Modeling and Precise Prediction of Thermophysical Attributes of Water/EG Blend-Based CNT Nanofluids by NSGA-II Using ANN and RSM

Abstract

In this study, the thermophysical properties of water–ethylene glycol mixture-based CNT nanofluid (NF) including relative thermal conductivity (TC), viscosity, and specific heat were investigated. These properties were optimized using the NSGA-II method, and the effects of temperature and volume fraction (VF) parameters were investigated on these properties. RSM and MLP methods were used to model these properties. Some correlations are presented in terms of temperature and VF to predict these properties. Convergence coefficients for relative TC, viscosity, and specific capacity are R2 = 0.9893, 0.9665, and 0.9977, respectively. The results showed that the MLP model is better than RSM for this prediction. It was found that the rate of viscosity enhancement reaches up to 3.5 times the amount of base fluid with the addition of nanoparticles (NP). The amount of increase in TC is increased by 17% after the addition of NPs to the base fluid. It was also found that temperature and VF of NPs have a direct effect on relative TC and viscosity. But relative specific heat is only affected by the VF. Both ANN and RSM models predicted the relative specific heat well. But MLP was more accurate in predicting relative viscosity than RSM. Also, both models had a good performance in predicting relative TC in some points. The optimization results were presented with the NSGA-II algorithm for different conditions, and these points do not have priority over each other and all are optimal. It was also found that the VF parameter is much more important than the temperature parameter.