Self-optimizing control of an LNG liquefaction plant. (February 2019)

• Record Type: Journal Article
• Authors: Verheyleweghen, Adriaen; Jäschke, Johannes
• Is Part Of: Journal of process control. Volume 74(2019)
• Page Start: 63
• Page End: 75
• Publication Date: 2019-02
• Subjects: Self-optimizing control -- LNG liquefaction -- Refrigeration cycle -- Optimal operation -- Control structure design; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2018.01.007
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: A model for a three refrigerant cascade LNG liquefaction plant is presented. Several control structures are investigated to minimize the operational loss. The exact local method finds the best controlled variables for SOC. The optimal subset of n measurements is selected using a branch and bound method. Abstract: In this paper we apply self-optimizing control (SOC) to a cascaded LNG liquefaction plant. We first introduce the model, and then define the operational objective, which is to achieve minimal energy consumption while satisfying operational constraints. Four control structures are compared; a "standard" temperature control structure, an SOC structure with two plant measurements, an SOC structure that uses a combination of plant measurements as controlled variable, and an SOC structure where we also include measurements of disturbances in addition to the plant measurements. We find that the SOC structures significantly reduce the average steady-state loss when the operating conditions change. We furthermore find that using more plant measurements in the SOC structure results in lowered losses. In particular, for the disturbances considered, the steady-state loss becomes acceptably low, such that there is no need for a supervisory real-time optimization layer. Finally, it has been found that including disturbance measurements results in somewhat reduced losses, although the improvement was insignificant for the studied case. The effectiveness of the SOC framework is shown by closed-loop step responses to selected disturbances.