Abstract: In the reliability analysis of slope stability, the deterministic analysis method is usually used to calculate the safety factor to evaluate the stability of slope. However, the inherent spatial variability of rock mass properties cannot be considered and described adequately in traditional deterministic method, resulting in the inaccurate calculation of slope failure probability. Based on Hoek-Brown criterion and random finite difference method (RFDM), the reliability analysis of slope stability and random response of pile are discussed in this paper considering spatial variability of rock mass. The uniaxial compressive strength and material constant for the intact rock are regarded as random field variables and geological strength index GSI is assumed to be a random variable. The results show that the spatial variability of rock mass parameters has a significant effect on slope failure probability and pile response. Ignoring the spatial variability of rock mass parameters will overestimate slope failure probability and the mean value of the maximum bending moment of anti-slide pile, and underestimate the mean value of displacement at pile head. The results can provide design guidance for slope reinforcement as well as layout optimization of anti-sliding piles.

Key words: reliability assessment of slopes, spatial variability, Hoek-Brown criterion, random finite difference method (RFDM), random response