest the presence of significant contained plastic deformation before
collapse and failure is often defined according to an arbitrary mle, such as load at a specified
86
displacement or according to Terzaghi's (1943) local yield criterion. By using finite
difference/finite element analysis, we can get a complete load deformation curve up to large
strains which is experimentally impossible. We need to estimate a practical definition of
failure for problems in which the full ultimate capacity is obtained only after extensive
contained plastic deformation.

Although many such definitions have been used a definition that is convenient and rational in
the present context is the one that is adopted by Rowe (1982 a&b). The failure load is
considered to have been reached when the displacement is a selected multiple of that which
would have been reached had the condition remained entirely elastic. This definition is
arbitrary in terms of the choice of multiple to be used; however it is not dependent on scale
and modulus and it gives a calculable limit on the displacement before failure, provided the
load path to failure is monotonic. In the present investigation 2 is chosen as the multiple and
the failure load is denoted as K2 failure load as shown in Fig. 4.10; and it corresponds to the
apparent stiffness of one half of the elastic stiffness (similarly Kg and Kq are loads
corresponding to an apparent stiffness of one third and one forth respectively of the elastic
stiffness). Rowe (1982a&b) proposed Kq failure theory for his analysis of model anchors.
But our analysis with large size anchors showed that for the present load-displacement curves,
it is impossible to adopt a Kq failure criteria, (the Kq secant line would go far beyond the load
displacement curve). This fact that the load-displacement patterns arevery different for small
size and large size anchors clearly indicates that to predict the behavior of large size anchors,
there is a need to analyze or conduct experiments directly on large size anchors. The adoption
of a multiple of 2 in conjunction with a typical factor of safety of 2 or more will generally
ensure that the working load is close to the linear range and hence displacement may be
estimated from elastic solutions.

-
4.7 Load Displacement behavior of Anchors

Displacement of anchor plate under different levels of pullout force will be of


paramount importance when design requirements limit such displacement. Many investigators
have proposed the techniques to estimate the uplift capacity of horizontal anchor plate
embedded in different soil media. However there has been only a limited effort to
characterize the load displacement response of these anchor plates. Using the method of
Pullout load (N)
present analysis, complete characterization of the load-displacement behavior has been carried
out.

Typical load displacement curves for different D/B ratio and different sizes of anchors (0.5m,
0.75m and lm) are shown in Figs 4.1 1 to 4.13. The pullout resistance increases significantly
with anchor embedment ratio and failure displacement also increases with embedment ratio as
shown in Figs 4.11 to 4.13. Typical load displacement curves for different dilation angles
with different embedment ratios for 0.5m size anchor are shown in Figs 4.13 to Fig 4.16. It
has been observed that for each soil type there is a characteristic relative depth beyond which
anchor plate starts behaving as a deep anchor. The higher the dilation angle, the effect of D/B
is more significant. Considerable increase in ultirrlate load is shown even beyond the
embedment ratio of 6 for higher values of dilation angle, where as for smaller values of
dilation angle, the rate of increase is small beyond D/B of 6. This shows that the shallow or
deep anchor condition depends on the dilation angle. Vesic (1977) and Das and Seely (1977)
hare reported that the characteristic relative depth will depend on the $ value and varies from
6 to 8 for loose sand having $=31°. The present analysis also indicates the existence of
characteristic relative depth, which depends on the value of v. The load displacement curves
are re-plotted in Figs 4. 17 to 4.34 to show the effect of dilation on load displacement curves
for different embedment ratios and for different sizes of anchor 0.5m, 0.751~1and 1m.

Figure 4.35 show the displacement vectors at failure for different embedment ratio for an
anchor of 1 m size. The surface heave is quite clearly indicated for a shallow anchor for D/B
ratio up to 4, while only a slight bulging of the soil surface is indicated for D/B ratio between
4 to 6, with a larger zone of heave for the shallow cases. However the surface heave was not
evident for D/B ratio of 8. Such observations are in broad agreement with those of Rowe
(1982b) and Dickin (1989).

Figures 4.36a,b&c show typically the load displacement curve, displacement vectors and
plastic regions for a l m width strip anchor for D/B ratio of 4 at different pullout loads of P,,
Pu/2, Pu/3, 3Pu/4 and 5P,/6. This clearly indicates the development of displacement vectors
and plastic regions with increase in pullout load.
m
0
+
d-
0
+-
d
+
0
d
0
+ B+
0
+ + od-+ d-
+
O E
+
S o
0
+
O
W W w W w w w w w w w
9
F
9
0
9
03
9
h
9
(0
9
Ln +0 9
'"
9
( N
9
-
9
0
O.OE+OO 1.OE-02 2.OE-02 3.OE-02 4.OE-02 5.OE-02 6.OE-02
Displacement (m)
Fig 4.14 Load-displacement curves for different D/B ratios
LO
0
+
w
9
N
m
0
+
W
a!
r-
LO
+
0
W
'9
T-
m
0
+
W
Y
r
0
+
w
'Y
T-
LO
+
0
W
9
T-
-T
0
+
W
9
a2
-
+
0
W
9
w
d-
o
+
W
9
d-
d-
0
, +
W
9
c\l
o
0
f
W
9
0
O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02
Displacement (m)

Fig 4.17 Load displacement curves for different dilation angles

O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02 3.OE-02 3.5E-02 4.OE-02


Dlsplacement (m)
-
Fig 4.18 Load displac~menlcurves for diflerent dilation angles
96
Displacement (m)

Fig 4.19 Load displacement curves for different dilation angles

O.OE+OO 2.OE-02 4.OE-02 6.OE-02 8.OE-02 1.OE-01 1.2E-01 1.4E-01


Displacement(m)
-
Fig 4.20 Load displacement curves for different dilation angles
Displacement (rn)

Fig 4.21 Load - displacement curves for different dilation angles

O.OE+OO 5.OE-02 1.OE-01 1.SE-01 2.OE-01 2.5E-01 3.OE-01 3.5E-01 4.OE-01 4.5E-01 5.OE-01
Displacement (m)

Flg 4.22 Load dlsplacernenct curves for different dilation angles


O.OE+OO 1.OE-03 2.OE-03 3.OE-03 4.OE-03 5.OE-03 6.OE-03 7.OE-03 8.OE-03 9.OE-03 1.OE-02
Displacement (m)

Fig 4.23 Load - diaplacement curves for different dilation angles

O.OE+OO 2.OE-03 4,OE-03 6.E-03 8.OE-03 1.OE-02 1.2E-02 1.4E-02 1.6E-02 1.BE-02 2.OE-02
Dlsplecement(m)
Fig 4.24 Load - displacement curves for different dilation angles
O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02 3.OE-02 3.5E-02
Displacement (m)

Fig 4.25 Load - displacement curves for different dilation angles

O.OE+OO 1.OE-02 2.OE-02 3.OE-02 4.OE-02 5.OE-02 6.OE-02 7.OE-02


Dlaplacernent (m)

-
Fig 4.26 Load displacement curves for different dilation angles
O.OE+OO v'
O.OE+OO 2.OE-02 4.OE-02 6.OE-02 8.OE-02 1.OE-01 1.2E-01
Displacement (rn)
Fig 4.27 Load - displacement curves for different dilation angles

O.OE+OO 5.OE-02 1.OE-01 1.5E-01 2.OE-01 2.55-01


Displacement (m)

Fig 4.28 Load - displacement curves for different dilation angles


O.OE+OO 5.OE-04 1.OE-03 1.5E-03 2.OE-03 2.5E-03 3.OE-03 3.5E-03 4.OE-03
Displacement (m)

Fig 4.29 Load - displacement curves for different dilation angles

O.OE+OO 1.OE-03 2.OE-03 3.OE-03 4.OE-03 5.OE-03 6.OE-03 7.OE-03 8.OE-03 9.OE-03 1.OE-02
Displacement (m)

Fig 4.30 Load - dispalcement curves for different dilation angles


O.OE+OO 2.OE-03 4.OE-03 6.OE-03 8.OE-03 1.OE-02 1.2E-02 1.4E-02
Displacement (m)
Fig 4.31 Load - displacement curves for different dilation angles

O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02


Displacement(m)

Fig 4.32 Load - displacement curves for different dilation angles


O.OE+OO I t I

O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02 3.OE-02 3.5E-02 4.OE-02 4.5E-02 5.0502
Displacement (rn)
Fig 4.33 Load - displacement curves for different dilation angles

O.OE+OO 1.OE-02 2.OE-02 3,OE-02 4.OE-02 5.OE-02 6.OE-02 7.OE-02 8.OE-02 9.OE-02 l.0E-01
Displacement (m)

Fig 4.34 Load - displacement curves for different dilation angles


t l l l
l t t l
t t l r
I
I
t
I
I
8
.
I
.

l t t l t I r
1 1 1 1 1 1

._ ._ _ ._ ._......,.................. . ..
t
! ! ! l I I I
1 1 1 1 I I 1
r
,, , r r , l l . r f l 1 t l t ~ I 1 ~ r l r I . I I , ,
l t t l
I t s 1
\ I l l
l
I
I
, ,,
I

,
r , , , 1 I ~ t l l l l t t l l l I l I I l l , \ , \ r.

Fig 4.35 Displacement vectors at failure in large size anchors


B=l m, @=32.5', Y=5"
Plasticity Indicator
* at yieldIn shear or vol.
X elastic, at yield in past
o at yield in tension

Fig. 4.36(b) Displacement vectors and plastic regions at different load levels
for Fig.4.36(a)
Plasticity Indicator '
* at yield in shear or vol.
X elastic, at yield in past
o at yield in tension

Fig. 4.36 (c) Displacement vectors and plastic regions at different load levels
for Fig.4.36(a)
-
) 1 3 0 0 3 ( ~
X>OO(
Plasticity indicator
* at Avield In shear 01
-

X elastic, at yield in
o at yield in tension
P"

Fig 4.36(d) Displacement vectors and plastic regions at different load levels
for Fig.4.36(a)
Figure 4.37a&b show the displacement vectors and plastic regions at failure for In1 width
strip anchor at DIB of 4 for different values of dilation angle varying from zero to
300.Practically, dilation angle of 30" may not be existing, and only for comparison purpose, it
has been presented. For loose sands (v<lOO)the displacement vectors and plastic regions are
vertical above the anchor plate. But for dense sands ( p l O O )the displacement vectors
plastic regions are extending laterally. Similar findings are reported by Rowe (1982b).

Figure 4.38 show the displacement vectors and plastic regions at failure of l m size of anchor
for a D/B of 2 for different dilation angles. Figures 4.39a&b show the displacement vectors
and the plastic regions for 0.5m width anchor for a D/B value of 6. It clearly shows that in
loose sand (v<lOO)at D/B=G there is no considerable displacements at the ground surface.
But for higher dilation angles greater than 10" (dense sand) there are significant
displacements at the ground surface at D/B equal to 6. Hence for the loose sand, the critical
embedment depth is less than 6 and for dense sand it is more than 6.

4.8 Generalization and Prediction of Load-Displacement behavior

There is a great need to generalize load displacement characteristics of anchors to


facilitate its use in designs. Rowe and Davis (1982 a&b) studied the behavior of anchor using
finite element method. Swami Saran et a1 (1986) developed an analytical procedure to predict
the load displacement characteristics of shallow anchors using non-linear constitutive
relationship. Das and Puri (1989) proposed a non-dimensional empirical load dispIacement
relationship for shallow plate anchors based on their laboratory investigations. Trautman and
Kulhawy (1988) have presented a mathematical formula utilizing a rectangular hyperbola to
represent the load displacement curves for horizontal anchor plates subjected to vertical
pullout for embedment ratio up to 3, for all types of soils. As reported by Rameshbabu
(1998), the variation observed between the actual load displacement response of the anchors
in sandy soil and the curve obtained by using the equation recommended by Trautman and
Kulhawy is very large. Rameshbabu (1998) developed two different equations to determine
load - displacement relationship, one for granular soils and other for cohesive soils. All the
above researchers have used the results of small-scale anchors for the characterization of load
-displacement behavior. In order to predict the response of horizontal plate anchors more
accurately, an attempt has been made in the present investigation to generalize the load
displacement curves obtained from FLAC analyses of large size anchors.
110
. . . , \ , \ \!\I ttttf t f l l f 1 I , d . .

Fig 4.37a Displacement vectors at failure for different diIation angles (y)
B=0.5m, DB=5
-xKn33WX
x%wE#wX

>000<X1(~
Plasticity Indicator
'at yield in shear or vol.
X elastic, at yield in past
o at yield in tension

Fig. 4.37b Plastic regions at failure for different dilation angles (y)
@=32.5", B=0.5m, D B = 4
Fig 4.38 Displacement vectors and plastic regions at failure for different dilation angles
B=l m, D/B=2
Scale
4.3 cm

- 3.95 crn ~
pb 3.81 cm

Fig. 4.39a Displacement vectors at failure for different dilation angles


B=OSm, DB=6
Plasticity Indicator
* at yieldin shear or vol.
X elastic, at yield in past
o at yield in tension

Fig 4.39b Plastic Regions at Failure for different dilation angles


B=0.5m, DA=6
The pullout load for different conditions from the analysis presented earlier is summarized in
Fig 4-40 to 4.43. And the displacement at failure for different conditions is summarized in
Fig 4.44 to 4.47. Pullout resistance increases with increase in width of the strip anchor. This
effect is more significant in deep anchors than the shallow anchors as shown in Figs 4.40 to
4.43 for different dilation angles. Failure displacements also significantly increase with
increase in width of the strip anchor as shown in Figs 4.44 to 4.47 for different dilation
angles.

Figure 4.48 show the normalized load displacement curve P/P, vs Z,Z, for the range of B
from 0.5m to Im, DL3 varying from 1 to 8 for different types of sands. It can be seen that the
plots for different cases fall in a narrow band, which can be considered to be unique for
practical purposes. Das and Puri (1989) have proposed this form of generalization for
shallow square and rectangular anchor plates embedded in medium and dense sands. The
same normalization form is adopted in the present analysis. From the above generalized
curve in Fig 4.48, if we know the ultimate pullout load P, and the corresponding ultimate
displacement, Z, then one can estimate the complete load displacement curve for a given
geometry and soil properties.

The pullout load is expressed as dimensionless pullout capacity factor (Pu/yBD), Fig 4.49
shows the normalized plot, of pullout capacity factor (P,IyBD) versus dilation angle (y) for
anchors of width 0.5m, 0.75m and 1 .Om, the data being extracted from the load displacement
curves for different cases presented earlier. By using this plot, one can estimate the ultimate
pullout capacity (P,) of the anchor for the given soil properties and geometric parameters.

The ultimate displacements (ZJ corresponding to P,, can be conveniently expressed in


dimensionless form as relative failure displacements (&/El)%. A general increase in Z,B
with DL3 occurs in all the tests, for width of anchors OSm, 0.75m and lm as shown in
Fig.4.50. Even though the displacements are decreasing with increase in dilation angles, the
variations are small. Figure 4.50 Summarizes data from different load displacement curves to
show the variation of &/B with DIB for size of anchors OSm, 0.75m and lm, with different
dilation angles. By using this compiled data, one can estimate ultimate displacements for
different sizes of anchors and soil properties.
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorpalte (Bin m)

Fig 4.40 Variation of P, with B

0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorplate (6 in m)
Fig 4.41 Variation of P, with B
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.65 0.90 0.95 1 .oo
Width of Anchor plate (€3 i n m)
Fig 4.42 Variation of P, with B

0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of the Anchorplate (B in m)
Fig 4.43 Variation of P, with B
I i
.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorplate (B in rn)
Fig 4.44 Variation of Z, with B

0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorplate (Bin m)
Fig 4.45 Variation of Z, with B
119
0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorplate (B in rn)

Fig 4.46 Variation of Z, with B

0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00
Width of Anchorplate (Bin m)

Fig 4.47 Variation of Z, with B


AS stated earlier, all the above analyses have been carried for a soil with a basic critical
friction value of 32S0, which is the most probable value for many sands in general. However,
this critical friction angle may vary over a small range for different sands. To examine the
effect of variation of critical friction angle 6," on the pullout behavior of anchors, analysis is
carried'out for different 0," values of 25", 27.5", 32S0, 35" for the case of y=0 only. It is
likely that for other higher values also, the effect of variation in (I,,will be of sirnilar
nature.

Figure 4.5 1a,b, &c show the variation of pullout capacity factor (P,/yBD) with critical friction
angle for sizes if anchors 0.5m, 0.75m and lm. The increase in pullout capacity factor is
more significant in higher embedment (DIB) ratios. For lower D/B ratios it is nearly constant.

25.0 27.0 29.0 31.0 33.0 35.0 37.0


Critical friction angle ,$,

Fig 4.51a Variation of pullout capacity factor with critical friction angle
25.0 27.0 29.0 31 .O 33.0 35.0 37.0
Critical friction angle (I,,
Fig 4.51 b Variation of pullout capacity factor with critical friction angle

25.0 27.0 29.0 31 .O 33.0 35.0 37.0


Critical friction angle 4,"
Fig 4.51 c Variation of Pullout capacity factor with critical friction angle
4.9 Summary
The load deformation behavior of large size horizontal strip anchors has been modeled
as plane strain problem using a two-dimensional finite difference program FLAC. The soil
has been modeled with a Mohr-Coulomb stain softeninghrdening soil model. Formulation
of the problem and the validation of the numerical program FLAC with experimental and
results have been presented. The frictional resistance of the soil is characterized in
terms of critical friction angle (9,") and dilation angle (y), being components of peak friction
angle (4). The dilatency of a soil mass gradually decreases with continued shearing from its
initial high value to zero after very Iarge shear strains, when the soil finally reaches a
constant, steady volume at critical states. Correspondingly the observed friction angle @
reduces from its peak value to $,, at a very large strains. A series of simulations have been
carried out.on large size strip anchors with parametric variation to study the pullout capacity
and load deformation behavior. The parameter varied are the geometric parameters of width
of anchor (B) and embedment depth (D/B), and the soil properties of critical friction angle
($,,) dilation angle ( y ) , unit weight (y) and Elastic modulus (E).

An attempt has been made to generalize the load displacement behavior of large size strip
anchors. The results are presented in the form of design charts, which can be used in hand
calculations to obtain an estimate of anchor capacity and deformation for a wide range of soil
types and size of anchors.
CHAPTER 5

ANALYSIS OF
HORIZONTAL STRIP ANCHORS IN CLAY

5.1 Introduction
This chapter deals with the analysis of the drained and undrained behavior of large
size horizontal strip anchors in clays using a two-dimensional explicit finite difference
program FLAC with modified Cam-clay model. Earlier investigators have studied the
undrained behavior of anchor plates in clays. No studies are reported in literature for the
drained behavior of anchors in clays. It is not clear whether, drained or undrained condition
will be critical for an anchor. A few investigators (Meyerhof and Adams; 1968 and Davie and
Sutherland; 1977) have proposed theoretical solutions for the pullout resistance of anchor
plates in clays under undrained condition. These theories have been developed by conducting
experiments on small size anchors. By using finite element analysis and Mohr-Coulomb soil
failure criterion, Rowe (1982a) studied the undrained behavior of model anchors in clays.

In the present investigation an attempt has been made to study the drained and undrained
behavior of large size horizontal strip anchor plates in both normally consolidated and
overconsolidated states using the Modified Cam-Clay model. Cam-clay model, which was
specially developed for clays, reflects the hydrostatic pressure or density dependent
hardening material response. Implementation of this model has been discussed in Chapter 3.

Normally consolidated soil is one, which has never been subject to an effective ~rkssure
greater than the existing overburden pressure. Clay is said to be over consolidated if it has
ever been subjected to a pressure in excess of its present overburden pressure. The maximum
overburden pressure to which the soil has been subjected to and under which it got
consolidated is known as pre-consolidation pressure.
The behavior of clays (being density dependent hardening materials) depends primarily on
the current state of the soil, which is in turn dependent on the past stress history experienced
by the soil.

In earlier days, clays used to be characterized by the strength parameters c and 4. Often,
under undrained conditions, @ would be even considered zero. But in the recent
developments, it is understood that all the strength of clays is frictional. There is nothing like
cohesion. The part of shear strength, which appears to be independent of normal stress, is
shown to be the effect of over-consolidation and the resulting dilation. Thus, although Cam-
clay model uses zero cohesion for all clays, it reflects this component of strength through
over-consolidation and in a more realistic way. Hence, it is appropriate to consider the pre-
consolidation pressure as a parameter in the analysis. Instead of using arbitrary parameters
we have selected known Cam-clay model parameters of London Clay to examine the
behavior.

The results have been presented in the form of plots of the drained pullout load versus
displacements for normally and overconsolidated clays and the undrained pullout load versus
displacements for normally consolidated clays.

5.2 Formulation of the problem

The same boundary conditions used for the analysis of anchors in sand discussed in
Chapter 4 (Fig. 4.1) has been used for the present investigation. In this case also, the
boundaries of the soil domain are located at 10 times the width of the anchor plate from the
center of the anchor plate. The chosen size of the domain is sufficiently large to have no
boundary effects on the results. A fine uniform grid has been adopted for the analysis since
there were some difficulties in using non-uniform grid, as used in the case of sands, due to
the sensitivity of Cam-clay model. The soil has been characterized using the modified Carn-
clay model, as explained in the earlier Chapter. The anchor plate is treated as a beam
element as before. The tie rod is rigidly connected to the anchor plate, but the frictional
interaction of the tie rod with the surrounding soil mass is neglected. Thus the load applied
on the top of the tie road is directly transferred to anchor plate.
Tfle soil domain is initially equilibrated to gravity stresses. An elastic model with a poison*^
of 0.4 is used during this stage. This results in a KO value of about 0.67, which is quite a
reasonable value. Then the soil model is changed over to Cam-clay, and equilibrated. This
constitutes the initial state for the normally consolidated cases. For cases where over-
corlsolidation is desired, the specific uniform stress (25kpa to 200kpa) is applied on the top
horizontal surface under KO conditions, (i.e. lateral displacements are not allowed), after
equiIibrium, the surface loading is removed and equilibrated again. This results in different
degrees of over-consolidation with depth. The value of KO also varies with depth. This is
more appropriate for natural soils.

The simulations are made under strain controlled test by applying a displacement at the top
node of the tie rod as done earlier for the case of sand.

The convergence criterion for FLAC is the nodal unbalanced force, i.e., sum of the forces
acting on a node from its neighboring elements. If a node is in equilibrium, these forces
should sum to zero. For this study, the unbalanced force of each node has been kept at less
than the 0.1 percent of the gravitational force.

5.3 Analysis of large size anchors

For the purpose of analysis, it is necessary to have defined Cam-Clay model


parameters, which can not be chosen arbitrarily, as usually done in a parametric study. For
this analysis we have selected as stated earlier, the model properties of London clay, which is
readily available in literature (Schofield and Wroth, 1968) are presented in Table 5.1

Table 5.1 Model Properties of London Clay

Soil Constants London Clay

h 0.161

K 0.062

r 2.759

M 0.888

G 7.92Mpa
Note: the value of I? is that for p ' = l . ~ k ~ m - ~
For the above properties the VL is equal to 2.83 and the dry density of the London clay is
O54kg/m3at 1kpa stress level.

Due to the limitations of time, the analysis could not be carried out on different soils.
Different soil types are reflected primarily in the value of their compressibility, i.e. h, K and
in the frictional parameter M or equivalently $. The value of effective angle of internal
friction for different clayey soils does not usually vary over a wide range, it may be in the
range of 20"-28'. However h and K could vary considerably. This is an area, which needs
further study.

FLAC simulations of pullout response of large size anchor embedded in homogeneous


London clay are carried out with different over consolidation ratios for drained condition and
undrained condition only for normally consolidated state. The geometric parameters
considered are the width of the anchor plate B and the depth of the anchor, D (specified as
D/B ratio):Three sizes of strip anchors, B=0.5m, 0.75m and l m are considered. Embedment
ratios (Dm)of 2 to 8 are considered. This may cover the practical range of sizes of anchors
in the field.

The material properties of the anchor plate are kept constant. It is assumed that the plate is
sufficiently stiff as not to affect the pullout response. The strain-controlled simulations are
made by applying a displacement at the top node of the tie rod and corresponding load is
recorded.

5.4 Discussion on Simulation Test Results


5.4.1 Undrained Simulations on Normally consolidated soils

Figures 5.1 to 5.3 show, the load displacement curves of normally consolidated soils
for different D/B values and different sizes of anchor. In this case also, to fix the uItimate
pullout capacity, P,, the K2 criteria as adopted in the case of sands has been adopted. For this
undrained case the ultimate pullout Puvalues appear to be linearly proportional to D B and B.

The typical displacement vectors and plastic regions at ultimate loads have been presented in
Figs. 5.4 and 5.5. Similar to earlier discussion on sand, the surface heave is indicated only
D/B = 6 D/B = 8

Fig 5.4 Displacement vectors at failures for undrained conditions


Plasticity Indicator
'at yield in shear or vol.
X elastic, at yield in past
o at yield in tension

Fig 5.5 Plastic regions at failure in normally consolidated soil for undrained case
h,r D/B values less than 4 and there is no surface heave for D/E3 greater than 4. This shows
that the critical depth ratio is less than 4. The displacement pattern indicates a characteristic
rotational field due to no volume change condition, as was also indicated earlier by Rowe
( 1982a).

5.4.2 Drained Simulations on Normally consolidated soils

Figures 5.6 to 5.8 show the pullout load versus displacement plots for different D/B
values for Im, 0.75m and 0.5m size of anchors. It can be seen that, the pullout load increases
alnlost linearly with DIB value.

It is interesting to note that, for a given identical condition, the undrained strength is higher
(40 to 50%) than the drained strength indicating that long-term stability is critical for anchors
in clays. Further, it may be more so for over-consolidated states, since the mobilized
negative pore pressure may indicate higher undrained strength initially which will not be
available in drained condition. Hence, for over consolidated condition only drained analysis
has been carried out.

5.4.3 Drained Simulations for over-consolidated clays

Figures 5.9 to 5.19 shows the plots of pullout load versus displacement for different
values of pre-consolidation pressure (PC). In these plots both sizes of anchors and D/B values
have been varied.

It can be seen that, for a higher PC(say 200kpa) there is substantial increase in pullout load
over that of with a lower Pc(say 25kpa). This increase is more significant for higher D/B
ratios. However, the failure displacement appears to increase up to D/B value of 4 and
decrease to a constant value beyond. This could be due to the difference in behavior
between shallow and deep anchor.

Further, it can be seen that, at lower values of D/B (say 2) and at PC greater than lookpa,
which means a very high over consolidation ratio (OCR) at the anchor level, soil exhibits
softening behavior which has lead to instability in the present simulations and is reflected in
the non-uniform load displacement curves shown in Figs, 5.9,5.13 and 5.17.
+
W
+
W
+
W
+
W
LO Ln LO LD * 8+ * 8+ 0
0
+
LLJ
0
w
+ 0
w
+ +
0
w
+
0
w W
+
0
w W
+
0
W
C9
.r
-?
7--
cu
T-
9
F
9
03
9
CD
9
9
9
CU
9
0
'd-
o
+
W
0
Lcj
PC= 50kPa

PC= 25kPa

NC

D/B = 2
B = 0.5m NC - Normal Consolidation Soil

O.OE+OO 1.OE-02 2.OE-02 3.OE-02 4.OE-02 5.OE-02 6.OE-02 7.OE-02 8.OE-02


Displacement (m)

Fig 5.17 Load displacement curves for drained case


NC

D/B = 4
B = 0,5m NC - Normal Consolidation Soit

O.OE+OO 1.OE-02 2.OE-02 3.OE-02 4.OE-02 5.OE-02 6.OE-02 7.OE-02 8.OE-02 9.OE-02 1.OE-01

Displacement (m)
Fig 5.18 Load displacement curves for Drained case
Figure 5.20a,b&c presents the variation of pullout load (P,) with D/B for different pre-
consolidation pressure (PC)values and different sizes of anchors. The P, also varies linearly
with DIB for given PC values. This plot is a summary of all the results.

Figure 5.21a,b,c,d&e presents the variation of pullout load (P,) with the size of anchor plate
(B) for different DfB ratios and pre-consolidation pressure PC. The pullout capacity increases
nearly linear with the increase in width of the anchor plate.

Figure 5.22 show the plots of percentage of normalized ultimate displacement (Z,/B%)
I ratios for different pre-consolidation pressure and for different sizes of anchor.
versus DB
The failure displacements appear to increase up to D/B value of 4 and decrease to a constant
value beyond. This could be due to the difference behavior in between shallow and deep
anchor.

Figure 5.23 show the normalized plots of all the simulations carried out in this investigation,
with P/P, versus Z / L . The best-fit curve, statistically obtained is also marked in the figure.
The general equation for the curve can be written as y=-0.9943x2+1.9523x+0.0192.

The above equation or curve (Fig. 5.23) can be used to obtained the load displacement curve
for any size and depth of anchor knowing ultimate load (P,) and ultimate displacement (&).

Figures 5.24a&b typically shows the displacement vectors and the plastic regions inside the
soil mass at ultimate failure loads for different PCvalues and for an anchor of size of 0.5m
and D/l3 of -2. It is seen that at PCequal to zero (i.e. normally consolidated), the displacement
vectors are nearly vertical and the surface heave is less than what it is observed for higher PC
values. The size of plastic zone is well spread near the surface for normally consolidated soil
and with increase in PC,the plastic zone shrinks in size.

Figures 5.25a&b and 5.26 show typically the displacement vectors and the plastic zones
inside the soil mass at ultimate failure loads at different D/B values for an anchor size of one
meter at a pre-consolidation pressure of 25kpa. With increase in Dm, the surface heave
reduces. The plastic zone is close to the surface for DIE3 of 4 or less, and confines to a zone
around the anchor for D/B greater than 4. Probably this could be the border between shallow
500.0
-+ NC
-+-- Pc=25kpa
400.0 -
-t- Pc=5Okpa

-e- Pc=l OOkpa


300.0 - -x-Pc=2OOkpa
3
e
200.0 -

100.0 -

0.0 2.0 4.0 6.0 8.0 10.0


D/B
Fig 5.20a Variation of Pu wiht D/B, B=lm

-t- Pc=50kpa

-+- Pc=l OOkpa


-+Pc=200kpa

Fig 5.20b Variation of Pu with D/B, B=0.75m


152
0.0 2.0 4.0 6.0 8.0 10.0
DIB
Fig 5 . 2 0 ~Faviration of Pu with DIB, B=0.5m

Width of anchor plate (m)

Fig 5.21a Variation of Pu with B for Normally consolidated soil


0.0 I I I I

0.5 0.6 0.7 0.8 0.9 1,O


Width of anchor plate (m)
Fig 5.21 b variation of pu with B for Pc=25kpa

I
250.0 -

-+- D/B=2
200.0 - +D/B=4
-t-- D/B=6
150.0 +D/B=8

0.5 0.6 0.7 0.8 0.9 1.O


Width of anchor plate (m)

Fig 5.21c Variation of Pu with B for Pc=50kpa


-+- D/B=2

-t- D/B=4
n
300.0 -.

Z It- D/B=6
5
t~ -o- D/B=8
m
2 200.0
+Ir
3
-03
0,

?
0.0 I I I J

0.5 0.6 0.7 0.8 0.9 1.O


Width of anchor plate (m)

Fig 5.21 d Varation of Pu with B for Pc=100kpa

0.0 -i 4 I
0.5 0.6 0.7 0.8 0.9 1.O
Width of anchor plate (m)
Fig 5.21 e Variation of Pu with B for Pc=200kpa
---"- - --- --- --*.-- . -
-+- NC
-u- Pc=25kpa
-t.Pc=5Okpa
+Pc=l OOkpa

D/B
Fig 5.22a Variation of ZU/B~/~
with D/B for B = l m

0.0 2.0 4.0 6.0 8.0 10.0


D/B
Fig 5.22b Variation of Zu/B% with D/B for B=0.75m
Fig 5 . 2 2 ~Variation of ZulBOhwith DIB, B=0.5m
Pc=50kpa PC=100 kpa

Pc=200kpa
Fig 5.24a Displacement vectors at failure for drained condition at different pre-
consolidation pressure, B=OSm, D/B=2
Normally consolidated soil Pc=25 kpa

P,=50 kpa PC=100 kpa

Plasticity Indicator
'at yield in shear or vol.
X elastic, at yield in past
o at yiaM in tension

P,=200 kpa

Fig 5.24b Plastic regions at failure for different pre-consoIidation pressure,


B=0.5m D/B=2
DA3=4

Fig 5.25a Displacement vectors at failure for drained case, B=lm, PC=25kpa
Fig 5.25b Displac:ement vectors
Plasticity Indicator
* at geld in shear or vd.
X elastic, at yield in past
o at yield in tension

D/B=6 D/B=8

Fig 5.26 Plastic regions at fa21.ire for Drained condition, B=lm, PC=25kpa
and deep anchor as reported by earlier investigators (Meyerhof and Adams; 1968 and Das;
1980).

5.5 Summary
Analysis of drained and undrained behavior of large size horizontal strip anchors in
clay using modified Cam-clay model has been carried out. It is not clear that whether drained
or undrained condition will be more critical for an anchor. A series of simulations have been
made on normally consolidated soil both drained and undrained conditions. It has been
observed that the undrained pullout capacity of an anchor in a soil of normally consolidated
stated will always be more than the drained capacity. This is in contrast to the usual
understanding that undrained behavior is more critical than the drained behavior. A series of
simulations have been carried out in normalIy consolidated soil and over consolidated soil
(25kpa, 50kpa, IOOkpa and 200kpa). The results have been presented in the form of plats of
the drained pullout load versus displacements for normally and over consolidated clays and
the undrained pullout load versus displacements for normally consolidated clays.
CHAPTER 6

ANALYSES OF
HORIZONTAL STRIP ANCHORS IN LAYERED SOILS

6.1 Introduction

During the past few years a great number of laboratory model and large-scale field test
results on the pullout resistance of horizontal anchor embedded in homogeneous sand has
been reported by many investigators. A review of related literature shows that not much work
has been done to analyze the behavior of anchor plates in layered soils a problem, which is
often encountered by the practicing engineer in the field. Stewart (1988) and Bowzza and
Finlay (1990) investigated experimentally the behavior of circular plate anchors buried in two
layered soil. Manjunatha (1997) has developed analytical expressions to estimate the uplift
capacity of strip and circular anchors in layered soils. Rameash Babu (1998) suggested an
expression to determine the pullout resistance of strip, square and circular anchors in layered
soils.

In the present investigation an attempt has been made to analyze the behavior of large size
anchros in two layered sand using an explicit two-dimensional finite difference program
FLAC. Soil is assumed to be a Mohr-coulomb strain softeningkardening material.

The geotechnical properties of backfill of anchor foundations are very sensitive to


construction and compaction techniques. There is no satisfactory method to analyze the
behavior of anchors in such inhomogeneous soil conditions.

An attempt has been made in this chapter to analyze the behavior of anchor plates having the
backfill of higher or lower strength than the native soils for different shapes of excavation.
6.2 Analyses of anchor behavior in two layered sand

The two layered soil for this analysis consisted of two cases a) a layer of loose sand
stratu~noverlaid by a dense sand layer and b) a layer of dense sand stratum overlaid by it

loose sand layer (Fig 6.1). For the present analysis we have chosen published properties of
Chattahoochee River Sand both at dense and loose corditions (Vesic and Clough; 1968). The
properties of the soil used for the analysis are shown in Table 6.1

Table 6.1 Soil properties used in the analyses


Property Loose condition Dense condition
Y 13.17 k ~ / m ~ 15.43 k ~ / r n ~

E 6 Mpa 19.6 Mpa I


In the analyses, width of the anchor plate (B) is considered as one meter and the embedment
ratio is varied from 2 to 8. The upper layer thickness (D,) is varied from minimum of B to
maximum of (D+2B). The material properties of the anchor plate are kept constant. It is
assumed that the plate is sufficiently stiff as not to affect the pullout response. The same grid
and boundary conditions used for the analyses of anchors in sands (Fig 4.2) is also used for
the present analyses. The anchor plate is treated as a beam element. The tie rod is rigidly
connected to the anchor plate, but the frictional interaction of. the tie road with the
surrounding soil mass is neglected. The strain-controlled simulations are made by applying a
displacement at the top node of the tie rod and the corresponding load is recorded. A
dimensionless parameter, upper layer thickness ratio Dt/B is introduced. This is the ratio of
upper layer thickness D,to width of anchor plate. D,/B is varied from I to 6 for D/B of 4, 1 to
8 for D/B of 6 and 1 to 10 for D/B of 8.

Fig. 6.1 Problem analyzed


166
Figures 6.2 to 6.4 shows the Ioad displacement relationship for an anchor cn~beddedi n a
layered soil bed. The bottom layer is dense sand and the top layer is loose sand for different
upper layer thickness ratio Dt/B . For all the cases of D,/B, the load displacement curves are
falling in between the load displacement curve of homogenous loose sand and that of dcnse
sand. It can be seen that even when D t B = DIB, i.e. when the top layer depth is upto the
anchor depth level, the load displacement behavior is not just equal to that for the case of
homogeneous soil, especially at s1na1l displacements, and for lower D/B cases. This shows
the soil beneath the anchor level also participates in the pullout resistance. However, the
difference in ultimate pullout load is not much for Dt/B values greater than D/lB . Hence it is
clear that in sand the Ioad deformation behavior depends on the layer of soil above the anchor
plate. For deep anchor (D/B>6) there is no difference in behavior of load deformation curve
up to D,/BIl, it is nearly following the same behavior of homogeneous sand.

Figure 6.5 shows the variation of P, with D,/B for different D/B ratios. The ultimate pullout
capacity is decreasing with increase in D,/B in the case where bottom layer is dense sand and
top layer is loose sand. Figure 6.6 shows the variation of ultimate displacement Z,,
with D,/B
for different D/B ratios. The ultimate displacement also decreases with the increase in Dm.

Figures 6.7 to 6.9 show the load displacement curves for D/B of 4, 6, and 8 at different upper
layer thickness ratios for the case of top layer dense sand and the bottom layer is loose sand.
In this case also for all the values of D,/B, the load displacement curves are falling in between
the load displacement curve of homogeneous loose and that of dense sand conditions. Even
by completely replacing the layer above the anchor from loose sand to dense sand
(DJ3 2 (Dm)) it is not following the same load displacement curve of the anchor in
homogeneous dense sand like in the earlier case with bottom layer of loose sand.

Figures 6.10 and 6.11 show the variation of Pu and Zu with upper layer thickness ratio DJB.
The ultimate pullout capacity and ultimate displacement are decreasing with the increase in
D1/B where the bottom layer is dense sand and top layer is loose sand, while it is the reverse
for the case where bottom layer is loose and top layer is dense sand.

Figure 6.12 shows the normalized load displacement curve for the all the cases where the top
layer is dense sand and bottom layer is loose sand and the top layer is loose sand and the
bottom layer is dense sand. Once we estimate the uItimate pullout load (P,) and ultimate

167
+
W
+
W i!i
+
W
+
W
+
W
+
W
+
W '
f
W
9
9
CO
9
b
9
CO
9
10 2 9
m CV
9
F 2
Top layer - Dense Sand
Bottom Lyer - Loose sand I

r I

Fig 6.5 Variation of P, Mth D@

1.OE-01
t------ Top layer - Dense Sand
Bottom Lyer - Loose sand

Fig. 6.6Variation of Z, with D@


Top layer - Loose Sand
-
Bottom Lyer Dense sard

B=l m

DB=6

w
Fig 6.10 Vaiiation of P,uith D@

6.0E-01
Toplayer - L m e W
Bottom L p r - Densesancl
5.0E-01 B=l m

0.0 1.O 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
4B
Fig 6.11 VariationofZ,vith i7B
Fig 6.12 Normalised load-displacement curve for layered sands
displacement (Z) (from plots 6.5, 6.6 and 6.10 & 6.1 I), one can get the complete load
deformation behavior of the anchors in a layered sands, using the above generalized load
displacement curve.

Figures 6.13a&b and 6.14a&b show the dispIacement vectors and plastic regions at failure for
D/B value of 4 at different DJB values. Fig 6.13a&b is for the case of top layer loose sand
and the bottom layer dense sand and Fig 6.14a&b is for the case of top layer dense sand and
the bottom layer loose sand. It can be clearly seen that the plastic zones are narrow and
vertical in a loose medium, while they are spread over a wider region in denser medium. The
spreading of failure surface is sharply seen at the level where the medium changes from loose
to dense sand.

6.3 Analysis of anchor plates with backfill having higher or lower


strength than the native soil

The backfill anchor foundations are very sensitive to construction techniques because
they must rely on the available soil, which is normally excavated and replaced. If the backfill
is loosely filled, then it may have a lower strength than the native soil. Conversely if the
backfill is compacted very well, then it is likely to have higher strength than the native soil.
There are several possible variations in the construction and backfilling. The excavation wall
may be vertical or inclined. For the present investigation, analysis is carried out for the two
cases, where the excavation wall is vertical and at an angle of 4.5' with the edge of the anchor
plate as shown in Fig 6.15.

Fig. 6.15 Shape of Excavation walls considered

177
Scale
-------+ 12.8cm

Plasticity Indicator.
'at yield in shear or vol,
X elastic, at yield in past
o at yield in tension

Fig 6.13a Displacement vectors and plastic regions at failure in layered soils.
top layer - dense sand
bottom layer- loose sand
B= 1m D/B=4
Scale

Plasticity Indicator
* at yield in shear or MI.
X elastic, at yldd in past

w m o at yield in tension N@xw


\/ \/

Fig 6.13b Displacement vectors and plastic regions at failure in layered soils.
top layer - dense sand
bottom layer- loose sand
B= 1m D/B=4
Scale
p
b 12.5cm

* at yield in shear or vol.


X elastic, at yield in past
o at yield in tension

Fig 6.14a Displacement vectors and plastic regions at failure in layered soils.
Top layer - loose sand
Bottom layer- dense sand
B=lm D/B=4
Scale
--------) 12.5 cm

Plasticity Indicator
at yield in shear or vol.
X elastic, at yield in past
o at yield in tension

Fig 6.14b Displacement vectors and plastic regions at failure in layered soils.
Top layer - loose sand
Bottom layer- dense sand
B= 1m D/B=4
The same finite difference grid and boundary conditions discussed in Chapter 4 for sands are
used here for the present investigations. Soil is assumed as a Mohr-Coulomb strain
softeninghardening material. It is assumed that proper trenching is provided at the time of
excavations. The anchor plate is treated as a beam element. Strain-controlled sin~ulationsare
made as before.

The present investigation is carried out for the two basic case, 1, where the basic soil is dense
homogeneous sand and the backfilling is done with the loose sand. This is the condition
where the backfill has a lower strength than the native soil. In case 2, in a homogeneous loose
sand, backfilling is done with the dense sand. This is a condition where the backfill has
higher strength than the native soil.

The width of the strip anchor is considered to be l m for the analysis and embedment depth
ratios (DB)
of 2,4,6, and 8 are used. Tables 6.2 and 6.3 show the values of ultimate pullout
load and ultimate displacements for the two cases at different D/B vales for excavation wall
vertical and inclined at 45".

Table 6.2. Dense native sand with Loose sand backfill


S.No. D/B Excavation PU(w ZU(cm)
Wall
1 2 Vertical 70 2.6

-
3 4 Vertical 200 9.3

I I I I

5 6 Vertical 395 24.5 .

7 8 Vertical 650 48
8 8 45" 420 31
Table 6.3 Loose native sand with Dense sand backfill

S.No D/B Excavation


Wall
PU (w ZJ( ~ 1 1 7 )

1 2 Vertical 63 4.6

2 2 4.5" 72 5

3 4 Vertical 170 12.8

4 4 45" 200 13.2

5 6 Vertical 335 28.2

6 6 45" 410 32

7 8 Vertical 550 53

8 8 45" 660 58

Figures 6.16 to 6.19 show the load displacement curves for different D/B ratios of 2,4, 6 and
8. Obviously, the load displacement curves for the excavated- backfilled conditions fall
within the extreme cases of homogeneous dense and loose sand conditions. Also, the effect
of 45"-backfilling is more than the vertical-backfilling. Further, the effect of backfilling is
significant only at large displacements close to the ultimate loads. At smaller displacement
levels, the behavior remains nearly the same as that of the native soil. This shows that it is the
soil below the anchor level that carries the load at small displacement levels. Vertical
backfilling appears to be effective by about 20% and 45"-backfilling by more than 50%
approximately.
i.e. For example, dense vertical backfill over loose sand will have a
Pu=P,-l,ose + 0.20 (Pu-dense - Pu-loose) (6.1)
Loose vertical backfill over dense sand gives
Pu=Pu-dense - 0.20 (Pu-dense - Pu-loose) (6.2)

Dense 45"-backfill over loose sand gives


Pu=Pu-lome + 0.5 (Pu-dense - Pu-100s~) (6.3)
Loose 45"-backfill over loose sand gives
Pu=Pu-derue - 0.5(Pu-dense - Pu-loose) (6-4)
The failure displacements are also correspondingly altered, higher displacements with
increase in the extent of loose soils.
183
Dense sand

BFDV
B
Loose sand -1

BFLV - Back fill with loose sand


(Excavation wall vertical)
BFLl - Back fill with loose sand
(Excavation wall inclined at 45')
BFDV - Back fill with dense sand
(Excavation wall vertical)
BFDl - Back fill with dense sand
(Excavation wall inclined at 45'1
O.OE+OO W
I I 1

O.OE+OO 5.OE-03 1.OE-02 1.5E-02 2.OE-02 2.5E-02 3.OE-02 3.5E-02 4.OE-02


Displacement (m)
Fig 6.1 6 Load displacement curves for Backfill with higher
or lower strength than the native soil
Displacement (m)
Fig 6.17 Load - displacement curves for Backfill with higher
or lower strength than the native soil
4.5E+05 - Dense sand
BFDl
4.OE+05 -

3.5E+05 -

n
5 3.OEt-05 -
u
m
- 2.5E+05
0
.c,
-

-0
3

3 2.0E+05 - BFLV - Back fill with loose sand


s
00 (Excavation wall vertical)
a
1.5E+05 -
BFLl - Back fill with loose sand
(Excavation wall inclined at 45')
1.OE+05 -
-
BFDV Back fill with dense sand
(Excavation wall vertical)
BFDl - Back fill with dense sand
5.OE+04 -
(Excavation wall inclined at 45')

Displacement (m)
Fig 6.18 Load - displacement curves for Backfill with higher
or lower strenght than the native soil
Displacement (m)
Fig 6.19 Load - displacement curves for backfill with higher
or lower strength than the native soil
6.4 Summary
An attempt has been made to analyze the pullout capacity and the load deformation
behavior of large size anchors in layered soils and for situations where the backfill soil has
higher or lower strength than the native soil. Two layered sands are considered as two cases:
a) top layer dense sand and bottom layer loose sand and b) top layer loose sand and bottom
layer dense sand. Results are presented in the form of load displacement curves. To study the
effect of backfill having lower or higher strength than the native soil, analysis has been made
for two conditions of excavation wall vertical and inclined at an angle of 45". Though the
values of P,, falls in between for homogeneous conditions of loose and dense sand based on
the results a simple method of calculations has been proposed for the two combinations of
backfilling.
CHAPTER 7

RESULTS AND DISCUSSIONS

This chapter presents in summary the critical examination of the present work
reported in this thesis. The important conclusions drawn from the present investigations are
also listed.

A number of investigators have used the results of small size model anchors to understand the
behavior and extrapolated the results to predict the behavior of large size anchors. This may
lead to unsatisfactory results. It has been clearly shown by Dickin (1989) that the failure
displacements and load deformation curve patterns are very different for small and large size
anchors, i.e. they are not just proportional to the size of the anchors. Critical pullout load and
load deformation behavior are required for the complete analysis of the anchors. Though
many theories have been presented to predict the uplift capacity with in the limits of accuracy
required at engineering level, at present no simple rational method is available for computing
deformations.

By using numerical techniques one can simulate the field conditions more realistically. Fast
Lagrangian analysis of continua (FLAC) is a two dimensional finite difference program and
can be used effectively to model the field conditions of anchors to analyze the load
deformation behavior of large size anchors, with a suitable constitutive model for soil.

Mohr-Coulomb strain softeninghardening model has been used in the analysis as a soil
constitutive model for sands. In this model the cohesion, friction angle, dilation and tensile
strength may harden or soften after the onset of plastic yield. The frictiona1 resistance of the
soil is characterized in terms of critical friction angle (4,") and dilation angle y, being a
component of peak friction angle 4. Modified Cam-clay model has been used in the analysis
as a constitutive model for clays. It reflects the hydrostatic pressure or density dependent
hardening material response.
In Chapter 4, a series of simulations have been carried out on large size strip anchors with
~arametricvariation to study the pullout capacity and load - deformation behavior. It has
been observed that the pullout capacity increases with increase in dilation angle and
embedment depth. The critical embedment depth depends on the dilation angle. The load
displacement behavior of large size strip anchors has been generalized and the results are
presented in the form of the design charts, which can be used in hand calculations to obtain
an estimate of anchor capacity and deformation for a wide range of soil types and anchors. .

Chapter 5 discusses the analysis of drained and undrained behavior of large size horizontal
strip anchors in clay using modified Cam-clay model. In earlier days, clay used to be
characterized by the strength parameters c and $. It is not clear from literature whether
drained or undrained condition will be more critical for an anchor. A series of simulations
have been made on normally consolidated soil under both drained and undrained conditions.
It has been observed that the undrained pullout capacity of an anchor in a soil of normally
consolidated state will always be more than the drained capacity. This is in contrast to the
usual understanding that undrained behavior is more critical than the drained behavior.
Hence a series of simulations have been made for drained condition in normally consolidated
soil and over consolidated soil with a preconsolidation pressure varying from 25kpa to
200kpa. The results have been presented in the form of plots of the drained pullout load
versus displacements for normally and over consolidated clays and the undrained pullout load
versus dispIacements for normally consolidated clays.

Due to the limitations of time, the analysis could not be carried out on different soils.
Different soil types are reflected primarily in the value of their compressibility, i.e. h, K and
in the frictional parameter M or equivalently 9. The value of effective angle of internal
friction for different clayey soils does not usually vary over a wide range, it may be in the
range of 20"-28". However h and K could vary considerably. This is an area, which needs
further study.

Results are presented in the form of load displacement curves. In Chapter 6, an attempt has
been made to study the pullout capacity and the load deformation behavior of large size
anchors in layered soils and for situations where the backfill is of higher or lower strength
than the native soil. In layered soils, for all the cases whether the top layer is loose sand or
dense sand and for all values of top layer thickness ratio (D,/B),the load displacement curves
are falling in between the load displacement curves of homogeneous loose and dense sand
conditions.

To analyze the effect of backfilling, simulations have been carried out for the two cases
where the excavation wall is vertical or inclined at an angle of 45"'with the vertical at the
edge of the anchor plate. The effect of backfilling is significant only at large displacements
close to ultimate loads. Backfilling of vertical excavations with dense sand in a loose medium
improves the behavior by about 20% while for a 45" excavation, the increase is as much as
50%. Similarly for backfilling with loose sand in dense medium the pullout load reduces.
1. Atkinson, J.H and Bransby, P.L. (1978). " The mechanics of soils, An introduction to
Critical State Soil Mechanics", McGraw-Hill Book Company, New York.
2. Balla, A. (1961). "The resistance to breaking out mushroom foundations for pylons", Proc.
Fifth Int. Conf. Soil Mech. And Found. Engg., Paris, France, Vol. I , pp.569-576.
3. Basudhar, P.K. and Singh, D.N. (1994). "A generalized procedure for predicting optimal
lower bound break-out factors of strip anchors", J1. of Geotech. Engg., Vo1.44, N0.2, pp.
..
307-3 18.
4. Bouazza, A. and Finlay, T.W. (1990). "Uplift capacity of plate anchors buried in two
layered sand", Geotechnique, Vo1.40, N0.2, pp. 293-297.
5. Chattopadyay, B.C., & Pise, P.J., (1986). "Breakout Resistance of Horizontal Anchors in
Sand", Soils and Found., Vo1.26, No.4, pp. 16-22.
6. Chen, W.F. and Mizuno, E. (1988). "Nonlinear Analysis in Geotechnical Engineering".
Elsevier science Ltd.,
7. Das, B.M. (1978). "Model tests for uplift capacity of foundations in clay", Soils and
Found., Japan, Vo1.18, No.2, pp. 17-24.
8. Das, B.M. (1980). "A procedure for estimation of uplift capacity of foundations in clay".
Soils and Found., Japan, Vo1.20, No. 1, pp. 77-82.
9. Das, B.M. (1987). "Advanced soil mechanics", McGraw-Hill Book Conlpany, New York.
10. Das, l3.M (1990). "Earth Anchors", Developments in Geotechnical Engineering, Elsevier.
New York.
11. Das, B.M. and Seeley, G.R.(1975). "Load-displacement relationship for vertical anchol
plates", J1. Of Geotech. Engg., Vol. 101, No. G 17, pp. 7 1 1-715.
12. Davie, J.R. and Sutherland, H.B. (1977). "Uplift resistance of shallow horizontal anchors"
J1. Geotech. Engg., ASCE, Vol. 10I, No.9, pp. 935-952.
13. Dickin, E.A. (1988). "Uplift behavior of horizontal anchor plates in sand". J1. Geotech
Engg., ASCE, Vo1114, No.11, pp. 1300-1317.
192
14. Downs, D.I. and Chieurzzi, R. (1966). "Transmission tower foundations", J1. Power Div.
ASCE, Vo1.88, No.2, pp.91-114.
15. Fast Lagrangian Analysis of Continua (1996), Version3.3, Reference Manual I,II and 111.
Itasca Consulting Group, Inc. Minnesota.
16. Frydn~an,S. and Shaham (1989). "Pullout capacity of slab anchors in sand", Canadian
Geotech. Jl., Vo1.26, pp.385-400.
17. Ghaly, A. and Hanna, A. (1994). "Ultimate pullout resistance of single vertical anchors",
Canadian Geotech. J1. Vo1.3 1, No.5, pp. 661-672.
18. Ghaly, A.M 91997). "Load-displacement prediction for horizontal loaded vertical plates",
J1. Of Geotech. And Geoenvironmental Engg., ASCE, Vol. 123, No. I, pp.74-76.
19. Hanna, A. and Gopal Ranjan (1992). "Pullout-Displacement of shallow vertical Anchor
Plates", Indian Geotechnical Journal, 22(1), 1992.
20. Ireland, H.O. (1963). "Uplift resistance of transmission tower footing", J1. Power Div.,
ASCE, Vo1.89, No.1, pp. 115-1 18.
21. Majdi, A. et al, (1993). "Force-displacement behavior of flexible plate anchors", J1. Of
Geotech. Engg., Vol. 119, No.3, pp. 590-597.
22. Manjunatha, K. (1998). "Uplift capacity of Horizontal strip and circular anchors in
homogeneous and layered soils", Ph.D., Thesis, Dept. of Civil Engg., Indian Institute of
Science, Bangalore.
23. Mariupol'skii, L.G. (1965). "The bearing capacity of anchor foundations; (English
translation)", Soil Mech. And Found. Engg. (in Russian), pp. 26-37.
24. Meyerhof, G.G. and Adarns, J.I. (1968). "The ultimate uplift capacity of foundations"
Canadian Geotechnical Jl., Vo1.5, No.4, pp. 225-244.
25. Mors, H. (1959). "The behavior of mast foundations subjected to tensile forces",
Bautechik, Vo1.36, No. 10, pp. 367-378.
26. Murray, E.J. and Geddes, J.D. (1987). "Uplift of anchor plates in sand", J1. Geotech.
Engg., ASCE, Vo1.113, No.3, pp. 202-2 15.
27. Ramesh Babu (1998). "Uplift capacity and behavior of shallow horizontal anchors in soil",
Ph.D. thesis, Dept. of Civil Engg., Indian Institute of Science, Bangalore.
28. Rowe, R.K. and Booker, J.R. (1980). "The elastic response of multiple underream
anchors", J1. Of Numerical and analytical methods in Geomechanics, Vo1.4, pp. 3 13-332.
29. Rowe, R.K. and Davis, E.H. (1982a). "The behavior of anchor plates in clay",
Geotechnique, London, Vo1.32, No. I, pp. 9-23.
30. Rowe, R.K. and Davis, E.H (1982b). "The behavior of anchor plates in sand",
Geotechnique, London, Vo1.32, No. 1, pp.25-4 I.
31. Saeedy, H.S. (1987). "Stability of circular vertical earth anchors:, Canadian Geotechnical
Journal, Vo1.24, No.3, pp.452-456.
32. Swami Saran, S. Gopal Ranjan and Nane, A.S. (1986). "Soil anchors and constitutive
laws", JI. Geotech. Engg., ASCE, Vol. 11, No. 12, pp. 1084-1100.
33. Stewart, W. 91985). "Uplift capacity of circular plate anchors in layered soil", Canadian
Geotechnical Journal, Vo1.22, pp.589-592.
34. Subbarao, K.S. and Jayant Kumar (1994). "Vertical uplift capacity of Horizontal anchors",
J1. Geotech. Engg., ASCE, Vol. 120, No.7, pp. 1134-1 147.
35. Suresh, P.P. (1992) "Experimental investigations on the uplift behavior of plate anchors
with geosynthetics", M.S Thesis, Dept. of Civil Engg, Indian Institute of Technology,
Madras.
36. Sutherland, H.B. (1988). "Uplift resistance of soils", Geotechnique, London, Vo1.38, No.4,
pp. 473-5 16.
37. Tagaya, K. et. al. (1983). "Application of finite element method to pullout resistance of
buried anchor", Soils and Foundations, Vo1.23, No.3, pp. 91-104.
38. Timoshenko, S.P. and Goodier J.N. (1970). "Theory of Elasticity", 3rdedition, McGraw-
Hill Book Company, New York.
39. Trautmann, C.H. and Kulhawy, F.H. (1988). "Uplift Load Displacement Behavior of
Spread Foundations", J1. Of Geotech. Engg., ASCE, Vol.114, No.2, pp. 168-184.
40. Veasaert, C.J and Clemence, S.P. (1977), " Dynamic pullout resistance of anchor", Proc.,
Int. Symposium of Soil- Structure Interaction, Roorkee, India, Vol. 1, pp. 389-397.
41. Vermeer, P.A. and Sutjiadi, W. (19g5). "The uplift resistance of shallow embedded
anchors", Proc. Eleventh Int. Conf. Soil Mech. And Found. Engg., San Francisco,
California, No.3, pp. 1635-1638.
42. Vesic, A.S. (1971). "Breakout resistance of objects embedded in ocean bottom", J1. Soil
Mech. And Found. Engg., ASCE, Vo1.97, No.9, pp. 1 183-1205.

You might also like