(740f) Robust Optimal Control of a Novel Process Intensification Strategy for Particulate Manufacturing

Spherical Agglomeration (SA) is a novel Process Intensification strategy for particulate manufacturing. In the context of Pharmaceutical Manufacturing, it has the potential to reduce the number of unit operations in downstream processing from 7 to 3, which significantly reduces the cost of drug manufacturing. Although there has been a huge interest in continuous manufacturing in the past decade, most pharmaceutical processes still operate in the batch paradigm. In this work, a first principle mechanistic model is constructed which captures the fundamental mechanistic processes underlying SA viz. particulate nucleation, growth and agglomeration. An open loop optimization problem is then formulated to control the end-of-batch Mean Particle Size (MPS) in the crystallizer. A closed loop Model Predictive Controller (MPC) is then formulated which significantly improves the controller performance under plant disturbances. However, when there are model parametric uncertainties, the controller performance degrades drastically resulting in huge mismatch between the end-of-batch MPS and the set-point MPS. A new surrogate model-based control framework is then constructed using Polynomial Chaos Expansion (PCE), which is found to give improved controller performance under model parametric uncertainties. It is further found that the system crosses into different regions of the solubility phase diagram for the three control problems which ultimately leads to differences in the quality attributes of the final end-of-batch product agglomerates.