Convergence issues during precipitation

[Izuto Takimoto]

Dear all,

I am modelling reactive transport in a 1-D packed column where oversaturated H₄SiO₄ enters from the inlet and precipitates as SiO₂(am). (attached my .in file)
When I enable both UPDATE_POROSITY and UPDATE_PERMEABILITY in the CHEMISTRY card, the simulation diverges after a few Newton iterations. Disabling both flags let the run complete.

Questions

1. Are there recommended solver/timestep settings for porosity–permeability feedback cases?

2. Do I need additional cards (AUTO_SCALE, SCALING, etc.) to stabilise the Newton solve?

3. When the inflowing solution is highly oversaturated, are there recommended guidelines for selecting the mineral RATE_CONSTANT, initial VOLUME_FRACTION, and SPECIFIC_SURFACE_AREA to maintain numerical stability?

[Hammond, Glenn E]

Izuto,

When I run the input deck that you sent, I see SiO2(am) precipitating on the inlet end of the column along with a slight porosity reduction and the simulation runs to completion. However, if the rate constant or specific surface area is large, it is possible that the SiO2(am) volume fraction will approach one with the porosity approaching zero...clogging. This is a challenging aspect of such a conceptual model. You can add "MINIMUM_POROSITY <float>" to the chemistry block to prevent full clogging. This will also prevent the reactive transport solver from fully failing due to zero porosity, but the SiO2(am) volume fraction will likely exceed one.

You can try setting the SURFACE_AREA_FUNCTION to POROSITY_VOLUME_FRACTION_RATIO (see

https://documentation.pflotran.org/user_guide/cards/subsurface/chemistry/mineral_kinetics_card.html along with Eq. 34 at 

https://documentation.pflotran.org/theory_guide/mode_reactive_transport.html#changes-in-material-properties

) to restrict the available surface area as the porosity approaches one.

If this does not work, a modified version of kinetic rate expressions (based on transition state theory) should be developed within a Reaction Sandbox where the rate approaches zero as the porosity approaches zero.

Always begin with a small RATE_CONSTANT and increase and/or small specific surface area.

You are using "expert" level capability in the code that is very sensitive to geochemical conditions. Maintaining numerical stability will be more challenging.

Glenn

From: 'Izuto Takimoto' via pflotran-users <pflotra...@googlegroups.com>
Date: Tuesday, July 8, 2025 at 1:31 AM
To: pflotran-users <pflotra...@googlegroups.com>
Subject: [pflotran-users: 8484] Convergence issues during precipitation

Check twice before you click! This email originated from outside PNNL.

--
You received this message because you are subscribed to the Google Groups "pflotran-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pflotran-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pflotran-users/f019e9eb-69ca-486b-96a4-db0938fa98d6n%40googlegroups.com.