Improvement of batch crystallization control under uncertain kinetic parameters by model predictive control

As the widespread use of a batch crystallization process in many industries, finding an optimal operating condition and effective control strategy are significant for improving product quality and downstream operations. To achieve these, an accurate model is required to predict the process behavior and to design an effective and robust controller. However, due to unknown disturbances and batch-to-batch variations, the kinetic parameters obtained from experimental study may not represent the real process resulting in poor control and estimation performances. In this work, improvement of batch crystallization control of a potassium sulfate production under uncertain kinetic parameters has been proposed. An extended Kalman filter (EKF) has been designed to estimate uncertain parameters and un-measurable states. An optimization model based feedback controller known as Model Predictive Control (MPC) technique has been implemented to achieve the desired crystal size distribution (CSD) subject to a product quality constraint i.e., the requirement of seeded crystal size. Simulation results demonstrate that the robustness of the batch crystallizer control satisfying the requirement of crystal quality has been improved by the MPC control integrated with the EKF in the presence of un-measurable states and uncertain parameters.