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 The information provided may be out of date.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.

Large fluctuations in simulation results

Please login with a confirmed email address before reporting spam

Hi everybody.

I'm simulating an electrostatic system in comsol+matlab 3.5a. My results shows large fluctuations when I change to geometry slightly, and I suspect that this has something to do with the meshing. Can anybody give me a hint of what to look into to minimize these fluctuations?

I have looked into meshsmooth(..) and meshrefine(..), but I have found that meshrefine(..) reduce the fluctuations but change the results qualitatively (to something qualitatively different from the analytical expr. it should approach). My geometry is cylindrical symmetric, so I think that the optimal would be to create a cylindrical symmetric mesh, however i cannot create this by revolving a 2D mesh because I have points on the revolving axis.

Help me! I'm in distress!

Thanks, Morten.

3 Replies Last Post 11.04.2011, 20:32 GMT-4
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 11.04.2011, 06:20 GMT-4
Hi

you can mostly cut your geometry with a few lines, creating hence interiour boundaries, such that it appears more symmetric / similar for a revolve mesh.
Then you points becomes the intersection of your lines and mostly one can revolve the mesh out (it depends on where you have your "points" ;)

As you want to do: always check by decreasing mesh size that your model is mesh insensitive (to the best you need/can manage). But normally it should CONVERGE against the analytical model, it could be that something else is wrong too

--
Good luck
Ivar
Hi you can mostly cut your geometry with a few lines, creating hence interiour boundaries, such that it appears more symmetric / similar for a revolve mesh. Then you points becomes the intersection of your lines and mostly one can revolve the mesh out (it depends on where you have your "points" ;) As you want to do: always check by decreasing mesh size that your model is mesh insensitive (to the best you need/can manage). But normally it should CONVERGE against the analytical model, it could be that something else is wrong too -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11.04.2011, 17:27 GMT-4
Hi Ivar. Thank you for your help! Just to be sure I have attached images of my geometry with and without inserted inner boundaries (by accident I named the images as subdomains. I don't hope that this causes any confusion). Is this what you meant?

Thanks, Morten.
Hi Ivar. Thank you for your help! Just to be sure I have attached images of my geometry with and without inserted inner boundaries (by accident I named the images as subdomains. I don't hope that this causes any confusion). Is this what you meant? Thanks, Morten.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11.04.2011, 20:32 GMT-4
Oooh thank you so very much Ivar. I just finished implementing your suggested change and it converges perfectly. You're the man! 2:30 am - going to bed now :D.

Thanks, Morten.
Oooh thank you so very much Ivar. I just finished implementing your suggested change and it converges perfectly. You're the man! 2:30 am - going to bed now :D. Thanks, Morten.

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.