(215b) A CFD-DEM Study to Quantify the Influence of Particle Clustering on Catalytic Reactions

Gas-solid catalytic reactors like risers are essential in the chemical industries. The multiphase flow in these reactors leads to the formation of particle clusters, influencing the contact between the gas and particles and reducing the reactant conversion. Three length scales can be identified in these systems – particle size ~O(mm), cluster size ~O(cm), and reactor size ~O(m). Quantitative models that can quantify the impact of particle clusters on the catalytic reactions are limited. To this end, we perform Computational Fluid Dynamics – Discrete Element Model (CFD-DEM) simulations of particle clustering in a triply periodic domain. This simulation setup is representative of the fully developed region in a riser reactor. The simulations are performed for various Damköhler numbers by changing the reaction rate coefficient. We utilize our recently developed DBSCAN-based methodology to obtain the cluster properties, such as volume, area, and chord length. DBSCAN is a commonly used unsupervised machine learning (ML) algorithm. A distribution of cluster size is found, which is characterized by a probability density function (PDF). The effective cluster size that controls the conversion of gas-solid reaction is estimated. A reduced-order model is developed based on the effective characteristic length of the cluster. The predictions of the reduced order model are compared with those of the CFD-DEM simulations.