Construction of poly-ellipsoidal grain shapes from SMT imaging on sand, and the development of a new DEM contact detection algorithm

Purpose - This paper aims to construct smooth poly-ellipsoid shapes from synchrotron microcomputed tomography (SMT) images on sand and to develop a new discrete element method (DEM) contact detection algorithm. Design/methodology/approach - Voxelated images generated by SMT on Colorado Mason sand are processed to construct smooth poly-ellipsoidal particle approximations. For DEM contact detection, cuboidal shape approximations to the poly-ellipsoids are used to speed up contact detection. Findings - The poly-ellipsoid particle shape approximation to Colorado Mason sand grains is better than a simpler ellipsoidal approximation. The new DEM contact algorithm leads to significant speedup and accuracy is maintained. Research limitations/implications - The paper limits particle shape approximation to smooth poly-ellipsoids. Practical implications - Poly-ellipsoids provide asymmetry of particle shapes as compared to ellipsoids, thus allowing closer representation of real sand grain shapes that may be angular and unsymmetric. When incorporated in a DEM for computation, the poly-ellipsoids allow better representation of particle rolling, sliding and interlocking phenomena. Originality/value - Method to construct poly-ellipsoid particle shapes from SMT data on real sands and computationally efficient DEM contact detection algorithm for poly-ellipsoids.