UCLA

In this work, we focus on multiscale modeling and control of a seeded batch crystallization process used to produce ibuprofen crystals. For the modeling of the crystal growth process, we consider kinetic Monte Carlo (kMC) simulations comprising of molecule adsorption, desorption, and migration type microscopic surface events. To account for growth rate variability, we propose a model for growth rate dispersion (GRD), based on available experimental data, which will be applied at the individual crystal growth level in the kMC simulations. Finally, a model predictive controller (MPC) is developed in order to control the crystal size distribution of ibuprofen in the batch crystallization process and the MPC closed-loop performance is compared against constant temperature control (CTC) and constant supersaturation control (CSC) policies. The proposed MPC is able to deal with the constraints of the control problem, in addition to minimizing the spread of the crystal size distribution in a superior fashion compared to the other control methodologies, which improves the crystal product quality at the end of the batch.