Application of artificial neural networks for prediction modeling and analysis of the thermophysical properties of Fe₃O₄/water ferrofluid using experimental measurement

Abstract

Abstract The use of nanofluids has attracted significant attention due to their improved thermal performance compared with conventional heat transfer fluids. Because nanoparticles are on the nanometer scale, problems such as channel blockage and severe abrasion can be reduced. Among these, Ferrofluid containing Fe 3 O 4 nanoparticles is of particular interest to researchers due to its unique properties, including favorable magnetic properties and environmental friendliness. In this study, an artificial neural network model is developed to predict the thermal behavior of Fe 3 O 4 /Water Ferro-nanofluid, and the effective factors, such as volume fraction (φ = 0.05% to 4.0%) and temperature (T = 24 °C to 50 °C), on thermal conductivity and dynamic viscosity are investigated. The performance of the ANN is validated by comparing the obtained results. The results show that when the $$\:\varphi\:$$ changes from $$\:\varphi\:$$ = 0.05% to 4.0%, the dynamic viscosity increases from 0 cP to 0.77 cP. The results showed that thermal conductivity increased with increasing nanoparticle concentration, while dynamic viscosity decreased with increasing temperature. In contrast, viscosity increased at higher nanoparticle volume fractions due to stronger particle interactions within the base fluid. The ANN models showed reasonable overall performance in estimating thermal conductivity, with correlation coefficients of 0.98 and errors of 1.56 × 10 − 5 .