Estimation and control in polymerization reactors. A review

Abstract

AbstractThe control of polymerization reactors is particularly arduous for several reasons; (a) difficulty in specification of the control objectives; (b) high exothermicities and viscosities involved: (c) complicated process dynamics: and (d) problems with measurement of the polymer structure. Computer control has now simplified many tasks such as sequential operations of batch reactors; but no less important, it has made feasible the implementation of elaborate state estimation and control algorithms. The tendency in the polymerization industry and in current research centers is centered on producing ‘tailor‐made’ polymers from the view of its microscopic and/or macroscopic structures. To this end, several solutions have been proposed. For example, in the ‘chemical approach’, open loop methods have been developed based on the detailed dynamic models of the processes. Any control technique requires a knowledge of the process to be controlled through mathematical models. These models may be developed either from detailed mass and energy balances, or through empirical input‐output or “black‐box” techniques. The most promising control techniques are probably optimal control, adaptive control, and their combinations. Optimal control is powerful because it utilizes directly the detailed non‐linear plant model. Adaptive control developed for linear plants has already proven satisfactory, in spite of the highly nonlinear nature of the polymerizations. Undoubtedly, improved solutions will be obtained when this technique is extended to non‐linear systems.