A novel statistical approach for prediction of thermal conductivity of CO[formula omitted] by Response Surface Methodology

Abstract

In the present article, a novel statistical approach was developed to accurately predict the thermal conductivity of CO2 using response surface methodology (RSM). The Artificial Neural Network (ANN) was also used for the modeling of the thermal conductivity of CO2. To develop the proposed models, 1910 experimental data were used including pressure, temperature and density of carbon dioxide as the input variables in the models. The performance of the models was fortified using statistical analysis and the values of 0.57, 1.25, 0.99993 and 0.99931 were obtained for AARD% and R2 of ANN and RSM models, respectively. The obtained results were compared with four conventional models to investigate the ANN and RSM accuracy. The results showed that the developed models were useful to predict CO2 thermal conductivity at wide ranges of temperature and pressure. It was found that the developed ANN model gives the best fit and satisfactory agreement with the experimental data. Also the proposed correlation presents higher accuracy compared with all previous correlations for prediction of CO2 thermal conductivity at different condition.