Title: A novel hybrid experimental - ANN framework for thermo-exergetic evaluation of salinity gradient solar ponds

Authors: Y. Özcan; E. Deniz; M. Gürdal; İ. Ekmekçi

Addresses: Department of Mechanical Engineering, Kastamonu University, 37150 Kastamonu, Türkiye ' Department of Mechanical Engineering, Karabuk University, 78050 Karabük, Türkiye ' Department of Mechanical Engineering, Kastamonu University, 37150 Kastamonu, Türkiye ' Faculty of Engineering, İstanbul Ticaret University, 34000 İstanbul, Türkiye

Abstract: In the hybrid experimental study conducted, the thermal and exergy performance of a salinity gradient solar pond (SGSP) was tested under real outdoor conditions and later neural network (ANN) method was developed to predict temperatures at various depths of the pond. The SCG algorithm was used in the optimised ANN model. The highest thermal efficiency was found to be around 20.68%, and the exergy efficiency was close to 0.81%. The ANN model performed well, reaching an average R² value above 0.998. These outcomes show that the proposed model can successfully forecast pond temperatures with high reliability.

Keywords: salinity gradient solar pond; energy and exergy analyses; machine learning.

DOI: 10.1504/IJEX.2026.153926

International Journal of Exergy, 2026 Vol.50 No.1, pp.38 - 56

Received: 17 Sep 2025
Accepted: 12 Nov 2025
Published online: 08 Jun 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article