A Multi-Parameter Correlation for Predicting the Seismic Displacement of an Earth Dam or Embankment

Abstract

Predictive displacement-based methods provide a useful index of the seismic performance of earth dams and embankments and can be used in preliminary assessments of these structures. In practice, simplified Newmark-type sliding block methods are commonly used for this purpose. Using a database of 122 previously published case histories of permanent deformations of earth dams and embankments, the performance of six simplified sliding block models was examined. The results show that all six simplified methods underpredict seismic displacement for many of the embankment and earth dam cases that were examined, sometimes by a significant amount. An empirical correlation was developed by performing linear multiple regression analysis utilizing multiple slope and ground motion input parameters. This approach is believed to more properly reflect strong ground motion characteristics than the use of a single ground motion parameter such as the peak ground acceleration, the approach that has been previously employed in other correlations of this type. After exploring numerous functional forms, the final resulting seismic displacement correlation that was proposed was determined to be a function of the critical acceleration, the critical acceleration ratio, the slope height, the peak ground acceleration, the peak ground velocity, the spectral acceleration, and the predominant period of earthquake shaking. The proposed empirical equation yields better correlation with the case history database than does other existing empirical correlations or simplified sliding block models.