Non convergence when considering non-isothermal or molecular diffusion in TOUGH2 ECO2N
I am using TOUGHREACT V3.32 to model CO2 injection in a radial 2D mesh with different layers in the z-direction. Before running the reaction part of the problem I wanted to test the flow part was working correctly. The model I have set up runs correctly under isothermal conditions with no molecular diffusion, for different CO2 injection rates. Unfortunately , when I run the same model under non-isothermal conditions and/or activate molecular diffusion it runs for a few days (depending on the rate of injection) and then always crashes (does not converge).
The model was first gravity initialized in 1D and these results where then used to gravity initialize the 2D model. If no CO2 is injected (steady state) the model runs with non-isothermal conditions and molecular diffusion.
I have tried different things.
1- Inject a small amount of water in the same cell in which the CO2 is injected.
2- Fix the temperature of the cell where CO2 is injected by considering an "infinite" grain density.
4- Instead of fixing the temperature, consider some enthalpy in the injection to keep the temperature of the injected block similar to the initial one.
3- Consider an "infinitesimal" distance between the injection block and adjacent blocks.
None of this seems to work and I am running out of ideas. Can someone help me?
I attach the flow.inp, Mesh, and INCON files that I am using.
Regards,
Hector
4 replies
-
Hector,
Upfront: I don't think I can help you, because:
(1) I could not run the file as sent: I needed to
(a) remove illegal tab stops in block DIFFU;
(b) remove all values after porosity in the INCON block (illegal format, unless these permeability values are given in a format that TOUGHREACT understands);
(c) increase the initial time step to avoid 10 consecutive steps that converge in a single iteration.
(2) Once I made the changes outlined above, it runs fine, for both isothermal and non-isothermal conditions (it stops again after 10 time steps that converge in a single iteration, which is due to the maximum time step size of 1e5, which is set too small).
I suggest you post this under the TOUGHREACT forum (not the TOUGH2 forum), in case there are differences between TOUGH2-ECO2N and TOUGHREACT-ECO2N.
Best,
Stefan
-
Yes, that is the format for permeabilities in TOUGHREACT (since they can vary over time and space owing to reaction or heterogeneity). Maybe there are some differences in TOUGH2-ECO2n and TOUGHREACT-ECO2n that should be checked. Some of the issues with convergence in a single iteration are fixed with a parameter setting in TOUGHREACT.
-
Dear Stefan and Eric,
Thanks a lot for looking at this.
Stefan: It is quite surprising and very "helpful" to know that this model does run in TOUGH2 ECO2N.
Eric: Could you let me know which is the parameter setting in TOUGHREACT that could help overcome convergence issues in a single iteration when the reaction part of the code is not activated?
Cheers,
Hector
-
Hi Hector,
That parameter is MOPR(4), but it is already set in your simulation. That was not the problem. It may have been the Tab stops in the file. Also, ECO2n is v1 in TOUGHREACT - I'm not sure if Stefan used V2. There are some differences regarding salt precipitation and the effect on permeability in TOUGH2 and TOUGHREACT as well. The other issues regarding permeability in the INCON file will not make a difference unless those permeabilities are different from the values you chose in ROCKS.
Eric