Experimental studies and ANN predictions on the thermal properties of TiO2-Ag hybrid nanofluids: Consideration of temperature, particle loading, ultrasonication and storage time

Abstract

The TiO2-Ag hybrid nanofluids were prepared by diluting high-concentration nanofluids and adding Span80 as surfactants. The stability of nanofluids was studied by taking photos of the nanofluids with storage time of 0–30 days. The thermal conductivity and dynamic viscosity were measured at different volume fraction, temperature and ultrasonication time. The results showed that compared with the base fluid, the maximum enhancement of thermal conductivity was 9.98%, and the maximum dynamic viscosity ratio was 2.78. And ultrasonication treatment is recommended to alleviate the sedimentation of nanofluids. Then, two artificial neural networks (ANNs) were established and the size of the networks was optimized according to the relative error to improve the ability to predict the thermal properties of nanofluids. The results show that the maximum relative error of the prediction of thermal conductivity and dynamic viscosity is 1.44% and 3.54%, respectively, which is beneficial to saving experiment time and cost.