Beruflich Dokumente
Kultur Dokumente
A R T I C LE I N FO A B S T R A C T
Keywords: The dynamic stability of a slope is an important problem in geotechnical engineering, especially when subjected
Three-dimensional slope to a strong earthquake. In this paper, an efficient calculation method for the time-history analysis of the dynamic
Dynamic stability stability of a 3D slope is developed, which is combined with the probability density evolution method (PDEM) to
Stochastic ground motion establish a reliability evaluation method for the dynamic stability of a 3D slope. A newly developed dimensional-
Probability density evolution
reduction spectrum-random function method combined with the stochastic process of ground motion provides
Reliability evaluation
Reinforcement range
an effective method for the stochastic seismic reliability analysis of complex nonlinear 3D slopes. Finally, the
stochastic dynamic response and reliability analysis of 2D and 3D slopes are performed based on a safety factor,
and then the extreme distribution probability is obtained. The results show that the seismic dynamic response of
the 3D slope is larger than that of a 2D slope because of the three-dimensional effect. In addition, the probability
density of the safety factor has a wider range of distribution in the three-dimensional situation. Therefore, the 3D
slope has a higher risk when subjected to seismic excitation, and it is necessary to study the dynamic stability
and reliability for some important slopes using 3D models in practical engineering. Moreover, the time-history
analysis method of the dynamic stability of a 3D slope established in this paper can directly obtain the range and
specific location of the slope instability failure and can provide a reliable basis for the prevention and re-
inforcement of landslides.
⁎
Corresponding author at: School of Hydraulic Engineering, Faculty of Infrastructure Engineering, Dalian University of Technology, Dalian 116024, China.
E-mail address: xubin@dlut.edu.cn (B. Xu).
1
The second author has the equal contribution as the first author to this paper.
https://doi.org/10.1016/j.soildyn.2019.02.006
Received 19 July 2018; Received in revised form 3 December 2018; Accepted 6 February 2019
0267-7261/ © 2019 Elsevier Ltd. All rights reserved.
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
30 28
27
25
25 23 23
22
21
20 20 20
20 19 19
16 16
15
15 13
10
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Fig. 1. Global earthquake frequency from 2000 to 2015 (magnitude of at least 7.0).
21st century, the occurrence of global earthquakes with a magnitude of stochastic earthquake action on the basis of the probability density
at least seven is 20.4 times/year on average, which has exceeded the evolution method and accurately calculated the seismic dynamic re-
annual average since 1900 (18.3 times/year). The Earth is currently in a liability. Therefore, a three-dimensional slope seismic stability analysis
seismic active period, and earthquake magnitudes are increasing. should be performed from the perspective of random vibration.
Therefore, the assessment of slope stability during an earthquake In this paper, a three-dimensional slope dynamic stability program
should be the focus of seismic disaster relief. is developed and verified via examples. Three-dimensional slope dy-
Currently, researchers worldwide have analysed and investigated namic stability is systematically investigated. A generalized probability
the seismic stability of two-dimensional slopes and made numerous density evolution method (GPDEM) is employed to perform the slope
achievements. Two-dimensional slope stability has a well-defined con- seismic reliability analysis. Random seismic excitation is introduced to
cept, mature theory and can be easily calculated. However, compared construct a virtual random process and corresponding GPDEM equa-
with a two-dimensional slope, a three-dimensional slope accurately tion. Based on an equivalent extreme value distribution probability
reflects the impact of slope width and boundary conditions and can be analysis, random dynamic analysis and deterministic finite element
directly applied to the investigation of the three-quantity distribution time-history analysis [26] are effectively integrated. Based on the slope
pattern of a three-dimensional slope. In addition, natural landslides are stability safety factor, the two-dimensional and three-dimensional slope
always three-dimensional; thus, simplified two-dimensional analysis dynamic stability and reliability are calculated. The feasibility and su-
results can be inaccurate [14–16]. A three-dimensional slope stability periority of this method in three-dimensional slope dynamic reliability
analysis can be performed to directly obtain the slope instability failure analysis are verified via examples. This method provides a new idea for
range and detailed location, which provides a basis for landslide pre- three-dimensional slope dynamic stability analysis, improves com-
vention and reinforcement. Numerous engineering practices demon- puting efficiency of three-dimensional slope system dynamic reliability
strate that the engineering structure safety index is solely based on analysis and provides effective technology for further investigation of
engineering experience and a single safety factor cannot quantify the three-dimensional slope dynamic stability and reliability.
safety margin or risk. The empirical safety factor cannot be extended
[17]. For rock soil, factors such as geological environment and load
condition are highly uncertain, which causes a highly variable slope. 2. Slope stability analysis
The reliability method is a perfect solution for this problem.
At present, the dynamic reliability analysis of slopes has become an 2.1. Three-dimensional slope stability analysis method
important topic in geotechnical engineering. Al-Homouda and
Tahtamonib [18] established a 3D probabilistic stability analysis model Using a sliding surface stress method to analyse three-dimensional
of a soil slope and embankment under a seismic load using a safety slope stability, the element stress of each component was calculated
factor and slope failure displacement. Abhijit and Subrata [19] ana- using a finite element method. Then, the stress component of a point on
lysed the reliability of a concrete dam under earthquake action based the sliding surface was calculated, and the surface normal and tan-
on the unconditional probability solved by the Taylor expansion gential stresses were determined.
method. However, the quasi-static method and Newmark finite sliding A general space problem contains 15 unknown functions, 6 of which
displacement algorithm [20,21] are mostly adopted. The quasi-static are stress components, 6 are deformation components and 3 are dis-
force method simplifies the seismic effect to an inertia force and cannot placement components, and all are functions of x, y and z coordinate
completely consider the seismic load randomness and variation of variables. The six stress components of point P on a rectangular co-
geotechnical material mechanical properties during earthquake. Cur- ordinate plane, σx , σy , σz , τxy = τyx , τxz = τzx , and τyz = τzy are known.
rent studies rarely consider the randomness of ground motion. In recent Select a plane ABC near an arbitrary point P on the slope passing
years, great progress has been made in the study of the dynamic re- through point P, and the three planes parallel to the coordinate plane to
liability combining structural randomness with stochastic excitation. Li form a tiny tetrahedron PABC, as shown in Fig. 2. When the tetrahedron
and Chen [22,23] developed a probabilistic density evolution theory of PABC decreases indefinitely towards point P, the stress on plane ABC is
stochastic dynamic system analysis, and this method provided a new the stress of the slope.
approach for the nonlinear stochastic dynamic reliability analysis of The cosine of the normal direction n′ external to plane ABC is:
large complex engineering structures. Xiong and Huang [24] analysed
the factors that affect the seismic stability of slopes, and the develop- cos(n, x ) = l , cos(n′, y ) = m , cos(n′, z ) = n.
ment prospect of the stochastic seismic response analysis method was
described in detail by describing the internal factors of this method. According to the equilibrium conditions of the tetrahedron,
Huang and Xiong [25] analysed the stability of a 2D slope under ∑ Fx = 0, ∑ Fy = 0, and ∑ Fz = 0 , the following can be obtained:
361
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
The normal stress and shear stress of any slope can be obtained from ment static horizontal stress, σx d is the element dynamic horizontal
Eqs. (1) and (2). The safety factor is expressed as [27]: stress, σy s is the element static vertical stress, σy d is the element dynamic
n
∑i = 1 (ci + σi tan φi ) Si vertical stress, τxy s is the element static shear stress; and τxy d is the
Fs = n element dynamic shear stress.
∑i = 1 τi Si (3)
where ci and φi are the adhesion angle and internal friction angle, re- 3. Generalized probability density evolution method (GPDEM)
spectively, of the i-th element; Si is the area of the i -th element passed
by a sliding arc; and σi and τi are the normal stresses and tangential In normal conditions, the slope nonlinear dynamic equilibrium
stresses on the sliding arc surface of the i -th element. equation in an earthquake is represented as follows [24,26,34,35]:
The three-dimensional slope stability analysis in this paper employs ¯ ̈ (t ) + CU̇ (t ) + KU (t ) = −MIx
¯ g̈ (Θ, t )
MU (7)
an enumeration method to search a sliding surface. The search is based
on a specified sphere centre space range and long/short axial radii of an where M̄ is the mass matrix; C is the damping matrix; K is the stiffness
ellipsoid. The calculation follows a predefined sequence. At each mo- matrix; Ü (t ) is the acceleration vector; U̇ (t ) is the velocity vector; U (t )
ment, the minimum safety factor of all sliding surfaces is selected as the is the displacement vector; I is the vector with all components being 1;
minimum safety factor of the slope. For the entire ground motion time- x g̈ (Θ, t ) is a stochastic process on an earthquake surface generated by a
history, this process is repeated to obtain the minimum safety factor of spectrum expression-random function; and Θ is a random vector.
slope dynamic stability during an entire earthquake. Considering the randomness of ground motion, assume that
Θ = (Θ1, Θ2 , ..., Θm) are random variables in a ground motion func-
2.2. Three-dimensional slope stability validation and analysis via example tion and m is the number of ground motion random variables.
To facilitate the calculation, the solution for formula (7) is expressed
To verify the three-dimensional slope stability program developed as follows:
in this paper, the three-dimensional stability of an ellipsoidal sliding
U (t ) = H (Θ , t ) (8)
surface (Example 1) in research paper [28] was analysed. The model
dimensions are shown in Fig. 3 (unit: m). The velocity process is represented as follows:
362
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
Table 1
Calculation results for an ellipsoid sliding surface.
(example source: Zhang Xing).
Example (year) Calculation method Sliding arc shape Safety factor
(1) In the basic random variable space Θ , the typical discrete point
θq (q = 1, 2, ..., nsel ) is selected, where nsel represents the total
number of discrete points [38].
(2) Based on the selected typical point θq , the slope nonlinear dynamic
equilibrium equation for an earthquake in formula (7) is solved. For
the given earthquake, the slope seismic response speed is obtained.
(3) The slope seismic response speed is substituted into the GPDEM
evolution equation to obtain a solution via the total variation di-
minishing (TVD) finite difference method [39].
(4) Based on formula (12), integral accumulation is performed [40] to
Fig. 4. Location of the sliding arc surface in an ellipsoid sliding surface. calculate the sum of results for q = 1, 2, ..., nsel and obtain the
(example source: Zhang Xing). probability density function.
defined as a random variable to analyse the slope dynamic force re- Ẍg (t ) = ∑ 2S Xg̈ (t , ωk ) △ω [cos(ωk t ) Xk + sin(ωk t ) Yk ]
k=1 (13)
liability. Based on the probability conservation principle [36,37], the
GPDEM evolution equation in the slope dynamic stability analysis is where ωk = kΔω; S Xg̈ is the bilateral evolution power spectrum density
represented as follows: function for an unsteady acceleration time-history of earthquake
ground motion; and {Xk , Yk }(k = 1, 2, ..., N ) are standard orthogonal
∂PFSΘ (FS , θ , t ) ∂PFSΘ (FS , θ , t )
+ FṠ ⎛⎜θ , t ⎞⎟ =0 random variables. Typically, Δω = 0.15 rad/s, and the number of
∂t ⎝ ⎠ ∂FS (10) truncation terms is N = 1600 .
Assume any two sets of standard orthogonal random variables X̄m
Initial condition:
and Ȳm , where (m = 1, 2, ..., N ) are functions of two independent basic
PFSΘ (FS , θ , t )|t = t0 = δ (FS − FS 0 ) PΘ (θ) (11) random variables Θ1 and Θ2 . The corresponding random functions are
expressed as follows [41,42]:
Joint probability density function (PDF) of FS (t ) :
X¯ m = cas (nΘ1), Y¯m = cas (nΘ2) (14)
PFS ⎛⎜FS , t ⎞⎟ = ∫ΩΘ PFSΘ ⎛⎜FS , θ , t ⎞⎟ dθ where cas (x ) = cos(x ) + sin(x ) is the Hartley function; and the basic
⎝ ⎠ ⎝ ⎠ (12)
random variables Θ1 and Θ2 are independent and follow a uniform
Solution process of the GPDEM evolution equation in the slope distribution in (0, 2π).
dynamic stability analysis is expressed as follows: This paper is based on an acceleration power spectrum model
Fig. 5. 3D elevation view of an ellipsoid sliding surface (example source: Zhang Xing): A-this study, B-reference [29].
363
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
2
a¯ max
S0 (t ) =
1
r¯2πωg (t ) ⎡2ξg (t ) + 2ξg (t )
⎤
⎣ ⎦ (19)
where ā max is the average ground acceleration peak, and γ = 2.6 is the
effective peak factor determined via seismic response spectrum optimal
fitting.
This paper is based on the spectrum in the latest hydraulic anti-
seismic specification and number theory. A series of non-stationary
ground motion are generated by modifying the evolution power spec-
Fig. 7. Safety factor calculation.
trum [44,45]. Finally, 144 acceleration time-history samples are ob-
tained discretely (the comparison between the mean values of the
samples and the target spectra before and after the correction is shown
proposed by Clough and Penzien [43]: in Fig. 8). The average acceleration peak is ā max = 200cm/s2 . The
ωg4 (t ) + 4ξg2 (t ) ωg2 (t ) ω2 combination of two statistical values (mean and standard deviation)
⎛ ⎞
S Xg̈ ⎜t , ω⎟ = A2 (t ) and the seismic response spectrum between the sampling and the target
[ω2 − ωg2 (t )]2 + 4ξg2 (t ) ωg2 (t ) ω2 determine the effectiveness and accuracy of the optical spectrum ex-
⎝ ⎠
ω4 pression-random function (Fig. 9).
⋅ 2
⋅S0 (t )
[ω − ωf (t )] + 4ξ f2 (t ) ωf2 (t ) ω2
2 2
(15)
5. Numerical example
where A (t ) is an intensity modulation function; the recommended
formula is expressed as follows: 5.1. Finite element analysis model
d
⎡t t ⎤
A (t ) = ⎢ exp ⎜⎛1 − ⎟⎞ ⎥ The geometric models of 2D and 3D slopes are shown in Fig. 10
⎣c ⎝ c ⎠⎦ (16) (unit: m). The slope model mesh dimension parameters and rock soil
parameters are listed in Tables 2, 3.
where c = 4 s is the mean arrival time of the peak acceleration; and
d = 2 is control index of A (t ) .
The parameters that reflect the unsteady characteristics of the 5.2. Slope random ground motion response analysis
evolution power spectrum are expressed as follows:
A deterministic dynamic time-history analysis is performed for a
t t
ωg (t ) = ω0 − a , ξg (t ) = ξ0 − b series of events with the average peak acceleration ā max = 200cm/s2 to
T T (17)
obtain the seismic safety factor time-history of 2D and 3D slopes. The
ωf (t ) = 0.1ωg (t ), ξf (t ) = ξg (t ) safety factor time-history is substituted into the GPDEM equation, and
(18)
the TVD finite difference is employed to obtain a solution. The random
where ωg (t ) is the circular frequency; and ξg (t ) is the rock soil damping response and dynamic reliability of the safety factor for various seismic
ratio. Initially, the circular frequency is ω0 = 25s−1, and the rock soil effects are also obtained.
damping ratio is ξ0 = 0.45. Based on the field classification and seismic Under static force, the minimum safety factor of the 2D slope is
environment category, the parameters are determined as follows: 1.1347, and that of the 3D slope is 1.2749. As shown in Fig. 11, under
a = 3.5 and b = 0.3. T = 15 s is the accumulated period of unsteady identical ground motions, the minimum dynamic safety factor for a 2D
acceleration processes. ωf (t ) and ξf (t ) are the corresponding filtering slope is 0.9603, and the value for a 3D slope is 0.8456. The stable safety
parameters. factor peak time and fluctuation distribution of 2D and 3D slopes are
The time evolution power spectrum also includes the earthquake not synchronized. Compared with a 2D slope, the 3D slope safety factor
intensity frequency spectrum parameter S0 (t ) , whose expression is ex- has a wider fluctuation range, more frequent fluctuations and is more
pressed as follows: susceptible to ground motion. In an earthquake, the slope instability
Fig. 8. Seismic acceleration response spectrum comparison: A- before iteration, B- after iteration.
364
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
Table 2
Slope model mesh dimension parameters.
Dimensionality Size (m) Elements Nodes
Table 3
Statistical characteristics of Example 1 soil parameters.
1800 20 25 0.3 20
2
1.8
1.6
1.4 2D
1.2 3D
1
0.8
0.6
0 2.5 5 7.5 10 12.5 15
Fig. 11. 2D and 3D slope safety factor time-history curves under identical
ground motion.
365
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
Fig. 13. Sliding arc depth and range of 3D slope under static and dynamic force.
Fig. 14. 2D and 3D slope safety factor mean and standard deviation time-history curves.
Fig. 15. 2D and 3D slope safety factor probability density functions at 4 s, 9 s and 10 s.
has a significant impact on the safety factor probability density func- in the range of 0.9–1.3. The 3D slope safety factor is primarily dis-
tion, and the safety factor fluctuates within an extensive range. As time tributed in the range of 1.0–1.7.
elapses, the impact of ground motion on the safety factor decreases and These results indicate that random ground motion is sensitive to the
the safety factor distribution increasingly concentrates around the mean slope stability safety factor. Compared with the 2D slope, the 3D slope
value. is more susceptible to ground motion. Rock soil under ground motion
Figs. 16 and 17 directly represent the evolution of 2D slope prob- exhibits a significant dynamic nonlinearity. Safety factor probability
ability density function and the 3D slope probability density function, density has a wider distribution. Around the mean value, the variation
respectively, with time from 9 to 10 s and the corresponding isoline range is wider, and the difference is more significant. This finding
distributions. Fig. 16 shows the peak and trough of the safety factor further proves the importance of analysing the dynamic stability of a 3D
probability density function evolution with time; the peak reflects the slope during an earthquake from the perspective of random ground
probability density under a higher probability. Fig. 17 shows the sur- motion.
face profile of the probability density function evolution with time, i.e.,
isoline distribution. The 2D slope safety factor is primarily distributed
366
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
Fig. 16. Slope safety factor probability density evolution: A-3D, B-2D.
Fig. 17. Slope safety factor probability density evolution isoline: A-3D, B-2D.
Fig. 18. PDF and CDF curves of 2D and 3D slope safety factor equivalent extreme values.
5.3. Slope dynamic reliability analysis the same safety factor determination criterion, the 3D slope dynamic
stability failure probability is higher than the 2D slope dynamic stabi-
To obtain the slope safety factor dynamic reliability, the equivalent lity failure probability.
extreme values of the minimal safety factors for various ground motion
events are simulated [34,36] to construct the “virtual” random process.
Based on the GPDEM evolution process, the reliability of the slope 6. Conclusions
dynamic stability analysis is obtained, i.e., PDF and CDF curves of the
equivalent extreme safety factor (Fig. 18). Fig. 18 shows the equivalent Slope dynamic stability is a complex issue. In an earthquake, slope
extreme-based PDF curve, which differs from Fig. 15, which shows the rock soil magnifies the seismic waves and demonstrates intense dy-
safety factor time-history PDF curve, which evolves with time. In namic nonlinearity, which has a significant impact on the slope dy-
Fig. 15, the PDF curve at different time points has a corresponding namic stability. From the perspective of random ground motion and
extensibility distribution, and the equivalent extreme events of the based on the equivalent extreme value distribution and GPDEM evo-
slope dynamic stability safety factor do not always follow a Gaussian lution, a slope dynamic reliability analysis method is proposed in this
distribution. paper. Numerical examples are used to analyse the dynamic stability of
As shown in Fig. 18, under identical random ground motion, the 2D 2D and 3D slopes and obtain the reliability of the slope dynamic sta-
slope dynamic stability analysis results are risky to some extent. Using bility. The following conclusions are obtained:
367
L. Song, et al. Soil Dynamics and Earthquake Engineering 120 (2019) 360–368
(1) Based on the slope dynamic response extreme value, a dynamic under extreme loadings; 2014.
random process with “virtual time parameters” is constructed in [14] Dai FC, Lee CF, Deng JH, et al. The 1786 earthquake-triggered landslide dam and
subsequent dam-break flood on the Dadu River, Southwestern China-Reply.
this paper to obtain the extreme value distribution probability of Geomorphology 2005;65(3/4):205–21.
the slope dynamic response and corresponding reliability in tandem [15] Huang RQ. et al. Geohazard assessment of the wenchuan earthquake; 2009.
with the GPDEM evolution process. This method is highly practical [16] Lin ML, Wang KL. Seismic Slope Behavior in a Large Scale Shaking Table Model test.
Eng Geol 2006;86(2):118–33.
for the random probability analysis of a complex structure and can [17] Li L, Wang Y, Cao ZJ, et al. Risk de-aggregation and system reliability analysis of
determine the slope dynamic reliability with high precision. slope stability using representative slip surfaces. Comput Geotech
(2) Compared with the 3D slope stability analysis, the 2D slope stability 2013;53(3):95–105.
[18] Al-Homoud AS, Tahtamoni WW. Reliability analysis of three-dimensional dynamic
analysis has numerous limitations. A 2D slope cannot accurately slope stability and earthquake-induced permanent displacement. Soil Dyn Earthq
reflect the influence of slope width and boundary conditions. The Eng 2000;19(2):91–114.
2D slope stability static force calculation result is conservative, [19] Abhijit C, Subrata C. Reliability of linear structures with parameter uncertainty
under non-stationary earthquake. Struct Saf 2006;28(3):231–46.
while the dynamic calculation result tends to provide more safety.
[20] Gazetas G. Seismic response of earth dams: some recent developments. Soil Dyn
Therefore, the simplified 2D dynamic analysis result may be ha- Earthq Eng 1987;6(1):2–47.
zardous. Furthermore, the 3D slope stability analysis can be per- [21] U.S. Department of Homeland Security. Federal guidelines for dam safety.
formed to directly obtain the slope instability failure range and Earthquake analyses and design of dams; 2005.
[22] Li J, Chen JB, Fan WL. The equivalent extreme-value event and evaluation of the
detailed location and provides a basis for landslide prevention and structural system reliability. Struct Saf 2007;29(2):112–31.
reinforcement. [23] Li J, Peng YB, Chen JB. A physical approach to structural stochastic optimal con-
(3) The random ground motion response analysis and reliability result trols. Probab Eng Mech 2010;25(1):127–41.
[24] Huang Y, Xiong M. Dynamic reliability analysis of slopes based on the probability
reveals the sensitivity of random ground motion to the slope sta- density evolution method. Soil Dyn Earthq Eng 2017;94:1–6.
bility safety factor. Compared with the 2D slope factor, the 3D slope [25] Xiong M, Huang Y. Stochastic seismic response and dynamic reliability analysis of
safety factor has a wider probability density distribution during an slopes: a review. Soil Dyn Earthq Eng 2017;100:458–64.
[26] Li J, Liu ZJ, Chen JB. Orthogonal expansion of ground motion and PDEM-based
earthquake; around the mean value, the variation range is wider, seismic response analysis of nonlinear structures. Earthq Eng Eng Vib
the difference is more significant, and rock soil under ground mo- 2009;8(3):313–28.
tion has a more significant dynamic nonlinearity. These results [27] Zhang ZY. Three-dimensional aseismic stability analysis of high concrete face
rockfill dam based on finite element sliding-surface stress method; 2018. [in
prove the importance of analysing the 3D slope dynamic stability Chinese]; 2018.
from the perspective of random ground motion. [28] Xing Z. Three-dimensional stability analysis of concave slopes in plan view. J
Geotech Eng 1988;114(6):658–71.
[29] Wang K. Improvement and application of slope three-dimensional limit equilibrium
Acknowledgements
method; 2013. [in Chinese].
[30] Fang JR. Slope stability analysis using direct search method based on 3D FEM and
This work was supported by National Key R&D Program of China its application in tunnel excavation; 2007. [in Chinese].
(2017YFC0404904) and the National Natural Science Foundation of [31] Zhang CL. Study on the three-dimensional limit equilibrium of slope stability; 2008.
[in Chinese].
China (Grant nos. 51679029, 51508071 and 51779034). These fi- [32] Zheng YR, Chen ZY. et al. Engineering treatment of slope & landslide; 2010.
nancial supports are gratefully acknowledged. [33] Chen ZY. Stability analysis of soil slope: principle-method-procedure; 2003.
[34] Chen JY, Jia QB, Xu Q. Study on probability density evolution method for time-
dependent dynamic reliability of concrete dams subjected to earthquake. J Hydraul
References Eng 2017;48(11):1–7. [in Chinese].
[35] Pang R, Xu B, Kong XJ, et al. Seismic reliability assessment of earth-rockfill dam
[1] Huang RQ. Large-scale landslides and their sliding mechanisms in china since the slopes considering strain-softening of rockfill based on generalized probability
20th century. Chin J Rock Mech Eng 2007;26(3):433–54. [in Chinese]. density evolution method. Soil Dyn Earthq Eng 2018;107:96–107.
[2] Au SWC. Rain-induced slope instability in Hong Kong. Eng Geol 1998;51(1):1–36. [36] Pang R, Xu B, Zou DG, Kong XJ. Stochastic seismic performance assessment of high
[3] Yin KL, Han ZS, Li ZZ. Progress of landslide researches in the world. Hydrogeol Eng CFRDs based on generalized probability density evolution method. Comput Geotech
Geol 2000;27(5):1–4. [in Chinese]. 2018;97:233–45.
[4] Schuster RL, Lynn MH. Socioeconomic impacts of landslides in the Western [37] Li J, Chen JB. Stochastic dynamics of structures. Structural safety; 2009.
Hemisphere. Reston VA, USA: United States Geological Survey; 2001. [38] Li J, Chen JB, Fan WL. The equivalent extreme-value event and evaluation of the
[5] United States Geological Survey. Landslide hazards. Reston, VA, USA: United States structural system reliability. Struct Saf 2007;29(2):112–31.
GS Fact Sheet (FS-071-00); 2000. [39] Hua LK, Wang Y. Applications of number theory to numerical analysis. Berlin:
[6] Parise M, Wasowski J. Landslide activity maps for landslide hazard evaluation: Springer; 1981.
three case studies from southern italy. Nat Hazards 1999;20(2/3):159–83. [40] Li J, Chen JB. Dynamic response and reliability analysis of structures with uncertain
[7] Zhang D, Wu ZH, Li JC, Jiang Y. An overview on earthquake-induced landslide parameters. Int J Numer Methods Eng 2005;62(2):289–315.
research. J Geomech 2013;19(3):225–41. [in Chinese]. [41] Liu ZJ, Liu W, Peng YB. Random function based spectral representation of sta-
[8] RadbruchHall DH, Colton RB, Davis WE. et al. Preliminary landslide overview map tionary and non-stationary stochastic processes. Probab Eng Mech 2016;45:115–26.
of the conterminous United States. Reston, VA, USA: United States Geological [42] Liu ZJ, Liu ZH, Peng YB. Simulation of multivariate stationary stochastic processes
Survey; 1983. using dimension-reduction representation methods. J Sound Vib 2018;418:144–62.
[9] Yamagishi H. Recent landslides in Western Hokkaido, Japan. Pure Appl Geophys [43] Deodatis G. Non-stationary stochastic vector processes: seismic ground motion
2000;157(6/8):1115–34. applications. Probab Eng Mech 1996;11(3):149–67.
[10] Lin PS, Lin JY, Hung JC, et al. Assessing debris-flow hazard in a watershed in [44] Tian JY, Liu HL, Wu XY. Evaluation perspectives and criteria of maximum aseismic
Taiwan. Eng Geol 2002;66(3/4):295–313. capability for high earth-rock dam. J Disaster Prev Mitig Eng 2013;33(S1):128–31.
[11] Bhasin R, Grimstad E, Larsen JO, et al. Landslide hazards and mitigation measures [in Chinese].
at Gangtok, Sikkim Himalaya. Eng Geol 2002;64(4):351–68. [45] Liu ZJ, Liu ZH, Liu W. Probability model of fully non-stationary ground motion with
[12] Li ZS. The state of the art of the research on seismic landslide hazard at home and the target response spectrum compatible. J Vib Shock 2017;36(2):32–8. [in
abroad. J Catastro 2003;18(4):65–71. [in Chinese]. Chinese].
[13] Wang GH. Dynamic response and damage mechanism of concrete gravity dams
368