On-line nonlinear model-based estimation and control of a polymer reactor

Authors

  • Rajendra K. Mutha,

    1. Dept. of Chemical Engineering, University of Toronto, Toronto, Ont., Canada M5S 3E5
    Current affiliation:
    1. Aspen Technology, Inc., 9896 Bissonnet, Houston, TX 77036
    Search for more papers by this author
  • William R. Cluett,

    Corresponding author
    1. Dept. of Chemical Engineering, University of Toronto, Toronto, Ont., Canada M5S 3E5
    • Dept. of Chemical Engineering, University of Toronto, Toronto, Ont., Canada M5S 3E5
    Search for more papers by this author
  • Alexander Penlidis

    1. Dept. of Chemical Engineering, University of Waterloo, Waterloo, Ont., Canada N2L 3G1
    Search for more papers by this author

Abstract

Polymer reactor control problems often lack frequent measurements of polymer properties, while other reactor environment measurements, such as temperatures, are readily available. The experimental application of a new estimator for handling such multirate measurement systems is presented. The estimator, in conjunction with a detailed nonlinear model and all measurements at their respective sampling rates, is able to provide frequent predictions of the process variables related to the quality (weight-average molecular weight) and quantity (monomer conversion) of the polymer product. The experimental application of a multivariable, nonlinear model-based predictive controller using reactor temperature and initiator flow rate as the manipulated variables for controlling weight-average molecular weight and monomer conversion is presented. The control strategy is designed to handle control nonaffine systems (nonlinear in the manipulated variables), while allowing for the incorporation of rate and amplitude constraints on the manipulated variables. A continuous methyl methacrylate solution polymerization pilot plant was used for the experiment.

Ancillary