Systematic model-based optimal design and operation strategies of multistage and integrated continuous crystallisation processes

Crystallisation is an essential purification process in the pharmaceutical industry commonly operated in batch. Despite the recent progress in the adoption of continuous manufacturing in pharma, the development of effective continuous crystallisation processes remains very challenging due to the underlying complex kinetic phenomena and lack of systematic and rigorous design and optimisation strategies particularly in the case of multistage crystallisation and integrated processes. This thesis attempts to address some of the critical challenges using rigorous model-based design and optimisation strategies. The thesis firstly revisits the Attainable Regions methodologies to help design multistage crystallisation processes. The optimal dynamic performance of such processes received particular attention to propose for the first time effective optimal strategies for start-up, steady state, and shutdown to help reduce the transitions times (e.g., start-up time) and maximise on spec productions. In addition, the thesis addresses the case of continuous multiproduct manufacturing and proposes effective methodologies for product grade transition. The thesis also proposes a holistic methodology to help assess the performance of batch vs continuous to guarantee more reliable decision making. Finally, the thesis explores superstructure optimisation with recycles and in presence of impurities and provides key insights into the design of integrated reaction and crystallisation processes. Different optimal control, multi- objective, and mixed integer nonlinear optimization strategies were implemented along with multicriteria decision aiding to deliver a holistic approach starting from early design and sizing to the integration and optimal operation. Cooling antisolvent crystallization of Aspirin was used as a case study due to the underlying complexity of the system and its inherent large degrees of freedom. Many scenarios were considered to deliver the intended optimization objectives by manipulating large sets of decision variables including cooling profiles, antisolvent federate, seeding policies. The proposed methodologies shows that the start-up time of the multistage continuous crystallizer can be reduced by 78%, dynamic performance relevant of product grade transition can be improved by 90%.