Suitable rolling resistance model for quasi-static shear tests of non-spherical particles via discrete element method

Abstract

Rolling resistance is a main consideration in quasi-static shear test simulations of particles via the discrete element method. However, not all rolling resistance models can satisfy the required objectivity and rate independence. A suitable model for spherical particles has been selected from five models in our previous works. In the current study, this model is combined with four normal and tangential contact models to confirm its applicability. After confirmation, the model is generalized to simulate direct shear tests on non-spherical particles. The stress–strain and dilatancy curves are rate-independent, and the relative rolling velocity between particles is objective. Furthermore, objectivity and rate independence for arbitrarily shaped particles are unchanged when normal stresses, volume fractions, or normal and tangential contact models are changed. Simulation results are also consistent with other experiment findings. For comparison, results are calculated for two other rolling resistance models; the shear curve at a single speed is consistent with the experiment, which has three stages: elastoplastic increasing, yielding, and keeping. However, the stress–strain curves at different shear rates do not coincide, which means that the models conflict with the rate independence of quasi-static granular systems. The virtues and defects of the five rolling resistance models are discussed from the perspectives of objectivity and rate independence. These two properties provide criteria for determining the appropriateness of a model, which has been rarely discussed in former studies.