Model Predictive Control of Continuous Crystallizers

Abstract

The problem of model predictive control of continuous isothermal crystallizers, using a detailed moment equation model is analysed. The mean size of the crystalline product and the variance of crystal size are the controlled variables, while the manipulated variables are the input concentration of the solute and the flow-rate. The controllability and observability, as well as the coupling between the inputs and the outputs are analyzed by simulation using the linearised model. The crystallizer has proved to be a nonlinear multi-input multi-output system with strong coupling between the state variables. It is shown that the mean crystal size and the variance of can be controlled nearly separately by the residence time and the inlet solute concentration, respectively. By seeding, the controllability of the crystallizer increases significantly. The linear model predictive controller synthesized using the moment equation model appears to be an efficient controller for continuous crystallizers.