Real-time Optimization with persistent parameter adaptation using online parameter estimation. (August 2018)

• Record Type: Journal Article
• Authors: Matias, José O.A.; Le Roux, Galo A.C.
• Is Part Of: Journal of process control. Volume 68(2018)
• Page Start: 195
• Page End: 204
• Publication Date: 2018-08
• Subjects: Real-time optimization -- Steady-state detection -- Nonlinear state estimation; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2018.05.009
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: New methodology for real time optimization (RTO) that avoids the stead-state wait. The methodology is called Real-time Optimization with Persistent Adaptation (ROPA). ROPA decouples the estimation problem in order to achieve plant-wide optimization. ROPA is compared to the Real-time standard RTO and presents improved economical results. Abstract: One of the major drawbacks of traditional Real-time Optimization (RTO) is the steady-state wait before estimating the parameters. This paper proposes an alternative solution called Real-time Optimization with Persistent Adaptation (ROPA), which integrates on-line parameter estimation in the optimization cycle, avoiding the SS detection step. Essentially, the idea is to use transient information to update the steady-state economic optimization problem and, then, by continuously solving it, the calculated optimal solution would reach the actual plant steady-state optimum in a given time horizon. ROPA provides an intermediary solution between static and dynamic optimization schemes. While it approximates the optimal trajectory, ROPA design enables the application of techniques to plant-wide optimization and the use of well-established static RTO commercial solutions. The new methodology benefits are illustrated with a case study, in which the traditional RTO and ROPA schemes are applied to the Williams–Otto reactor. Their performance is compared based on profit loss and deviation from the actual optimal decisions. The results show that the refinement of the prediction capacity by decreasing the time between two sequential optimization leads to a better economic performance and enhances the disturbance detection of the optimization cycle.