Probabilistic analysis of width-limited 3D slope considering spatial variability of Hoek-Brown rock masses

The inherent spatial variability of rock mass strength has been explicitly considered in reliability analyses of slopes governed by the Hoek-Brown(HB) criterion. However, previous studies have primarily focused on extensive slopes where plane strain analysis is applicable, neglecting rock slopes constrained by boundaries that exhibit significant 'end effects' and are unsuitable for two-dimensional(2D) analysis. To bridge this gap, this research presents a novel three-dimensional(3D) reliability framework designed specifically for width-limited slopes in spatially variable rock masses. Considering the efficiency limitation of previously widely adopted numerical simulations, which struggle to accommodate extensive computation task, this study constructs the deterministic model using upper bound limit analysis (UBLA). A discretized mechanism is developed to determine the safety factor of spatially variable HB slopes. The integration of accelerated search strategies enables the determination of a solution to safety factor within a mere 5 min, mitigating the computational burden associated with high-dimensional stochastic problems or scenarios with a low probability of failure. Probabilistic analysis is conducted using a metamodel Sparse Polynomial Chaos Expansion (SPCE) in conjunction with Monte Carlo Simulation (MCS). Parametric analysis is employed to investigate the influence of various factors on slope reliability, including the autocorrelation length, coefficient of variation of strength, and correlation coefficient. This research presents a novel, computationally efficient deterministic model for slopes characterized by spatial variability in HB strength parameters. Furthermore, the employed principles show promising applicability in adjacent fields, such as tunneling and foundation engineering. Develops a novel 3D reliability framework for width-limited slopes in spatially variable rock masses governed by the Hoek-Brown criterion.Develops an efficient UBLA-based deterministic model with a 3D spatial discretized technique and accelerated search strategies, reducing Fs computing time to within 5 mins.Integrates SPCE metamodel and MCS for probabilistic analysis, enabling investigation of various factors on slope reliability.The employed principles demonstrate promising applicability in adjacent fields, such as tunneling and foundation engineering.