Investigation of Numerical Modelling Techniques for Predicting Highly Nonlinear Extreme Waves in Shallow and Deep Water

This paper aims to investigate the accuracy and computational efficiency of three CFD-based numerical codes to accurately simulate extremely large regular and irregular waves of different steepness in deep and shallow water conditions. The work assesses the performance of three numerical techniques with different formulations of the fluid dynamic equations. Firstly, an open-sourced smoothed particle hydrodynamics (SPH) code; secondly, a finite difference method (FDM) based 3D numerical model with the assumption of inviscid and incompressible fluid flow; and thirdly, a commercial CFD code that uses a finite volume method (FVM) to solve the Reynolds-averaged Navier-Stokes (RANS) equations. A suite of metrics and methodologies, considering three key performance parameters: accuracy, computational requirements and available features for providing a consistent framework for the quantitative assessment of different techniques, has been presented. Numerically simulated free surface elevations, wave periods, and spectrum (for irregular waves only) are compared with experimental data previously acquired at an Offshore Engineering Basin (OEB) facility. Extensive convergence studies were carried out for each numerical tool for a selected large wave before predictions were model for all waves. All three models reproduced waves with an accuracy comparable to physical wave makers in the wave basin experiments for the deep-water regular and irregular waves; however, the SPH model performed better than the other two models for the shallow water waves. The challenge remains for wave basins to reduce unwanted basin effects and numerical facilities to accurately model waves with proper account for boundary effects and numerical diffusions. In addition, only flat-bottom domains were considered in the investigation, leaving the wave modelling for uneven bottom for future studies.