Discrete Element Simulations
In our group, we introduced numerical and theoretical approaches. Our researchers cooperatively develop DEM, DEM-LB (Combining Lattice Boltzmann Method and DEM) and continuous approaches to describe various particulate systems across several length scales accurately. On the other hand, general-purpose computation on graphics hardware (GPGPU) has recently become a promising alternative for parallel computing on clusters or supercomputers. Our group has gained experience in developing hybrid GPU-CPU codes in NVIDIA-CUDA architecture, examining particles of different shapes. In some examples, we have obtained significant speed-ups by factors of up to hundreds compared to serial CPU codes.
In the past, we explored some applied scenarios: granular systems, crowd dynamics, and complex fluids. We have conducted parametric studies, examining the influence of particle shape, contact models, and other specific details on the particle scale on the predictions for large-scale applications. Our numerical efforts have helped in identifying the most relevant parameters facilitating calibrations, using experimental data and analytical predictions. Moreover, we have used coarse-graining approaches, linking the behavior of those systems at the smallest scale (contacts between elementary constituents) with its macroscopic response. The synergic contribution (experimental-theoretical) of the collaborative approaches for many diverse applications has enhanced the generality and establish the applicability of our methods and findings. Our ultimate scientific aim is to achieve robust continuum mechanical descriptions of particulate flows based on micromechanics details, which sheds light on how the local micromechanical details affect the constitutive response and jamming of complex granular flows.
More information in: