Connectionist technique estimates H2S solubility in ionic liquids through a low parameter approach

• Evolving low parameter and easy-to-use model for estimation H2S solubility in various ionic liquids.
• Extensive H2S solubility in various ionic liquids data banks has been used.
• Statistical analysis was performed to the results gained from GA-LSSVM approach.

Adequate knowledge of solubility of acid gases in ionic liquids (ILs) at different thermodynamic conditions is of great importance in the context of gas processing and carbon sequestration. Thus, a precise estimation of this key parameter seems inevitable in the design prospective of IL-based separation processes. This paper introduces another interesting application of least square support vector machine (LSSVM) to forecast hydrogen sulfide (H2S) solubility in various ILs. Genetic algorithm (GA) is also employed to obtain optimal magnitudes of hyper parameters (including γ and σ2) which are embedded in the LSSVM technique. Utilizing 465 data samples (e.g., where 11 ionic liquids are included), the new strategy presented in this study demonstrates great predictive performance so that the coefficient of determination (R2) and mean squared error (MSE) are determined to 0.997594 and 6.6507E−05, respectively. Provided accurate solubility, such a competent tool has high potential to be combined with existing PVT and chemical engineering software packages for the proper design of process equipment in gas sweetening operations.

Figure optionsDownload as PowerPoint slide