Drag Measurements from Particle-Resolved CFD Simulations of Flow Past Random Assemblies of Stationary Ellipsoidal Particles

Currently-available literature on fluid-particle drag models for non-spherical shapes are focused on reporting the drag coefficients of individual particles, with very few studies on the drag experienced by an assembly of non-spherical particles. For multiphase-flow applications (e.g., fluidized beds), these single-particle drag coefficients are useful at the very-dilute limit but are generally inaccurate at larger solid fractions. New drag models must be developed at larger solid fractions—as has been done for spherical particles, but not for non-spherical shapes.

In this work, we have performed single-phase computational fluid dynamics (CFD) simulations of fluid flow past random assemblies of stationary ellipsoidal particles. This approach follows the method used in previous studies [Koch and Ladd, J. Fluid Mech. (1997), 349, 31-66; Tenneti et al., Int. J. Multiph. Flow (2011), 77, 1072-1092, to name a few works] to derive drag laws for spherical particles, which is now adapted to ellipsoidal shapes. Our simulations are performed using Gerris, an open-sourced CFD code, which allows automated mesh-refinement of the particle-boundary layer. Using our simulation setup, we probe the influence of varying particle aspect ratio, orientation, solid fraction, and Reynolds number on the fluid-particle drag. We demonstrate that employing a spherical drag model for non-spherical particles misrepresents the actual drag. The particle orientation also plays a key role—the drag is lowest for ellipsoids aligned perfectly with the flow direction and, as expected, and increases non-linearly as the angle between the particles’ axes and the flow increases.

Current simulations of non-spherical particles, typically utilizing CFD-DEM models, apply ad-hoc corrections to spherical drag laws to account for non-sphericity. For example, Zhou et al. [Chem. Eng. Sci. (2011), 66, 6128-6145] calculate fluid-particle drag by substituting the drag coefficient C_d in a spherical drag law with a C_d correlation for non-spherical shapes. We show that such corrections are generally not accurate and, therefore, emphasize the need to develop new drag models specific to beds of non-spherical particles.