Convergence issues of montmorillonite.

[Aditya-Puru]

Hello,
We are working on capturing some cations using the montmorillonite mineral and modeling it on react. The fluid has a high concentration of sodium which the mineral is supposed to capture. The surface data has been attached which has been developed to capture sodium ions according to three different sites as was the case in the same FeOH data files. The thermo.tdat file has been attached with the added mineral being used in the system along with the surface data file and x1t file. The issue is we have tried a multitude of combinations of concentration of ions but still encountering convergence issues. The main aim is to have high sodium concentrations in the fluid for the mineral to capture the ions when the fluid flows through the aquifer. Please find all the necessary files attached.

Thank You

Montmorillonite_capture.sdat thermo (2).tdat Montmorillonite.x1t

[Jia Wang]

Hello,

To start troubleshooting your issues, I performed a Run -> "go initial" calculation and it failed immediately. This indicates that your initial fluid chemistry needs to be troubleshooted. I would suggest that you do this in React.

Could you explain what you are using to set your initial composition? When you perform a mineral swap, like what you have done for Quartz and Montmor-Na, this specifies the mass in the system and tells the program to calculate the composition of the dissolved solutes in equilibrium with these minerals. Is this a correct assumption for your starting system? With the extremely small amount of Na+, I think it is unlikely that Montmor-Na is in equilibrium with your initial fluid. I tried swapping in another aluminosilicate mineral, Kaolinite, which triggered no convergent issue. If you aim to include Montmor-Na as a surface for sorption only, you can always add it as a kinetic mineral instead and set the reaction rate to 0. This way, your mineral will not dissolve or precipitate but the surface area will be available for sorption reactions. Also, a mineral swapped in cannot be set at 0 mass because of the nature of the numerical solver but a kinetic mineral can be. It seems like that was what you were trying to do with the node-by-node setting. More on how to set a kinetic reactant in section 4 Kinetic Reaction Paths of the GWB Reaction Modeling Guide.

I would suggest taking a closer look at the chemistry of the boundary fluid that you specified in the "Fluids" tab as well. When I added the fluid composition to a React instance, it also failed to converge.

I would also suggest that you double check your units to make sure they are correct. Typically, dissolved concentrations are given in units of per mass solution or solvent. If you are using per volume units instead, please check that the values are correct. I also noticed that the concentration for Mg++ is set as an absolute unit (mmol), which is typically not recommended. Absolute units do not scale with the size of the domain. For more information on setting initial conditions, please see section 2.7 in the GWB Reactive Transport Modeling User Guide.

Also, in the attached X1t input file, the surface dataset is Resin.sdat and not the attached surface dataset. I am not sure if this was intentional or a mistake.

Hope this helps,
Jia Wang
Aqueous Solutions LLC