Robust event-driven particle tracking in complex geometries

Abstract

Particle tracking, that is, the repeated localization of particles within a grid by means of tracking the particles’ trajectories, is routinely applied in particle-based schemes where the domain is described by an unstructured polyhedral grid. A range of tracking algorithms are available in the literature, which are inherently similar to algorithmic approaches common both in event-driven particle dynamics (EDPD) and ray-tracing methods. We propose a reformulation of existing particle tracking algorithms in the context of EDPD. On the one hand, this resolves inconsistencies in the mapping between particle positions and grid cells triggered, e.g., by imperfect grids. More importantly, it allows the specification of solid objects via constructive solid geometry (CSG), a standard technique for the modeling of solids in computer-aided design. While usually considered contrary approaches, our description of the computational domain as the combination of a bounding volume defined by an unstructured grid and solids modeled via CSG embedded into this volume can be highly advantageous. The two different approaches of modeling the computational domain complement each other perfectly, as the CSG representation is not only efficient in terms of memory and computing time, but also avoids the challenges of generating finely resolved unstructured grids in the presence of complicated boundaries. These benefits, as well as the positive impact of several algorithmic optimizations of the extended tracking algorithm, are exemplified via a particle-based simulation of a gas flow through a highly porous medium.