Modeling of thermal conductivity of MWCNT-SiO2 (30:70%)/EG hybrid nanofluid, sensitivity analyzing and cost performance for industrial applications

• Published in: Journal of thermal analysis and calorimetry 2018-01, Vol.131 (2), p.1437-1447
• Main Authors: Esfe, Mohammad Hemmat, Esfandeh, Saeed, Mousa Rejvani
• Format: Article
• Language: English
• Subjects: Concentration (composition); Correlation analysis; Cost analysis; Ethylene glycol; Heat conductivity; Heat transfer; Industrial applications; Mathematical models; Modelling; Multi wall carbon nanotubes; Nanofluids; Neural networks; Sensitivity analysis; Silicon dioxide; Thermal conductivity
• ISSN: 1388-6150; 1588-2926
• DOI: 10.1007/s10973-017-6680-y

In the present study, the thermal conductivity of SiO2-MWCNT/EG hybrid nanofluid has been investigated experimentally at solid volume fraction range from 0.025 to 0.86% and temperatures range from 30 to 50 °C. SiO2 particles and multi wall carbon nanotubes (MWCNTs) dispersed with the ratio of 70:30% by mass in ethylene glycol (EG) as the base fluid. The thermal conductivity ratio of mentioned hybrid nanofluid increased to 20.1% more than EG thermal conductivity at 50 °C and the solid volume fraction of 0.86%. Also in the present study, a new correlation was proposed to predict experimental TCR (thermal conductivity ratio) based on the solid volume fraction and the temperature. The R-squared for the proposed correlation is equal to 0.9864. The sensitivity of nanofluid’s thermal conductivity was increased with temperature and solid volume fraction increasing. Also, an ANN was designed for TCR data modeling and forecasting. The most optimal topology was an ANN contains two hidden layers and four neurons in each hidden layer. The R-squared, MSE, and AARD for proposed ANN are equal to 0.9989, 6.8344e−06, and 0.0105, respectively. The results indicated that the neural network is stronger than the correlation in the estimating and predicting experimental thermal conductivity ratio.