Estimating CH4 and CO2 solubilities in ionic liquids using computational intelligence approaches

In this work, we aimed to evaluate the high potential of different ionic liquids (ILs) in the field of gas separation. The gas solubility values of natural gas comprised of methane (CH4) and carbon dioxide (CO2) has been determined for 11 dissimilar ILs. To develop the predictive model, the mole fraction of CH4, the mole fraction of CO2, % CH4 in the feed, % CO2 in the feed, temperature, critical pressure (PC), acentric factor (ω) and critical temperature (TC) of ionic liquids were assumed as the inputs, while the bubble-point pressures of CH4 and CO2 were calculated as the desired output. Two intelligent methods namely CSA-LSSVM and PSO-ANFIS were proposed to correlate the inputs and outputs. Moreover, to assure the reliability of the models, several graphical and statistical diagrams were precisely plotted and analyzed. Based on the obtained results, it became obvious that the improved models had high accuracy and showed more precision compared to other intelligent algorithms in the prediction of actual gas solubility data. In fact, the LSSVM model demonstrated better performance and more precision compared to other developed models. The results of the current work confirmed that intelligent-based algorithms are influential and advantageous substitution for complex and time-consuming investigational procedures in the field of estimating gas solubility.