Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

failure to find solution in newer version of COMSOL

Please login with a confirmed email address before reporting spam

Hi,

I am a student with very minimal experience with COMSOL (most of what I know, I have learned just from experimenting with the program). I am modeling fluid flow through a flow chamber (see attached model). COMSOL was able to find a solution when I ran the study for this model in an older version (either 3.5 or 3.5a, I can't remember). However, now that I have updated to 4.2a, running the same model fails to find a solution (Error message is "Maximum number of Newton iterations reached.") I have tried different mesh sizes, without success.

Since I am not familiar with the commands and capabilities of 4.2a, I am hoping someone can direct me toward where I am going wrong. I have attached images of some results I obtained from running the model in the older version (Reynolds number and velocity field slices). I need these results to be verified when I run the same model in 4.2a, before I can move on to redesigning the chamber.

Thanks!


7 Replies Last Post 17 juin 2016, 02:23 UTC−4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 janv. 2012, 04:22 UTC−5
Hello there,

I am also having the same problem. I was given a model that was constructed on a previous version of Comsol, which apparently ran fine in that version. Now when I run it in Comsol 4.2a I get the error:

Failed to find a solution.
In segregated group 2:
The relative error (1.3e+002) is greater than the relative tolerance.
Returned solution is not converged.


Is there something I have to change in the model to make it compatible with my version of Comsol?

Thank you for your time,
Carolyn.
Hello there, I am also having the same problem. I was given a model that was constructed on a previous version of Comsol, which apparently ran fine in that version. Now when I run it in Comsol 4.2a I get the error: Failed to find a solution. In segregated group 2: The relative error (1.3e+002) is greater than the relative tolerance. Returned solution is not converged. Is there something I have to change in the model to make it compatible with my version of Comsol? Thank you for your time, Carolyn.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 janv. 2012, 06:30 UTC−5
Hi

I experience often issues when loading older version , at lest for rather complex ones, so I tend to rebuild them fully.

But you have a few possibilities, try to rebuild new solvers (or make new Study nodes)) and solve with them, and in any case check the segregated solver sequence, sometimes, if you edit the solver sequence, COMSOl flags it a s a custom solver sequence and expect the user to re-edit it regularly, sometimes dependent variables get missed out and you only need to add them in again

--
Good luck
Ivar
Hi I experience often issues when loading older version , at lest for rather complex ones, so I tend to rebuild them fully. But you have a few possibilities, try to rebuild new solvers (or make new Study nodes)) and solve with them, and in any case check the segregated solver sequence, sometimes, if you edit the solver sequence, COMSOl flags it a s a custom solver sequence and expect the user to re-edit it regularly, sometimes dependent variables get missed out and you only need to add them in again -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 janv. 2012, 07:35 UTC−5
Hi

Your problem interested me so much that I downloaded your model file. My first impression was that your mesh is rather sparse. I changed the user-defined mesh, "General physics" to "Fluid dynamics" with the "Coarse" size. Then there was 108305 elements, hence 10-fold of yours. Then I changed in Solver configuration => Solver 1 => Stationary solver => Iterative 1 => Multigrid 1 => Mesh coarsening factor: 1. Thanks to Jukka in Comsol Helsinki office for telling me this hint!

You inlet velocity, 0.25 m/s sounded so high that I changed that to 0.025 m/s. The solution was reached in 10 minutes, but when I tried with 0.25 m/s, the solution did not converge. Hence, the reason might be physical. There may some turbulence in the corners of the thin pipe line, but your original mesh was so sparse that it did not see it. Maybe I am wrong, but you should have at least 4 layers in z direction in the larger domain to make velocity profile meaningful. Due to the five pipelines into the large domain, a swept mesh cannot be done (at least I was not able to accomplish it).

Fluid dynamics is, according to my experience, one of the most difficult phenomena to simulate. Therefore, there are several software packages around just for CFD.

Keep the faith!
Lasse
Hi Your problem interested me so much that I downloaded your model file. My first impression was that your mesh is rather sparse. I changed the user-defined mesh, "General physics" to "Fluid dynamics" with the "Coarse" size. Then there was 108305 elements, hence 10-fold of yours. Then I changed in Solver configuration => Solver 1 => Stationary solver => Iterative 1 => Multigrid 1 => Mesh coarsening factor: 1. Thanks to Jukka in Comsol Helsinki office for telling me this hint! You inlet velocity, 0.25 m/s sounded so high that I changed that to 0.025 m/s. The solution was reached in 10 minutes, but when I tried with 0.25 m/s, the solution did not converge. Hence, the reason might be physical. There may some turbulence in the corners of the thin pipe line, but your original mesh was so sparse that it did not see it. Maybe I am wrong, but you should have at least 4 layers in z direction in the larger domain to make velocity profile meaningful. Due to the five pipelines into the large domain, a swept mesh cannot be done (at least I was not able to accomplish it). Fluid dynamics is, according to my experience, one of the most difficult phenomena to simulate. Therefore, there are several software packages around just for CFD. Keep the faith! Lasse

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 24 janv. 2012, 10:41 UTC−5
Thanks, Ivar. The model is simple, so I may just try rebuilding it for the sake of convenience. Thank you again for your help.
Thanks, Ivar. The model is simple, so I may just try rebuilding it for the sake of convenience. Thank you again for your help.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 15 juil. 2012, 11:13 UTC−4
Dear Lasse,

I am facing similar issues while solving for piezo and acoustic coupling. WHat do you suggest?
I do not find the
" in Solver configuration => Solver 1 => Stationary solver => Iterative 1 => Multigrid 1 => Mesh coarsening factor: 1." option in my system.
Would be glad if someone can help!

Thanks,
Regards,
Megha
Dear Lasse, I am facing similar issues while solving for piezo and acoustic coupling. WHat do you suggest? I do not find the " in Solver configuration => Solver 1 => Stationary solver => Iterative 1 => Multigrid 1 => Mesh coarsening factor: 1." option in my system. Would be glad if someone can help! Thanks, Regards, Megha

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 oct. 2013, 11:56 UTC−4
I am doing a Laminar flow problem in a 3-D micro-channel.
My convergence plot doesn't even start niether I get any error.

I am doing a Laminar flow problem in a 3-D micro-channel. My convergence plot doesn't even start niether I get any error.

Please login with a confirmed email address before reporting spam

Posted: 8 years ago 17 juin 2016, 02:23 UTC−4
Have you solved the problem “relative error ist bigger than tolerence"?
Have you solved the problem “relative error ist bigger than tolerence"?

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.