Specific Ion Interaction Theory (SIT) Database

[elg0086]

 

I am working with a series of 50 highly concentrated solutions in SpecE8 and an element of interest is Uranium (up to 80 g/L in solution). All 50 solutions were successfully speciated using thermo_minteq.tdat. Due to a range of "higher than normal" ionic strengths (0.2 Molal - 0.9 Molal) and differences in the aqueous species and solid phases associated with uranium, I am also working with the thermo_nea.tdat database. From the GWB Essentials guide, the Specific Ion Interaction Theory (SIT) activity model is "accurate to higher ionic strengths than the Debye-Hückel equations, but not to the salinities achievable using the Pitzer equations." Given the resulting ionic strengths, the SIT model appears to be an appropriate database to use. Unfortunately, the same solutions that converged using thermo_minteq.tdat will not converge using the NEA database. The error produced is:

       -- Internal error:  calc_sit: Ionic strength is out of range

As an alternative approach, I successfully ran the solutions through PHREEQC using the same NEA database.

Why does the error message reference ionic strength as being "out of range?" Is there a maximum range associated with the SIT activity model? Also, why do solutions speciate without issue in PHREEQC but not in GWB. The NEA database for each software was checked and they are identical.

Any assistance is greatly appreciated.

[Jia Wang]

Hello,

From your description, it doesn't sound like your fluid is really out of the ionic strength range. I am not sure what might be the exact issue without seeing the input file though. Was there any additional information provided regarding the error in SpecE8?  In some cases, other users have run into this error when they choose an aqueous species in the basis species that's very small in concentration. In that case, it would help the program to converge on a solution if a more abundant species is chosen to represent the component concentration in the basis. 

Hope this helps,
Jia Wang
Aqueous Solutions 

[elg0086]

Hi Jia,

Thank you for the response. Prior to posting a forum message, I attempted basis swapping for several of the major elements. I also removed a few of the minor elements with little to no success. Convergence was achieved by lowering the U concentration to 50 g/L and removing several metals, however, uranium is the element of interest and should not be altered. Unfortunately, convergence achieved by tweaking individual concentrations defeats the purpose of using GSS and SpecE8 to process ~400 solutions. Attached is one of the solutions along with the current NEA database. Any additional thoughts or comment are greatly appreciated.

Thank you!

 

[Jia Wang]

Hello,

Thank you for providing the input file and your dataset. I noticed that the issues seem to arise with your lower concentration anion components. I think the issue here is that the free concentration of the aqueous species for AsO4-- and SeO4-- is in a very small concentration given the condition of your fluid. When this is the case, the program may have difficulty converging in its Newton-Raphson iterations. For more information on this, please see section 4.3.7 Optimizing the Starting Guess in the Geochemical and Biogeochemical Reaction Modeling textbook. When I removed both components from your basis pane, SpecE8 had no issue converging. I returned to your input file and added both components back into the Basis at very small amounts (1e-5 mg/l) and then ran the calculation to find the most abundant arsenic and selenium species, which is calculated to be NiHAsO4 and FeSeO3+ respectively. Then I returned to the Basis and swapped them in for AsO4-- and SeO4-- with the same concentration you had in your original file. When I ran the calculation this time, SpecE8 was able to converge successfully. 

By the way, your input script was using the default thermo_nea.tdat dataset and not the customized version you attached. I also noticed that you had decoupled the UO2+/U+++ redox couple. This allows you to set the component concentration for U+++ independently from UO2+, but looking at thermo_nea.tdat, there are no species or minerals that are in the redox form of U+++. If uranium speciation is your interest here, I am not sure how useful this would be. If you expect uranium to speciate across different redox state, then you would need to allow UO2+/U+++ remain in equilibrium. For more information on redox disequilibrium and examples, please see section 7.3 in the GWB Essentials User Guide. 

Hope this helps,
Jia 

[Jia Wang]

Hello,

Just a few more thoughts for your consideration. The SIT dataset accounts for fluids where the dominant electrolytes are Na+, Cl-, K+, ClO4-, and NO3-. Interaction parameters are provided for these major electrolyte ions to work accurately at higher ionic strength. Since your fluid is dominated by sulfates and uranium, this dataset doesn't have the interaction parameters for these species to account for speciation at the higher ionic strength as you were expecting. 

I should clarify that when swapping in As and Se species, the concentration should be constrained with equivalent concentrations. You can do so by finding the elemental concentration matching the original constraint and then using the "as" unit, similar to what you have done for H2(PO4). 

Also note that since there are fewer U species in this model than Minteq's database, so I suspect the ionic strength is going to be relatively different. You might also want to consider the various redox equilibria / disequilibria for each redox component to make sure that couples are decoupled appropriately. When you add a redox species in GSS, the redox species is automatically disabled in your SpecE8 calculations. 

If you can post your PhreeqC script, we can take a look to see how it is setup compared to your GWB input script. 

Best regards,
Jia

[elg0086]

Hi Jia,

Thank you for the reply and I appreciate your efforts towards resolving this issue. I apologize for the delayed response and thought it would benefit the community to complete correspondence. There’s quite a bit of detail in your responses and I will do my best to address each point. First, I should summarize the objectives, as I can now see from my previous posts that the details were a bit vague.

The goal of speciating a large number of solutions (200+) was to investigate mineral saturation of each solution, specifically (super)saturated minerals. Ideally, GSS would launch SpecE8 for all solutions (and converge), followed by a Python post-processing script that would selectively find saturated minerals in each output file. The results would be dumped into excel, where they would be graphed along with pH, ionic strength, etc. The results of this exercise would help guide the development of a custom database for a project by identifying minerals to migrate from the NEA database to a modified MINTEQ database. The NEA database contains more uranium solid phases than other available databases.

Ionic Strength: The project solutions have higher ionic strength than a typical natural water, or even impacted mine waters. You are correct that the chemistry I provided is missing Cl- and NO3-, but several of the epsilon parameters only require Na+, K+ and SO4--. The project also has some Cl- data that has not been distributed yet and NO3- is not present in the system.

Basis Swap: Prior to posting the GWB file I was able to speciate other project solutions using basis swapping, similar to the methods you described. However, I was hoping to avoid investigating each solution independently to find the most abundant species of a particular element and then basis swapping. You are correct that As and Se are relatively small compared to other elemental concentrations; however, these concentrations are not insignificant in other solutions.

Redox: Redox reactions will receive more scrutiny in subsequent speciations and transport modeling. As part of the database review, redox reactions are coupled and rely on the system master Eh. The decoupled model (Fe and U) is an error and likely part of a troubleshooting sequence, sorry about that.

I also apologize for the script file having the incorrect database. Lately, swapping thermo databases using the GUI (file à preferences) has been unsuccessful and  requires a work-around by opening the script in a text editor. Perhaps I am missing something though?

I think my original post reflects some optimism, where I was secretly hoping that a hidden “knob” could assist with convergence errors (i.e., epsilon, itmax, timax). If the majority of solutions were able to converge, this task would only take an hour or two to complete.

In summary, it was bit frustrating to see immediate convergence with PHREEQC and not GWB. Although basis swapping solves most of the convergence errors, the effort was not worth it for this particular task and I ended up using PHREEQC.

Thank you again for your assistance!

-Erik