Reliability-Based Design and Control of Crystallization Systems Under Parametric and Process Uncertainties

Population balance model-based approaches are widely employed in crystallization process design to meet the industry-specific critical quality attributes (CQAs). However, their reliability is often limited by parameter uncertainties arising from noisy experimental data, which if, unaccounted for, can lead to suboptimal or failed designs. This study introduces a novel reliability-based design optimization (RBDO) framework for open-loop design of crystallization systems under uncertainty. The framework formulates CQA constraints as probabilistic constraints with user-defined reliability thresholds, enabling quantified, risk-informed process design. To solve the resulting RBDO problems efficiently, a surrogate modeling-based, nested simulation-optimization workflow is developed. The proposed methodology is applied to both batch and continuous pharmaceutical crystallization systems, demonstrating its ability to generate robust design trajectories with significantly higher probabilities of meeting CQAs under both parametric and operational uncertainties compared to the deterministic designs. The computational efficiency and accuracy of the approach are further validated through benchmarking against back-mapping-based solutions of the underlying chance constrained formulations.