CFD-DEM simulation of clustering behavior in a riser: effect of the collision model

Risers play a crucial role as reactor units where important catalytic reactions occur. However, one of the drawbacks is the formation of particle clusters, which hinder the effective gas–solid interaction. This problem is primarily driven by energy dissipation due to non-ideal particle collisions. This study aims to compare the effectiveness of two common collision modeling approaches in Computational Fluid Dynamics - Discrete Element Model (CFD-DEM), namely hard and soft sphere model, in predicting particle clustering in a riser system in the fast fluidization regime at several superficial velocities. A pseudo-2D lab scale riser is used with dimensions 1.59 m × 0.07 m × 0.006 m. A comparison of these models is made based on the following three criteria: the cluster properties (e.g. the spatial distribution and the size of the clusters), the global variables in the riser, such as the solids volume fraction and the mass flux and finally collision variables such as collision frequency, CPU-time and dissipated energy are recorded. This study reveals significant differences in collision frequency and CPU-time between the hard and soft sphere models, especially at lower superficial gas velocities. Despite these differences, the total dissipated energy remains relatively consistent, leading to similar predictions of solids hold-up and clustering behavior. These findings hold promise for practical applications of collision models within riser systems, with the exception of scenarios approaching pneumatic transport. The simulation results show that the usage of the soft sphere model is preferred when considering dense risers, while the use of the hard sphere model can be beneficial in very dilute riser. Using this insight, an optimal CFD-DEM model can be created for optimizing reactor performance and gas–solid interaction.