Experimental investigations and the modeling approach for CO2 solubility in aqueous blended amine systems of monoethanolamine, 2-amino-2-methyl-1-propanol, and 2-(butylamino)ethanol

Abstract

In this work, new CO₂ solubility data on three types of aqueous amine blends were reported to complement existing databases. The experiments were conducted at temperatures of 313 K (absorption condition) and 363 K (desorption condition). The effect of the MEA concentration on the CO₂ solubility in several amine blends at low CO₂ partial pressure (8 to 50.65 kPa) were studied in this work, including 0.1, 0.3, 0.5 mol/L MEA + 2 mol/L AMP; 0.1, 0.3, 0.5 mol/L MEA + 2 mol/L BEA; and 0.1, 0.3, 0.5 mol/L MEA + 1, 2 mol/L AMP + 1, 2 mol/L BEA. Besides, an additional group of equilibrium CO₂ solubility data were conducted at 298 K in order to estimate the heat of CO₂ absorption of the blended solvents at a temperature range from 298 to 313 K. A new simplified Kent-Eisenberg model was developed for the predictions of blended solvents, and a multilayer neural network model with Levenberg-Marquardt backpropagation algorithm was developed upon five hundred reliable published experimental data. The predictions from two methods are both in good agreement with the experimental CO₂ solubility data.

Keywords: Amine blends; CO2 capture; CO2 solubility; Kent-Eisenberg; Machine learning; Thermodynamics.