Control of batch and semibatch free radical solution polymerization using GPC molecular weight monitoring

The control of polymer molecular weight distribution during both batch and semibatch free radical solution polymerization using GPC molecular weight monitoring was studied in this work. A mathematical model expressing the polymerization reaction in terms of the three molecular weight distribution moments has been developed. The sensitivity of the system output variables to the control variables was examined by sensitivity analysis and the variations of weight and number average molecular weights and the molecular weight distribution at different reaction conditions for isothermal, temperature perturbed and initial reactant composition error cases have been simulated. The model was used for the estimation of the status of the reaction and to predict the weight and number average molecular weights of polymer at the end of the reaction to determine the best control policy. The control strategies were developed by applying the maximum principle incorporated with the gradient method.;The control strategies studied consisted of single variable control by using reaction temperature or initiator concentration as the manipulated variable, and multivariable control by using the combinations of initiator and monomer addition as well as reaction temperature and initiator addition as control variables. These control strategies were applied to different reaction cases such as temperature perturbation and initial reactant composition error and experimentally shown to provide satisfactory control performance for most cases.;The application of this kind of control strategy requires exact information regarding the status of the state variables in the reactor. In order to obtain the optimal estimates from the noise-upset process model prediction data and the error corrupted measurements and to obtain the state of unmeasurable variables from easily accessed available measurements, the extended Kalman filter method was used for the estimation and reconstruction of the measurements. Two different cases were studied. In the first case only the measurements of reaction temperature and monomer concentration were considered. In the second case, the measurement of weight and number average molecular weights of the polymer by GPC were taken into account in the filter.