Optimal state estimation in the transesterification stage of a continuous polyethylene terephthalate condensation polymerization process

Abstract

This paper describes an application of the extended Kalman filter algorithm to the state estimation of the melt transesterification stage of a continuous polyethylene terephthalate condensation polymerization process. When only two on-line measurements of reaction variables are used for state estimation, the prediction of reaction rates and product concentrations are unsatisfactory. When such limited on-line measurements are supplemented by five additional off-line measurements of various functional group concentrations, the system is completely observable and the overall performance of the state estimator is greatly improved. It has been shown that the analysis delay of 24 h is quite adequate for accurate state estimation of the process. In particular, the concentration of unwanted side product such as diethylene glycol (DEG) was predicted precisely. The simulation results indicate that the extended Kalman filter can be used successfully with a dynamic transesterification process model for precise control of reaction product quality.