(55c) Verification of Sub-Grid Drag Models for Gas-Particle Fluidized Beds With Immersed Horizontal Cylinder Arrays

Cylinder arrays are commonly used in fluidized beds to achieve heat transfer—hundreds of immersed cylinders, typically a few centimeters in diameter, can be present in large commercial devices that are many meters tall.  Grid-independent CFD simulations of commercial-scale fluidized beds, where fine particle clusters and flow around individual cylinders must be explicitly resolved, are computationally infeasible.  An alternate strategy is to use faster coarse-grid simulations where the cylinder array is viewed as a stationary, uniform porous medium.  The influence of the cylinders on the suspension is modeled as a filtered cylinder-suspension drag force, and on the cluster/bubble formation is captured via sub-grid corrections to the gas-solid drag.  Filtered models for the cylinder-suspension drag and gas-solid drag for beds with cylinder arrays have been developed in a previous study¹.  This work addresses verification of these sub-grid drag models: the fidelity of the filtered sub-grid model predictions to highly-resolved simulation results is evaluated. 

Results from highly-resolved simulations of a large bed with an immersed cylinder array are generated for two operating regimes—bubbling and turbulent fluidization.  These computationally expensive, grid-independent results are considered to be the ‘exact’ solutions.  Two approaches are followed to perform model-on-model verification.  First, the large bed CFD results are analyzed in an attempt to reconstruct the filtered drag models.  The reconstructed models are then compared against the previously developed filtered relationships reported in Sarkar et al.¹ Second, the filtered constitutive models are implemented as sub-grid corrections in coarse-grid simulations of the large fluidized bed.  Predictions for the macroscopic particle distributions and velocity profiles are compared against the exact solutions (i.e., the highly-resolved simulations).  Addition of sub-grid corrections is found to improve the predictive capabilities of the coarse-grid simulations, which otherwise over-estimate the bed height.  Differences observed between the coarse-grid CFD predictions and the exact solutions may be resolved by further refinement of the filtered constitutive models. 

¹Sarkar, A., Sun, X., Sundaresan, S.  A Sub-grid Model for an Array of Immersed Cylinders in Coarse-grid Multiphase Flow Simulations of a Carbon Capture Device.  AIChE Annual Meeting, Pittsburgh, October-November 2012.