Abstract
Considering that granular material is second only to water in how often it is handled in practical applications, characterizing its dynamics represents a ubiquitous problem. However, studying the motion of granular material poses stiff computational challenges. The underlying question in this contribution is whether a continuum representation of the granular material, established in the framework of the smoothed particle hydrodynamics (SPH) method, can provide a good proxy for the fully resolved granular dynamics solution. To this end, two approaches described herein were implemented to run on graphics processing unit (GPU) cards to solve the three-dimensional (3D) dynamics of the granular material via two solution methods: a discrete one, and a continuum one. The study concentrates on the case when the granular material is packed but shows fluid-like behavior under large strains. On the one hand, we solve the Newton–Euler equations of motion to fully resolve the motion of the granular system. On the other hand, we solve the Navier–Stokes equations to describe the evolution of the granular material when treated as a homogenized continuum. To demonstrate the similarities and differences between the multibody and fluid dynamics, we consider three representative problems: (i) a compressibility test (highlighting a static case); (ii) the classical dam break problem (highlighting high transients); and (iii) the dam break simulation with an obstacle (highlighting impact). These experiments provide insights into conditions under which one can expect similar macroscale behavior from multibody and fluid dynamics systems governed by manifestly different equations of motion and solved by vastly different numerical solution methods. The models and simulation platform used are publicly available and part of an open source code called Chrono. Timing results are reported to gauge the efficiency gains associated with treating the granular material as a continuum.
