The current state-of-the-art for industrial process operation is a cascaded structure of Model Predictive Control (MPC), Real Time Optimization (RTO) and Scheduling and Planning (S&P). In this structure RTO represents the weakest link because it is based on steady state models; infrequent model updates (only at steady state!) result in slow ineffective feedback [1]. So there is an opportunity to extend MPC and obtain economic dynamic integration with S&P. It is exactly this integration that will allow MPC and especially NMPC to develop its full economic potential. The research focuses on two questions: 1. How to formulate MPC to achieve dynamic economic integration with S&P? 2. Does this formulation fix all degrees of freedom? Both questions will be dealt with in an open-loop setting.
The formulation boils down to the following constrained optimization problem:
Minimize integrated operational costs from now to tf. Subjected to process behavior, required product quantity/quality at tf and operational limitations.
Here tf stands for the time horizon; tf and the required product quantity/quality at tf are supplied by S&P (schedule!). The objective (F1) is solely based on economics. The last research question was addressed in two numerical experiments [2]. Both experiments show the existence of multiple (non-unique) solutions. So there are still degrees of freedom left to optimize operation further. An attractive way to do this is by introducing a 2nd non-economic objective (F2) and lexicographic ordering; F1, F2. Lexicographic optimization was applied successfully to both experiments. It resulted in unique solutions and the remaining degrees of freedom were used to achieve other operational objectives.
References
[1] D.C. White, Online Optimization: What have we learned?, Hydrocarbon Processing, 77:55-69, 1998.
[2] A.E.M. Huesman, O.H. Bosgra and P.M.J. Van den Hof, Degrees of Freedom Analysis Of Economic Dynamic Optimal Plantwide Operation, Preprints 8th IFAC International Symposium on Dynamics and Control of Process Systems (DYCOPS), Vol. 1, pp. 165-170, Mexico, 2007.