Enhancing Polyethylene Reactor Operation, Efficiency & Process Control Using Online Inference of Polymer Properties: Melt Index & Density

Abstract

The main obstacle to controlling product quality in polymerization reactors is the lack of appropriate online polymer property measurements. In the UNIPOL Polyethylene reactor, a method for inferring melt index and density presented. From the available on-line temperature and gas composition readings, theory-based models are developed to anticipate quality characteristics. Online parameter updates are performed for these models’ movable parameters using infrequent laboratory measurements and back propagation-based parameter estimates. Operating data from LLDPE2 (UNIPOL PE) reactor is used to illustrate how this methodology is applied. These methodologies, when combined with the most recent recursive algorithms, allow inferring crucial polymer parameters, thus enabling way for Advanced Process Control by giving operators the ability to designate set point for the parameter once it has been fully validated. Knowledge of product property deviations from desired targets is required so that plant operators can take corrective actions to reduce the quality of off-grade material and produce consistent quality product.