(448ae) Multi-Scale Approach to Developing Non-Spherical Particle Models for Realistic Granular Flow Simulations

The simulation of granular flows can be performed with either a Lagrangian or Eulerian approach. The Lagrangian approach, i.e. discrete element method (DEM) simulations, has the ability to simulate realistic particle shapes, but only a limited number of particles can be simulated due to the high computational cost. The Eulerian approach, i.e. two-fluid simulations, can simulate large-scale granular flows, but the models in this approach typically assume that the particles are spherical. The spherical-particle assumption hampers the accuracy of two-fluid simulations, due to the fact that particle shape has a significant impact on flow behavior. As a result, better Eulerian models should be developed to account for the shape effect. This work focuses on the development of a workflow for the numerical modeling of non-spherical particle flow at a large scale. In this workflow, the DEM simulations are used to determine the terms in the granular kinetic theory for the description of non-spherical particle flow. The developed non-spherical particle models are then plugged into the two-fluid model for the large-scale simulations.  

Granular kinetic theory terms that are important for two-fluid simulations include: solid stress, collisional dissipation rate, and granular conductivity. The former two terms are dominant for dense granular flows. DEM simulations with the glued-sphere particle model and true cylinder model have been utilized to develop the solid stress and collisional dissipation rate models for elongated, rod-like particles. The developed models are implemented into a two-fluid model code for the simulations of rod-like particle flows. A validation study is presented that utilizes an experiment with monodisperse beds of cylindrical particles. The experiment involves the impingement of an air jet onto a particle bed, which produces a crater.  Varying flow rates and particle sizes with aspect ratios from 2 to 8 have been used. The depth of the crater over time is measured for each case. The air jet-induced particle bed cratering is simulated using the two-fluid code and the comparison of the experimental and simulated data is presented.