Numerical investigation of segregation and mixing in bidisperse systems using the coarse-grained CFD-DEM approach

Abstract

Bi- and polydisperse granular materials are widely used in various industries and are an essential subject of current research. The modeling of such systems using the Discrete Element Method (DEM) is computationally very demanding. Therefore, it is limited to lab-scale systems. A common approach to solve this issue is to summarize a specific number of original particles into large grains using the so-called coarse-grain approach (CG). This study examines the accuracy of the CG approach in bidisperse systems. First, a mechanically agitated system is studied under wall and periodic boundary conditions, ranging from minimal segregation to pronounced segregation with a visible Brazil nut effect. The ascending velocity of large grains increases by a factor of 2.1, marking this transition. Second, a fluidized bed with different particle diameter ratios and fluidization velocities is simulated, showing mixing or segregation depending on the settings. The mixing index ranges from 27% to 98%. The simulations are repeated for varying levels of coarsening, keeping either the CG factor or the grain diameter constant for the respective particle types, to assess the limitations and effectiveness of scaling strategies for bidisperse granular systems. Scaling with a constant grain diameter more accurately represents fluidized bed systems with a low particle-diameter ratio ($d_S/d_L=0.2$), while the impact of the coarsening strategy diminishes as particle sizes become more similar ($d_S/d_L>0.2$). In mechanically agitated systems, scaling with a constant CG factor amplifies the Brazil nut effect, whereas a fixed grain diameter leads to a weaker prediction.