Objective Three-dimensional (3D) slope reliability analysis encounters two primary challenges: 1) the stochastic modeling of soil properties involves complex spatial variability, requiring extensive data processing and imposing substantial computational demands; and 2) brute-force trial-and-error methods for locating critical failure surfaces are computationally inefficient, hindering real-world implementation despite the need for precise stability evaluations in civil infrastructure. Therefore, this study employs the covariance matrix decomposition method to generate 3D lognormal random fields for soil parameters, enabling efficient modeling of spatial variability. The particle swarm optimization (PSO) algorithm is refined with enhanced termination criteria and integrated with the 3D Bishop method to search for the minimum factor of safety (F). This approach significantly improves the accuracy and computational efficiency of slope reliability analysis. Methods The core of this framework lies in employing a stepwise covariance matrix decomposition method to generate 3D lognormal random fields for slope soil parameters. The stepwise covariance matrix decomposition efficiently partitions the problem into tractable components, drastically reducing computational demands while enabling high-resolution random field generation with limited resources by systematically decomposing the covariance matrix. This decomposition facilitated rapid simulation of spatially variable soil parameters without excessive overhead. An enhanced PSO algorithm was proposed for 3D slope stability analysis to complement the random field modeling. The algorithm integrated the 3D Bishop method, a well-established limit equilibrium technique, into PSO's global search mechanism, combining PSO's exploratory capacity with the Bishop method's precise stability calculations. This interaction enhanced the identification of critical slip surfaces. Refined iteration termination criteria were incorporated to expedite convergence toward the minimum factor of safety (F) and the corresponding critical surface, improving computational efficiency for reliability analysis. The failure probability of 3D slopes was assessed through Monte Carlo simulation, which accounted for inherent soil property uncertainties. The method's accuracy and effectiveness were validated through numerical examples that systematically analyzed diverse slip surface geometries (for example, cylindrical and cylindrical-ellipsoidal combinations). Results and Discussions The proposed method was applied to analyze Example 1. Excluding spatial variability of soil parameters and assuming a cylindrical sliding surface, a deterministic analysis yielded an F of 1.352 4, which closely matched previous result of 1.352 5. For a sliding surface combining cylindrical and ellipsoidal shapes, F decreased as the failure surface width (B) increased, gradually approaching the two-dimensional analysis outcome. When spatial variability of soil parameters was incorporated, the calculated failure probability (Pf) range of 0.090 5‒0.091 6 (average 9.11%) for Example 1 closely aligned with previous finding of 9.25%, with a relative error of approximately 1.5%. The comparative analysis of safety factors and failure probabilities confirmed the accuracy of the proposed method and the computational procedure. Systematic studies were conducted for various slip surface forms that encompassed combinations of cylindrical and cylindrical + ellipsoidal shapes, as well as different correlation lengths and coefficients of variation. When the sliding surface was cylindrical, and the characteristic length scale in the out-of-plane direction (ly ) was set to 20 m, Pf gradually decreased as B increased. This observation indicated that the spatial variability of soil parameters was one of the pivotal factors contributing to the enhanced stability assessment accuracy of 3D slope analysis compared to its two-dimensional counterparts. When ly approaches infinity (ly →+∞), as B increases, Pf gradually rises while the average safety factor (Fav ) decreases. These findings indicated that B significantly influences the stability of 3D slopes. When the coefficient of variation for cohesion (Vc ) increased from 0.1 to 0.6, Pf rose from 0.0 to 0.25, reflecting an approximate increment of 0.25. When the coefficient of variation for the internal friction angle (Vφ ) increased from 0.1 to 0.4, Pf increased from 0.04 to 0.14, corresponding to an approximate increment of 0.10. When the correlation coefficient between cohesion and the internal friction angle (ρc,φ ) increased from -0.5 to 0.5, Pf rose from 0.055 to 0.070, indicating an approximate increment of 0.015. These findings demonstrated that Vc, Vφ, and ρc,φ all exert significant influences on the failure probability. When the characteristic length scale in the x-direction (lx ) increased from 10 m to 40 m, Pf increased by approximately 0.03. In contrast, when ly increased from 10 to 40 m, Pf increased by approximately 0.075. This result indicated that ly had a more pronounced impact on slope failure probability compared to lx . In addition, when the characteristic length scale in the z-direction (lz ) increased from 1 to 4 m, Pf increased by approximately 0.05, indicating an influence on failure probability. Conclusions The results demonstrate that the enhanced PSO algorithm, when integrated with the 3D Bishop method, markedly improves the computational efficiency of 3D slope reliability analysis without compromising accuracy. When only spatial variability is considered, the failure probability decreases as the width of the failure surface increases. In contrast, when only the 3D effect is considered, the failure probability increases with the width of the failure surface. When both factors are considered simultaneously, their effects on the failure probability tend to offset each other, resulting in a significantly lower failure probability for 3D slopes compared to 2D slopes. When accounting for the variability of soil parameters and 3D effects, the failure probability of slopes initially increases and then decreases with the widening of the failure surface. This pattern indicates that 3D effects dominate when the failure surface is narrow, because geometric constraints and stress redistribution play a critical role. In contrast, spatial variability of soil properties becomes the primary factor when the failure surface exceeds a critical width, because material heterogeneity governs the failure process. The analysis reveals a critical failure surface width that marks the transition from geometry-controlled to material-controlled failure mechanisms in slope stability assessment.
ChoS E.Probabilistic assessment of slope stability that considers the spatial variability of soil properties[J].Journal of Geotechnical and Geoenvironmental Engineering,2010,136(7):975‒984. doi:10.1061/(asce)gt.1943-5606.0000309
[3]
WuXingzheng, JiangLiangwei, LuoQiang,et al.Analysis of model uncertainty for stability reliability of embankment slope[J].Rock and Soil Mechanics,2015,36(Supp2):665-672.
LiuLeilei, DengZhiping, ZhangShaohe,et al.Simplified framework for system reliability analysis of slopes in spatially variable soils[J].Engineering Geology,2018,239:330‒343. doi:10.1016/j.enggeo.2018.04.009
[8]
WangYuke, ShaoLinlan, WanYukuai,et al.Reliability analysis of earth rock dam slope considering spatial variability of soil parameters based on K‒L expansion method[J].Journal of Hydraulic Engineering,2024,55(12):1417‒1427.
ZhangLiang, GongWenping, LiXinxin,et al.A comparison study between 2D and 3D slope stability analyses considering spatial soil variability[J].Journal of Zhejiang University:Science A,2022,23(3):208‒224. doi:10.1631/jzus.a2100139
[13]
WuDi, WangYuke, ChenXin,et al.Limit analysis of 3d soil slopes considering the earthquake and nonlinear strength[J].Chinese Journal of Theoretical and Applied Mechanics,2024,56(5):1426‒1438.
GriffithsD V, HuangJinsong, FentonG A.On the reliability of earth slopes in three dimensions[J].Proceedings of the Royal Society A:Mathematical,Physical and Engineering Sciences,2009,465(2110):3145‒3164. doi:10.1098/rspa.2009.0165
[16]
HicksM A, SpencerW A.Influence of heterogeneity on the reliability and failure of a long 3D slope[J].Computers and Geotechnics,2010,37(7/8):948‒955. doi:10.1016/j.compgeo.2010.08.001
[17]
HicksM A, NuttallJ D, ChenJ.Influence of heterogeneity on 3D slope reliability and failure consequence[J].Computers and Geotechnics,2014,61:198‒208. doi:10.1016/j.compgeo.2014.05.004
[18]
KasamaK, FurukawaZ, HuLihang.Practical reliability analysis for earthquake-induced 3D landslide using stochastic response surface method[J].Computers and Geotechnics,2021,137:104303. doi:10.1016/j.compgeo.2021.104303
[19]
VarkeyD, HicksM A, VardonP J.An improved semi-analytical method for 3D slope reliability assessments[J].Computers and Geotechnics,2019,111:181‒190. doi:10.1016/j.compgeo.2018.12.020
[20]
ShuShuang, GeBin, WuYongxin,et al.Probabilistic assessment on 3D stability and failure mechanism of undrained slopes based on the kinematic approach of limit analysis[J].International Journal of Geomechanics,2023,23:06022037. doi:10.1061/(asce)gm.1943-5622.0002635
[21]
JiangShuihua, LiJianping, MaGuotao,et al.Probabilistic assessment of 3D slope failures in spatially variable soils by cooperative stochastic material point method[J].Computers and Geotechnics,2024,172:106413. doi:10.1016/j.compgeo.2024.106413
[22]
HuLihang, TakahashiA, KasamaK.Effect of spatial variability on stability and failure mechanisms of 3D slope using random limit equilibrium method[J].Soils and Foundations,2022,62(6):101225. doi:10.1016/j.sandf.2022.101225
[23]
WangYuke, ShaoLinlan, WanYukuai,et al.Three-dimensional reliability stability analysis of earth-rock dam slopes reinforced with permeable polymer[J].Probabilistic Engineering Mechanics,2023,74:103537. doi:10.1016/j.probengmech.2023.103537
[24]
WangYuke, ShaoLinlan, WanYukuai,et al.Reliability analysis of three-dimensional reinforced slope considering the spatial variability in soil parameters[J].Stochastic Environmental Research and Risk Assessment,2024,38(4):1583‒1596. doi:10.1007/s00477-023-02636-5
[25]
LeeS W, ChingJ.Simplified risk assessment for a spatially variable undrained long slope[J].Computers and Geotechnics,2020,117:103228. doi:10.1016/j.compgeo.2019.103228
[26]
LiDianqing, XiaoTe, ZhangLimin,et al.Stepwise covariance matrix decomposition for efficient simulation of multivariate large-scale three-dimensional random fields[J].Applied Mathematical Modelling,2019,68:169‒181. doi:10.1016/j.apm.2018.11.011
[27]
NgC W W, QuChuanxiang, NiJunjun,et al.Three-dimensional reliability analysis of unsaturated soil slope considering permeability rotated anisotropy random fields[J].Computers and Geotechnics,2022,151:104944. doi:10.1016/j.compgeo.2022.104944
[28]
WanYukuai, XuRenhao, YangRong,et al.Application of the improved particle swarm optimization method in slope probability analysis[J].Marine Georesources & Geotechnology,2024,42(10):1531‒1541. doi:10.1080/1064119x.2023.2291173
[29]
ZhouX P, HuangX C, ZhaoX F.Optimization of the critical slip surface of three-dimensional slope by using an improved genetic algorithm[J].International Journal of Geomechanics,2020,20(8):04020120. doi:10.1061/(asce)gm.1943-5622.0001747
[30]
SunCong, ZhengHong, LiChunguang,et al.Critical slip surface search by using rigorous maximum residual thrust method based on ant colony algorithm[J].Rock and Soil Mechanics,2014,35(10):3021‒3026.
ChengY M, LiL, ChiS C.Performance studies on six heuristic global optimization methods in the location of critical slip surface[J].Computers and Geotechnics,2007,34(6):462‒484. doi:10.1016/j.compgeo.2007.01.004
[33]
KalatehjariR, AliN, HajihassaniM,et al.The application of particle swarm optimization in slope stability analysis of homogeneous soil slopes[J].International Review on Modelling and Simulations,2012,5(1):458‒465.
[34]
ShinodaM, MiyataY.PSO‒based stability analysis of unreinforced and reinforced soil slopes using non-circular slip surface[J].Acta Geotechnica,2019,14(3):907‒919. doi:10.1007/s11440-018-0678-x
[35]
HimanshuN, BurmanA.Determination of critical failure surface of slopes using particle swarm optimization technique considering seepage and seismic loading[J].Geotechnical and Geological Engineering,2019,37(3):1261‒1281. doi:10.1007/s10706-018-0683-8
[36]
KumarV, BurmanA, HimanshuN,et al.Rock slope stability charts based on limit equilibrium method incorporating Generalized Hoek‒Brown strength criterion for static and seismic conditions[J].Environmental Earth Sciences,2021,80(6):212. doi:10.1007/s12665-021-09498-6
[37]
KalatehjariR, ArefniaA, Rashid A SA,et al.Determination of three-dimensional shape of failure in soil slopes[J].Canadian Geotechnical Journal,2015,52(9):1283‒1301. doi:10.1139/cgj-2014-0326
[38]
HungrO, SalgadoF M, ByrneP M.Evaluation of a three-dimensional method of slope stability analysis[J].Canadian Geotechnical Journal,1989,26(4):679‒686. doi:10.1139/t89-079
[39]
ZhuLei, GaoXinyue, WanYukuai,et al.Effect of seismic force on slope reliability[J].Science Technology and Engineering,2023,23(10):4324‒4330.
WangBin, LiuLeilei, LiYuehua,et al.Reliability analysis of slopes considering spatial variability of soil properties based on efficiently identified representative slip surfaces[J].Journal of Rock Mechanics and Geotechnical Engineering,2020,12(3):642‒655. doi:10.1016/j.jrmge.2019.12.003