Robust adaptive process control of a polymerization reactor

Abstract

Two different approaches to robust adaptive control are applied to the control of monomer conversion (i.e. productivity) of a simulated continuous solution polymerization reactor in the presence of reactive impurities with drifting concentration levels. The robustness of these two controllers is with respect to unmodeled process dynamics and unmeasured disturbances. The first of these approaches uses a predictive control law with a dead-zone and normalization in the parameter estimator. The second approach is based on an augmented predictive control law which incorporates control weighting. A single-input, single-output (SISO) strategy is used to evaluate these approaches using the initiator concentration in the feed as the control variable.