An adaptative discretization to model boundary value problems with discrete element method

Abstract

An adaptive discretization method is introduced to develop numerical models for boundary value problems using the Discrete Element Method. This method involves discretizing a domain, with smaller particles in regions of interest—where the material undergoes large displacements and irreversible deformations—while continuously increasing particle sizes elsewhere, as distance from these regions of interest increases. While maintaining uniform mechanical properties within the whole simulation domain through appropriate scaling of contact model parameters, this approach presents the main benefit to substantially reduce the number of particles in the model, thereby lowering computational costs, without making the numerical method itself more cumbersome. This method is applied and assessed in the context of modeling a Cone Penetration Test within a calibration chamber.