Combined cooling and antisolvent crystallization is a critical unit operation in pharmaceutical manufacturing, especially for heat-sensitive or poorly soluble active pharmaceutical ingredients. The model-based design of these systems relies on the accuracy of the underlying growth and nucleation kinetic parameters. Unlike temperature where these kinetic parameters are well-known to follow an Arrhenius relation, their dependency on solvent composition still remains unclear, especially in continuous mixed-suspension, mixed-product removal (MSMPR) systems. In this paper, we use population balance modeling coupled with nonlinear regression to estimate growth and nucleation parameters as a function of both temperature and solvent composition. As solvent composition increases from 44 vol % to 66 vol % solvent, both growth and nucleation rates were observed to decrease monotonically with their values reduced by almost one-third. It was also shown that, if the solvent dependency is ignored, the yield can be overpredicted or underpredicted by as much as 15%.

Combined cooling and antisolvent crystallization enables crystallization of many pharmaceutical products, but its process design typically neglects solvent composition influences on crystallization kinetics. This paper evaluates the influence of solvent-dependent nucleation and growth kinetics on the design of optimal, multistage mixed-suspension, mixed-product removal (MSMPR) crystallization cascades. The ability to independently select temperature and solvent compositions in each stage of the cascade serves to greatly expand the attainable region for a two-stage cascade, with diminishing returns for additional stages. Failure to include solvent-dependent kinetics can result in simulating incorrect attainable regions, active pharmaceutical ingredient (API) yields, and crystal size distributions. This work also demonstrates that commonly employed crystallization process design heuristics, such as equal antisolvent addition and decreasing temperature in successive stages, can result in suboptimal process design if kinetics are strongly solvent dependent.

No abstract

In a standard optimization approach, the underlying process model is first identified at a given set of operating conditions and this updated model is, then, used to calculate the optimal conditions for the process. This "two-step" procedure can be repeated iteratively by conducting new experiments at optimal operating conditions, based on previous iterations, followed by reidentification and re-optimization until convergence is reached. However, when there is a modelplant mismatch, the set of parameter estimates that minimizes the prediction error in the identification problem may not predict the gradients of the optimization objective accurately. As a result, convergence of the "two-step" iterative approach to a process optimum cannot be guaranteed. This paper presents a new methodology where the model outputs are corrected explicitly for the mismatch such that, with the updated parameter estimates the identification and optimization objectives are properly reconciled. With the proposed corrections being progressively integrated over the iterations, the algorithm has guaranteed convergence to the process optimum and also, upon convergence, the final corrected model predicts the process behavior accurately.The proposed methodology is illustrated in a run-to-run optimization framework with a fed-batch bioprocess as a case study. Highlights An iterative optimization algorithm is proposed. The model outputs are corrected iteratively to account for model-plant mismatch. Parameters are estimated to satisfy both identification and optimization objectives. The proposed algorithm is illustrated using a fed-batch bioprocess.