Modeling the thermal conductivity ratio of an antifreeze-based hybrid nanofluid containing graphene oxide and copper oxide for using in thermal systems

Abstract

According to laboratory data, the thermal conductivity ratio of an antifreeze-based hybrid nanofluid containing graphene oxide (GO) and Copper oxide (CuO) was modeled using mathematical methods, one is based on an artificial brain structure model, and the other is based on curve-fitting method. A two-variable empirical based correlation (R2 = 0.996) as a function of temperature and volume fraction suggested from the curve-fitting method. In the brain structure-based section, an artificial neural network employed by applying temperature and concentration as input variables and thermal conductivity ratio as the desired output. The correlation coefficient (R) values of designed ANN are 0.999963, 0.999409, and 0.999103 for train, validation and test, respectively. Mean squared errors (MSE) values of designed ANN are 1.01743e-6, 5.01019e-5, and 2.90237e-5 for train, validation and test, respectively. The findings indicated that the artificial neural network and the proposed correlation can predict the thermal conductivity ratio of GO-CuO (50:50%)/EG-Water (50:50%) hybrid nanofluid with high accuracy.

Keywords

• Correlation
• Artificial neural network
• Antifreeze
• GO-CuO
• Thermal conductivity
• Hybrid nanofluid