Enhanced Predictions for Peak Outflow from Breached Embankment Dams

A study was conducted to enhance the historic regression relations that predict the peak discharge from breached embankment dams. Forty-four dam breach case studies were collected and added to an existing database resulting in a composite database of 87 case studies. The composite database was evaluated and a statistical analysis performed using regression techniques. Peak outflow $\left(Q_p\right)$ prediction expressions from breached embankment dams were developed as a function of the height of the dam $\left(H\right)$ , the volume of water behind the dam $\left(V\right)$ , the embankment length $\left(L\right)$ , the average embankment width $\left(W_{\text{ave}}\right)$ , and a combination of these variables. The multivariate regression analysis indicated that a series of expressions may be formulated relating peak outflow as a function of $H\cdot V\cdot L$ and $H\cdot V\cdot W_{\text{ave}}$ . The newly developed expressions derived from the expanded database appear to reduce the conservatism in predicting the peak discharge from a breached embankment, reduce the prediction error, and reduce the uncertainty bandwidth while improving the prediction correlation. The available data are limited and the quality of the composite database is quite variable. The study results strongly suggest that the art and science of dam breach forensics (i.e., data acquisition, quality, and availability) has not changed since inception and must be improved to enhance regression prediction credibility.