Modelling Granular Flows in Large-Scale Industrial Processes
Despite constantly growing computing power, discrete element method (DEM) simulations of granular systems can hardly reach the time and length scales required to describe industrial applications. The coarse-grain (CG) model of the DEM alleviates the hardware requirements representing a number of particles by a coarser (pseudo) particle, thus effectively reducing the amount of particles involved in the calculations. However, due to the violation of geometric similarity introduced by the upscaling of particles, this approach fails for effects that intrinsically depend on particle size.
For processes operating at multiple length-scales, we thus have developed a technique which concurrently couples multiple different CG levels to adjust the resolution of the simulation as needed. Starting out with a coarse simulation of the overall system, finer resolved sub-regions are embedded successively where required. This design allows capturing the details of the particle system in spatially confined regions of interest while retaining the performance benefits of the CG method where a lower resolution is sufficient. The coupling of differently resolved DEM domains is established by exchanging volumetric properties of the granular flow to impose proper boundary conditions in each sub-region. Furthermore, the detailed information of the finer resolved parts of the simulation can be used to improve the overall behaviour of the coarser simulation by an analogous mechanism. The speedup of this multi-level coarse-grain (MLCG) model compared to a fully resolved simulation is nearly proportional to the saved number of particles.