You are on page 1of 9




World Tunnel Congress 2013 Geneva
Underground – the way to the future!
G. Anagnostou & H. Ehrbar (eds)
© 2013 Taylor & Francis Group, London
ISBN 978-1-138-00094-0







A limit equilibrium method for the assessment of the tunnel face
stability taking into account seepage forces
P. Perazzelli(1), T. Leone(1), G. Anagnostou(1)
(1)
ETH Zurich, Switzerland

ABSTRACT: The present paper investigates the problem of the stability of the tunnel face under seepage flow
conditions based upon the so-called “method of slices”. This computational model improves the limit equilibrium
method of Anagnostou & Kovári (1996) by treating the equilibrium in the wedge consistently with the overlying
prism and does not need an a priori assumption concerning the distribution of the vertical stresses. Assuming a
simplified distribution of the hydraulic head in the ground ahead of the face, a closed-form solution is derived
which can be used for assessing face stability.

1 Introduction
Seepage flow conditions increase the risk of a collapse of the tunnel face because they are associated
with the occurrence of pore pressure gradients in the ground ahead of the face. The pressure gradient
acts as a body force, so-called “seepage force”, which is directed towards the face and is, therefore,
unfavorable with respect to its stability. The effect of seepage flow on the stability of the face can be
investigated by considering the limit equilibrium of a mechanism (Anagnostou & Kovári 1996, Broere
2001), by methods based on plasticity theorems (Lee et al. 2001, 2003, 2006) or by numerical stress
analyses (Ströhle and Vermeer 2009, Vermeer et al. 2002).
The present paper investigates tunnel face stability under drained conditions by considering a wedge
and prism mechanism (Fig. 1a, cf. Anagnostou and Kovári 1996) and analysing the equilibrium of the
wedge based upon the method of slices (Anagnostou 2012). In analogy to the silo theory, the method
of slices assumes proportionality between the horizontal stress ’y and the vertical stress ’z:
'y 'z , (1)

where is the coefficient of lateral stress which is assumed to be constant. In order to calculate the
distribution of the vertical stresses ’z inside the wedge, the equilibrium of an infinitesimally thin slice is
considered (Fig. 1b). The method of slices represents an improvement of the model of Anagnostou &
Kovári (1996) by eliminating the need for an a priori assumption of the distribution of the vertical stress
’z in the wedge and offering the possibility of analyzing cases with non-uniform face support,
heterogeneous ground consisting of horizontal layers and non-uniform distribution of the seepage
forces along the height of the face. In this paper we consider a homogenous soil obeying the Mohr-
Coulomb failure criterion, a uniform support pressure and an approximate distribution of the hydraulic
head, which is obtained by fitting the results of three dimensional seepage flow analyses (Fig. 2b).
The computational predictions of the method of slices agree very well with published results of
experimental tests in dry soil when the coefficient of lateral stress is taken equal to 1.0 (Anagnostou
2012). For this reason, the calculations according to the method of slices will be carried-out for = 1.0
in the present paper.
Section 2 describes the seepage-flow analysis and presents the simplified distribution of the hydraulic
head, which will be introduced in the limit equilibrium analysis. Section 3 outlines the limit equilibrium
analysis of the considered mechanism. Section 4 compares the computational results with those
obtained based upon Anagnostou and Kovári (1996).
(a) (b)

Figure 1. (a) Failure mechanism. (b) Forces acting upon an infinitesimal slice

(a) (b)

Figure 2. (a) Finite element mesh for the computation of the hydraulic head field. (b) Contour lines of the
hydraulic head for an example with B = H = 10 m, T/H = 5, h = 100 m and h0 > T+H (blue: head h close to
the head hF of the face, red: head h close to the initial head h0)
2 Seepage flow analysis
We determine numerically the three dimensional, steady state hydraulic head field around the tunnel
face assuming Darcy’s law with a uniform ground permeability. The permeability coefficient does not
influence the hydraulic head field. A no-flow boundary condition and a constant piezometric head hF
are prescribed to the tunnel wall (impervious lining) and to the tunnel face, respectively. At the far-field
boundary, the piezometric head is taken equal to the water table elevation h0. This condition applies
also to the water table (no draw-down, i.e. sufficient groundwater recharge from the surface).
Figure 2a shows the central part of the finite element mesh adopted for the calculations, which were
performed by the finite element program COMSOL®. A square tunnel cross-section is considered for
simplicity (analogue to the limit equilibrium model). The computational domain consists of one half of
the system due to the vertical symmetry plane.
Figure 3 shows the normalized distributions of the hydraulic head ahead of the face and above the
tunnel along two characteristic lines (the tunnel axis and the vertical axis z) for a square tunnel (H=B)
with overburden T=5H assuming that the water table is located above the soil surface (h0 > T+H). The
normalized distribution of the hydraulic head depends in general on the normalized overburden T/H,
but remains practically constant for T/H values higher than 10 (Zingg and Anagostou 2012). In the
comparative calculations of the following sections, T/H is taken equal to 5.
The numerically computed hydraulic head ahead of the face (i.e., inside the wedge) and above the
tunnel (i.e., inside the prism) is given approximately by the equations
x
-b
h x, y , z h x hF 1- e H
h (2)

and
x z
-b a 1-
H H
h x, y , z h x, z hF 1- e h, (3)

respectively, where h is the difference between the far field hydraulic head h0 and the hydraulic head
at the face hF. The constants a and b are obtained by curve fitting the analytical solution to the
numerical results and depend in general on the normalized overburden T/H. For the comparative
calculations in the following Sections, the constants a and b are taken to 2.85 and 1.64, respectively,
which apply to T/H = 5 (Fig. 3).

z/H [-] 4 (h-hF)/(h0-hF) [-]

3.5 numerical results


approximate distribution
3

2.5

1.5

Tunnel 0.5

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(h-hF)/(h0-hF) [-] x/H [-]

Figure 3. Distribution of the normalized hydraulic head (h-hF)/(h0-hF) along two characteristic lines (B = H,
T/H = 5, h0 > T+H)
3 Computational model

3.1 Outline
The mechanism under consideration (Fig. 1a) fails, if the load exerted by the prism upon the wedge
exceeds the force which can be sustained by the wedge at its upper boundary. At the limit equilibrium
the prism load is equal to the bearing capacity of the wedge. The prism load is calculated based on
the silo theory (Section 3.2), while the bearing capacity of the wedge is calculated by considering the
equilibrium of an infinitesimal slice (Section 3.3). Both, the load of the prism and the bearing capacity
of wedge, depend on the inclination of the slip plane. The critical value of the angle (Fig. 1a), i.e. the
value that maximizes support requirements, will be determined iteratively.

3.2 The prism loading


Assuming that the water level is above the soil surface and that the ground is homogeneous and
obeys the Mohr-Coulomb failure condition with cohesion c and angle of internal friction , the effective
vertical load of the prism reads as follows:
T
R '- c - tan
V 'silo max 0, 1- e R
w h BH tan , (4)
tan

where the coefficient R is the ratio of the area to the circumference of a horizontal cross-section of the
prism, while the coefficient depends on the hydraulic head distribution (Anagnostou and Kovári
1996). The coefficient can be determined either numerically on the basis of the numerically
computed hydraulic head distribution (Anagnostou and Kovári 1996) or analytically assuming the
approximate distribution of Eq. (3):
a tan
a T
H R b tan
1 e 1 e . (5)
a tan
Hb tan
H R

3.3 The bearing capacity of the wedge


In order to determine the bearing capacity of the wedge we consider the equilibrium of an infinitesimal
slice (Fig. 1b). The following forces act upon the slice: its submerged weight dG’; the “supporting”
force V’(z) resulting from the effective normal stress exerted by the underlying ground; the “loading”
force V’(z) + dV’ resulting from the effective normal stress exerted by the overlying ground; the
effective normal force dN’ resulting from the effective normal stress at the inclined slip surface; the
shear force dT on the inclined slip surface; the shear force dTs on the two vertical slip surfaces; the
resultant seepage force dFx and dFz (dFy is equal to zero due to the symmetry) and the supporting
force dS’ resulting from the effective support pressure s’ at the tunnel face. The equilibrium equations
parallel and perpendicular to the sliding direction are:
dTs dT dS ' dFx sin (dV ' dG ' dFz )cos (6)

and
dN ' ( dV ' dG ' dFz ) sin dS '- dFx cos , (7)
where
z tan B/2
h x, y , z
dFx w dx dy dz (8)
0 B/2
x

and
z tan B/2
h x, y , z
dFz w dx dy dz . (9)
0 B /2
z

Expressions for the other forces appearing in Eqs. (6) and (7) can be found in Anagnostou (2012). The
equilibrium conditions lead to the following differential equation for the effective vertical force V’(z):
dV ' z dFz dFx
B V' M +P B BPs . (10)
dz B dz dz
The coefficients , M, P and Ps were introduced by Anagnostou (2012). The effective vertical force
V’(z) can be determined by solving the differential equation (10) for the boundary condition V’(0) = 0.
For the approximate distribution of hydraulic head (Eq. 2), the effective vertical force is equal to
V '( z ) Cs ( ) B 2 s ' C c ( ) B 2 c - C ( ) B 3 '- C h B2 w h, (11)

where is the normalized z coordinate ( =z/H), the coefficients Cs( ), Cc( ) and C ( ) can be found in
Anagnostou (2012), while the additional coefficient

H
1 1 H 1 H b tan 1
C h( ) Ps eB e (12)
H B H B
b tan b tan
B B
takes into account the seepage forces. The bearing capacity of the wedge is identical to the effective
vertical force at z = H ( =1).
From the equilibrium equation (7), we can obtain the following expression for the distribution of the
effective normal stress ’n on the sliding surface:
dN ' z dV '( z ) 1 b tan
'n z ' H tan sin s' w h 1 e cos . (13)
B dz dz B

One can readily verify, that ’n becomes negative (tension), if the hydraulic head difference h is
sufficiently high. This result is expected because the seepage forces are directed towards the face and
their magnitude increases with h. Note that even if the ground exhibits a tensile strength, tensile
stresses higher than c/tan (in combination with the Mohr-Coulomb failure criterion) would lead to
negative shear strength values, which does not make sense. In this case, which may happen at high
h values, the solution for the effective vertical force V’(z) is clearly wrong.
The minimum effective support pressure s’1, which ensures that the tensile effective stress does not
exceed c/tan at any point of the sliding surface can be determined from Eq. (13):
s '1 P1 w h P2 'H P3 c , (14)

where
b tan
1 e C h (1)
P1 , (15)
Cs (1) s cot Cs (1)
1
s cot

C (1)B
P2 , (16)
s cot Cs (1) H

c H
c Cc (1) cot
P3 B s
. (17)
s cot Cs (1) (Ps sin cos ) s cot Cs (1) tan

The coefficients Ps, Pc and Mc can be found in Anagnostou (2012).


For soft ground without tensile strength, the effective normal stress ’n must be higher than 0, which is
a more strict condition than the one mentioned above. The minimum effective support pressure, which
is needed in order to satisfy this condition, is given by the following equation:
s '2 P1 w h P2 'H P4 c , (18)

where

c H
c Cc (1)
P4 B . (19)
s cot Cs (1)
Eq. (11), which gives the effective vertical force inside the wedge (and for z=H corresponds to the
bearing capacity), is only correct for effective support pressures s’ higher than s’1 (or s’2 in the case of
a no-tension material).

3.4 The effective support pressure


In order for the face to remain stable, the bearing capacity of the wedge has to be higher than the
vertical load exerted by the prism. To ensure this, the effective support pressure has to be higher than
a minimum value s’3 which can be determined from the limit equilibrium condition V’(H)=V’silo:

s '3 F1 w h F2 ' H F3 c , (20)

where
1 H tan
F1 C h (1) , (21)
Cs (1) B

1 C (1) B tan R (1 e tan T / R


)
F2 , (22)
Cs (1) H B tan

tan T / R
1 tan H (1 e )
F3 Cc (1) . (23)
Cs (1) B tan

The necessary effective support pressure must ensure all criteria mentioned above, i.e.:
s ' max (s1 or s2 ), s3 . (24)

4 Comparative calculations
Figure 4 shows the effective support pressure as a function of the hydraulic head difference h for the
example of a tunnel crossing cohesionless soil with a friction angle of 15°, 25° or 35°. Figure 5 shows
the effect of cohesion on the support pressure for the same example and for a fixed value of the
hydraulic head difference ( h = 100 m).
The solid lines are calculated according to the method of slices. The dashed lines are based on the
method of Anagnostou and Kovári (1996) for a uniform coefficient of lateral stress = 1 (as in the
method of slices). The dotted lines consider the conservative assumption of = 0.8 for the prism and
w = 0.4 for the wedge, which was proposed by Anagnostou and Kovári (1996). Consequently, the
difference between dotted and dashed lines is due to the different values of , while the reason for the
difference between the solid and the dashed lines is that the two computational models (equilibrium of
the entire wedge vs. equilibrium of slices) lead to different horizontal stresses at the lateral sliding
surfaces of the wedge. As the horizontal stresses generate frictional resistance, their effect increases
with the friction angle. The differences between the two models are, therefore, smaller for low to
moderate friction angles and bigger for the case of = 35°.
Method of Slices ( =1)
Anagnostou and Kovári 1996 ( = 1)
Anagnostou and Kovári 1996 ( = 0.8, w= 0.4)

800

700

600
s' [kPa]

500

400

300

200

100

0
0 20 40 60 80 100 120 140
h [m]

Figure 4. Effective support pressure s’ as a function of the hydraulic head difference h for a
cohesionless soil (B = H = 10 m, T = 50 m, h0 > T + H, ’ = 12 kN/m3)

800

700

600

500
s' [kPa]

400

300

200

100

0
0 5 10 15 20 25 30 35 40 45 50
c [kPa]
Figure 5. Effective support pressure s’ as a function of the cohesion c for a hydraulic head difference of
h = 100 m (B = H = 10 m, T = 50 m, h0 > T + H, ’ = 12 kN/m3)
5 Closing remarks
According to the results of comparative calculations, the method of slices leads to lower effective
support pressures than the method of Anagnostou and Kovári (1996), but the differences in the model
predictions are small (< 20%) for low to moderate friction angles. These conclusions are true for
tunnels crossing soft ground at shallow or moderate depths. In the case of an open tunnel face at
great depth below the ground water table, very high hydraulic head gradients develop in the ground
ahead of the face. In this case equilibrium is only possible if the face is supported or if the ground
exhibits a sufficiently high tensile strength. If the cohesion of the ground is high (which may be true for
hard soils or weak rocks) and its tensile strength is negligible, the necessary face support pressure
may exceed by far the pressure, which would be needed just in order that the wedge is able to sustain
the load of the overlying prism. In this case, tensile failure rather than sliding becomes the critical
mode for the determination of the support pressure, which means that the standard limit equilibrium
model may underestimate the necessary support pressure (cf. Anagnostou and Kovári 1994).

6 References
Anagnostou, G., Kovari, K. 1994. Die Stabilität der Ortsbrust bei Erddruckschilden. Mitteilungen der Schweiz.
Gesellschaft für Boden- und Felsmechanik, 129, 27-34.
Anagnostou, G., Kovári, K. 1996. Face stability conditions with Earth Pressure Balanced shields. Tunnelling and
Underground Space Technology, 11 (2), 165-173.
Anagnostou G. 2012. The contribution of horizontal arching to tunnel face stability. Geotechnik, Vol. 35, No. 1, 34-
44
Broere, W. 2001. Tunnel Face Stability & New CPT Applications. PhD thesis, Delft University of Technology,
Delft.
COMSOL®. Subsurface Flow Module (Comsol Multiphysics 4.2). COMSOL 1998-2011.
Lee, I.-M., Nam, S.-W. 2001. The study of seepage forces acting on the tunnel lining and tunnel face in shallow
tunnels. Tunnelling and Underground Space Technology, No.16, pp. 31-40.
Lee, I.-M., Nam, S.-W., Ahn, J.-H. 2003. Effect of seepage forces on tunnel face stability. Canadian Geotechnical
Journal, No. 40, pp. 342-350.
Lee, I.-M., Nam, S.-W. 2006. Seepage force considerations in tunnelling. International Symposium on
Underground Excavation and Tunnelling, Bangkok, Thailand.
Ströhle, P., Vermeer, P. A. 2009. Die Stabilität der Ortsbrust bei strömendem Grundwasser. Kolloquium Bauen in
Boden und Fels, TAE, Ostfildern 2010, pp. 117-123 (in German).
Vermeer, P. A., Ruse, N., Marcher, Th. 2002. Tunnel Heading Stability in Drained Ground. Felsbau, 20, No. 6, 8 -
18.
Zingg, S., Anagnostou, G. 2012. The effects of advance drainage on face stability in homogeneous ground. ITA-
AITES World Tunnel Congress, WTC 2012, Bangkok

You might also like