CFD simulations of flow in random packed beds of spheres and cylinders: analysis of the velocity field

Abstract

In this work, we aim to better understand the flow patterns in a random arrangement of particles that might affect local mass transfer and effective reactor performance. Using the DEM code Grains3D, spherical or cylindrical particles are randomly inserted inside a horizontally biperiodic container and fall under gravity. Hydrodynamic simulations are performed with PeliGRIFF, a Fictitious Domain/Finite Volume numerical model. Simulations parameters are the bed height and particulate Reynolds number. Effect of random packing on the flow field is analysed in terms of the probability distribution function (PDF) of the normalized vertical velocity. A higher Reynolds number makes more backward flow zones and changes the PDF curves that we interpret as thinner boundary layers. Unexpectedly, internal variability is independent of bed height. We propose that the probability of occurrence of random structures increases with bed volume in opposition with volume averaging effects. Internal and external variability are similar for beds of spheres and cylinders of aspect ratio (< 2). However, for longer cylinders (higher aspect ratio), subdomains with same thickness are statistically different from one bed to another. We propose that the subdomain thickness required to average out sources of variability increases with high particle aspect ratio.