Adaptive meshing methodology based on topometric data for dambreak inundation assessments

Flood maps are the final products of dam failure studies that are required by dam safety regulations. A flood limit, which represents the maximum envelope reached by flood waves, is generally the result of a dam-break scenario simulated by a hydraulic numerical model. However, the numerical model uses only a limited portion of the available bathymetry data to build the terrain model (2D mesh plus topometric elevation at nodes). This is particularly so in the cases where the topo-metric data recorded by LIDAR was estimated in several million points. But the hydraulic numerical models rarely exceed hundreds of thousands of nodes, in particular because of the computer constraints and time associated with the operation of these models. The production of the final flood map requires consistency between projected levels and elevations for all points on the map. This verification may be tedious for a large area with several small secondary valleys of tributary streams that have not been represented by the original hydraulic numerical model. The aim of this work is to propose an automatic remeshing strategy that uses the envelope of the maximum dimensions reached by the original model coupled with the available LIDAR data to produce an improved mesh that can accurately capture the wet/dry fronts and the overflows of the secondary valleys. This model helps us to consider the maximum slope inside each element on the basis of the real data, instead of controlling the slope for not having negative depth or controlling the velocity. The algorithm is based on a few basic steps: (i) find the elements cut by the envelope of the wet/dry interfaces; (ii) project the topometric points onto the cut elements; (iii) if these points are very close to the interface, if they are found in a valley, or if they are more elevated than the corresponding cut elements, then these points will be added to the previous nodes and included in a subsequent triangulation step; and (iv) re-run the simulation on the new mesh. This algorithm has been implemented and validated in the study of a dambreak flow with a complex river topography on the Eastmain River and the Romaine-Puyjalon River.