Hybrid Particle Cluster CPFD Simulation in the Acceleration and Stabilized Sections of a Downflow Circulating Fluidized Bed

Abstract

This study reports a numerical hybrid cluster simulation of a downer reactor unit using CPFD Barracuda VR software. This model includes an experimentally determined particle cluster size distribution to describe the particle cluster morphological changes as well as the cluster acceleration. Gas–solid flows are considered for a fluid catalytic cracking catalyst with an apparent particle density of 1722 kg/m3, flowing in a 5.08 cm-ID column with air at 300 K and with a pressure close to atmospheric. Conditions for the simulations consist of 1.0–1.6 m/s superficial gas velocity and 30–50 kg/m2 s solid mass flux ranges. Calculations developed involve the Hölzer–Sommerfeld and Ganser drag correlations. The proposed numerical model is validated by comparing experimental and simulated radial cluster particle velocities and particle volume fractions. Simulations developed show that the Hölzer–Sommerfeld drag function in conjunction with a proper selection of cluster geometry provides good cluster velocities and particle volume fraction values both in the cluster acceleration and cluster-stabilized flow unit sections.

Artificial Intelligence for Hybrid Modeling in Fluid Catalytic Cracking (FCC)

“…Thus, the CPFD methodology involves a discrete particle method, where the fluid phase is treated as a continuum and the catalyst particles (solids) are modeled using Lagrangian computational groups of particles, also called "clouds" or "computational parcels" [26]. These computational parcels are used to represent a certain number of actual catalyst particles that share the same properties, including uniform size, velocity, density, and temperature.…”

“…This method has rapidly gained significant attention in recent years. This is the case, considering its capabilities for modeling scaled-up units as well as its ability to assist in the optimization of multiphase gas-solid catalytic reactors [26]. HMs for FCC units have primarily been developed using machine learning (ML) algorithms, combined with FPMs, to create simpler, more flexible, accurate, and interpretable models for industrial-scale units [20][21][22].…”

“…This method has rapidly gained significant attention in recent years. This is the case, considering its capabilities for modeling scaled-up units as well as its ability to assist in the optimization of multiphase gas-solid catalytic reactors [26].…”

Optimization of Yield and Conversion Rates in Methane Dry Reforming Using Artificial Neural Networks and the Multiobjective Genetic Algorithm

“…The multiphase particle in cell (MP-PIC) method adopts an alternative tracking approach by treating groups of particles’ properties as parcels bridging from the Langrangian coordinates to a Eulerian grid, before recycling it back through stress tensors into individual particles. − It was first deployed for fluidized beds in 1993 . Barracuda is a commercial computational particle fluid dynamics software that adopts the MP-PIC approach to solve multiphase flows . The attributes of the MP-PIC approach have recently gained considerable interest in various processes, as it has been validated with experimental data, empirical correlations, and industrial applications. − Lu et al employed both the two-fluid model (TFM) and the MP-PIC approach for simulating the carbon dioxide photoreduction in annular CFB by varying the initial catalyst bed height and inlet gas velocity, among other parameters, for envisioning the rise in bed height and size of formed bubbles and determined that the MP-PIC approach surpassed the TFM for the motion of the catalyst within a single cell .…”

The CREC Fluidized Riser Simulator a Unique Tool for Catalytic Process Development

“…These studies are relevant for the full development of industrial scale fluid bed processes. Furthermore, to be successful, the integration of rigorously derived kinetics with CPFD (Computerized Particle Fluid Dynamic) simulations of large-scale downer units is needed, with the inclusion of particle clusters, as developed by the CREC research team and as reported in Table 1 [3][4][5][6][7][8][9].…”