Experimental investigation of fluvial dike breaching due to flow overtopping

The failure of fluvial dikes (levees) often leads to devastating floods that cause loss of life and damages to public infrastructure. Overtopping flows have been recognized as one of the most frequent cause of dike erosion and breaching. Fluvial dike breaching is different from frontal dike (embankments) breaching, because of specific geometry and boundary conditions. The current knowledge on the physical processes underpinning fluvial dike failure due to overtopping remains limited. In addition, there is a lack of a continuous monitoring of the 3D breach formation, limiting the analysis of the key mechanisms governing the breach development and the validation of conceptual or physically-based models. Laboratory tests on breach growth in homogeneous, non-cohesive sandy fluvial dikes due to flow overtopping have been performed. Two experimental setups have been constructed, permitting the investigation of various hydraulic and geometric parameters. Each experimental setup includes a main channel, separated from a floodplain by a dike. A rectangular initial notch is cut in the crest to initiate dike breaching. The breach development is monitored continuously using a specific developed laser profilometry technique. The observations have shown that the breach develops in two stages: first the breach deepens and widens with the breach centerline being gradually shifted toward the downstream side of the main channel. This behavior underlines the influence of the flow momentum component parallel to the dike crest. Second, the dike geometry upstream of the breach stops evolving and the breach widening continues only toward the downstream side of the main channel. The breach evolution has been found strongly affected by the flow conditions (i.e. inflow discharge in the main channel, downstream boundary condition) and floodplain confinement. The findings of this work shed light on key mechanisms of fluvial dike breaching, which differ substantially from those of dam breaching. These specific features need to be incorporated in flood risk analyses involving fluvial dike breach and failure. In addition, a well-documented, reliable data set, with a continuous high resolution monitoring of the 3D breach evolution under various flow conditions, has been gathered, which can be used for validating numerical models.