Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Wrong stress distribution

Please login with a confirmed email address before reporting spam

I'm modeling ice flow using a user defined creep strain rate tensor. I'm getting close to the expected velocities but the stresses are much too high and have an odd distribution that I think indicates some numerical error (see attached screenshot of the xy component of the deviatoric stress tensor). I've tried a few different solvers (MUMPS and PARADISO are the ones I've tried that solve in a reasonable time) and I've tried decreasing the mesh size and have the same problem. I'd really appreciate any ideas on what to try to fix this. I've attached the mph file.



1 Reply Last Post 2020年1月9日 GMT-5 02:21
Henrik Sönnerlind COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 5 years ago 2020年1月9日 GMT-5 02:21

Hi Claire,

What you see is a postprocessing artefact.

The main problem here is the location at which the stresses are evaluated during postprocessing. The stresses are computed from the elastic strains, which is the difference between the total strains (a linear field within each element) and the creep strains (defined pointwise in the Gauss points). These two sets of strains are almost equal, so computing the difference is a sensitive operation.

If you instead use an expression where you evaluate the stresses at the Gauss points, gpeval(4,solid.sdevxy), you will get rid of the artefacts.

These effects are discussed in detail in the section Result Presentation in the Structural Mechanics User's Guide (version 5.5).

You can also check out https://www.comsol.com/blogs/introduction-to-numerical-integration-and-gauss-points .

In your model, the effect is extremely severe because of the singularity introduced at the end of the Fixed constraint. The (almost) infinite stress causes very high creep strains locally. Maybe a Roller condition is better.

Regards,
Henrik

-------------------
Henrik Sönnerlind
COMSOL
Hi Claire, What you see is a postprocessing artefact. The main problem here is the location at which the stresses are evaluated during postprocessing. The stresses are computed from the elastic strains, which is the difference between the total strains (a linear field within each element) and the creep strains (defined pointwise in the Gauss points). These two sets of strains are almost equal, so computing the difference is a sensitive operation. If you instead use an expression where you evaluate the stresses at the Gauss points, *gpeval(4,solid.sdevxy)*, you will get rid of the artefacts. These effects are discussed in detail in the section **Result Presentation** in the Structural Mechanics User's Guide (version 5.5). You can also check out . In your model, the effect is extremely severe because of the singularity introduced at the end of the Fixed constraint. The (almost) infinite stress causes very high creep strains locally. Maybe a Roller condition is better. Regards, Henrik

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.