Author(s): Andre Paquier
Linked Author(s): André Paquier
Keywords: Breach; Numerical model; Dam failure; Risk analysis
Abstract: During the European project IMPACT (2001-2004), five field tests and about 20 laboratory tests were performed to study the erosion of an embankment caused by a breach. Some of the tests were used for benchmarking during IMPACT project. Several numerical models were used; among these models, RUPRO model developed by Cemagref proved to provide suitable results except for the width of the breach that was too high in the calculations. Consequently, it was decided to test modifications to RUPRO model in order to obtain a better estimate of breach width. Several options are implemented; they change the way the sediment discharge and the distribution of erosion between walls and bottom are calculated. The new versions of the model are tested on four field tests and eight laboratory tests that were used for blind tests during IMPACT project. Two of the field tests deal with piping while all the other tests concern overtopping. Comparisons are based on the three parameters that were measured for all these 12 test cases: peak discharge hydrograph, time to peak discharge and final breach width. For these three parameters, one of the proposed methods provides ratios of calculated value to measured value close to 1 (0. 97 to 1) with an average deviation less than 30% ; particularly, it provides a suitable estimate for the final breach width and the calculation can follow the evolution of the breach width. With the proposed changes to RUPRO model, the results obtained for IMPACT test cases lead to conclude that the proposed model can be a useful tool to estimate the consequences of the failure of an earthen dam. Moreover, it seems that RUPRO model can be used for the failure of levees, which requires a suitable estimate of the breach width.