Maqola

Evolutionary machine learning for thermophysical property modeling and materials design of CNT–metal oxide hybrid nanofluids

S. BourazzaDepartment of Mathematics, College of Science, Jazan University, P.O. Box. 114, Jazan 45142, Saudi ArabiaAs’ad AlizadehDepartment of Civil Engineering, College of Engineering, Cihan University-Erbil, Erbil, IraqMuneera AltayebFaculty of Engineering, Hourani Center for Applied Scientific Research, Al-Ahliyya Amman University, Amman, JordanAnshy SinghDepartment of Computer Engineering and Applications, GLA University, Mathura, (UP), IndiaHanen KaramtiDepartment of Computer Sciences, College of Computer and Information Sciences, Princess Nourah Bint Abdulrahman University, P.O. Box 84428, Riyadh 11671, Saudi ArabiaAbdellatif M. SadeqFaculty of Agricultural Mechanization, TIIAME National Research University, Kori Niyoziy 39, 100000 Tashkent, UzbekistanHusam RajabCollege of Engineering, Department of Mechanical Engineering, Najran University, King Abdulaziz Road, P.O Box 1988, Najran, Saudi ArabiaHeena AroraDivision of Research and Development, Lovely Professional University, Phagwara, Punjab, IndiaNarinderjit Singh Sawaran SinghDepartment of Computer Engineering and Applications, GLA University, Mathura, (UP), IndiaMohamed ShabanDepartment of Physics, Faculty of Science, Islamic University of Madinah, Madinah 42351, Saudi Arabia

2026en

Ochiq kirishOltin

Litsenziya: cc-by

ABI

Annotatsiya

Hybrid nanofluids incorporating carbon nanotubes (CNTs) and metal oxide nanoparticles have emerged as advanced functional nanomaterials for thermal management because of their tunable thermophysical properties and synergistic nanoparticle interactions. However, the strong nonlinear dependence of their density, viscosity, thermal conductivity, and specific heat capacity on nanoparticle composition and concentration complicates accurate property prediction and materials design. In this study, an evolutionary machine learning framework is developed to establish composition–property relationships for aqueous CNT–metal oxide hybrid nanofluids. A multilayer perceptron neural network (MLPNN) coupled with a genetic algorithm (GA) is employed to optimize the network architecture and learning parameters and to predict four key thermophysical properties, namely density, viscosity, thermal conductivity, and specific heat capacity. The predictive performance of the evolutionary GA–MLPNN model is systematically benchmarked against Gaussian process regression (GPR) and classification and regression tree (CART) models. The results demonstrate that GA–MLPNN provides highly accurate predictions for all investigated properties, achieving correlation coefficients above 0.998 for density, viscosity, and thermal conductivity and reaching 0.9999 for specific heat capacity. The analysis further reveals distinct composition–property relationships: nanoparticle volume fraction is the principal factor controlling density, viscosity, and thermal conductivity, whereas the specific heat capacity is strongly associated with the thermal capacity contributions of the constituent phases. These results demonstrate that evolutionary learning can effectively capture the complex property landscape of CNT–metal oxide hybrid nanofluids and provide an efficient route for thermophysical property estimation without extensive additional experimentation. The proposed framework establishes a materials-oriented computational strategy for screening and designing hybrid nanofluids with tailored thermophysical properties for advanced thermal-management and energy-related applications.

Hali tarjima qilinmagan

Identifikatorlar

DOI: 10.1016/j.matdes.2026.117322

Iqtiboslar va manbalar

0 ta iqtibos65 ta foydalanilgan manba