Phase equilibrium modeling for binary systems containing CO2 using artificial neural networks

In this study, two mathematical models based on the Feed-Forward Back Propagation Artificial Neural Network (FFBP-ANN) are employed for the prediction of CO2 mole fraction in liquid (x(1)) and vapor (y(1)) phases in the Vapor Liquid Equilibrium (VLE) for fifty-six CO2-containing binary mixtures. 2104 data sets from the open literature have been used to construct the models. Furthermore, some new experimental data (not applied in ANN training) have been used to examine the reliability of the model. Predictions using ANN were compared with available literature data and the results confirm that there is a reasonable conformity between the predicted values and the experimental data. The average absolute deviation percent (AAD (%)) for ANN model I (x(1) prediction) and ANN model II (y(1) prediction) are 1.572 and 0.848 respectively. The study shows that the neural network model is a good alternative method for the estimation of phase equilibrium properties for this type of mixtures.