Within the frame work of the realization of the 'Sigmaplan' for the river Schelde in Flanders (Belgium), a large-scale dike breaching experiment following overflow was organized at Lillo (Antwerp) in 2012. The outcomes of the breach test serve now to unveil the impact of selected breaching models and provide guidance for proper model selection and usage.
The main use of breaching models includes two tasks: predicting breach characteristics and estimating flow through the breach. The breach growth models considered in this study are the following: breaching module of Mike 11, HR Breach (InfoWorks RS), Verheij-van der Knaap breach formulae, BRES-Visser and AREBA-TUD. Besides a short description on how the dike breaching processes are taken into account by each model, selected model options are explained in the paper.
The main conclusion of this modelling exercise would be that useful results can be obtained with all models starting from the same input regarding the soil and strength parameters of the dike. However, in some cases foreknowledge is needed when executing the simulations as well was interpreting the results. In addition, as most models take only a (limited and sometimes different) set of breaching mechanisms into account, the use of today’s state-of-the-art breach growth models is not (yet) dummy proof. Whether and how retrograde surface erosion and/or upstream headcut migration are accounted for will have an important influence on the timing of the peak flow through the breach. Furthermore, even when stated that block failure is included in a model, a stepwise increase of the breach width is never observed in the model output due to implicit assumptions in some models.