Hydrodynamic modeling of porous hollow fiber anti-solvent crystallizer for continuous production of drug crystals

Abstract

This study introduces a Computational Fluid Dynamics (CFD)-based hydrodynamic model for a novel porous hollow fiber-based anti-solvent crystallizer (PHFAC). Here the anti-solvent, water, is introduced from the hollow fiber lumen into the acetone-based feed solution of a drug flowing on the shell side where crystals of the drug Griseofulvin are formed and swept away. Such a device is very useful for achieving continuous anti-solvent crystallization with significant crystal size control. It is necessary to predict the hydrodynamics and other mixing characteristics of the PHFAC device to facilitate optimization and prediction of their behavior. In this study, a 3D physical model for a PHFAC device was first constructed using the software GAMBIT. After meshing and optimization, a 3D computational simulation-based modeling was conducted by the CFD software FLUENT for understanding the hydrodynamics and the mixing characteristics of a miscible aqueous-organic system in the PHFAC module without any drug being present. Different combinations of characteristic parameters e.g., inlet flow rates of the shell side and tube side and numbers of hollow fiber membranes, were investigated to optimize the conditions from the perspective of shell-side mixing of the two fluids. The CFD-based study revealed an optimized flow rate combination by examining the interior mixing pattern in the module shell-side under different flow conditions. The optimized flow rate ratio was found numerically to be the same for two different sized modules with one having twice the number of hollow fiber membranes than the other. The CFD simulation results of mixing conditions suggest that the optimized flow rate combinations for shell-side mixing are also coincident with previously obtained experimental results of smaller crystal size of the drug, Griseofulvin and the morphology of the precipitated drug crystals. However, the absence of any crystal formation dynamics in the CFD model suggests the need for a more comprehensive model which can predict the crystal size distribution (CSD).