Modeling of CO2 solubility in aqueous amine solutions using hybrid neural network

Abstract

The use of chemical absorption with amine aqueous solutions has become of great interest as potential post-combustion CO2 removal process. In such processes, knowledge of solution equilibrium conditions is essential and is necessary to design CO2 treating equipment. Model of solubility of CO2 in N-methylidiethanolamine (MDEA) aqueous solution is presented. Model, based on well-known Kent-Eisenberg structure, was combined with neural network. Such combination forms hybrid neural network model. Neural network was used to determine amine protonation equilibrium constant and further was employed in hybrid neural network to predict equilibrium partial pressure of CO2 over MDEA aqueous solution in different temperatures and for various solution concentrations. Results show very good agreement between model and experimental data.