Artificial Neural Networks for Accurate Prediction of Physical Properties of Aqueous Quaternary Systems of Carbon Dioxide (CO2)-Loaded 4-(Diethylamino)-2-butanol and Methyldiethanolamine Blended with Monoethanolamine

作者：Pouryousefi Fatemeh; Idem Raphael^*; Supap Teeradet^*; Tontiwachwuthikul Paitoon

来源：Industrial & Engineering Chemistry Research, 2016, 55(44): 11614-11621.

DOI：10.1021/acs.iecr.6b03018

Physical and heat transport properties such as density, viscosity, refractive index, heat capacity, thermal conductivity, and thermal diffusivity of aqueous carbon dioxide (CO2)-loaded and unloaded 4-(diethyl amino)-2-buthanol (DEAB) and methyldiethanolamine (MDEA) as single amines and each blended with a primary amine (MEA) were measured at different ranges of temperature (25-60 degrees C), amine concentrations (0.5-2 M for tertiary amine and 5 M for primary amine), and CO2 loading (0-0.6 mol/mol amine). Results showed an increasing trend of CO2 loading on density, viscosity, and refractive index and a decreasing trend on the heat transport properties. Two artificial neural network techniques, back propagation neural network (BPNN) and radial basis neural network (RBFNN) as well as some well-known semi-empirical correlations from literature, were applied to correlate and predict these physical properties for two quaternary systems: MEA-i-DEAB+water+CO2 and MEA+MDEA +water+CO2. Results from the correlation showed that artificial neural network techniques gave the least deviation for the prediction of all physical properties of both amine systems with less than 1% AAD. The correlation coefficient between the experimental and predicted values in terms of R-2 value was in the range of 0.98-0.99.

• 出版日期2016-11-9