Menu

Discussion Forum

New Discussion
Filter

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.

How to set epsilon/mu parameters in PML(4.0 RF)

Posted Jul 29, 2010, 4:26 p.m. EDT 10 Replies

Jie Xiang
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Hi,

I tried to simulate a TM plane wave by setting background electric field Ey=exp(-j*emw.k0*x).
PML is surrounded by the rectangle simulatd area.

When I set parameters epsilon/mu/conductivity as air (epsilon=1, mu=1,conductivity=0)all over the geometry(PML+simulated area), I can get pefect plane wave. However, when I set parameters as other material(epsilon=2, mu=1,conductivity=0) all over the geometry,I can not get pefect plane wave.

BTW, the thickness of PML is one wavelength.
The attached images are the results of these two different settings.

My question is:
in order to get perfect plane wave, how to set PML parameters if the simulated area is not air?
10 Replies Last Post Sep 23, 2010, 11:38 a.m. EDT
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Aug 1, 2010, 7:49 a.m. EDT
Hi

for me it looks like your lateral sides are not "Perfectly" matched and you have reflexion from mthe lateral sides interacting, but this is just a guess.

In which version are you ? use the new forum tags, it helps us others udnerstand better too ;)

--
Good luck
Ivar
Hi for me it looks like your lateral sides are not "Perfectly" matched and you have reflexion from mthe lateral sides interacting, but this is just a guess. In which version are you ? use the new forum tags, it helps us others udnerstand better too ;) -- Good luck Ivar
Christian Matyssek
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 8, 2010, 11:25 a.m. EDT
I am experiencing the same issue in a scattering model using Comsol 4.0.0.937. The particle is located at the origin, then we have some space and the outer subdomain is a PML boundary (See attached file pml_air.png).

If I use air in the surrounding and the PML subdomains, the solution looks nice. If I switch both permittivities to some value unequal 1, I get severe errors in the solution. (See pml_glass.png). I also tried setting the permittivity of the PML region equal to 1 and the surrounding of the particle to some value unequal 1. Same wrong results.

Seems like these are UPML (UnPerfectly Matched Layers ;o)). Unfortunately, the documentation is not of much help here. Has anyone experienced the same issue or does anyone have an idea?

Best regards!
I am experiencing the same issue in a scattering model using Comsol 4.0.0.937. The particle is located at the origin, then we have some space and the outer subdomain is a PML boundary (See attached file pml_air.png). If I use air in the surrounding and the PML subdomains, the solution looks nice. If I switch both permittivities to some value unequal 1, I get severe errors in the solution. (See pml_glass.png). I also tried setting the permittivity of the PML region equal to 1 and the surrounding of the particle to some value unequal 1. Same wrong results. Seems like these are UPML (UnPerfectly Matched Layers ;o)). Unfortunately, the documentation is not of much help here. Has anyone experienced the same issue or does anyone have an idea? Best regards!

Attachments:

Kaiting He
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 10, 2010, 1:39 a.m. EDT
For the expression of incident wave in 'Scalar Variables', the wavevector k0 should be corrected as sqrt(epsilon)*k0, in which the epsilon is the permittivity of surrounding. My COMSOL version is 3.5a, and I did not know if this solution would work in COMSOL4.0. Good luck!
For the expression of incident wave in 'Scalar Variables', the wavevector k0 should be corrected as sqrt(epsilon)*k0, in which the epsilon is the permittivity of surrounding. My COMSOL version is 3.5a, and I did not know if this solution would work in COMSOL4.0. Good luck!
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 10, 2010, 5:20 a.m. EDT
Hi

an check that you have latest patch, fo me today this means 4.0.0.993, see tha main COMSOL web site

--
Good luck
Ivar
Hi an check that you have latest patch, fo me today this means 4.0.0.993, see tha main COMSOL web site -- Good luck Ivar
Christian Matyssek
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 15, 2010, 5:35 a.m. EDT
Hi!

Thanks for hinting me at the wave vector definition. sqrt(epsilon)*k0 indeed did the trick. This is somehow inconvenient, because when changing the material of the surrounding domain, you'll also have to change the incident field definition. The only solution I see so far is to adding a global variable with the refractive index.

By the way: Is there a complete variable reference for emw.*? Haven't been able to find one so far.

Thanks and best regards!

Christian
Hi! Thanks for hinting me at the wave vector definition. sqrt(epsilon)*k0 indeed did the trick. This is somehow inconvenient, because when changing the material of the surrounding domain, you'll also have to change the incident field definition. The only solution I see so far is to adding a global variable with the refractive index. By the way: Is there a complete variable reference for emw.*? Haven't been able to find one so far. Thanks and best regards! Christian
Kaiting He
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 15, 2010, 12:49 p.m. EDT
I also have not seen variable of this kind. :)


He Kaiting
I also have not seen variable of this kind. :) He Kaiting
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 16, 2010, 2:44 a.m. EDT
Hi

but epsilon and "n" the refractive index are linked to the spped of light no ?

--
Good luck
Ivar
Hi but epsilon and "n" the refractive index are linked to the spped of light no ? -- Good luck Ivar
Christian Matyssek
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 16, 2010, 3:15 a.m. EDT
Hi!

In a medium with refractive index n the speed of light is c=c0/n, where c0 is the _vacuum_ speed of light. In terms of the wave vector, in a medium with refractive index n, the wave vector is k=k0*n. The refractive index and the permittivity epsilon are related by sqrt(epsilon)=n.
So in the Comsol model, we use the free space wave vector emw.k0 to specify the wave. This is somewhat confusing, because I would have expected Comsol to scale the wave vector automatically in a medium with n!=1, but it seems like it does not.
Hi! In a medium with refractive index n the speed of light is c=c0/n, where c0 is the _vacuum_ speed of light. In terms of the wave vector, in a medium with refractive index n, the wave vector is k=k0*n. The refractive index and the permittivity epsilon are related by sqrt(epsilon)=n. So in the Comsol model, we use the free space wave vector emw.k0 to specify the wave. This is somewhat confusing, because I would have expected Comsol to scale the wave vector automatically in a medium with n!=1, but it seems like it does not.
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 16, 2010, 5:21 a.m. EDT
Hi

ideally we should have this confirmed by COMSOL support, no ?
(PS I'n not by COMSOL ;)

--
Good luck
Ivar
Hi ideally we should have this confirmed by COMSOL support, no ? (PS I'n not by COMSOL ;) -- Good luck Ivar
Christian Matyssek
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 23, 2010, 11:38 a.m. EDT
Yes, you are right. I had this issue in 3.5a already and the support told me to rescale the wave vector. I'll contact the support.
Yes, you are right. I had this issue in 3.5a already and the support told me to rescale the wave vector. I'll contact the support.

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.

Suggested Content