Assessment of Thermal Conductivity and Viscosity of Alumina-Based Engine Coolant Nanofluids using Random Forest Approach

Abstract

Thermal conductivity and viscosity are crucial thermophysical properties of nanofluids. They play a pivotal role in industries involved with heat transfer applications. Alumina (Al2O3) nanoparticles are known to be a good additive for thermophysical properties enhancement with favorable results in countless research. However, the measurement of thermophysical properties of nanofluids through experimental is expensive. Therefore, the random forest (RF), an advanced computational intelligence approach, is proposed to correctly predict the thermal conductivity and viscosity of alumina-based engine coolant nanofluids in this research. Experimental data from previous literature are utilized as input parameters for the development of the random forest models. The input parameters for the prediction of thermal conductivity are temperature and concentration, whereas the input parameters for the prediction of viscosity are temperature, concentration, and shear rate. Error metrics consisting of R-squared (R2), Adjusted R-squared (A-R2), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) are used to analyze and determine the performance of each random forest model. Based on the results, it is observed that all random forest models exhibit significantly high and consistent predictive accuracy with R2 of 0.9877 for thermal conductivity prediction and R2 of 0.9974 for viscosity prediction. The predictive accuracy of the random forest models can be enhanced by training it with multiple datasets which include several thermophysical properties of diversified nanofluids.