Comparison of CO2 solubility between Neqsim and CMG_WinProp and Spycher-Pruess Model

[nwpro9]

Title: CO₂ Solubility in NaCl Solution Significantly Deviates from CMG_WinProp and Spycher-Pruess Models

Description:

I'm excited to explore Neqsim for CCS fluid modeling due to its flexibility. This issue pertains to the following notebook:

📄 notebooks/thermodynamics/solubility_of_gases_in_water.ipynb

Context

I'm comparing CO₂ solubility in the aqueous phase across three tools:

1. CMG_WinProp
2. Spycher-Pruess model
3. Neqsim

Observations

CO₂-H₂O System (Pure Water):
Neqsim matches very well with CMG_WinProp and Spycher-Pruess in terms of CO₂ mole fraction vs. pressure at a fixed temperature.

CO₂-H₂O-NaCl System (Salt Added):
When Na⁺ and Cl⁻ are introduced, Neqsim's CO₂ solubility results deviate significantly—by 2 orders of magnitude—from the other two tools. CMG_WinProp and Spycher-Pruess remain consistent with each other.

Troubleshooting Attempts

I tested various configurations in Neqsim:

Equations of State (EoS):
-Electrolyte-CPA-EoS
-Electrolyte-ScRK-EoS
-CPA
-SRK
-PR

Ion Concentration Variations:
-Adjusted mole/mass values for Na⁺ and Cl⁻ based on salinity used in CMG_WinProp and Spycher-Pruess
-Tried back-calculating mole values to match the solubility results, but had to reduce the values in Neqsim artificially by a factor of ~5, which seems physically inconsistent

Questions

1. Are there known limitations in Neqsim when modeling CO₂ solubility in NaCl solutions? The sample jupyter above uses methane. If so, is there anyway to correct any deviations for CO2?
2. In the sample notebook, Na⁺ and Cl⁻ are set to 10 mg/kg each. Is this example fluid equivalent to a 0.02 weight fraction NaCl solution in Neqsim's salinity/unit conversion?

Any guidance or clarification would be greatly appreciated!

[EvenSol]

Hi!

Thanks for reporting this issue. I got help from @Sviatose to update the notebook.

1. Are there known limitations in Neqsim when modeling CO₂ solubility in NaCl solutions? The sample jupyter above uses methane. If so, is there anyway to correct any deviations for CO2?

Please see the updated examples with comparison to experimental data. The fit to experimental data seems ok (even though electrolyte CPA and electrolyte ScRk does not give perfect match).

2. In the sample notebook, Na⁺ and Cl⁻ are set to 10 mg/kg each. Is this example fluid equivalent to a 0.02 weight fraction NaCl solution in Neqsim's salinity/unit conversion?

The example was just for illustration purpose of how to use the model. As you point out the is not reprecentable for real NaCl solution. I added a 4.3 salinity example with correct individual weight of Na and Cl.

[nwpro9]

Thank you for updating the Jupyter notebook with more NaCl–CO₂ system examples — much appreciated!

Using the SW model did improve the match from what I had previously, which is great to see. I currently don't have any experimental data to compare directly. My fluid salinity is around 4.3 molal, which aligns with one of the available experimental data points. However, my system temperature is approximately 32 °C, which differs from the conditions in that experimental dataset.

The mismatch between the Neqsim-SW model and the experimental data in the example seems to occur at higher salinities and higher pressure ranges. Similarly, I’ve noticed that the mismatch between Neqsim-SW and both WinProp and Spycher–Pruess models that I ran also becomes more pronounced under those same conditions. The magnitude of mismatch between Neqsim-SW and WinProp is shown in the figure below. Any comments here why am I still seeing this amount of mismatch? I understand CO2 solubility is still pretty minimal after all at this P, T, and salinity conditions, but it would be great to note if this is caused by a correlation limitation (handling NaCl solution), I/O in the code (not that I found any), or else.

Thanks again for checking into this!

[Sviatose]

Hi, thanks for your observations. SW model uses original binary interaction parameters defined in Phase Behavior Monograph by Curtis H. Whitson and Michael R. Brule. However, some newer articles showed that the Soreide and Whitson model underestimates the solubility of carbon dioxide at high molalities, and they proposed a better correlation for SW parameters by refitting it to solubility data of CO2 in water and NaCl brine. We will see if we can review these parameters any time soon.

Ref. Salaheddine Chabab, Pascal Théveneau, Jérôme Corvisier, Christophe Coquelet, Patrice Paricaud, et al.. Thermodynamic study of the CO 2 -H 2 O -NaCl system: Measurements of CO 2 solubility and modeling of phase equilibria using Soreide and Whitson, electrolyte CPA and SIT models. In�ternational Journal of Greenhouse Gas Control, 2019, 91, pp.102825. ff10.1016/j.ijggc.2019.102825ff. ffhal-02310963f