Enhanced Thermophysical Properties of 40:60 % Ethylene Glycol and Water Mixture Based MgO + Fe3O4 Hybrid Nanofluids: An Experimental and ANN Predictions

Abstract

The growing demand for efficient heat transfer and energy conversion systems require advanced working fluids with superior thermal transport properties. In this work, MgO, and Fe₃O₄ nanoparticles were synthesized by using chemical coprecipitation method. Then, 40:60 % ethylene glycol and water mixture based MgO + Fe₃O₄ hybrid nanofluids were produced by the two-step method. Three different compositions like 1:2, 1:1, and 2:1 of MgO and Fe₃O₄ were used in this study. The hybrid nanofluids of 0.2–1.0 vol % were prepared, and thermal conductivity, viscosity, density, and specific heat were measured over temperatures ranging from 20 °C to 60 °C. Sodium dodecylbenzene sulfonate surfactant was used to enhance the stability of the hybrid nanofluids. Transmission electron microscopy, and X-ray diffraction techniques were used for the morphological characterization of the nanoparticles. Results show that thermal conductivity increases with increase of nanoparticles concentration as well as temperatures, whereas the viscosity decreases with temperature, but increases with nanoparticle concentration. Moreover, the density increases with nanoparticles concentration but decreased with temperatures, and the specific heat decreases with increase of nanoparticles, but increases with temperature. The same trend is observed for the three nanoparticles composition. At 1.0 vol % and 60 °C, thermal conductivity improved by 22.09 %, 25.12 %, and 29.07 % relative to base fluid, whereas, at 1.0 %, and 20 °C the viscosity improved by 51.3 %, 53.9 %, and 56.8 %, relative to base fluid at nanoparticles composition of 1:2, 1:1, and 2:1 in the base fluid. The useful empirical correlations were developed for all the prepared hybrid nanofluids for thermal conductivity, viscosity, density, and specific heat, which demonstrates the high predictive accuracy, with R² values between 0.942 and 0.993. Artificial Neural Network (ANN)- Levenberg–Marquardt (LM) algorithm were used to train the experimental datasets to model nonlinear interactions between temperature, concentration, and thermophysical properties. Among all the datasets, the particle composition of 1:2 based hybrid nanofluids yielded the highest accuracy coupled with the lowest error values of the studied properties, e.g. thermal conductivity, MSE = 2.5036e⁻⁵, R = 0.99768; viscosity, MSE = 0.0035, R = 0.99925; density, MSE = 9.8186, R = 0.99142; specific heat, MSE = 28.941, R = 0.99783, respectively. Overall, the combined experimental and intelligent-modelling framework provides robust predictive capability for optimizing MgO + Fe₃O₄ hybrid nanofluids.