Bayesian regularization optimization algorithm for the experimental thermophysical property for 80:20% water and ethylene glycol based ZrO2 nanofluids

Abstract

Abstract. In the current study, water and ethylene glycol (W/EG 80:20%) are used as the base fluid, and sodium dodecyl benzene sulfonate is used as a surfactant to create nanofluids using ZrO2 nanoparticles prepared using the sol-gel technique. For temperatures ranging from 20 oC to 60 oC and various volume loadings of nanoparticles, 0.2, 0.4, 0.6, 0.8, and 1.0%, respectively, the thermal conductivity, dynamic viscosity, density, and viscosity of these ZrO2 nanofluids are experimentally evaluated. Artificial neural network based Bayesian regularization algorithm was used to find the correlation coefficient R2 and root-mean square error. New correlations were also suggested for each of the thermophysical properties. Experiments show that temperatures and concentrations of nanoparticles have a significant impact on the thermophysical properties of nanofluids. In fact, it is shown that, at 20 oC and 60 oC, respectively, increasing the thermal conductivity of nanofluids by 1.0 vol% leads to increases of almost 10.16% and 24.53%. Additionally, at 1.0 vol and 20 oC to 60 oC, the dynamic viscosity is reduced from 61.94% to 50.79%. The correlations and outcomes of the developed artificial neural network are in perfect agreement with the experimental data.