A Modified SPH Model for Simulating Water Surface Waves

Xiaojing Niu,Jialin Yu

DOI: https://doi.org/10.1016/j.proeng.2015.08.288

2015-01-01

Procedia Engineering

Abstract:As a flexible Lagrangian particle method, smoothed particle hydrodynamics (SPH) can easily capture large interface deformation, breaking, merging and splashing, which is quite suitable for the simulation of water surface waves. The classic weakly compressible smoothed particle hydrodynamics (WCSPH) algorithm for incompressible fluid flow usually needs small time steps to ensure numerical stability, which makes it computational time consuming. The major reason is that the pressure is linked with the density deviation which should be zero everywhere in incompressible fluid flows. Generally, the equation of state for a weakly compressible fluid is adopted to simulate incompressible fluid. To enforce the incompressibility, the stiffness of fluid is set to be large enough to keep the density deviation less than an allowable error. That will lead to a problem that a tiny error of computed density will cause a large pressure error and further numerical instability. In order to obtain rational pressure distribution and good computational efficiency as well, a new algorithm is proposed considering that flow field is globally incompressible from the view of spatial averaged flow field and weakly compressible locally due to the computational error from the view of particles. The basic conception of the proposed modified SPH is to split the pressure into a global part and local fluctuating part. The global part of pressure or the spatial averaged pressure on the resolution of background mesh is obtained by solving the pressure Poisson equation based on the velocity divergence free condition. The local fluctuating pressure is related to the local particle density variation by solving the local pressure Poisson equation, which is applied to maintain particles’ spacing in relative equilibrium. Based on this idea, formulas for the spatial averaged pressure and local fluctuating pressure are derived. The spatial averaged pressure is used as a primary estimation of the pressure gradient force and then the local fluctuating pressure is added as a correction of local density deviation. The proposed algorithm is verified by simulating the classic dam broken problem and then applied to simulate water wave breaking. A comparison of the particle configurations with previous studies at different times is shown. Good agreement between the present results and the previous results can be found. A discussion on computational efficiency is also carried out. The result shows that the proposed algorithm has better stability and can greatly reduce the computational time cost.