Real-Time optimization as a feedback control problem – A review

• Published in: Computers & chemical engineering Vol. 161; p. 107723
• Main Authors: Krishnamoorthy, Dinesh, Skogestad, Sigurd
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2022
• Subjects: Active constraint switching; Feedback control; Gradient estimation; Real-time optimization; Real-time optimization; Feedback control; Active constraint switching; Gradient estimation
• ISSN: 0098-1354; 1873-4375
• DOI: 10.1016/j.compchemeng.2022.107723

•Comprehensive overview and classification of the different feedback optimizing control approaches that aims to achieve asymptotic optimal operation without the need to solve numerical optimization problems online.•Systematic design procedure of ǣwhat to controlǥ such that the economic objectives are translated to feedback control objectives.•Review of different model-based and model-free gradient estimation approaches.•Comparative study of the different approaches on a benchmark Williams-Otto reactor example. Feedback optimizing control aims to achieve asymptotic optimal operation by directly manipulating the inputs using feedback controllers, without the need to solve numerical optimization problems online. The main question when designing feedback optimizing control is then what to control such that the economic objectives are translated into control objectives. Over the past two decades, active research has resulted in several different approaches to feedback optimizing control stemming from different application areas, leading to a rich source of literature. The different methods differ on the choice of the controlled variables, the need for detailed process models, type of measurements used, convergence speed, accuracy, ease of implementation etc. This paper aims to provide a survey of these different approaches under the unified umbrella of feedback optimizing control, and provides an overview and comparison of the different feedback-based real-time optimization approaches for continuous processes.