You are on page 1of 8

The 12

th
International Conference of
International Association for Computer Methods and Advances in Geomechanics (IACMAG)
1-6 October, 2008
Goa, India

Undrained Stability of Dual Square Tunnels
D. W. Wilson, A. J. Abbo, S. W. Sloan, A. V. Lyamin
Centre for Geotechnical and Materials Modeling
School of Engineering, University of Newcastle, Australia
Keywords: Tunnel stability, limit analysis, plasticity
ABSTRACT: The design of tunnels for roads and railways often utilise separate tunnels to carry traffic in opposite
directions. In this paper numerical limit analysis and semi-analytical rigid block techniques are used to investigate
the effect of inter-shaft distance on the stability of two square tunnels constructed side by side. The tunnels are
modelled under conditions of plain strain assuming that the tunnels are infinitely long. Bounds on the stability of
the tunnels are obtained using finite element limit analysis, the numerical formulation of which is based upon the
upper and lower bounds theorems of classical plasticity. Solutions are obtained using advanced conic
programming schemes to solve the resulting optimisation problems. Upper and lower bound estimates on the
stability of the tunnels are obtained for a range of geometries. These bounds, which bracket the true collapse
load from above and below, are found to differ by at most 5%. Results from this study are summarised in stability
charts for use by practising engineers.
1 Introduction
This paper investigates the undrained stability of two parallel square tunnels running side by side. The tunnels,
which are separated by an inter-shaft distance S, are modelled under plane strain conditions and are shown in
Figure 1. The soil surrounding the tunnels is modelled as a uniform Tresca material with an undrained cohesion
(c
u
) and a unit weight (). The soil unit weight and the cohesion are usually known and are necessary to
determine the stability of the tunnels. When the soil has uniform properties, it is useful to describe the stability of
the tunnels in terms of the dimensionless parameter (
s

t
)/c
u
which is a function of H/B, B/c
u
and S/B. Using
this set of dimensionless parameters allows a compact set of stability charts to be obtained which are useful for
design purposes.


Figure 1: Plane strain dual square tunnel setup in a uniform Tresca material.
This paper describes the application of finite element limit analysis to calculate upper and lower bounds on the
stability (
s

t
)/c
u
for the tunnel stability problem shown in Figure 1. These techniques utilise linear finite

t

t

s



S
Plane Strain Undrained shear strength =c
u

Unit Weight =
4284

elements to formulate an optimisation problem that is solved using second order conic programming. Safe
estimates for the true solution of (
s

t
)/c
u
were obtained using the lower bound method based on the principle
that any set of loads supported by a statically admissible stress field will underestimate the true collapse load.
The upper bound method, which uses the fact that any kinematically admissible velocity field will provide an
unsafe solution on the true collapse load, provides an estimate of the collapse load that is above the true solution.
Using both methods in tandem enables the true collapse load to be bracketed from above and below, and as the
accuracy of each of the bounds is increased, the true solution is known with more certainty. A semi-analytical
method using a series of rigid block mechanisms, as described by Chen (1975), was also used to find upper
bounds on the stability.

Before 1990, most published research on tunnel stability focused on circular tunnels. Analytical, numerical and
experimental investigations on the stability of circular tunnels has been carried out at Cambridge since the 1970s.
Papers published from this work include those of Cairncross (1973), Atkinson and Cairncross (1973) and
Seneviratne (1979). Later, theoretical solutions for circular tunnel problems were given by Muhlhaus (1985) as
well as Leca and Dormieux (1990). The undrained stability of circular tunnels in heterogeneous soils, whose
undrained shear strength increases linearly with depth, were investigated by Sloan and Assadi (1992), while more
recently Lyamin and Sloan (2000) studied the collapse of circular tunnels in cohesive-frictional soils.

Previous studies on the collapse of square tunnels have concentrated chiefly on the undrained stability of single
tunnels. Theoretical investigations in this direction have been undertaken at the University of Newcastle. Assadi
and Sloan (1991) produced a series of results using the finite element limit analysis methods developed by Sloan
(1988, 1989) for both active and passive failure of single square tunnels in uniform undrained clay. Attempts
were also made at this time to classify the modes of failure. This work was extended to include heterogeneous
undrained clay where the shear strength increases linearly with depth by Sloan and Assadi (1991). More
recently, Lyamin, Jack and Sloan (2001) considered the stability of square tunnels in a cohesive-frictional soil.

All the theoretical studies mentioned so far have only investigated the stability of single tunnels. It would appear
that there is very little information available on the affects of interaction between tunnels located close to one
another. This problem is studied here using methods recently developed by Lyamin and Sloan (2002a,b) and
Krabbenhoft et al. (2005, 2007) and considers two square tunnels located side by side, at the same depth, in an
undrained clay. It should be noted that the techniques applied in this paper do not consider buckling of the wall or
pillar between the tunnels which would be a possible mode of failure for tunnels with a small inter-shaft distance.
2 Finite Element Limit Analysis
The upper and lower bound theorems are useful tools when applied to geotechnical stability problems. Finite
element formulations of these theorems allow them to be applied to a wide variety of complex cases. The upper
bound theorem is based on the notion that the imposed loads cannot be carried by the soil mass if for any
kinematically admissible failure mechanism (velocity field) the rate of work done by the external forces exceeds
the internal rate of dissipation (Chen, 1975). A kinematically admissible velocity field is one which satisfies the
flow rule and the velocity boundary conditions.

The lower bound theorem is based upon the notion that if an equilibrium state of stress can be found that satisfies
the stress boundary conditions and the yield criterion, then the imposed loads can be safely carried by a soil
mass. If such a state of stress can be found it gives a lower bound (or safe) solution and underestimates the true
collapse load. The stress field that meets all of the above criteria is known as a statically admissible stress field.

Limit analysis is most useful when both the upper and lower bounds are calculated as the true solution is then
bracketed from above and below. This allows the error in the solution to be computed as simply the difference
between the bounds. Even though lower bounds are more desirable from a practical point of view, it is generally
easier to find upper bounds which is why they are seen more widely in the literature.

The finite element limit analysis formulation used in this investigation is based upon the methods originally
developed by Sloan (1988,1989). These techniques have evolved significantly over the past two decades and the
current implementation follows the techniques described in Lyamin and Sloan (2002a,b) and Krabbenhoft et al
(2005,2007).

In this study finite element limit analysis is used to compute upper and lower bounds on the stability of the dual
tunnels for a range of different values of H/B, B/c
u
and S/B. The primary interest in these analyses was to
investigate the effect of the inter-shaft distance (S/B) between the tunnels. Although a range of tunnel depths
have been considered in this research, generally only the stability of a relatively shallow tunnel with H/B = 1 and
that of a deeper tunnel with H/B = 5 are presented. The finite element mesh used in the lower bound analysis of
the tunnels with H/B = 1 and S/B = 1 is shown in Figure 2, together with the stress boundary conditions (on the
normal stress,
n
, and the shear stress, ). This mesh, which is typical of those used for other geometries,
contains a minimum of 100,380 elements. The meshes for deeper tunnels contain more elements reflecting the
larger volume of soil. Note that the meshes have statically admissible stress discontinuities along all shared
4285

element edges. The lower bound analysis is performed by solving an optimization problem to find a statically
admissible stress field which maximizes the quantity (
s

t
)/c
u
for various values of H/B, B/c
u
and S/B.

The finite element mesh shown in Figure 2 is also used for the upper bound analyses, with the velocity boundary
conditions on the horizontal velocity, u, and vertical velocity, v, as indicated. The upper bound analysis is
performed by solving an optimization problem to find a kinematically admissible velocity field which minimizes the
quantity (
s

t
)/c
u
.

Figure 2: Finite element mesh for H/B = 1 showing boundary conditions for lower and upper bound analysis.
3 Rigid Block Analysis
Upper bound estimates of the stability of the dual tunnels were also determined using rigid block analysis. The
various rigid block mechanisms considered are shown in Figure 3. In addition to these, various single tunnel
mechanisms (see Assadi and Sloan, 1991) were also considered which are relevant for tunnel geometries with
higher S/B ratios. The minimum upper bound for each of the mechanisms was obtained by optimising the
geometry of the mechanism using the Hook-Jeeves algorithm.
u = v = 0
u

=

v

=

0


=

u

=

0

n
= = 0
= 0
4286



Figure 3: Rigid block mechanisms used to find upper bound solutions.
4 Results and Discussion
For the plane strain tunnel geometry shown in Figure 1, collapse is driven by a combination of the action of
gravity () and the surcharge pressure (
s
) on the ground surface while resistance is provided by the internal
tunnel pressure (
t
). Since the stability is defined by the dimensionless parameter (
s

t
)/c
u
, it is possible to
simplify the problem by setting the surcharge pressure to zero. Further simplification is achieved by setting the
dimensions of the tunnels (B) and the undrained shear strength of the soil (c
u
) to unity. Using the dimensionless
parameters described above (H/B, B/c
u
and S/B), this reduces the variables in the parametric study to the tunnel
depth (H), the soil unit weight () and the spacing between the tunnels (S).

Results obtained for the stability of the tunnels with H/B=1 and H/B = 5 are summarised in Figures 4 and 5 as
stability plots of (
s

t
)/c
u
versus S/B. These charts show that the finite element upper and lower bounds lie, for
the most part, within a few percent of each other. The rigid block upper bounds also give a close approximation
to the finite element solutions and provide a reasonably good approximation to the stability curves. For the
majority of the cases the rigid block mechanism shown in Figure 3(b), which is able to model pillar, roof and outer
wall collapse, produced the best upper bound. This mechanism also reduced to a trapdoor failure mechanism for
the shallower tunnels.

The stability of the tunnels would be expected to increase as they move further apart. However, the results show
that for small inter-shaft distances, the stability of the tunnels actually decreases as the spacing between the
tunnels is increased. Indeed, for higher values of B/c
u
and H/B, the curves show a distinct increase in stability as
the tunnel spacing S/B approaches 0.25. Possible causes for this include the fact that the model used does not
consider an asymmetric mode of failure in the pillar. Another possible cause is that with very narrow pillars and
higher unit weights, the increase in the soil weight is not completely countered by the small increase in resistance
to failure as the pillar gets wider. It is also important to appreciate the changing geometry of the problem. For
very small inter-shaft distances for which the strength of the pillar can be ignored, the stability can be
approximated as that of a single tunnel of width 2B+S. Clearly the stability of such a tunnel decreases as S
increases.
(a) (b)

(e) (d)

(c)
4287

-2.50
-2.00
-1.50
-1.00
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00
S/B
S
t
a
b
i
l
i
t
y

N
u
m
b
e
r

Figure 4: Stability bounds for H/B = 1.
-12.00
-10.00
-8.00
-6.00
-4.00
-2.00
0.00
2.00
4.00
6.00
8.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
S/B
S
t
a
b
i
l
i
t
y

N
u
m
b
e
r

Figure 5: Stability bounds for H/B = 5.

A typical failure mode for very shallow tunnels is shown in Figure 6 for the case with H/B = 1, B/c
u
= 1 and S/B =
1. It can be seen that the failure is localised to the roof and pillar and that the failure mechanism at the ground
surface is approximately 2B + S wide. This continues to be the case for shallow tunnels as the S/B ratio is
increased until the point is reached where the two tunnels are no longer interacting. After this distance reached,
Upper Bound (RB)
Upper Bound (FEM)
Lower Bound (FEM)
B/c
u
=3
B/c
u
=2
B/c
u
=1
B/c
u
=0

Upper Bound (RB)
Upper Bound (FEM)
Lower Bound (FEM)
B/c
u
=3
B/c
u
=2
B/c
u
=1
B/c
u
=0
4288

the failure is consistent with that of two completely separate shallow tunnels.

Figure 6: Upper bound collapse mechanism and power dissipation (H/B = 1, B/c
u
= 1, S/B = 1).
The failure mode associated with deeper tunnels is demonstrated Figure 7 for the case where H/B = 4, B/c
u
= 2
and S/B = 1. Here the failure mechanism changes from a simple pillar and roof collapse to a much more complex
mechanism which is much wider at the ground surface due to collapse of the outside walls of the tunnels.

Figure 7: Upper bound collapse mechanism and power dissipation (H/B = 4, B/c
u
= 2, S/B = 1).
Figure 8 shows the failure mechanism for the case where H/B = 5, B/c
u
= 3 and S/B = 5. The failure mode now
extends under the floor of the tunnel and is even wider at the ground surface. This is typical for deeper tunnels
with higher B/c
u
values.
4289


Figure 8: Upper bound collapse mechanism and power dissipation (H/B = 5, B/c
u
= 3, S/B = 5).
While the tunnels are close enough to interact, the majority of the failures for the range of H/B, B/c
u
and S/B
ratios that were considered can be broadly categorized into the 3 mechanisms shown above. For the situations
in which S/B was large, the two tunnels act as single tunnels with no interaction and gave solutions approximately
the same as those in Assadi and Sloan (1991).

It is important to note the meaning of the stability numbers shown in the stability charts (Figures 4 and 5). Since
the stability number is defined to be (
s

t
)/c
u
this means that, for the special case where
s
is zero, it actually
refers to the required pressure on the inside of the tunnels to prevent collapse. A negative stability number
implies that a compressive normal stress must be applied to the tunnel faces to maintain stability, while a positive
stability number means that no tunnel support is required to prevent collapse (in fact the tunnel faces can support
a tensile normal stress). For the general case where
s
is compressive and nonzero, a negative stability number
implies that
t
must exceed
s
for stability to be maintained.
5 Conclusions
The stability of dual square tunnels has been investigated using both finite element limit analysis and rigid block
upper bound methods. Stability charts have been generated for a variety of tunnel depths, material properties
and inter-shaft distances. The stability charts show that the stability of the tunnels does not simply increase with
the spacing between the tunnels. For tunnels which are very close together, it was observed that the stability
may in fact decrease as the inter-shaft distance is increased. Several possible causes of this behaviour have
been discussed. The stability charts show that the numerical upper and lower bounds are generally within a few
percent of one another, with the true solution lying between the two bounds.
6 References

Assadi, A. & Sloan, S. W. 1991. Undrained stability of a shallow square tunnel. Journal of the Geotechnical Division, ASCE
117, 1152-1173.

Atkinson, J. M. & Cairncross, A. M. 1973. Collapse of a shallow tunnel in a Mohr-Coulomb material. In A. C. Palmer (ed.), Role
of Plasticity in Soil Mechanics: 202-206. Proceedings of Symposium, Cambridge, September: 13-15.

Cairncross. A. M. 1973. Deformation Around Model Tunnels in Stiff Clay. PhD Thesis, University of Cambridge.

Chen, W.-F. 1975. Limit Analysis and Soil Plasticity. Amsterdam: Elsevier Scientific Publishing Company.

Krabbenhoft, K, Lyamin, A.V., Hjiaj, M and Sloan, S.W. 2005. A new discontinuous upper bound limit analysis formulation,
International Journal for Numerical Methods in Engineering, 63, 1069-1088.

Krabbenhoft, K., Lyamin, A. V., & Sloan, S. W. 2007. Formulation and solution of some plasticity problems as conic programs.
International Journal of Solids and Structures, 44, 1533-1549.

Leca, E. & Dormieux, L. 1990. Upper and lower bound solutions for the face stability of shallow circular tunnels in frictional
material. Gotechnique 40(4), 581-606.
4290


Lyamin, A.V, Jack D and Sloan SW. 2001.. Collapse analysis of square tunnels in cohesive-frictional soils. First Asian-Pacific
Congress on Computational Mechanics (APCOM 01), Sydney, Australia, 405-414.

Lyamin, A. V. and Sloan, S. W. 2000. Stability of a plane strain circular tunnel in a cohesive-frictional soil. Proceedings J. R.
Booker Memorial Symposium, Sydney, Australia, 139-153.

Lyamin, A.V. and Sloan, S.W. 2002a. Lower bound limit analysis using nonlinear programming. International Journal for
Numerical Methods in Engineering, 55, 573-611.

Lyamin, A.V. and Sloan, S.W. 2002b. Upper bound limit analysis using linear finite elements and nonlinear programming.
International Journal for Numerical and Analytical Methods in Geomechanics, 26(2), 181-216.

Muhlhaus, H. B. 1985. Lower bound solutions for circular tunnels in two and three dimensions. Rock Mechanics and Rock
Engineering 18, 37-52.

Seneviratne, H. N. 1979. Deformations and Pore-Pressures Around Model Tunnels in Soft Clay. PhD Thesis, University of
Cambridge.

Sloan, S. W. 1988. Lower bound limit analysis using finite elements and linear programming. International Journal for
Numerical and Analytical Methods in Geomechanics 12: 61-67.

Sloan, S. W. 1989. Upper bound limit analysis using finite elements and linear programming. International Journal for
Numerical and Analytical Methods in Geomechanics 13, 263-282.

Sloan, S. W. & Assadi, A. 1991. Undrained stability of a square tunnel in a soil whose strength increases linearly with depth.
Computers and Geotechnics 12, 321-346.

Sloan, S. W. and Assadi, A. 1992. The stability of tunnels in soft ground. Proceedings of Peter Wroth Memorial Symposium on
Predictive Soil Mechanics, Oxford, 644-663.

4291

You might also like