The evaluation of factors affecting the thermophysical properties of nanofluids via principal component analysis

Traditional heat transfer fluids exhibit relatively inefficient thermophysical properties, including density, viscosity and thermal conductivity. However, due to the fact that metals possess significantly higher thermophysical properties, the utilization of nanofluids in heat transfer applications has become a widely studied approach. It is evident that the incorporation of nanoparticles into conventional heat transfer fluids has been shown to result in enhanced thermophysical properties which varies with the predominant factors of the types of nanoparticle and base fluid employed, nanoparticle concentration, and temperature. Despite a substantial number of studies that have examined the effects of these factors on variations in thermophysical properties, inconsistency in the results can be observed, particularly due to the wide ranges of nanoparticle and base fluid types, various concentrations and especially the difference in measurement systems. Therefore, the objective of this study was to synthesize copper oxide and iron oxide nanoparticles of comparable sizes and suspend them in base fluids comprising ethylene glycol and water in varying volume fractions. Subsequent analysis involved the measurement of the relevant thermophysical properties as a function of concentration, temperature and volume fraction, and these measurements along with the data presented in the literature were then subjected to PCA analysis for determining the correlations between the factors. Furthermore, empirical prediction models were investigated with the specified variables. The results demonstrate that colloidally stable CuO and Fe3O4 nanofluids, comprising a mixture of ethylene glycol and water as the base fluid, have been successfully obtained. Furthermore, it was determined that an increase in nanoparticle concentration resulted in enhanced viscosity and density, while thermal conductivity exhibited deterioration. The PCA demonstrated that all of the thermophysical property variations were highly affected by nanoparticle concentration and subject to variable responses to temperature changes. Finally, the application of non-linear regression to the measurement data yielded the observation that empirical prediction models can be utilized for a particular type of nanofluid. Nevertheless, the accuracy of these models is subject to deterioration when a substantial amount of data from diverse nanofluids is incorporated.