(299a) Calibration of DEM Material Models

As industry increases its use of DEM for simulation of particulate handling and processing operations, attention is increasingly focused on model calibration and methods for dealing with scaling and multi-scale modeling issues. Here we examine the basis for choosing a DEM model relative to the bulk particulate material and the process being simulated, and outline strategies for calibrating DEM models against measured particle and bulk properties and behavior.

A DEM model is a combination of the particle inertial properties (size, shape, density) and mechanical properties (stiffness, elasticity, plasticity) and interaction terms between particles and other objects ? particles or boundary surfaces. The latter depends on the choice of contact models which may include parameters such as friction, restitution coefficient and cohesion for example, and the body force model, usually gravity, but also could include fluid drag, electrostatic force, and magnetic force, and so on. Together these properties and parameter values form what we describe as the DEM material model.

The choice of DEM material model depends on the focus and objectives of the simulation. In bulk transport problems where the objective is to predict local mass flow rates and load transfer to boundary surfaces, the choice and calibration of the material model is dependent only on bulk flow characteristics. In applications where the individual particle dynamics and interactions, for example, particle orientation, number and area of contacting surfaces, load distribution through the bulk are important, then the particle inertial properties and contact physics will need to be closer to the actual system. The third category of DEM simulation is one where the objective is to relate correlate (particle-scale) interactions with process-scale conditions, usually by simulating a sub-region of the system. In that case the focus is usually on the interactions between particles, with equipment surfaces or with surrounding media, which influences the choice of material model to ensure that governing particle properties and interaction terms, such as specific mass, surface area and contact area and deformation are more closely matched. The approach to material model calibration is likely to differ for each of these categories of DEM simulation.

This paper will discuss some of the methods currently employed for calibrating DEM material models and look at developments in computational, experimental and optimization techniques to advance this important aspect of DEM simulation.