Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

A fully nonlinear coupled seismic displacement model for earth slope with
multiple slip surfaces
Jian Song, Zhuxi Lu, Jian Ji, Yufeng Gao *
Key Laboratory of Ministry of Education for Geomechanics and Embankment Engineering, Hohai University, Nanjing, 210098, China

A R T I C L E I N F O A B S T R A C T

Keywords: For slopes consist of layered soil profiles, the earthquake-induced deformation may occur in various locations
Earthquakes and the slope may exhibit a set of plastic slip surfaces. In this paper, a fully nonlinear coupled sliding
Seismic sliding model displacement model is presented to evaluate the seismic performance of slopes considering multiple slip surfaces.
Slip surface
Both the nonlinear behavior in the cyclic response of soils and the sliding process at different slip surfaces are
Nonlinear
Coupled analysis
described based on a multi-degree-of freedom lumped parameter model. The nonlinear hysteretic soil response is
represented by the hyperbolic stress-strain curve and the Masing rules. The yield coefficients obtained from
pseudo-static analysis of slopes are used to represent the limit shear strength at each slip surface. The dynamic
equations of motion of the system in the cases of various sliding modes at shallow, median and deep slip surfaces
are derived, and the sliding displacements at different locations of slopes during the earthquake motion can be
computed. The code is validated by degrading the developed model to the conventional seismic site response
analysis and seismic sliding analysis models. The sliding displacement at different slip surfaces and the stress-
strain results of soil are analyzed to highlight the coupling effects of the plastic shear deformations at
different slip surfaces and the cyclic responses of soil. The comparisons of displacement results for different
slopes are then performed between the proposed model and the linear soil model and the single sliding analysis
model. Finally, an application to the Bainigou slope is presented to evaluate the seismic performance of the slope.

1. Introduction sliding block was used as an analog to the actual sliding mass, and thus
there are some limiting assumptions. Recent applications of
Earthquake-induced landslides have been an important hazard in Newmark-type sliding block method have implemented some im­
seismically active mountainous areas. During the last decades, various provements that are designed to obtain more accurate estimates of slope
methods were used to evaluate the seismic performance of slopes, sliding displacement by accounting for the dynamic slope response
including the pseudo-static analysis method, sliding displacement during earthquake shaking. The common modifications on the Newmark
analysis method and numerical analysis method (Ref. [1]). Among rigid-block analysis that are used to capture the dynamic response of
them, the sliding displacement analysis method bridges the gap between slope materials include decoupled and coupled analysis procedures. In
overly simplistic pseudo-static analysis and complex numerical analysis the decoupled approach, the dynamic slope response and the plastic
methods, and it is still widely used to assess the seismic stability of slopes sliding along the pre-defined potential failure surface are analyzed
at both site-specific and regional scales. separately (e.g. Refs. [3,4]). The equivalent acceleration time history
The sliding displacement analysis method is original form the New­ which represents the average seismic acceleration within the sliding
mark sliding block method, which models a landslide as a block that mass is calculate without considering the plastic deformation, and then
slides on an inclined plane under earthquake ground motions (Ref. [2]). it is input into the Newmark rigid-block analysis to calculate the seismic
Several increments of sliding displacement would occur along the shear displacement. In contrast, the coupled approach models together the
surface when the earthquake ground acceleration exceeds its limit shear dynamic response and the potential slip episodes of the sliding mass, and
strength (characterized by the yield coefficient ky). The final permanent the effect between the plastic sliding and the dynamic response of soil
sliding displacement can be served as an index to assess the seismic profile is taken into account (e.g. Refs. [5–14]). In these studies, the soil
performance of the slope. In the Newmark sliding block method, a rigid was modeled as linear (e.g. Refs. [5–7,11]) or nonlinear materials (e.g.

* Corresponding author.
E-mail address: yfgao66@163.com (Y. Gao).

https://doi.org/10.1016/j.soildyn.2022.107353
Received 30 November 2021; Received in revised form 18 April 2022; Accepted 9 May 2022
Available online 16 May 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 1. Analytical model representation of the coupled seismic sliding system.

Refs. [8–10,12–14]), and the nonlinear analysis would yield more ac­ In this paper, a fully nonlinear coupled sliding displacement model is
curate estimates of soil response and seismic displacement. developed to evaluate the seismic performance of slopes considering
The above-mentioned coupled seismic sliding methods focused on multiple slip surfaces. The nonlinear cyclic response of soils and the
slopes with a single slip surface, and the dynamic response was not plastic shear sliding process at different slip surfaces are described based
considered for soils below the slip surface (i.e., single basal sliding on a multi-degree-of freedom lumped parameter model. The dynamic
mode) for most studies. Among the slopes in mountainous areas, there is equations of motion of the system in the cases of various sliding modes at
a common type of complicated slopes composed of multi-layer soil de­ shallow, median and deep slip surfaces are derived, and the sliding
posits with different physical and mechanical properties. For these displacements at different locations of slopes during the earthquake
slopes, the earthquake-induced deformations may occur in various lo­ motion can be computed. The code is validated by degrading the
cations and a set of plastic slip surfaces may develop. This landslide type developed model to the conventional models, i.e., the seismic site
with multiple slip surfaces has also been mentioned in field in­ response analysis model and the seismic basal sliding model. The sliding
vestigations (Ref. [15]) and in experimental studies (Ref. [16]). The displacement at different slip surfaces and the nonlinear stress-strain
traditional seismic stability analysis and evaluation methods for slopes results of soil are analyzed. The proposed model provides a rigorous
with single sliding surface may not be suitable for the slopes with methodology to fully describe the coupling effects of the plastic shear
multiple sliding surfaces. deformations at different slip surfaces and the cyclic soil responses of
Leschinsky [17] presented a nested Newmark model by discretizing slopes.
the slope into many independent rigid wedges, and the seismic
displacement of each wedge was computed using the Newmark method. 2. Analytical formulation of nonlinear coupled multi-sliding
In this method, the distributed snear deformation of slope can be derived response
and the effect of failure geometry on the final slope deformation was
considered. Song et al. [18] developed a multi-block sliding approach 2.1. Multiple stick-slip model
for a limited number of rigid blocks to calculate the seismic sliding
displacement of slopes by considering the interaction among the sliding The soil profile of slope is discretized into individual layers by using a
process of each block. Recently, this approach was further extended to multi-degree-of freedom lumped parameter model, which is shown in
the slope cases with an arbitrary number of slip surfaces (Ref. [19]). Fig. 1. The soil profile has an inclination (θ), a depth (Hs), and each layer
These improvements have extended the applicability of the is represented by a corresponding mass (m), nonlinear spring (k) and
displacement-based methodology from both theoretical and practical dashpot (c). The base of the soil layers is modeled as a visco-elastic
aspects, but the rigid-block assumption was still used. For slopes with bedrock with a viscous damping coefficient Cr (Cr is equal to the prod­
non-unique potential slip surfaces, the evolution of one slip surface and uct of the density and shear-wave velocity of the bedrock). There are
the process of shear plastic deformation on this slip surface may alter the three potential slip surfaces (i.e., shallow, median and deep slip sur­
dynamics properties of slope and influence the wave propagation faces) considered in the lumped parameter model, which divide the soil
through the slope. This would in turn affect the development of sliding profile of slope into four sub-models, referred to as s1, s2, s3 and s4. The
displacement along the other slip surfaces. However, the complex number of discretization layers for the four sub-models is n1, n2, n3 and
coupling effects between the nonlinear cyclic response of soil and n4, respectively. During earthquake loading, the slope would respond as
multi-sliding behavior in various locations of slopes are not addressed in a nonlinear vibration system (stick mode), until the shear strength is
these studies. exceeded by the inertial force at one or more slip surfaces (stick-slip

2
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

mode). Because the occurrence of multiple sliding is considered, there ( )


1
would be different dynamic response modes for the slope. üb1 = − ky1 g − üs2 − 1 + üg − ⋅ 1T Ms1 ⋅üs1 (8)
In the stick stage without sliding occurrence, the relative seismic soil MT1
response to the base of soil profile can be solve by During the slip stage at the shallow surface, the subsequent sliding
Mü + Cu̇ + Ku = − M⋅1⋅üg (1) along the median or deep surface would be initiated in the case of either
Eq. (9a) or (9b) as
where M, C and K are the mass matrix, damping matrix and stiffness ( )
matrix, respectively; ü, u̇ and u are relative accelerations, relative ve­ − MT2 üs3 − 1 + üg − 1T Ms2 ⋅ üs2 + ky1 MT1 g ≥ ky2 MT1 T2 g (9a)
locities and relative displacements; üg is the acceleration at the base of
( )
soil profile; and 1 is a unity vector.
During the stick stage of vibration response, the inertial force acting − MT2 T3 üs4 − 1 + üg − 1T Ms2 s3 ⋅ üs2 s3 + ky1 MT1 g ≥ ky3 MT1 T2 T3 g (9b)
on each slip surface will be checked to see it exceeds the resistance.
Similar to Refs. [8,10] for single slip surface, this is expressed as In the case that sliding occurs at both shallow and median slip sur­
( ) faces, the equation of motion for soil mass of sub-model s1 is described as
Shallow ⋅ slip ⋅ surface: ⋅ − MT1 üs2 − 1 + üg − 1T ⋅ Ms1 ⋅ üs1 ≥ ky1 MT1 g (2a) ( )
Ms1 üs1 + Cs1 u̇s1 + Ks1 us1 = − Ms1 ⋅ 1⋅ üs2 − 1 + üb1 + üs3 − 1 + üg + üb2 (10)
( )
Median ⋅ slip ⋅ surface: ⋅ − MT1 T2 üs3 − 1 + üg − 1T ⋅ Ms1 s2 ⋅ üs1 s2 ≥ ky2 MT1 T2 g and the equilibrium on the shallow slip surface is governed by
( )
(2b)
− MT1 üs2 − 1 + üb1 + üs3 − 1 + üg + üb2 − 1T Ms1 ⋅ üs1 = ky1 MT1 g (11)
( )
Deep ⋅ slip ⋅ surface: ⋅ − MT1 T2 T3 üs4 − 1 + üg − 1T ⋅ Ms1 s2 s3 ⋅ üs1 s2 s3 Thus, the equation of motion for the sub-model s1 is the same as Eq.
(5).
≥ ky3 MT1 T2 T3 g (2c)
The equation of motion for soil mass of sub-model s2 is described as
( )
in which MTi is the total mass of the soil body of sub-model si, (üsi − 1 +üg )
Ms2 üs2 + Cs2 u̇s2 + Ks2 us2 = − Ms2 ⋅ 1 ⋅ üs3 − 1 + üg + üb2 + ky1 MT1 g⋅J (12)
represents the absolute acceleration response at layer si-1 (üsi − 1 is the
relative acceleration obtained from Eq. (1)), the term 1T ⋅ Msi ⋅ üsi rep­
The equilibrium on the median slip surface is governed by
resents the inertial force induced by the non-uniform relative accelera­
( )
tion profile (üsi ) of sub-model si to the underlying slip element bi, and the
− MT2 üs3 − 1 + üg + üb2 − 1T ⋅ Ms2 ⋅ üs2 + ky1 MT1 g = ky2 MT1 T2 g (13)
term kyi MTi g indicates the shear strength of the slip surface, and kyi is its
yield coefficient.
If sliding occurs first along the shallow slip surface, the soil mass and the final form of the equation of motion for sub-model s2 would be
above and below the shallow slip surface would respond in different [
MT ( )
]
modes. For the soil mass above the slip surface (sub-model s1), the dy­ M*s2 üs2 + Cs2 u̇s2 + Ks2 us2 = Ms2 ⋅ 1 ⋅ ky2 g + 1 ky2 − ky1 g + ky1 MT1 g⋅J
MT2
namic equation of motion is described as
(14)
( )
Ms1 üs1 + Cs1 u̇s1 + Ks1 us1 = − Ms1 ⋅ 1⋅ üs2 − 1 + üg + üb1 (3) Then, the equation of motion for soil mass of sub-model s3 and s4 is
Ms3 s4 üs3 s4 + Cs3 s4 u̇s3 s4 + Ks3 s4 us3 s4 = − Ms3 s4 ⋅1⋅üg + ky2 MT1 T2 g⋅J (15)
where übi is the relative sliding acceleration of the slip surface. On the
By solving Eqs. (5), (12) and (15) to obtain the relative acceleration
slip surface, in this case, the equilibrium is governed by the condition
response, the sliding acceleration-time history at shallow and median
( )
slip surfaces can be derived from Eqs. (11) and (13) as
− MT1 üs2 − 1 + üg + üb1 − 1T Ms1 ⋅ üs1 = ky1 MT1 g (4)
( )
1
üb1 + üb2 = − ky1 g − ⋅ 1T Ms1 ⋅ üs1 − üs2 − 1 + üs3 − 1 + üg (16a)
Substituting equation (4) in equation (3), the equation of motion is MT1
obtained as
[ ] ( )
MT ( ) 1
M∗s1 üs1 + Cs1 u̇s1 + Ks1 us1 = Ms1 ⋅1⋅ky1 g (5) üb2 = − ky2 g + 1 ky2 − ky1 g − ⋅ 1T Ms2 ⋅ üs2 − üs3 − 1 + üg (16b)
MT2 MT2
where M∗si is given as During the stage of both shallow and median sliding, the subsequent
sliding along the deep surface would be initiated in the condition
1
M∗si = Msi − ⋅Msi ⋅1⋅1T ⋅Msi (6) ( )
MTi
− MT3 üs4 − 1 + üg − 1T Ms3 ⋅ üs3 + ky2 MT1 T2 g ≥ ky3 MT1 T2 T3 g (17)
For the soil mass below the slip surface (sub-model s2, s3 and s4), the
force transmitted at the shallow slip surface from the shallow sliding is If sliding occurs simultaneously at shallow, median and deep sur­
considered and the equation of motion is described as faces, all the four sub-models would respond in different modes. The
equation of motion for sub-model s1 and s2 is the same as Eq. (5) and Eq.
Ms2 s3 s4 üs2 s3 s4 + Cs2 s3 s4 u̇s2 s3 s4 + Ks2 s3 s4 us2 s3 s4 = − Ms2 s3 s4 ⋅1⋅üg + ky1 MT1 g⋅J (7)
(14). For the sub-model s3, the equation of motion is expressed as
( )
where J is a vector containing the number of elements equals to the total
Ms3 üs3 + Cs3 u̇s3 + Ks3 us3 = − Ms3 ⋅ 1 ⋅ üs4 − 1 + üg + üb3 + ky2 MT1 T2 g⋅J
number of discretization layers of the sub-model (n2+n3+n4), where the
first equals to one and the others equal to zero. By solving Eqs. (5) and (18)
(7) to obtain the acceleration response of üs1 and üs2 − 1 respectively, the
sliding acceleration-time history is derived from Eq. (4) as and the equilibrium on the deep slip surface is governed by

3
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

( )
( )
− MT3 üs4 − 1 + üg + üb3 − 1T Ms3 ⋅ üs3 + ky2 MT1 T2 g = ky3 MT1 T2 T3 g (19)
− MT1 üs2 − 1 + üs3 − 1 + üg + üb2 − 1T Ms1 ⋅ üs1 ≥ ky1 MT1 g (28a)
The equation of motion for soil mass of sub-model s3 is written as ( )
[ ]
MT T ( ) − MT3 üs4 − 1 + üg − 1T Ms3 ⋅ üs3 + ky2 MT1 T2 g ≥ ky3 MT1 T2 T3 g (28b)
M*s3 üs3 + Cs3 u̇s3 + Ks3 us3 = Ms3 ⋅ 1 ⋅ ky3 g + 1 2 ky3 − ky2 g + ky2 MT1 T2 g⋅J
MT3
(20) If sliding occurs at both shallow and median slip surfaces, it is the
same as the above case. If sliding occurs at both median and deep slip
Finally, the equation of motion for soil mass s4 is surfaces, the equation of motion for the soil mass above the median
Ms4 üs4 + Cs4 u̇s4 + Ks4 us4 = − Ms4 ⋅1⋅ug + ky3 MT1 T2 T3 g⋅J (21) surface is the is the same as Eq. (26), the equation of motion for the soil
mass s3 is the same as Eq. (20), and the equation of motion for the soil
By solving the above dynamic equations for all four sub-models, the mass s4 is the same as Eq. (21). The sliding acceleration at median and
sliding acceleration-time history at shallow, median and deep slip sur­ deep slip surfaces can be derived as
faces can be derived as ( )
1
1
( ) üb2 + üb3 = − ky2 g − ⋅ 1T Ms1 s2 ⋅ üs1 s2 − üs3 − 1 + üs4 − 1 + üg (29a)
üb1 + üb2 + üb3 = − ky1 g − ⋅ 1T Ms1 ⋅ üs1 − üs2 − 1 + üs3 − 1 + üs4 − 1 + üg MT1 T2
MT1
[ ] ( )
(22a) MT T ( ) 1
üb3 = − ky3 g + 1 2 ky3 − ky2 g − ⋅ 1T Ms3 ⋅ üs3 − üs4 − 1 + üg
[ ] MT3 MT3
MT ( ) 1
üb2 + üb3 = − ky2 g + 1 ky2 − ky1 g − ⋅ 1T Ms2 ⋅ üs2 (29b)
MT2 Mt2
( )
and the subsequent sliding along the shallow surface is initiated as
− üs3 − 1 + üs4 − 1 + üg (22b)
( )
[ ] ( ) − MT1 üs2 − 1 + üs3 − 1 + üb2 + üs4 − 1 + üg + üb3 − 1T Ms1 ⋅ üs1 ≥ ky1 MT1 g (30)
MT T ( ) 1
üb3 = − ky3 g + 1 2 ky3 − ky2 g − ⋅ 1T Ms3 ⋅ üs3 − üs4 − 1 + üg
MT3 MT3 Finally, for the case that sliding occurs first along the deep slip sur­
(22c) face, the equation of motion for the soil mass above the deep surface is
derived as
Similarly, in the case that sliding occurs at both shallow and deep slip
surfaces, the equation of motion for soil mass of sub-model s1 is the same M∗s1 s2 s üs1 s2 s3 + Cs1 s2 s3 u̇s1 s2 s3 + Ks1 s2 s3 us1 s2 s3 = Ms1 s2 s3 ⋅1⋅ky3 g (31)
as Eq. (5), and the equation of motion for soil mass s2 and s3 can be
3

derived as and the equation of motion for the soil mass below the deep surface is
[
MT1 ( )
] the same as Eq. (21). The sliding acceleration-time history along the
M*s2 s3 üs2 s3 + Cs2 s3 u̇s2 s3 + Ks2 s3 us2 s3 = Ms2 s3 ⋅ 1 ⋅ ky3 g + ky3 − ky1 g deep surface is derived as
MT2 T3
( )
+ ky1 MT1 g⋅J (23) 1
üb3 = − ky3 g − üs4 − 1 + üg − ⋅ 1T Ms1 s2 s3 ⋅üs1 s2 s3 (32)
MT1 T2 T3
The equation of motion for the sub-model s4 is the same as Eq. (21).
The sliding acceleration at shallow and deep slip surfaces can be derived and the initiation of sliding on either shallow or median surface condi­
as tioned on the occurrence of deep sliding is given by
( ) ( )
1
üb1 + üb3 = − ky1 g − ⋅ 1T Ms1 ⋅ üs1 − üs2 − 1 + üs4 − 1 + üg (24a) − MT1 üs2 − 1 + üs4 − 1 + üg + üb3 − 1T Ms1 ⋅ üs1 ≥ ky1 MT1 g (33a)
MT1
[ ] ( ) ( )
MT1 ( ) 1
üb3 = − ky3 g + ky3 − ky1 g − ⋅ 1T Ms2 s3 ⋅ üs2 s3 − üs4 − 1 + üg − MT1 T2 üs3 − 1 + üs4 − 1 + üg + üb3 − 1T Ms1 s2 ⋅ üs1 s2 ≥ ky2 MT1 T2 g (33b)
MT2 T3 MT2 T3
(24b)
The subsequent sliding process would the same as the above
At this stage, sliding at the median surface would be further initiated derivations.
in the condition
( )
− MT2 üs3 − 1 + üs4 − 1 + üg + üb3 − 1T Ms2 ⋅ üs2 + ky1 MT1 g ≥ ky2 MT1 T2 g (25) 2.2. Soil nonlinearity

Similar analysis can be performed for the case that sliding occurs first In the presented multi-stick-slip model of seismic slope displace­
along the median slip surface. The dynamic equation of motion for the ment, the nonlinear hysteretic soil behavior is characterized. The
soil mass above the median surface is described as modified Kondner-Zelasko (MKZ) hyperbolic model (Ref. [20]) was used
to describe the nonlinear stress-strain curve of soil under cyclic loading.
M∗s1 s2 üs1 s2 + Cs1 s2 u̇s1 s2 + Ks1 s2 us1 s2 = Ms1 s2 ⋅1⋅ky2 g (26) The backbone curve which defines the stress-strain relationship for
loading is given by:
For the soil mass below the median surface (sub-model s3 and s4), the
equation of motion is the same as Eq. (15). The sliding acceleration-time G0 γ
τ= ( )s (34)
history along the median surface is derived as 1+β γ
γr
( )
1
üb2 = − ky2 g − üs3 − 1 + üg − ⋅ 1T Ms1 s2 ⋅üs1 s2 (27)
MT1 T2 where τ is the shear stress, γ is the shear strain, G0 is the initial shear
modulus, γr is the reference shear strain, β and s are the dimensionless
and the initiation of subsequent sliding would occur either on the parameters allowing for control over the shape of the stress-strain curve.
shallow or on the deep surface. This can be described as The stress-strain relationship for unloading-reloading path was rep­
resented by the modified hyperbolic model and the Masing rules, and is
given by (Ref. [21]):

4
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 2. Shear modulus reduction and damping ratio data for clay (PI = 30) proposed by Vucetic and Dobry [24] and the fitted curves to match the data.

G0 ((γ − γrev )/2) all soil layers is less than the tolerance value. In addition to the hys­
τ=2 ( )s + τrev (35)
1 + β (γ−2γγrrev ) teretic damping, the viscous damping at small strain was added using
the full Rayleigh damping formulation, in which the damping matrix C
is proportional to the mass matrix and the stiffness matrix and expressed
where γ rev and τrev are the shear strain and shear stress at the reversal
as
point between the unloading and reloading paths, which were identified
for each soil layer during analysis. C = αR ⋅M + βR ⋅K (38)
The parameters in the hyperbolic model γr, β and s can be determined
by matching both the modulus reduction and damping curve of the soil where αR and βR are Rayleigh damping coefficients, which were deter­
to find the best fit values. The modulus reduction and damping curve of mined based on the small strain damping of soil, and the first and second
the soil can be obtained by the laboratory data points or by the empirical natural periods of the slope profile (Ref. [20]) as
curves of different soil types provided from literatures. 4π ξmin T1 T2 ξmin
For different sliding modes at shallow, median and deep slip sur­ αR = , βR = (39)
T1 + T2 π(T1 + T2 )
faces, the seismic responses of the slope were calculated using the in­
cremental form of the dynamic equations of motion that are derived where T1 and T2 are the natural period of the first and second modes, and
above. At each time step, the Newmark β method with the integration ξmin is the minimum soil damping ratio.
coefficients of unconditionally stable (Ref. [22]) was used to numeri­ In summary, the inputs for the developed code include the soil pa­
cally solve all the equations of motion. The code was built in the rameters (i.e., the depth, unit weight, shear-wave velocity, the minimum
MATLAB environment to take advantages of its matrix operations. The soil damping ratio and the nonlinear soil parameters of each layer, and
stiffness of the ith layer is defined as the location and yield coefficient of each slip surface), the bedrock pa­
Gi Δτi rameters (i.e., the unit weight and shear-wave velocity) and the earth­
ki = = (36) quake ground motion. The computed output includes the dynamic
hi hi Δγi
responses and shear stress-strain results at various soil depths, and the
where Gi is the tangent shear modulus at current time step, hi is the permanent sliding displacement along different slip surfaces as well.
thickness of the ith layer, and Δτi and Δγi are the increment of shear Note that the presented formulation models the coupled seismic sliding
stress and shear strain of the ith layer at this time step. The shear strain of slope with three slip surfaces, but similar derivation could be per­
of the ith layer is computed from the relative displacement response of formed for a more generalized condition (i.e., slope with an arbitrary
the layer, and is given by number of slip surfaces). To this end, all the possible stick-slip modes
should be firstly identified and the governing equations of motion for all
ui − ui+1
γi = (37) sub-systems are then derived. The initiation of different sliding and the
hi
sliding accelerations in different cases would be mathematically
Because the soil response is nonlinear, an iterative scheme is determined.
required to be done at each time step. The stiffness from the previous
step is used as initial value to obtain the shear strain at current step, and
the shear stress at current step is calculated based on Eq. (34) or (35).
Then the stiffness will be updated to repeat the process, until the
maximum difference of strain between the two consecutive solutions for

Fig. 3. Acceleration time history and 5% damped spectral acceleration at ground surface computed from the code in this study and from DEEPSOIL.

5
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 4. Profiles of the maximum shear strain and shear stress, and the stress-strain plots at different depths computed from the code in this study and from DEEPSOIL.

3. Code validation with the conventional models earthquake, which is also from Rathje and Bray [8].
Fig. 2 shows the shear modulus reduction and damping ratio data for
3.1. Validation of seismic site response clay (PI = 30) proposed by Vucetic and Dobry [24] and the fitted curves
to match the data. The obtained parameters from the fitting procedure
The code of the proposed analysis method was firstly checked by are γr = 0.0768%, β = 0.615 and s = 0.705. The small strain damping
degrading the developed model to the seismic site response analysis used in the Rayleigh damping coefficients is 1%. Fig. 3 presents the
model without consideration of shear failure sliding (by setting large acceleration response at ground surface computed from the code in this
yield coefficients to all slip surfaces). A simplified uniform site consists study and from DEEPSOIL. Observe that both acceleration time history
of 60 m thick layer of a medium plasticity clay soil from Rathje and Bray and the 5% damped spectral acceleration are consistent from the two
[8] was used to conduct the seismic response analysis, and the results codes. Fig. 4 also shows the profiles of the maximum shear strain and
were compared with those from DEEPSOIL platform (Ref. [23]). The shear stress, and the stress-strain cycles at different depths computed
unit weight and shear-wave velocity are 18 kN/m3 and 240 m/s for the from the code in this study and DEEPSOIL. The results show a good
soil, and are 23 kN/m3 and 3000 m/s for the underlying elastic bedrock. agreement for the nonlinear site responses, which validate the perfor­
The strain-dependent shear modulus reduction and damping curves for mance of the code in describing the stick stage of nonlinear vibrations of
the soil were those from Vucetic and Dobry [24] with a plasticity index slope.
(PI) of 30. The input ground motion was the Superstition Mountain
strong motion record (135 component) from the 1987 Superstition Hills

Fig. 5. Sliding displacement from the nonlinear coupled basal sliding analysis along with the results from Rathje and Bray [8].

6
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 6. Difference of sliding displacements from nonlinear decoupled and Fig. 7. Acceleration time history of the simplified wavelet motion.
coupled basal sliding analyses (ky = 0.05) along with the results from Rathje
and Bray (2000).

3.2. Validation of sesimic basal sliding displacement

The code of the proposed method was then checked by degrading the
developed model to the traditional nonlinear stick-slip analysis model
considering only one slip surface at the bedrock location of the site (by
setting large yield coefficients to the other slip surfaces). This type of
seismic sliding model of slope has been widely used in the previous
studies (e.g. Refs. [6–8]). The nonlinear sliding displacements calculated
from Rathje and Bray [8] were adopted for the validation. The soil pa­
rameters and the input ground motion are presented in Section 3.1,
which are the same as those in Rathje and Bray [8]. A large value of
shear-wave velocity (3000 m/s) is used for the underlying elastic
bedrock to approximate the rigid bedrock assumption in Rathje and Bray
[8].
Coupled sliding displacements calculated with nonlinear soil prop­
erties are plotted versus Ts/Tm (by changing the height of the sliding
mass) in Fig. 5 for different yield coefficients of the basal slip surface.
The period ratio Ts/Tm is the ratio of the initial fundamental period of
the sliding mass to the mean period of the input ground motion Fig. 8. Profiles of the maximum shear strain for different sliding cases.
(Ref. [25]). It can be seen that the calculated displacements are gener­
ally agree favorably with the literature. The maximum displacements ky3 = 0.15, respectively. To allow the investigation of single frequency
are all observed at a period ratio Ts/Tm less than one because of the input motion, the acceleration time history of the wavelet motion was
nonlinearity introduced by the soil response and the sliding process. In adopted, which is shown in Fig. 7. This type of wavelet motion includes
addition, the sliding displacement was also calculated using the the build-up and decay of shaking in time history that is similar to real
decoupled procedure as that in Rathje and Bray [8], and the difference of earthquake ground motions (Ref. [26]). In addition, the wavelet motion
sliding displacements from nonlinear decoupled and coupled analyses is symmetrical, and this allows for a better observation on the effect of
for a yield coefficient of 0.05 at the basal slip surface are displayed in one-direction sliding on the cyclic response of soil. The frequency of the
Fig. 6. The displacement results shown in Figs. 5 and 6 validate the wavelet motion is 2 Hz and the amplitude is 0.4 g.
performance of the code in describing the nonlinear soil response and Variable sliding modes of the slope were considered by assigning
the seismic basal-slip behavior of slope. different yield coefficient values to the three slip surfaces, i.e., the no
sliding mode (ky1 = 10, ky2 = 10, ky3 = 10), shallow sliding mode (ky1 =
4. Cyclic response of soil under different sliding modes 0.05, ky2 = 10, ky3 = 10), shallow and median sliding mode (ky1 = 0.05,
ky2 = 0.1, ky3 = 10), and three-sliding mode (ky1 = 0.05, ky2 = 0.1, ky3 =
Based on the developed analytical model, the coupling interactions 0.15). Fig. 8 shows the profiles of the maximum shear strain for different
between the cyclic nonlinearity of soil and the shear sliding behavior in sliding cases. Note that this shear-strain plot is only the result of
various locations of slope were examined. A slope consists of 30 m thick nonlinear response and does not include the plastic strain induced by the
layer of clay soil was used to perform the analysis. The parameters of the shear sliding. It can be seen that consideration of different sliding modes
soil are the same as in Section 3.1 (thus the fundamental period of the would result in different distribution of shear-strain response. Compared
slope is 0.5 s), and the shear-wave velocity of the underlying elastic with the no sliding case, the occurrence of shallow sliding would reduce
bedrock is 760 m/s. The shallow, median and deep slip surfaces were the shear strain at shallow-to-median depth of the slope, while increase
considered, which are located at 10 m, 20 m, 30 m depths of the slope, the shear strain of the lower soil layers. This is because the dynamic
and the yield coefficient of the three surfaces are ky1 = 0.05, ky2 = 0.1, characteristic of the system changes with the occurrence of the shallow

7
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 9. Sliding displacements for different sliding modes (a) shallow sliding, (b) shallow and median sliding (c) shallow, median and deep sliding.

sliding. In the nonlinear site, the resonance period ratio would shift to and the case of coupled shallow and deep sliding. Observe that the
less than one, and the nonlinear resonance is more obvious for the sub- sliding displacement at the deep surface becomes much larger with
model in Eq. (7) because the fundamental period for this sub-model considering the shallow sliding. This is the result of the larger shear
would be less than 0.5 s (the period of the input motion). When stress and strain developed at the location of deep slip surface when the
sliding behaviors occur in more slip surfaces, this increase in shear strain shallow sliding occurs (Fig. 10c). Therefore, the developed model is able
would not be observed because the nonlinear resonance would shift to capture the coupling effects between the nonlinear cyclic responses
away from the sub-models. In addition, the shear strain is similar be­ and the plastic shear deformations in various locations of slope, which
tween the shallow and median sliding mode, and three-sliding mode, as are important for the accurate prediction of the seismic response and
shown in Fig. 8. This is due to that only small sliding displacement oc­ sliding displacement of slope.
curs at the deep surface, as can be seen from the sliding displacement
results for different sliding modes shown in Fig. 9. It is observed from 5. Comparison of sliding displacements from different
Fig. 9 (a) and (b) that the sliding displacement at the shallow slip surface approaches
decreases when considering the occurrence of subsequent median
sliding. This is related to the influence of the median sliding process on 5.1. Nonlinear and linear analysis
the transmission of the earthquake force to the upper sliding mass. When
both the shallow and median sliding occur, the inertial force transmitted The sliding displacement calculated from linear and nonlinear
to the shallow sliding mass is somewhat limited by the sliding behavior coupled analyses are compared to examine the effects of soil nonline­
at the median surface. By comparing Fig. 9 (b) and (c), it can be seen that arity on the multi-sliding displacements. The Superstition Mountain
the sliding displacements at shallow and median surfaces are similar. strong motion record was still used as the input motion. Soil profile of
This is also due to that only small sliding displacement occurs at the deep slopes with various thicknesses were studied and the sliding displace­
surface, and its effect on the median and deep sliding displacements is ments from linear and nonlinear analyses are plotted versus Ts/Tm in
limited. Fig. 12. The soil parameters are the same as the above, and the shear-
Fig. 10 shows the stress-strain cycles of soil layer at various depths wave velocity of the underlying elastic bedrock is 760 m/s. For each
for different sliding modes. Observe that when sliding occurs, the slope site, the location of shallow, median and deep surfaces is Hs/3,
development of stress and strain at the location of the surface will be 2Hs/3 and Hs of the slope, respectively. Different combinations of the
limited in the sliding direction. Different sliding modes result in yield coefficient values for the slip surfaces were considered, repre­
different cyclic response and nonlinearity of slope soils. The shear strain senting different distribution for the shear strength of the multiple slip
is generally reduced with the sliding occurrence, except for the location surfaces. For linear analysis, because the shear modulus (i.e., the initial
of deep slip surface with shallow sliding mode (the second plot in shear modulus) and the damping ratio are constants values, different
Fig. 10c), and this is because of the change in the dynamic characteristic damping ratios (5% and 15%) are examined.
of the system which has been explained above. Fig. 11 shows the sliding It can be seen from Fig. 12 that the linear analysis results generally
displacement at the deep slip surface for the case of single deep sliding underpredicted the displacements at smaller period ratios while

8
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 10. Stress-strain plots of soil at different depths for different sliding modes (a) location of 10 m, (b) location of 20 m, (c) location of 30 m.

overpredicted the displacements at larger period ratios, which is stick-slip model and the single stick-slip model were further compared to
consistent with the results of the single basal sliding analysis from Rathje examine the effects of the multi-sliding process on the seismic perfor­
and Bray [8]. The underprediction of displacement at small period ratios mance of slope, and the results are shown in Fig. 13. In the figure, sliding
is more obvious for a larger value of damping ratio. However, for some displacements from two types of single sliding model are presented. In
particular cases, e.g., the case shown in Fig. 12d, the linear analysis with the first model, the displacement of the three slip surfaces was calculated
a damping ratio of 15% would provide underpredictions for all period independently, and this was done by setting a large shear strength to
ratios. This is related to the much more complex interactions between other slip surfaces when analyzing the displacement for one slip surface.
the dynamic response and sliding process in the multi-sliding system. In this model, the site condition is considered but the multi-sliding
Therefore, the damping ratio in the linear analysis affects the displace­ behavior is ignored. The second model is that used in the traditional
ment results, but there is not a rigorous basis to guide the selection of the studies. The displacement of the three slip surfaces was also calculated
damping ratio used in the linear analysis. The nonlinear analysis model independently, but each sliding mass was located at the bedrock (i.e.,
proposed in this study can describe more realistic dynamic behaviors of the basal sliding mode). In this model, both the site condition below the
soil and slope under earthquakes, and provide more reasonable pre­ slip surface and the multi-sliding behavior are ignored.
dictions of multi-sliding displacement. It is observed from Fig. 13 that the two types of single-sliding anal­
ysis methods predict different displacements compared to the coupled
multi-sliding method. The single sliding method generally obtains larger
5.2. Multiple and single sliding analysis displacements than the multi-sliding method, and in some cases, this
overestimation could be very significant. This can be explained by the
The sliding displacement calculated from the nonlinear multiple

9
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

method can be either larger or smaller than the multi-sliding method.


This is related to the neglection of the site condition below the slip
surface in this method. In this case, the seismic site response would be
totally different from the full-site analysis. In addition, the variation of
sliding displacement with the site depth (Hs) also exhibits different
patterns for different analysis methods. The displacements peak at
different site depths although the soil nonlinearity is included in all
methods. In summary, the conventional single-sliding model, which
ignores either the site condition below the slip surface or the multi-
sliding coupling behavior, may deviate from the realistic conditions of
slopes with multiple slip surfaces, and this ignoration also has significant
effects on the prediction of sliding displacement.

6. Example application of the proposed model

An example study on the Bainigou slope is presented to show the


application of the developed method to the realistic cases. The Bainigou
slope is located at the Qiaojia county, Yunnan province of China, where
it is the seismically active region. The deposit slope includes silty clay
and strong weathered limestone layers, and there is a weak clay inter­
Fig. 11. Comparison of sliding displacement at the deep slip surface for the layer under the strong weathered rock layer (Ref. [27]). Fig. 14 shows
case of single deep sliding and the case of coupled shallow and deep sliding.
the typical section of the Bainigou slope, which was modified from Wang
et al. [27]. The analysis was performed using the Superstition Mountain
influence of the multi-sliding process on the transmission of the earth­ motion above. The soil parameters required for the seismic analysis is
quake force. When multi-sliding occurs, the inertial force from earth­ listed in Table 1 by combining the natural and saturated soil states. The
quake ground motion may be limited by the sliding behavior at each slip nonlinear soil parameters (γ r, β, s) were determined from the Vucetic and
surface. Note that this observation is only for this particular case. For Dobry [24] based on the value of PI. Although most of the
other cases of input motion and soil profile, the single sliding method displacement-based methods reduced the 2D problem to a 1D scheme,
may also predict smaller displacements than the multi-sliding method the predictions were found to provide reasonable approximations of the
(Fig. 11) because of the effects of sliding behavior on the fundamental displacement range by using several 1D soil models to represent the
period of the system (which has been explained above). The displace­ main sections of the slope geometry (Ref. [28]). For the Bainigou slope,
ment predictions from the more conventional single basal sliding

Fig. 12. Sliding displacement calculated from linear and nonlinear analyses (a) ky1 = 0.05, ky2 = 0.1, ky3 = 0.15, (b) ky1 = 0.05, ky2 = 0.15, ky3 = 0.1, (c) ky1 = 0.15,
ky2 = 0.05, ky3 = 0.1, (d) ky1 = 0.1, ky2 = 0.15, ky3 = 0.05 (The solid lines represent the nonlinear results, the dotted and dashed lines represent the linear results with
a constant damping ratio of 5% and 15%, respectively).

10
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Fig. 13. Sliding displacement calculated from coupled multi-sliding and conventional single sliding analyses (a) ky1 = 0.05, ky2 = 0.1, ky3 = 0.15, (b) ky1 = 0.1, ky2 =
0.15, ky3 = 0.05.

Fig. 14. Section of the Bainigou slope (Ref. [27]).

11
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

Table 1
Physical and mechanical parameters of soils.
Soil layer Unit weight (kN/m3) Shear-wave velocity (m/s) Cohesion (kPa) Friction angle (◦ ) PI γr (%) β s

Silty clay 19.4 180 20 12 15 0.0638 0.900 0.720


Clay 18.9 140 17 12.5 30 0.0768 0.615 0.705
Strongly weathered limestone 22.0 1200 45 27 / / / /

Table 2
Factor of safety, yield coefficient and thickness of sliding masses.
Fos ky Thickness at S1 (m) Thickness at S2 (m) Thickness at S3 (m) Thickness at S4 (m) Thickness at S5 (m)

Shallow mass 1.28 0.081 8 12 9 9 7


Deep mass 1.07 0.032 5 10 16 20 6

7. Conclusions

This paper presents a fully nonlinear multi-sliding model to analyze


the seismic response and sliding behaviors of slopes with a set of slip
surfaces. The analytical formulation is developed by deriving the dy­
namic equations of motion of the soil system in the cases of various
sliding modes at shallow, median and deep slip surfaces. Specifically, the
equations that describe the dynamic response, the initiation of different
sliding states and the sliding displacement along the different slip sur­
faces are obtained. The nonlinear hysteretic soil behavior is character­
ized by the modified hyperbolic model for the backbone curve and the
Masing rules for the unloading-reloading stress-strain relationship.
The performance of the code is examined by reducing the developed
model to two types of traditional seismic analysis model, i.e., nonlinear
seismic site response analysis model without consideration of shear
failure sliding and the nonlinear basal sliding analysis model. The
comparisons validate the accuracy of the code in describing the
nonlinear soil response and the seismic sliding behavior on the slip
Fig. 15. Seismic sliding displacement of the example slope calculated from the
developed approach.
surface. The detailed analyses of the sliding displacement and stress-
strain results of soil highlight the coupling effects of the plastic shear
sliding at various slip surfaces and the nonlinear cyclic response of soil.
there were five sections selected as the typical 1D models to calculate the
The developed analytical model is then applied to the Bainigou slope to
sliding displacement under the earthquake motion. According to the
evaluate the seismic performance of the slope.
survey, two potential slip surfaces exist, i.e., the shallow surface is along
The developed nonlinear coupled seismic displacement models
the interface between the silty clay and strong weathered limestone
provide an alternative to describe more realistic seismic soil behaviors of
layers, and the deep one is along the weak clay interlayer. The thickness
slopes when sliding deformation may occur along a set of surfaces. In
of the weak clay interlayer is about 1 m. For the underlying elastic
addition, the developed model does not bring any additional computa­
bedrock, the unit weight and shear-wave velocity are 25 kN/m3 and
tion and input parameters that is difficult to acquire compared to the
2400 m/s.
traditional displacement-based methods.
The factor of safety (Fos) and the yield coefficient (ky) determine
from the Spencer limit equilibrium analysis method (Ref. [29]) are given
in Table 2, along with the thicknesses of the shallow and deep soil Author statement
masses for each section. The deep slip surface has a lower stability level
than the shallow one. The seismic sliding displacement of Bainigou slope Jian Song: Conceptualization, Methodology, Software, Writing -
calculated from the developed approach is shown in Fig. 15, in which Original Draft Zhuxi Lu: Investigation, Resources, Validation Jian Ji:
both the displacement of each section and the average displacement are Visualization, Supervision Yufeng Gao: Writing - Review & Editing,
shown. The average sliding displacements along the shallow and deep Supervision, Project administration.
slip surfaces induced by the earthquake ground motion are 8.2 cm and
136.9 cm, respectively. Therefore, the earthquake-induced landslide Declaration of competing interest
probably exhibits a deep failure mode, but a relatively small sliding
deformation may also occur along the shallow slip surface. The authors declare that they have no known competing financial
It should be noted that factors such as build-up of pore water pres­ interests or personal relationships that could have appeared to influence
sure at the soil profile (e.g. Ref. [30]), rotation of sliding mass (e.g. Refs. the work reported in this paper.
[31,32]) and the strain softening of the slip surface (e.g. Ref. [32]) are
not considered in the developed method. These factors may also be
Acknowledgements
important and affect the seismic shear response for some slope condi­
tions. Because the method is still based on the multi-degree-of freedom
The study is supported by the National Natural Science Foundation
lumped parameter model, similar efforts may be introduced to include
of China (Grant No. 52178325 and 51879091), and the Fundamental
the pore water pressure based on the model in Refs. [23,30]. The rota­
Research Funds for the Central Universities in China (Grant No.
tion of sliding mass and the strain softening of the slip surface may also
B210202046). These supports are gratefully acknowledged. Special
be introduced by the similar manner as in Refs. [31,32].
thanks to Prof. Ellen M. Rathje in Univ. of Texas, Austin for valuable

12
J. Song et al. Soil Dynamics and Earthquake Engineering 159 (2022) 107353

discussions regarding the code implementation of the nonlinear coupled [16] Wartman J, Seed RB, Bray JD. Shaking table modeling of seismically induced
deformations in slopes. J Geotech Geoenviron Eng 2005;131(5):610–22.
model. The authors would also like to thank the journal editor and
[17] Leshchinsky BA. Nested Newmark model to calculate the post-earthquake profile of
anonymous reviewers for their valuable comments. slopes. Eng Geol 2018;233:139–45.
[18] Song J, Fan QQ, Feng TG, Chen ZQ, Chen J, Gao YF. A multi-block sliding approach
References to calculate the permanent seismic displacement of slopes. Eng Geol 2019;255:
48–58.
[19] Song J, Rodriguez-Marek A, Feng T, Ji J. A generalized seismic sliding model of
[1] Jibson RW. Methods for assessing the stability of slopes during earthquakes—a slopes with multiple slip surfaces. Earthq Eng Struct Dynam 2021;50:2595–612.
retrospective. Eng Geol 2011;122:43–50. [20] Matasovic N. Seismic response of cosmic horizontally-layered soil deposits. PhD
[2] Newmark NM. Effects of earthquakes on dams and embankments. Geotechnique dissertation. Los Angeles: Dept. of Civ. and Envir. Engrg., University of California;
1965;15(2):139–60. 1993.
[3] Makdisi FI, Seed HB. Simplified procedure for estimation dam and embankment [21] Hashash, Y.M.A., Phillips, C, Groholski, D.R. Recent advances in non-linear site
earthquake induced deformations. J. Geotech. Eng Div, ASCE 1978;104(GT7): response analysis, 2010. Fifth International Conference on Recent Advances in
849–68. Geotechnical Earthquake Engineering and Soil Dynamics, San Diego, California..
[4] Song J, Gao YF, Feng TG. Influence of interactions between topographic and soil [22] Newmark NM. A method of computation for structural dynamics. J. Eng. Mech.
layer amplification on seismic response of sliding mass and slope displacement. Div,ASCE 1959;85(3):67–94.
Soil Dynam Earthq Eng 2020;129:105901. [23] Hashash YMA, Musgrove MI, Harmon JA, Ilhan O, Xing G, Numanoglu O,
[5] Lin JS, Whitman RV. Decoupling approximation to the evaluation of earthquake- Groholski DR, Phillips CA, Park D. DEEPSOIL 7, user manual. Urbana, IL: Board of
induced plastic slip in dams. Earthq Eng Struct Dynam 1983;11(5):667–78. Trustees of University of Illinois at Urbana-Champaign; 2020.
[6] Kramer SL, Smith MW. Modified Newmark model for seismic displacements of [24] Vucetic M, Dobry R. Effect of soil plasticity on cyclic response. J Geotech
compliant slopes. J Geotech Geoenviron Eng 1997;123(7):635–44. Geoenviron Eng 1991;117(1):89–107.
[7] Rathje EM, Bray JD. An examination of simplified earthquake-induced [25] Rathje EM, Abrahamson NA, Bray JD. Simplified frequency content estimates of
displacement procedures for earth structures. Can Geotech J 1999;36(1):72–87. earthquake ground motions. J Geotech Geoenviron Eng 1998;124(2):150–9.
[8] Rathje EM, Bray JD. Nonlinear coupled seismic sliding analysis of earth structures. [26] Tripe R, Kontoe S, Wong TKC. Slope topography effects on ground motion in the
J Geotech Geoenviron Eng 2000;126(11):1002–14. presence of deep soil layers. Soil Dynam Earthq Eng 2013;50(7):72–84.
[9] Song J, Rodriguez-Marek A. Sliding displacement of flexible earth slopes subject to [27] Wang Z, Su L, Shi W, Zhang YJ, Ling XZ. Reinforcement effect of landslide on
near-fault ground motions. J Geotech Geoenviron Eng 2015;141(3):04014110. different support structures under earthquake. J Eng Geol 2020. https://doi.org/
[10] Tropeano G, Chiaradonna A, D’Onofrio A, Silvestri F. An innovative computer code 10.13544/j.cnki.jeg.2020-487 [In Chinese)].
for 1D seismic response analysis including shear strength of soils. Geotechnique [28] Rathje EM, Bray JD. One-and two-dimensional seismic analysis of solid-waste
2016;66(2):95–105. landfills. Can Geotech J 2001;38(4):850–62.
[11] Song J, Wu KL, Feng TG, Gao YF. Coupled analysis of earthquake-induced [29] Spencer E. A method of analysis of the stability of embankments assuming parallel
permanent deformations at shallow and deep failure planes of slopes. Eng Geol inter-slice forces. Geotechnique 1967;e17(1):11–26.
2020;274:105688. [30] Tropeano G, Chiaradonna A, D’Onofrio A, Silvestri F. A numerical model for non-
[12] Katsenis LC, Stamatopoulos CA, Panoskaltsis VP, Di BF. Prediction of large seismic linear coupled analysis of the seismic response of liquefiable soils. Comput Geotech
sliding movement of slopes using a 2-body non-linear dynamic model with a 2019;105:211–27.
rotating stick-slip element. Soil Dynam Earthq Eng 2020;129:105953. [31] Jafarian Y, Lashgari A. Simplified procedure for coupled seismic sliding movement
[13] Ji J, Wang CW, Cui HZ, Li XY, Song J, Gao YF. A simplified nonlinear coupled of slopes using displacement-based critical acceleration. Int J GeoMech 2016;16
Newmark displacement model with degrading yield acceleration for seismic slope (4):04015101.
stability analysis. Int J Numer Anal Methods GeoMech 2021;45(10):1303–22. [32] Di BF, Katsenis L, Stamatopoulos CA, Panoskaltsis VP. Sliding element simulating
[14] Du W, Wang G, Huang D. Influence of slope property variabilities on seismic the response of slip surfaces and its application for the prediction of earthquake-
sliding displacement analysis. Eng Geol 2018;242:121–9. induced landslide movement using one-dimensional dynamic analyses. Landslides
[15] Stewart JP, Bray JD, McMahon DJ, Smith PM, Kropp AL. Seismic performance of 2021;18(6):2297–307.
hillside fills. J Geotech Geoenviron Eng 2001;127(11):905–19.

13

You might also like