Prediction of thermal conductivity of various nanofluids using artificial neural network

This paper presents a 5-input artificial neural network (ANN) model for the prediction of the thermal conductivity ratio of nanofluids to the base fluid (knf/kf) of various nanofluids based on water and ethylene glycol (EG) and a type of transformer oil. The studied nanofluids are Al2O3–Water, Al–Water, TiO2–Water, Cu–Water, Cuo–Water, ZrO2–Water, Al2O3–EG, Al–EG, Cu–EG, Cuo–EG, Mg(OH)2–EG, Al2O3–Oil, Al–Oil, Cuo–Oil and Cu–Oil (15 nanofluids). The network is designed and trained using a total of 776 experimental data points collected from 21 sources of experimental data available in the literature. Average diameter, volume fraction, thermal conductivity of nanoparticles and temperature as well as some appropriated numbers for both nanoparticle and base fluid are chosen as input variables of the network, whereas the corresponding value of (knf/kf) is selected as its target. The developed optimal ANN model shows a reasonable agreement in predicting experimental data with mean absolute percent error of 1.26% and 1.44% and correlation coefficient of 0.995 and 0.993 for training and testing data sets, respectively.