Akademik Veri Yönetim Sistemi
INTERNATIONAL JOURNAL OF EXERGY, cilt.49, sa.4, ss.1-32, 2026 (SCI-Expanded, Scopus)
This study experimentally and computationally investigates the thermo-exergetic performance of a salinity gradient solar pond (SGSP) constructed in Karabük, Türkiye. The system was evaluated under natural solar conditions over six days, during which temperature and density distributions, solar 13 radiation, and ambient parameters were recorded. The heat storage zone (HSZ) reached a peak temperature of 49°C, while the system’s highest daily energy and exergy efficiencies were 23.61% and 2.09%, respectively. Heat and exergy inputs/outputs were quantified hourly to assess thermal storage performance and second-law effectiveness. To enhance predictive capacity and minimize experimental costs, an artificial neural network (ANN) model using a scaled conjugate gradient algorithm was developed. The model used time, solar radiation, and ambient temperature as inputs to predict water temperatures at five different depths. Performance metrics confirmed high model accuracy, with average R² values exceeding 0.998 and maximum reaching 0.99956. The mean absolute and squared errors remained low, particularly at lower-temperature zones. The study’s originality lies in integrating experimental data with ANN-based forecasting to evaluate SGSP performance under real conditions. This hybrid approach demonstrates a robust framework for thermal behavior prediction and system optimization in renewable energy applications. The results support ANN as a reliable tool for rapid evaluation of SGSP systems, offering a sustainable pathway for energy modeling and design in regions with strong solar potential.