DEM simulations of polydisperse media: efficient contact detection applied to investigate the quasi-static limit

T Shire and KJ Hanley and K Stratford, COMPUTATIONAL PARTICLE MECHANICS, 8, 653-663 (2021).

DOI: 10.1007/s40571-020-00361-2

Discrete element modeling (DEM) of polydisperse granular materials is significantly more computationally expensive than modeling of monodisperse materials as a larger number of particles are required to obtain a representative elementary volume, and standard contact detection algorithms become progressively less efficient with polydispersity. This paper presents modified contact detection and inter-processor communication schemes implemented in LAMMPS which account for particles of different sizes separately, greatly improving efficiency. This new scheme is applied to the inertial number (I), which quantifies the ratio of inertial to confining forces. This has been used to identify the quasi-static limit for shearing of granular materials, which is often taken to be I = 10(-3). However, the expression for the inertial number contains a particle diameter term and therefore it is unclear how to apply this for polydisperse media. Results of DEM shearing tests on polydisperse granular media are presented in order to determine whether I provides a unique quasi-static limit regardless of polydispersity and which particle diameter term should be used to calculate I. The results show that the commonly used value of I = 10(-3) can successfully locate the quasi-static limit for monodisperse media but not for polydisperse media, for which significant variations of macroscopic stress ratio and microscopic force and contact networks are apparent down to at least I = 10(-6). The quasi-static limit could not be conclusively determined for the polydisperse samples. Based on these results, the quasi-staticity of polydisperse samples should not be inferred from a low inertial number as currently formulated, irrespective of the particle diameter used in its calculation.

Return to Publications page