Error Analysis of Predicted Seismic Displacement of Earth Dams Using Simplified Sliding Block Methods

Simplified displacement-based methods developed based on sliding block theory are widely used in geotechnical engineering practice to predict seismic displacements of earth dams and embankments. These methods are typically selected because of the relatively low computational effort that they require, in combination with their perceived tendency to provide a reliable index of seismic slope performance. In this paper, a previously published database of observed seismic displacements for 122 earth dams and embankments was used to evaluate the accuracy of predicted seismic displacements from six simplified sliding block methods. Residual error analysis was performed to compare the model-predicted values with those that were observed after seismic shaking. Results show that all of the models that were examined have the tendency to underpredict seismic displacement, which is of concern as it can lead to unconservative engineering design. This tendency for underprediction cannot be clearly attributed to a single phenomenon — the authors hypothesize that it may be due to a combination of factors including: the inherent limitations and simplifying assumptions of the underlying sliding block theories that were used to develop the simplified sliding block models, the inherent inability of sliding block models to capture the earthquake-induced volumetric compression that occurs in an embankment during shaking, and/or cyclic strain softening during earthquake shaking, which can decrease the critical acceleration during a seismic event.