An Investigation of Three Probabilistic Approaches for Levee Underseepage Analysis

Abstract

A variety of probabilistic approaches can be applied when performing levee underseepage analyses, as part of larger assessments of the reliability or probability of failure of a levee system. Historically, much of the work that has been conducted in this area has used probabilistic approaches that utilize a mean value first order second moment (MVFOSM) technique, due to the inherent simplicity of this probabilistic approach. This paper investigates the effect of three probabilistic approaches for assessing the likelihood of failure of a levee via an underseepage mechanism: MVFOSM, advanced first order second moment (AFOSM), and Monte Carlo simulation (MCS). For each of these approaches, two limit state criteria are explored: safety margin and safety factor. Analyses are conducted in accordance with the current U.S. Army Corps of Engineers levee design code, and representative results from some classic USACE design cases are presented. For the various design cases that were examined, results were found to be sensitive to both the probabilistic approach that was utilized and the limit state criterion that was selected. The results provide useful guidance for practicing engineers that are trying to select a probabilistic approach for these types of analyses that is both accurate and as computationally efficient as possible.