Rename shear stress postprocessor

[anne-glerum]

The shear stress postprocessor outputs the tensor
-2*eta*deviatoric_strain_rate.
I think this is the deviatoric stress (deviatoric_stress = stress - 1/3*trace(stress)*unity_tensor), where the full stress is -2*eta*deviatoric_strain_rate+pressure*unity_tensor.

Shear stresses are the off-diagonal components of the stress tensor.

I would suggest to rename the postprocessor deviatoric stress, but this would not be backwards compatible.

Also, in the shear stress postprocessor, the elastic stresses are added to -2*eta*deviatoric_strain_rate and I'm not sure that is correct. For a deviatoric stress postprocessor, I would suggest to compute the full (including elasticity when switched on) stress and then compute the deviator.

[bangerth]

Do you happen to have a concise definition of the shear stress? Does it just involve setting the diagonal entries to zero?

[anne-glerum]

After discussion with @cedrict, the naming of the postprocessor is okay. For incompressible elasticity, it is also okay to add the stress contribution to the shear stress tensor, instead of building the full stress tensor (including the elastic contribution) and then take the deviator.

[bangerth]

OK. So should we close this? Do you feel like we ought to better describe what the postprocessor does?