An efficient algorithm for detecting particle contact in non-uniform size particulate system

Abstract

Algorithms for detecting particle collision play an important role in the discrete element method (DEM) for the simulation of granular flow systems since the time taken to detect the contact pairs usually occupies a considerable proportion of the total CPU time for the simulation. In this study, we developed a new octree algorithm called multi-octree algorithm, for detecting candidate contact pairs. The so-called multi-octree algorithm adopts the topology of octree for detecting possible contacts, in which the particles are managed by hierarchical classification according to particle size and then mapped to different level nodes of an octree, instead of mapping all particles into the leaf nodes of an octree in Vemuri-octree algorithm. The present algorithm speeds up the detecting process by reducing the number of candidate particles for possible contacts. A non-uniform size particulate system with three particle sizes was simulated by the multi-octree and the Vemuri-octree algorithms simultaneously. The simulation results showed that the multi-octree algorithm is relatively insensitive to the volumetric fraction of particles and particle shapes, and takes much shorter time to find candidate particles for possible contacts than the Vemuri-octree algorithm for non-uniform size particulate systems.