Recent decades have witnessed considerable research efforts in the development and application of discrete element method (DEM) to granular material. One key aspect which appears to have not been addressed adequately is the choice or determination of the DEM input parameters to produce satisfactory quantitative predictions. In many DEM studies, the input parameters were often not measured and the values were sometimes assumed without any proper justification. Experimental validations of the numerical results were also very rarely performed, so it is often difficult to ascertain whether the DEM computations have produced satisfactory quantitative agreement with the corresponding physical systems.

This paper describes a validation type study in which DEM was used to model confined compression experiments of spherical and non-spherical particles. The key properties of the particles were independently measured. The experimental setup consists of an instrumented Perspex cylinder which is filled with particles and vertically compressed via a rigid platen. The experimental rig was devised to provide data on the bulk compressibility of the granular assembly and the corresponding load transfer to the containing walls and the bottom platen, thus providing quantitative information for direct comparison with the DEM results. Initial particle packing arrangement is expected to influence the bulk behaviour of the assembly, so several methods of particle generation were used to investigate the sensitivity of particle generation in DEM. The influence of the DEM input parameters on the bulk mechanical response was explored and the relative importance of these parameters is discussed.