Simulation and visualization of discrete particles on clusters, grids and clouds

Abstract

This paper presents the development of parallel algorithms and visualization techniques for particle systems simulated by the discrete element method on clusters, grids and clouds. The dynamic domain decomposition strategies based on the moving planes, the multilevel k-way graph-partitioning and the recursive coordinate bisection were applied to the simulation of hopper discharge. The parallel speedup of discrete element computations measured on Docker containers of the OpenStack cloud was compared with that obtained by using the native hardware. The cell cut-, Voronoiand cell centre-based techniques were developed for extraction and visualization of crack surfaces propagating in monodispersed particulate media, which was simulated by the discrete element method. The quantitative comparison of three visualization techniques based on local space decompositions was performed. The trials from the case study showed that the developed parallel algorithms and visualization techniques were able to satisfy needs of researchers simulating large discrete particle systems.