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.

Infinite Elements - Electrostatics

Please login with a confirmed email address before reporting spam

Hi all,

I try to simulate an charged object inside an unbounded region. I use the AC/DC module --> Electrostatics in Version 4.1.
I attached the file. The outer bounds are ground and the charged 'cat' has a given charge density. The outer region is an infinite element.

When I start the simulation, I got the following message: Undefind value found.

Has anyone any idea, what I may make wrong? I try the same in Version 3.5a and there is no error.

Thanks a lot.

Best regards.
Irina Stockem


6 Replies Last Post Aug 22, 2013, 12:11 p.m. EDT
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 Jan 7, 2011, 4:09 a.m. EST
Hi

I notice a few things:
1) you are still with the initial 4.1, get the PATCH on the main COMSOL site, there are still a few equation error in your version
2) your Bezier Polygone 6 is empty, I'm not sure it has any influence, but as a general rule, get rid of non used features, COMSOL is already complicated like that

3) I do not think you really need the infinite elements, your volume is already "large" like that, try just to disable the Infinite elements.

I believe the issue is that your ground is so far away "behind" the infinite value that you are missing a BC like that.

You should even use a finer mesh (or analyze the mesh density relation to the results)

If you turn on a Countour plot you will notice that your square is influencing the field shape by its corners. So I would advise to rather use a large circle centered on your cat, then your results would be even more precise over the hole domain. This is also worth a comment as it can have its importance in many simulation cases, the shape of "large" "infinite" domains are also important

One more:

Once you have replaced the two squares by a circle, then 1) subtract the cat from the circle, no need to mesh inside the cat, 2) then you inverse your system: a) put the cat boundary at GND, b) remove your charge and c) put 1000V around on the circle border. This will better show you the spiking effects, use extra fine meshing. (all without infinite elements OK

Note the potential around the spiky ears and tail, compare to the applied voltage !

Nice cat ;)


--
Good luck
Ivar
Hi I notice a few things: 1) you are still with the initial 4.1, get the PATCH on the main COMSOL site, there are still a few equation error in your version 2) your Bezier Polygone 6 is empty, I'm not sure it has any influence, but as a general rule, get rid of non used features, COMSOL is already complicated like that 3) I do not think you really need the infinite elements, your volume is already "large" like that, try just to disable the Infinite elements. I believe the issue is that your ground is so far away "behind" the infinite value that you are missing a BC like that. You should even use a finer mesh (or analyze the mesh density relation to the results) If you turn on a Countour plot you will notice that your square is influencing the field shape by its corners. So I would advise to rather use a large circle centered on your cat, then your results would be even more precise over the hole domain. This is also worth a comment as it can have its importance in many simulation cases, the shape of "large" "infinite" domains are also important One more: Once you have replaced the two squares by a circle, then 1) subtract the cat from the circle, no need to mesh inside the cat, 2) then you inverse your system: a) put the cat boundary at GND, b) remove your charge and c) put 1000V around on the circle border. This will better show you the spiking effects, use extra fine meshing. (all without infinite elements OK Note the potential around the spiky ears and tail, compare to the applied voltage ! Nice cat ;) -- Good luck Ivar


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 Jan 7, 2011, 4:31 a.m. EST
Hi Again

Your solution is working in fact, but you should use a "general" Type infinite element, and not a cartesian, because your boxes are not centered on 0,0 and your Cartesian infinite domain conditions are defined at 0,0

You can also either center your 2 boxes on 0,0, or replace the infinite element center and use Cartesian. But then you will notice the effect of the corners

Again the shape of your infinite elements should be close to the one of your field for best reply

--
Good luck
Ivar
Hi Again Your solution is working in fact, but you should use a "general" Type infinite element, and not a cartesian, because your boxes are not centered on 0,0 and your Cartesian infinite domain conditions are defined at 0,0 You can also either center your 2 boxes on 0,0, or replace the infinite element center and use Cartesian. But then you will notice the effect of the corners Again the shape of your infinite elements should be close to the one of your field for best reply -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Jan 7, 2011, 5:11 a.m. EST
Hi Ivar

Thanks a lot for you fast and detailed response.
I will change the mentioned points.

Regards,
Irina
Hi Ivar Thanks a lot for you fast and detailed response. I will change the mentioned points. Regards, Irina

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Aug 22, 2013, 5:03 a.m. EDT
Hi,

How do you display the boundary identification which is very helpful to select the boundary in your graph window?

Thank you,

Qize,
Hi, How do you display the boundary identification which is very helpful to select the boundary in your graph window? Thank you, Qize,

Jeff Hiller COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Aug 22, 2013, 8:01 a.m. EDT
Hello Qie,
In the Model Builder, under Definitions>View, check the "Show geometry labels" checkbox.
Best,
Jeff
Hello Qie, In the Model Builder, under Definitions>View, check the "Show geometry labels" checkbox. Best, Jeff

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Aug 22, 2013, 12:11 p.m. EDT
Hi Jeff,

Thank you for your help,

with best regards,

Qize
Hi Jeff, Thank you for your help, with best regards, Qize