Reliability analysis of soil-rock composite slopes based on dimensionality reduction response surface method

To address the “curse of dimensionality” caused by high-dimensional parameters and the computational bottlenecks of 3D numerical simulations in the reliability evaluation of “overlying silty clay and underlying weathered mylonite” composite strata slopes, a reliability analysis framework combining 2D limit-equilibrium screening and a response-surface surrogate constructed from quasi-3D FLAC3D simulations is proposed. First, by combining the limit equilibrium method with an orthogonal experiment and multivariate analysis of variance (ANOVA), the 9-dimensional random variable space is reduced, identifying the cohesion, internal friction angle, and unit weight of the moderately weathered mylonite as the dominant parameters. Secondly, based on the equivalent continuum assumption for the fractured weathered rock mass, a quasi-3D finite-difference model implemented in FLAC3D is constructed, which demonstrates the progressive development of a deep-seated rotational failure mechanism in the composite slope. According to this, a second-order response surface surrogate model with a coefficient of determination R 2 = 0.9990 is developed using the Box–Behnken Design (BBD), which effectively captures the nonlinear coupling effects among the controlling parameters. Lastly, 100,000 Monte Carlo simulations using truncated-normal input distributions yield a baseline slope failure probability of 2.767% and an equivalent reliability index of 1.916. A Gaussian-Copula sensitivity analysis further shows that the estimated failure probability is influenced by the assumed dependence between cohesion and internal friction angle. The results demonstrate that probabilistic parameter variability can produce a non-negligible failure risk even when the deterministic mean - value condition gives F o S > 1.0 . This dimensionality-reduction response-surface framework provides an efficient approach for the probabilistic reliability assessment of heterogeneous Soil-Rock composite slopes.