The Analysis of Parallelized SPH Particle Flow

Abstract

Nowadays, numerical simulation is widely used to solve complex physical problems involving fluids or solids in engineering. Meshfree methods, such as Smoothed Particle Hydrodynamics (SPH), are receiving a lot of attention to simulate fluid flow due to their advantages over grid-based methods in problems dealing with some complicated phenomena. To simulate fluid flow in the SPH method, a great number of particles are necessary because the entire domain is expressed by a finite number of particles. Therefore, parallelization of SPH using Message Passing Interface (MPI) is applied to the problems involving large computational domains for improving the numerical efficiency. For the parallel computing in the SPH method, the simulation domain is decomposed into subdomains with a group of particles using the new proposed parallel method for the continuity of subdomain boundary under the Single Instruction Multiple Data (SIMD) processor. The distribution is also implemented based on the procedure of the SPH computations. By the parallel computing using the methods just mentioned, the SPH method is to be more flexible for better performance. The performance of the parallel computing of the SPH method is evaluated by the simulation of fluid flow.