Probabilistic stability analyses of the landslide-stabilizing piles system considering the spatial variability of geotechnical parameters

In the reliability analysis of the landslide-stabilizing piles system (LSP), the most commonly used uncertainty models of geotechnical parameters mainly include the homogeneity model (HM) and unconditional random field (UCRF). However, neither of these models considers the existence of borehole data, which may lead to overestimation of the degree of site uncertainty. Based on borehole data in situ, the conditional random field (CRF) of the shear strength of the sliding surface was established through random field theory and the Kriging estimation method. By combining the limit equilibrium method (LEM) and Monte Carlo simulation (MCS), a novel procedure was introduced that can calculate the stability coefficient and design thrust of pile stabilization at the same time. By considering multiple failure modes comprehensively, a calculation flowchart for the failure probability of the LSP was proposed. The spatial layout parameters of pile stabilization, including the pile location, pile spacing, and pile length, significantly impact the calculation results. Compared with using HM and UCRF, the failure probability of the LSP may be lower when using CRF. Based on the allowable failure probability, feasible design solutions were provided. Further comparisons were made on the number of feasible solutions under these three uncertainty models. The research results in this paper might provide some assistance for the optimization of pile stabilization.