Model predictive control of industrial crystallizers
|Project members:||dr.ir. A. Mesbah (Ali), ir. A.E.M. Huesman (Adrie), prof. P.J. Jansens , H.J.M. Kramer|
|Keywords:||Model predictive control, Batch processes, Crystallization, Model-based control|
|Sponsored by:||SenterNovem, BASF, BP|
A substantial amount of materials in the pharmaceutical, food, and fine chemical processes are produced in crystalline, i.e. solid, form. Batch crystallization is a key separation and purification unit in such industries, with a significant impact on the efficiency and profitability of the overall process. Improved control of such processes offers many possibilities to achieve the stringent requirements on the final product quality, namely crystal size, purity and morphology, and also enhance the process efficiency. Nonetheless, the control of batch crystallization processes is a challenging task due to their highly non-linear behaviour, plant-model mismatch, irreproducible start-up, unmeasured process disturbances, and lack of reliable measurements for the system states. It is, therefore, likely that the effectiveness of the optimal operating policies computed offline degrades in real applications.
In the face of recent advances in the field of optimal and model predictive control, as well as the current developments in crystallization modelling, this multi-disciplinary research project primarily concerns the design and real-time implementation of model-based control systems for batch crystallization processes. The viability of the developed controllers for real-time applications has to be verified experimentally by closed-loop implementations on the semi-industrial facilities available at Process and Energy department (TU Delft). Furthermore, plant-wide optimization of an existing industrial particulate processing system will also be investigated in this research project.