Case Studies in Chemical and Environmental Engineering (Jun 2023)
Solubility prediction of CO2 in ionic liquids under subcritical and supercritical carbon dioxide conditions by the adaptive neuro-fuzzy inference system
Abstract
In this study, an adaptive neuro-fuzzy inference system (ANFIS) was employed to check the solubility of carbon dioxide in 26 different ionic liquids with different ionic structures under subcritical and supercritical carbon dioxide conditions during a wide range of pressure (0.34–14596 kPa) and temperature (278.1–368.4 K). Two models were introduced: Model#1 with a critical temperature (Tc), melting temperature (Tm) and molecular weight (Mw) of ionic liquids as input variables, and Model#2 with a critical temperature (Tc), critical pressure (Pc), and acentric factor (ω) of ionic liquids as input variables. The results showed that the proposed models could predict the behavior of laboratory data for both training and test data, mainly in critical conditions. The statistical factor, R2, for Model#1 was 0.95, which was a little less than that of Model#2, 0.9950. It confirmed that although both mathematical models have high accuracy, Model#2 predicts the solubility of carbon dioxide in ionic liquids better than Model#1. The R2 value for the test data in Model#2 was equal to 0.9895, indicating that the obtained model can be generalized to all ionic liquids-carbon dioxide systems. The sensitivity analysis illustrated that the pressure of the systems had the greatest effect on the solubility prediction of carbon dioxide in ionic liquids under subcritical and supercritical carbon dioxide conditions, causing the solubility to increase. Against the molecular weight, increasing the acentric factor reduced the solubility of carbon dioxide in ionic liquids. The effect of temperature on the carbon dioxide solubility compared to the other variables was negligible.
