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 The information provided may be out of date.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.

relativ error: point charge (in grounded shell) ~2E-4

Please login with a confirmed email address before reporting spam

Hallo everybody,
i try to figure out how accurate results from Comsol are by testing simple analyitcal known problems.

The first problem i tried is the point charge in free space (using infinite elements) and a point charge in a grounded shell of 1m radius. I achieved the best results with physical controlled mesh on the finest setting with up to 4 refinement steps and 80GB of available memory, 2 days of computing time on a big machine and my relativ error oszilates with peaks at 2E-4 (or even higher) which is - in the opinion of my supervisor - way to high.

I am desperate to know if i do conceptional misstakes and how to achieve a higher accuray by some oders of magnitude and what your experiences with lower limit of relativ errors are.

I export with full precision :)

Thank you!

2 Replies Last Post 2012年6月5日 GMT-4 16:15
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 2012年6月5日 GMT-4 14:58
Hi

indeed that sounds rough ;) but perhaps you are overdoing the mesh density ? anyhow you have also to play with the discretization level, and probably an uniform mesh is not even optimum

--
Good luck
Ivar
Hi indeed that sounds rough ;) but perhaps you are overdoing the mesh density ? anyhow you have also to play with the discretization level, and probably an uniform mesh is not even optimum -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2012年6月5日 GMT-4 16:15
I've seen errors below 10-8 achieved in less than 60 sec of computational time :)

Your test case is not the best for FEM as the method intrinsically has problems with points. You need extremely fine meshing around them, especially if there is singularity at the point. I'm not sure how you simulate the point charge, but what I found is that a charged sphere or very small radius gives much better results than the point charge.
I've seen errors below 10-8 achieved in less than 60 sec of computational time :) Your test case is not the best for FEM as the method intrinsically has problems with points. You need extremely fine meshing around them, especially if there is singularity at the point. I'm not sure how you simulate the point charge, but what I found is that a charged sphere or very small radius gives much better results than the point charge.

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.