Study of capabilities of the ANN and RSM models to predict the thermal conductivity of nanofluids containing SiO2 nanoparticles.

In this paper, thermal conductivity prediction of nanofluids is discussed by the RSM and ANN models. The nanofluids contain SiO2 nanoparticles, and their thermal conductivity is measured in the temperature range of 30–60 °C. The effect of SiO2 nanoparticles on thermal conductivity depends on the type of base fluid. For the ethylene glycol (EG) base fluid, SiO2 nanoparticles improve the thermal conductivity by 12%. For glycerol base fluid, thermal conductivity is increased by 6%. The thermal conductivity of both nanofluids depends on temperature and volume fraction so that as the volume fraction and temperature increase, their positive effect on thermal conductivity is enhanced. ANN and RSM models are used to estimate the thermal conductivity ratio of nanofluid to base fluid, i.e., TCR = k nf k bf . Both techniques are well able to predict the amount of TCR for both nanofluids. A cubic function with R 2 = 0.997 and 0.994 is proposed for SiO2-EG and SiO2-G nanofluids, respectively. It is found by trial and error that the neural network with 8 neurons is suitable for both nanofluids. For Ann statistical calculations demonstrate that the R 2 for SiO2-EG nanofluid is 0.995 and for SiO2-G nanofluid is 0.994. The error of both methods is less than 0.3%, which indicates that both methods can well estimate TCR value. [ABSTRACT FROM AUTHOR]