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.
discretization of multiple domains
Posted 22.09.2010, 03:14 GMT-4 1 Reply
Please login with a confirmed email address before reporting spam
Hi ,
I am modeling a multiple domain model to solve for different physics. How is it possible to determine whether an element node on an interior domain surface belongs to that domain or to the surrounding domain?
In a special case, the tangential current density on some of the surface nodes of a plate surrounded by air is zero, that means theses nodes obviously are assigned to the surrounding air but they should be part of the plate with non-zero currents.
How can I tell COMSOL to assign these nodes to the interior domain?
(The geometry sequence is: 1. Block_plate, 2. Block_air, 3. Form union; the mesh sequence is in the same order)
Maybe is it coming from some kind of mesh optimization that COMSOL does automatically before solving?
I am thankful for any suggestions!
Regards, Marcel
I am modeling a multiple domain model to solve for different physics. How is it possible to determine whether an element node on an interior domain surface belongs to that domain or to the surrounding domain?
In a special case, the tangential current density on some of the surface nodes of a plate surrounded by air is zero, that means theses nodes obviously are assigned to the surrounding air but they should be part of the plate with non-zero currents.
How can I tell COMSOL to assign these nodes to the interior domain?
(The geometry sequence is: 1. Block_plate, 2. Block_air, 3. Form union; the mesh sequence is in the same order)
Maybe is it coming from some kind of mesh optimization that COMSOL does automatically before solving?
I am thankful for any suggestions!
Regards, Marcel
1 Reply Last Post 22.09.2010, 14:29 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
My understanding, but I'm not 100% sure it's correct (I'm not from Comsol and do not know more of it's interiour than anybody else here on the user forum) is that physics and boundary conditions are assigned onto the analysed geometry, then the meshing is made and the corresponding physics attached to the nodes depending on which domain, boundary, edge or point they are linked to.
Now I have also understood that basically boundaries inherite settings from the domains, as edges from the surrounding boundaries, and points from the adjacent edges. Provided this makes sens, or average values may be used.
Interiour boundaries to domain with same physics are by default continuous boundaries. But, if we use the "assembly" mode there is no continuous boundary, and there are in fact two overlapping boundaries, with "up" and "down" normals to distinguish both sides.
If we have different physics on each domain around a common "interiour boundary (w.r.t. geometry)" the nodes on the boundaries should have variables attached from both/each physics (for me "continuity does not apply to different physics, to be checked).
Remains the question if one should rather use "assembly" mode between different adjacent domains where one apply different physics, I do not believe so, (to be checked again).
One example I have noticed were inheritance is not 100% is with an ALE domain with a fixed boundary, i.e. in 2D the edge boundary limit points do not necesarily inherite the prescribed displacement of the edge/boundary, and may want to follow the mesh motion. In this case one might need to add a prescribed displacement explicitely on the edge points.
Other comments appreciated
--
Good luck
Ivar
My understanding, but I'm not 100% sure it's correct (I'm not from Comsol and do not know more of it's interiour than anybody else here on the user forum) is that physics and boundary conditions are assigned onto the analysed geometry, then the meshing is made and the corresponding physics attached to the nodes depending on which domain, boundary, edge or point they are linked to.
Now I have also understood that basically boundaries inherite settings from the domains, as edges from the surrounding boundaries, and points from the adjacent edges. Provided this makes sens, or average values may be used.
Interiour boundaries to domain with same physics are by default continuous boundaries. But, if we use the "assembly" mode there is no continuous boundary, and there are in fact two overlapping boundaries, with "up" and "down" normals to distinguish both sides.
If we have different physics on each domain around a common "interiour boundary (w.r.t. geometry)" the nodes on the boundaries should have variables attached from both/each physics (for me "continuity does not apply to different physics, to be checked).
Remains the question if one should rather use "assembly" mode between different adjacent domains where one apply different physics, I do not believe so, (to be checked again).
One example I have noticed were inheritance is not 100% is with an ALE domain with a fixed boundary, i.e. in 2D the edge boundary limit points do not necesarily inherite the prescribed displacement of the edge/boundary, and may want to follow the mesh motion. In this case one might need to add a prescribed displacement explicitely on the edge points.
Other comments appreciated
--
Good luck
Ivar