| Discrete element method (DEM) has been recognized as one of the most effective tools for the study of granular materials. However, the huge computational time and effort required to simulate large number of particles has been a major hindrance in extending DEM applications in real engineering practices.; The main objective of the study is to develop algorithms to improve computational efficiency on discrete element analysis as well as providing realistic simulation results for large scale granular material simulations. Methods and algorithms are presented from several aspects of DEM including neighbor search, contact detection, contact points and forces determination, and contact damping, etc. Some featured developments include the two level search (TLS) algorithm for neighbor search, fast common plane (FCP) method and shortest link (SL) method for contact detection, contact zone method for determination of contact points, a nonlinear elastic model for the calculation of normal contact force, and a dual damping criterion for shear contact damping. All these proposed methods devote the efforts to enhance available algorithms and promote engineering DEM applications. Also these methods are general enough to be applied to any discrete element applications, but not limited to granular flows or tool-soil interactions. They are implemented in the discrete element code BLOKS3D, which is newly developed for the simulation of granular material behaviors, such as granular flows and tool-soil interactions for earth moving equipment, with 3D polyhedral particles of any size.; The performance of these methods has been validated and demonstrated by several simulation series including the simulation of push test, energy-damping relationship and energy conservation test, earth pressure distribution test, angle of repose test, direct shear test and bucket-soil interactions. The simulation results provide confidence in the reliability of DEM simulations with polyhedral particles.; BLOKS3D is parallelized with OpenMP on NCSA supercomputers to seek higher computational speed, and a speed up ratio of 3∼5 has been achieved. The largest size simulations involve more than 97,000 particles. |
