Numerical modelling of wave agitation in ports and access channels

Abstract

The determination of wave penetration in ports is of great importance for the design of coastal structures and for the efficiency of port operations. Nowadays, coastal engineers can make use of numerical models for the assessment of wave climates in ports. Various numerical models exist, including the phase resolving models SWASH (TU Delft) and MIKE 21bw (DHI). Van Oord Dredging and Marine Contractors are currently using the numerical model MIKE 21bw. To get a bit more familiar with SWASH and to see how the two numerical models relate to one another, they initiated this master project together with TU Delft. Therefore, the main aim of this thesis is to compare the numerical models SWASH and MIKE 21bw in terms of their accuracy, computational requirements and numerical stability for waves encountering approach channels and waves propagating into ports. A theoretical analysis between the two numerical models is presented first which shows that they are fundamentally different, i.e. based on different equations and assumptions. The main difference is that SWASH uses the non-hydrostatic non-linear shallow water equations, including the vertical coordinate, while MIKE 21bw uses the (enhanced) Boussinesq equations which exclude the vertical coordinate. Another difference is that SWASH may use multiple vertical layers, improving its dispersion characteristics. Therefore, the application range of SWASH is larger in terms of the maximum water depth to deep water wave length. Furthermore, SWASH can include more physical phenomena and does not require a minimum water depth as MIKE 21bw does. Beside the theoretical analysis, a comparison study was performed making use of a physical experiment of an approach channel (Case 1) and of a harbour basin (Case 2). From the comparison of the measurements with the numerical simulation of the approach channel (Case 1), it was found that both numerical models are capable of resolving wave-channel interaction for long waves (f < 0.13 Hz) and the accuracy in both models is comparable, although the significant wave height based on the total wave spectra is under-estimated in both numerical models. Numerical damping cannot (fully) explain the under-prediction of the significant wave height. Further, it is believed that inaccuracies/uncertainties in the physical experiment regarding the measurements and the setup were quite severe. This may explain the observed differences between the numerical models and physical model results. A comparison of the numerical simulations of the harbour (Case 2) with the physical experiment shows that both numerical models can accurately model wave penetration of primary waves into ports when simulating on prototype scale. However, the simulations performed on physical model scale (same scale as the physical experiments) resulted in critical numerical instabilities in both numerical models, especially in MIKE 21bw. The accuracy in predicting the significant wave height and energy period is comparable for both numerical models. Furthermore, SWASH turned out to be numerically more stable while the computational requirements are less in MIKE 21bw when two vertical layers are used in SWASH. Finally, a general guideline is presented for the selection of SWASH or MIKE 21bw for wave modelling purposes.