One of the ways to boost the carbon dioxide (CO2) capture performance of amine based solvents is to blend a primary or secondary amine with a tertiary amine. This has been shown to be true in the blending of monoethanolamine (MEA) with either methyldiethanolamine (MDEA) or diethylaminobutanol (DEAB), resulting in quaternary amine systems of MEA-MDEA-H2O-CO2 and MEA-DEAB-H2O-CO2, respectively. However, one of the challenges is ensuring that the blending ratio remains constant throughout the entire duration of the CO2 capture operation considering the relative volatilities of the components of the blended amine solvent. In this work, two quaternary amine systems: MEA-DEAB-H2O-CO2 and MEA-MDEA-H2O-CO2 were used as examples of formulated solvents to develop an on-line analytical technique. This technique was used to determine the composition of the formulated solvent being used to capture CO2 from a flue gas stream. The development of the on-line analytical system was based on their liquid phase thermo physical properties. To achieve this, thermo physical properties, namely, density (), viscosity (η), refractive index (nD), heat capacity (Cp), thermal conductivity (k), and thermal diffusivity (D) of single amine components, binary system made up of aqueous solutions of single amines, ternary systems made up of aqueous single amines loaded with CO2 or aqueous MEA-DEAB or MEA-MDEA, and quaternary system made up of MEA-MDEA-H2O-CO2 and MEA-DEAB-H2O-CO2 were first measured. Selection of these properties was based on the availability of instruments and fast on-line measurements. The measurements of the liquid systems’ thermo physical properties were then correlated with the corresponding systems’ compositions using well known correlations of Arago-Biot, Gladston-Dale for refractive index, Nissan Grunberg for viscosity, Weiland and non-additive equation for density, and Redlich-Kister for heat transport properties, as well as two artificial neural network (ANN) techniques, namely, BPNN and RBFNN. The properties were measured at temperatures ranging from 22 ºC to 60 ºC, using amines in the concentration range of 0.5-5 mol/L, molar mixing ratio of MEA/tertiary amine of 5/1, 5/1.25, 5/1.5, and 5/2 and CO2 loading from 0-0.6 mole CO2/ mole amine. The existing models were used to correlate experimental data of the amine systems. Also, the two artificial neural network techniques (BPNN and RBFNN) were trained and applied to correlate the experimental data, and then used to predict these thermo physical properties. The overall objective was to use the predictive power of ANN or correlation models to determine the composition of the liquid quaternary or more components system based on on-line measurements of their thermo physical properties. Results obtained from both BPNN and RBFNN showed excellent correlation between the thermo physical properties as well as their ability to determine the composition with high accuracy, very low deviations (less than 1 %) and faster analytical time compared to existing correlations (such as Redlich-Kister, Weiland, and etc.) and conventional analytical methods (such as HPLC and titration). The ANN methods were then used to design an analytical system unit for determining the composition of any formulated amine solvent, which could include degradation products as well as corrosion inhibitors, degradation inhibitors and other additives. This design pinpoints the most logical spots to locate the unit in a typical CO2 capture plant. It also illustrates the working principle of the unit from measurement of physical properties and using these properties to quickly determine the composition of the amine system at that instant.