Model Predictive Control of the Mojave solar trough plants. (June 2022)

• Record Type: Journal Article
• Authors: Gallego, Antonio J.; Macías, Manuel; de Castilla, Fernando; Sánchez, Adolfo J.; Camacho, Eduardo F.
• Is Part Of: Control engineering practice. Volume 123(2022)
• Publication Date: 2022-06
• Subjects: Solar energy -- Model Predictive Control -- Solar trough -- Large scale; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2022.105140
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: The size of the current commercial solar trough plants poses new challenges in the applications of advanced control strategies. Ensuring safe operation while maintaining the temperature around an adequate set-point can lead to substantial gains in power production. Furthermore, the controller has to take into account the steam generator constraints to avoid trips leading to production losses. Model Predictive Control algorithms have proved to perform well when controlling solar trough plants. In particular, many MPC strategies were developed and tested at the old experimental solar trough plant of ACUREX at the Plataforma Solar de Almería with excellent results. In this paper, a Model Predictive Control algorithm is presented to control the average temperature of the large scale solar trough plants Mojave Alpha and Mojave Beta. This controller takes into account steam generator constraints in order to ensure safe operation. Several tests under different conditions have been carried out at the actual plants. Results show that the controller performs well on clear and cloudy days in spite of the great size of these plants. Highlights: Advanced control strategies play a decisive role in the improvement of the operation in CSP plants. MPC control techniques have demonstrated a good performance when controlling solar trough plants. A MPC control algorithm is presented and tested for the Mojave solar plants. It is tested at the real plants with excellent results. The controller is currently installed and used by human operators.