Akademik Veri Yönetim Sistemi
Journal of Thermal Analysis and Calorimetry, 2025 (SCI-Expanded, Scopus)
The growing demand for materials used in automobile manufacturing, whether for electric or internal combustion vehicles, stems from the continuous rise in vehicle production. To meet these demands, reducing vehicle mass while maintaining performance has become a key objective. In this research, the heat transfer characteristics of Al2O3–water nanofluids are examined within automotive radiator systems by combining laboratory-scale experimentation with artificial intelligence modeling techniques. Experiments were conducted across varying nanoparticle concentrations, flow rates, and inlet temperatures. The results were used to develop artificial neural network (ANN) and physics-informed neural network (PINN) models to predict key performance parameters, namely the Nusselt number and convection coefficient. The novelty of this study lies in two key aspects: (i) experimental investigation of Al2O3 nanofluid behavior in a plate-fin heat exchanger with offset strip fins under laminar flow conditions and (ii) comparative evaluation of PINN and ANN modeling approaches for thermal performance prediction—a first in the context of nanofluid-cooled automotive radiators. The optimal thermal performance was achieved at 0.05% Al2O3 concentration, resulting in up to a 56.3% increase in the convection coefficient. At higher concentrations, performance declined due to increased viscosity and particle agglomeration. In model comparisons, the PINN architecture demonstrated superior accuracy, outperforming ANN with approximately 2.4% and 3.5% higher R2 scores (0.9567 and 0.9762) for Nusselt number and convection coefficient predictions, respectively. These findings highlight the potential of combining nanofluid-enhanced heat exchangers with physics-informed AI models to achieve accurate and efficient predictions in automotive thermal management applications.