Calibrating friction coefficients in discrete element method simulations with shear-cell experiments

Abstract Discrete Element Method (DEM) simulations coupled with shear cell experimental results have been used to investigate the flow behaviour of a dry particle assembly of glass beads in the quasi-static regime. Experimental studies have been undertaken using an FT4 powder shear cell apparatus, in parallel with extensive DEM simulations of both homogeneous simple shear and the FT4 shear cell itself. The findings show that it is not possible to accurately predict the bulk friction coefficient with homogeneous simple shear simulations unless both rolling and sliding friction are considered. There are, however, multiple pairs of sliding and rolling friction coefficients which can reproduce the experimental bulk friction coefficient. Sliding test experiments were conducted to yield the coefficient of sliding friction, and hence minimise the set of potentially correct pairs. Simulations of the full FT4 shear cell with two different calibration pairs, along with a pair without rolling friction, were then undertaken to understand the effect of their selection on realistic wall-bounded shearing conditions. Discrepancies were mainly found in the obtained radial contact number and velocity profiles, with increasing friction coefficients - particularly sliding friction - found to inhibit packing and particle velocity in the shear deformation zone. Comparison between homogeneous simple shear and shear cell simulation results showed a significant effect of the wall on the obtained force network, with almost a complete absence of the weakest structures which were seen supporting the strong structures in the simple shear scenario.