470926 Modeling an Industrial-Scale Fluidized Bed Wurster Coating Process Using CFD-DEM

Thursday, November 17, 2016: 4:09 PM
Golden Gate (Hotel Nikko San Francisco)
Peter Böhling1, Dalibor Jajcevic2, Conrad Davies3, Alan Carmody4, Pankaj Doshi5, Johannes G. Khinast6, Mary T. am Ende5 and Avik Sarkar5, (1)Research Center Pharmaceutical Engineering GmbH, Graz, Austria, Graz, Austria, (2)Research Center Pharmaceutical Engineering (RCPE), Graz, Austria, (3)Worldwide Research and Development, Pfizer Inc., Sandwich, Kent, UK, (4)Material Science and Oral Products Center of Emphasis, Worldwide Research and Development, Pfizer Inc., Sandwich, Kent, UK, Sandwich, United Kingdom, (5)Worldwide Research and Development, Pfizer Inc., Groton, CT, (6)Institute of Process and Particle Engineering, Graz University of Technology, Graz, Austria

Industrial fluid-bed coating operations involve the flow of several hundred million beads in a large-scale Wurster-coating device. Experimental measurements available from such large scale operations are limited—very few beads per sample (10-100) can be analyzed for coating thickness uniformity. In such scenarios, computational fluid dynamics (CFD) based modeling models become important for developing a better understanding of the process and identifying the key parameters that influence coating-thickness uniformity. With recent advances in computational hardware and software, including graphics card (GPU) based processing, simulations of larger industrial-scale systems can be performed.

Previous results have shown that the CFD-DEM (discrete element method) approach is better-suited to obtain particle trajectories compared to the CFD-TFM (two-fluid model) approach. With GPU-enabled codes, ~20 million beads can be realistically simulated. Therefore, exploiting the azimuthal symmetry of the geometry to simulate a smaller 3-D sector (pie-slice) of the full unit, the flow results are shown to be reasonable with this idealization. Single-phase CFD simulations were performed to determine the air flow distribution through the distributor plate, which is subsequently used as the inlet boundary conditions for the CFD-DEM models. The bead-size dispersity is also incorporated in the model.

The influence of controllable process parameters, such as air-flow rates, and material properties can be virtually investigated using these CFD-DEM simulations. The residence time distributions of the beads in the active coating region (inside the Wurster column) and the non-coating regions are compared for the different conditions explored. These residence time distributions, especially inside the Wurster column, provide qualitative and quantitative measures of the bead-coating uniformity at present. In future, these residence time measurements can be extrapolated to predict long-term coating uniformity using compartment-based and/or population-balance models.

Extended Abstract: File Not Uploaded