Correlation and prediction of surface tension in single and mixed aqueous electrolyte solutions based on the mean ionic activity coefficient: A comparative analysis of Pitzer, E-NRTL and E-UNIQUAC models

Abstract

Li and Lu's model to calculate the surface tension of single and mixed aqueous electrolyte solutions, coupled with Pitzer, e-NRTL and e-UNIQUAC activity coefficient models, is critically evaluated in this paper. Parameters of these activity coefficient models were taken from vapor-liquid equilibrium data available in the literature. The effect of the mean ionic activity coefficient model on the surface tension calculation is investigated in order to obtain low deviations from experimental data with emphasis on salt reservoir applications. An evaluation is made of the magnitude of the percentage surface tension increment relative to the surface tension of pure water, which occurs due to an increase in salt concentration. The feasibility of interpolation and extrapolation of the surface tension model with respect to the temperature was investigated for single electrolyte solutions at several temperatures. A sensitivity analysis of the initial guesses of the Li and Lu's model when adjusting the model parameters was performed. A new approach is also proposed to estimate the Pitzer parameters for some systems. This study explores the Li and Lu's model as never found in the literature before and is generally able to provide more accurate surface tension calculations compared with previous works. The results show that for single electrolyte solutions, the Pitzer and e-NRTL models provide similar deviations with an overall AAPD of 0.49%. For mixed electrolyte solutions, the e-NRTL model performed better than the Pitzer and e-UNIQUAC, providing an overall AAPD of 0.42%. The simulated increments of surface tension due to addition of salt are in accordance with the literature. The sensitivity analysis demonstrates the relevance of well adjusting the initial guesses for parameter estimation. Interpolation and extrapolation analysis yielded quite acceptable deviations with all overall AAPDs lower than 1.40% and also indicated that e-NRTL model is the most accurate. The proposed approach to fit the Pitzer model parameters provided an average deviation of 0.75%.