Abstract
In a probabilistic slope stability analysis, the failure probability associated with the most critical slip surface (the one with the minimum reliability index) is known to be smaller than that for the system that comprises all potential slip surfaces. The first order reliability method (FORM) targets the minimum reliability index related to the critical slip surface, and thus cannot be used to predict the system reliability of slopes, except when all possible slip surfaces are perfectly correlated. It is shown in this paper that the random finite element method (RFEM), which uses elastoplastic finite elements combined with random field theory in a Monte-Carlo framework can accurately predict the system probability of failure (pf) of slopes.
