A multivariate regression parametric study on DEM input parameters of free-flowing and cohesive powders with experimental data-based validation

One of the key challenges in the implementation of discrete element method (DEM) to model powder's flow is the appropriate selection of material parameters, where empirical approaches are mostly applied. The aim of this study is to develop an alternative systematic numerical approach that can efficiently and accurately predict the influence of different DEM parameters on various sought macroscopic responses, where, accordingly, model validation based on experimental data is applied. Therefore, design of experiment and multivariate regression analysis, using an optimized quadratic D-optimal design model and new analysis tools, i.e., adjusted response and Pareto graphs, are applied. A special focus is laid on the impact of six DEM microscopic input parameters (i.e., coefficients of static and rolling friction, coefficient of restitution, particle size, Young's modulus and cohesion energy density) on five macroscopic output responses (i.e., angle of repose, porosity, mass flow rate, translational kinetic energy and computation time) using angle of repose tests applied to free-flowing and cohesive powders. The underlying analyses and tests show, for instance, the substantial impact of the rolling friction coefficient and the minor role of the static friction coefficient or the particle size on the angle of repose in cohesive powders. In addition, in both powders, the porosity parameter is highly influenced by the static and rolling friction coefficients.

Automated summary

This study aims to develop a numerical method to predict the influence of DEM parameters on various macroscopic responses in powder flow simulations. By using design of experiment and multivariate regression analysis, the impact of different material parameters on macroscopic output responses is revealed, highlighting the important role they play in powder flow.