Probabilistic assessment of 3D slope failures in spatially variable soils by cooperative stochastic material point method
Shui‐Hua Jiang, Jianping Li, Guotao Ma, Mohammad Rezania
Abstract
Accurate assessment of slope failures and their large deformations is critical for effective landslide mitigation. This study introduces the new Cooperative Stochastic Material Point Method (CSMPM), addressing challenges in probabilistic characterization of slope large deformations considering three-dimensional (3D) soil heterogeneities. The method employs an enhanced Karhunen-Loève (KL) expansion to model 3D soil spatial variability efficiently. By using rough and refined grids, derived through the enhanced KL expansion, the study achieves computational efficiency without compromising accuracy. By combining the computational advantages of the rough grid with the precision of the refined grid, the CSMPM enables efficient probabilistic analysis of 3D heterogeneous slopes. The results demonstrate its capability to identify slope large deformation failure modes and quantify the associated failure probability. Notably, the shallow failure mode exhibits fan-shaped horizontal diffusion, introducing uncertainty, while the compound failure mode presents challenges in landslide prevention. The progressive failure mode poses the highest hazard. Horizontal heterogeneities significantly influence both large deformation likelihood and failure modes, emphasizing the importance of 3D soil spatial variability in geotechnical reliability assessments. The CSMPM, with its innovative approach, proves to be a practical tool for enhancing our understanding of geohazards and associated uncertainties, as well as large deformations. It provides valuable insights for improving risk assessment of slope hazards.