The effect of mesh discretisation on damage and wear predictions using the Discrete Element Method

Abstract

Assessing wear, damage and time to failure of industrial granular systems using the Discrete Element Method (DEM) is an area of limited research. In this paper, wear and damage are calculated using the Archard, Finnie, Bitter and Clark wear models of abrasion, ductile, brittle and a combined erosion model respectively using a DEM simulation with an underlying triangular surface mesh to capture and aggregate wear effects. These are compared for varying surface mesh discretisations to determine the effect of mesh refinement on wear distribution and prediction. It was found that the wear pattern remains similar for different discretisations, but the prediction of the maximum wear location varies. Therefore, the ratio between the particle area to surface element area needs to be carefully considered when looking at wear in an industrial setting using DEM. The amount of wear remained approximately constant with 1 -4 % and 1 - 2 % variations for abrasion and ductile erosion respectively. A linear damage accumulation model is presented based on the Palgrem-Miner model which shows a large variation in lifetime prediction for different surface mesh discretisations. The lifetime prediction needs to be compared to a real-world industrial system for both calibration and validation due to the non-converging nature of wear and damage modelling inside of DEM. To assist in determining a point for accurate wear and damage assessment the formation of holes in the model is studied. The total cumulative hole area is tracked and is found to converge at the impact area of the particles impacting the plate as the mesh is refined. The point of convergence is found to be when particle and element size are approximately equal. This particle to element size ratio is suggested to be a good choice for this type of DEM simulation as it balances accuracy and computational time for both wear and damage assessment.