Rolling friction as a technique for modelling particle shape in DEM

Shape is a computationally expensive property to model in DEM, and so rolling friction is often employed in order to simulate its effects. However, the precise way in which rolling friction models shape has never been properly outlined. This paper presents an attempt to do so by estimating the coefficient of rolling friction for simplified shapes using a simple geometric argument based on the eccentricity of contact. This method was tested by running a series of back-to-back simulations of a simple angle of repose test with clumped particles (simulating shape) and idealised spherical particles with rolling friction. Some similarities were observed, however, quantitatively, there was a significant discrepancy. Approximately half as much rolling friction was required to reproduce the effects of an equivalent amount of shape estimated by the method. This discrepancy is thought to be due to the fact that shape can cause rotation as well as resist it, while rolling friction always acts to oppose rolling motion. This paper presents an attempt to quantify the coefficient of rolling friction for DEM simulations in terms of a typical eccentricity of contact for non-spherical particles. This approach was tested by running a series of back-to-back simulations of a simple angle of repose test using clumped non-spherical particles, compared to idealised spherical particles with rolling friction. It was found that rolling friction qualitatively models shape well, but its effects were much stronger than an equivalent amount of shape. [Display omitted] ► Equivalent rolling friction coefficients are estimated for non-spherical particles. ► Angle of repose test simulations is used to evaluate the estimated coefficients. ► Qualitatively, rolling friction models the effects of shape reasonably well. ► Rolling friction has a much stronger influence than an equivalent amount of shape.