DEM simulation of shear box test: discussion about the influence of initial packing.

Resumo

The Discrete Element Method (DEM) assumes a discontinuous medium, made of particles within a granular system. Particles are represented assuming a geometric model, such as polyhedra, spheres, or another class of entities. Particles can move and interact with each other. Since the pioneer works of Cundall and Strack, the DEM has become a popular tool in scientific and industrial communities that study granular materials. Currently, despite its expensive computational cost, the utilization of DEM for the comprehension of real-world problems has gained attention. However, reproducing real problems in DEM is not a straightforward process, mainly due to several quantities of parameters such as the particle shape, materials properties, dimensions of boundaries, and number of particles. All these properties need to be defined or chosen, and they can have influence in results. Collaborating to recent discussions about good practices to model real-world problems in DEM, this paper presents the scenario of a large shear box test. A comparative analysis is done, utilizing three different particle shapes: a sphere, a convex polyhedron, and a non-convex polyhedron, addressing the effects that alterations in the initial packing of these particles have in numerical simulations results. Based on the outcomes, we can see that for the three particle shapes, the utilization of different initial packing did not present a significant impact on the variability of maximum stress and initial stiffness of the simulations. However, the modification of initial packing influenced the volumetric variation (ΔV) of samples with convex and non-convex particle shapes.