An acceleration-based stable approach transferring information from fluid to solid in the SPH-based FSI solver

Abstract

In this work, the information transfer in the SPH-based FSI solver is discussed and analyzed. The traditional arithmetic mean and distance-weighted way in the pressure integration method easily overestimate the influence of distant particles. Hence, two improved schemes, named the distance-squared weighted and normal distance-squared weighted methods, are proposed to solve this problem. The traditional and improved schemes are compared by interpolating pressure on the solid boundary from different particle distribution configurations, in which the normal distance-squared weighted method achieves the minimum relative error in reproducing the pressure value at the given position. Moreover, a novel stabilized method is proposed to overcome the fluctuated pressure-induced instability by using the acceleration and Newton`s law to transfer information. The classical dam breaking and hyper elastic gate flows are considered as the benchmark tests for their strong nonlinearities and extreme variations in free surface and pressure. The results show that the proposed acceleration-based scheme can always stabilize the simulation, giving smooth and reliable predictions in complex FSI situations.