(91g) CFD Simulations for Prediction of Scaling Effects In Pharmaceutical Fluidized Bed Processors At Three Scales

Effective operation of fluidized bed processors requires an even gas distribution and good mixing of the particles.  Scaling up a drying process from lab scale to production scale presents unique challenges, such as preventing granule attrition, fines loss, or segregation.  In addition, current methods make it difficult to optimize processor performance without  large scale experimentation due to the scale dependence of influences on the fluidization and related effects on the fluidized particle (e.g. attrition).     Fluidization behavior is also dependent on the particle size distribution and mechanical properties of the powder, so it is necessary for these to be considered in simulating a fluidized bed processor.  As the subject model was intended to capture fluidized bed behavior, it was necessary to choose a numerical method capable of accounting not only for the particle-fluid effects (e.g., drag) but also for particle-wall impacts and particle-particle interactions.

The particle-fluid dynamics within fluidized bed processors of approximate diameters 140mm, 300mm, and 900mm were simulated using a commercial CFD code with a unique model for particle-fluid simulation.  In this novel application of the computational technique to fluidized bed processors in this configuration, for each processor multiple flow rates were simulated at isothermal conditions for processor loadings between 0.15kg and 45 kg of a powder with a size distribution from 2 to 1500 microns in diameter.  The simulations modeled both start-up and steady operation of the processors on a full 3D, time-transient basis within a reasonable calculation time (1-2 days).

This study was performed on a “blind test” basis, with no results shared with the CFD analyst until the simulations had been completed.  Simulation results were compared with experimental data for dense phase bed height, elutriated particle trajectory, and patterns in the fluidization flow structure.  Due to scale-dependent effects, the flow structure varied from a pattern approaching spouted flow to a homogeneously fluidized bed.  Observations through the sight glasses of the dense phase and elutriated particle flow fluid beds were also compared to model predictions.  The simulation results compared well with both the laboratory measurements and qualitative observations.  The numerical method and simulation approach were found to successfully predict the flow behavior of several fluid bed scales, yielding qualitatively correlated information based solely on material properties, processor loading, inlet air flow rate and processor geometry.  This implies that CFD of this kind may be used to assist in determining select operating parameters for pharmaceutical processors and in scaling processes to different sizes to meet production demands.