Numerical investigation and neural network modeling of the performance of a dual-fluid parabolic trough solar collector containing non-Newtonian water-CMC/Al2O3 nanofluid. (December 2021)

• Record Type: Journal Article
• Authors: Al-Rashed, Abdullah A.A.A.; Alnaqi, Abdulwahab A.; Alsarraf, Jalal
• Is Part Of: Sustainable energy technologies and assessments. Volume 48(2021)
• Publication Date: 2021-12
• Subjects: Artificial neural network -- Non-Newtonian nanofluid -- Numerical simulation -- Parabolic trough collector -- Two-phase solution; Renewable energy sources -- Periodicals; Energy development -- Technological innovations -- Periodicals; Electric power production -- Periodicals; Energy storage -- Periodicals; 333.79
• DOI: 10.1016/j.seta.2021.101555
• Languages: English
• ISSNs: 2213-1388

Abstract

Highlights: Changes of PTC efficiency with Reynolds number has an ascending-descending pattern. Thermal efficiency of the dual-fluid PTC is higher than a conventional collector. The best performance occurs at φ = 1.5%, Re =20000 and insulation angle of 90°. The R 2 value for the developed model for efficiency of PTC using ANN is 0.9998. Abstract: The performance of a novel parabolic trough collector (PTC) equipped with a non-circular absorber tube and a solid insulation is investigated numerically using the Eulerian-Eulerian two-phase method. The water-CMC/Al2 O3 nanofluid is considered as the working fluid. The impacts of the Reynolds number, nanoparticle volume concentration, nanoparticle diameter and insulation angle on the performance metrics are examined. The results showed that the changes of collector efficiency with Reynolds number has an ascending-descending pattern. In addition, it was found that the highest collector efficiency is 61.7%, which belongs to the case of dual-fluid collector with a novel tube and insulation angle of 90° containing nanofluid with φ = 1.5% and nanoparticle diameter of 100 nm. Finally, the artificial neural network was employed to provide a predictive model for the collector efficiency. It was found that the neural network consisting of eight neurons has a high potential in forecasting the efficiency of the novel collector. The maximum deviation was less than 0.2% and this high accuracy caused the R-square for the neural network to be 0.9998.