Proceedings of the 7th The 3rd National Conference of GEO-Institute - ASCE.

June 13-17, 1999, University of Illinois Urbana Champaign, Illinois, USA. 1999
Modication has been made using the
materials from my Ph.D Thesis June-2000 (Titled: ANALYSIS OF TUNNEL
SUPPORTS USING THE FINITE ELEMENT METHOD)
- Many thanks, Wilham G. Louhenapessy Thursday 29 June 2000

Tunnel Support Design with Finite Element Analysis

W.G. Louhenapessy 123
Abstract

One of the major shortcomings of the Q classication system is that

it does not take into account the orientation of rock joints with respect to
the exposed surface of the tunnel excavation. This paper proposes a rational
methodology by taking into consideration the in uence of the orientation of
joint sets. Using the nite element method and the multilaminate model for
jointed rock masses (Zienkiewicz & Pande, 1977), a set of design charts (rose
diagrams) useful to the practising engineers for circular tunnels are presented.
Introduction

Rock mass classication systems, such as the Q classication system
(Barton et al., 1974) and RMR system (Bieniawski, 1990), as well as others are
eorts to classify rock mass properties and the rock condition as a single number.
Advanced numerical methods have been frequently adopted but are not suitable
for routine analysis especially when unforeseen conditions are encountered onsite during construction.
Stability in rock tunnels

Collapse of the tunnel roof, sides or face takes place when the stresses
which are imposed on the rock mass due to excavation exceed its strength.
Thus, the factors which in uence the collapse of a tunnel are, strength of the
1 Department of Civil Engineering, University of Wales, Swansea, Singleton Park, Swansea, SA2 8PP,
2
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
3

Address: 45 GWYDR CRESCENT UPLANDS, Swansea, SA2 8PP, UK

Email: wilham.george@gmail.com or
wilham@signtegra.com

UK
WILHAM G. LOUHENAPESSY

Address now (2016) is at : SIGNTEGRA

7th Floor Plaza Sentral,
JL. JENDRAL SUDIRMAN Kav. 47
JAKARTA 12930
INDONESIA

jointed rock mass and factors aecting stresses imposed due to excavation.
The strength of jointed rock masses is aected by the strength of the
intact rock, the presence of joints, mechanical properties of rock joints and also
the presence of water. The properties or characteristic of rock joint are: dip
and orientation, spacing of parallel joint set, number of joint sets and surface
roughness. Some factors aecting stresses imposed by excavation on the rock
mass are insitu stress ratio and the depth of excavation.
Constitutive models and Fundamental equations

as.

The Mohr-Coulomb constitutive model for rock joints has been adopted

j = Cj + n tan

(1)

where  and Cj are friction angle and cohesion respectively for the joint, j
is the shear stress on the joint plane and n is the normal stress on the joint
plane. The failure criterion for intact rock used is the Mohr-Coulomb criterion
as follows,
i = ni tano + Co
(2)
where n is the normal stress on the failure plane and o and Co are material
constants for intact rock.
A general framework for constitutive models for jointed rock masses

Multilaminate framework for developing a constitutive model for jointed

rock masses has been discussed in detail in various publication (Zienkiewicz &
Pande, 1977; Pande & Williams, 1990). The elastic constants of a jointed rock
mass and rock joint can be determined from large scale in-situ experiments
or large scale triaxial tests. However, these experiments are expensive and
time consuming. An alternative approach is to derive the elasticity matrix of
the jointed rock mass from the constitutive properties of its constituents. The
philosophy here is to treat jointed rock mass as a composite material with intact
rock and rock joints as its constituents. It is to nd the normal stiness and
shear stiness of the rock joint. The elasticity matrix of rock mass, D rm
e , is

Drm
e

" n
X

i=1

Cjl

TT

Die

(3)

where "n" is number of joint sets, T is a transformation matrix, C is the compliance matrix contain joint stiness data and D ie is the conventional elasticity
matrix of the intact rock (Pande & Williams, 1990).
2

Methodology of computation of pressure on tunnel supports

The methodology proposed here for the computation of pressure on tunnel supports is based on a practical approach which is commonly used in many
areas of engineering design. Since the behaviour of a jointed rock mass is highly
non-linear, considerable computational eort is needed in solving the complex
rock-structure interaction problem. This is not practical and not economic from
a numerical modelling point of view. An alternative is to apply the so called
'stress path method' in which an estimate is made of the stress path experienced at a few typical points in the structure (Pande & Williams, 1990). The
stability of these points in the rock mass is considered based on the adopted
failure criterion and the support pressure is computed, if required, in such a
way that the rock mass is prevented from collapse.
For example, consider a point such as A on the roof of the tunnel (Figure 1) excavated at a certain depth in a jointed rock mass. Before excavation
this point experiences geostatic stresses. The stresses at this point after excavation which may be in stages can be computed assuming jointed rock mass
as an anisotropic multilaminate material having the elasticity matrix given by
equation 3. The deviation of stress from the geostatic condition is readily obtained and gives the stress path to which a rock mass will be subjected at point
A. This stress path at a point is imposed on rock mass and computation made
to judge if failure in any of the following modes is possible:
a. failure of intact rock: The strength parameters of the intact rock are examined and the failure function is checked.
b. failure of joint sets: The strength parameters are examined and failure in
shear or tension is checked. A numerical algorithm which determines the place
of failure onset, ie. intact or jointed rock is presented elsewhere (Louhenapessy,
1998).
If failure is observed in any of the modes, a pressure (p) normal to the
periphery of tunnel is computed which would prevent the failure of the rock
mass at that point. The above procedure is repeated at a number of points
on the periphery of the tunnel and simple engineering calculations are made to
determine the spacing of passive rock bolts of a given diameter.
Numerical Examples

In this section, analysis of a 12.8 m diameter circular tunnel excavated

at various depths in the jointed rock having one or two sets of joints is presented for the illustration of the methodology of computing support pressures.
Here, a two-dimensional idealisation is adopted, though extension to a threedimensional situation is straight forward. The notation for describing the fabric
of the the rock joints is shown in the inset. Figure 2(a) depicts the geometry
3

(a) geometry

(b) normal support pressure

(c) stress path

h
A

Rock Mass = intact rock +

rock joint

DE = stress path
D = geostatic stress
E = stress after
excavation

: Pressure applied normal to the tunnel periphery at point A to prevent failure of

joint rock. (a) geometry, (b) application of normal support pressure to prevent failure under
the stress path experienced by point A, (c) stress path experienced by point A in principal
stress space

Figure 1

of the problem. The material parameters assumed for illustration are shown in
Tables 1.
Support pressure has been computed for the rock mass having one set of
joints at various orientations as well as for two sets of joints. The results for one
set and two sets of joints are presented here. Some of the results for multiple
sets are presented elsewhere (Louhenapessy & Pande, 1997; Louhenapessy &
Pande, 1998). For the calculation of stress paths, elastic nite element analysis
is undertaken. Figure 2(b) shows the typical nite element (FE) mesh used
for the analysis which consists of 736 nodes and 224 eight-noded isoparametric
elements. In view of the approximate nature of the method of calculation, the
density of the mesh is not crucial and it is assumed that the mesh shown in
Figure 2(b) gives accurate stress paths for practical purposes.
Eight points have been chosen on the circumference of the tunnel for
studying the requirement of support pressure. Three cases of in-situ stress
corresponding to Ko = 0.333, 1.00 and 2.00 have been studied. It is noted that
cohesion for joints is adopted as zero and the friction angle is varied between
10o to 50o .
Rose diagrams of support pressure

Rose diagrams are useful tools for presenting results of parametric studies
of tunnel support pressure analysis. Here the support pressure required at a
point on the periphery of the tunnel is plotted as a radial line, the length of
4

(a)

(b)
ground surface

ground surface

one set of joint

x
Figure 2

160m

two sets of joint

160 m

: (a) Tunnel Geometry and fabric of rock joint (inset). (b) Finite Element mesh

which represents the support pressure. Such diagrams are shown in Figures 3
to 6. The support pressure has been normalised with reference to geostatic
stress at the centre of the tunnel before excavation ( h). Figures 4(a) and (b)
are for sd tunnel for the case when the joints are inclined at  = 0o and 45o
respectively. Figure 5 and 6 are for two sets of joint. For reason of symmetry,
it is sucient to look at only half side of rose diagram (Figures 4 to 6).
Normalised Support Pressure is calculated from the following equation,

Np = Ph

(4)

where Np is the normalised support pressure, P is the tunnel support pressure

obtained from nite element analysis, is the unit weight of rock and h is the
depth of tunnel. From these rose diagrams, support pressure can be obtained
based on the depth of tunnel, joint friction angle , in situ stress ratio, Ko , and
orientation of the joints, . The requirement for support pressure varies from
point to point on the periphery of the tunnel. Obviously, engineering judgement
has to be used and provision should be made for maximum required support
pressure in any section. It should be noted thats in most cases failure takes
place due to sliding on joints but there are also situations in which joint open
or intact rock fails, the latter case arising at very deep tunnel (vd).
Comparison with the Q classication system
Here we examine two specic cases of the support pressure requirements
and compare them with those obtained from the Q classication system. The

rst case is that of a shallow tunnel, sd (CASE I), with Ko being 0.333, 1.0
and 2.0, while the second case is that of very deep tunnel, vd, (CASE II). The
tunnels are in sandstone with one set of joint and intact rock having compresive
strength (c) of 100 MPa.
The rock mass quality, Q has been proposed based on 6 parameters
(Barton et al., 1974). The values of Q and Proof are dened by,
Jr Jw
Q = RQD
Jn : Ja : SRF

and

P roof = 23 Jn2 Jr 1Q

1
3

where RQD is the Rock Quality Designation, Jn is the joint set number, Jr is the
joint roughness number, Ja is the joint alteration number, Jw is the joint water
reduction factor, SRF is the stress reduction factor and Proof is permanent roof
support pressure.
The following data been assumed for comparison (Deere, 1968; Oberti
et al., 1986; Hoek et al., 1995; Natau et al., 1995; Palmstrom, 1995; Louhenapessy
& Pande, 1998):
 RQD=72 %, Jn = 2 (one joint set), Jr =1.5, Ja =1.0 and Jw =1.0 (dry),
 1 is the maximum principal stresses (from FE analysis).
 for sd tunnel: 1  1.09 MPa. c  91.5 (medium stress, SRF = 1.0)
1
 for vd tunnel: 1  26.67 MPa. c  3.75 (high stress, SRF = 7.0)
1
Based on the above parameters:
for CASE I, Q = 54.00 and Proof = 16.6 kPa and,
for CASE II, Q = 7.71 and Proof = 31.8 kPa.
Assuming, 25 mm (diameter) steel bolts, their spacing is obtained as
2950 mm c/c. and 2100 mm c/c respectively. Tables 2 and 3 show the comparison of spacing of rock bolts for cases I and II respectively computed on the
basis of the proposed methodology. It is obvious that whilst the Q system gives
a single spacing for each case, the spacing based on the theory of this paper
varies depending primarily on the orientation of joints. The Q system does not
always give a safe spacing.
Discussion

In general rose diagram give more extensive information i.e. the zone
and extent of area to be rock bolted is indicated. Moreover, they provide
a more rational and practical solution as compared to that proposed by any
classication system.
It may be noted that rock bolt normal to the periphery of the tunnel
may not be eective in certain situations i.e. the orientation of the joints may
6

be such that no amount of normal pressure would prevent joint failure or the
spacing of rock bolts may be too small for adoption in practice, in such a case a
combination of shotcrete lining and rock bolts of at inclination maybe required.
It is shown that, as expected, the use of Q classication system leads
to an over-conservative design in some cases whilst leading to unsafe design in
others.
Conclusion

Analsysis and design of tunnel support system is a complex problem

of rock sturcture analysis. In this paper a rational but practical method of
computing support pressure has been suggested. It is based on the 'stress path'
method of analysis. The stress path at a number of point on the periphery of
the tunnel is computed using an elastic nite element method.
A multilaminate theory is used to compute the support pressure which
would prevent the collapse of the rock mass. The methodology is explained by
a set of rose diagram. It is proposed that the engineers should develop similar
diagram for the tunnel based on actual laboratory / eld data. These design
charts can be readily read for any situation during construction.
The methods of excavation of support pressure based on a classication
system lack rationale and should be used with caution.
REFERENCES
Barton, N., Lien, R., & Lunde, J. 1974. Engineering classication of rock masses
for the design of tunnel support. Rock Mech., 6, 189{236.
Bieniawski, Z.T. 1990. Engineering rock mass classication. 1 edn. Chichester:
Wiley.
Deere, D.U. 1968. Geological consideration. Pages 1{20 of: Stagg, K.G., &
Zienkiewicz, O.C. (eds), Rock Mechanics in Engineering Practice. London:
John Wiley & Sons.
Hoek, E, Kaiser.P.K., & Bawden, W.F. 1995. Support of Underground Excavations in Hard Rock. 1 edn. Rotterdam: A.A.Balkema.
Louhenapessy, W, & Pande, G. N. 1997. A Rational Finite Element Analysis
(FEA) Based Procedure for The Analysis of Pressure on Tunnel Supports.
Internal report no: CR/964/97. Department of Civil Engineering: Univ. of
Wales, Swansea.
7

Louhenapessy, W.G. 1998. A Rational Finite Element Analysis Based Procedure for The Analysis of Pressure on Tunnel Supports. Pages 236-243 of:
Proc. Canadian Soc. Mech. Engineering FORUM 1998, vol. 2. Ryerson
Polytechnic University, Toronto.
Louhenapessy, W, & Pande, G. N. 1998. On a rational method of analysis and
design of tunnel supports based on the nite element technique. J. of Rock
Mech. & Tunnelling Tech., 4(2), 97{124.
Natau, O, Buhler, M, Keller, S, & Mutschler, T. 1995. Large scale triaxial test
in combination with a FEM analysis for the determination of the properties
of a transversal isotropic rock mass. Pages 635{643 of: Fuji, T (ed), 8th
International Congress on Rock Mechanics, vol. 2. ISRM, Tokyo.
Oberti, G, Bavestrello, F, Rossi, P, & Flamigni, F. 1986. Rock Mechanics Investigation, Design and Construction of the Ridracoli Dam. Rock Mechanics
& Rock Engineering, 19, 113{142.
Palmstrom, A. 1995. RMi - a system for characterizing rock mass strength for
use in rock engineering. J. of Rock Mech. & Tunnelling Tech., 1(2), 69{108.
Pande, G.N., & Williams, J.R. 1990. Numerical Methods in Rock Mechanics.
Chichester: John Willey. 327p.
Zienkiewicz, O.C., & Pande, G.N. 1977. Time dependent multi-laminate model
of rocks - a numerical study of deformation and failure of rock masses. Int.
J. Numerical and Analytical Meth. in Geomech., 1(1), 219{247.

Table 1

: Properties for tunnel in jointed rock

Intact E = 7 x 10 kPa
Joint Cn = 1 x 10 7 kPa 1
rock Rocktype: SANDSTONE rock Cs = 2 x 10 7 kPa 1
7

 = 0.3
Ci = 28870 kPa
i = 30 o
= 24.5 kN/ m3

Ko = 0.333 , 1.0 and 2.0

Cohesion = 0
 = 10o, 20o, 30o, 40o, 50o
rock joint spacing = 1 m

Tunnel type sd : depth, h= 80 m

Tunnel type vd : depth, h = 1600 m

Table 2

Orientation


0o

45o
90o

: Roof Bolt Spacing (Diameter = 25 mm)

Circular tunnel : CASE I (sd) ; depth , h = 80 m

insitu
Q
Proposed Method
stress system
(mm)
ratio (mm)
joint friction angle, 

Ko

0.333
1.0
2.0
0.333
1.0
2.0
0.333
1.0
2.0

10o

20o

30o

40o

2950 2150 2150 2150 2150

2950 NO NO NO NO
2950 NO NO NO NO
2950 970 960 930 850
2950 220 220 220 210
2950 X  X  X  X 
2950 NO NO NO NO
2950 NO NO NO NO
2950 NO NO NO NO

50o

2150
NO
NO
NO
NO
NO
NO
NO
NO

NO = no support required X= spacing too small, (not practical)

Table 3

: Roof Bolt Spacing (Diameter = 25 mm)

Circular tunnel : CASE II (vd) ; depth , h = 1600 m

insitu
Q
Proposed Method
Orien- stress system
(mm)
tation ratio (mm)
joint friction angle, 


0o

45o
90o

Ko

0.333
1.0
2.0
0.333
1.0
2.0
0.333
1.0
2.0

2150
2150
2150
2150
2150
2150
2150
2150
2150

10o

20o

30o

40o

50o

X 
X 
X 
X 
NO
NO
X 

X 
X 
X 
X 
NO
NO
X 

440 440 440 440

280 280 280 280

X 
X 
X 
X 
NO
NO
X 

NO = no support required

X  X 
X  X 
X  X 
X  X 
NO NO
NO NO
X  X 

X = spacing

440
280

too small, (not practical)

90
20
30
40

1.5

0.5

180
2

0.5

1.5

2
270

(a) Joint orientation, = 0o

1.5

90
20
30
40

0.5

180
1.5

0.5

1.5
270

(b) Joint orientation, = 45o

Figure 3

: Normalised Support Pressure for "sd" tunnels, for one set of joint, Ko = 0.333

10

(a) Joint orientation, = 0o

2.0

(b) = 45o

90

1.5

1.0
20
30
40

90

0.8

20
30
40

0.6
1
0.4
0.5

0.2
0

0.2

0.5

0.4
1.0
0.6
1.5

0.8

2
270

Figure 4

1.0
270

: Normalised Support Pressure for "sd" tunnels, for one set of joint, Ko = 1.0

11

(a) Joint orientations, 1 =0o and 2 =90o

4

90

2
20
30
40

0.5

20
30
40

0.5

1.5

4
270

2
270

: Normalized Support Pressure for "sd" tunnels, for two sets of joint, Ko = 1.0

(a) Joint orientations,

2 =150o
2

1 =30o

and

90

1.5

(b) 1 =60o and 2 =150o

2

20
30
40

0.5

0.5

0.5

0.5

1.5

1.5

2
270

90

1.5

Figure 6

90

1.5

Figure 5

(b) 1 =45o and 2 =135o

12

20
30
40

2
270

: Normalised Support Pressure for "sd" tunnels, for two sets of joint, Ko = 1.0