Advanced wave generation systems for numerical modelling of coastal structures

Accurate generation of wave climates in the context of numerical models (and in particular CFD models) is a challenging problem, as these are increasingly used to provide design support to coastal engineering projects. In this paper we will briefly present a technique that addresses the generation (and active absorption) of non-repeating wave sequences for modelling storm events in a meaningful manner. This technique includes a spectral window preprocessing method that is used to reduce the computational costs associated with wave generation algorithms. These can be particularly cumbersome for generating storm events. It was demonstrated that numerical cost can be reduced by about 40 times by using O(101) frequencies for wave reconstruction, rather than O(104) which current methods would need to accurately reproduce long wave series, without any noticeable difference in terms of the generated wave signal. The technique is already in use within the context of the computational toolkit Proteus (https://github.com/erdc/proteus) and is it is combined with both the CFD and shallow water module of the model. The methodology is also fit with a 2nd order correction for generating nonlinear random wave series. Case studies are also presented that prove i) the capability of the technique to reproduce meaningful sea states in the context of numerical modelling of coastal structures and ii) the improvement of computational cost, when compared to currently available techniques. These case studies comprise modelling of random waves in a numerical wave tank to acquire wave statistics by using both CFD and shallow water models, as well as modelling coastal structures such as a low-crested levees and a caisson breakwater using random sea states