Verification and Use of CFD Modeling in Stirred Tank Reactors for Pharmaceutical Applications

Mixing often determines the success or failure of pharmaceutical process scale-up. Many unit operations, including heterogeneous reactions, fed-batch processes, liquid-liquid extractions, and crystallizations are susceptible to changes in performance due to variable mixing. To predict mixing performance between scales, we have adopted a computational fluid dynamics (CFD) modeling approach using MixIT, a CFD package specialized for mixing in stirred tanks. In this talk, we present a comparison of CFD modeling with experimental measurements. To assess the model's accuracy, we measured agitator power consumption, a representative mixing attribute in stirred tanks, up to kilo-scale and compared the measurements with the model's predictions. The measurements were taken across a variety of common agitator types and baffle configurations, with a wide range of power per unit volume values. The results demonstrated good agreement between experiments and simulations, showing the model’s accuracy in predicting mixing attributes. We then demonstrated the use of CFD to predict performance of a mixing sensitive liquid-liquid reaction as it was scaled up to 20L. In this example, reaction monitoring with process analytical technology (PAT) was combined with CFD predictions to develop correlations between mixing attributes and the mass transfer coefficient.