You are on page 1of 13

Soil Dynamics and Earthquake Engineering 139 (2020) 106310

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

Rock slope reliability analysis using Barton-Bandis failure criterion with


modified pseudo-dynamic approach
Lianheng Zhao a, b, Chenghao Yu a, Liang Li a, Aijun An c, Zhihong Nie a, Anping Peng d,
Shi Zuo a, *
a
School of Civil Engineering, Central South University, Changsha, Hunan, 410075, China
b
Key Laboratory of Heavy-haul Railway Engineering Structure, Ministry of Education, Central South University, Changsha, Hunan, 410075, China
c
China Road and Bridge Engineering Co Ltd, Beijing, 100011, China
d
Hunan Road and Bridge Construction Group Ltd, Hunan Communication & Water Conservancy Group Ltd, Changsha, Hunan, 410075, China

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

Keywords: A modified pseudo-dynamic approach was used for a reliability-based stability analysis of plane failure rock
Rock slope slopes. The Barton-Bandis nonlinear failure criterion and the limit equilibrium principle were adopted to account
Modified pseudo-dynamic method for the nonlinear characteristics of rock joint shear strength and safety factor of slope, respectively. The Monte
Barton-bandis failure criterion
Carlo simulation method was used to analyze the rock slopes. Comparisons of the present method to a numerical
System reliability
Monte Carlo
simulation and to a pseudo-static case, were implemented. The influence of different parameters and tension
cracks on slope stability was analyzed. Results showed that the stability of the rock slope decreases with
increasing rock unit weight, horizontal seismic coefficients and vertical seismic coefficients, and increases with
increasing residual friction angle, joint roughness coefficient and joint compressive strength. The location of the
tension crack has significant influence on the stability of rock slopes, while the influence of inclination of the
tension crack is negligible.

that the B–B joint model is the most realistic empirical model for pre­
1. Introduction dicting the shear failure behavior of rough joints; hence, it is widely used
in multi aspect slope stability analysis [8–10]. Furthermore, Prassetyo
Earthquakes are destructive natural disasters. In mountainous re­ et al. [11] discussed the linearization of the B–B failure criterion and
gions, earthquakes are a major cause of rock slope collapse [1–3]. developed a method to determine the equivalent M − C model
Consequently, the analysis of rock slope seismic stability is critical and parameters.
researchers have developed multiple methods to estimate the stability of Plane failure is a type of rock failure that is used to reflect the
rock slopes under seismic conditions. Traditionally, rock slope stability sensitivity of rock slope stability to changes in shear strength, seismic
analysis has employed predefined parameters to obtain a slope safety force, and other forces. To estimate the influence of uncertainties on
factor. However, uncertainties in rock mass, environmental changes, slope failure probability, studies have introduced reliability theory into
and design parameter assumptions require consideration when evalu­ slope stability analysis. Simulation methods, point-estimate methods,
ating rock slope stability. The slope reliability analysis method is an and the first order-second moment (FOSM) method have been variously
effective technique for quantifying slope engineering uncertainty. applied to slope stability problems [12–15]. Low [16] divided slope
Disturbed joints in rock masses create weak structural surfaces, failures into four failure modes and investigated slope stability using
significantly impacting rock slope stability. The Mohr-Coulomb (M − C) system reliability theory. Jimenez-Rodriguez [17] discussed several
joint model has been widely used to estimate the shear strength of joint failure modes and the system reliability of rock slopes comprised of two
surfaces in slope stability analysis. However, studies have shown that blocks linked by joints. The Monte Carlo simulation method is widely
the M − C criterion over-estimates the shear strength of rocks under used in slope stability analysis because it can directly determine slope
normal, low stress conditions [4,5]. Using extensive experimental reliability using a computer. For example, Tamimi et al. [12] investi­
studies, Barton et al. [4–6] proposed the Barton-Bandis (B–B) nonlinear gated the reliability of a rock slope with plane failure using Monte Carlo
shear strength criterion for rock joints. Current research [5–7] suggests simulations, while Zhao et al. [18,19] used Monte Carlo simulations and

* Corresponding author.
E-mail address: zuoshi@csu.edu.cn (S. Zuo).

https://doi.org/10.1016/j.soildyn.2020.106310
Received 23 January 2020; Received in revised form 23 June 2020; Accepted 6 July 2020
Available online 23 August 2020
0267-7261/© 2020 Elsevier Ltd. All rights reserved.
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

List of notations Qh , ​ Qv Horizontal and vertical inertial force acting on the whole
slope
ah (z, t), ​ av (z, t) Horizontal and vertical acceleration at any depth, z t Time
and time, t T Period of the harmonic seismic acceleration
AA The area of contact of block A and the failure plane uh (z,t), ​ uv (z,t) Horizontal and vertical displacement at any depth, z
AB The area of contact of block B and the failure plane and time, t
c The cohesion of rock Vs , ​ Vp Velocity of P - and S - waves in the rock
D Damping ratio W, ​ WA , ​ WB The weight of the rock slope, block A and block B
E The elastic modulus of rock respectively
Fdriving , ​ Fresist The driving force and resisting force of block X Length of rock slope
FS Safety factor against plane failure mode XA , XB Location of tension crack
FSA Safety factor of block A y1 ks1 H
FSB Safety factor of block B y2 ks2 H
g Acceleration due to gravity z Depth from slope top
G The shear moduli of rock β The slope face angle
h The distance between the slope top and the bottom of the δ The dip of the tension crack
tension crack γ The unit weight of rock
H Slope height ϕ The friction angle of rock
JCS The joint compressive strength ϕr The residual friction angle
JRC The joint roughness coefficient θ The dip of the slope failure plane
kh , ​ kv Horizontal and vertical seismic acceleration coefficient σn The effective normal stress of joint
kn , ​ ks The normal and shear stiffness of the joints τ The shear stress of joint
K The bulk moduli of rock υ Poisson’s ratio
l The distance between the slope top and the top of tension ω Angular frequency of motion = 2π/T
crack ξA XA /H
m(z), ​ mB (z) The mass of different elemental strip at depth z ξB XB /H
QAh , ​ QAv Horizontal and vertical inertial force acting on block A ξh h/H
QBh , ​ QBv Horizontal and vertical inertial force acting on block B ξl l/H

the B–B model to analyze hydraulic impacts on the stability of plane rock 2. The modified pseudo-dynamic method
slopes with inclined cracks.
The dynamic response of engineering structures is complex; hence, Bellezza [24] re-derived the horizontal and vertical seismic accel­
many simplified methods have been adopted when conducting slope eration. The rock slope is assumed as a visco-elastic material and
stability analysis. In the seismic stability analysis, magnitude of the modeled as a Kelvin–Voigt medium, the velocities of shear and primary
dynamic force increment due to shaking is calculated using the Mono­ waves in the considered visco-elastic rock slope are VS and VP respec­
nobe–Okabe approach or pseudo-static method [20,21]. However, the tively. Energy dissipation characteristics of rock slope are considered by
pseudo-static method considers seismic response in a very simplified incorporating into damping ratio. A horizontal seismic base shaking of
manner and does not account for time. To overcome this, Steedman and period T is taken into account, the following two boundary conditions
Zeng [22,23] proposed a simplified pseudo-dynamic method and are assumed [25,26]: (1) shear stress at the ground surface is zero (z =
compared their theoretical results with the test results of a centrifugal 0). (2) displacement at the base is uhb = uh0 cos (ωt), so the horizontal
model to verify the accuracy of the method. However, the Steedman and displacement can be written:
Zeng method does not satisfy boundary conditions and ignores material uh0
damping. Based on visco-elastic behavior of the geomaterials, Bellezza uh (z, t) = [(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)] (1)
CS2 + SS2
[24] improved the Steedman and Zeng method, overcoming some of the
limitations of the original method. The modified pseudo-dynamic where:
method satisfies the zero-stress boundary condition at the free ground (y z) (y z)
surface and considers acceleration amplification. (2a)
s1 s2
CSZ = cos cosh
H H
The review of the above literature shows that the seismic reliability
assessment of plane failure rock slopes, considering the nonlinear (y z) (y z)
(2b)
s1 s2
SSZ = − sin sinh
characteristics of the shear strength of rock and characteristics of H H
earthquake ground motions, has not received any attention. Therefore,
this study adopted the Barton-Bandis failure criterion and proposed an CS = cos(ys1 )cosh(ys2 ) (2c)
efficient approach for evaluating the reliability of rock slopes under
SS = − sin(ys1 )sinh(ys2 ) (2d)
seismic load in the framework of a modified pseudo-dynamic method.
Firstly, limit state functions for a rock slope with plane failure were √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅
established. Then, Monte Carlo simulation was implemented to evaluate ys1 =
ωH 1 + 4D2 + 1
( ) (2e)
the probability of failure and reliability indices of the rock slope. The Vs 2 1 + 4D2
numerical simulation method and the pseudo-static method were
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅
applied for comparison with the presented method to validate the ac­ ωH 1 + 4D2 − 1
curacy of the presented method. Finally, reliability analysis examples ys2 = − ( ) (2f)
Vs 2 1 + 4D2
were carried out to investigate the influence of various parameters
including the location and inclination of tension cracks on the stability
of the rock slope.

2
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

kh g 3.2. Method of limit analysis


ah (z, t) = [(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)] (3)
CS2 + SS2
Following Jimenez [17], a simple sliding mass composed of two
where khg = − ω 2uh0, kh-horizontal seismic coefficient. ys1 and ys2 are blocks separated by a tension crack is considered, the tension crack is
respectively a function of the normalized frequency ωH/Vs and damping considered inclined in this study, as shown in Fig. 1. W is the weight of
ratio D, H is height, z is depth, Vs is velocity of S-wave propagating, ω = the rock slope. Qh(t) and Qv(t) are respectively horizontal and vertical
2π/T is angular frequency, T is period, t is time, ah (z, t) is horizontal inertial force derived using the modified pseudo-dynamic methods.
acceleration. Generally, for plane failure of the considered rock slope to occur, four
The whole horizontal seismic inertial force can be expressed as (for geometrical conditions must be satisfied [29]: (a) a continuous plane on
rock): which sliding occurs must strike parallel or nearly parallel (within
∫H approximately ±20◦ ) to the slope face, (b) the dip of the failure plane
Qh (t) = m(z)ah (z, t)dz (4) must be less than the dip of the slope face, (c) the dip of the failure plane
0
must be greater than the angle of friction of this plane, and (d) surfaces
The vertical seismic acceleration is shown as: of separation that provide negligible resistance to sliding must be pre­
sent in the rock mass to define the lateral boundaries of the failing block.
kv g
av (z, t) = [(CP CPZ + SP SPZ )cos(ωt) + (SP CPZ + CP SPZ )sin(ωt)] (5) In the stability analysis of a rock slope with plane failure, the limit
CP2 + SP2
equilibrium method is employed. The safety factor for the considered
rock slope with plane failure is calculated by resolving all forces acting
where
on the slope into components parallel and normal to the sliding plane.
(y z) (y z)
CPZ = cos
p1
cosh
p2
(6g) The vector sum of the shear forces acting down the plane is termed the
H H driving force Fdriving. The product of shear strength obtained by Barton-
(y z) (y z) Bandis failure criterion and area of sliding plane is termed the resisting
(6h)
p1 p2
SPZ = − sin
H
sinh
H force Fresist. the safety factor of the rock slope is the ratio of the resisting
forces to the driving forces [28,29]:
( ) ( )
CP = cos yp1 cosh yp2 y (6i) Fresist
FS = (9)
( ) ( ) Fdriving
SP = − sin yp1 sinh yp2 (6j)
In this analysis, two different cases should be identified, depending
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√√ ̅
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ on the interaction between blocks A and B as follows. In case 1, block B is
ωH 1 + 4D2 + 1
yp1 = ( ) (6k) stable, and there is no interaction between blocks. The safety factor of
Vp 2 1 + 4D2
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√√ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅
ωH 1 + 4D2 − 1
yp2 = − ( ) (6l)
Vp 2 1 + 4D2

where ω = 2π/T, kvg = − ω2uv0, kv-vertical seismic acceleration coeffi­


cient at the base. yp1 and yp2 are respectively a function of the normal­
ized frequency ωH/Vp and damping ratio D, H is height, z is depth, Vp is
velocity of P-wave propagating, ω = 2π/T is angular frequency, T is
period, t is time, av (z, t) is vertical acceleration.
The whole vertical seismic inertial force can be expressed as (for
rock):
∫H
Qv (t) = m(z)av (z, t)dz (7)
0

3. Model formulation

3.1. The nonlinear Barton-Bandis Failure criterion

Barton [4–6]proposed the Barton-Bandis failure criterion after con­


ducting a large number of shear tests on the surface of a natural struc­
ture. In the non-linear Barton-Bandis failure criterion, the shear strength
of joints is explained with roughness, compressive strength, and residual
friction angle of joint. The nonlinear formula for joint shear strength is:
[ ( )]
JCS
τ = σ n tan φr + JRClog10 (8)
σn

where τ is the shear stress of joint, σn is the effective normal stress of


joint, JRC is the joint roughness coefficient, JCS is the joint compressive
strength, and ϕr is the residual friction angle of joint. The values of the
parameters JRC, JCS, and ϕr were determined as suggested in the
existing literature [27,28].

Fig. 1. Geometrical definition of the rock slope stability model: (a) tension
crack at slope face, (b) tension crack at slope top.

3
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

block A and block B are calculated independently. In case 2, block B is where γ is the unit weight of rock, β is the slope face angle, δ is the dip of
unstable, an interaction force IF due to block B being unstable is acting the tension crack, m(z) and mB(z) are the mass of elemental strip in
on block A. different depth, ξh and ξl are geometry parameters as shown in Fig. 1,
other different parameters are shown in the above Eqs. (1)–(6). m(z) and
3.2.1. Case 1. No interaction between blocks mB(z) are given by:
Zhao et al. [19] proposed safety factor of rock slope based on γ
Barton-Bandis failure criterion. According to the approach proposed by m(z) = (H − z)(cot θ − cot β)dz (15)
g
Zhao et al. the safety factor of block B based on Barton-Bandis failure
criterion is determined: γ
mB (z) = (h − z)(cot θ − cot δ)dz (16)
⎧ ⎡ ⎤⎫ g

⎪ ⎪

⎨ ⎢ ⎥⎬ For the tension crack located at the top of the slope as shown in Fig. 1
[(WB − QBv )cos θ − QBh sin θ]⋅tan ϕr + JRClog10 ⎢ JCS ⎥


⎣(WB − QBv )cos θ− QBh sin θ⎦⎪
⎪ (b), WB, QBh and QBv are given by:
⎩ AB

FSB = 1
(WB − QBv )sin θ + QBh cos θ WB = γH 2 ξh csc θ csc δ sin(δ − θ) (17)
2
(10)
∫h
where JRC is the joint roughness coefficient of the failure plane, JCS is
QBh = mB (z)ah (z, t)
the joint compressive strength of failure plane, and ϕr is the residual
friction angle of failure plane, θ is the dip of the slope failure plane, AB is
0

the area of contact with the failure surface, QBh is the horizontal inertial ∫h
kh γ(h − z)(cot θ − cot δ)
force acting on block B, QBv is the vertical inertial force acting on block = *[(CS CSZ + SS SSZ )cos(ωt)
CS2 + SS2
B. AB is given by: 0

+(SS CSZ + CS SSZ )sin(ωt)]dz (18)


AB = h csc θ (11)
∫h
where h is the distance between the slope top and the bottom of the
tension crack. QBv = mB (z)av (z, t)
The weight of block B and seismic inertial force acting on block B 0

depend on the location of tension crack. For the tension crack located at ∫h
the slope face as shown in Fig. 1(a), WB, QBh and QBv are given by: =
kv γ(h − z)(cot θ − cot δ)
*[(CP CPZ + SP SPZ )cos(ωt)
CP2 + SP2
1 0
WB = γH 2 (ξh + ξl − ξh ξl )csc θ csc β sin(β − θ) (12) +(SP CPZ + CP SPZ )sin(ωt)]dz (19)
2
Similarly, the safety factor of block A is expressed as:

∫l ∫h
QBh = m(z)ah (z, t) + mB (z)ah (z, t)
0 l

∫l
kh γ(H − z)(cot θ − cot β)
= *[(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)]dz (13)
CS2 + S2S
0

∫h
kh γ(h − z)(cot θ − cot δ)
+ *[(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)]dz
CS2 + SS2
l

∫l ∫h
QBv = m(z)av (z, t) + mB (z)av (z, t)
0 l

∫l
kv γ(H − z)(cot θ − cot β)
= *[(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)]dz (14)
CS2 + S2S
0

∫h
kv γ(h − z)(cot θ − cot δ)
+ *[(CS CSZ + SS SSZ )cos(ωt) + (SS CSZ + CS SSZ )sin(ωt)]dz
CS2 + SS2
l

4
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

⎧ ⎡ ⎤⎫

⎪ ⎪

⎨ ⎢ ⎥⎬
⎢ JCS ⎥
[(WA − QAv )cos θ − QAh sin θ]⋅tan ϕr + JRClog10 ⎣(WA − QAv )cos θ− QAh sin θ⎦

⎪ ⎪

⎩ AA

FSA =
(WA − QAv )sin θ + QAh cos θ
(20)

where AA is the area of contact of the failure plane, QAh is the horizontal
inertial force acting on block A, QAv is the vertical inertial force acting on
block A. AA is given by:
AA = (H − h)csc θ (21)
Again, the weight of block A and seismic inertial force acting on
block A depend on the location of tension crack. For the tension crack
located at the slope face, WA, QAh and QAv are given by: Fig. 2. Model geometry and boundary conditions of the rock slope model
1 using UDEC.
WA = γH 2 (1 − ξh − ξl − ξh ξl )csc θ csc β sin(β − θ) (22)
2

∫H where ψ = θ − ϕAB − δ is the angle between the direction of interaction


QAh = Qh − QBh = m(z)ah (z, t) − QBh force IF and the normal of the failure plane. All the parameters in Eqs.
0 (25) and (26) are as defined previously. In case 2, IF is calculated by
making FSB = 1 in Eq. (25), and safety factor of block A FSA can be
∫H
kh γ(H − z)(cot θ − cot β) obtained by substituting IF into Eq. (26). The slope is considered to be
= *[(CS CSZ + SS SSZ )cos(ωt)
CS2 + SS2 stable if FSA > 1 and unstable otherwise.
0

+(SS CSZ + CS SSZ )sin(ωt)]dz − ​ QBh ​ (23)


4. Numerical simulation analysis
∫H
QAv = Qv − QBv = m(z)av (z, t) − QBv The universal distinct element code (UDEC) is commonly used to
study the failure mechanisms of jointed slopes. UDEC solves the contact
0
problem between blocks very well, and the constitutive models provided
∫H
kv γ(H − z)(cot θ − cot β) in UDEC can be used to simulate the failure of intact rock. Therefore, it
= *[(CP CPZ + SP SPZ )cos(ωt) was employed in this study to validate the accuracy of the presented
CP2 + SP2
0 method.
+(SP CPZ + CP SPZ )sin(ωt)]dz − ​ QBv ​ (24)

where Qh and Qv are horizontal and vertical inertial force acting on 4.1. Linearization of the Barton-Bandis failure criterion
whole slope, QBh and QBv are obtained by Eqs. 13–19.
UDEC applies the strength reduction method to calculate the safety
3.2.2. Case 2. interaction between blocks factor for a variety of different structures, and this method can also be
When block B is unstable (i.e., FSB < 1 in Eq. (10)), there will be a applied to the strength properties for joints. However, due to the
extra load due to block B acting on block A. If block A under the extra nonlinear characteristics of the B–B failure criterion, the safety factor of
load due to block B remains stable, then the slope is stable. The other is the slope with joints described by the B–B failure criterion cannot be
that block A is unstable, and consequently, the slope fails. The safety calculated by strength reduction technology. Given the need for realistic
factor corresponding to blocks A and B are similar to those presented in joint modeling, Prassetyo et al. [11] developed a linearized imple­
Eqs. (10) and (20) with consideration of interaction force IF between mentation of the B–B failure criterion as follows:
blocks. It is assumed that the interaction force IF has a direction inclined ϕt = arctan(A − B) (27)
at an angle ϕAB with respect to the surface of the tension crack as shown
in Fig. 1. Similar to the derivation above, the safety factor of block A and ct = σ n B (28)
block B are expressed as:

⎧ ⎡ ⎤⎫

⎪ ⎪

⎨ ⎢ ⎥⎬
[(WB − QBv )cos θ − QBh sin θ − IF cos ψ ]tan ϕr + JRClog10 ⎢ JCS
⎣(WB − QBv )cos θ− QBh sin θ+IF

cos ψ ⎦

⎪ ⎪

⎩ AB

FSB = (25)
(WB − QBv )sin θ + QBh cos θ + IF sin ψ
⎧ ⎡ ⎤⎫

⎪ ⎪

⎨ ⎢ ⎥⎬
[(WA − QAv )cos θ − QAh sin θ − IF cos ψ ]tan ϕr + JRClog10 ⎢ JCS
⎣(WA − QAv )cos θ− QAh sin θ− IF

cos ψ ⎦

⎪ ⎪

⎩ AA

FSA = (26)
(WA − QAv )sin θ + QAh cos θ − IF sin ψ

5
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

Table 1 5. Reliability analysis


Basic parameters of rock slopes adopted in the numerical simulation.
Parameters Value 5.1. Limit state function and system composition
The friction angle of rock, ϕ( )

45
The cohesion of rock, c (Mpa) 3.0 Once the safety factor is identified, the limit state function of the rock
The elastic modulus of rock, E (GPa) 45 slope system can be obtained as:
The bulk moduli of rock, K (Gpa) 30
The shear moduli of rock, G (Gpa) 15 g(x) = FS − 1 (31)
Poisson’s ratio, υ 0.25
The normal stiffness of the joints, kn (GPa) 50 where x is a vector of the variables and FS is the safety factor. If g(x) < 0,
The shear stiffness of the joints, ks (GPa) 25 then the rock slope is in a failure domain, and if g(x)≥0, then the rock
slope is stable. The limit state functions based on the related equations
[ ( ) ] are given in Table 3 and the system composition of the overall slope
JCS
A = tan JRClog10 + ϕr (29) system is shown in Fig. 3.
σn

( ) JRC 5.2. Reliability analysis with Monte Carlo simulation


B = 1 + A2 (30)
ln 10
In reliability theory, physical systems are composed of multiple
where ϕt is the equivalent friction angle and ct is the equivalent components, including series, parallel, and combined systems. Com­
cohesion. bined systems can be classified into cut-set (series-parallel) and link-set
(parallel-series) systems. Further details regarding these systems are
4.2. Model configuration available in the literature [31–33].
We consider a slope system as an assembly of components. Such
The pseudo-static analysis is a special case of the pseudo-dynamic combined systems can be analyzed by calculating the reliabilities for the
analysis. Hence, it can be readily transformed from the pseudo- individual series and parallel sections and then combining them in the
dynamic analysis. The numerical simulation applied the pseudo-static appropriate manner. Therefore, the probability of failure (Pf) of each
method to conduct a seismic stability analysis. Using UDEC, a 15 m subsystem is first evaluated, then the overall Pf is usually obtained as
rock slope model was created to simulate seismic slope failure, as shown follows:
in Fig. 2.
The parameters for the numerical simulation are summarized in
Table 1. The properties of rock mass (e.g. cohesion, friction angle, tensile
strength, elastic modulus, and Poisson’s ratio) and the properties of Table 3
joints are referenced from Zheng et al. [30]. It should be noted that the Physical interpretation of limit state functions.
cohesion and friction angle of the intact rock are irrelevant for the slope Limit state function Physical interpretation Eqs
failure; the stability of the rock slope is controlled by the strength of the
g1 = FSB − 1 ≤ 0 Block B is unstable (32)
joints.
g2 = {FSA |g1 > 0} − Block A is unstable, given block B stable (no (33)
1≤0 interaction occurs)
4.3. Comparison between numerical simulation and presented method g3 = {FSA |g1 ≤ 0} − Block A is unstable, given block B unstable (34)
1≤0 (interaction occurs)
The following parameters are adopted in the theoretical and nu­
merical analysis: H = 15 m, γ = 25 kN/m3, ϕr = 30◦ , ϕAB = 20◦ , JCS = 50
Mpa, JRC = 9, T = 0.5 s, D = 0.3, θ = 30◦ , β = 60◦ , δ = 70◦ , and ξA = 0.5
(i.e., ξB = 0.5). In the numerical simulation, strength parameters of
joints (e.g. JRC, JCS, ϕr) are used to acquire the equivalent friction angle
ϕt and the equivalent cohesion ct by Eqs. 27 and 28. A comparative study
of the results is given in Table 2 for the range of values of kh = 0.0–0.3
and kv/kh = 0.0–0.1.
It can be noted from Table 2 that the theoretical results are
marginally different than the results obtained by the numerical simu­
lation. This marginal difference in the values of the safety factor of the
rock slope can be attributed to the different definition of safety factor
and linearization of the B–B failure criterion adopted in the numerical
Fig. 3. System formulation of plane rock slide stability problem.
simulation.

Table 2
A comparison of safety factor obtained by two methods.
Safety factor of rock slope (FSA/fos)

kh kv/kh = 0.0 kv/kh = 0.5 kv/kh = 1.0

Theoretical Numerical Theoretical Numerical Theoretical Numerical

0.00 1.940 1.902 1.940 1.902 1.940 1.902


0.05 1.740 1.738 1.741 1.738 1.741 1.738
0.10 1.569 1.574 1.564 1.566 1.564 1.566
0.15 1.422 1.416 1.404 1.404 1.404 1.404
0.20 1.294 1.287 1.260 1.271 1.260 1.271
0.25 1.181 1.193 1.129 1.143 1.129 1.143
0.30 1.081 1.096 1.009 1.018 1.009 1.018

6
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

Table 4
Statistics of stochastic variables.
Random variable Distribution Mean Coefficient of variable (COV) %

ϕr Log-Normal 30◦
15
ϕAB Log-Normal 20◦ 15
JCS Log-Normal 50 Mpa 20
JRC Log-Normal 9 10
Vs Normal 2500 m/s 5
Vp Normal 4700 m/s 5


Pf = P(g(x) ≤ 0) = ​ f (x) ​ dx (35)
g(x)≤0

where f(x) is the probability density function of variables.


Using the Monte Carlo simulation method, we generated a sequence
of N input vectors xi (i = 1, …, N) based on the assumed probability
distribution of the variables. The limit state function g (xi) (i = 1, …, N) Fig. 5. Probability of failure Pf against slope height H for different kh.
gives the state of the system under these random samples, and Ni, Nf,i
represent the numbers of occurrences and failures, respectively,
observed for subsystem i (when the factor of safety is less than 1.0). The
frequency of the subsystem Pi, as well as the likelihood of failure, Pf,i are
determined as follows [17]:
Ni
Pi = (36)
N

Nf ,i
Pf ,i = (37)
N
The failure probability of the system Pf,i is.

Nf ,i
Pf ,sys = (38)
N

where N is the total number of simulations.

5.3. Numerical implementation

In this study, a computer program was developed to obtain numer­


ical values of the proposed method. The program is coded by MATLAB to
calculate the seismic inertial force, reliability index and probability of
failure of the rock slope. According to the input parameters, the program
calculates the seismic inertial forces acting on the rock slope and im­ Fig. 6. System reliability indices β against unit weight of rock γ for different kh.
plements the Monte Carlo simulation for the reliability analysis. In order
to obtain a highly accurate result, the failure probability is solved by the 6. Results and discussion
Monte Carlo simulation with a simulation time of 5 * 106 (i.e., N = 5 *
106 in Eqs. 36–38). For the purpose of illustrating the type of result that can be obtained
using the methodology, a series numerical analyses were implemented.
The following numerical analyses include comparison of present study
and pseudo-static solution, the influence of seismic load, parameters of
rock mass properties and inclination of tension cracks on the reliability
of the rock slope. The stochastic parameters are given in Table 4.

6.1. Comparison of modified pseudo-dynamic method and pseudo-static


method

The pseudo-static method is widely used in seismic stability analysis


of slopes [20]. To verify the accuracy of the present method, a com­
parison between the pseudo-static case and the present method is
implemented. In the pseudo-static case, the seismic inertial forces of the
rock slope are considered as the product of the seismic acceleration
coefficient kh, kv and weight of block W (i.e., Qh = khW, Qv = kvW). Fig. 4
shows the comparison of reliability indices of the rock slope obtained by
the modified pseudo-dynamic method with those obtained by the
pseudo-static method, and Fig. 5 shows the comparison of the proba­
bility of failure of the rock slope obtained by the modified
pseudo-dynamic method with those obtained by the pseudo-static
Fig. 4. Reliability indices β against slope height H for different kh. method. The following values are used for the deterministic

7
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

method are slightly higher than the present method and the reliability
indices β is reversed. The modified pseudo-dynamic result is more
conservative than the pseudo-static result due to the consideration of the
dynamic effects of seismic loads.

6.2. Influence of self-weight on reliability of rock slope

Fig. 6 shows the influence of the unit weight γ of rock masses on the
reliability indices of rock slopes. The following values are used for the
deterministic parameters: kv = 0.1, H = 15 m, T = 0.5 s, D = 0.3, θ = 30◦ ,
β = 60◦ , δ = 70◦ , and ξA = 0.5 (i.e., ξB = 0.5). The stochastic parameters
are given in Table 4.
It can be observed from Fig. 6 that the value of system reliability
indices β decreases gradually as the unit weight of rock γ increases,
considering that the value of rock unit weight γ is limited, so conclusions
can be drawn that the unit weight of rock γ only has a little influence on
Fig. 7. Probability of failure Pf against residual friction angle ϕr for different kh. the stability of rock slopes.

6.3. Influence of parameters of rock mass properties ϕr and JRC on the


reliability of rock slope

To investigate the influence of the parameters of rock mass proper­


ties ϕr and JRC on the reliability of rock slopes, the following values are
used for the deterministic parameters: kv = 0.1, H = 15 m, γ = 25 kN/m3,
T = 0.5 s, D = 0.3, θ = 30◦ , β = 60◦ , δ = 70◦ , and ξA = 0.5 (i.e., ξB = 0.5).
The stochastic parameters are given in Table 4. Fig. 7 shows the influ­
ence of the residual friction angle ϕr on the probability of failure of the
rock slope and Fig. 8 shows the influence of the joint roughness coeffi­
cient JRC on the probability of failure of the rock slope.
Notably the residual friction angle has a significant influence on the
probability of failure of the rock slope in Fig. 7. When the value of ϕr
varies from 15◦ to 35◦ , for kh = 0.2, the value of Pf decreases from 0.24 to
4 × 10− 6. That is, the increase in the value of ϕr increases the stability of
the rock slope. Moreover, it can be observed in Fig. 8 that the joint
roughness coefficient JRC also has a significant influence on the stability
Fig. 8. Probability of failure Pf against joint roughness coefficient JRC for
different kh.
of the rock slope. For kh = 0.3, for example, as the value of JRC increases
from 4 to 14, the value of Pf decreases from 0.92 to 1.8 × 10− 5. Namely,
the increase in the value of JRC increases the stability of the rock slope.
parameters: kv = 0.1, γ = 25 kN/m3, T = 0.5 s, D = 0.3, θ = 30◦ , β = 60◦ ,
This discussion clearly demonstrates that the increase of the residual
δ = 70◦ , and ξA = 0.5 (i.e., ξB = 0.5). The stochastic parameters are given
friction angle ϕr and joint roughness coefficient JRC increases the sta­
in Table 4.
bility of the rock slope significantly.
As can be seen in Figs. 4 and 5, the value of the probability of failure
Pf and reliability indices β obtained by the present approach are in close
agreement with the value obtained by the pseudo-static method. Hence, 6.4. Influence of residual friction angle JCS on the reliability of rock slope
it confirms the validity of the present formulation. It can be seen that the
values of the probability of failure Pf obtained by the pseudo-static To investigate the influence of the joint compressive strength JCS on
the reliability of the rock slope, the following values are used for the

Fig. 9. System reliability indices β against horizontal seismic coefficient kh for Fig. 10. System reliability indices β against horizontal seismic coefficient kh for
different JCS. different JCS.

8
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

stochastic parameters are given in Table 4. Fig. 10 depicts the effect of


the vertical seismic coefficient kv on the reliability indices of the rock
slope.
It can be observed from Fig. 10 that as the value of kv increases from
0 to kh, the value of the system reliability indices β decreases slightly. For
example, when the value of kv increases from 0 to kh for kh = 0.25, the
value of β decreases from 1.41 to 0.72, so it can be concluded that the
vertical seismic coefficient kv has little effect on the stability of the rock
slope.

6.6. Influence of inclined tension crack on the reliability indices of rock


slope

To investigate the influence of location and inclination of tension


cracks on the reliability of the rock slope, the following values are used
for the deterministic parameters: kh = 0.2, kv = 0.1, H = 15 m, γ = 25
Fig. 11. System reliability indices β against location of tension crack ξA for kN/m3, T = 0.5 s, D = 0.3, θ = 30◦ , and β = 60◦ . The stochastic pa­
different kh. rameters are given in Table 4. Fig. 11 shows the influence of the location
of the tension crack ξА on the reliability indices of the rock slope and
Fig. 12 shows the influence of the inclination of the tension crack δ on
the reliability indices of the rock slope.
It can be obseeved in Fig. 11 that the location of the tension crack has
significant influence on the system reliability of the rock slope; the value
of β increases as the value of ξA increases. For example, for kh = 0.1, as
the value of ξA increases from 0.1 to 0.6, the value of β increases by
about 55%, which demonstrates that ξA has significant influence on the
stability of the rock slope. Fig. 12 shows that the inclination of the
tension crack has a negligible influence on the stability of the rock slop.
For example, for kh = 0.3, as the value of δ increases from 70◦ to 90◦ , the
value of β increases about 19%; while for kh = 0.2, the value of β in­
creases approximately by 11%. This discussion clearly demonstrates that
the location of the tension crack ξA has significant influence on the
stability of the rock slope while the inclination of the tension crack δ has
a negligible effect on the stability of the rock slope.

7. Conclusion

Fig. 12. System reliability indices β against inclination of tension crack δ for In this study, the Barton-Bandis failure criterion is adopted, and a
different kh.
modified pseudo-dynamic method is presented to perform a reliability
analysis of a plane failure rock slope. A Monte Carlo simulation is
deterministic parameters: kv = 0.1, H = 15 m, γ = 25 kN/m3, T = 0.5 s, D implemented to evaluate the reliability of the rock slope. The accuracy
= 0.3, θ = 30◦ , β = 60◦ , δ = 70◦ , and ξA = 0.5 (i.e., ξB = 0.5). The sto­ of the presented method is verified by comparison with numerical
chastic parameters are given in Table 4. Fig. 9 depicts the effect of joint simulation and pseudo-static results. The influence of the rock self-
compressive strength JCS on the reliability indices of the rock slope. weight γ, horizontal seismic coefficient kh, vertical seismic coefficient
It can be observed that the horizontal seismic coefficient has a sig­ kv, residual friction angle ϕr, joint roughness coefficient JRC, joint
nificant influence on the system reliability of rock slopes in Fig. 9. For compressive strength JCS, and location and inclination of the tension
example, for JCS = 50 Mpa, the value of the system reliability indices β crack ξA, δ are investigated, and the conclusions are as follows:
decreases by about 93% when the value of kh is increased from 0.1 to
0.3. Namely, the increase in the value of kh decreases the stability of rock (1) The results obtained by the present method are in close agree­
slopes. Meanwhile, Fig. 8 also shows that the influence of the joint ment with the case that adopted pseudo-static method. The re­
compressive strength JCS on the stability of the rock slope is significant. sults obtained by the modified pseudo-dynamic method are more
For kh = 0.2, for example, as the value of JCS increases from 30 Mpa to conservative than the pseudo-static results.
90 Mpa, the value of β increases by about 37%. That is, the increase in (2) The increase in the value of the rock unit weight γ decreases the
the value of JCS increases the stability of rock slopes. Thus, it can be stability of the rock slopes, but the influence of the unit weight of
concluded that the influence of the horizontal seismic coefficient kh and rock γ on the stability of the rock slopes is negligible.
joint compressive strength JCS on the stability of rock slopes is (3) The stability of the rock slopes improves as the values of the re­
significant. sidual friction angle ϕr and joint roughness coefficient JRC in­
crease. The residual friction angle ϕr and joint roughness
6.5. Influence of vertical seismic acceleration coefficient kv of rock on the coefficient JRC have a significant influence on the stability of the
reliability of rock slope rock slopes.
(4) The increase of the horizontal seismic coefficient kh decreases the
When influence of the vertical seismic acceleration coefficient kv on stability of the rock slopes, while the increase of joint compres­
the reliability of the rock slope is considered, the following values are sive strength JCS increases the stability of the rock slopes.
used for the deterministic parameters: H = 15 m, γ = 25 kN/m3, T = 0.5 (5) The vertical seismic coefficient kv has a limited influence on the
s, D = 0.3, θ = 30◦ , β = 60◦ , δ = 70◦ , and ξA = 0.5 (i.e., ξB = 0.5). The stability of the rock slopes, and the stability of the rock slopes

9
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

decreases as the value of the vertical seismic coefficient kv Declaration of competing interest
increases.
(6) The influence of the location of the tension crack ξA on the sta­ The authors declare that they have no known competing financial
bility of the rock slopes is significant while the inclination of the interests or personal relationships that could have appeared to influence
tension crack δ has little effect on the stability of the rock slopes. the work reported in this paper.
Therefore, the tension cracks many research studies are consid­
ered vertical. Acknowledgments

CRediT authorship contribution statement This work was supported by the Guizhou Provincial Department of
Transportation Foundation (No.2013121015, 2018123040), the Funded
Lianheng Zhao: Software, Formal analysis, Writing - original draft. projects of The Fifth Civil Engineering Co., Ltd of CREC Shanghai Group
Chenghao Yu: Software, Formal analysis. Liang Li: Software, Writing - (No. GYSLHKG-JSFW-03) and the Fundamental Research Funds for the
review & editing. Aijun An: Formal analysis, Writing - review & editing. Central Universities of Central South University (No. 2019zzts009). The
Zhihong Nie: Writing - review & editing. Anping Peng: Writing - re­ financial supports are gratefully acknowledged.
view & editing. Shi Zuo: Writing - review & editing, Supervision.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.soildyn.2020.106310.

Appendix

A1. Geometry of rock slope model

With reference to Fig. 1, the following equations can be derived: for Fig. 1(a):
( cot β cot δ )
ξh = (tan β − tan θ)⋅ − ξA cot θ (A.1)
cot θ − cot δ

ξl = tan θ tan β − ξA (A.2)

for Fig. 1(b):


( )
1 + cot θ
ξh = (1 − ξA )⋅ (A.3)
cot θ − cot δ

A2. Seismic inertial force

The expression of seismic inertial forces acting on blocks are reported in this part. The expression of horizontal and vertical inertial forces Qh, Qv
acting on whole slope are given by:
[ ]
CS Is1 + SS Is2 SS Is1 − CS Is2
Qh = kh γ(cot θ − cot β) 2 2
cos(ωt) + 2 2
sin(ωt) (A.4)
CS + SS CS + SS
[ ]
CP Ip1 + SP Ip2 SP Ip1 − CP Ip2
Qv = kv γ(cot θ − cot β) cos(ω t) + sin(ωt) (A.5)
CP2 + SP2 CP2 + SP2

where Is1, Is2, Ip1 and Ip2 are given by:


∫H
Is1 = CSZ (H − z)dz
0
(A.6)
H2 { ( ) ( )}
=( )2 2ys1 ys2 sin(ys1 )sinh(ys2 ) − y2s1 − y2s2 cos(ys1 )cosh(ys2 ) + y2s1 − y2s2
y2s1 + y2s2

∫H
Is2 = SSZ (H − z)dz
0
(A.7)
H2 {( ) }
=( )2 y2s1 − y2s2 sin(ys1 )sinh(ys2 ) + 2ys1 ys2 cos(ys1 )cosh(ys2 ) − 2ys1 ys2
y2s1 + y2s2

10
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

∫H
Ip1 = CPZ (H − z)dz
0
(A.8)
H2 { ( ) ( ) ( ) ( ) ( ) ( )}
=( )2 2yp1 yp2 sin yp1 sinh yp2 − y2p1 − y2p2 cos yp1 cosh yp2 + y2p1 − y2p2
y2p1 + y2p2

∫H
Ip2 = SPZ (H − z)dz
0
(A.9)
H2 {( ) ( ) ( ) ( ) ( ) }
=( )2 y2p1 − y2p2 sin yp1 sinh yp2 + 2yp1 yp2 cos yp1 cosh yp2 − 2yp1 yp2
y2p1 + y2p2

The horizontal and vertical seismic inertial force acting on block B QBh, QBv is depended on the location of tension crack. For the tension crack
located at slope face, QBh and QBv are given by:
[ ]
CS Ils1 + SS Ils2 SS Ils1 − CS Ils2
QBh = kh γ(cot θ − cot β) cos( ω t) + sin(ω t)
CS2 + SS2 CS2 + SS2
[ ] (A.10)
CS Ihs1 + SS Ihs2 SS Ihs1 − CS Ihs2
+kh γ(cot θ − cot δ) cos( ω t) + sin(ω t)
CS2 + SS2 CS2 + SS2
[ ]
CP Ilp1 + SP Ilp2 SP Ilp1 − CP Ilp2
QBv = kv γ(cot θ − cot β) 2 2
cos(ωt) + 2 2
sin(ωt)
CP + SP CP + SP
[ ] (A.11)
CP Ihp1 + SP Ihp2 SP Ihp1 − CP Ihp2
+kv γ(cot θ − cot δ) 2 2
cos(ωt) + 2 2
sin(ωt)
CP + SP CP + SP

where Ils1, Ils2, Ilp1 and Ilp2 are given by:


∫l
Ils1 = CSZ (H − z)dz
0

=(
1 { ( 2
2ks1 ks2 sin(ks1 l)sinh(ks2 l) + ks2 2
− ks1
) ( 2
cos(ks1 l)cosh(ks2 l) + ks2 2
− ks1
)} (A.12)
2 2 2
)
ks1 + ks2

l− H
− 2 2
{ks2 cos(ks1 l)sinh(ks2 l) + ks1 sin(ks1 l)cosh(ks2 l)}
ks1 + ks2

∫l
Ils2 = SSZ (H − z)dz
0

=(
1 {( 2
ks2 2
− ks1
)
sin(ks1 l)sinh(ks2 l) + 2ks1 ks2 cos(ks1 l)cosh(ks2 l) − 2ks1 ks2
} (A.13)
2 2 2
)
ks1 + ks2

l− H
+ 2 2
{ks2 sin(ks1 l)cosh(ks2 l) − ks1 cos(ks1 l)sinh(ks2 l)}
ks1 + ks2

∫l
Ilp1 = CPZ (H − z)dz
0

1 { ( ) ( ) ( 2 ) ( ) ( ) ( 2 )}
=( 2 2
)2 2kp1 kp2 sin kp1 l sinh kp2 l + kp2 − kp1 cos kp1 l cosh kp2 l + kp2 − kp1 (A.14)
2 2
kp1 + kp2

l− H { ( ) ( ) ( ) ( )}
− 2 2
kp2 cos kp1 l sinh kp2 l + kp1 sin kp1 l cosh kp2 l
kp1 + kp2

11
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

∫l
Ilp2 = SPZ (H − z)dz
0

1 {( ) ( ) ( ) ( ) ( ) }
=( )2
2
kp2 2
− kp1 sin kp1 l sinh kp2 l + 2kp1 kp2 cos kp1 l cosh kp2 l − 2kp1 kp2 (A.15)
2 2
kp1 + kp2

l− H { ( ) ( ) ( ) ( )}
+ 2 2
kp2 sin kp1 l cosh kp2 l − kp1 cos kp1 l sinh kp2 l
kp1 + kp2

where ks1 = ys1/H, ks2 = ys2/H, kp1 = yp1/H, kp2 = yp2/H. Ihs1, Ihs2, Ihp1 and Ihp2 are given by:
∫h
Ihs1 = CSZ (h − z)dz
l
⎧ ( 2 ) ⎫

⎨ 2ks1 ks2 sin(ks1 h)sinh(ks2 h) + ks2 − ks1 cos(ks1 h)cosh(ks2 h) ⎪
2

1 (A.16)
=( )2
2
ks1 2
+ ks2 ⎩ − 2ks1 ks2 sin(ks1 l)sinh(ks2 l) − ( k2 − k2 )cos(ks1 l)cosh(ks2 l) ⎪
⎪ ⎭
s2 s1

l− h
+ 2 2
{ks2 cos(ks1 l)sinh(ks2 l) + ks1 sin(ks1 l)cosh(ks2 l)}
ks1 + ks2

∫h
Ihs2 = SSZ (h − z)dz
l
⎧( ) ⎫

⎨ ks2 − ks1 sin(ks1 h)sinh(ks2 h) + 2ks1 ks2 cos(ks1 h)cosh(ks2 h) ⎪
2 2

1 (A.17)
=( )2
2
ks1 2
+ ks2 ⎩ − ( k2 − k2 )sin(ks1 l)sinh(ks2 l) − 2ks1 ks2 cos(ks1 l)cosh(ks2 l) ⎪
⎪ ⎭
s2 s1

l− H
− 2 2
{ks2 sin(ks1 l)cosh(ks2 l) − ks1 cos(ks1 l)sinh(ks2 l)}
ks1 + ks2

∫h
Ihp1 = CPZ (h − z)dz
l

⎧ ( ) ( ) ( 2 2
) ( ) ( )⎫
⎨ 2kp1 kp2 sin kp1 h sinh kp2 h + kp2 − kp1 cos kp1 h cosh kp2 h ⎪
⎪ ⎬
1 (A.18)
=( )2
2
kp1 + 2
kp2

⎩ ( ) ( ) ( 2 2
) ( ) ( )⎪

− 2kp1 kp2 sin kp1 l sinh kp2 l − kp2 − kp1 cos kp1 l cosh kp2 l

l− h { ( ) ( ) ( ) ( )}
+ 2 2
ks2 cos kp1 l sinh kp2 l + ks1 sin kp1 l cosh kp2 l
kp1 + kp2

∫h
Ihp2 = SPZ (h − z)dz
l

⎧( 2 2
) ( ) ( ) ( ) ( )⎫
⎨ kp2 − kp1 sin kp1 h sinh kp2 h + 2kp1 kp2 cos kp1 h cosh kp2 h ⎪
⎪ ⎬
1 (A.19)
=( )2
2
kp1 2
+ kp2 ⎩ ( 2
⎪ 2
) ( ) ( ) ( ) ( )⎪

− kp2 − kp1 sin kp1 l sinh kp2 l − 2kp1 kp2 cos kp1 l cosh kp2 l

l− H { ( ) ( ) ( ) ( )}
− 2 2
kp2 sin kp1 l cosh kp2 l − kp1 cos kp1 l sinh kp2 l
kp1 + kp2

For the tension crack located at slope top, QBh and QBv are given by:
[ ]
CS Ibs1 + SS Ibs2 SS Ibs1 − CS Ibs2
QBh = kh γ(cot θ − cot δ) cos( ωt) + sin(ωt) (A.20)
CS2 + SS2 CS2 + SS2
[ ]
CP Ibp1 + SP Ibp2 SP Ibp1 − CP Ibp2
QBv = kv γ(cot θ − cot δ) 2 2
cos(ωt) + 2 2
sin(ωt)
CP + SP CP + S P

where Ibs1, Ibs2, Ibp1 and Ibp2 are given by:

12
L. Zhao et al. Soil Dynamics and Earthquake Engineering 139 (2020) 106310

∫h
Ibs1 = CSZ (h − z)dz
0
(A.21)
1 { ( 2 2
) ( 2 2
)}
=(
2 2
)2 2ks1 ks2 sin(ks1 h)sinh(ks2 h) − ks1 − ks2 cos(ks1 h)cosh(ks2 h) + ks1 − ks2
ks1 + ks2

∫h
Ibs2 = SSZ (h − z)dz
0
(A.22)
1 {( 2 2
) }
=( )2 ks1 − ks2 sin(ks1 h)sinh(ks2 h) + 2ks1 ks2 cos(ks1 h)cosh(ks2 h) − 2ks1 ks2
2 2
ks1 + ks2

∫h
Ibp1 = CPZ (h − z)dz

(A.23)
0

1 { ( ) ( ) ( 2 ) ( ) ( ) ( 2 )}
2 2
=( )2 2kp1 kp2 sin kp1 h sinh kp2 h − kp1 − kp2 cos kp1 h cosh kp2 h + kp1 − kp2
2 2
kp1 + kp2

∫h
Ibp2 = SPZ (h − z)dz

(A.24)
0

1 {( ) ( ) ( ) ( ) ( ) }
2 2
=( )2 kp1 − kp2 sin kp1 h sinh kp2 h + 2kp1 kp2 cos kp1 h cosh kp2 h − 2kp1 kp2
2 2
kp1 + kp2

References [17] Jimenez-Rodriguez R, Sitar N, Chacon J. System reliability approach to rock slope
stability. Int J Rock Mech Min Sci 2006;43(6):847–59.
[18] Zhao LH, Cao JY, Zhang Y. Effect of hydraulic distribution form on the stability of a
[1] Garwood NC, Janos DP, Brokaw N. Earthquake-caused landslides: a major
plane slide rock slope under the nonlinear Barton-Bandis failure criterion.
disturbance to tropical forests. Science 1979;205(4410):997–9.
Geomech Eng 2015;8(3):391–414.
[2] Meunier P, Hovius N, Haines AJ. Regional patterns of earthquake-triggered
[19] Zhao LH, Zuo S, Li L, Lin YL, Zhang YB. System reliability analysis of plane slide
landslides and their relation to ground motion. Geophys Res Lett 2007;34(20):
rock slope using Barton-Bandis failure criterion. Int J Rock Mech Min Sci 2016;88:
161–72.
1–11.
[3] Dai FC, Xu C, Yao X, Xu L, Tu XB, Gong QM. Spatial distribution of landslides
[20] Belghali M, Saada Z, Garnier D, Maghous S. Pseudo-static stability analysis of rock
triggered by the 2008 Ms 8.0 Wenchuan earthquake, China. J Asian Earth Sci 2011;
slopes reinforced by passive bolts using the generalized hoek-brown criterion.
40(4):883–95.
J Rock Mech Geotech Eng 2017;9. 670.
[4] Barton NR. The shear strength of rock and rock joints. Int J Rock Mech Min Sci
[21] Qin C, Chian SC. Pseudo-static/dynamic solutions of required reinforcement force
Geomech Abstr 1976;13(9):255–79.
for steep slopes using discretization-based kinematic analysis. J Rock Mech
[5] Bandis S, Lumsden AC, Barton NR. Experimental studies of scale effects on the
Geotech Eng 2019;11(2):289–99.
shear behaviour of rock joints. Int J Rock Mech Min Sci Geomech Abstr 1981;18(1):
[22] Steedman RS, Zeng X. The influence of phase on the calculation of pseudo-static
1–21.
earth pressure on a retaining wall. Geotechnique 1990;40(1):103–12.
[6] Barton NR, Choubey V. The shear strength of rock joints in theory and practice.
[23] Steedman RS, Zeng X. On the behaviour of quay walls in earthquakes.
Rock Mech 1977;10(1):1–54.
Geotechnique 2015;43(3):417–31.
[7] Barton N, Bandis S, Bakhtar K. Strength, deformation and conductivity coupling of
[24] Bellezza I. A new pseudo-dynamic approach for seismic active soil thrust. Geotech
rock joints. Int J Rock Mech Min Sci Geomech Abstr 1985;22(3). 121e40.
Geol Eng 2014;32(2):561–76.
[8] Nagpal A, Basha BM. Reliability analysis of anchored rock slopes against planar
[25] Rajesh BG, Choudhury D. Generalized seismic active thrust on retaining wall with
failure. In: Proceedings of Indian geotechnical conference. New Delhi, India; 2012.
submerged backfill using modified pseudo-dynamic method. Int J GeoMech 2016;
p. 1–4.
17(3):261–73.
[9] Basha BM, Moghal AAB. Load Resistance Factor Design (LRFD) approach for
[26] Pain A, Choudhury D, Bhattacharyya SK. Seismic rotational stability of gravity
reliability based seismic design of rock slopes against wedge failures. Proceedings
retaining walls by modified pseudo-dynamic method. Soil Dynam Earthq Eng
of the design, analysis and performance of rock slopes and rock fill. ASCE
2017;94(3):244–53.
Geotechnical Special Publication; 2013. p. 582–91. SanDiego.
[27] Liu MW, Fu H, Wu JL. Current status and thinking of determination method of
[10] Feng P, Lajtai EZ. Probabilistic treatment of the sliding wedge with ezslide. Eng
shear strength parameters of rock mass structural plane. J Chongqing Jianzhu Univ
Geol 1998;50(1):153–63.
2005;24(5):65–7.
[11] Prassetyo SH, Gutierrez M, Barton NR. Nonlinear shear behavior of rock joints
[28] Wyllie DC, Mah CW. Rock slope engineering. Spon Press; 2004. London.
using a linearized implementation of the barton-bandis model. J Rock Mech
[29] Li DQ, Jiang SH, Chen YF, Zhou CB. System reliability analysis of rock slope
Geotech Eng 2017;9(4):671–82.
stability involving correlated failure modes. KSCE J Civ Eng 2011;15(8):1349–59.
[12] Tamimi S, Amadei B, Frangopol DM. Monte Carlo simulation of rock slope stability.
[30] Zheng Y, Chen C, Liu T, Zhang H, Sun C. Theoretical and numerical study on the
Comput Struct 1989;33(6):1495–505.
block-flexure toppling failure of rock slopes. Eng Geol 2019;263. 105309.
[13] Wang J, Tan W, Feng S, Zhou R. Reliability analysis of an open pit coal mine slope.
[31] Ang HS, Tang WH. Probability concepts in engineering planning and design:
Int J Rock Mech Min Sci Geomech Abstr 2000;37(4):715–21.
design, risk and reliability, vol. 2. Wiley; 1984. New York.
[14] Christian JT, Baecher GB. The point-estimate method with large numbers of
[32] Der Kiureghian A. First-and second-order reliability methods. In: Engineering
variables. Int J Numer Anal Methods GeoMech 2002;26(15):15–29.
design reliability handbook. CRC Press; 2005.
[15] Duzgun HSB, Yucemen MS, Karpuz C. A methodology for reliability-based design
[33] Ang A, Tang W. Probability concepts in engineering: emphasis on applications to
of rock slopes. Rock Mech Rock Eng 2003;36(2):95–120.
civil and environmental engineering, vol. 1. John Wiley Publishers; 2007.
[16] Low BK. Reliability analysis of rock slopes involving correlated nonnormals. Int J
Rock Mech Min Sci Geomech Abstr 2007;44(6):922–35.

13

You might also like