Computers and Geotechnics 122 (2020) 103544

Contents lists available at ScienceDirect

Computers and Geotechnics

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

Research Paper

Stability analysis of 3D cracked slope reinforced with piles T

X.L. Yang , S. Zhang
⁎

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

ARTICLE INFO ABSTRACT

Keywords: The presence of cracks in soil slopes is of major concern to researchers and engineers, as this phenomenon is
Cracks typically considered to be an early indication of instability and potential landslides. In this study, the kinematical
Stabilizing piles approach of limit analysis was applied to evaluate the three-dimensional (3D) stability of soil slopes with pre-
Soil slope existing cracks and the reinforcement effects of piles. A vertical crack was accommodated in the classical 3D
Limit analysis
rotational failure mechanism by introducing a velocity discontinuity surface. Formulas for the external work
Factor of safety
rates and internal energy dissipation rates of the failure mechanism were derived within the framework of limit
analysis. An optimization program was developed in combination with the strength reduction method to search
for the most critical failure mechanism and the corresponding upper-bound factor of safety. A new parameter,
the maximum crack depth coefficient, was introduced for evaluating the sensitivity of slopes to the presence of
cracks. The numerical results indicated that slopes with large inclination angles are extremely sensitive to
natural cracks, whereas gentle slopes are less likely to be affected. It was also found that stabilizing piles could
effectively restrict the depth of the crack opening and thus, improve the stability of fissured slopes; the re-
inforcement effect was extremely dependent on the installation location and the density of the pile group.

1. Introduction engineering [8–11]. In recent years, several studies have investigated

the two-dimensional (2D) stability of fissured slopes using the limit
Under the influence of various environmental factors such as de- analysis method. A preliminary study was presented by Utili [12]. The
siccation, weathering, or drying-wetting cycles, soil slopes often present failure mechanisms were considered for slopes with three types of
natural cracks [1–4]. This issue is of major concern for researchers and cracks: cracks with known depths but undetermined locations, cracks
engineers, as the presence of cracks in slopes is typically considered as with known locations but undetermined depths, and cracks with un-
an early indication of instability and potential failure. The influence of determined locations and depths. Then, given that slopes with cracks
the cracks on slope stability has been investigated using different are more vulnerable under seismic conditions, the proposed method
methods, which can be generally classified into three categories: the was extended with the pseudo-static method to include seismicity [13].
limit equilibrium method, limit analysis method, and numerical Michalowski proposed a mechanism for considering the crack opening
methods. The limit equilibrium method has long been used in the field process within the strict framework of limit analysis, and studied the
of slope stability analysis, owing to its practicality and simplicity. An influences of pre-existing cracks and formation cracks (cracks formed as
early study of this issue was conducted by Baker [5], in which the a part of the failure process) [14]. A comparison study on the stability
depths of tension cracks and their influence on the safety of soil slopes of slopes with tension cracks was presented by Gao et al. [15]. It was
were first investigated. Chowdhury and Zhang [6] developed an opti- found that the upper-bound solutions obtained by the limit analysis
mization approach based on Janbu’s generalized procedure of slices to method were quite consistent with the limit equilibrium solutions.
search for the critical slip line, as well as the critical depth and location Stabilizing pile group is one of the most commonly installed re-
of a crack. It was found that the depth of the water in a tension crack inforcement measures in engineering practice and is used to enhance
has a significant effect on the stability of slopes. By assuming a full- the stability of natural slopes. A study examined the reinforcement ef-
height dry tension crack in an embankment, researchers studied the fect of drilled shafts on soil slope stability using the limit equilibrium
effects of berms and tension cracks on the stability of embankments for method [16]. The soil arching effect, i.e., the stress transfer from the
soft soils [7]. The limit analysis method is another effective approach yielding zone to the adjacent non-yielding zone, was considered by
for stability analysis, and has been used extensively in geotechnical incorporating the load transfer characteristic curves into the traditional

⁎
Corresponding author.
E-mail addresses: yangky@aliyun.com (X.L. Yang), 174811044@csu.edu.cn (S. Zhang).

https://doi.org/10.1016/j.compgeo.2020.103544
Received 30 December 2019; Received in revised form 12 March 2020; Accepted 12 March 2020
0266-352X/ © 2020 Elsevier Ltd. All rights reserved.
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

slice technique. Martin and Chen investigated the responses of piles O

undergoing lateral soil movements [17]. They argued that a pile’s

θh
θ0
θD
failure mode was highly related to the relative stiffness between the pile

θB
and soil. A numerical study of the stability of pile-reinforced slopes was

θC
θP
presented by Won et al. [18]. The results showed that neglecting the
M
coupling effect of the pile-slope system would lead to a more con- r'0
servative solution, and that the pile head conditions and bending A' 2φ
stiffness have important influences on the reinforcement effect of the
piles. Ausilio et al. conducted an upper-bound stability analysis of E' r0
slopes reinforced with piles [19]. The piles were considered by as- D
rm B A
suming that a lateral force and a moment were both applied at the
rh
depth of the potential sliding surface. A similar study was presented on P' h x
the influences of pile location and deployment density, and estimated

R
the resistance force acting on piles using an equation proposed by Ito p(z) C

α1*
hp H
and Matsui [20–21]. d1
y
In the previous literature concerning slope stability, few studies
have examined the stability of 3D slopes with cracks. Moreover, the β P x

R
reinforcement effects of stabilizing piles on fissured slopes have not E
α2
*

been fully investigated. Based on the upper bound limit analysis the- d2
orem, this study investigated the stability of a homogeneous frictional/ XF y
cohesive soil slope with a pre-existing vertical crack and pile re-
inforcement. A novel 3D rotational failure mechanism was postulated Fig. 1. Rotational failure mechanism for slopes with cracks and stabilizing
for a slope containing a vertical velocity discontinuity surface. The piles.
results for the upper-bound factor of safety were calculated through an
iteration program and are presented in the figures. The coupling effects
of the pre-existing crack and pile reinforcement were illustrated to fa- Crack plane

cilitate further engineering design. B

b
Acting face of piles
2. Description of the problem L2

A homogeneous frictional/cohesive soil slope with a pre-existing L1

vertical crack on its top surface was considered in this study. It was
reinforced by a row of stabilizing piles to mitigate the risk of landslides. H
Three subproblems of the analysis, i.e., the 3D failure mechanism of a Plane strain insert
fissured slope, maximum depth of the vertical crack, and reinforcement β
effects of the pile group, are discussed as follows.
LX
2.1. Failure mechanism for slopes with natural cracks
Fig. 2. Schematic diagram of 3D failure mechanism with plane strain insert.

Under the framework of limit analysis, one primary issue concerns

establishing a kinematically admissible velocity field (i.e., the failure The calculations for f1 and f2 are presented in the Appendix.
mechanism) [22]. Initially, a 3D slope with slope height = H, slope As illustrated in Fig. 2, the 3D failure mechanism was modified with
width = B, and slope angle = β was postulated. Based on the classical a plane insert block, allowing the transition to a plane-strain me-
3D rotational failure mechanism, a velocity discontinuity surface was chanism (provided that its width b was large enough). Put another way,
introduced to accommodate the vertical crack. As shown in Fig. 1(a), the entire sliding body is composed of two parts, i.e., a curvilinear cone
region DBEC (the sliding body) rotates around the axis passing through and a plane insert mechanism. The sliding surface was assumed to be
point O at angular velocity ω, whereas region ADC remains static as the passing through the slope toe, so the aforementioned 3D rotational
remaining part of the slope. The vertical crack (with a depth of h) is mechanism could be determined by four independent variables, i.e., θ0,
represented by DC, and its location is determined by the parameter θC. θh, θC, and r'0/r0.
The sliding surface is a sector of a curvilinear cone with an apex angle
of 2φ, where φ is the friction angle of the soil. The lower and upper 2.2. Maximum depth of a vertical crack
contours of the symmetry plane are defined by two logarithmic spirals.
The lower curve AE and the upper curve A'E' are expressed as follows: Owing to the limitations of boundary conditions, cracks beyond a
certain depth cannot exist naturally. To form a reasonable and ad-
AE : r = r0 e ( 0 ) tan (1)
missible failure mechanism, the depths of the cracks need to be re-
A E : r = r0 e ( 0 ) tan (2) stricted during optimization. This issue can be solved by considering an
unsupported vertical cut, where the critical height of the cut is equal to
Here, OA = r0 and OA' = r'0; θ is the generic angle of the loga- the maximum depth of a vertical crack.
rithmic spirals, and θ0 is the angle between OA and the horizontal di- Following this approach, several solutions have been obtained
rection. The cross-section of the cone is a circle with a radius of R, and under the framework of limit equilibrium. For example, Fellenius ob-
the distance between the cone axis and rotation center O is rm. R and rm tained an improved solution by using a circular arc as failure surface as
can be expressed as follows: a substitute for the inclined plane (translational mechanism) used by
Terzaghi [23,24], as follows:
rm = (r + r ) 2 = r0 f1 ( ) (3)
3.85c
R = (r r ) 2 = r0 f2 ( ) (4) hcr = tan +
4 2 (5)

2
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

Table 1 d
Comparison of the results of γhcr/c.
γhcr/c φ = 10° φ = 20° φ = 30° φ = 40°
pile
B
Chen (1975) 4.56 5.47 6.63 8.21
Michalowski (2013) 4.58 5.50 6.69 8.29 π /8+φ/4
This study (2D) 4.58 5.52 6.69 8.31 π/4+φ/2
This study (B/H = 5) 4.64 5.58 6.81 8.48 A E
This study (B/H = 2) 4.73 5.73 7.02 8.75
This study (B/H = 0.8) 5.03 6.11 7.60 9.71 plastic
D2 D1 x deformation
zone
In the above, c is the soil cohesion, and is the unit weight of the soil.
However, the solution derived from the limit equilibrium method is not A' E'
a lower or upper bound, as only the equilibrium condition is satisfied.
Chen [22] employed the plastic limit theorems and found that the π /4- φ/2
critical height of an unsupported vertical cut is as follows: B'
pile
3.83c
hcr = tan +
4 2 (6)
Fig. 3. Plastic deformation zone around piles (after Ito and Matsui, 1975).
In contrast with Eq. (5), the equilibrium condition, boundary con-
ditions, and stress–strain relationship are all satisfied for Eq. (6); thus, it p (z )
is a valid upper bound solution. After some rearrangements, the critical (N 1 2 tan + N 1)
height of an unsupported cut can be written in a dimensionless form, as D1
= cD1
γhcr/c. The results for γhcr/c, as obtained by different methods, are D2
listed in Table 1. 1 D1 D2
To better demonstrate the effects of natural cracks on slope stability, exp N tan tan + 2N1 2 tan 1
N tan D2 8 4
a maximum crack depth coefficient kmax was introduced in this study. It
is defined as kmax = hcr/H, where hcr is the maximum depth of the 2 tan + 2N1 2
+N 1 2
2 tan + 2N1 2
+N 1 2
+ c D1 2D2
crack. As the results for the critical slope height obtained by the kine- N1 2 tan +N 1 N1 2 tan +N 1
matical approach are upper-bound solutions, the hcr of a 3D slope could
be obtained by optimizing the formula for the critical height of a ver- N 1 2

tical slope. The least value of hcr is the maximum crack depth.

(N 1 2 tan + N 1)
z D D1 D2
2.3. Reinforcement with a row of stabilizing piles + D1 1 exp N tan tan + D2
N D2 D2 8 4

Stabilizing piles are commonly used in slope engineering to re- (9)

inforce unstable slopes. The piles are typically installed in a row within Here, N = 4 + 2) , and D1 and D2 are the center-to-center
tan2 (
the slope, known as the pile group. The location of the pile group is interval and clear interval between piles in a row, respectively. Hence,
denoted by a coefficient n and is defined as follows: (D1–D2) represents the diameter of the stabilizing piles, denoted as d in
Fig. 3. γ is the unit weight of the soil, and z represents an arbitrary
XF X tan
n= = F depth of the soil layer from the ground surface. The soil around the piles
LX H (7)
was assumed to be in plastic state to form an arching zone, as shown in
Fig. 3. The interval between piles should not be too large; otherwise,
Here, XF and LX are the horizontal distance from the slope toe to the
Eq. (9) may no longer be valid for estimating the lateral force dis-
pile group and slope crest, respectively, as illustrated in Figs. 1 and 2. It
tribution of the piles.
should be noted that when n = 0/1, the pile group is located at the toe/
crest of the slope. The intersection of the pile group and sliding surface
is denoted as P, and a relationship can be observed, as follows: 3. the kinematical approach of limit analysis

XF + rh cos h = rp cos p (8) Based on the theory of perfect plasticity, the kinematical approach
assumes that the soil around the slip surface follows the flow rule, and
In the above, rh/rp and θh/θp are the coordinates of point E/P in the thus solves for the critical parameters without the complexity of a step-
polar system, respectively. by-step elastic–plastic analysis [25–28]. The upper bound solutions for
Although many research studies have been conducted on the effects ultimate failure can be determined by equating the external work rate
of reinforcing piles, the previous research presents relatively limited to the internal dissipation rate in a kinematically admissible velocity
theoretical solutions regarding lateral force distribution models for the field [23], as follows:
piles. In this respect, a widely employed estimation equation was pro- W = Dint + Dp (10)
posed by Ito and Matsui [21], in which two types of plastic states
around piles, namely, the theory of plastic deformation and the theory Here, W is the external work rate, andDint andDp are the energy
of plastic flow, were respectively considered, so as to derive analytical dissipation rates owing to internal friction and the resisting force of the
solutions for the lateral force acting on piles. As shown in Fig. 3, they piles, respectively.
assumed that the soil in the area of AEBB E A undergoes plastic de-
formation and the angle between EB and the × axis is ( 4 + 2) , and 3.1. Calculation of external work rate
it was estimated that per unit thickness of soil layer, the lateral force
acting on each pile is as follows: The crack is assumed to be dry, and groundwater pressure is not

3
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

considered in this study. Thus, the rate of external work is determined D3D = D SACE D SAC = cwr03 (g5 g6 ) (21)
by the soil weight only, i.e., W = Wγ. The work rate based on self-
weight consists of two portions which are calculated separately, as Details regarding g5 and g6 are reported in the Appendix. D SACE is
follows: equal to the internal energy dissipation of an intact slope (i.e., without a
crack), and can be obtained. D SAC can be obtained using the same for-
W =W 3D +W insert (11) mula as that for DACE with the slope angle modified to be 90°. Likewise,
In Eq. (11), Wγ-3D and Wγ-insert are the work rates for the curvilinear the energy dissipation for the plane insert mechanism is expressed as
cone and plane insert mechanism, respectively. For the former, as the follows:
block ADC stays static, Wγ-3D is determined by subtracting the work of Dinsert = D ACE D AC (22)
block ADC from the work of block ABE, as follows:
In the above, the integral domains ΓACE and ΓAC represent the cy-
W 3D = WABE WADC (12) linder sliding surfaces ACE and AC in Fig. 1, respectively. The energy
For the sake of brevity, the derivations and formulas for calculating dissipation on an infinitesimal element is calculated as follows:
the work rates are presented in the Appendix. By introducing a local rd
coordinate system x-y, WABE can be obtained through multiple in- Dinsert = b cv cos = bcw r 2d
cos (23)
tegrations. After some rearrangements, WABE can be expressed as fol-
lows: Here, is a generic integral domain, v denotes the velocity at any
point on the sliding surface, and v = r .
WABE = r0 4g1 ( 0, h, r 0 r0) (13) By substituting Eq. (1) into Eq. (23) and performing some re-
WADC is calculated in the same way, except that the slope angle is arrangement, D ACE and D AC can be obtained as follows:
90° as the crack is assumed to be vertical. The formula of WADC is ex- e 2( h 0 ) tan 1
pressed as follows: D = bcwr02
(24)
ACE
2 tan
WADC = r0 4g2 ( 0 , C, r 0 r0 ) (14)
e 2( C 0 ) tan 1
Thus, the external work rate of the 3D portion is as follows: D = bcwr02
(25)
AC
2 tan
W 3D = r0 4 (g1 g2) (15) Thus, the total energy dissipation rate for the plane insert me-
Here, g1 and g2 are two parameters determined by the geometry of chanism is as follows:
the failure mechanism. The calculations for g1 and g2 are presented in e 2( h 0) tan e 2( C 0 ) tan
the Appendix. Dinsert = bcwr02
2 tan (26)
Following the same methodology, the external work rate of the
plane insert mechanism is expressed as follows:
3.3. Work dissipation rate of stabilizing piles
W insert = W ABE W ADC (16)
In Eq. (16), W'ABE and W'ADC are the work rates of the plane insert The work dissipation per unit thickness of the layer for a single pile
blocks ABE and ADC, respectively. As the plane insert mechanism is in a is equal to the product of the lateral force at a certain depth and the
plane strain state, its external work rates can be obtained by multi- corresponding velocity:
plying the corresponding work rates under a 2D situation by its width b.
dDp = p (z ) l (z ) dz (27)
According to Chen, the external work rate of block ABE under 2D
analysis is as follows [22]: In the above, p(z) is the lateral force along the piles, and l(z) is the
2D arm of the lateral force. As shown in Fig. 4, P P is the active length of the
WABE = r03 (f6 f7 f8 ) (17)
stabilizing piles (piles intercepted by the slope surface and sliding
Therefore, surface), and dp is the active width of a pile group at any depth.
W ABE = r03 b (f6 f7 f8 ) = r04 g3 (18) O
θ θ θ' p active length of
The work rate of the block ADC can be obtained in the same way, as p
stabilizing piles
follows: P'
r pc
os

W ADE = r03 b (f9 f10 f11 ) = r04 g4 (19)

θp
/co

l(θ) z
Additional details regarding g3, g4, and f6–f11 are presented in the
sθ

Appendix. dθ

r pcosθ pdθ/cos2θ
3.2. Calculation of internal energy dissipation rate

rp x
As the rotational mechanism is assumed to be a rigid body (volume active width
R

strain is neglected), the internal energy is only dissipated on the sliding of piles in a row
surface. Moreover, the crack opening process is not included here, as
the crack forms before the slope failure. Therefore, the internal energy
dissipation rate is expressed as follows: y
Dint = D3D + Dinsert (20) dp
P
Here, D3D and Dinsert are the work dissipated on the 3D portion (arc
surface) and 2D portion (cylinder surface), respectively. D3D is calcu- Fig. 4. Illustration of work dissipation on an infinitesimal length of stabilizing
lated by subtracting the energy dissipation rate on 3D sliding surface piles.
SAC from that on sliding surface SACE and is expressed as follows:

4
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

According to geometric relations, the following equations can be de- Start

termined:
Input lower/upper bound factors of safety
z = rp cos p tan rh sin h + XF tan (28) FS1/FS2 and calculation accuracy k

Factor of safety
XF + rh cos h rp cos p =0 (29) FS=(FS1+FS2)/2

XF tan rh sin h + rp cos p tan p =0 (30) Calculating work rates

with optimization variables
Here, θp (θ'p) are the angles between OP (OP') and the horizontal
direction, respectively. Substituting Eq. (28) into Eq. (27) and in- max[W - Dint+D p ]>0?
tegrating along the active section of the pile group, the dissipation rate Yes No
for a row of piles can be obtained as follows: FS1=FS, FS2=FS2 FS2=FS, FS1=FS1
p p( ) rp cos p
Dp3D = l ( ) dp d No FS2-FS1 k ?
p D1 cos2 (31)
Yes
p p( ) rp cos p Factor of safety
Dp2D = l ( )b d FS=(FS1+FS2)/2
p D1 cos2 (32)
End
In the above, θp and θ'p can be solved for from the implicit Eqs. (29)
and (30). The expressions for dp and l(θ) are given as follows: Fig. 5. Flowchart of the optimization scheme.

2
rp cos p
dp = 2 R2 rm
cos (33)
4. Comparison and parametric analysis
l ( ) = rp cos p tan (34)
4.1. Comparison with 2D conditions

As is well known, when the width-to-height ratio B/H is large en-

3.4. Factor of safety ough, the 3D rotational failure mechanism can be regarded as in a
plane-strain state, and the upper bound solutions of a 3D stability
An optimization program combined with the strength reduction analysis are the same as those of a 2D analysis. To verify the effec-
method (SRM) was developed to search for the upper-bound factor of tiveness of the proposed analysis, the value of the dimensionless critical
safety (FS). According to the SRM, the FS is defined as the ratio of the height γH/c in this study was calculated and compared with the results
actual strength of the soil to the reduced strength at the critical state, as of Michalowski under 2D conditions, as shown in Table 2. Evidently, for
follows: either intact slopes or slopes with a pre-existing crack, the γH/c value
c tan obtained from this study shows good agreement with previous results.
Fs = ~ = Thus, the present approach proved to be effective for analyzing the
c tan (35)
stability of fissured slopes.
where c , are the actual shear strength parameters of the soil and ~c, The maximum crack depth coefficient kmax values under different B/
are the shear strength parameters of the soil that brings the slope to H values are presented in Table 3. As can be seen from Table 3, kmax
failure. Based on the above analysis, the FS can be captured through an decreases along with an increase in φ, and the positive effect of φ is
iteration process, by solving the implicit equation of FS, as follows: more pronounced for slopes with greater inclination angles. Moreover,
the maximum depth coefficient kmax of the 3D failure mechanism is
FS = min f ( 0, h, C , r0 r0 |c, , , n, D1 D2 ,B, H, ) (36) smaller than that of the 2D slopes, meaning that long slopes (with large
To make the failure mechanism geometrically meaningful, the op- B/H) are more likely to be affected by cracks.
timization variables should satisfy the following constraint conditions:
4.2. Comparison with intact slopes
0< < B< C< h<
0
0 < r 0 r0 < 1 Fig. 6 illustrates the FS values for intact and fissured slopes under
0 < (b + B max ) H < B H (37) different width-to-height ratios and slope angles. The decreasing curves
represent the FS values of the slopes, and the rising curves correspond
Here, B'max is the maximum width of the rotational failure me- to the decreasing rates of FS, denoted as η in Fig. 6. The comparative
chanism, and B is the finite width of the slopes. Notably, the results of results show that the impact of the pre-existing crack on the slope
the kinematic method are upper bounds to the actual solutions, so the stability is mainly dependent on the slope inclination angle. For slopes
lowest result is the optimal solution. The problem turns out to be the with inclination angles < 40°, their stability is almost unaffected by the
optimization of an objective function with multiple constraints and presence of natural cracks, with the FS reduced by no more than 5%,
variables. As shown in Fig. 5, a nested loop structure was employed in even if the internal friction angle of the soil is relatively small. As shown
the optimization scheme to determine the minimum FS. When a certain in Table 3, for slopes with = 30° , their maximum crack depth coeffi-
set of optimization variables (θ0, θh, θC, r'0/r0) violate the constraint cients are rather small, meaning that in gentle slopes, the propagation
conditions or yield a reasonable failure mechanism, the optimization of natural cracks is restricted. However, once the slope angle exceeds
variables are changed sequentially in a single computation loop. The 60°, the influences of cracks become much more evident. The FS values
procedure is repeated until the lowest upper-bound solution among all of fissured slopes could decrease by as much as 30% as the slope angle
reasonable failure mechanisms is captured. approaches vertical, indicating that the presence of natural cracks in

5
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

Table 2
Comparison of dimensionless critical height γH/c under 2D conditions.
γH/c no crack pre-existing crack

φ = 10° φ = 20° φ = 30° φ = 10° φ = 20° φ = 30°

β = 30° Michalowski (2013) 13.499 41.215 — 12.969 40.566 —

This study 13.502 41.256 — 12.968 40.562 —
β = 45° Michalowski (2013) 9.309 16.160 35.540 8.524 15.283 34.555
This study 9.311 16.163 35.547 8.523 15.234 34.717
β = 60° Michalowski (2013) 7.257 10.390 16.035 6.233 9.212 14.696
This study 7.258 10.392 16.038 6.260 9.255 14.649

Table 3 (a) 3.0 50

Comparison of maximum crack depth coefficient kmax under different B/H va- open crack
lues. 2.5 intact slope
40
kmax B/H = 0.8 B/H = 2 B/H = 5 2D
2.0 2D
β = 30° φ = 10° 0.2305 0.3086 0.3320 0.3477 B/H=0.8 B/H=2 30
φ = 20° 0.1836 0.2305 0.2539 0.2617
B/H=0.8

(%)
β = 45° φ = 10° 0.3555 0.4336 0.4492 0.4648 1.5

FS
φ = 20° 0.2852 0.3477 0.3633 0.3711
B/H=2
φ = 30° 0.2461 0.2930 0.3008 0.3086 20
β = 60° φ = 10° 0.5039 0.5820 0.5898 0.6055 1.0 2D
φ = 20° 0.4180 0.4883 0.5039 0.5898
φ = 30° 0.3633 0.4258 0.4414 0.4648 10
0.5

0.0 0
steep slopes is highly risky, and must be prevented. 30 40 50 60 70 80 90
The decreasing rate of FS values for slopes with smaller B/H values Slope angle (°)
is less than that for 2D slopes, indicating that the presence of natural
cracks has a slight impact on slopes with restricted widths, in ac-
(b) 3.0 50
open crack
cordance with the results for kmax shown in Table 3. Furthermore, it can
2.5 intact slope
be seen from Fig. 6 that as compared to a 3D analysis, a traditional 2D B/H=0.8 40
analysis tends to overestimate the influence of cracks on the 3D slope
stability. This is consistent with the expectation that the solutions from 2.0 2D
B/H=2 B/H=2 30
a 2D stability analysis will be more conservative.

(%)
B/H=0.8
1.5
FS

2D
20
4.3. Influence of reinforcing density of pile group 1.0

To investigate the reinforcement effects of stabilizing piles on naturally 10

0.5
fissured slopes, two illustrative examples are considered in this section.
Selected parameters for the two examples are given as follows. For case 1,
0.0 0
the slope height H = 13.7 m, and slope inclination angle β = 30°, and the 30 40 50 60 70 80 90
cohesion intercept c, internal friction angle φ, and unit weight γ of the soil
Slope angle (°)
are 23.94 kPa, 10°, and 19.63 kN/m3, respectively. For case 2, the slope
height H = 10 m and slope inclination angle β = 45°, and the soil (c) 3.0 50
properties are the same as those in case 1. open crack
The density of a pile group is defined as the ratio of the center-to- 2.5 intact slope
B/H=0.8 40
center interval to the diameter of the pile, and is denoted as D1/d.
Notably, the greater the value of D1/d, the sparser the group piles are. 2.0 2D
B/H=2
As shown in Fig. 7, FS decreases with increases of B/H and D1/d. B/H=2 30
(%)

Moreover, the reinforcement effect of the piles is more pronounced 1.5 B/H=0.8
FS

2D
when D1/d < 2.5. Therefore, it is suggested that piles placed in a row
20
should be placed close enough to form a compact pile group, so as to 1.0
achieve a better reinforcement effect.
Fig. 8 shows the maximum crack depth coefficient along with the 10
0.5
density of the pile group. It is indicated that the reinforcement from the
pile group could effectively restrict the depths of cracks, and thereby
improve the stability of fissured slopes. As combined with Fig. 7, the 0.0 0
40 50 60 70 80 90
numerical results revealed a significant negative correlation between
Slope angle (°)
the FS and kmax. In this sense, kmax could be regarded as an effective
index for evaluating the sensitivity of slopes to natural cracks. Fig. 6. Comparison of FS for intact slopes and fissured slopes with H = 10 m,
γ = 20 kN/m3, c = 20 kPa, n = 0 (a) φ = 10°; (b) φ = 20°; (c) φ = 30°.

6
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

(a) 3.0 (b) 3.0

Case 1 Case 2 B/H=0.8
B/H=0.8
2.7 B/H=2 2.7 B/H=2
B/H=5 B/H=5
2D 2D
2.4 2.4

2.1 2.1
FS

FS
1.8 1.8

1.5 1.5

1.2 1.2
1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0
D1/d D1/d

Fig. 7. FS versus D1/d with different B/H values for n = 0.7 (a) Case 1; (b) Case 2.

4.4. Optimal installation location of the pile group in fissured slopes to the numerical results, the main conclusions are drawn as follows.

Based on the illustrative examples described above, the values of FS for 1) The impact of the pre-existing crack on the slope stability is mainly
different installation locations of pile groups are presented in Fig. 9. It can dependent on the slope inclination angle. When slope inclination
be observed that the ‘end effect’ has a significant influence on fissured angle β < 40°, the stability of slopes is almost unaffected by natural
slopes, with the FS increasing by nearly 20% and 50% when the failure cracks. However, for steep slopes (β greater than 60°), the stability
mechanism changed from the 2D condition to B/H = 2 and B/H = 5, could be severely weakened owing to the presence of cracks.
respectively. The curves in Fig. 9 show evident unimodal characteristics, Therefore, the presence of natural cracks in steep slopes is highly
and the FS reaches its peak when n is approximately 0.7 or 0.8. Accord- risky and needs to be prevented.
ingly, reinforcing piles should be installed within the upper half of slopes 2) The FS of the slopes shows a significant negative correlation with
to attain their maximum effectiveness. Moreover, the curves are almost the maximum crack depth coefficient kmax. In this sense, kmax could
parallel to each other, indicating that the optimal installation location of a be regarded as an effective index for evaluating the sensitivity of
pile group is independent from the value of B/H. slopes to natural cracks.
3) For slopes with a smaller width-to-height ratio, the maximum crack
5. Conclusions depth coefficient is smaller, and the differences in the FS values
between intact slopes and fissured slopes are also smaller. This in-
As common phenomena in slope engineering, the presence of cracks dicates that under the same slope height, slopes with a smaller
and their corresponding impacts on slope stability are of vital im- width-to-height ratio are less likely to be affected by the presence of
portance in stability analysis. Considering that natural slopes are typi- cracks.
cally limited by neighboring rock formations or existing structures, the 4) The reinforcement effect of a pile group is closely related to its in-
3D stability of slopes with a vertical crack and pile reinforcements were stallation location and deployment density. It is suggested that piles
studied, based on the kinematical approach of limit analysis. An opti- placed in a row should be placed close enough to form a compact
mization program was developed to search for the most unfavorable pile group, to attain their maximum effectiveness. The optimal lo-
failure mechanism. The coupling effect of the pre-existing crack and cation for reinforcing piles is located within the upper half of slopes.
reinforcing piles was investigated using a parametric study. According

(a) 0.30 (b) 0.45

B/H=0.8 Case 1 B/H=0.8 Case 2
B/H=2 B/H=2
B/H=5 0.40 B/H=5
0.25
2D 2D

0.35
kmax
kmax

0.20
0.30

0.15
0.25

0.10 0.20
1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0
D1/d D1/d

Fig. 8. kmax versus D1/d with different B/H values for n = 0.7 (a) Case 1; (b) Case 2.

7
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

(a) 2.2 (b) 2.2

2D Case 1 2D Case 2
2.0 B/H=5 2.0 B/H=5
B/H=5
B/H=2 B/H=2
B/H=0.8 B/H=0.8
1.8 1.8

1.6 1.6

FS
FS

1.4 1.4

1.2 1.2

1.0 1.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
n n

Fig. 9. FS versus pile location coefficient n with D1/d = 2.5 (a) Case 1; (2) Case 2.

Acknowledgment For Postgraduate (2019zzts028). The financial support is greatly ap-

preciated.
The study is funded by the Hunan Provincial Innovation Foundation

Appendix

The horizontal lengths of the slope crest AB and AD are denoted by L1 and L2, respectively. According to geometric relations, H/r0, h/r0, L1/r0,
and L2/r0 can be expressed as follows.
H r0 = sin he
( h 0 ) tan sin 0 (A1)

h r0 = sin Ce
( C 0) tan sin 0 (A2)

sin( 0 + ) sin( h + )
L1 r0 = e( h 0 ) tan
sin sin (A3)

L2 r0 = cos 0 e( C 0 ) tan cos C (A4)

Substituting Eqs. (1) and (2) into Eqs. (3) and (4), the following expressions are obtained.

1 ( 0 ) tan
r0 ( 0 ) tan
f1 ( ) = e + e
2 r0 (A5)

1 ( 0 ) tan
r0 ( 0 ) tan
f2 ( ) = e e
2 r0 (A6)
WABE and WADC are integrated in a local coordinate system x-y first, as shown in Fig. 1(a), which gives their values as follows:
B x1 y h x2 y
WABE = 2 (rm + y )2cos dydxd + (rm + y ) 2cos dydxd
0 0 a B 0 d1 (A7)

D x1 y C x2 y
WADC = 2 (rm + y )2cos dydxd + (rm + y ) 2cos dydxd
0 0 a B 0 d2 (A8)

Here, x1 = R2 a2 , x2 = R2 di2 (i = 1, 2) and y = R2 x2 . The parameters a and di are given as follows:

sin 0
a= r0 rm = r0 f3 ( )
sin (A9)

sin( + h) ( 0 ) tan
d1 = r0 e h rm = r0 f4 ( )
sin( + ) (A10)

cos C ( C 0 ) tan
d2 = e rm = r0 f5 ( )
cos (A11)

sin 0 1 ( 0) tan
r0 ( 0) tan
f3 ( ) = e + e
sin 2 r0 (A12)

sin( + h ) ( 0 ) tan
1 ( 0 ) tan
r0 ( 0 ) tan
f4 ( ) = e h e + e
sin( + ) 2 r0 (A13)

cos C ( C 0 ) tan
1 ( 0 ) tan
r0 ( 0 ) tan
f5 ( ) = e e + e
cos 2 r0 (A14)

8
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

Substituting the above expressions into Eq. (A7) and Eq. (A8), the following expressions are obtained.
B
g1 ( 0 , h, B, r 0 r0 ) = 2 (f22 f3 8 f33 4 2f1 f32 3 f12 f3 2 + 2f1 f22 3) (f22 f32 )
0

+ (f 24 8 + f12 f22 2)arc cos (f3 f2 )cos d

h
+2 (f22 f4 8 f43 4 2f1 f42 3 f12 f4 2 + 2f1 f22 3) (f22 f42 )
B

+ (f 24 8 + f12 f22 2)arc cos (f4 f2 )cos d (A15)

D
g2 ( 0 , C, D, r 0 r0 ) = 2 (f22 f3 8 f33 4 2f1 f32 3 f12 f3 2 + 2f1 f22 3) (f22 f32 )
0

+ (f 24 8 + f12 f22 2)arc cos (f3 f2 )cos d

C
+2 (f22 f5 8 f53 4 2f1 f52 3 f12 f5 2 + 2f1 f22 3) (f22 f52 )
D

+ (f 24 8 + f12 f22 2)arc cos (f5 f2 )cos d (A16)

In the above, as shown in Fig. 1, angles B and D can be obtained from the geometrical relations as follows:
cos 0 L1 r0
B = arccos
2
(cos 0 L1 r0 ) 2 + sin 0 (A17)

cos C e( C 0 ) tan
D = arccos
2
cos2 (A18)
2( 0) tan
Ce + sin
C
0

According to Chen (1975), the following expressions are obtained.

1 3( h 0) tan
f6 ( 0, h) = [(3 tan cos h + sin h)e (3 tan cos 0 + sin 0 )]
3(1 + 9tan2 ) (A19)

1 L1 L1
f7 ( 0, h) = sin 0 2cos 0
6 r0 r0 (A20)

1 ( 0 ) tan
L1 L1 ( h 0 ) tan
f8 ( 0, h) = e h sin( h 0) sin h cos 0 + cos he
6 r0 r0 (A21)
The formula of g3 is expressed as follows:
b ( h 0) tan
g3 ( 0, h, b H) = (f f7 f8 )[sin he sin 0]
H 6 (A22)
Similarly, g4 is expressed as follows:
b ( C 0 ) tan
g4 ( 0 , C, b H) = (f f10 f11 )[sin Ce sin 0]
H 9 (A23)
Here, f9, f10, and f11 are expressed as follows:
1 3( C 0) tan
f9 ( 0 , C) = [(3 tan cos C + sin C )e (3 tan cos 0 + sin 0 )]
3(1 + 9tan2 ) (A24)

1 L2 L2
f10 ( 0, C) = sin 0 2cos 0
6 r0 r0 (A25)

1h
f11 ( 0, C) = cos2 Ce
2( C - 0) tan
3 r0 (A26)
The expressions g5 and g6 are obtained as follows.
D
g5 = f2 (4f1 + f3 ) f22 f32 + f2 (2f12 + f2 2 ) arc cos(f3 f2 ) d
0

h
+ f2 (4f1 + f4 ) f2 2 f4 2 + f2 (2f12 + f22 ) arc cos(f4 f2 ) d
D (A27)
B
g6 = f2 (4f1 + f3 ) f22 f32 + f2 (2f12 + f2 2 ) arc cos(f3 f2 ) d
0

C
+ f2 (4f1 + f5 ) f2 2 f5 2 + f2 (2f12 + f2 2 ) arc cos(f5 f2 ) d
B (A28)

9
X.L. Yang and S. Zhang Computers and Geotechnics 122 (2020) 103544

References [14] Michalowski RL. Stability assessment of slopes with cracks using limit analysis. Can
Geotech J 2013;50(10):1011–21.
[15] Gao Y, Song W, Zhang F, Qin H. Limit analysis of slopes with cracks: Comparisons of
[1] Tang CS, Cui YJ, Tang AM, Shi B. Experiment evidence on the temperature de- results. Eng Geol 2015;188:97–100.
pendence of desiccation cracking behavior of clayey soils. Eng Geol [16] Zeng SP, Liang R. Stability analysis of drilled shafts reinforced slope. Soils Found
2010;114(3–4):261–6. 2002;42(2):93–102.
[2] Ayalew L, Yamagishi H, Marui H, Kanno T. Landslides in Sado Island of Japan: part [17] Martin GR, Chen CY. Response of piles due to lateral slope movement. Comput
I. Case studies, monitoring techniques and environmental considerations. Eng Geol Struct 2005;83(8–9):588–98.
2005;81(4):419–31. [18] Won J, You K, Jeong S, Kim S. Coupled effects in stability analysis of pile–slope
[3] Li JH, Zhang LM, Li X. Soil-water characteristic curve and permeability function for systems. Comput Geotech 2005;32(4):304–15.
unsaturated cracked soil. Can Geotech J 2011;48(7):1010–31. [19] Ausilio E, Conte E, Dente G. Stability analysis of slopes reinforced with piles.
[4] Proceedings of the Institution of Civil Engineers - Water Management Comput Geotech 2001;28(8):591–611.
2009;162(3):221–32. https://doi.org/10.1680/wama.2009.162.3.221. [20] Li XP, He SM, Wang CH. Stability analysis of slopes reinforced with piles using limit
[5] Baker R. Tensile strength, tension cracks, and stability of slopes. Soils Found analysis method. In Advances in earth structures: research to practice 2006; pp.
1981;21(2):1–17. 105-112.
[6] Chowdhury RN, Zhang S. Tension cracks and slope failure. In Slope stability en- [21] Ito T, Matsui T. Methods to estimate lateral force acting on stabilizing piles. Soils
gineering developments and applications. Proceedings of the international con- Found 1975;15(4):43–59.
ference on slope stability organized by the Institution of Civil Engineers and held on [22] Chen WF. Limit analysis and soil plasticity. Elsevier; 1975.
the Isle of Wight on 15–18 April 1991. Thomas Telford Publishing; 1991. p. 27–32. [23] Fellenius W. Erdstatische Berechnungen. revised edition Sons, Berlin: W. W Ernst u;
[7] Kaniraj SR, Abdullah H. Effect of berms and tension crack on the stability of em- 1927.
bankments on soft soils. Soils Found 1993;33(4):99–107. [24] Terzaghi K. Theoretical Soil Mechanics. New York: John Wiley & Sons; 1943.
[8] Li TZ, Yang XL. Stability of plane strain tunnel headings in soils with tensile strength [25] Huang F, Zhang M, Wang F, Ling TH, Yang XL. The failure mechanism of sur-
cut-off. Tunn Undergr Space Technol 2020;95:103138. rounding rock around an existing shield tunnel induced by an adjacent excavation.
[9] Zhang DB, Zhang B. Stability analysis of the pressurized 3D tunnel face in aniso- Comput Geotech 2020;117:103236.
tropic and nonhomogeneous soils. Int J Geomech 2020;20(4):04020018. [26] Li LC, Tang CA, Zhu WC. Liang, ZZ. Numerical analysis of slope stability based on
[10] Li ZW, Yang XL, Li TZ. Static and seismic stability assessment of 3D slopes with the gravity increase method. Computers and Geotechnics 2009;36(7):1246-1258.
cracks. Eng Geol 2020;265:105450. [27] Huang F, Feng Y, Zhang ZQ, Yang XL, Ling TH. Upper Bound Solution of the Safety
[11] Zhong JH, Yang XL. Kinematic stability of tunnel face in non-uniform soils. KSCE J Factor for a Shield Tunnel Face Subjected to the Hoek-Brown Failure Criterion. Int J
Civ Eng 2020;24(2):670–81. Civil Eng 2019;17(12):1941–50.
[12] Utili S. Investigation by limit analysis on the stability of slopes with cracks. [28] Qin CB, Chian SC. Bearing capacity analysis of a saturated non-uniform soil slope
Geotechnique 2013;63(2):140. with discretization-based kinematic analysis. Comput Geotech 2018;96:246–57.
[13] Utili S, Abd AH. On the stability of fissured slopes subject to seismic action. Int J
Numer Anal Meth Geomech 2016;40(5):785–806.

10