Experimental examination of the properties of Fe3O4/water nanofluid, and an estimation of a correlation using an artificial neural network

Currently, considering that the addition of solid particles to normal fluids directs to a considerable increase in thermal conductivity, the utilization of nanofluids rather than typical liquids is extremely cus-tomary. This paper provides an experimental evaluation of the viscosity and thermal conductivity of Fe3O4/water nanofluid and provides an estimation model using an artificial neural network (ANN) and margin of deviation. For this purpose, Fe3O4 nanoparticles were synthesized by the co-precipitation method, and the size of nanoparticles was investigated by the XRD test.Also, the Fe3O4/water nanofluid was prepared in a two-step method and the stability of the nanofluid was evaluated by the zeta potential test.Then, thermal conductivity and viscosity of nanofluid were measured at temperatures of 25, 30, 35, 40, 45, and 50 degrees Celsius and volume fractions of 0.1, 0.2, and 0.3%. The outcomes of this experiment indi-cated that the volume fraction of nanoparticles and temperature have an immediate relationship with the thermal conductivity coefficient of nanofluids because, with the increase of these two parameters, the thermal conductivity coefficient increases and vice versa. Also, the viscosity of the nanofluid decreases with the decrease in volume fraction of nanoparticles and increases with the decrease in temperature. Because of the absence of an exact and proper correlation to predict the viscosity and thermal conductiv-ity of this nanofluid, two correlations are presented as a subordinate of temperature and volume fraction. Also, a analogy was made among the results of the experiment and the output of the proposed relation-ship and it was found that the maximum margin of deflection for thermal conductivity is equal to 0.42% and for viscosity is 0.06%. Toward the end of this research, the laboratory results were analyzed by neural networks. The designed neural network was compared with the values obtained from the suggested rela-tionship and the outcomes indicated that in predicting the viscosity and thermal conductivity of nanoflu-ids, the neural network has fewer blunders and higher accuracy.(c) 2022 Elsevier B.V. All rights reserved.