Abstract: How to effectively assess the system reliability of slope containing unlimited number of potential slip surfaces and how to accurately identify key slip surfaces significantly contributing to slope failure are critical questions in slope engineering. To address these questions, this study proposed an approach to assess slope system reliability and to identify the representative slip surfaces based on generalized subset simulation (GSS). The equations for the evaluation of slope and quantification of the relative contribution of individual slip surface to slope are derived in this study. Using the reliability analysis results of GSS, the independent representative slip surfaces (RSSs) are identified from possible slip surfaces using probabilistic network evaluation technique (PNET). Finally, the proposed approach is verified using examples of a two-layered cohesive slope and a case of Congress Street Cut in Chicago. Results show that the proposed approach provides proper estimates of failure probabilities of individual slip surfaces and the slope system simultaneously by a single GSS run. This can avoid repeated simulations for each failure mode. The proposed approach is more efficiency than Monte Carlo simulation (MCS) especially for failure modes at low probability level and simultaneously overcome the limitations of subset simulation (SS). The RSSs are selected using PNET after obtaining the failure probabilities of the slope system and its corresponding potential slip surfaces. The slope system can be represented effectively by the RSSs identified in this study. The proposed approach effectively avoids the dependence of the accuracy of system failure probability on the number of RSS and the accuracy of failure probability of RSS.

Key words: slope, generalized subset simulation, probabilistic network evaluation technique, system reliability, potential slip surfaces, representative slip surfaces