Efficient modelling of lateral discharge through a dike breach

Breaches in fluvial dikes can lead to major flooding in the hinterland with severe societal and economic consequences. The discharge partitioning at the location of a dike breach is a complex flow phenomenon with 2D and 3D flow features that needs to be predicted accurately for the estimation of flood hazard. Determining the exact location of a potential breach is highly uncertain and so are the circumstances in which it could appear. Therefore, many scenarios should be investigated. Fast and accurate modelling of the discharge partitioning with appropriate simplifications and parameterizations are required to allow for a large number of simulations within reasonable computational time. To achieve this, spatially lumped or one-dimensional flow models have been used in combination with side weir equations. For the first time, the present study systematically assesses the performance of eleven side weir equations for the determination of the lateral discharge through a breach in a dike that is parallel to the flow direction along a straight river reach. These side weir equations were implemented in a zero-dimensional spatially lumped flow model and in a one-dimensional spatially distributed flow model. Both models were evaluated against experimental data from laboratory tests with a side opening that was either fixed, or dynamically evolving. The performance of the side weir equations varied with the experimental data, highlighting the empirical nature of most of these equations. The coupling of the side weir equations with the spatially distributed flow model did not always generate better results than the coupling with the lumped model, which implies that increasing the model complexity does not systematically lead to better predictions of the dike breach discharge.