Closed-loop stochastic model predictive control in a receding horizon implementation on a continuous polymerization reactor example

We evaluate closed-loop stochastic model predictive control techniques on a nonlinear high density polyethylene fluidized bed example. Closed-loop MPC is a control strategy in which one optimizes feedforward signals while maintaining back-off to inequality constraints on the process variables. This back-off is kept minimal by using so-called closed-loop model prediction in which control plays a central role. The objective in this paper is to illustrate these novel techniques on a realistic simulator of an industrial size HDPE plant by enabling a grade change under persistent disturbances in the feeds.