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 The information provided may be out of date.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.

Ignore geometry boundaries for meshing

Please login with a confirmed email address before reporting spam

Hi all.

I am new to comsol and have a problem I could not find a solution to in the manuals.

I am using the Electrostatics package (in rotation symmetry 2D) to compute a potential (field) map. I used geometry building blocks for the electrodes and the casing and this works very well.

Then i wanted a much higher mesh resolution in a small part of setup, so that i can extract a highly precise fieldmap from this area. Therefore I defined a rectangular geometry and used a seperate mapped mesh + distribution with a high number of points on the geometry borders (maybe there is a better way to do this?).

My Problem now is, that i need to introduce a elliptically shaped space charge inside of this mentioned area. I defined an elliptic geometry and assigned the domain a spacecharge. But because of this new geometric object, the automatic meshing takes the elliptic border into account and distorts the original mesh, with the added effect that the calculated potential in the space charge area is very "spikey" (=discontinous, therefore high fields where actually no field should be present at all and i think this comes from the mesh, as it looks like it changes its properties stongly with different meshes) .

My question now is: Can i somehow tell comsol, that it should ignore the elliptic area (domain) for meshing purposes? I want to retain the ortogonal mesh from the rectangular geometry and it should penetrate the elliptic area as if it is not there at all.

Thank you.


PS: If I export my data in spreadsheet format and use the grid option, so that the extracted points do not coincide with the mesh points, what kind of interpolation is used in that case? Nearest neighbor, linear, n-spline?

2 Replies Last Post Nov 18, 2013, 10:25 p.m. EST
Sven Friedel COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 8, 2013, 8:44 a.m. EST
Hi Michael,

> Can i somehow tell comsol, that it should ignore the elliptic area (domain) for meshing purposes?

Have a look if Geometry > Virtual Operations may be of help to you. There you can hide entities from the mesher. Another way of defining an ellipsiod with special properties (apart from defining it as a domain) is by a 3D interpolation function. It would be automatically interpolated on the exsiting (e.g. mapped) mesh.

If this does not help you further, feel free to contact us in support
www.ch.comsol.com/support

Best regards,
Sven Friedel

Hi Michael, > Can i somehow tell comsol, that it should ignore the elliptic area (domain) for meshing purposes? Have a look if Geometry > Virtual Operations may be of help to you. There you can hide entities from the mesher. Another way of defining an ellipsiod with special properties (apart from defining it as a domain) is by a 3D interpolation function. It would be automatically interpolated on the exsiting (e.g. mapped) mesh. If this does not help you further, feel free to contact us in support http://www.ch.comsol.com/support Best regards, Sven Friedel

Robert Koslover Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Nov 18, 2013, 10:25 p.m. EST
Here's another way: A region of space charge doesn't actually have to have an explicit geometric boundary. Instead, you can define the charge density as a function of (x,y,z). This approach will generally have no impact on the mesh generated. You can use step functions to create ~hard boundaries, if you really want them. But a smooth function (such as a gaussian) will perform better numerically.
Here's another way: A region of space charge doesn't actually have to have an explicit geometric boundary. Instead, you can define the charge density as a function of (x,y,z). This approach will generally have no impact on the mesh generated. You can use step functions to create ~hard boundaries, if you really want them. But a smooth function (such as a gaussian) will perform better numerically.

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.