Examining the experimental effects of concentration and temperature on the viscosity of nanofluid containing graphene oxide, suggesting a correlation, and developing a neural network.

The application of nanotechnology represents a recent initiative to explore thermal sciences across various sectors. With substantial investments from developed nations in nanotechnology, extensive research has been conducted on the thermophysical properties of nanofluids, including viscosity and thermal conductivity. The present study measured the viscosity of Graphene oxide/water nanofluids at concentrations ranging from 5 to 8 g/L and temperatures between 20 °C and 35 °C. The objective was to establish an empirical relationship using Origin software to encompass all data with minimal error and a maximum correlation coefficient. The findings revealed that as the concentration of nanoparticles in the base fluid increased, the viscosity correspondingly rose. Conversely, at elevated temperatures, viscosity decreased. Furthermore, the viscosity of the Graphene oxide nanofluid was modeled using a Perceptron Artificial Neural Network with a 2-7-1 architecture, with concentration and temperature as the two inputs and viscosity as the output. The results for the proposed empirical relationship and the artificial neural network demonstrated that the mean squared error, mean absolute error, root mean squared error, and correlation coefficient were 0.002188, 0.0172564, 0.04677, and 0.98704, respectively, indicating a successful evaluation. The maximum positive and negative errors for the acquired data were 0.0375% and 0.068%, respectively. The values obtained were − 7.71% < Margin of Deviation (Correlation) < + 7.5% and − 9.55% < Margin of Deviation (Artificial Neural Network) < + 7.7%.