Fix stress

[anne-glerum]

As first mentioned on the forum by Sungho, the stress should be computed as 2*eta*strain_rate-P*I or -2*eta*strain_rate+P*I, instead of 2*eta*strain_rate+P*I as is currently done in the visualization postprocessors that output stresses. Here I fixed the sign, sticking to the convention of positive compressive stresses, so -2*eta*strain_rate+P*I.

See also discussion in #3878 .

For all pull requests:

• I have followed the instructions for indenting my code.

For new features/models or changes of existing features:

• I have tested my new feature locally to ensure it is correct.
• I have created a testcase for the new feature/benchmark in the tests/ directory.
• I have added a changelog entry in the doc/modules/changes directory that will inform other users of my change.

[naliboff]

Looks good, thanks @anne-glerum! The only key remaining item would be to update the test output.

A general questions: Is there somewhere in the manual that discusses visualizing stress that should be updated?

Once merged, the forum post should also be updated.

[bobmyhill]

The "compressible_strain_rate" variable is actually the incompressible part of the strain rate. Ditto for the other two files. Suggest changing the name to incompressible_strain_rate?

This is my only quibble :)

[anne-glerum]

Thanks for having a look. It depends on what the material model returns for (this->get_material_model().is_compressible() whether the strain rate in compressible_strain_rate is in fact compressible or the incompressible part. So the name is not always correct. But that's not really related to this PR, so I'll leave that for now, OK?

[naliboff]

Hmm, the relevant code above is:

= (this->get_material_model().is_compressible()
?
strain_rate - 1./3 * trace(strain_rate) * unit_symmetric_tensor()
:
strain_rate);

Even if the material is compressible, the pressure is removed (-1/3 * trace(strain_rate)) and thus you have the deviatoric strain rate. If an incompressible formulation is used, then should the strain rate by default be deviatoric?

If the above is correct, would deviatoric_strain_rate be the most accurate name?

No preference on whether this is resolved now or in a follow-up PR, but will leave that up to @bobmyhill.

[bobmyhill]

@naliboff Yes, that's what I was trying to say. Thanks for clarifying!
I'm fine for the rename to be a separate PR.

[anne-glerum]

• Alright, I've checked that the failing tests failed because of the stress postprocessors and updated the results.
• I couldn't find an instance in the manual about stress visualization that needs to be updated.
• I'll update the forum post when this is merged.

[naliboff]

@anne-glerum - This is good to go, thanks! I'll make a follow-up PR that renames compressible_strain_rate to 'deviatoric_strain_rate', per @bobmyhill suggestion.