Inferential control of a dextrose column crystallizer based on an Extended Kalman Filter

The monitoring of the CSD of the crystal product from an industrial crystallizer requires measuring devices which are seldom applied. This difficulty may be overcome estimating the interesting variables by a soft sensor. The present work describes the application of an Extended Kalman filter for the estimation of the fines content in the crystal product of a continuous column crystallizer for the production of dextrose monohydrate and the subsequent development of an inferential control scheme of the crystallizer, by using the INCA control environment. A rigorous model of the crystallizer was used both to implement the EKF and to check the control process scheme performances.