Modele simplifie de formation de breche par debordement

Breach formation in a rockfill or earthfill dike due to overtopping of the crest is a complex process involving interaction between the hydraulics of the flow and the erosion characteristics of the fill material. This document is devoted to the description and validation of a simple parametric model for breach formation due to overtopping, with an attempt being made to model as closely as possible the physical processes involved within the restriction of the simplified analysis. The objective is to predict the relevant timescales for the phenomenon leading to a prediction of the outflow hydrograph.; Breach initiation and evolution have been well documented on the basis of field observations and laboratory experiments. Most breaches caused as a result of overtopping follow the general sequence: (1) Initial overtopping results in rivulet type erosion on the downstream face of the embankment and formation of a headcut; (2) An eroded channel forms and increases in depth and width; (3) As depth increases, sidewall failure as well as slope failure is possible; (4) The headcut steadily advances upstream until the crest is bridged to reach the upstream water level resulting in a sudden increase in discharge; (5) The breach crest then descends rapidly accompanied by widening. The process continues until the foundation of the dam is reached. From this point on, the breach widens, with no further downward erosion, until the reservoir is evacuated.; The detailed behavior depends on the soil properties and compaction protocol. For example, a purely non-cohesive soil will slump continuously with no appearance of a headcut. Field observations have indicated that the breach geometry tends to be invariably trapezoidal.; The model developed here (AMBRE) uses a broad-crested weir formula for the hydraulics assuming a trapezoidal breach shape. Crest erosion as well as headcut progression is computed using available empirical relations derived from carefully conducted experiments published in the literature. The lateral stability of the sideslopes that may lead to failure due to tensile stress is evaluated based on recent geotechnical criteria. The possible presence of an impermeable core with significantly different material properties is also included in the formulation as is a simple level pool routing technique to compute reservoir water levels. Computational times are extremely low, this being an advantage of the present model over more complicated ones requiring numerical integration of the full differential equations of motion.; AMBRE has been validated on the Oros dam failure in Brazil, Banqjao dam failure in China, Teton Dam in the United States as well as on the recent five and six meter embankment tests carried out at Rosvatn in Norway as part of the CADAM project funded by the European Union. The results are extremely encouraging.