The study of violent wave-structure interaction (WSI) is crucial for the safety and operation of offshore and marine structures. As the mesh-free approach, the incompressible Smoothed Particle Hydrodynamics (ISPH) method is emerging as a potential tool for simulating the WSI problems [1]. The pressure in the conventional ISPH method is obtained by solving the pressure Poisson’s equation (PPE), which is the most time-consuming part. Recently, the machine learning (ML) techniques have been used in the fluid dynamics. In this study, the graph neural network (GNN) is combined with ISPH (ISPH_GNN) [2] and used to predict the fluid pressure instead of solving the PPE directly. The hybrid ISPH_GNN method with one trained GNN model based on training data generating from relatively simple sloshing and dam-breaking cases without any structure will be extended to simulate different violent WSI problems. It will be shown that the ISPH_GNN method gives satisfactory results when compared with experimental data, indicating its good generalization properties. In addition, this method will be demonstrated to requires much less computation time than the conventional ISPH for estimating pressure large-scale violent WSI simulations involving a large number of particles.
 

ISPH,violent wave-structure interaction,Graph neural network (GNN),hybrid method,PPE,Pressure prediction