DOE-Based CFD Optimization of Pharmaceutical Mixing Processes

Fluid mixing and homogenization are key manufacturing processes in the pharmaceutical industry that in an industrial setting are typically optimized and adapted using empirical techniques rather than numerical methods. In the recent years, in silico techniques have increasingly attracted interest due to the many advantages and the increased information content. Computational fluid dynamics, for example, have often been applied to mixing problems. Although numerical flow simulations are nowadays common for simple applications, more complex cases (e.g., industrial mixing) still require much work to achieve reliable results with reasonable resources. In our work, we present an efficient procedure for optimizing the mixing performance of an unbaffled tank for pharmaceutical applications. The optimization objectives were the position of an impeller in the tank defined by the bottom clearance, the eccentricity of the impeller, the angle of the impeller shaft, and the impeller rotational speed. In order to generate a regression model for prediction of the optimal performance, design of experiments was used. Our optimization study showed that the impeller eccentricity had significantly more impact on mixing performance than the shaft angle, that the impeller speed was the main driver for the power input and the average shear forces, and that the bottom clearance may have strongly impacted the flow in the bottom tank area.