Predicting peak breach discharge due to embankment dam failure is of vital importance for dam failure prevention and mitigation. Because, when dams fail, property damage is certain, but loss of life can vary depending on flood area and population. Many parametric breach models based on regression techniques have been developed so far. In this study, an efficient model is proposed to forecast peak discharge from the height of the water and volume of water behind the dam at failure, respectively, by using the Kriging approach. The previous studies, which consist of 13 numerical models, are used as a benchmark for testing the proposed new model, by employing five different error criteria. Moreover, a new database is compiled by extending the previous one. In addition, it is demonstrated that R2, which only quantifies the dispersion between measurements and predictions, should not be considered alone for checking the model capabilities. At least, the other criteria should be employed together with R2. As a result, it is shown that one can forecast the peak flow discharge with more significant accuracy by the proposed model than other previous models developed so far.
Review of Brumadinho tailings dam failure, the subsequent loss of life and how it could have been reduced
