PARALLELIZATION OF 3D IMMERSED BOUNDARY LATTICE BOLTZMANN METHOD FOR LARGE SCALE SOLID PARTICLE SUSPENSION FLOWS USING MPI AND OpenMP

Discrete particle simulations consider the issue of coupling liquid flow and solid particles. The coupling methods and the corresponding algorithms for realizing the couple are still developed. The applications of Lattice Boltzmann Method (LBM) have been widely practiced for the simulations of multiphase flows through computational fluid dynamics (CFD) technique. In this study, the LBM was used to simulate the suspensions of solid particles in the fluid using the 3D immersed-boundary LBM method. The in-house code of the 3D immersed-boundary LBM method was parallelized using the naturally parallelized merits of the LBM. The high performance computing (HPC) was tested and analyzed at a supercomputer. The parallel programming was based on OpenMP and MPI respectively. In OpenMP parallel programming the fork-join technique was employed for do-loop running and in MPI parallel programming the domain decomposition technique was employed for data partitioning and communicating. Through the high performance computing (HPC) studies on large scale parallel computing, it shows that OpenMP parallel programming is able to lead the immersed-boundary LBM method on the parallel speed-up for large scale solid particle suspension flows.