Advancing nanofluid technology: Experimental thermal conductivity measurements and accurate predictive models for yttrium oxide-water nanofluid using neural networks and mathematical correlations

Abstract

This work involved the experimental measurement of thermal conductivity (TC) values for a novel nanofluid. This nanofluid was created by incorporating yttrium oxide nanoparticles into pure water, which served as the base fluid. The TC values of the nanofluid consisting of yttrium oxide and water were empirically determined at five distinct concentrations. These measurements were conducted within the temperature range of 10–65 degrees Celsius. The data obtained from the experiment demonstrated a 1.47% enhancement in the TC of the nanofluid consisting of yttrium oxide and water, in comparison to the base fluid. Experimental results were compared with correlations commonly used in the literature. A multilayer neural network model and a novel mathematical correlation were built utilizing a dataset of 60 experimental observations to ascertain the relationship between temperature, concentration, and the TC of yttrium oxide-water nanofluid. Nine data sets were used for validation of the network model, and in addition, the model was validated with various performance parameters. The values derived from the network model and mathematical correlation were juxtaposed with both the empirical findings and the outcomes documented in existing literature. The R value for the network model was calculated as 0.99980 and the MSE value was 5.89E-07. The average deviation rate for the network model was determined to be −0.0007%, whereas the mathematical correlation yielded a value of 0.049%. The study’s results demonstrated the development of a neural network model and a novel mathematical connection for accurately calculating the TC of yttrium oxide-water nanofluid.