Gallery: Spot Model / DEM comparison
DEM Simulation
Spot Model
The top sequence of images shows a Discrete Element Method simulation of a granular drainage experiment in a thin rectanglar container, fifty particle dimeters across and eight particle diameters deep. Each particle is accurately modeled according to Newton’s laws and a realistic friction model is employed to capture particle interactions. The simulation matches experiments with glass beads to a high degree of accuracy, but is extremely computationally intensive, and can take up to a week on twenty processors.
The bottom sequence of images shows a simulation of the same experiment, using a method based on the Spot Model. In this simulation, particles move in response to spots introduced at the orifice, that diffuse upwards through the material according to a random walk. Five parameters determining the size of the random walk are calibrated from DEM. Individual frictional interactions between the particles are not considered, and thus the Spot Model runs much more rapidly, taking only eight hours on a single processors, representing a speedup of order 100 over DEM.
The images show that there is an extremely good match between the two simulations. While the Spot Model does not reproduce the exact track of every particle, it faithfully reproduces many of the statistical quantities of the DEM simulation; for example, we see the correct amount of mixing between the two-color interfaces. More importantly, the Spot Model is in good agreement with DEM for various microscopic statistical quantities, such as the radial distribution function, and the bond angle distribution. Thus the Spot Model generates remarkably realistic flowing random packings, and demonstrates the possibility of multiscale modeling in amorphous materials.
For more details, see “Dynamics of random packings in granular flow” (2006).