Computers and Geotechnics 110 (2019) 82–93

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Three-dimensional face stability of shallow-buried tunnels with tensile T

strength cut-off
T.Z. Li, X.L. Yang
⁎

School of Civil Engineering, Central South University, Hunan 410075, China

ARTICLE INFO ABSTRACT

Keywords: In terms of a very small or even zero tensile strength of soils, the tensile strength cut-off is introduced to modify
Tunnel the classical Mohr-Coulomb failure criterion which allows a nonlinear strength envelope of soils in tensile re-
Face stability gime. Based on the modification, an improved three-dimensional (3D) rotational failure mechanism is proposed
3D rotational failure mechanism to assess the face stability of a soil tunnel in the framework of limit analysis. The shallow-buried case is in-
Tensile strength cut-off
vestigated to account for the influence of tensile strength cut-off on face stability of tunnels with different buried-
Pore water pressure
depths. The influences of the unit weight of soils γ, the ratio ξ, the cohesion c and the internal friction angle φ on
the critical support pressure are researched using the proposed approach. In order to assess the face stability of
tunnels under water table, the coefficient ru is employed to describe the distribution of pore water pressure. It is
found that the proposed approach can give a more critical support pressure against tunnel face failure, and the
presence of pore water pressure makes the influence of tensile strength cut-off more prominent.

1. Introduction of limit analysis, the failure mechanisms of tunnel face that are often
employed in literature can be divided into two categories, namely the
With the increase of population density in metropolitan areas, the translational failure mechanism and rotational failure mechanism.
ground space is gradually being developed and exhausted for human Mollon et al. [8] presented a 3D translational failure mechanism which
living and entertainment. The traffic congestion that comes with it has is composed of several rigid cones with circle cross-sections. It is an
become one of the major problems that restrict the improvement of improvement to the two-block collapse mechanism proposed by Leca
urban life experience. To overcome this issue, engineers turn to the and Dormieux [9]. It was assumed that the internal energy dissipation
development of underground space in an attempt to make cities run was generated not only along the outer contour of the whole failure
more smoothly. Under this circumstance, the tunnel face stability has block, but also from the relative sliding between blocks. This kind of
attracted much attention of researchers and engineering designers as a computational model divides the failure block ahead of tunnel face into
practical problem in underground excavations [1–3]. Due to the con- small pieces which, however, increases the difficulties of building a
centration of workers and excavation equipment, the tunnel collapse at kinematically admissible velocity field to some extent. Subrin and
the face always poses a potential risk to people's lives and engineering Wong [10] proposed the 3D rotational failure mechanism of a tunnel
loss. Especially for shallow-buried tunnels, the face failure may cause face. The upper and lower contours of its symmetrical plane, namely
surface settlements or even collapse, which in turn leads to the de- the vertical cross-section along the tunneling direction, were defined by
struction of buildings on the ground. log-spirals. Unlike the former, the failure block was considered as a
In the realm of geotechnical engineering, stability of tunnel face is a rigid block rotating around the center point, so the internal energy
classical problem that has been discussed by many scholars. Several dissipation was only produced along the outer contour. This collapse
approaches are applied to estimate the face stability of tunnels, in- mechanism was also extended to handle the 3D stability problems of
cluding limit analysis method, limit equilibrium method, numerical slopes and retaining walls [11,12]. Mollon et al. [13] introduced a
simulation and experimental method [4–7]. Among them, the limit novel rotational failure mechanism of tunnel face by proposing a spatial
analysis method has been widely recognized as an efficient tool to re- discretization technique to generate the 3D failure block point by point.
solve the stability of geotechnical structures, such as slope stability, Pan and Dias [14,15] estimated the safety factor of a tunnel face with a
support pressure and bearing capacity of foundations. In the framework non-circular section using the spatial discretization technique, and

⁎
Corresponding author.
E-mail address: [email protected] (X.L. Yang).

https://doi.org/10.1016/j.compgeo.2019.02.014
Received 26 September 2018; Received in revised form 8 February 2019; Accepted 12 February 2019
0266-352X/ © 2019 Elsevier Ltd. All rights reserved.
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

discussed the influence of seepage forces on the stability of tunnel face.

Another important issue affecting the face stability assessment is the
yield criterion that is adopted to describe the strength of geomaterials.
The classical Mohr-Coulomb yield criterion is often used to calculate
the critical face pressure of a tunnel driven in soils in previously pub-
lished papers. However, they usually failed to recognize that the
strength of soils in tensile regime is remarkably reduced, or even is
taken as zero. This phenomenon can commonly be observed in slope
failure who sometimes has a very steep failure surface at the top. In
some works involving tunnel collapse, it forms a gentle corner or a
rounded corner at the top of the failure block instead of a sharp corner
described by two intersecting log-spirals, which probably verifies the
strength reduction in tensile regime of soils [16,17]. In reality, the soils
in tensile regime cannot afford the tensile strength that prescribed by
the linear Mohr-Coulomb function. In order to deal with this problem,
Drucker and Prager [18] defined a concept of tension cut-off to limit the
tensile strength of soils to zero which allows a non-linear strength en-
velop of soils subjected to tensile stress. On this basis, Paul [19] pro-
posed a strategy to limit the tensile strength by using three mutually (a)
perpendicular planes to cut the Mohr-Coulomb yield surface in the
principle stress space. This research allows the tensile strength of soils
ranges from zero (full tension cut-off) to a certain value (partial tension
cut-off). Recently, Michalowsk [20,21] investigated the 2D slope sta-
bility with tensile strength cut-off in which the pore water pressure is
also taken into account. Park and Michalowsk [22] extended this study
into a 3D slope which seems more convincing for landslides with sig-
nificant spatial characteristics. However, the investigation on tunnel
face stability with tensile strength cut-off is rarely reported in existing
works.
With the development of the cities and other infrastructure con-
structions, near-surface excavations are increasingly common in en-
gineering, which prompts people to consider the influence of tensile
strength cut-off in tunnel design. To fulfill this demand, this paper is
devoted to the 3D face stability of a shallow-buried tunnel driven in
soils in which two failure modes, namely outcrop and not outcrop, are
considered. In order to give a better estimation of face stability, the
tensile strength cut-off is introduced to modify the linear Mohr-
Coulomb yield criterion. A novel procedure is presented to calculate the
critical face pressure by combining the rotational failure mechanism of
tunnel face and the modified Mohr-Coulomb yield criterion.
(b)
2. Tensile strength cut-off

In previous studies, the strength of soils is typically characterized by

linear Mohr-Coulomb yield criterion whose parameters are obtained
from soil mechanics tests in the compressive regime and extrapolation
into the tensile regime. Accordingly, the uniaxial and triaxial tensile
strength of soils are 2ccosφ/(1 + sinφ) and c/tanφ respectively where c
and φ refer to the cohesion and internal friction angle of soils. In fact,
soils cannot afford so large a tensile strength, especially for the highly
weathered rock masses or unbounded granular soils whose tensile
strength is very close to zero. To describe the strength reduction of soils
in tensile regime, the tensile strength cut-off is introduced which allows
a non-linear strength envelop of soils when the normal stress is close to
zero or negative (tensile stress). Paul [19] elaborated the process to
limit the tensile strength of geomaterials in the principle stress space.
To make it better understood, Fig. 1 interprets the tensile strength cut-
off in the τ-σn space (τ and σn respectively represent the shear stress and
normal stress). T is a vector representing the traction on the failure
surface in the physical space; v is the velocity vector on the failure (c)
surface whose direction is determined by the associated flow rule; δ is
the dilatancy angle described by the nonlinear portion of the strength Fig. 1. Strength envelopes of Mohr-Coulomb soils with tension cut-off: (a)
partial tension cut-off; (b) full tension cut-off; (c) minimum tension cut-off.
envelope; fc and ft respectively denote the uniaxial compressive strength
and reduced uniaxial tensile strength. As a result, the internal energy
dissipation rate per unit area along the velocity discontinuity whose
stress state is defined by the nonlinear portion of the modified Mohr-

83
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

Coulomb yield condition can be readily calculated as a dot product of be expressed as

the traction vector T and the velocity vector v [23], namely
n
( )= ( 0) +
P0 = |v| fc
1 sin
+ ft
sin sin E 0 (4)
2 1 sin (1)
For a shallow-buried tunnel, the tunnel collapse probably extends to
where |v| is the magnitude of velocity vector. the ground which forms a failure mechanism with outcrop. When the
According to the classical Mohr-Coulomb function, the uniaxial buried-depth h is smaller than the vertical distance measured the tunnel
compressive and tensile strengths, denoted as fc and fm respectively, can roof to point E (the top of the failure block), the failure block outcrops
be expressed as and intersects with the surface at point C and D. It is an improved ro-
2c cos tational failure mechanism with tensile strength cut-off and the radial
fc = 1 sin cross-section is circular for any rotation angle. For ease of numerical
fm =
2c cos calculation, a local coordinate system with its origin at the center of the
1 + sin (2) circular section is built. As shown in Fig. 2, the circular section of the
For ease of comparison with the classical Mohr-Coulomb failure truncated curvilinear cone can be classified into three types with re-
criterion, the reduced tensile strength ft is denoted in the form of fm, spect to different rotation angles, namely I-I, II-II and III-III. r refers to
namely ft = ξ·fm. So Eq. (1) can be rewritten as the rotation radius and its subscript is used to distinguish different
rotation angles.
1 sin sin sin
P0 = c |v| cos +2
1 sin cos (3)
3.2. Kinematical analysis
The coefficient ξ ranges from 0 to 1. ξ = 0 represents a full tension
cut-off with zero tensile strength as shown in Fig. 1b, while ξ = 1 re- In the framework of limit analysis, the kinematically admissible
presents a minimum tension cut-off with ft = fm as shown in Fig. 1c. It is velocity field is necessary for establishment of the assessment model of
called as partial tension cut-off when 0 < ξ < 1 as shown in Fig. 1a. tunnel face stability. According to the geometric relationship in Fig. 2,
When δ is constantly equal to φ, the internal energy dissipation is de- rA and rB can be expressed as
graded to the traditional case without tension cut-off. It can be inferred
from Fig. 1 that the range of δ is from φ to 90°. rA =
d sin B
sin( A B)
d sin A
rB =
3. The 3D stability of tunnel face with tensile strength cut-off sin( A B) (5)

3.1. Computational model So the log-spiral curves AF, BG and the curves FC, GD can be de-
noted as
Unlike geotechnical structures on the ground, such as infinite slopes
rAF ( ) = rA exp[( A) tan ]
and retaining walls, tunnel excavations are limited to a small space
which leads to a remarkable 3D effect. As a consequence, the 3D rFC ( ) = rF exp tan ( ) d
0
computational model of face failure is essential for a reliable estimation
rBG ( ) = rB exp[( B ) tan ]
of face stability [24–26]. In this paper, the 3D rotational failure me-
chanism proposed by Subrin and Wong [10] is employed to perform rGD ( ) = rG exp tan ( ) d
0 (6)
face stability analysis of tunnels in soils governed by the Mohr-Coulomb
yield condition. Notice that this kind of failure mechanism cannot cover where
the whole tunnel face, which probably underestimates the critical
support pressure. By taking into account the tensile strength cut-off, a rF = rA exp[( 0 A) tan ]
modification to the linear Mohr-Coulomb yield criterion is introduced rG = rB exp[( B 0) tan ] (7)
which allows a non-logarithmic spiral failure surface of the collapse
mechanism. Due to the influence of tensile strength cut-off, it more The extension lines of FC, GD intersect at point E where the dila-
likely forms a gentle corner (angle CED in Fig. 2) at the top of the tancy angle δ reaches its maximum value δn. So the relationship be-
failure block rather than a sharp corner (angle CE'D in Fig. 2). Based on tween θ0, θE and δn can be derived as
the upper bound theorem of limit analysis, a novel procedure is in-
troduced to numerically determine the pattern of the failure mechanism n rA exp[( 0 A) tan ]
cos n = exp ln cos
that is in tensile regime under the objective function of maximum 2( E 0) rB exp[( B 0) tan ] (8)
pressure against tunnel face failure.
As shown in Fig. 2, a circle tunnel with a diameter of d and a buried When the tunnel collapse extends to the ground, the relationship
depth of h is excavated in Mohr-Coulomb soils. O is the rotation center between rA, rC, rD and h can be obtained as follows.
and ω is the angular velocity. θA, θB, θ0, θC, θD, θE denote the corre-
rA cos A rC cos C =h
sponding rotation angle as listed in Fig. 2. When the rotation angle θ
rA cos A rD cos D =h
ranges from θB to θ0, the failure surface has the shape of a curvilinear
cos ( C )
cone with upper and lower contours defined by log-spirals AF and BG.
E 0
rC = rF exp ln cos
n
The curvilinear cone is truncated by a vertical plane with θ ranging cos ( D)
E 0
from θB to θA, namely the tunnel face. A is the tunnel roof and B is the rD = rG exp ln
n cos (9)
tunnel invert. It is assumed that the failure block is in tensile regime
when θ is bigger than θ0. So FE and GE are not log-spiral curves, but the Obviously, Eq. (9) is only for the case that the part of the failure
curves determined by the dilatancy angle δ. In this part, the velocity block subjected to tensile stress is partially above the ground. In fact, it
direction forms an angle of δ with the velocity discontinuity which is possible that the top part of the failure block subjected to tensile
ranges from φ to δn as θ changes from θ0 to θE. As suggested by Mi- stress is completely cut away by the ground surface. This eventually
chalowski [20,21], the linear variation of δ as a function of the rotation leads to a face failure without tensile strength cut-off. In this case, Eq.
angle θ can obtain the best estimation of tunnel face stability. So δ can (9) is replaced by the following equation.

84
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

Fig. 2. The 3D rotational failure mechanism of tunnel face with tensile strength cut-off.

rA cos A rC cos C =h Likewise, Pγ2 can be obtained by

rA cos A rD cos D =h
D R R2 y 2
rC = rA exp[( C A ) tan ] P2 = 2 (rm + y )2sin dxdyd
0 R 0
rD = rB exp[( B D ) tan ] (10) l2 R2 y 2
C
+2 (rm + y )2sin dxdyd
In order to facilitate 3D analysis, the distance between O and the D R 0
center of circular cross-section is denoted as rm and the diameter of = rA4 [g13 ( 0 , D) + g14 ( C , D )] (15)
circular cross-section as R for any rotation angle. So rm and R can be
written as and the upper limit of integration l2 can be computed by

rA f1 ( ), < < rC cos C

B 0 l2 = rm = rA f6 ( )
rm = cos (16)
rA f3 ( ), 0 < < E (11)
where g13 ( 0, D ), g14 ( C , D ), f6 ( ) are given in Appendix A.
rA f2 ( ), B < < 0 When the failure block in tensile regime is completely above the
R=
rA f4 ( ), 0 < < E (12) ground, the upper limit of integration θ0 in Eq. (13) is replaced by θD
and Pγ2 is replaced by the following equation.
where f1 ( ), f2 ( ), f3 ( ), f4 ( ) are provided in Appendix A.
The work rate done by gravity can be divided into two parts, namely C l2 R2 y 2
P2=2 (rm + y ) 2sin dxdyd = rA4 g14 ( C, D)
the part described by log-spirals Pγ1 and the part in tensile regime Pγ2. D R 0

Pγ1 can be expressed as (17)

A R R2 y 2 where
P1 = 2 (rm + y )2sin dxdyd
l1 0
B rC cos C
l2 = rm = rA f6 ( )
0 R R2 y 2 cos (18)
+2 (rm + y )2sin dxdyd
A R 0
and g14 ( C , D ), f6 ( ) can be seen in Appendix A.
= rA4 [g11 ( A , B) + g12 ( A , 0 )] (13) Another part of external work rate is done by the support pressure
γ refers to the unit weight of soils and the lower limit of integration against tunnel face. Suppose that σT is a uniform pressure applied on
l1 can be calculated with the following equation. tunnel face, the work rate PT can be calculated as

rA sin A R2 l12 2 ·(rm + l1) 2cos

l1 = A
rm = rA f5 ( ) PT = T· dxd = rA3 T g2 ( A , B)
sin (14) B 0 sin (19)

where g11 ( A , B ), g12 ( A , 0 ), f5 ( ) are given in Appendix A. where g2 ( A , B) can be seen in Appendix A.

85
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

can be found in Appendix A.

3.3. Critical support pressure

The upper bound theorem of limit analysis states that the load
calculated by equating internal energy dissipation rate to external work
rate is no less than the actual collapse load when the boundary condi-
tions are satisfied in any kinematically admissible velocity fields. So the
critical support pressure against tunnel face failure can be computed
using the following equation.
P1 + P2 PT = Pd1 + Pd2 (23)

By submitting Eqs. (13), (15), (19)–(21) into Eq. (23), σT can be

calculated as
rA (g11 + g12 + g13 + g14) c (g31 + g32 + g33 + g34 )
T =
g2 (24)
Fig. 3. The critical support pressure versus h/d.
In order to ensure a reasonable failure mechanism of tunnel face, an
optimization program is coded to search the critical support pressure
Based on the assumption of rigid body, the internal energy dis- under the following constraints.
sipation is only produced along the failure surface. For the part of
failure block in compressive regime, the energy dissipation Pd1 can be < n < /2
written as 0< B < /2
B < A < /2
A 1
Pd1 = 2 cR (rm + R cos )2d d A < 0 <
B 0
0 0 < E < (25)
+ 2 cR (rm + R cos ) 2d d
A 0

= c r A3 [g31 ( A, B) + g32 ( A , 0 )] (20) 4. Numerical results and comparisons

where 1 = arccos(f5 / f2 ) and g31 ( A, B ), g32 ( A , 0 ) can be seen in
Appendix A. 4.1. Comparisons with existing findings
For the part of failure block in tensile regime, the tensile strength
cut-off has a significant influence on the strength of the soils which In order to validate the proposed approach, the critical support
leads to a non-logarithmic spiral failure surface. So the energy dis- pressures against tunnel face are calculated using the proposed ap-
sipation Pd2 can be calculated using Eq. (3), namely proach with respect to different values of h/d. The parameter settings
are as follows: γ = 18 kN/m3, d = 10 m, ξ = 0. Two cases of c = 7 kPa,
Pd2 =
D
2 cR (rm + R cos ) 2 cos
1 sin ( ) φ = 17° and c = 10 kPa, φ = 25° are taken into account for ease of
0 0 1 sin comparison with the findings of Mollon et al. [8] and Leca and Dor-
sin ( ) sin 1 mieux [9]. As shown in Fig. 3, the critical support pressure firstly grows
+2 d d with the increase of h/d (h/d < 0.5) and then becomes unchanged (h/
cos cos ( )
d > 0.5). The critical value of h/d at which the critical support pres-
C 1 sin ( ) sure keeps unchanged is relatively small for the case with a big value of
+ 2 cR (rm + R cos )2 cos
D 2 1 sin φ. It is because the height of the failure block gets smaller as the in-
sin ( ) sin 1 ternal friction angle becomes larger. The difference of the critical
+2 d d
cos cos ( ) support pressures with and without tensile strength cut-off is very close
to 0 when h/d is smaller than 0.3 in the case of c = 7 kPa. Then the
= c rA3 [g33 ( 0 , D) + g34 ( C , D )] (21) difference continuously increases to 0.93% and keeps unchanged when
where 2 = arccos(f6 / f4 ) and g33 ( 0, D ), g34 ( C , D ) can be seen in h/d increases to 0.5 or more. A similar tendency can be observed in the
Appendix A. case of c = 10 kPa. So it can be inferred that the tensile strength cut-off
Similarly, for the case that the failure block in tensile regime is has little influence on tunnel face stability when the failure mechanism
completely above the ground, Pd2 is replaced by the following equation. outcrops. In comparison with the previously published works, the cri-
tical support pressures are obviously improved according to the pre-
C
Pd2 = 2 cR (rm + R cos )2d d = c rA3 g34 ( C , D) sented failure mechanism, which shows the benefit of the proposed
(22)
approach.
D 2

where the lower limit of integration 1 = arccos(f6 / f2 ) and g34 ( C , D) A more direct comparison for tensile strength cut-off is performed

Table 1
Comparisons with the critical support pressures (kPa) given by Mollon et al. [13].
Presented solution Solution by Mollon et al. [13] Difference

Case 1 With tension cut-off 33.92 37.44 −9%

Without tension cut-off 33.61 37.07 −9%
Case 2 With tension cut-off 11.34 14.23 −20%
Without tension cut-off 10.63 13.71 −21%
Case 3 With tension cut-off 14.01 18.22 −23%
Without tension cut-off 10.77 15.20 −28%

86
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

Table 2
The critical support pressures with and without tensile strength cut-off.
c/kPa φ/° γ/kN·m−3 h/d ξ σT/kPa (with tension cut-off) σT/kPa (without tension cut-off) Increment

5 5 18 3 0 179.36 178.36 0.56%

5 10 18 3 0 82.45 81.94 0.62%
5 15 18 3 0 48.31 48.16 0.32%
5 20 18 3 0 31.37 31.28 0.31%
10 5 18 3 0 126.29 121.07 4.31%
10 10 18 3 0 56.03 53.53 4.67%
10 15 18 3 0 30.55 29.46 3.70%
10 20 18 3 0 18.20 17.51 3.92%
15 5 18 3 0 76.98 63.78 20.69%
15 10 18 3 0 31.38 25.12 24.93%
15 15 18 3 0 14.01 10.77 30.10%
15 20 18 3 0 6.25 3.75 66.65%
20 5 18 3 0 32.54 6.49 401.45%
20 10 18 3 0 9.32 0 –
20 15 18 3 0 0.16 0 –
20 20 18 3 0 0 0 –

by referring to the results given by Mollon et al. [13] as shown in

Table 1. Three cases are included according to Mollon et al. [13],
namely Case 1: c = 7 kPa, φ = 17°; Case 2: c = 10 kPa, φ = 25°; Case 3:
c = 15 kPa, φ = 15°. The critical support pressures with or without
tensile strength cut-off are given in one column. The results obtained by
two methods are listed in one row which is followed by the difference
between two methods. In total, the critical support pressures estimated
by the proposed approach are smaller than those given by Mollon et al.
[13]. The difference becomes bigger as the critical support pressure gets
smaller, but the effects of tensile strength cut-off estimated by two
methods are similar.
To further highlight the influence of tensile strength cut-off on
tunnel face stability, the critical support pressures with tensile strength
cut-off are calculated using the proposed approach and compared with
those without tensile strength cut-off. The numerical results are listed in
Table 2 in which only the case without outcrop is considered, namely h/
d = 2. ξ = 0 indicates that the case of full tension cut-off is adopted
here. σT = 0 shows that the tunnel face is self-stable without support
structures. The critical support pressures are computed with respect to
different shear strength parameters of soils where c ranges from 5 kPa (a)
to 20 kPa and φ ranges from 5° to 20°.
As shown in Table 2, the critical support pressure with tensile
strength cut-off is increased in comparison with those without tensile
strength cut-off. The increase of c makes the influence of tensile
strength cut-off more prominent. For the case of φ = 5°, the critical
support pressure is increased by 0.56% with c = 5 kPa, but it is in-
creased by 401.45% with c = 20 kPa. Unlike this situation, the tension
cut-off-induced increment of the critical support pressure changes
slightly with the variation of φ, but it becomes obvious when c is big
enough. In addition, it can be found that the smaller the critical support
pressure (a more stable tunnel face) is, the more remarkable the in-
fluence of tensile strength cut-off will be. It should be noticed that when
the tunnel face is close to limit state, the face stability is probably
mistakenly estimated by the conventional approach. For example, the
tunnel face is considered stable without considering the tensile strength
cut-off in the case of c = 20 kPa and φ = 10°, but actually it can't be
self-stabilizing due to the strength reduction in tensile regime of soils.
This phenomenon should be treated with caution in tunnel construc-
tion.

(b)
4.2. Influence of unit weight of soils
Fig. 4. Critical support pressure versus c considering the variation of unit
weight of soils (φ = 10°): (a) the absolute variation of critical support pressure;
In order to investigate the influence of unit weight of soils on the
(b) the relative increment of critical support pressure.
critical support pressure with tensile strength cut-off, three cases of
γ = 21 kN/m3, γ = 18 kN/m3 and γ = 15 kN/m3 are discussed in this

87
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

(a) Fig. 6. Critical support pressures for different values of ξ.

4.3. Influence of the parameter of ξ

In engineering, the fact that the soil cannot withstand any tensile
stress is not necessarily inevitable. In some cases, the partial tension
cut-off is more desirable to describe the reduction of soil strength in
tensile regime. For this purpose, the influence of the parameter of ξ is
discussed in this section considering different shear strength parameters
of soils. As shown in Fig. 6, γ = 18 kN/m3, h/d = 2, c ranges from 4 kPa
to 20 kPa and φ ranges from 5° to 20°. The critical support pressures are
calculated using the proposed approach as ξ varies from 0 to 1. The
results are compared with those derived from the case without tensile
strength cut-off.
It is shown that compared with the case without tension cut-off, the
critical support pressure varies distinctly when ξ changes from 0 to 0.5
but slightly when ξ changes from 0.5 to 1. This phenomenon can be
more obviously observed when c get bigger. The variation of φ only has
a small influence on the tension cut-off-induced increment of critical
support pressure. No matter ξ is equal to 0 or equal to 1, the critical
support pressure is bigger than that without tensile strength cut-off. So
(b) it can be concluded that by taking into account of the tensile strength
cut-off, we can obtain a higher critical pressure to retain the face sta-
Fig. 5. Critical support pressure versus φ considering the variation of unit
weight of soils (c = 10 kPa): (a) the absolute variation of critical support bility. It is of great importance for a reliable tunnel design, especially
pressure; (b) the relative increment of critical support pressure. for those tunnels excavated in hard soils or fractured rock masses.

section. Fig. 4a and b respectively present the absolute variation and

4.4. Discussions on the critical failure mechanism
relative increment of critical support pressure induced by tensile
strength cut-off with the increase of c (φ = 10°). It is shown that the
Fig. 7 presents the critical failure mechanisms of tunnel face with
effect of tensile strength cut-off becomes more obvious as c increases,
tensile strength cut-off considering the case of (a) c = 5 kPa and
which is in good agreement with the discussions above. This phenom-
φ = 15 kPa (b) c = 10 kPa and φ = 15 kPa (c) c = 15 kPa and
enon is intensified when the value of γ becomes smaller. As shown in
φ = 15 kPa where γ = 18 kN/m3, ξ = 0, d = 10 m and h/d = 2. The
Fig. 4b, the relative increment of critical support pressure grows sharply
variation of φ is not researched in this section in view of the fact that it
with the case of γ = 15 kN/m3.
has little influence on the tension cut-off-induced increment of critical
Fig. 5a and b present the absolute variation and relative increment
support pressure. It can be observed that the critical failure mechanism
of critical support pressure with the increase of φ respectively
leans backward and becomes shorter as the increase of c, which means a
(c = 10 kPa). It can be seen that φ has an obvious influence on the
more stable tunnel face simultaneously. Not surprisingly, the failure
critical support pressure, but it has only a little influence on the effect of
mechanism in tensile regime is enlarged as c increases, which just ex-
tensile strength cut-off. According to Fig. 5b, the relative increment of
plains why the influence of tensile strength cut-off is more obvious with
critical support pressure induced by tensile strength cut-off basically
a bigger value of c. Unlike the critical failure mechanism plotted ac-
remain unchanged as φ increases. With respect to the parameter of γ,
cording the conventional methods which always has a sharp corner at
the relative increment grows slightly when γ becomes smaller. In
the top, the critical failure block forms a round corner at the top due to
summary, it can be drawn that the tensile strength cut-off should be
tensile strength cut-off. This is a more intuitive explanation about the
seriously considered for tunnels excavated in hard soils (large cohesion)
influence of tensile strength cut-off on tunnel face stability.
with small unit weight.

88
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

y/m Failure block in

20 20 y/m y/m
(a) tensile regime 20
18 18 18
16 16 16
14 14 14
12 12 12
10 10 10
8 8 8
6 6 6
4 4 4
2 Failure block in 2 2
compressive regime
0 0 0
x/m
x/m
8 10 x/m
0 2 4 6 8 10 2 4 6
0 0 2 4 6 8 10
(b) Failure block in
20 y/m tensile regime 20 y/m 20 y/m
18 18 18
16 16 16
14 14 14
12 12 12
10 10 10
8 8 8
6 6 6
4 4 4
2 2 2
Failure block in
0 compressive regime 0 0
x/m
x/m
6 8 10 x/m
0 2 4 6 8 10 0 2 4
0 2 4 6 8 10

Failure block in
16 y/m tensile regime 16 y/m 16 y/m
(c)
14 14 14
12 12 12
10 10 10
8 8 8
6 6 6
4 4 4
2 2 2
Failure block in
0 compressive regime 0 0
x/m
x/m
2 4 6 8 10 x/m
0 2 4 6 8 10 0 0 2 4 6 8 10

Fig. 7. The critical failure mechanism of tunnel face with tensile strength cut-off: (a) c = 5 kPa and φ = 15 kPa; (b) c = 10 kPa and φ = 15 kPa; (c) c = 15 kPa and
φ = 15 kPa.

89
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

influence of tensile strength cut-off more prominent. The increase of c

leads to the decrease of the critical support pressure, but it is not the
case for the increase of φ especially when the tunnel face is affected by
a high pore water pressure. When ru is bigger than 0.2, there exists an
intersecting point of those curves. When c exceeds the point, the in-
crease of φ leads to an increase of the critical support pressure. More-
over, the intersecting point appears earlier with a higher pore water
pressure. This phenomenon can be explained as follows. The increase of
φ helps to enhance the stability of tunnel face while the increase of ru
weakens its stability. But the increase of φ also enlarges the component
of the pore water pressure in the movement direction of the failure
mechanism as is shown in Eq. (26). In fact, under the influence of pore
water pressure, the increase of φ is weakening the stability of tunnel
face simultaneously. And the weakening effect is more obvious than its
strengthening effect of tunnel face stability with a high pore water
pressure.
In addition, the critical support pressure linearly varies with the
increase of c when the tensile strength cut-off of soils is not considered.
But it becomes nonlinear when considering the influence of tensile
Fig. 8. The comparison with existing findings. strength cut-off. The curves become gentler and more concentrated for
a relatively stable tunnel face with the increase of ru, which probably
means that the stability of tunnel face is more affected by pore water
pressure than by the variation of shear strength parameters.
5. Stability of an undrained tunnel face with tensile strength cut-
off 6. Conclusions

Tunnel excavations are often subjected to pore water pressure in This paper is devoted to a novel 3D rotational failure mechanism
practical engineering. A predetermined distribution of pore water for stability assessment of tunnel face in soils governed by the
pressure u is often used to analyze the influence of pore water pressure modified Mohr-Coulomb failure criterion. A reduced or zero tensile
on tunnel face stability in the framework of limit analysis. Such a dis- strength of soils is introduced to modify the classical Mohr-
tribution is firstly proposed by Bishop and Morgenstern [27] for sta- Coulomb failure criterion. Accordingly, a novel procedure for the
bility analysis of slopes, which is also adoptable in the calculations of calculation of critical support pressure of tunnel face is presented
the critical support pressures of tunnel face under water table. In the based on the kinematical approach of limit analysis. The coefficient
literature, the distribution of pore water pressure is approximately r u is introduced to investigated the tunnel face stability with pore
described by the form of u = ruγH, where ru is the coefficient, γ refers to water pressure using the proposed approach. The following con-
the unit weight of the overlying soils and H denotes the distance from a clusions can be drawn.
point being considered to the water table. Guided by this idea, the work For shallow-buried tunnels whose failure mechanisms have a chance
rate done by pore water pressure Pu is included into the limit state to outcrop, the influence of tensile strength cut-off is less significant in
equation as a part of external work rate [28], which can be calculated comparison with the case of no outcropping. The tensile strength cut-
as off-induced increment of critical support pressure grows remarkably
with the increase of c, but it varies slightly with the increase of φ. This
Pu =
St
uv sin dSt
(26) tendency is also affected by the variation of unit weight of soils. It
should be noticed that a stable tunnel face evaluated by the conven-
where v is the velocity and St is the velocity discontinuity surface. tional method is proved to be unstable when considering the influence
Combined with the proposed 3D failure mechanism of tunnel face, the of tensile strength cut-off, which maybe leads to an incorrect assess-
specific calculation formulas are given in the Appendix B in which only ment.
the case without the failure mechanism outcropping is considered. As The parameter of ξ have influences on the stability of tunnel face.
suggested in literature [20,29], ru empirically takes 0–0.5. Compared with the case without tension cut-off, the critical support
To find out whether the distribution of pore water pressure pro- pressure varies distinctly when ξ changes from 0 to 0.5 but slightly
posed by Bishop and Morgenstern [27] is suitable to assess the tunnel when ξ changes from 0.5 to 1. So it can be inferred that ξ = 0 can give a
face stability, a comparison with Lee et al. [30] is performed as shown safer design of tunnel face.
in Fig. 8 where Hw/d = 0 means a dry case without the influence of As the variation of c have an obvious influence on tensile strength
pore water pressure. As we can see, when ru takes 0.4 or 0.5, the cal- cut-off, the views of the critical failure mechanism of 3D tunnel face are
culated results are close to those given by Lee et al. [30]. In literature plotted to give an insight into the proposed approach. It is shown that
[30], the pore water pressure is calculated using the numerical software the critical failure mechanism leans backward and becomes shorter as
which is subsequently added to the pressure caused by soils ahead of the increase of c, and meanwhile the 3D failure mechanism in tensile
tunnel face. Undoubtedly, this method is closer to the real situation, but regime is enlarged.
the numerical simulation inevitably consumes a lot of time. So using ru The influence of pore water pressure on tunnel face stability with
to determine the pore water pressure is an acceptable method in tunnel tensile strength cut-off is investigated. Results show that the increase
face stability assessment under water table. In project, ru can be de- of pore water pressure makes the influence of tensile strength cut-off
termined empirically or based on engineering detection, water pressure more prominent. With the influence of pore water pressure, the in-
test and other techniques. crease of c leads to the decrease of the critical support pressure. But
A further discussion is given in Fig. 9 where γ = 18 kN/m3, h/d = 2, the increase of φ probably leads to the increase of the critical support
H = 10 m, ξ = 0. The calculated results are also compared with those pressure in an indirect way, which depends on the magnitude of pore
without tensile strength cut-off with respect to different values of c and water pressure. It is because the increase of φ not only enhances the
φ. It can be observed that the increase of pore water pressure makes the stability of tunnel face, but also causes the failure surface to withstand

90
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

(a) (b)

(c) (d)

(e) (f)
Fig. 9. The influence of pore water pressure on tunnel face stability.

greater pore water pressure due to the shape change of the failure high a pore water pressure. In view of the curves of critical support
block. So the increase of φ weakens the tunnel face stability si- pressure versus c, the stability of tunnel face is more affected by pore
multaneously with the influence of pore water pressure. And the water pressure than by the variation of shear strength parameters for a
weakening effect is more obvious than its strengthening effect with relatively stable tunnel face.

91
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

Acknowledgment Innovation Foundation for Postgraduate of Central South University,

China (2018zzts185), and National Natural Science Foundation
The preparation of the paper has received financial supports from (51378510). The financial supports are greatly appreciated.

Appendix A. 3D failure mechanism of tunnel face with tensile strength cut-off

1 sin A
f1 ( ) = exp[( B ) tan ] + exp[( A) tan ]
2 sin B (A1)

1 sin A
f2 ( ) = exp[( B ) tan ] exp[( A) tan ]
2 sin B (A2)

1 sin A E 0 cos ( ) E 0 cos ( )

f3 = exp ( B 0) tan + ln + exp ( 0 A) tan ln
2 sin B n cos n cos (A3)

1 sin A E 0 cos ( ) E 0 cos ( )

f4 = exp ( B 0) tan + ln exp ( 0 A) tan ln
2 sin B n cos n cos (A4)

sin A
f5 ( ) = f1 ( )
sin (A5)

cos C E 1 cos ( C )
f6 ( ) = exp ( 0 A) tan ln f3 ( )
cos n cos (A6)

cos C
f6 ( ) = exp[( A) tan ] f1 ( )
cos
C
(A7)

A 4 2 4 2 1 1 3 f12 f22 f 24 f 24 f5
g11 ( A , B) = f12 f5 + ff f f + f22 f5 f f22 f52 + + f12 f22 + arcsin sin d
B 31 2 315 4 2 5 2 8 4 f2 (A8)

0 1 4
g12 ( A , 0) = f12 f22 + f sin d
A 4 2 (A9)

D 1 4
g13 ( 0, D) = f32 f42 + f sin d
0 4 4 (A10)

C 4 2 4 1 2 1 f32 f42 f 44 f 44 f6
g14 ( C , D) = f32 f6 f f + f4 f62 f f + f63 f42 f62 + + + f32 f42 + arcsin sin d
D 33 4 3 4 4 6 2 2 8 4 f4 (A11)

C 4 2 4 1 2 1 f22 f6
g14 ( C , D) = f12 f6 f f + f1 f6 2 f f + f6 3 f22 f6 2
+ f12 + f22 arccos sin d
D 31 2 3 4 2 6 2 4 f2 (A12)

A 2(f1 + f5 )2cos
g2 ( A , B) = f22 f52 d
B sin (A13)

A f5
g31 ( A , B) = (2f12 f2 + f23 ) arccos + f22 f52 (f2 f5 + 4f1 f2 ) d
B f2 (A14)
0
g32 ( A , 0) = [ f2 (2f12 + f22 )] d
A (A15)

D 1 sin ( ) sin ( ) sin 1

g33 ( 0, D) = f4 (2f32 + f42 ) cos +2 d
0 1 sin cos cos ( ) (A16)

C f6 1 sin ( ) sin ( ) sin 1

g34 ( C , D) = (2f32 f4 + f43 ) arccos f42 f62 (f4 f6 + 4f3 f4 ) cos +2 d
D f4 1 sin cos cos ( ) (A17)

C f6
g34 ( C, D) = (2f12 f2 + f23 ) arccos f22 f6 2 (f2 f6 + 4f1 f2 ) d
D f2 (A18)

92
T.Z. Li and X.L. Yang Computers and Geotechnics 110 (2019) 82–93

Appendix B. Calculation of work rate done by pore water pressure

A 1
Pu = 2 0
ru H (rm + R cos )2R tan d d
B
0
+2 0
ru H (rm + R cos ) 2R tan d d
A
E
+2 0
ru H (rm + R cos ) 2R tan ( ) d d
0

= ru rA4 [g41 ( A, B) + g 42 ( A , B) + g43 ( A , 0) + g44 ( A , 0) + g 45 ( 0 , E) + g 46 ( 0 , E )] (B1)

where the first two terms refer to the work rate done by the pore water pressures that are applied on the failure mechanism in compressive regime
and the third term denotes that in tensile regime.
1 = arccos(f5 / f2 ) (B2)

H = (rm + R cos ) cos + Hw rA cos A (B3)

Hw = Hw / rA (B4)
where Hw represents the distance measured from the tunnel roof to the water table.

g41 ( A , B) = A
(f1 cos + Hw cos A ) f2 [(2f12 + f22 )·
B

arccos ( )+f5
f2
f22 f52 (f5 + 4f1 ) tan d
(B5)

g42 ( A , B) = A
B
f2 (2f 1
2 4
+ 3 f22 + 3 f52 + 3f1 f5 ·
2
)
f22 f52 + 2f1 f22 arccos ()f5
f2
cos tan d
(B6)
0 2
g43 ( A , 0) = (f1 cos + Hw cos A ) f2 (2f1 + f22 ) tan d
A (B7)
0
g44 ( A , 0) = 2f1 f23 cos tan d
A (B8)
E 2
g45 ( 0, E) = (f3 cos + Hw cos A ) f4 (2f3 + f42 ) tan ( ) d
0 (B9)
E
g46 ( 0, E) = 2f3 f43 cos tan ( ) d
0 (B10)

References Anal Meth Geomech 2016;40(15):2123–36.

[15] Pan Q, Dias D. Safety factor assessment of a tunnel face reinforced by horizontal
dowels. Eng Struct 2017;142:56–66.
[1] Huang M, Tang Z, Zhou W, Yuan J. Upper bound solutions for face stability of [16] Huang M, Li S, Yu J, Tan JW. Continuous field based upper bound analysis for
circular tunnels in nonhomogeneous and anisotropic clays. Comput Geotech three-dimensional tunnel face stability in undrained clay. Comput Geotech
2018;98:189–96. 2018;94:207–13.
[2] Paternesi A, Schweiger HF, Scarpelli G. Numerical analyses of stability and de- [17] Zou J, Chen G, Qian Z. Tunnel face stability in cohesion-frictional soils considering
formation behavior of reinforced and unreinforced tunnel faces. Comput Geotech the soil arching effect by improved failure models. Comput Geotech 2019;106:1–17.
2017;88:256–66. [18] Drucker DC, Prager W. Soil mechanics and plastic analysis or limit design. Q Appl
[3] Zhang R, Yang XL. Limit analysis of active and passive mechanisms of shallow Math 1952;10(2):157–65.
tunnels in nonassociative soil with changing water table. Int J Geomech [19] Paul B. A modification of the Coulomb-Mohr theory of fracture. J Appl Mech
2018;18(7):04018063. 1961;28(2):259–68.
[4] Zhang F, Gao Y, Wu Y, Wang Z. Face stability analysis of large-diameter slurry [20] Michalowski RL. Stability of intact slopes with tensile strength cut-off.
shield-driven tunnels with linearly increasing undrained strength. Tunn Undergr Géotechnique 2017;67(8):720–7.
Space Technol 2018;78:178–87. [21] Michalowski RL. Failure potential of infinite slopes in bonded soils with tensile
[5] Zhang Z, Shi X, Wang B, Li H. Stability of NATM tunnel faces in soft surrounding strength cut-off. Can Geotech J 2017;55(4):477–85.
rocks. Comput Geotech 2018;96:90–102. [22] Park D, Michalowski RL. Three-dimensional stability analysis of slopes in hard soil/
[6] Song Y, Zhang N, Huang G, Sun Z. Experimental study on motions of tunnel element soft rock with tensile strength cut-off. Eng Geol 2017;229:73–84.
during immersion standby stage in long wave regime. Ocean Eng 2018;161:29–46. [23] Michalowski RL. Limit analysis of quasi-static pyramidal indentation of rock. Int J
[7] Perazzelli P, Anagnostou G. Stress analysis of reinforced tunnel faces and compar- Rock Mech Mining Sci Geomech Abstr 1985;22(1):31–8.
ison with the limit equilibrium method. Tunn Undergr Space Technol [24] Li YX, Yang XL. Soil-slope stability considering effect of soil-strength nonlinearity.
2013;38:87–98. Int J Geomech 2019;19(3):04018201.
[8] Mollon G, Dias D, Soubra AH. Probabilistic analysis and design of circular tunnels [25] Ukritchon B, Yingchaloenkitkhajorn K, Keawsawasvong S. Three-dimensional un-
against face stability. Int J Geomech 2009;9(6):237–49. drained tunnel face stability in clay with a linearly increasing shear strength with
[9] Leca E, Dormieux L. Upper and lower bound solutions for the face stability of depth. Comput Geotech 2017;88:146–51.
shallow circular tunnels in frictional material. Geotechnique 1990;40(4):581–606. [26] Qin C, Chian SC. 2D and 3D stability analysis of tunnel roof collapse in stratified
[10] Subrin D, Wong H. Stability of the front of a tunnel in a rubbing environment: a new rock: a kinematic approach. Int J Rock Mech Min Sci 2017;100:269–77.
3D break mechanism. Rendus Mécanique 2002;330(7):513–9. [27] Bishop AW, Morgenstern NR. Stability coefficients for earth slopes. Geotechnique
[11] Xu JS, Yang XL. Seismic stability of 3D soil slope reinforced by geosynthetic with 1960;10(4):129–50.
nonlinear failure criterion. Soil Dyn Earthquake Eng 2019;118:86–97. [28] Viratjandr C, Michalowski RL. Limit analysis of submerged slopes subjected to
[12] Yang XL, Li ZW. Kinematical analysis of 3D passive earth pressure with nonlinear water drawdown. Can Geotech J 2006;43(8):802–14.
yield criterion. Int J Numer Anal Meth Geomech 2018;42(7):916–30. [29] Li ZW, Yang XL. Kinematical analysis of active earth pressure considering tension
[13] Mollon G, Dias D, Soubra AH. Rotational failure mechanisms for the face stability crack, pore-water pressure and soil nonlinearity. KSCE J Civ Eng 2019;23(1):56–62.
analysis of tunnels driven by a pressurized shield. Int J Numer Anal Meth Geomech [30] Lee IM, Nam SW, Ahn JH. Effect of seepage forces on tunnel face stability. Can
2011;35(12):1363–88. Geotech J 2003;40(2):342–50.
[14] Pan Q, Dias D. The effect of pore water pressure on tunnel face stability. Int J Numer

93