Model Requirements for Next Generation Integrated MPC and Dynamic Optimization

This paper outlines the requirements imposed upon models and modeling techniques applied for high performance model based control systems and model based optimizers for support of the operation of (chemical) processes. An overview is given of on-going developments in the area of integrated high performance model predictive control and model based dynamic optimization (INCOOP and IMPACT projects). To enable tight control of processes at given Cpk values and within imposed 6-sigma intervals on a variety of process variables, the models applied for control and for optimization have to be sufficiently accurate both as a function of frequency as well as a function of the time varying operating conditions. Requirements on the models are summarized. Modeling of relevant process behavior for control and optimization is a very significant cost factor in overall application development. Reduction of these costs by extensive use of a-priori knowledge and by integration of various modeling techniques is discussed.