A Multiple Response-Surface based Subset Simulation for Slope Reliability Analysis in Spatially Variable Soils

摘要

This paper proposes a multiple response-surface based Subset simulation approach for slope reliability analysis at low-probability levels in spatially variable soils.An example of reliability analysis of the Congress Street cut slope is presented to illustrate the effectiveness of the proposed method.The results indicate that the proposed approach possesses the following advantages: (1) it can properly evaluate the slope reliability at low-probability levels (e.g., pf ＜10-6) in spatially variable soils; (2) greatly improve the computational efficiency in parametric sensitivity analysis, and provide an effective way to investigate effects of statistics (e.g., probability distribution, scale of fluctuation) on the slope reliability.Additionally, the probabilities of slope failure associated with undrained shear strengths following the commonly-used Gaussian and lognormal distributions differ considerably.