Skip to:

Publication Abstract

Large-scale DEM-LBM Modeling towards Off-road Mobility

Jelinek, B., Mason, G. L., Johnson, D., Carrillo, A., Goodman, C. C., Priddy, J., & Vahedifard, F. (2019). Large-scale DEM-LBM Modeling towards Off-road Mobility. HPCMP User Group Meeting. Vicksburg, MS.

Numerical models for dry granular media and saturated particle-fluid systems based on Large-scale Discrete Element Method (DEM) and Lattice-Boltzmann Method (LBM) are presented. The capabilities of MPI-parallelized LBM and DEM models to represent macroscopic behavior of dry and saturated granular media are demonstrated through a micromechanical study of shear thickening in granular media suspension. A calibration study using biaxial compression test to identify coarse-grained DEM parameters is presented. In view of the cone penetrometer test status as the primary method used by the U.S. Army for evaluating mobility off-road mobility, the initial focus for DEM simulation is a group of experiments in a cone penetration calibration chamber. To define the resolution required, a convergence study on the effect of DEM particle size on penetration resistance is first performed. Large-scale simulations of the calibration-chamber cone penetration test at the size and time scale matching the calibration chamber measurements are shown using parameters calibrated to match the penetration resistance for Bayou Pierre sand in dense and very dense states. These demonstrations include numerical evaluations of soil stress components, particle rotations and sand porosity changes throughout the probe insertion process. The DEM is then applied to model wheel-soil interaction for mobility assessments. Prescribed slip test simulations predict pull-slip curve of a rigid wheel in dense sand. The effects of the DEM parameters on the wheel response are examined. Results are compared to those in the extensive DROVE (Database Records for Off-road Vehicle Environments) database. Parallel performance of the present DEM high performance computing (HPC) code is evaluated and compared with alternative parallelization schemes. The significant reduction of the computational time afforded by HPC for these studies demonstrate the potential for simulation to reduce the need for expensive and time-demanding physical experiments.