(368v) Developing Digital Design Frameworks for Enhancing Pharmaceutical Crystallization Processes

Modeling, Optimization, Process optimization under uncertainty, Machine learning, Pharmaceutical manufacturing, Crystallization process development

Research Abstract:

The pharmaceutical industry demands well-designed and controlled crystallization processes to consistently produce active pharmaceutical ingredient (API) crystals with target purity, yield, polymorphic form, and crystal size distribution (CSD). Although the advantages of model-based design and control methodologies over the traditional, resource-intensive design of experiment (DoE)-based approaches for crystallization process design are well-researched, the development of systematic frameworks for constructing model-based digital twins remains an ongoing pursuit. This research aims to address this gap by developing integrated modeling and experimental design frameworks for constructing digital twins of pharmaceutical crystallization processes.

In this work, a systematic framework for the digital design of a continuous cooling crystallization process is presented. The key features of the framework include the experimental investigation of operating space, kinetic parameter estimation using population balance modeling, and kinetic parameter translation studies from batch to continuous crystallization processes. A novel parameter uncertainty-propagation-based methodology was also developed to justify the robustness of the parameter estimation procedure.¹

Recognizing the importance of uncertain model parameters, the framework is further extended to incorporate the impact of kinetic parameter uncertainty in optimal process design through a chance-constrained programming approach. This methodology identifies optimal operating conditions with a higher probability of achieving desired crystal quality attributes compared to designs that ignore model parameter uncertainty.² To ensure accurate models with reduced parameter uncertainty, iterative model-based experimental design (IMED) frameworks were also developed, facilitating automated mechanistic model identification and optimal experimental design to enhance parameter precision.³

The application and experimental validation of these frameworks are demonstrated using a commercial API, Diphenhydramine hydrochloride.

References:

1. Barhate Y., Kilari H., Wu, W.-L. & Nagy, Z. K. Population balance model enabled digital design and uncertainty analysis framework for continuous crystallization of pharmaceuticals using an automated platform with full recycle and minimal material use. Eng. Sci. 287, 119688 (2024).
2. Barhate Y., Nagy Z.K., Reliability-based optimal control of crystallization systems under uncertainty. IFAC –PapersOnLine, 2024.
3. Barhate Y., Kilari H., Nagy Z.K., Automated model-based experimental design procedure for robust digital twin development for continuous crystallization systems. AIChE Annual Meeting, 2023.