An efficient approach to generating DEM samples having coincident void ratios with actual sand

The ranges of initial void ratios that can be achieved for numerical samples with the same grading as actual sand under gravity are explored using three-dimensional discrete element method (DEM). A corrected compressible accumulation model is proposed to predict the packing density of sand. Compared with the measured results, the errors of the prediction results are small. For three-dimensional DEM samples, when the ratio of sample size to particle size L/R is equal to or larger than 30, the maximum and the minimum void ratios are minimally affected by the sample size L. The maximum void ratios and the minimum void ratios of numerical samples with spherical particles are much smaller than those of actual sands. The generation method for numerical samples with non-spherical particles is proposed based on the theory of CPM. The variations of void ratios of numerical samples with non-spherical particles are wider than actual sands. So it can meet all the need of DEM simulation on the void ratio of actual sand.