Integrating Numerical Methods to Assess Failure Probability of Rock Slopes Considering Uncertainties in Mechanical and Hydraulic Properties

Deep-seated landslide formations in rock slopes are common in areas with steep hillside geophysical features and torrential rainfall. These slopes commonly experience heavy rainfall during typhoons and extreme weather conditions, which reduce rock mass strength, leading to the failure of slopes. The Lushan slope in the middle of Taiwan has continuously slid due to typhoons and heavy rainfall for recent decades. The geological conditions and analysis parameters of natural slopes are difficult to grasp causing uncertainties and affecting the slope stability results. Considering these uncertainties, analyzing its collapse probability can provide a more objective assessment of the stability of the slope. This study will use the Finite Element Method (FEM) software PLAXIS 2D Mohr-Coulomb (MC) model and Van-Genuchten (VG) unsaturated model combined with rainfall infiltration displacement coupling analysis to establish and simulate the slope model of the Lushan landslide area from rainfall duration and groundwater level data. The rock mass strength, unsaturated and saturated parameters were back-calculated and sensitivity analyses were performed to explore the impact of these parameters on the rise of groundwater levels. The probability density functions (PDFs) of dependent parameter groups and independent parameters were determined to consider their uncertainties. Stochastic Finite Element Method (SFEM) analysis was conducted by combining Monte Carlo Simulation (MCS) method with FEM to perform random sampling and determine different parameter combinations of the chosen parameters as random variables with uncertainty. Finally, the probability of slope collapse was evaluated by considering the safety factor as the criterion for judgment. The PDF of the safety factor is used to infer the collapse probability of Lushan slopes under the conditions of different return periods and rainfall delays. In this study, the uncertainty of mechanical and hydraulic parameters is considered to explore the probability of deep collapse which can be used as a reference for the risk assessment and warning systems of large-scale collapse.