240a A Metric for Monitoring Mpc Optimization Strategy

William Canney, Advanced Process Control, Aspen Technology, Inc., 2500 CityWest Blvd, Suite 1500, Houston, TX 77042

A key element of Model Predictive Control (MPC) applications is the determination of steady-state operating targets based on consideration of the full set of manipulation variables, control variable constraints, and process economics. Since these targets are generally determined as the result of a multivariable calculations, most often formulated as a linear program steady-state target calculation, it is frequently difficult for the user of MPC to detect and understand suttle changes in operating strategy. While active constraint sets can change frequently, the predominate strategies of the MPC application rarely change, thus monitoring constraints is a difficult method for monitoring changes in MPC strategy. This paper defines a new metric for monitoring MPC controller strategy and discusses simulation results, and experiences using this metric on an operating processing unit.