Predictive control of a nonlinear open-loop unstable polymerization reactor

Abstract The temperature control of a solution polymerization reactor with different control algorithms is considered. The plant is characterized by strong nonlinear static and open-loop unstable dynamic behavior; additional difficulties arise from control-variable (cooling oil flow rate) constraints and the impossibility of heating. For this rather complicated object, conventional PID control is compared with two different predictive control methods—adaptive generalized predictive control (GPC) and a newly developed heuristic predictive controller (HPC). The comparison takes into account such aspects like closed-loop behavior and control effort, necessary a priori information about the plant, requirements of software and devices for design and implementation. The simulations show surprisingly good results with the simple but effective HPC algorithm and this demonstrate the usefulness of application of advanced (predictive) control methods to reactor control.