The Eulerian-Eulerian model was used to solve the equations for both phases, gas and glass beads. The continuity and momentum equations with closure models were solved in all computational cells of the domain. Then the variables for different phases such as velocities and volume fractions are known at every grid point. The continuity and momentum equations for different phases interact through mass transfer and the interaction forces representing the momentum transfer between the phases. The Syamlal-O'Brien model was used to show the drag force between the two phases.
The experimental cold-flow fluidized bed was fabricated from a 152 mm ID clear acrylic tube and aerated with a perforated stainless steel plate containing 132 uniformly distributed 1 mm diameter holes, providing an aeration plate open area of 0.57%. The X-ray computed tomography (CT) system utilized in this study has been described elsewhere (Heindel, T.J., Gray, J.N., and Jensen, T.C., “An X-ray System for Visualizing Fluid Flows,” Flow Measurement and Instrumentation, 19(1): 67-78, 2008). In this study, the CT spatial resolution in the fluidized bed is ~0.6 mm.
A two-dimensional grid study was completed revealing there are two higher velocity regions of air within the bed, and they appear to be axially symmetric. The air velocity is uniformly dispersed without forming large recirculation zones. The glass beads are well-distributed throughout the bed. The volume fraction of glass is smaller in the regions of higher air velocity. The glass velocity distribution shows two large recirculation zones are generated near the wall, while two smaller zones are produced just above the aeration plate. The volume fraction distribution of glass seems to be inconsistent with that of the glass velocity. Three-dimensional simulations were completed with a mesh size of 4 mm. Assuming a uniform inlet velocity profile, 2D and 3D simulations of the fluidized bed with no side port air injection provided similar results, with a relative error within 5%. In both cases, the simulated bed pressure drop agrees with experimental results.
Average gas holdup of the 2D simulations without side air injection were compared to local time-averaged gas holdup maps experimentally acquired with good qualitative agreement. The 3D simulations were used to model side air injection and good qualitative agreement with CT imaging was also obtained. The results show that the local time-averaged gas holdup is similar with that of the experimental x-y, x-z, and y-z CT imaging planes. The CFD simulations gave a detailed view of the velocity distribution throughout the entire bed, especially for glass. However, in order to improve the flow condition and enhance the mass, moment, and energy exchange, more analysis is needed.