Wave propagation over a submerged bar: benchmarking of VoF, sigma transformation, and SPH numerical models against physical wave flume tests

Abstract Accurate prediction of wave transformation is key in the design of coastal and nearshore structures which typically depends on numerical models. Turbulent and rotational effects call for the use of Computational Fluid Dynamics (CFD) solvers of which a large range of formulations including free surface treatments exists. Physical wave flume tests of wave propagation over a submerged bar with various levels of nonlinearity, regularity, and wave-breaking, dedicated to numerical model benchmarking or validation, were carried out in the Ocean and Coastal Engineering Laboratory of Aalborg University. Three fundamentally different CFD models each widespread within their category are benchmarked against the experimental data. The CFD models are based on (i) the Volume of Fluid (VoF) based interFoam solver of OpenFOAM, (ii) the sigma-transformation solver of MIKE 3 Waves Model FM, and (iii) the weakly compressible delta-SPH solver of DualSPHysics. Accuracy of the numerical models is assessed from surface elevation time series, evaluation metrics (averaged errors on surface elevations, amplitudes, phases, and wave set-up), and spectral analyses to calculate the amplitude and phase contents of primary and higher-order components along the wave flume. Applicability is assessed from computational costs and ease-of-use factors such as the effort to configure the numerical models and achieve convergence. In general, the numerical models have high correlation to the physical tests and are as such suitable to model complex wave transformation with an accuracy sufficient for most coastal engineering applications. The VoF model performs more accurately under the turbulent conditions of breaking waves, increasing its relative accuracy in the prediction of downwave surface elevation. The sigma transformation model has simulation times one to two orders of magnitude lower than those of the VoF and SPH models.