| Constrained Model Predictive Control (CMPC) is a model-based multivariable control strategy. It combines feedback/feedforward controls with process optimization using Linear Programming (LP). Its ability to handle constraints of a complex system makes CMPC superior to PID, and it has received more and more attention.; Polymerization is one of the most important commercial processes in modern chemical industry. It also poses great challenges for effective and optimal process control due to its complex nature and, especially, its unusual dynamic behaviors such as gel effects, multiple steady states and reaction auto-acceleration coupled with high exothermic reaction. Solution polymerization of methyl-methacrylate (MMA) in a continuous stirred tank reactor (CSTR) is a typical example which exhibits such great challenges.; As one of only a few attempts to implement advanced control strategies to polymerization, this study has applied CMPC to the complex PMMA process. This non-square system has been simulated on SPEEDUP, a simulator developed by AspenTech. The dynamic model developed includes the leading moments of the distribution of polymer chain length, with CMPC directly controlling the polydispersity and the number average chain length (NACL). A multi-loop PID controller is also designed and applied to the PMMA (modified square system). Simulation results indicate that in most cases CMPC has outperformed PID, in terms of overshoot and movement of manipulated variables. CMPC, due to its capability for process optimization, has performed well over the non-square system with manipulated variables outnumbering controlled ones. CMPC is also successfully applied to three CSTR's-in-series, a system with more inputs than outputs. The maximization of monomer feed flow rate through process optimization has been demonstrated. |
