Evolving LSSVM and ELM models to predict solubility of non-hydrocarbon gases in aqueous electrolyte systems

•ELM and LSSVM models are used for solubility estimation of non-hydrocarbon gases.•Sensitivity analysis and outlier detection are applied.•ELM indicates better capability for solubility estimation. Due to lack of a comprehensive and accurate predictive model for estimation of non-hydrocarbon (N2 and CO2) solubility in the aqueous system, the main aim of the present study is the suggestion of new estimation tools for non-hydrocarbons solubility in the different aqueous solutions. To this end, two computational-based models, including extreme learning machine (ELM) and least-squares support vector machine (LSSVM), have been implemented. These models have excellent backgrounds in the estimation of behaviors of fluids. The non-hydrocarbon solubility values of LSSVM and ELM algorithms have been compared with this dataset in visual and mathematical methods. Furthermore, the sensitivity analysis has been employed to identify the amount of impacts of aforementioned parameters on solubility of non-hydrocarbons. These reliable investigations can help researchers to successfully estimate the main thermodynamic parameters which have important roles in optimization of design of industrial plants such as natural gas processing units.