Abstract
Traditional rigorous limit equilibrium methods satisfy all equilibrium conditions and usually have high accuracy, however, which are less efficient for slope reliability analysis. The main reason is that the limit state functions are highly nonlinear implicit functions of safety factor. Complex numerical iterations are required, which may sometimes lead to computational convergence problems. A new method for computing slope reliability calculation with high efficiency and accuracy was proposed. This method was based on the rigorous limit equilibrium method by modifying normal stresses over the slip surface. The critical horizontal acceleration factor
1 Introduction
The value of safety factor cannot truly reflect the safety state of the slope, due to the inherent variability of geotechnical materials and the uncertainty of loads, as well as the additional assumptions of inter-slice forces required by the limit equilibrium methods. The slope with safety factor greater than one is not necessarily safe [1]. As a matter of fact, some slopes with high safety factors have been damaged [2,3]. Slope reliability is another index that can reflect slope stability more effectively than safety factor. The purpose of slope reliability analysis is to solve the problem of stability uncertainty caused by the randomness of geotechnical parameters. The calculations of failure probability and reliability index must adopt a calculation method with sufficient accuracy. Otherwise, the reliability calculation results will be less objective than that of the safety factor. Therefore, it is necessary to develop a slope reliability analysis method with simple algorithm, high calculation accuracy, and high efficiency.
In practical engineering, the reliability analysis of slope stability is often based on the limit equilibrium method. Generally, the difference between the safety factor and one is used as the limit state function, which is combined with the reliability calculation method to determine the slope reliability. The accuracy and efficiency of slope reliability calculation are closely related to the complexity of safety factor expression and the effectiveness of reliability algorithm. At present, there are two kinds of limit equilibrium methods. One is the “simplified” method that does not rigorously meet the static equilibrium conditions, such as Fellenius method [4], simplified Bishop method [5], and simplified Janbu method [6]. The other is “rigorous” method that fully satisfies all the force and moment equilibrium conditions, such as the rigorous Janbu method [7], Spencer method [8], and Morgenstern–Price method [9,10]. Many studies have shown that only the safety factor that satisfies all equilibrium conditions is reliable for complex slopes with noncircular slip surfaces [11,12]. The calculation accuracy of safety factors of “rigorous” methods is generally higher than that of “simplified” methods. However, the expression of safety factor of the “rigorous” method is usually not explicit. Multiple complex numerical iterations are required to obtain the results. Sometimes, the calculation results do not converge [13].
Some scholars have proposed many efficient slope reliability analysis methods by constructing the approximate explicit expression of performance function and/or reducing the number of samplings [14]. The surrogate model method simplifies the expression of performance function from a complex nonlinear implicit format to an approximately equivalent explicit format. Nonintrusive stochastic finite element method (NISFEM) and response surface method (RSM) are representative methods [15,16]. These methods can improve the accuracy of the complex slope reliability calculation by combining with machine learning methods such as artificial neural network [17] and support vector machine [18]. However, a large number of data samples are needed to obtain the surrogate models. The calculation accuracy and efficiency mainly depend on the selection of parameters and the location of sample points. For the slope reliability problem with highly nonlinear implicit performance function and small failure probability, the number of sample points required to fit the surrogate model is huge. At present, there is no consensus about the effective point selection method.
Sarma proposed a new method to characterize the slope stability by using critical horizontal acceleration factor
Even if the performance functions are implicit, as long as there are enough samples, the results of slope reliability calculation by Monte Carlo simulation (MCS) method are accurate. However, for complex slopes with failure probability less than
An efficient calculation method of slope reliability based on the rigorous limit equilibrium method was proposed in this article. First, based on the rigorous limit equilibrium method by modifying the normal stress on the sliding surface, the explicit expression of the critical horizontal acceleration factor
2 Slope reliability calculation method based on rigorous limit equilibrium
2.1 Limit equilibrium equations
A two-dimensional slope with a general shape slip surface is shown in Figure 1. The sliding body boundary consists of the ground surface
Assuming that the sliding body is in a limiting state, the safety factor
where
The centroid coordinate of the sliding body is
where
By substituting equation (2) into equations (3a)–(3c), the equations thus become:
where
2.2 Modified normal stresses over the slip surface
It can be seen from equations (4a)–(4c) that if the distribution of normal stress
To satisfy the three equilibrium equations, the correction function can be taken as a linear function with two auxiliary variables:
where
2.3 Rigorous limit equilibrium equations by modifying normal stresses over the slip surface
By substituting equations (6), (7), (8), (9a), and (9b) into equations (4a)–(4c), we can obtain:
where
When the initial normal stresses distribution
If
where
2.4 Limit state performance function of slope stability characterized by
K
c
The accuracy and efficiency of the slope reliability calculation depend on the complexity of the limit state performance function expression. Sarma [19] proposed a method for calculating the critical horizontal acceleration factor
After calculations, Sarma found that there was a monotonically decreasing function relationship between the horizontal acceleration coefficient
When there is no seismic force (i.e., the known horizontal seismic coefficient
According to the aforementioned idea, the critical horizontal acceleration factor
where
It can be seen from equation (18) that
The method of calculating
It should be noted that the calculation results must be checked to satisfy the conditions that the horizontal thrust of the entire sliding body is zero, there is no tension between the soil slices, and the normal stresses over the sliding surface are positive.
2.5 Slope failure probability calculation based on SS method
It is assumed that the random variables affecting the slope stability are
where
Among many slope reliability calculation methods, MCS method is recognized as an accurate method. Based on the law of large numbers, MCS method uses the frequency of failure events to approximate the probability of failure events. When the number of samples is large enough, the estimated failure probability can converge to the true value. Let the sampling number be N, and the unbiased estimation of slope failure probability can be obtained:
where
The concept of MSC method is simple, but its computational efficiency is very low for small failure probability problems. Au and Wang [22] improved the MCS method and proposed an SS method that can effectively solve small failure probability problems. SS method divides the probability space into a series of subsets with a sequence inclusion relationship by introducing reasonable intermediate failure events. Therefore, the small failure probability can be expressed as the product of a series of large conditional probabilities that can be efficiently obtained.
In this method, the target failure event is
where
Let
The outline of calculating failure probability by the SS method is as follows:
Step 1. MCS method is used to generate
Step 2. The performance function values
Step 3. The
Step 4. The performance function values corresponding to this
Step 5. Repeat Steps 3–4 until the failure domain
Step 6. The failure probability of slope is estimated by equation (28).
The improved Metropolis algorithm is used to perform MCMCS to generate conditional samples of intermediate failure events. The detailed algorithm and calculation steps can be seen in previous studies [28–30]. The number of samples produced in the whole process is as follows:
The reliability of the structure can be expressed by failure probability or reliability index. There is a one-to-one correspondence between failure probability and reliability index. When the failure probability is obtained, the reliability index
where
A flowchart of coupling the proposed method with the SS method to calculate the slope failure probability is shown in Figure 3.
3 Examples
In order to verify the effectiveness of the proposed method, the reliability indexes, failure probability, and computer CPU time of two slope examples were compared. The explicit expression of
3.1 Example 1
The first example slope is taken from a test question of the Australian CAD Association [32]. The slope has three soil layers and a known circular sliding surface, as shown in Figure 4. The unit weight of each soil layer is constant, and cohesion and friction coefficients are normally distributed. The statistical parameters are given in Table 1. There is no seismic force on the slope, so
Soil layer number | c (kPa) | f = tan φ | γ (kN/m3) | ||
---|---|---|---|---|---|
μ c | σ c | μ f | σ f | ||
① | 0 | 0 | 0.781 | 0.1 | 19.5 |
② | 5.3 | 0.7 | 0.424 | 0.05 | 19.5 |
③ | 7.2 | 0.2 | 0.364 | 0.05 | 19.5 |
Note: In soil layer ①, μ c = 0, σ c = 0 means that the soil is cohesionless soil, and its cohesion is minimal and can be ignored.
Performance functions | Reliability methods | Sampling number N | β | P f | Computation time t (s) |
---|---|---|---|---|---|
|
MCS | 1,000,000 | 3.183 | 0.000730 | 1557.60 |
|
MCS | 512,000 | 3.185 | 0.000725 | 788.39 |
|
MCS | 1,000,000 | 3.172 | 0.000756 | 33.36 |
|
MCS | 512,000 | 3.169 | 0.000761 | 16.48 |
|
SS | 1,000,000 | 3.183 | 0.000730 | 31.98 |
|
SS | 64,000 | 3.184 | 0.000727 | 3.78 |
Bishop | 3.281 [33] |
Note: MCS: Monte Carlo simulation; SS: subset simulation.
After one million samplings, reliability indexes and failure probability of three methods were all very close, and their reliability index errors with the simplified Bishop method [33] were all less than 4%. The CPU time of the
According to Figure 5, when the calculation results tended to be stable, the reliability index and failure probability calculation results of the
Although the sampling times of the
3.2 Example 2
Another slope is James Bay Dam in northern Quebec, Canada [34]. As shown in Figure 6, the slope contains four soil layers, and the slip surface is noncircular. The statistical values of soil parameters of each layer are listed in Table 3. The calculation results of slope reliability are presented in Table 4. The comparison curves of reliability indexes, CPU time, and sampling times are shown in Figure 7.
Soil layer number | c (kPa) | f = tan φ | γ (kN/m3) | |||
---|---|---|---|---|---|---|
μ c | σ c | μ f | σ f | γ | σ γ | |
① | 0 | 0 | 0.577 | 0.031 | 20 | 1.1 |
② | 41 | 0 | 0 | 0 | 19 | 0 |
③ | 34.5 | 3.95 | 0 | 0 | 19 | 0 |
④ | 31.2 | 6.31 | 0 | 0 | 20.5 | 0 |
Note: In soil layer ①, μ c = 0, σ c = 0 means that the soil is cohesionless soil, and its cohesion is minimal and can be ignored.
Performance functions | Reliability methods | Sampling number N | β | P f | Computation time t (s) |
---|---|---|---|---|---|
|
MCS | 1,000,000 | 1.477 | 0.0698 | 37441.00 |
|
MCS | 32,000 | 1.476 | 0.0700 | 1233.70 |
|
MCS | 1,000,000 | 1.463 | 0.0718 | 166.25 |
|
MCS | 64,000 | 1.476 | 0.0700 | 5.89 |
|
SS | 1,000,000 | 1.464 | 0.0716 | 122.55 |
|
SS | 32,000 | 1.464 | 0.0716 | 3.53 |
FEM | RSM | 1.49 [34] |
Note: FEM: finite element method; RSM: response surface method.
The results of the three methods after one million samplings were compared with that of the combination method of FEM and RSM. The reliability index errors were all less than 2%. Among them, the error of the
It can be seen from Figure 7 that the sampling times of the
According to the analyses of the aforementioned two examples, when the reliability calculation method was the same, the reliability results of performance function using
In summary, the accuracy of slope reliability calculation results using the
4 Conclusion
For slopes with general shape slip surfaces, based on the limit state performance functions of safety factor expressions of traditional rigorous limit equilibrium methods, the accuracy of the reliability analysis results is high, but the calculation efficiency is very low. Based on the rigorous limit equilibrium method by modifying normal stresses over the slip surface, a slope reliability calculation method using the explicit expression containing the critical horizontal acceleration factor as the limit state performance function was proposed in this article. Through the case studies, the following conclusions were drawn:
This method is based on the limit equilibrium method by modifying the normal stress over the slip surface, which strictly satisfies all the equilibrium conditions and is suitable for the reliability analysis of two-dimensional slope with arbitrary shape slip surface.
The accuracy of calculation results of this method was very close to that of the limit state performance function using the safety factor implicit expression of the Morgenstern–Price method, but the computational efficiency was significantly improved.
This method is especially suitable for the reliability calculation of small failure probability slope with complex soil layer and a large number of random variables. It can be used as an effective tool for engineering designers to carry out a rapid slope reliability analysis.
In this article, only the reliability analyses of slopes with known slip surfaces were carried out. In practical engineering applications, this method can be easily combined with the optimization method to calculate the slope system reliability, which will be the focus of the next research work of this method.
-
Funding information: This work was supported by the National Natural Science Foundation of China (NSFC Grant No. 52079121).
-
Author contributions: Conceptualization, methodology, software, formal analysis, investigation, data curation, writing – original draft preparation, writing – review and editing, visualization. Conceptualization: Juxiang Chen and Dayong Zhu; methodology: Juxiang Chen and Dayong Zhu; software: Juxiang Chen; formal analysis: Juxiang Chen, Dayong Zhu, and Yalin Zhu; investigation: Juxiang Chen; data curation: Juxiang Chen; writing – original draft preparation: Juxiang Chen; writing – review and editing: Juxiang Chen, Dayong Zhu, and Yalin Zhu; visualization: Juxiang Chen; all authors have read and agreed to the published version of the manuscript.
-
Conflict of interest: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.
-
Ethical approval: The conducted research is not related to either human or animal use.
-
Data availability statement: All data generated or analyzed during this study are included in this published article (and its supplementary information files).
References
[1] Cho SE. Effects of spatial variability of soil properties on slope stability. Eng Geol. 2007;92(3):97–109. 10.1016/j.enggeo.2007.03.006.Search in Google Scholar
[2] He K, Wang S, Du W, Wang S. Dynamic features and effects of rainfall on landslides in the three gorges reservoir region, China: using the Xintan landslide and the large Huangya landslide as the examples. Environ Earth Sci. 2010;59(6):1267–74. 10.1007/s12665-009-0114-5.Search in Google Scholar
[3] Johari A, Lari AM. System probabilistic model of rock slope stability considering correlated failure modes. Comput Geotech. 2017;81:26–38. 10.1016/j.compgeo.2016.07.010.Search in Google Scholar
[4] Fellenius W. Calculation of the stability of earth dams. In: Proceedings of 2nd Congress on Large Dams. Vol. 4. Washington DC: 1936. p. 445–63. Conference paper.Search in Google Scholar
[5] Bishop AW. The use of the slip circle in the stability analysis of earth slopes. Géotechnique. 1955;5(1):7–17. 10.1680/geot.1955.5.1.7.Search in Google Scholar
[6] Janbu N, Bjerrum L, Kjaernsli B. Soil mechanics applied to some engineering problems. Oslo, Norway: Norwegian Geotechnical Institute. Publication 16; 1956.Search in Google Scholar
[7] Janbu N. Slope stability computations. In: Hirschfield E, Poulos S, editors. Embankment dam engineering, casagrande memorial volume. New York: Wiley; 1973. p. 47–86.Search in Google Scholar
[8] Spencer E. A method of analysis of the stability of embankments assuming parallel inter-slice forces. Géotechnique. 1967;17(1):11–26. 10.1680/geot.1967.17.3.296.Search in Google Scholar
[9] Morgenstern NR, Price VE. The analysis of the stability of general slip surfaces. Géotechnique. 1965;15(1):79–93. 10.1680/geot.1965.15.1.79.Search in Google Scholar
[10] Chen ZY, Morgenstern NR. Extensions to the generalized method of slices for slope stability analysis. Can Geotech J. 1983;20(1):104–19. 10.1139/t83-010.Search in Google Scholar
[11] Duncan JM. State of the art: limit equilibrium and finite-element analysis of slopes. J Geotech Eng ASCE. 1996;122(7):577–96. 10.1061/(ASCE) 0733-9410.Search in Google Scholar
[12] Zhu DY, Lee CF, Jiang HD, Kang JW. Solution of slope safety factor by modifying normal stresses over slip surface. Chin J Rock Mech Eng. 2004;23(16):2788–91. 10.3321/j.issn:1000-6915.2004.16.023.Search in Google Scholar
[13] Zhu DY, Lee CF. Explicit limit equilibrium solution for slope stability. Int J Numer Anal Methods Geomech. 2002;26(15):1573–90. 10.1002/nag.260.Search in Google Scholar
[14] Li DQ, Tang XS. Review and prospect of basic research in the field of reliability and risk mitigation in hydraulic geotechnical engineering. Bull Natl Nat Sci Found China. 2021;35(3):440–50. 10.16262/j.cnki.1000-8217.20210611.010.Search in Google Scholar
[15] Jiang SH, Li DQ, Zhang LM, Zhou CB. Slope reliability analysis considering spatially variable shear strength parameters using a non-intrusive stochastic finite element method. Eng Geol. 2014;168:120–8. 10.1016/j.enggeo.2013.11.006.Search in Google Scholar
[16] Li DQ, Jiang SH, Cao ZJ, Zhou W, Zhou CB, Zhang LM. A multiple response-surface method for slope reliability analysis considering spatial variability of soil properties. Eng Geol. 2015;187:60–72. 10.1016/j.enggeo.2014.12.003.Search in Google Scholar
[17] Huang FM, Zhang J, Zhou CB, Wang Y, Huang J, Zhu L. A deep learning algorithm using a fully connected sparse auto encoder neural Network for landslide susceptibility prediction. Landslides. 2020;17(5):217–29. 10.1007/s10346-019-01274-9.Search in Google Scholar
[18] Ma CH, Yang J, Cheng L, Ran L. Research on slope reliability analysis using multi-kernel relevance vector machine and advanced first-order second-moment method. Eng Computers. 2022;38:3057–68. 10.1007/S00366-021-01331-9.Search in Google Scholar
[19] Sarma SK. Stability analysis of embankments and slopes. Géotechnique. 1973;23(3):423–33. 10.1680/geot.1973.23.3.423.Search in Google Scholar
[20] Zhu DY, Lee CF, Jiang HD. Generalised framework of limit equilibrium methods and numerical procedure for slope stability analysis. Géotechnique. 2003;53(4):377–95. 10.1680/geot.2003.53.4.377.Search in Google Scholar
[21] Melchers RE. Importance sampling in structural systems. Struct Saf. 1989;6(1):3–10. 10.1016/0167-4730(89)90003-9.Search in Google Scholar
[22] Au SK, Wang Y. Engineering risk assessment with Subset Simulation. Singapore: John Wiley & Sons; 2014.Search in Google Scholar
[23] Huang JS, Fenton G, Griffiths DV, Li DQ, Zhou CB. On the efficient estimation of small failure probability in slopes. Landslides. 2017;14(2):491–8. 10.1007/s10346-016-0726-2.Search in Google Scholar
[24] Grooteman F. Adaptive radial-based importance sampling method for structural reliability. Struct Saf. 2007;30(6):533–42. 10.1016/j.strusafe.2007.10.002.Search in Google Scholar
[25] Sarma SK. Stability analysis of embankments and slopes. J Geotech Geoenviron Eng Div, ASCE. 1979;105(12):1511–24. 10.1061/AJGEB 6.0000903.Search in Google Scholar
[26] Halatchev RA. Probabilistic stability analysis of embankments and slopes. In: Proceedings of 11th International Conference on Ground Control in Mining. Australia. NSW Wollongong: 1992 July 7–10. p. 432–7. Conference paper.Search in Google Scholar
[27] Sarma SK, Bhave MV. Critical acceleration versus static factor of safety in stability analysis of earth dams and embankments. Géotechnique. 1974;24(4):661–5. 10.1680/geot.1974.24.4.661.Search in Google Scholar
[28] Metropolis N, Rosenbluth AW, Rosenbluth MN, Teller AH, Teller E. Equation of state calculations by fast computing machines. J Chem Phys. 1953;21(6):1087–92. 10.1063/1.1699114.Search in Google Scholar
[29] Hastings WK. Monte Carlo sampling methods using Markov chains and their applications. Biometrika. 1970;57(1):97–109. 10.2307/2334940.Search in Google Scholar
[30] Au SK, Beck JL. Estimation of small failure probabilities in high dimensions by subset simulation. Prob Eng Mech. 2001;16(4):263–77. 10.1016/S0266-8920(01)00019-4.Search in Google Scholar
[31] Zhu DY, Lee CF, Huang MS, Qian QH. Modifications to three well-known methods of slope stability analysis. Chin J Rock Mech Eng. 2005;24(2):183–94. 10.1007/s11769-005-0030-x.Search in Google Scholar
[32] Giam PSK, Donald IB. Example problems for testing soil slope stability programs. Civil Eng Res Rep No. 8/1989. Melbourne, Australia: Monash University.Search in Google Scholar
[33] Cheng YM, Li L, Liu LL. Simplified approach for locating the critical probabilistic slip surface in limit equilibrium analysis. Nat Hazards Earth Syst Sci. 2015;15(10):2241–56. 10.5194/nhess-15-2241-2015.Search in Google Scholar
[34] Xu B, Low BK. Probabilistic stability analyses of embankments based on finite-element method. J Geotech Geoenviron Eng. 2006;132(11):1444–54. 10.1061/(ASCE)1090-0241(2006)132:11(1444).Search in Google Scholar
© 2023 the author(s), published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.