Predicting Dynamic Viscosity in Nanofluids of Graphene Nanoplatelets and SAE10W Oil Utilizing Artificial Neural Networks

Abstract

Viscosity is an essential factor when selecting nanofluids, as it significantly impacts their thermal behavior and heat transfer characteristics. This study aims to develop a prediction model for the dynamic viscosity of damper oil (Grade: SAE10W) nanofluid containing graphene nanoplatelets (GnPs) using an artificial neural network (ANN) based on experimental data. With high precision, ANN accurately predicts the dynamic viscosity variations with nanoparticle volume concentration and temperature. The use of a network with one hidden layer and 10 neurons resulted in a regression coefficient of 0.9998, indicating high accuracy with a simple structure. Furthermore, a mathematical correlation derived using the curve fitting method resulted in a coefficient of determination value of 0.9990. These models were evaluated in terms of percentage error to determine their accuracy. The error range for the ANN model was between −0.89% and 0.66%, and for the mathematical correlation, it was between −6.74% and 5.27%. In comparison to the mathematical correlation, the ANN model predicts better the dynamic viscosity of GnPs-SAE10W oil nanofluids. Hence, this model has the potential in the development of applications related to heat transfer.