10.1021/ef401815q.s001 Zhaomin Li Zhaomin Li Shuhua Wang Shuhua Wang Songyan Li Songyan Li Wei Liu Wei Liu Binfei Li Binfei Li Qi-Chao Lv Qi-Chao Lv Accurate Determination of the CO<sub>2</sub>–Brine Interfacial Tension Using Graphical Alternating Conditional Expectation American Chemical Society 2015 monovalent cation molalities formation water salinity IFT CO 2 impurities ACE CH 4 mole fraction Spearman correlation coefficients SD bivalent cation molalities reservoir pressure CO 2 streams rank correlation coefficients AARE 2015-12-17 00:14:32 Dataset https://acs.figshare.com/articles/dataset/Accurate_Determination_of_the_CO_sub_2_sub_Brine_Interfacial_Tension_Using_Graphical_Alternating_Conditional_Expectation/2027661 A newly developed CO<sub>2</sub>–brine interfacial tension (IFT) correlation based on the alternating condition expectation (ACE) algorithm has been successfully proposed to more accurately estimate the CO<sub>2</sub>–brine IFT for a wide range of reservoir pressure, temperature, formation water salinity and injected gas composition. The new CO<sub>2</sub>–brine correlation is expressed as a function of reservoir pressure, temperature, monovalent cation molalities (Na<sup>+</sup> and K<sup>+</sup>), bivalent cation molalities (Ca<sup>2+</sup> and Mg<sup>2+</sup>), N<sub>2</sub> mole fraction and CH<sub>4</sub> mole fraction in injected gas. This prediction model is originated from a CO<sub>2</sub>–brine IFT database from the literature that covers 1609 CO<sub>2</sub>–brine IFT data for pure and impure CO<sub>2</sub> streams. To test the validity and accuracy of the developed CO<sub>2</sub>–brine IFT model, the entire dataset was divided into two groups: a training database consisting of 805 points and a testing dataset consisting of 804 points, which was arbitrarily selected from the total database. To further examine its predicted capacity, the new CO<sub>2</sub>–brine IFT correlation is validated with four commonly used pure CO<sub>2</sub>–pure water IFT correlations in the literature, it is found that the new CO<sub>2</sub>–brine IFT correlation provides the comprehensive and accurate reproduction of the literature pure CO<sub>2</sub>–pure water IFT data with an average absolute relative error (% AARE) of 12.45% and standard deviation (% SD) of 18.57%, respectively. In addition, the newly developed CO<sub>2</sub>–brine IFT correlation results in the accurate prediction of the CO<sub>2</sub>–brine IFT with a % AARE of 10.19% and % SD of 13.16%, respectively, compared to two CO<sub>2</sub>–brine IFT correlations. Furthermore, sensitivity analysis was performed based on the Spearman correlation coefficients (rank correlation coefficients). The major factor influenced on the CO<sub>2</sub>–brine IFT is reservoir pressure, which has a major negative impact on the CO<sub>2</sub>–brine IFT. In contrast, the effects of CO<sub>2</sub> impurities and salt components in the water on the CO<sub>2</sub>–brine IFT are in the following order in terms of their positive impact: bivalent cation molalities (Ca<sup>2+</sup> and Mg<sup>2+</sup>), CH<sub>4</sub>, N<sub>2</sub>, and monovalent cation molalities (Na<sup>+</sup> and K<sup>+</sup>).