Comparison of two soft computing approaches for predicting CO2 solubility in aqueous solution of piperazine. (October 2016)

• Record Type: Journal Article
• Authors: Tatar, Afshin; Barati-Harooni, Ali; Najafi-Marghmaleki, Adel; Mohebbi, Armin; Ghiasi, Mohammad M.; Mohammadi, Amir H.; Hajinezhad, Ahmad
• Is Part Of: International journal of greenhouse gas control. Volume 53(2016:Oct.)
• Page Start: 85
• Page End: 97
• Publication Date: 2016-10
• Subjects: CO2 capture -- Piperazine (PZ) -- Model -- ANFIS -- LSSVM -- Prediction; Greenhouse gases -- Environmental aspects -- Periodicals; Air -- Purification -- Technological innovations -- Periodicals; Gaz à effet de serre -- Périodiques; Gaz à effet de serre -- Réduction -- Périodiques; Air -- Purification -- Technological innovations; Greenhouse gases -- Environmental aspects; Periodicals; 363.73874605
• DOI: 10.1016/j.ijggc.2016.07.037
• Languages: English
• ISSNs: 1750-5836

Abstract

Highlights: Two models have been developed for prediction of CO2 solubility in piperazine, PZ aqueous solution. The models are based on CHPSO-ANFIS and CSA-LSSVM approaches. The experimental data gathered from the literature were used for model development. The effectiveness and accuracy of the models were evaluated by statistical and graphical approaches. CHPSO-ANFIS model yields more accurate predictions compared to CSA-LSSVM model. Abstract: Removal of sour gases such as carbon dioxide (CO2 ) and hydrogen sulfide (H2 S) is of great importance in various processes such as streams of sour natural gas, production of ammonia by treating synthesis gas, and hydrogen purification processes. Hence, developing accurate tools for prediction of solubility of CO2 in aqueous solutions of different substrates seems to be of great importance. This study aims to develop two reliable models namely CHPSO-ANFIS and CSA-LSSVM for prediction of CO2 solubility in aqueous solution of mixture of 1, 4-diazacyclohexane (piperazine, PZ). The experimental data were gathered from several published works in the literature. The effectiveness and accuracy of the proposed models were evaluated by both statistical and graphical approaches. Results show that the developed CHPSO-ANFIS model presents more accurate and reliable predictions compared to the outcomes of CSA-LSSVM model. The overall R 2 and AARD% values for CHPSO-ANFIS model were 0.9902 and 3.76 which indicate accuracy and robustness of the developed model.