Viscosity correlations of MEA and water mixtures

“The nature of the model depends on the solution characteristics. Generally, the liquid viscosity decreases with the increase of temperature, and it increases with the increase of pressure. For pure MEA, an exponential model was frequently used to correlate the temperature dependence of viscosity. Table 13 summarizes the various published correlations for the viscosity of different MEA solutions. The relation between the viscosity of pure MEA with temperature can be represented by the Arrhenius equation shown in equation (30) and Teng et al. [59] calculated the activation energy for viscous flow from the data presented in DiGuilio et al. [9]. DiGuilio et al. [9] used a modified Andrade from (1934) viscosity model [60] by Vogel [61] as shown in equation (31).”

Unlike ideal density, several mathematical relations have been proposed to determine ideal viscosity in a liquid mixture in the literature.

Kendall and Monroe [62]:

Bingham [63]:

Cronauer et al. [64] for ideal kinematic viscosity:

And the following expression is frequently used in recent publications [65].

The viscosity of aqueous MEA deviates from the ideal mixture viscosity. This deviation of excess viscosity has been studied to make correlations fit the measured viscosity of the mixture. Accordingly, correlations are built to fit the viscosity deviation that is the viscosity difference between a real solution and an ideal solution.

McAllister presented a model to calculate kinematic viscosity in a binary mixture [26667]. It is a semiempirical model, which is based on Eyring’s absolute rate theory [68]. This model is given in two forms as the McAllister Three-Body Model and Four-Body Model considering different intermolecular interactions with neighbouring molecules. Lee and Lin [14] and Amundsen et al. [2] adopted the Three-Body model as shown in equation (32) to fit viscosity data of aqueous MEA at different temperatures. Arachchige et al. [33] used a correlation suggested by Teng et al. [59] given in equation (33) to correlate measured viscosity of aqueous MEA at different temperatures. This correlation uses the viscosity of pure water and a polynomial to fit the viscosity of the binary mixture. The polynomial coefficients were found through a regression analysis at different temperatures. A Redlich–Kister type correlation as illustrated in equation (34) was proposed by Islam et al. [17] to determine  (excess viscosity), and the parameters were found for different temperatures through a regression. A similar work was performed by Hartono et al. [1]. Then a Redlich–Kister type model was proposed to fit  given in equation (35). The main advantage of Hartono’s aqueous MEA viscosity correlation is that it comprises the temperature dependence of viscosity that is not considered in Islam’s correlation. Idris et al. [34] discussed the applicability of correlations based on the work by Heric-Brewer [69], Jouyban-Acree [70], Herráez et al. [71], and Redlich-Kister [57] as given in equations (36)–(39) respectively. The fitting parameters are in the form of a second-order polynomial of temperature to correlate temperature dependency of the viscosity as given in equation (40).

Limited attempts have been made to build correlations for the viscosity data of CO2-loaded aqueous MEA mixtures. Accordingly, more measurements are still required to validate the existing data and correlations. Weiland et al. [26] developed a correlation for CO2-loaded aqueous MEA for viscosity under different CO2 loadings, MEA concentrations, and temperatures as described by equation (41). It is applicable for viscosities up to 40 mass% of MEA aqueous solutions at CO2 loading of 0.6 mol CO2/mol MEA to a maximum temperature of 298.15 K. Amundsen et al. [2] adopted Weiland’s correlation to fit the measured viscosities at different amine concentrations, CO2 loadings, and temperatures. Hartono et al. [1] developed a correlation for different CO2 loadings and temperatures by making a relation between viscosities of CO2-loaded and unloaded aqueous MEA solutions as given in equation (42). The correlation was fit for CO2-loaded viscosities of 30 and 40 mass% MEA and claimed 3.9% maximum AARD. Idris et al. [34] adopted a modified Setschenow-type [72] correlation as shown in equations (44) and (45) to fit CO2-loaded aqueous MEA data at high MEA concentrations. This approach has been tested for the physical properties of amine solutions by Shokouhi et al. [7374]. A new approach was taken by Matin et al. [58] using Eyring’s absolute rate theory [68] as illustrated in equations (46)–(49). Therefore, assuming the equivalence between the Gibbs free energy of activation for viscous flow and the equilibrium Gibbs free energy of mixing, the concepts of classical thermodynamics can be extended to the viscous flow behaviour of liquid mixtures [65]. The electrolyte-NRTL model is used to calculate the excess Gibbs free energy. Having tested for different terms, Matin et al. [58] revealed that the Gibbs free energy of mixing is the appropriate thermodynamic quantity to substitute for the excess free energy of activation for viscous flow for CO2-loaded aqueous MEA mixture. The absolute rate theory with a reliable thermodynamic model is applicable for viscosity estimation of strong electrolyte systems, such as CO2-loaded alkanolamine solutions.

The proposed correlations for viscosity of aqueous MEA were examined for accuracies compared to literature viscosity data. Table 14 lists the calculated AARD and maximum deviation for the McAllister model based on fitted parameters by Amundsen et al. [2] and Hartono’s correlation for the considered three data sources. It was observed that the AARD for viscosity correlations are greater than the AARD for density correlations for aqueous MEA. For Amundsen’s correlation, the highest AARD of 5.66% was reported for data presented by Ma et al. [27] and a maximum deviation was observed as 0.871 MPa·s at  and T = 298.15 K for data given by Maham et al. [35]. Hartono’s correlation showed a highest AARD of 4.35% for the viscosity data presented by Ma et al. [27] and a maximum deviation of 0.854 MPa·s for data presented by Maham et al. [35] at  and T = 303.15 K.

Leave a Comment