Add Viscosity Limit to Reactive Fluid Transport

[danieldouglas92]

In the evaluate function of Reactive fluid transport model, the base material model is first evaluated and then a fluid weakening term is multiplied to the viscosity computed in the base model. For base models like Visco plastic, this means that the viscosity that is returned by Reactive fluid transport can be much lower than the Minimum viscosity parameter set within the base model.

This PR adds a new parameter to the Reactive fluid transport model called Minimum weakened viscosity that sets a lower bound on the viscosity after applying this fluid weakening factor. I was hoping that there might be a better way to do this without defining an essentially duplicated parameter, but the problem is that not all base models contain a parameter that sets a minimum viscosity value (i.e. the Simple material model). If there is an alternative way to generalize this, I would love to hear other ideas!

[danieldouglas92]

FYI @naliboff @xlia62

[gassmoeller]

Very nice, just two minor comments.

[gassmoeller]

Suggested change

	const double weakend_viscosity = out.viscosities[q] * (1.0 - porosity) * std::exp(- alpha_phi * porosity);
	const double weakened_viscosity = out.viscosities[q] * (1.0 - porosity) * std::exp(- alpha_phi * porosity);

[gassmoeller]

I know the other variable were around already, but can you change this to make it easier to read:

Suggested change

	double min_weakened_visc;
	double min_weakened_viscosity;

[danieldouglas92]

@gassmoeller Thanks for the quick review and for catching that spelling mistake! I opened up a quick follow up PR to change the parameter name of the compaction viscosity limiters to be consistent with the new parameter in this pr

[naliboff]

@danieldouglas92 - Nice, thank you for putting this PR together! I only have to minor suggested changes on top of those already noted by @gassmoeller.

[naliboff]

Suggested change

	# This model tests the minimum viscosity within the reactive fluid transport model
	# This model tests that the minimum viscosity within the reactive fluid transport model

[naliboff]

Suggested change

	# the parameter Minimum weakened viscosity to 1e10 Pa s, we ensure that the viscosity
	# the parameter Minimum weakened viscosity to 1e19 Pa s, we ensure that the viscosity

[danieldouglas92]

@naliboff Thanks for catching the typos! Just addressed your comments

[danieldouglas92]

I'm not sure why the indent tester is failing? I've run make indent locally and no files are changed, and the indent artifact that the tester outputs is empty

[naliboff]

/rebuild

[gassmoeller]

The code looks all good now.

The tester is failing, because since #6881 we also check in this tester if the parameter documentation needs to be updated, and you added a new parameter here. You can update the parameter documentation manually by running ./update_parameters.sh ../build/aspect from the doc/ directory, or you can download a changes file from the tester. Unfortunately, the tester had a bug that didnt let it upload the changes file correctly. Can you take a look at #6902 and merge that (if it passes the tests)? Then rebase this PR and you should get a changes file prepared on the results page of the tester (just like for the indent and results tester).

[danieldouglas92]

The first commit worked for models that I was running (which did not use the Katz2003 melting model), but @xlia62 pointed out that it wasnt working when using the katz2003 melting model. This is because the katz2003 melting model calculates this weakening factor on it's own within reaction_model/katz2003_melting_model.cc.

I pushed a second commit which removes this code from katz2003_melting_model.cc and moves the code in reactive_fluid_transport.cc around so that the weakening is applied regardless of the reaction model being used. I think this makes sense to avoid needing to potentially make individual viscosity limiters for each reaction model, and I think this should just work, but we'll see if the testers pass

[danieldouglas92]

@gassmoeller Ah that makes sense! I'll push more changes in a minute with the updated parameter.

[danieldouglas92]

The changes to katz2003 caused two of the tests (viscoplastic_melt_blob_freezing and shear_heating_with_melt) to fail. viscoplastic_melt_blob_freezing failing is confusing to me because it uses the reactive fluid transport model, but rising_melt_blob_freezing passes which uses the melt simple material mode. shear_heating_with_melt also uses the melt simple material model, and so it would make sense why this test might fail since the melt weakening is now only being calculated if the katz2003 reaction model uses the reactive fluid transport model.

Clearly this is why there was an if statement in reactive_fluid_transport that excluded the fluid weakening from being calculated if the reaction model was equal to katz2003. Because the katz2003 interfaced with material models outside of reactive fluid transport, the weakening had to be determined internally within the reaction model itself.

An easy solution to this would be to just add an identical parameter to the katz2003 reaction model that allows the user to limit the reduction of the shear viscosity, but this feels pretty hacky.