Modeling, optimization, and sensitivity analysis of a continuous multi-segment crystallizer for production of active pharmaceutical ingredients
Abstract
We have investigated the simulation-based, steady-state optimization of a new type of crystallizer for the production of pharmaceuticals. The multi-segment, multi-addition plug-flow crystallizer (MSMA-PFC) offers better control over supersaturation in one dimension compared to a batch or stirred-tank crystallizer. Through use of a population balance framework, we have written the governing model equations of population balance and mass balance on the crystallizer segments. The solution of these equations was accomplished through either the method of moments or the finite volume method. The goal was to optimize the performance of the crystallizer with respect to certain quantities, such as maximizing the mean crystal size, minimizing the coefficient of variation, or minimizing the sum of the squared errors when attempting to hit a target distribution. Such optimizations are all highly nonconvex, necessitating the use of the genetic algorithm. Our results for the optimization of a process for crystallizing flufenamic acid showed improvement in crystal size over prior literature results. Through the use of a novel simultaneous design and control (SDC) methodology, we have further optimized the flowrates and crystallizer geometry in tandem. We have further investigated the robustness of this process and observe significant sensitivity to error in antisolvent flowrate, as well as the kinetic parameters of crystallization. We have lastly performed a parametric study on the use of the MSMA-PFC for in-situ dissolution of fine crystals back into solution. Fine crystals are a known processing difficulty in drug manufacture, thus motivating the development of a process that can eliminate them efficiently. Prior results for cooling crystallization indicated this to be possible. However, our results show little to no dissolution is used after optimizing the crystallizer, indicating the negative impact of adding pure solvent to the process (reduced concentration via dilution, and decreased residence time) outweighs the positive benefits of dissolving fines. The prior results for cooling crystallization did not possess this coupling between flowrate, residence time, and concentration, thus making fines dissolution significantly more beneficial for that process. We conclude that the success observed in hitting the target distribution has more to do with using multiple segments and having finer control over supersaturation than with the ability to go below solubility. Our results showed that excessive nucleation still overwhelms the MSMA-PFC for in-situ fines dissolution when nucleation is too high.
Degree
Ph.D.
Advisors
Nagy, Purdue University.
Subject Area
Mathematics|Chemical engineering|Pharmacy sciences
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.