An Industrial Perspective on Applying Process Knowledge and Mathematical Programming for RTO, MPC, NMPC, and Power Scheduling

This paper presents an industrial perspective on implementing real-time optimization (RTO), model predictive control (MPC), nonlinear model predictive control (NMPC), and power scheduling in chemical and power plants. It emphasizes integrating process knowledge with mathematical programming for sustained value creation and operational excellence. Three important classes of problems are examined with an emphasis on attaining process objectives in an industrial setting with measured and unmeasured disturbances. Process modeling details are chosen based on plant objectives and available process information for sustained value creation. Mathematical programming techniques and approximations are made to ensure global optimality and robustness. Solution implementation details are chosen after taking process characteristics and distributed control system (DCS) for plant into account.