Abstract. We present experimental and numerical results of the effect that a partial discharge has on the morphological and micro-mechanical properties of non-spherical, convex particles in a silo. The comparison of the particle orientation after filling the silo and its subsequent partial discharge reveals important shearinduced orientation, which affects stress propagation. For elongated particles, the flow induces an increase in the packing disorder which leads to a reduction of the vertical stress propagation developed during the deposit generated prior to the partial discharge. For square particles, the flow favors particle alignment with the lateral walls promoting a behavior opposite to the one of the elongated particles: vertical force transmission, parallel to gravity, is induced. Hence, for elongated particles the flow developed during the partial discharge of the silo leads to force saturation with depth whereas for squares the flow induces hindering of the force saturation observed during the silo filling.

The cooling dynamics of a 2D granular gas of elongated particles is analyzed. We perform simulations on the temporal evolution of soft particles, using a molecular dynamics algorithm. For weakly dissipative particles, we found a homogeneous cooling process where the overall translational kinetic energy decreases analogously to viscoelastic circular particles. In contrast, for strongly dissipative particles we observed an inhomogeneous cooling process where the diminishing of translational kinetic energy notably slows down. The rotational kinetic energy, however, always decays in agreement with Haff's prediction for the homogeneous cooling state of inelastic particles. We mainly found that the cooling kinetics of the system is controlled by the mechanisms that determine the local energy dissipation (collisions). However, we detected a strong influence of particle shape and inelasticity on the structure of the clusters which develop in the inhomogeneous cooling regimes. Our numerical outcomes suggest that strong dissipation and particle anisotropy induce the formation of ordered cluster structures that retards the relaxation to the final asymptotic regime.

We report numerical results of effective attractive forces on the packing properties of two-dimensional elongated grains. In deposits of non-cohesive rods in 2D, the topology of the packing is mainly dominated by the formation of ordered structures of aligned rods. Elongated particles tend to align horizontally and the stress is mainly transmitted from top to bottom, revealing an asymmetric distribution of local stress. However, for deposits of cohesive particles, the preferred horizontal orientation disappears. Very elongated particles with strong attractive forces form extremely loose structures, characterized by an orientation distribution, which tends to a uniform behavior when increasing the Bond number. As a result of these changes, the pressure distribution in the deposits changes qualitatively. The isotropic part of the local stress is notably enhanced with respect to the deviatoric part, which is related to the gravity direction. Consequently, the lateral stress transmission is dominated by the enhanced disorder and leads to a faster pressure saturation with depth.

Abstract. General-purpose computation on Graphics Processing Units (GPU) on personal computers has recently become an attractive alternative to parallel computing on clusters and supercomputers. We present the GPU-implementation of an accurate molecular dynamics algorithm for a system of spheres. The new hybrid CPU-GPU implementation takes into account all the degrees of freedom, including the quaternion representation of 3D rotations. For additional versatility, the contact interaction between particles is defined using a force law of enhanced generality, which accounts for the elastic and dissipative interactions, and the hard-sphere interaction parameters are translated to the soft-sphere parameter set. We prove that the algorithm complies with the statistical mechanical laws by examining the homogeneous cooling of a granular gas with rotation. The results are in excellent agreement with well established mean-field theories for low-density hard sphere systems. This GPU technique dramatically reduces user waiting time, compared with a traditional CPU implementation.

No abstract