Study of surface tension of CO2+water and CO2+ethanol solutions from combined CPA and PC-SAFT EoSs with gradient theory and artificial neural network

The gradient theory of the interface was combined with the cubic plus association and perturbed chain statistical association fluid theory equations of state to describe the surface tension of (CO2+ethanol) and (CO2+water) systems. Two methods of phase equilibrium and two forms of influence parameters were applied to these systems. A novel influence parameter was also suggested for the gradient theory. The results of this study showed that the new proposed influence parameter results in the accuracy of the surface tension model. The lowest % AADs of surface tension were 2.37 and 6.02, for (CO2+ethanol) and (CO2+water) systems, respectively. Therefore, the accurate results of the surface tension were obtained for both systems. Then an artificial neural network model was developed to model the surface tension of the applied mixtures. The best results were obtained with 5 layers and 4 layers and using "trainlm" and "tansig" functions.