Probabilistic Stability Frameworks of Stone Column–Supported Embankments Considering Failure Mechanisms

Column-supported embankments can exhibit various failure mechanisms when suffering from instabilities. This study presents a novel framework for the probabilistic assessment of stone column–supported embankment stability, focusing specifically on two primary failure modes: lateral sliding and deep-seated failure. The framework introduces the first-order reliability subset simulation method (FORSSM), which integrates the subset simulation (SS) method with the first-order reliability method (FORM). This method substantially improves both accuracy and computational efficiency. On the basis of the framework, variations in input parameters such as the embankment friction angle (ϕemb), embankment unit weight (γemb), soil cohesion (cs), and column friction angle (ϕcol) are considered in terms of failure probabilities and failure mechanisms. Compared with the failure mechanism in the deterministic analysis, the probabilistic results quantitatively show that a high failure probability with a large variation in embankment stability is induced by changes in the failure mechanism. A greater variability of slip surface sizes for various failure mechanisms is revealed in deep-seated failure than in lateral sliding, which is linked with a high failure probability. Furthermore, this study explores the impact of input parameters on the factor of safety via Sobol indices in conjunction with the FORSSM results.