Fluvial dike breaching induced by overtopping: lessons from experimental and numerical studies

Overtopping of fluvial dikes (i.e. dykes or levees) can lead to the initiation of breaching and potentially brutal dike failure and inundation of the protected area. Flood risk management and prevention require a precise quantification of the hazard. Accurate estimate of the flow through the breach is paramount, for which a precise understanding of the breach formation and expansion is required. Existing methods are often the result of investigations conducted on overtopping of frontal dikes (embankment dams). The application of such approaches to fluvial dikes is not reliable and processes underpinning breach expansion are still under research. An experimental program was conducted to fill this gap by investigating the physical processes involved in fluvial dike gradual breaching induced by overtopping. Experiments were conducted within the framework of a collaboration between the National Laboratory of Hydraulics and Environment (LNHE) of the R&D division of EDF and the research team of Hydraulics in Environmental and Civil Engineering (HECE) of University of Liège. Experiments were conducted in two distinct experimental setups, each consisting of a main channel and a floodplain area separated by an erodible fluvial dike. The focus was made to study the spatial erosion of homogenous, non-cohesive dikes induced by overtopping. Measurements included continuous scanning of the dike geometry using a non-intrusive method (Laser Profilometry Technique), designed and developed specifically for the present work. Tests conducted under controlled flow and dike configurations allowed assessing the effects of channel inflow discharge, downstream channel regulation system, and floodplain confinement on the breach development and outflow. Effects of main channel size, dike material size, apparent cohesion, and bottom erodibility were studied as well. In the present paper, the most important findings of the experimental study are presented. Additionally, the flow and dike breaching were simulated using the two-dimensional depth averaged code TELEMAC-2D. This allowed assessing the performance of a detailed morphodynamic code for dike breaching processes.