A novel particle decomposition scheme to improve parallel performance of fully resolved particulate flow simulations

TL;DR

This paper introduces a new particle decomposition scheme that enhances the parallel scalability of surface-resolved particulate flow simulations, enabling more efficient modeling of large, complex particle systems.

Contribution

The study presents a novel particle decomposition scheme that reduces communication overhead and improves scalability in parallel particulate flow simulations.

Findings

Significant performance improvement in large particle systems

Reduced communication constraints between processors

Enhanced scalability demonstrated through hindered settling experiments

Abstract

This study addresses the challenge of simulating realistic particle systems by proposing a novel particle decomposition scheme that improves the parallel performance of surface resolved particle simulations. Realistic particle systems often involve large numbers of particles and complex particle shapes. The resulting need to account for shape factors requires the inclusion of even more particles to obtain statistically meaningful results. However, the computational cost increases with the number of particles, making efficient parallelization crucial. Therefore, the proposed scheme aims to improve the scalability by optimizing the communication and data management between processors. Through hindered settling experiments, the applicability and performance of the novel particle decomposition scheme are thoroughly investigated using the homogenized lattice Boltzmann method. The results show that the proposed method significantly improves the performance, especially in scenarios with a large number of particles, by reducing communication constraints and improving scalability. This research contributes to the advancement of computational methods for efficient study of complex particle systems and provides valuable insights for future developments in this field.