Multifactor Regression Analysis for Predicting Embankment Dam Breaching Parameters

Abstract

Embankment dams are common worldwide, and the majority of past dam failures are related to such dams. It is imperative to improve the prediction accuracy of embankment dam failure to better estimate the potential impacts on downstream areas and create effective emergency response plans. A substantial database of 162 case studies was established based on previous studies. Three groups of regression equations relating to peak outflow discharge Qp, mean breach width Bm and breach height Hb were developed as a function of two hydraulic conditions (volume of water above breach bottom Vw, and depth of water above breach bottom Hw). These proposed prediction formulas have good prediction accuracy and statistical significance. Comparative analyses are presented to further validate the accuracy of the prediction formulas developed in this study. The results of relative weights analysis indicate that both Vw and Hw exert a significant impact on Qp, and the effect of Vw on Bm is significantly higher than that of Hw on Bm, while Hw is considered a prominent factor of Hb. In addition, two main factors leading to embankment dam failure, overtopping and piping, were proven to cause a significant difference in the three breaching parameters (Qp, Bm, and Hb).