Product Quality Control in Semibatch Reactors Using Midcourse Correction Policies

Abstract

A practical approach to the control of final product quality in semibatch reactors is proposed. It is based on the use of readily available on-line measurements such as temperatures plus a few off-line analyses obtained from one or more samples taken from the reactor throughout the course of the batch run. These measurements are used either with a theoretically-based model or with simple empirical regression models to predict the final product properties. If the predictions fall outside of a defined no-control region, then a midcourse correction is made to bring the product quality closer to target. The approach is illustrated for the control of molecular weight and cross-link density in the simulated semibatch emulsion polymerization of styrene−butadiene rubber (SBR).