STATE ESTIMATION FOR POLYMERIZATION REACTORS

The use of extended Kalman Filters for state estimation in polymerization reactors is discussed. An important factor in making these filters work well in practice is to properly account for the stochastic disturbances in the system through the introduction of meaningful stochastic states. In particular, since reactive impurities are the major source of these stochastic disturbances, it is natural to incorporate the stochastics in the model through the reactive impurity terms. This approach is illustrated experimentally for the emulsion polymerization of styrene in batch and continuous reactors. Using on-line density measurements the extended Kalman Filter is shown to track conversion and latex particle size very closely, and to provide good estimates of the impurity levels in the reactors.