Artificial Intelligence-based prediction of nanofluid thermal conductivity: a comparative evaluation of machine learning models
Thermal Science and Engineering Progress, cilt.77, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 77
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.tsep.2026.104857
- Dergi Adı: Thermal Science and Engineering Progress
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC
- Anahtar Kelimeler: Heat transfer, Machine learning, Nanofluids, Prediction, Thermal conductivity, Thermophysical properties
- Karadeniz Teknik Üniversitesi Adresli: Evet
Özet
Nanofluids have attracted considerable attention as advanced heat transfer fluids due to their superior thermophysical properties. Among these properties, thermal conductivity is one of the most important properties affecting heat transfer performance. However, experimental measurement of nanofluid thermal conductivity is usually expensive, time-consuming, and limited to specific operating conditions. Therefore, accurate data-driven prediction models offer an attractive alternative. Accordingly, in this study, machine learning models are used to predict the thermal conductivity of nanofluids. The nanoparticle concentration, nanoparticle type, base fluid, and temperature are used as input parameters. A comprehensive dataset comprising 1064 experimental data points is collected from the literature to train and validate different machine learning models. The used models are Linear Regression, Support Vector Machines, Gaussian Process Regression, Decision Trees, and Artificial Neural Networks. The predictive performance of these models is evaluated by computing the coefficient of determination, root mean square error, and mean absolute error. Among the investigated models, the Fine Tree model provides the highest predictive accuracy. The results indicate that machine learning provides a fast, reliable, and computationally efficient tool for predicting nanofluid thermal conductivity within the experimental domain represented by the compiled database, thereby reducing the reliance on extensive experimental measurements. However, it should be noted that the developed models are applicable only within the experimental domain represented by the compiled dataset.