Nicolin Govender
Department of Mechanical Engineering, University of Johannesburg, South Africa;
Research Center Pharmaceutical Engineering GmbH, Graz 8010, Austria
Particle shape plays a crucial role in the resulting macroscopic behavior of granular material yet the majority of DEM simulations use simple spheres with non-physical constructs such as rolling friction without fully understanding the implications thereof. While there have been shape studies using super quadric’s and ellipsoids over the years, they still have the same single point contact resolution as spheres. The majority of industrially relevant granular materials are however non-curve linear and have shapes that is best captured by polyhedral shapes. Apart from mechanical behavior, thermal conduction as well as cohesion between these particles occurs over a finite-contact area that is more faithfully captured by polyhedra. In this talk the role of particle shape will be explored for a number of industrial cases such as mixers, screw feeders and silos along with prescriptions for when shape and size simplifications can be used. The underlying theme of the talk will be on the use of GPU computing in the past decade to advance the physics fidelity of DEM simulations along with reduced run-times. Finally an outlook to the future of using high fidelity DEM to generate data for statistical models and physics informed neural networks that can provide answers in the age of digital twins will be presented.
