Computational Fluid Dynamics for Determining the Interplay between Stirring Conditions and Crystal Size Distribution in Small Laboratory Devices

A computational fluid dynamics method has been developed to describe the hydrodynamics of different stirring devices, which are commonly used in the development stage of protein crystallization. The experimental setup consisted of minicrystallizers in which the crystallization cocktail was stirred using a magnetic stirrer, mechanical agitator, shaker, or rotating mixer at different rotational frequencies. The hydrodynamics of the setups was determined with respect to local and average values of flow velocity, turbulent dissipation rate (epsilon), turbulent kinetic energy (k), and shear rate ((gamma)over dot). The calculations were performed by using the ANSYS Fluent software, in which the k-epsilon RNG turbulence model was implemented. The model was used to determine the probability density functions (PDFs) of the hydrodynamic parameters, which were correlated with the morphology of the protein crystalline phase obtained in the crystallization experiments. Lysozyme was used as a model protein, which was crystallized from sodium chloride solutions. For the shaker and rotating mixer, the dependence of the crystal size distribution on the k-PDF or (gamma)over dot-PDF distributions followed a similar pattern. For the mechanical agitator, a minor discrepancy was observed from that pattern at higher rotational frequencies. The correlation could not be validated for the magnetic stirrer, in which no crystalline phase was obtained under the crystallization conditions used. This was attributed to the pulverization of the solid phase caused by the magnetic bar.