Optimal Grade Transition Control for Liquid-propylene Polymerization Reactor

A nonlinear control system integrating an off-line optimizer and a nonlinear model-based controller is developed to perform optimal grade transition operations in a continuous pilot plant reactor. A simple black-box model is developed and used to determine optimal trajectories of inputs and outputs for a series of three polypropylene grades. The simplified model is also used to develop a nonlinear controller. This controller is similar to generic model control; however, the integral action is omitted and an on-line updating scheme is incorporated to update pre-specified model parameters using delayed process measurements. The time optimal inputs, which are calculated by the off-line optimizer, are introduced to the plant in a feedforward manner. At the same time, the deviations from the optimal output are corrected using the feedback nonlinear controller. The simulations on a complex mechanistic model of the process reveal that the nonlinear control scheme performs well for both set point tracking and disturbance rejection. This paper integrates well-known methodologies, such as the generic-model control algorithm, parameter update schemes, and off-line optimization, together to develop an applicable and robust control technique for continuous polymerization reactors.