MODEL CORRECTION FACTOR METHOD BASED APPROACH FOR RELIABILITY ANALYSIS OF SPATIALLY VARIABLE SLOPES

Reliability analysis of spatially variable slopes involves repeatedly evaluating the slope stability using a deterministic analysis method such as the limit equilibrium method (LEM) or the finite element method (FEM). The LEM is conceptually simple and computationally efficient, while the FEM tends to give a more realistic prediction of slope failure mechanism, particularly when considering the spatial variability of soil properties. Thus, it is of great interest to adopt advantages of both LEM and FEM in estimating the reliability of slope stability. This paper aims to propose a model correction factor method (MCFM) based approach for reliability analysis of spatially variable slopes. In this approach, first-order reliability method is adopted for slope reliability analysis considering spatial variation. A model correction factor is introduced to modify the idealized LEM-based limit-state surface to the more accurate FEM-based limit-state surface. Finally, a line sampling is adopted to estimate the probability of slope failure based on the corrected LEM model with the model correction factor. The reliability assessment of a cohesive-frictional slope example is studied to investigate the performance of the proposed approach considering the spatial variability of the soil strength parameters. The results indicate the proposed approach not only provides an accurate estimation of probability of failure consistent with that obtained from the FEM-based subset simulation, but also significantly reduces the number of finite element analyses of slope stability. Thus, it provides an effective and versatile tool for slope reliability analysis at low-probability levels considering the spatial variability of the soil strength parameters.