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 The information provided may be out of date.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.

general and cartesian infinite elements

Please login with a confirmed email address before reporting spam

Hi all,

I have trouble understating how Cartesian infinite elements works.

Let's say, I design a sphere with radius 10, and position 0,0,0. I also add a layer with thickness 2m.

Therefore, the center of this object is at 0,0,0, and I can assign this center at the infinite elements by using type: spherical and as center 0,0,0.




Now if I design a block with 10x10x10 and position corner at 0,0,0, layer thickens 2 m, the center of the object is 5,5,5.

1) When I use Cartesian type infinite elements, what is considered as center?
2) If I change the type to general, I can understand the scaling direction (some areas are extent to x, some to x,y, some to x,z etc). But what is coordinate at interface? I remember that comsol 3.5 had as coordinates, the center of the domain. How could I use the general type, with blocks?

Thanks in advance.

5 Replies Last Post 12.06.2012, 00:46 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 08.06.2012, 15:00 GMT-4
Hi

I do not believe you need to define any centre for cartesian PMLs, but you should not apply cartesian PML on a spherical or cylindrical outer geoemetrical shape, that will not work well.

Use the spherical or cylindrical PMLs on corresponding shaped geometries such that a wave emerging from "the centre" will hit the PML surfaces as close as possible in a "parallel" field way (perpendicular propagation)
This will alsways be the case, more or less, for a cartesian type independent where the source is in the volume

Another trick I have learnt about the PML or infinite elements: use the layer feature of the outer air/vacuum geometrical shape, then mesh the central part and mesh the PML is to use the advanced layer with a sweep mesh (or a few sweep meshes in cartesian) such that these PML region meshes have the type of "onion shells", and do not use regular thets (it works too, but I find less efficient. I mostly use about layer thickness of 1/10 of the total air size and at least 3 wavelengths, whatever is greatest, I tend to optimise both if possible

--
Good luck
Ivar
Hi I do not believe you need to define any centre for cartesian PMLs, but you should not apply cartesian PML on a spherical or cylindrical outer geoemetrical shape, that will not work well. Use the spherical or cylindrical PMLs on corresponding shaped geometries such that a wave emerging from "the centre" will hit the PML surfaces as close as possible in a "parallel" field way (perpendicular propagation) This will alsways be the case, more or less, for a cartesian type independent where the source is in the volume Another trick I have learnt about the PML or infinite elements: use the layer feature of the outer air/vacuum geometrical shape, then mesh the central part and mesh the PML is to use the advanced layer with a sweep mesh (or a few sweep meshes in cartesian) such that these PML region meshes have the type of "onion shells", and do not use regular thets (it works too, but I find less efficient. I mostly use about layer thickness of 1/10 of the total air size and at least 3 wavelengths, whatever is greatest, I tend to optimise both if possible -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11.06.2012, 16:42 GMT-4
Hi Ivar,

Thanks for the reply.
I am trying with the spherical elements, but I still have no luck. The "anomalous" body, should extented in infinity.

Any ideas?
Hi Ivar, Thanks for the reply. I am trying with the spherical elements, but I still have no luck. The "anomalous" body, should extented in infinity. Any ideas?


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 11.06.2012, 17:00 GMT-4
Hi

but you have GND on the external shell so I do not believe you can get cte charge all the way out, and anyhow it's not good practive (was in v3.5, perhaps that have chaned in v4?) to change the "material" and BCs of the PMLs.

And your cylinder is violating the spherical symmetry.

But have you tried with a cylinder type PML and geometry shape ?

--
Good luck
Ivar
Hi but you have GND on the external shell so I do not believe you can get cte charge all the way out, and anyhow it's not good practive (was in v3.5, perhaps that have chaned in v4?) to change the "material" and BCs of the PMLs. And your cylinder is violating the spherical symmetry. But have you tried with a cylinder type PML and geometry shape ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11.06.2012, 17:08 GMT-4
Thanks again.

I am trying to duplicate a comsol 3.5 model, where the cylinder was extending to infinity. This is something I cannont do with version 4? If not, then what is an alternative way besides extending the domain far far away?

I have two options for the infinite elements, where for these cylinders I have assigned cylindrical BC. In both cases, I have the same issues.




P.S.

(see this thread www.comsol.com/community/forums/general/thread/25481/)

Thanks again. I am trying to duplicate a comsol 3.5 model, where the cylinder was extending to infinity. This is something I cannont do with version 4? If not, then what is an alternative way besides extending the domain far far away? I have two options for the infinite elements, where for these cylinders I have assigned cylindrical BC. In both cases, I have the same issues. P.S. (see this thread http://www.comsol.com/community/forums/general/thread/25481/)

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 12.06.2012, 00:46 GMT-4
Hi

if you want a constant distribution, what about considering a cylindrical volume with symmetry conditions at the end caps ?

--
Good luck
Ivar
Hi if you want a constant distribution, what about considering a cylindrical volume with symmetry conditions at the end caps ? -- Good luck Ivar

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.