One of the major problems in DEM simulations is the extremely high computational cost when tracking a large number of particles. Coarse grain models are an increasingly popular method for reducing computational cost where the particle size used in the simulation is artificially increased. We have proposed a novel coarse grain model named Scaled-Up Particle (SUP) model, which is based on the continuum assumption to derive scaling laws of forces and torques and is applicable to a variety of systems. The example above is original and coarse grained simulations of gas-solid fluidised beds where the particles are wet and bonded with liquid bridge forces.

In this example, the SUP model is applied to a gas-liquid-solid three-phase flow where the capillarity is also important. The particles used are hydrophilic and the liquid initially penetrates the bed due to capillary action.

In DEM, the particle stiffness is often reduced artificially to permit a large time step interval and decrease computational cost. However, it is commonly known that particles may become more cohesive when stiffness is reduced if attraction forces are present between the particles. In our research group, we have developed generic scaling laws for attraction forces named Reduced Particle Stiffness (RPS) scaling to accurately represent the original particle behaviour using particles with reduced stiffness. The examples above show simulations with the original stiffness, reduced stiffness without attraction force scaling and reduced stiffness with the RPS scaling.

In DEM, parallelisation by domain decomposition using the Message Passing Interface (MPI) is often employed to accelerate simulations. For MPI computing, dynamic load balancing is an commonly used method to ensure the parallelisation efficiency.