Control strategies for multivariable processes: Distillation columns & polymerization reactors

Most industrial multivariable processes are controlled by multiloop (single-input single-output) controllers. This thesis discusses the implementation of the recently developed "Conservative Model Based Control (CMBC)" algorithm on several multivariable processes. CMBC has a single tuning constant compared to three tuning constants in a PID controller. When implemented on simulated distillation columns, multiloop CMBC performed better than the multiloop PI controllers, as designed by the BLT method (Luyben 1985). In some cases, responses obtained by PID controllers were superior to those of the PI controllers and comparable to the ones obtained by CMBC method. In one case tested, PID controller performed poorly in comparison to the PI and CMBC control strategies.;The successful implementation of the multiloop CMBC algorithm led to its testing on a simulated reactor used for continuous solution polymerization of methyl methacrylate. The performance of multiloop CMBC is compared with that of multivariable "Simplified Model Predictive Control". Good regulatory and servo behavior was observed in most of the cases.