On-line data reconciliation and parameter estimation for an industrial polypropylene reactor

Abstract

This work presents the implementation of a simple methodology for dynamic data reconciliation and simultaneous estimation of quality and productivity parameters using data from an industrial bulk Ziegler-Natta propylene polymerization process. A simple model of the real process, based on mass and energy balances, was developed including only the variables of greater significance for the desired parameter estimation problem. The resulting nonlinear dynamic optimization problem was solved using a sequential approach on a time moving window. Resuslts have shown that including the energy balance increases information redundancy, leading to better estimations than the ones obtained when energy contribution are not considered.