Modelling density segregation in flowing bidisperse granular materials

Abstract

Preventing segregation in flowing granular mixtures is an ongoing challenge for industrial processes that involve the handling of bulk solids. A recent continuum-based modelling approach accurately predicts spatial concentration fields in a variety of flow geometries for mixtures varying in particle size. This approach captures the interplay between advection, diffusion and segregation using kinematic information obtained from experiments and/or discrete element method (DEM) simulations combined with an empirically determined relation for the segregation velocity. Here, we extend the model to include density-driven segregation, thereby validating the approach for the two important cases of practical interest. DEM simulations of density bidisperse flows of mono-sized particles in a quasi-two-dimensional-bounded heap were performed to determine the dependence of the density-driven segregation velocity on local shear rate and particle concentration. The model yields theoretical predictions of segregation patterns that quantitatively match the DEM simulations over a range of density ratios and flow rates. Matching experiments reproduce the segregation patterns and quantitative segregation profiles obtained in both the simulations and the model, thereby demonstrating that the modelling approach captures the essential physics of density-driven segregation in granular heap flow.

Granular flow in a wedge‐shaped heap: Velocity field, kinematic scalings, and segregation

“…Beginning at the free surface ( z / δ = 0), the profile decreases steeply from a maximum velocity, then gradually transitions to quasi‐static creeping flow below z / δ = −0.7. Like the quasi‐2D bounded heap, an exponential fit provides a reasonable approximation to the velocity profile, shown by the solid curve in Figure 6b. Similar results occur for the other conditions tested as listed in Table .…”

“…An asymmetry in the profiles with respect to the centerline is evident in Figure 5a, with slightly greater velocities to the left of the centerline, and is consistent with that mentioned with respect to z/δ = −0.7. Like the quasi-2D bounded heap, 12,20,26,27 an exponential fit provides a reasonable approximation to the velocity profile, shown by the solid curve in Figure 6b. Similar results occur for the other conditions tested as listed in Table 1.…”

Asymmetric concentration dependence of segregation fluxes in granular flows

“…We have successfully applied the model using a segregation velocity linear with concentration (Eq. 5) for a variety of situations including bi-and polydisperse size segregation in bounded heap flow [7,8,39], bidisperse segregation in tumbler flow [32], tri-disperse size segregation in chute flow [41], bidisperse density segregation in bounded heap flow [16], and even segregation of rod-like particles in bounded heap flow [40]. Here we compare the predictions of the model using a segregation velocity linearly dependent on c s (Eq.…”

A fast adhesive discrete element method for random packings of fine particles

“…In multiphase and granular flows, discrete element method (DEM) has been widely used to model particle-particle interaction and accurately predict the motion of individual particles (Cundall & Strack, 1979;Tsuji et al, 1993;Zhu et al, 2008;Marshall & Li, 2014;Sundaresan et al, 2018;Xiao et al, 2016). For soft-sphere DEM, Young's modulus of particles used in the simulation is usually much smaller than its real value.…”