Studies on crystallization process; monitoring and control

Abstract

Crystallization is an oldest unit operations used by industries for the separation as well as purification of a solid product. The mathematical modelling, control and monitoring of crystallization process is a significant research area both from academic as well as industrial point of view. A temperature trajectory that improves the crystal size distribution in a batch crystallizer is proposed and compared with some of the basic cooling modes like natural, linear and controlled cooling. The properties of the crystalline product and dynamic behavior of the crystallizer, obtained by numerical experimentation are presented and analyzed here.The Mathematical modeling and simulation of a non-isothermal continuous cooling crystallizer followed by a control strategy to improve the crystal size distribution has been proposed. The model developed here is used to monitor the process using various multivariate statistical process control techniques. Present dissertation furnishes the successful implementation of various chemometric techniques. Two methods were chosen for process monitoring and control: Clustering of time series data and moving window based pattern matching. PCA similarities and dissimilarity index were chosen as the index of clustering as well as pattern matching. These methods have been successfully implemented in continuous crystallizer for fault detection and to differentiate among various operating conditions. Both the methods produced promising results.