Despite their wide applicability, batch crystallization and continuous reactive extrusion processes are very difficult to control. A dynamic and highly non-linear behavior of both processes, as well as the lack of complete understanding of all process phenomena, represent main challenges for implementing a suitable control strategy. Design, implementation and evaluation of real-time model-based control onto these two processes are the key topics of this thesis.

Within the realized research work, a systematic procedure was adopted to implement the model-based control strategy for each process. The procedure consisted of model development (either from first principles or by model identification technique), utilization of model in the controller design, and implementation of the control strategy to the process. Process model was developed from first principles for crystallization process, and obtained by process identification technique for extrusion process, respectively. The crystallization model was based on population balance equation and a kinetic model for nucleation and growth. Models were experimentally validated and incorporated into the control strategy. The experiments were performed within the scope of design of experiments statistical methodology. Real-time optimal control was chosen for crystallization, and model predictive control (MPC) for control of extrusion process. Both control strategies were developed and implemented through integration of state-of-art online instrumentation and high-level computational engines. The novelty of both projects consisted of first time implementation of these model-based control strategies in real-time for both, crystallization of pharmaceuticals and extrusion of thermoplastic vulcanizates. The results of implementation of model based control strategy showed a substantial improvement of the product qualities for both processes, crystallization and extrusion.