Real-time estimation of polymer properties in an industrial polyethylene reactor

In this paper, we describe and present results from the online, real time implementation of polymer property estimation techniques for an industrial polymerization reactor. We use extended Kalman filtering to keep an online, dynamic model of the INSITE/sup TM/ Technology solution polyethylene process accurately tracking process variables through upsets and product grade change transitions. The predictions of this corrected model are then used by a second Kalman filter to provide real time estimates of the current melt flow index of the polymer being produced in the polymerization reactor. When imbedded within a suitable multivariate control scheme, these state estimation techniques in conjunction with the dynamic model have the potential to significantly improve process performance by reducing the amount of offgrade material produced during type changes, and minimizing the effect of process upsets and disturbances.