Modeling and optimizing mixing efficiency in a two-impeller stirred tank

For commercial scale reactors, insufficient mixing is a common phenomenon. Imperfect mixing can have significant impact on product quality. Therefore, improving the efficiency of mixing is an important part of process design and optimization. This study intends to develop a computationally feasible model of a stirred tank reactor, and use this model to improve the mixing efficiency. In this study, a scaled-down version of a real stirred tank reactor is modelled. A computational fluid dynamics (CFD) based compartmental model was developed through finite volume discretization of mass balance equation. A resolution sensitivity test was conducted to determine the discretization mesh scheme. This model demands less computation than dynamic CFD simulation while shows satisfactory predictive power. A benchmark chemical reaction system was integrated to characterize the mixing efficiency, which is composed of a first-order decay and a parallel second order coupling. Two operating policies were studied to optimize the mixing efficiency. The first is a static policy which employs a constant feeding rate in a semi-batch process, and the second is a dynamic policy which adjusts the feeding rate dynamically. Feeding point and feeding rate profile are the decision variables. Mixed-integer surrogate optimization was implemented to find the optimum, and the solution was validated with interior point method. The results obtained indicate that the choice of feeding point is independent of feeding rate profile. Additionally, it has been concluded that the static policy results in poor utilization of feed. By implementing the dynamic policy, through the extra degree of freedom, the waste of material is reduced and improvement in the economy of process is achieved.M.S.Includes bibliographical referencesby Shu Yan