See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276353358

Numerical Simulation of Seismic Slope Stability Analysis based on Tension-

Shear Failure Mechanism

Article  in  Geotechnical Engineering · June 2012

CITATIONS READS

16 732

5 authors, including:

Yingbin Zhang Lu Zheng

Southwest Jiaotong University Sichuan University
91 PUBLICATIONS   551 CITATIONS    28 PUBLICATIONS   365 CITATIONS   

SEE PROFILE SEE PROFILE

Xiaoying Zhuang
Tongji University
162 PUBLICATIONS   3,056 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

earthquake induced landslides View project

Excellence Cluster - PhoenixD View project

All content following this page was uploaded by Yingbin Zhang on 30 August 2015.

The user has requested enhancement of the downloaded file.

 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Numerical Simulation of Seismic Slope Stability Analysis Based on Tension-Shear Failure

Mechanism

Yingbin Zhang1, Guangqi Chen1*, Jian Wu1, Lu Zheng1, Xiaoying Zhuang2

1
Department of Civil and Structure Engineering, Kyushu University, Fukuoka, JAPAN
2
Department of Geotechnical Engineering, Tongji University, Shanghai, CHINA
E-mail address: [email protected], [email protected] ( *Corresponding author)

ABSTRACT: Most slope stability analysis approaches regard the failure mechanism of seismic slope as single shear failure while ignoring
the influence of tension failure. However, many model testes and a large number of post-earthquake investigations provided supporting
evidence of the significant influence of tension failure in seismic slope instability. To estimate the effects of tension failure on seismic slope
stability, a numerical modeling considering both shear and tension failure is performed using FLAC3D. After discussions of failure
mechanism, strength reduction techniques and the definition of slope failure, a homogeneity slope under a modified transverse earthquake
load is analyzed. The results obtained from the simulation are presented in terms of permanent displacement, factor of safety and failure
surface. Finally, the outcomes compared with those from various existing methods. The results show that the influence of tension failure is
significant and consideration of it is necessary.
KEYWORDS: Seismic slope stability analysis, Numerical simulation, Failure mechanism, Tension failure, Factor of safety.

1. INTRODUCTION the center of gravity of each slice, respectively, and the results of
factor of safety were 1.21 and 1.32, respectively. Pseudo-static
Landslide is one of the worst natural disasters. As one of the method can be simply and directly used to identify the factor of
main trigger, the earthquake can easily induce collapse of slopes safety F and the critical seismic coefficient kc but it cannot simulate
and produce landslides that can result in serious damage to life and the transient dynamic effects of earthquake shaking, because it
property. For example, the Wenchuan earthquake (Ms=8.0) assumes a constant unidirectional pseudo-static acceleration. In
occurred in Sichuan Province, China at 14:28 CST on 12 May 2008 addition, performance of slope is closely related to permanent
induced as many as 56,000 landslides [1]. It is estimated that over displacement, but the results of pseudo-static method are difficult to
one third of the total lost from the 2008 Wenchuan earthquake was interpret the performance of slope after a seismic event because this
caused by the earthquake induced landslides. Hence, the evaluation method provides no information about permanent displacement.
of stability of seismic slope is the most important aspect of
geotechnical earthquake engineering, especially when the slopes are Table 1 Pseudo-static coefficient from several studies
situated close to residential areas.
There are at least four different formulations of the seismic Recommended Factor Permanent
Original
slope stability analysis problem: in terms of (1) the factor of safety pseudo-static of displacement Ref.
application
F, (2) the critical seismic coefficient kc, (3) the permanent coefficient (k/g) safety (D/m)
displacement D, and (4) the shape of slip surface. In order to solve 0.1 (R-F=IX) Nature or
these problems, several methods can be used in evaluating the 0.2 (R-F=X) F>1.0 - man-made [3]
stability of slopes subject to earthquake loading. These methods can 0.5 (R-F=XI) slope
fall into three general categories: (1) pseudo-static method, (2) 0.1 (M=6.5) F>1.1
<1 Earth dams [4]
permanent displacement method, and (3) stress-strain method. Each 0.15 (M=8.25) 5
of these types of analysis has strengths and weaknesses, and each (1/3~1/2)PGA F>1.0 - - [5]
can be appropriately applied in different situations [2]. These 1/2 PGA F>1.0 <1 Earth dams [6]
methods are briefly retrospective as follow: 1/2 PGA
The majority of seismic slope stability analyses performed in (M=8.25)
practice still use pseudo-static approaches where the effect of 1/3 PGA
earthquake on a potential failure soil mass is represented in an (M=7.5)
approximate manner by a static force. As we know, magnitude, F>1.0 - - [7]
1/4 PGA
direction and point of application are three elements of a force. In (M=7.0)
generally, the pseudo-static force equal to the product of the 1/5 PGA
gravitational force and a coefficient, k, the pseudo-static seismic (M=6.5)
coefficient and act on the center of gravity of a potential sliding
0.15 F>1.1 - Dams [8]
mass or of a typical vertical slice in various slice method at
Soild-
horizontal direction. Pseudo-static has been studied by many (0.6~0.75)
F>1.0 <0.15~0.3 waste [9]
researchers in the aspects of magnitude, direction and point of PGArock
landslides
application of pseudo-static force, respectively. Firstly, because
(0.25~0.75) Urbanized
magnitude of the pseudo-static force on a mass of weight W is kW, F>1.0 <0.05~0.15 [10]
PGArock slopes
selection of the pseudo-static coefficient k is the most important
aspect of pseudo-static method. Some researcher proposed many Note: R-F is Rossi-Forel earthquake intensity scale, IX: severe
useful studies. Table 1 lists several recommendations for selecting a earthquake, X destructive earthquake, XI catastrophic
pseudo-static coefficient. Secondly, analyses performed by several earthquake; M is earthquake magnitude; PGA is peak
investigators with an inclined seismic force (i.e. coupled with ground acceleration, in terms of acceleration of gravity.
vertical component of the earthquake force) have shown that the
inclination can have a significant influence on the seismic slope Permanent displacement is a useful index of slope performance,
stability analysis [11, 12]. Thirdly, different points of application of especially for those man-made slopes constructed for special
pseudo-static force can induce a significant difference in the result. purposes like dams, embankments, etc. Newmark’s method
Seed [4] provided a well-known example, the analysis of Sheffield provides a measure of permanent displacement along a fixed slip
Dam. In his study the seismic forces were applied at the base and surface based on a double integration of acceleration of a

18
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

potentially sliding soil block with simple input parameters and description for tension segment of slope failure surface based on
relatively low computational complexity. In fact, the Newmark’s post-earthquake investigations. Zheng et al. [21], Zhang et al. [28]
method also involves some assumptions about the manner in which and Yan et al. [29] have certificated the existing of tension failure
the earthquake loading. Considerable attention has been focused zone by analyzing the mechanism of seismic slope using numerical
over the last decades on developing procedures to more accurately simulation. Zhang et al. [30] showed that significant effect of
analyze the seismic performance of a slope for dams, embankments tension failure on slope stability analysis using upper bound limit
or other important structures by modeling the dynamic slope analysis. These studies, however, most focus on the description or
response more rigorously. Jibson [2] grouped analytical procedures explanation of phenomenon, but few on deep research of stability
for estimating permanent co-seismic landslide displacements into analysis subjected to the tension failure.
three types: (1) rigid-block [13], (2) decoupled [14, 15], and (3) Hence, to investigate how the tension failure effect on seismic
coupled [16-19] (see figure 1). stability analysis, a full dynamic analysis is carried out in this paper
with an emphasis on seismic slope stability using finite difference
method through a homogeneity slope.

(a)Steep scarp with coarse cracks of Donghekou landslide,

Qingchuan

Figure 1 Calculated models of permanent displacement: (a) rigid

block model, (b) decoupled model, and (c) coupled model.
Stress-strain method represents a powerful alternative
approach for seismic slope stability analysis which is accurate,
versatile and requires fewer a priori assumptions, especially,
regarding the shape of failure surface. With the developments of
computer technology and simulation approach in recent decades,
the numerical simulation method is becoming used increasingly in (b)Steep scarp with coarse cracks of Pingxicun landslide, Pingwu
engineering practice and more and more popular for the real
dynamic analysis. The most commonly applied numerical methods
can be categorized into continuous and discontinuous approach.
This paper focus on homogeneity slope stability analysis, i.e.
continuous slope material is assumed. Stress-strain methods have
been used successful in seismic slope stability analysis by earlier
researchers, e.g. finite element method (FEM) [20], finite difference
method (FDM) [21, 22].
However, almost all of the above methods consider the failure
mechanism of seismic slope as a completely shear failure. The
frequently occurring earthquake events in recent years in New
Zealand, Japan and China have led to a renewed knowledge in
mechanism of instability of slopes. A large number of evidences
from investigations of earthquake induced landslide and shaking (c)Tension cracks in the top of Shiziliang
table test show that tension failures appear in top of almost all
landslides or potential sliding slopes. For example, figure 2 shows Figure 2 Evidences of the existence of tension failure from
steep scarp with coarse crack of Donghekou landslide and investigation of Wenchuan earthquake induced landslides or
Pingxicun landslide induced by Wenchuan earthquake occurred in potential landslide. Steep scarp with coarse cracks of (a)
2008 and figure 3 shows a result of shaking table test carried out by Donghekou landslide, Qingchuan and (b) Pingxicun landslide,
Wakai et al. [23] in which tension crack can be obviously found in Pingwu and (c) tension cracks in the top of Shiziliang landslide.
top of slope. The existing methods what usually only based on shear (Adopted from reference [26])
failure mechanism while ignoring the tension failure may lead to
inaccuracy especially when the slope shook by a strong earthquake
load. There are a large number of described, analytical and
numerical studies which have provided supporting evidences of the
existence of tension failure in slope stability analysis. Huang et al.
[24, 25], Xu et al. [26] and Yin et al. [27] have gave much detailed

19
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Figure 3 Supporting evidence of the existence of tension failure

from shaking test (Adopted from reference [23])
2. IMPORTANT ASPECTS IN SEISEMIC SLOPE
Figure 4 Stress state of a point in slope mass at conditions of (a)
STABILITY ANALYSIS
static and (b) horizontal seismic shear stress loadings.
In this section, three major aspects that influence seismic slope
Many failure criterions have been presented for modeling the
stability analysis are discussed. The first is about the failure
strength of soil, the Mohr-Coulomb criterion remains the one most
mechanism of seismic slope. The second is the concept of factor of
widely used in geotechnical practice. A modified Mohr-Coulomb
safety and the third is the definition of the slope failure.
failure criterion is used in this study. The representation of the
2.1 Failure mechanism subjected to tension failure failure criterion in the (σ, τ) plane is sketched in figure 5. The
At present, the failure mechanism of seismic slope follows the failure envelope is defined from point A to B by the Mohr-Coulomb
static slope failure mechanism, i.e. the main reason that caused yield function
 '   3'  '   3'
seismic slope instable is shear failure while ignoring the influence f shear  1  c cos   1 sin   0 (6)
of tension failure. In fact, with the reason of small tensile strength 2 2
and the action of earthquake loading, the slope in reciprocating and from B to C by the tension yield function
motion is more easily to be tensioned. Stress states of a point in ftension   3'- t  0 (7)
slope mass at two situations, static and dynamic, are illustrated as
where c is cohesion, φ is friction angle of soil material, σt is tension
figure 4. As the description of most soil mechanics textbooks,
strength of soil material.
compressive stresses are considered positive, and tensile stresses are
negative. The major and minor principal stresses σ1 and σ3 of static
state obtained from
x  z x  z 2 2
1   ( )   xz (1)
2 2

x  z x  z 2 2
3   ( )   xz (2)
2 2
Where σx σz are stress in horizontal and vertical direction
respectively; and τxz is shear stress. Figure 5 Failure criterion combined tension-shear failure
When the slope is effecting by a horizontal earthquake loading mechanism.
what propagate from bottom upward to top of the slope. A shear
stress τs caused by the earthquake loading can added into the Because tensile strength of soil material is much smaller than
existing state stress state. The magnitude and direction of dynamic their shear strength, we can take the opinion that once the minor
stress τs are time dependent. If not consider the influence of wave principal stress σ3 equals to tensile strength σt, the tension failure
reflection and refraction, the stress state in this situation can be will occur and the tension failure surface parallel to minor principal
simply indicated as figure 4(b). The combined major and minor stress plane.
principal stresses σ1' σ3' calculated from The failure functions ftension and fshear can be interpreted as
follows
x  z x  z 2 2 ● ftension >0 and fshear <0 note that the normal and shear stresses
1  '  ( )  ( xz   s ) (3) on a plane in a soil mass inside failure envelope, green zone in
2 2 figure 5. Failure will not occur in this situation;
x  z x  z ● ftension =0 or fshear =0 notes that the normal and shear stresses
2 2
3  '  ( )  ( xz   s ) (4) on a plane in a soil mass on failure envelope, yellow line in figure 5.
2 2 The material is yielding state in this situation.
where τs obtained from: ● ftension <0 or fshear >0 notes a state of stress plotting as red
zone in figure 5 that cannot exist, since the stresses outside failure
 s  Cs vs (5)
envelope, failure would have occurred before this condition was
where ρ is mass density of soil material, Cs is speed of s-wave reached.
propagation through material, and vs is input shear particle velocity For a slope stability analysis problem, tensile failure should be
caused by earthquake loading in horizontal direction. Equations (3) first considered, because tension strength of soil material is usually
and (4) show an opposite change pattern of major and minor much smaller than shear strength. If take cut-through of the tension
principal stresses σ1' σ3'. In the (σ, τ) plane, this change pattern failure zone and shear failure surface as definition of slope failure,
presented as expand or narrow of the Mohr circle (see figure 5). compute process can be illustrated as figure 6. Definition of slope
Note that if the influence of wave reflection and refraction are failure is discussed in the third subsection. Note that the procedure
considered, the expressions of major and minor principal stressesσ1' of tension failure and shear failure automatically included in FLAC
σ3' will be very complex. This point will be detail description by program while the global failure need own judgment based on
monitoring records through the time history of seismic excitations. different definitions.

20
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Note that, permanent displacement in these definitions of

failure notes residual displacement after an earthquake event. In the
example studied in here, the first definition of failure, cut-through
of plastic zone from bottom to top surface of a slope, is taken as
being key indicator of seismic slope failure. In addition, comparison
of slope failure definitions on seismic slope stability analysis is
presented in section 5.

3. DYANAMIC FORMULATION

Figure 6 Flow chart of seismic slope stability analysis based on 3.1 Brief description of dynamic analysis using FLAC3D
tension-shear failure mechanism. The program used in this paper is finite difference program
FlAC3D version 3.10 [38]. The calculation is based on the explicit
2.2 Factor of safety F and strength reduction factor SRF finite difference scheme to solve the full equations of motion, using
Factor of safety F is a value that is used to examine the lumped grid point masses derived from the real density of
stability state of slopes and of great interest for engineering practice. surrounding zones rather than fictitious masses used for optimum
A generally accepted definition of F is that first given by Bishop convergence in the static solution scheme. And a flow chart of
[31]: the ratio of the available strength of soil material to that dynamic analysis for a slope is illustrated in figure 8.
required to maintain equilibrium. This definition used by In the whole process, there are several important aspects
Zienkiewicz [32], Ugai [20], Matsui & San [33], Griffiths & Lane should be considered while preparing a FLAC model for dynamic
[34], Dawson & Roth [35], Zheng et al. [21] and others. For c-φ-σt analysis: in terms of (1) dynamic loading, (2) boundary conditions,
material studied in this paper, factor of safety F against slope is (3) mechanical damping, and (4) wave transmission through the
simply calculated as model. They are discussed in this section.
c tan  
F   t (8)
c ' tan  '  t '
where c, φ and σt are the actual values of soil material. They are
cohesion, friction angle and tensile strength, respectively, and c', φ'
and σt' are the relevant parameters required to maintain the limit
equilibrium. To achieve the correct factor of safety F, it is essential
to trace the strength parameters by a coefficient called strength
reduction factor SRF until the limit state of slope is achieved. The
flow chart of calculation of factor of safety is shown in figure 7.
Strength reduction factor SRF defined as
c tan  t
SRF    (9)
cm tan  m  tm
where cm, φm and σtm are the calculated values of strength.

Figure 7 Flow chart of calculation of factor of safety.

It should be note that factor of safety F and strength reduction
factor SRF are two different concepts. Only when the slope is in
limit equilibrium state, the value of SRF equals to factor of safety F.
2.3 Definition of slope failure
When strength reduction technique is used to analyze slope
stability, one typical problem is how to define the state of slope
failure. At present, in static situation, there are several possible
definitions of slope failure in numerical simulation, e.g. cut-through
of plastic zone from toe to top of slope, non-convergence of the
solution [36], mutation of displacement of potential failure mass [37]
and so on. In static case, similar outcomes can be obtained based on
these definitions of slope failure. Although these definitions of
failure have been widely used in numerical simulation, it is difficult
to implement with the current dynamic simulation. In dynamic
numerical simulation, few studies focus on definition of slope
failure was carried out. From the early researchers, three possible
definitions of slope failure can be used in a dynamic analysis: Figure 8 Flow chart of seismic slope stability analysis using FLAC
(1) Cut-through of plastic zone from bottom to top surface of a program (Modified from [39]).
slope,
(2) Non-convergence of permanent displacement of potential 3.2 Modeling with FLAC3D
failure mass, and This paper analyzes a classical homogeneity slope with the
(3) Mutation of permanent displacement of potential failure height of 20m and the incline angle of 45°that has studied by some
mass. earlier researchers [21, 28]. A schematic illustration of the 2D
analyzed mesh and the boundary conditions is provided in figure 9.
21
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

In order to study the development process of slope failure surface, simple and pragmatic method. The local damping coefficient αL
one square meters of grid size is meshed. Both x and y defined as
displacements are fixed at the base of the model. And x αL=πD (10)
displacements are fixed on either side of the model along the y-axis. where D is fraction of critical damping. Although the actual value
The slope is allowed free to move in both the directions. Free field given to the local damping has a profound influence on the dynamic
boundary is used in the present model to minimize the wave wave transmission, if it chooses from a certain range, it has little
reflection. This boundary condition provides a better representation influence on the predicted factor of safety in seismic slope stability
when quiet boundaries are used in conjunction with external seismic analysis. Hence, local damping of 0.157 (i.e. fraction of critical
sources. The size of slope ensures the assumption of free field damping is 5%) is used in the model as suggested by other studies
boundary. For a dynamic analysis, FLAC3D version 3.10 program for these kinds of problems.
provides several mechanical damping in which local damping is a

Figure 9 Mesh generation and boundary conditions of finite difference model for dynamic slope stability analysis.

Table 2 Property parameters of slope material

Parameters Value
Modulus of elasticity E 77.48×106 Pa
Poisson's ratio μ 0.3
Density ρ 2000 kg/m3
Cohesion c 40000 Pa
Inter friction angle φ 20°
Figure 10 Acceleration time history of earthquake loading applied
Dilation angle ψ 20°
in study.
Tensile strength σt 0
3.3.2 Baseline correction
Table 2 shows the properties parameters of soil material used If the acceleration record shown in figure 10 is directly used as
in the analysis. It should be note that some unlisted parameters e.g. an earthquake loading time history, the FLAC3D model may exhibit
unit weight γ, modulus of volume K and shear modulus G, are also continuing velocity or residual displacements after the motion has
used in this paper. They can be translated from the parameters listed finished as shown in figure 11(a) and figure 11(b). This arises from
in table 2. The density ρ assigned to the soil decides the total unit the fact that the integral of the complete acceleration time history
weight γ and to move forward the self-gravity load. The parameters may not be zero. In this paper, a low frequency velocity wave
c and φ note the effective cohesion and inter friction angle of the which, when added to the original history, produces a final
soil material. The dilation angle ψ affects the volume change of the displacement which is zero is used. The action will not affect the
soil during yielding. In this study, we take ψ=φ, i.e. the plasticity mechanics of the deformation of the model. The process of baseline
flow rule is associated and direct comparisons with theorems from correction and corrected displacement time histories are shown in
classical plasticity can be made. In addition, it should be note that figure 11.
the tensile strength σt is taken as 0, because it is so unreliable that
can be ignored.
3.3 Earthquake loading
3.3.1 Original earthquake loading
The dynamic load applied in here is the transverse component
of the acceleration time history (record name: KJM-0º) modified
from the Kobe earthquake, occurred in Japan, 1997 as shown in
figure 10. The total duration of the earthquake loading is 15s with a
time step of 0.02s. The amax value of the recorded earthquake is
0.2046g at time of 3.52s. The earthquake loading is applied at
bottom of the whole stratum. From the acceleration time history
record, velocity and displacement time histories can be computed
by once and twice integration respectively.

22
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Figure 12 Input stress time history shifted from velocity time

history. (a) Velocity time history shifted during single integration
from original seismic accelerations; (b) Shifted stress time history
from velocity time history.
4. RESULTS
In this section, three major results of seismic slope stability
analysis are presented. The first is the validation of stress state and
relationship with tension failure. The second is about the permanent
displacement in actual strength case. The third is the factor of safety
under different failure mechanism and the forth aspect is the shape
of slip surface.
4.1 Stress and tension failure
Figure 13 shows the major and minor prince stresses records of
a monitoring point set away from the slope surface and with a
certain depth from top surface in where the influence of wave
reflection and refraction is small. From the figure, an opposite
Figure 11 The baseline correction process change pattern of major and minor principal stresses σ1' σ3' can be
3.3.3 Translation of stress time history found that consist with the description of section 2.1.
In order to avoid the effect of the quiet boundary be nullified, a
stress boundary condition is used. This stress history is transformed
from a velocity record using the formula
 s  2(  Cs )vs (11)
where Cs is the speed of s-wave propagation through medium; vs is
input shear particle velocity time history of earthquake loading. The
factor of two in the equation accounts for the fact that the applied
stress must be doubled to overcome the effect of the viscous
boundary. Cs is given by
Cs  G/ (12)
where G is the shear modulus of material and given by
Figure 13 Prince stresses records of a point in slope under an
G  E /  2(1   )  (13) earthquake loading.
Figure 12 shows the input stress time history shifted from
velocity time history. As previous yield criterion, tension-shear failure criterion,
tension failure firstly judged based on the minor prince stress. The
depth of tension failure zone can be judged. Figure 14 shows a
serious minor prince stresses records of points that have different
depth from top of slope. As the tension failure criterion, once the
minor prince stress smaller than tensile strength of soil, tension
failure will occur. In figure 14, black dash line notes tensile strength
of soil material, stress records of points in depth of 1~5 m outstrip
the tensile strength line. So we can take the conclusion that depth of
tension failure zone is 5~7m. Figure 15 shows the block tension
state in which 5~7m depth of tension failure zone can be found.

23
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Figure 14 Minor prince stress records of monitoring points.

Figure 17 Horizontal displacement time histories of monitoring

points.

Figure 15 Tension state of slope after the shaking.

4.2 Permanent displacement
The contour of permanent displacement of the slope after the
complete dynamic event is shown in figure 16. It can be found that
a maximum displacement (Dmax) of 0.2896m is observed near the
toe. This displacement is the accumulated permanent displacement
due to earthquake. Figure 16 also depicts that the displacements are Figure 18 Vertical displacement time history of monitoring points.
nearer to the toe of the slope.
4.3 Factor of safety F based on tension-shear failure mechanism
As previous mentioned, factor of safety F can be computed by
reducing the soil strength in stages to bring the slope into a state of
limiting equilibrium. The stability analysis of slope is carried out
based on tension-shear failure mechanism.
Take the cut-through of tension failure zone and shear failure
surface as definition of seismic slope failure, the failure surface
contains tension and shear segment. Tension segment depended on
the depth of tension failure zone, and shear segment decided by the
Figure 16 Contour of permanent displacement of the slope after the shear strain increment, i.e. cumulant of shear strain. After a series of
complete seismic event. trying calculation, it can be obtained that cut-through of tension
plastic zone and shear failure zone occurred in the smallest SRF of
In addition, the displacements along the slope face are
0.99. The figure 19 shows the contours of tension plastic zone and
captured using the history command in FLAC library. Typical
shear strain increment at situations of SRF=0.98 and SRF=0.99. The
variations of horizontal and vertical displacements with time at
maximum values of shear strain increment (SSImax) are 0.03219 and
various monitoring points (see figure 9) during the earthquake are
0.03593, respectively. It can be found that the significant shear
shown in figure 17 and figure 18, respectively. It can be seen from
strain increment not increases to the tension zone at SRF=0.98,
figure 10 that the horizontal permanent displacements are about
while cut-through of tension failure and shear failure is obtained at
0.33m, 0.20m, 0.12m and 0.00m at points P1, P2, P3 and P4,
SRF=0.99. Hence, we can take the opinion that the slope is in the
respectively and with little change after the motion. It can be seen
limit state at SRF=0.98. As the previous definition of factor of
from figure 11 that the vertical permanent displacements are about
safety, it is 0.98 based on the tension-shear failure mechanism.
0.33m, 0.20m, 0.12m and 0.00m at points P1, P2, P3 and P4,
respectively and with continue increase after the motion expect the
point P4.

24
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Figure 19 Block tension state and contour of shear strain increment at SRF=0.98 and SRF=0.99.

5. COMPARISONS AND DISCUSSIONS displacement also be proposed, i.e. the slope is not stable at
SRF=1.11. In opposite, figure 23 and 24 show the velocity that
5.1 Traditional failure mechanism: completely shear failure equals to 0 and convergence of permanent displacement,
If only consider the shear failure mechanism, failure surface of respectively, i.e. the slope is stable at SRF=0.98. These results
slope is just induced by the shear failure. Take the cut-through of confirm the correctness and reasonable of tension-shear failure
shear plastic zone as definition of seismic slope failure as well, after mechanism.
a series of trying calculation, it can be obtained that cut-through of
shear plastic zone occurred in the smallest SRF of 1.12. The figure 5.2 Definition of seismic slope failure
20 shows the contours of shear strain increment at SRF=1.11 and As previous description, there are three definitions of slope
SRF=1.12, respectively. The maximum values of shear strain failure can be considered in a dynamic analysis. The previous
increment (SSImax) are 0.07264 and 0.07963 at SRF=1.11 and analysis mainly used the first definition. Some researcher have used
SRF=1.12, respectively. These results show that the slope is in the the other two definitions for seismic slope stability analysis and
limit state at the situation of SRF=1.11. As the previous definition obtained some useful results, but research of definition of seismic
of factor of safety and the cut-through definition of slope failure, slope failure is still in primary stage and much research should be
we can come to the conclusion that factor of safety F is 1.11 based done to certify the correctness and applicability of every definitions.
on single shear failure mechanism. Based on the example studied in here, primary discussion is given
below.

Figure 21 Horizontal velocity records of monitoring points at slope

Figure 20 Contours of shear strain increment at (a) SRF=1.11 and (b) surface at SRF=1.11.
SRF=1.12.
The factor of safety 1.11 calculated based on traditional failure
mechanism is larger than that 0.98 based on tension-shear failure
mechanism. For the studied slope in here, factor of safety reduce
about 12% from traditional failure mechanism to new failure
mechanism, i.e. tension-shear failure mechanism.
According to the previous analysis, different factor of safety F
are obtained from different failure mechanism. In order to provide
evidence for the correctness of tension-shear failure mechanism, the
second definition of slope failure, non-convergence of permanent
displacement of potential failure mass, is used in here. Permanent
displacements of potential failure mass are recorded by set some
key monitoring points alone the surface of slope (see figure 9).
Horizontal velocity and displacement time histories of three Figure 22 Horizontal displacement records of monitoring points at
monitoring points at SRF=1.11 and SRF=0.98 are proposed in slope surface at SRF=1.11.
figure 21~24. From figure 21, we can found that the residual
velocity of monitoring points after the shaking is not equals zero at
SRF=1.11, and from figure 22, non-convergence of permanent
25
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

of safety should be 1.28. Compare to the results from other reliable

methods, we can clearly found that the mutation definition of
seismic slope failure is not reliable in some situation, e.g. the slope
studied in here.

Figure 23 Horizontal velocity records of monitoring points at slope

surface at SRF=0.98.

Figure 26 Horizontal permanent displacement of monitoring points

at slope surface at SRF=0.90~1.30.
Based on the above discussion, the authors recommend the
first two definitions of slope failure and take the influence of
tension failure.
5.3 Shape of slip surface
The shape of slip surface is an important aspect of seismic
slope stability analysis. Because it has a close relationship with
volume of slip mass that plays a determination role on the starting
Figure 24 Horizontal displacement records of monitoring points at
velocity and travel distance once slope collapse occurred and
slope surface at SRF=0.98.
formed a landslide. The figure 27 shows the failure surfaces of
5.2.1 Non-convergence of permanent displacement of potential slope in different cases: static case, dynamic cases based on shear
failure mass mechanism and tension-shear mechanism. From the figure 27, we
Figure 25 shows the horizontal displacement time histories of can found that in dynamic cases, shapes of slip surface are
key point P1 under different strength reduction factor SRF. The shallower than that under static case (dense green dotted line). More
permanent displacements after motion at SRF=0.90~1.30 show a important, the failure surface under tension-shear failure
change of from convergence to non-convergence. If take the non- mechanism (blue solid line) contains two segments: tension failure
convergence of permanent displacement as definition of seismic and shear failure which consist with the result of model test [23, 40]
slope failure, from the figure 25, we can found that factor of safety and post-earthquake investigation [24, 26].
F of the studied slope should be a value between 0.95~1.02 that
consist with the result of cut-thought definition of slope failure
consider tension-shear failure mechanism. Other monitoring points,
e.g. P2, P3, have the same change trend.

Figure 27 Failure surfaces under static and dynamic situations based

on different failure mechanisms.
5.4 Comparison of factor of safety
As previous description, the calculation of factor of safety F is
one of the most important aspects of seismic slope stability analysis.
The table 3 shows the factors of safety F from various methods.
From the table 3, it can be found that the result presented in this
paper similar with those obtained from pseudo-static method. The
results based tension-shear failure mechanism is reasonable and
Figure 25 Horizontal displacement records of monitoring point at smaller than that obtained from traditional single shear failure
toe of slope at SRF=0.90~1.30. mechanism.

5.2.2 Mutation of permanent displacement of potential failure 5.5 Progressive failure

mass The failure progress of slope has an important significance for
If take mutation of permanent displacement of potential failure design of reinforce slope against earthquake loading. Considering
mass as the definition of seismic slope failure, only need to record the tension-shear failure mechanism, the figure 28 shows the failure
the residual displacements of monitoring points after the seismic progress of seismic slope at different dynamic time. From the figure
event. Figure 26 shows the horizontal residual displacement of 28, it can be obviously found that the instability of slope is a
monitoring points at different strength reduction factor SRF. From progressive failure in which the shear failure zone is expand slowly
the figure 26, it can be found that the residual displacement increase from the toe upward to the top of slope and achieve a cut-through
exponentially with the increase of SRF and mutate at with tension failure zone. In this example, the slope is not collapse
SRF=1.28~1.30. As the third definition of slope failure, the factor at dynamic time t=3s at which the first peak of wave of the
26
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

earthquake loading go through. After the largest peak of wave method a Sc 0.96 **
shaking at time of 6s, the slope have a significant performance but TS 1.05 **
k=1/2PGA
the cut-through of tension zone and shear failure zone is not S 1.10 **
achieved. The global instability occurred at time t=9s at which the TS 0.98
largely peak of wave have went through. From the figure of tension Cut-through d
Present S 1.11
state of block, we can found that the tension failure occurred at time method Non-convergence e 0.95~1.02
of largest peak of wave go through. It should be noted that the result Mutation f 1.28
presented here is just a tentative results. The research of tension * The result is calculated using limit analysis method
failure mechanism of seismic slope is at the primary stage and there ** The results are calculated using limit analysis method presented
is still plenty work to do. in reference [29]
a
k is pseudo-static seismic coefficient, PGA is peak ground
Table 3 Factors of safety F calculated from various methods. acceleration
b
Method F TS notes tension-shear failure mechanism
c
Static situation 1.27 * S notes shear failure mechanism
def
Pseudo-static k=PGA a TS b 0.93 ** Note three definitions of slope failure presented in section 2

Figure 28 Process of seismic slope failure.

6. CONCLUSIONS 3) The shapes of failure surface of seismic slope under

traditional single shear failure mechanism and tension-shear failure
This paper analyzed the stability of seismic slope using the mechanism have an obvious difference. Failure surface of seismic
finite difference program FLAC3D. Based on the analysis above, the slope considering tension-shear failure mechanism contains two
following main conclusions are drawn: segments and shallower than that just considering the shear failure
1) Tension failure has a significant influence on seismic slope while ignoring the influence of tension failure.
stability analysis. The FLAC3D program can take the influence of 4) The failure of seismic slope is progressive with the
tension failure into consideration, so as to make the analysis proceeding of earthquake. The instability of slope is a progressive
technique more reasonable for practical application. It can be used failure in which the shear failure zone is expand slowly from the toe
to determine reasonable factor of safety F which smaller than that upward to the top of slope and achieve a cut-through with tension
obtained from traditional single shear failure mechanism. The failure zone.
widespread use of this mechanism should now be seriously It should be note that seismic slope stability analysis is a
considered to traditional failure mechanism. complex problem, especially about the seismic slope failure
2) Definition of failure is an important aspect in seismic slope mechanism. The conclusions listed in here are based on limited
stability analysis. As the example studied in here, non-convergence observations. More in-depth and more extensive research should
of permanent displacement of potential failure mass is a reliable continue to be studied in future.
definition of seismic slope failure. If take the influence of tension
failure into consideration, the first definition of failure, cut-through
7. ACKNOWLEDGMENTS
of plastic zone from bottom to top surface of a slope under tension-
shear failure mechanism, gives results very similar to the second This study has received financial support from the Global
definition of failure, non-convergence. In addition, the third Environment Research Found of Japan (S-8), and from Grants-in-
definition of slope failure can resulted in unreasonable outcome. Aid for Scientific Research (Scientific Research (B), 22310113, G.
27
 Geotechnical Engineering Journal of the SEAGS & AGSSEA Vol. 43 No.2 June 2012 ISSN 0046-5828

Chen) from Japan Society for the Promotion of Science. And the [20] Ugai, K., A method of calculation of total safety factor of
first author acknowledges the support of China Scholarship Council. slopes by elasto-plastic FEM, Soils Found, 29 (1989) 190-195.
These financial supports are gratefully acknowledged. (in Japanese)
[21] Zheng, Y.R., Ye, H.L., Huang, R.Q., Li, A.H., Xu, J.B., Study
8. REFERENCES on the seismic stability analysis of a slope, Journal of
Earthquake Engineering and Engineering Vibration, 30 (2010)
[1] Dai, F., Xu, C., Yao, X., Xu, L., Tu, X., Gong, Q., Spatial 173-180. (in Chinese with English abstract)
distribution of landslides triggered by the 2008 Ms 8.0 [22] Latha, G.M., Garaga, A., Seismic stability analysis of a
Wenchuan earthquake, China, J Asian Earth Sci, 40 (2011) Himalayan rock slope, Rock Mech Rock Eng, 43 (2010) 831-
883–895. 843.
[2] Jibson, R.W., Methods for assessing the stability of slopes [23] Wakai A, Ugai K, Sato M, and Tazo T. “Numerical analysis
during earthquakes—A retrospective, Eng Geol, 122 (2011) 43– for seismic behavior of a slope based on a simple cyclic loading
50. model,” 4th International Conference on Recent Advances in
[3] Terzaghi, K., From Theory to Practice in Soil Mechanics: Geotechnical Earthquake Engineering and Soil Dynamics, Paper
selections from the writings of Karl Terzaghi, with bibliography No. 5.06 in CD-ROM, San Diego, 2001.
and contributions on his life and achievents, L. Bjerrum, et al. [24] Huang, R.Q., et al., Geohazard Assessment of the Wenchuan
(Eds.), New York, John Wiley and Sons, 1960, pp. 202-245. Earthquake, Science Press, Beijing, China, 2009. (in Chinese)
[4] Seed, R.B., Considerations in earthquake-resistant design of [25] Huang, R.Q., Xu, Q., Huo, J., Mechanism and geo-mechanics
earth and rockfill dams, Nineteeth Rankin Lecture, Geotecnique, models of landslides triggered by 5.12 Wenchuan earthquake, J
29 (1979) 13-41. Mountain Sci, 8 (2011) 200-210.
[5] Marcuson, W.F., Moderator's report for session on Earth Dams [26] Xu, Q., Pei, X.J., Huang, R.Q., et al. Large-scale landslides
and Stability of Slopes under Dynamic Loads. Proceedings, induced by the Wenchuan earthquake, Science Press, ISBN
International Conference on Recent Advances in Geotechnical 978-7-03-026906-5, Beijing, China, 2009. (in Chinese)
Earthquake Engineering and Soil Dynamics, St. Louis, MO, 3, [27] Yin, Y.P., Wang, F.W., Sun, P., Landslide hazards triggered by
1981, 1175. the 2008 Wenchuan earthquake, Sichuan, China, Landslides, 6
[6] Hyness-Griffin, M.E., Franklin, A.G., Rationalizing the seismic (2009) 139-152.
coefficient method, Miscellaneous Paper GL-84-13, U.S. Army [28] Zhang, Y.B., Chen, G.Q., Zen, K., Kasama, K., Wu, J., Zheng,
Corps of Engineers Waterways Experimental Station, Vicksburg, L., Seismic slope stability analysis subjected to tension failure,
Mississippi, 1984, 21. Western Regional Division Report of Natural Disaster Research
[7] Pyke, R.M., Selection of seismic coefficients for use in Council, No. 36 (2012) 21-24.
pseudostatic slope stability analyses, [29] Yan, Z.X., Zhang, S., Zhang, X.D., Duan, J., Failure
http://www.tagasoft.com/TAGAsoft/TAGAsoft/Discussion/artic mechanism and stability analysis of slope under earthquake, J
le2_html, 1991. Eng Geol, 18 (2010) 844-849. (in Chinese)
[8] California Department of Conversation, Division of Mines and [30] Zhang, Y.B., Chen, G.Q., Zen, K., Kasama, K., Dong, S.M.,
Geology, Guidelines for evaluating and mitigating seismic Limit analysis of seismic slope stability based on tension-shear
hazards in California, CDMG Special Publication, 1997, 117. failure mechanism, Proceedings, Slope Stability 2011:
[9] Bray, J.D., Rathje, E. M., Earthquake-induced displacements of International Symposium on Rock Slope Stability in Open Pit
solid-waste landfills, J Geotech Geoenviron, 124 (1998) 242- Mining and Civil Engineering. 18-21, September 2011
253. Vancouver, Canada.
[10] Stewart, J.P., Blake, T.F., Hollingsworth, R.A., A screen [31] Bishop, A.W., The use of slip circle in the stability analysis of
analysis procedure for seismic slope stability, Earthq Spectra, slopes, Geotechnique, 4 (1955) 7-17.
19 (2003) 697–712. [32] Zienkiewicz, O.C., Humpheson, C., Lewis, R.W., Associated
[11] Chopra, A.K., The importance of vertical component of and non-associated visco-plasticity and plasticity in soil
earthquake motion, Bull Seism Soc Am, 56 (1966) 1163-1175. mechanics, Geotechnique, 25 (1975) 671–689.
[12] Ling, H.I., Leshvhinsky, D., Effects of vertical acceleration on [33] Matsui, T., San, K.C., Finite element slope stability analysis by
seismic design of geosynthetic-reinforced soil structures, shear strength reduction technique, Soils Found, 32 (1992) 59-
Geotechnique, 48 (1998) 347-373. 70.
[13] Newmark, N.M., Effects of earthquakes on dams and [34] Griffiths, D.V., Lane, P.A., Slope stability analysis by finite
embankments, Geotechnique, 15 (1965) 139-159. elements, Geotechnique, 49 (1999) 387-403.
[14] Makdisi, F.I., Seed, H.B., Simplified procedure for estimating [35] Dawson, E.M., Roth, W.H., Drescher, A., Slope stability
dam and embankment earthquake-induced deformations, ASCE analysis by strength reduction, Geotechnique, 49 (1999) 835-
J of Geotech Eng Div, 104 (1978) 849-867. 840.
[15] Bray, J.D., Rathje, E.M., Earthquake-induced displacements of [36] Zienkiewicz, O.C., Tayor, R.L., The finite element method,
soild-waste landfills. J Geotech Geoenviron, 124 (1998) 242- Vol.1, 4th edn, London, New York: McGraw-Hill, 1989.
253. [37] Zheng, Y.R., Tang, X.S., Zhao, S.Y., Deng, C.J., Lei, W.J.,
[16] Lin, J.S., Whitman, R.V., Earthquake induced displacements of Strength reduction and step-loading finite element approaches in
sliding blocks, J Geotech Eng, 112 (1983) 44-59. geotechnical engineering, J Rock Mech and Geotech Eng, 1
[17] Rathje, E.M., Bray, J.D., An examination of simplified (2009) 21-30.
earthquake-induced displacement procedures for earth structures, [38] Itasca, FLAC3D fast Lagrangian analysis of continua in 3
Canadian Geotech J, 36 (1999) 72-87. dimensions V.2.0, Itasca Consulting Group, Inc., Minneapolis.
[18] Rathje. E.M., Bray, J.D., Nonlinear coupled seismic sliding [39] Chen, Y.M., Xu, D.P., FLAC/FLAC3D foundation and
analysis of earth structures, J Geotech Geoenviron, 126 (2000) engineering project, China WaterPower Press, Beijing, China,
1002-1014. 2009. (in Chinese)
[19] Bray, J.D., Travasarou, T., Simplified procedure for estimating [40] Lin, M.L., Wang, K.L., Seismic slope behavior in a large-scale
earthquake-induced deviatoric slope displacements, J Geotech shaking table model test, Eng Geol, 86 (2006) 118-133.
Geoenviron, 133 (2007) 381-392.

28

View publication stats