Evaluation of simplified methods for predicting earthquake-induced slope displacements in earth dams and embankments

This paper provides a review and comparison of existing simplified displacement-based sliding block models. Analyses were performed to evaluate the relative accuracy of fifteen of these simplified models for predicting earthquake-induced displacements. To accomplish this task, the predictive capability of the models was assessed by comparing model predictions with the actual displacements that were observed after earthquake shaking in 122 case histories of earth dams and embankments. The results indicate that the model predictions of displacement were less than the observed displacement for a large majority of the case histories that were examined. The difference between the observed and predicted displacements was relatively large for a significant percentage of the cases, for each model that was examined. The shapes and positions of the models' relative error cumulative distribution functions did not change significantly if the case histories were filtered to include only those with intermediate levels of observed displacement (i.e., 0.01 m < observed displacement < 1 m), which indicates that the simplified models may exhibit the same behavior for cases of small and/or large displacements as they do for cases in the intermediate range, provided that a percentage-based approach such as relative error is used to compare the results from different models.

► Simplified sliding block models can predict earthquake-induced slope displacements.
► Case history database was used to assess prediction accuracy of existing models.
► Existing models predicted displacements that were typically less than observed values.
► Difference between observed and predicted displacements was often large.