(35d) Experimental Determination, Computational Prediction, and Modeling-Based Analysis of Mixing Time in a Fully Baffled Stirred Vessel Equipped with a High Shear Homogenizer

Shear sensitive batch processes are commonly found in the chemical and pharmaceutical industries. These processes can be enhanced by introducing a high shear homogenizer in a stirred tank system agitated by a conventional primary impeller. This arrangement combines the blending/suspending ability of the primary impeller with the shear effects introduced by the homogenizer in a single device. In this work, the mixing time required to homogenize the system's liquid content in a fully baffled, stirred tank equipped with a centrally mounted low speed primary impeller (disk or pitched blade turbine) and an additional high-shear rotor-stator homogenizer was experimentally determined using a colorimetric technique coupled with image analysis. The effect of the presence of the homogenizer on mixing time was assessed for different operating variables, including tank size, type of primary impeller and the agitation speeds of both the homogenizer and the low speed impeller. The experimental results were compared with predictions based on two different computational/mathematical modeling approaches. The first was a CFD-based approach where mixing time was predicted by computationally determining the rate of dispersion of a tracer. This approach utilized CFD simulations to predict the velocity distribution, which were also experimentally validated. The second approach was based on the development of a compartmental model in which the flow between compartments was assumed to be directly proportional to the pumping capacity of all the mixing devices in the system. Experiments data show that the overall blend time produced by the primary impeller alone was significantly reduced by the presence of the high shear homogenizer in the tank. The same trend was observed at different primary impeller speeds, homogenizer speeds, and total tank volumes. Good agreement between the experimentally measured mixing times and the predictions of both the CFD-based model and the compartmental model was found.