Neural Network to Separate Carbon Dioxide from different Gas Mixtures Using Semi-Clathrate Hydrates in the Presence of Promoters

Excessive carbon dioxide (CO2) emission from various systems such as biogas (CH4+CO2), fuel gas (H2+CO2), and flue gas (N2+CO2) is one of the main reasons of global warming and environmental problems. In recent years, special attention devoted to the hydrate based gas separation (HBGS) processes. This study attempts to propose a comprehensive neural network intelligent model, to predict hydrate formation conditions in various systems and in the presence of different promoters of quaternary ammonium and Phosphonium salts to separate CO2 from various systems. Finally, experimental data compared with predicted data, which neural network model has hydrate conditions prediction capability with acceptable accuracy (R2~0.98). Other error analysis results for network training dataset (MSE=0.24 and MEAE%=7.19), indicate the acceptable performance of the proposed model.