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 The information provided may be out of date.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.

Hollow cylindrical optical cavity

Please login with a confirmed email address before reporting spam

Hi,

I'm fairly new to COMSOL, I was trying to model the optical modes in a hollow silica cylinder - which is easy enough to do with PEC boundaries, but I was hoping to see the modes in the presence of a surrounding air atmosphere. I was hoping there was a simple way to set up the equivalent of an infinite air domain surrounding it. Setting up a small air boundary around the cylinder causes problems because there's modes within the air cavity, setting up a very large domain causes problems with meshing and solving time.

Advice greatly appreciated.

-I tried to attach the file but I was given a file size error

15 Replies Last Post 31 déc. 2010, 04:30 UTC−5
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 21 déc. 2010, 02:21 UTC−5
Hi
first of all to attach files, clear them from the solution eventually also the mesh (no problem in v.4) and do a reset model, then you should have no difficulties to upload it. OK we wouldneed to resolve it cannot have everything ;)

in RF you have the PML isnt that enough (V4.1 at least) ?

--
Good luck
Ivar
Hi first of all to attach files, clear them from the solution eventually also the mesh (no problem in v.4) and do a reset model, then you should have no difficulties to upload it. OK we wouldneed to resolve it cannot have everything ;) in RF you have the PML isnt that enough (V4.1 at least) ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 21 déc. 2010, 20:33 UTC−5
Thanks, I've attached the file

I'm using v.4, when I try defining the air domain as a PML I either get an undefined value error or if it compiles I just get an imaginary solution

help appreciated
Thanks, I've attached the file I'm using v.4, when I try defining the air domain as a PML I either get an undefined value error or if it compiles I just get an imaginary solution help appreciated


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 22 déc. 2010, 15:33 UTC−5
Hi

butI do not see any electric field defined anywhere. With E=0 you will not solve much apart enhance some numerical noise, no ?

then I'm not suree how to interpreste those modes (once you select some E scatter field BC as the defined "air" cylinder modes and the glass defined modes will certainly interact.

You could use some periodic or symmetry boundaries perhaps, and some PML/infinite elements too, no ?

--
Good luck
Ivar
Hi butI do not see any electric field defined anywhere. With E=0 you will not solve much apart enhance some numerical noise, no ? then I'm not suree how to interpreste those modes (once you select some E scatter field BC as the defined "air" cylinder modes and the glass defined modes will certainly interact. You could use some periodic or symmetry boundaries perhaps, and some PML/infinite elements too, no ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 22 déc. 2010, 20:52 UTC−5
Thanks for the response

I'm a little confused, right now I have the electric field defined under the physics node by the refractive index
Is there some other way to define it? Do you mean define a initial value?

And you're suggesting setting the air domain BC as a scattering boundary?
I can try that, thanks

kee s.
Thanks for the response I'm a little confused, right now I have the electric field defined under the physics node by the refractive index Is there some other way to define it? Do you mean define a initial value? And you're suggesting setting the air domain BC as a scattering boundary? I can try that, thanks kee s.

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 23 déc. 2010, 03:03 UTC−5
Hi

I get E=0 when solving so I see no electric field in there, I added a scatter boundary field with Ez or Ex=1[V/m] on the lower circular boundary and then I see some modes build up

--
Good luck
Ivar
Hi I get E=0 when solving so I see no electric field in there, I added a scatter boundary field with Ez or Ex=1[V/m] on the lower circular boundary and then I see some modes build up -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 déc. 2010, 04:25 UTC−5
Thanks,

I was able to find the modes I was interested in by putting a scatter boundary around the air cylinder and making a plane wave

I had a weird issue though - I saved the model that worked and when I came back later and ran i again I got a different result and the error message

UMPS allocation factor increased to 1.44.
The relative error is greater than the relative tolerance.
Relative residual 1.5e-014
Relative error 0.0016
The relative error is greater than the relative tolerance.
Relative residual 4.3e-014
Relative error 0.0019
The relative error is greater than the relative tolerance.
Relative residual 1.4e-013
Relative error 0.00077
The relative error is greater than the relative tolerance.
Relative residual 1.2e-014
Relative error 0.0011
The relative error is greater than the relative tolerance.
Relative residual 1.5e-014
Relative error 0.0016
The relative error is greater than the relative tolerance.
Relative residual 2.2e-014
Relative error 0.0031
The relative error is greater than the relative tolerance.
Relative residual 2.4e-014
Relative error 0.0014
The relative error is greater than the relative tolerance.
Relative residual 3.8e-014
Relative error 0.0017
The relative error is greater than the relative tolerance.
Relative residual 9.9e-015
Relative error 0.0021
The relative error is greater than the relative tolerance.
Relative residual 1.7e-014
Relative error 0.0035
The relative error is greater than the relative tolerance.
Relative residual 2e-014
......

any idea what this is?

thanks again for your help so far
Thanks, I was able to find the modes I was interested in by putting a scatter boundary around the air cylinder and making a plane wave I had a weird issue though - I saved the model that worked and when I came back later and ran i again I got a different result and the error message UMPS allocation factor increased to 1.44. The relative error is greater than the relative tolerance. Relative residual 1.5e-014 Relative error 0.0016 The relative error is greater than the relative tolerance. Relative residual 4.3e-014 Relative error 0.0019 The relative error is greater than the relative tolerance. Relative residual 1.4e-013 Relative error 0.00077 The relative error is greater than the relative tolerance. Relative residual 1.2e-014 Relative error 0.0011 The relative error is greater than the relative tolerance. Relative residual 1.5e-014 Relative error 0.0016 The relative error is greater than the relative tolerance. Relative residual 2.2e-014 Relative error 0.0031 The relative error is greater than the relative tolerance. Relative residual 2.4e-014 Relative error 0.0014 The relative error is greater than the relative tolerance. Relative residual 3.8e-014 Relative error 0.0017 The relative error is greater than the relative tolerance. Relative residual 9.9e-015 Relative error 0.0021 The relative error is greater than the relative tolerance. Relative residual 1.7e-014 Relative error 0.0035 The relative error is greater than the relative tolerance. Relative residual 2e-014 ...... any idea what this is? thanks again for your help so far

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 23 déc. 2010, 04:42 UTC−5
Hi

funny, have you tried to save and restart COMSOl ? and which version 4.0 or 4.1 ? similar things did happen for me in v4.0 and "a", hardly with the new version

--
Good luck
Ivar
Hi funny, have you tried to save and restart COMSOl ? and which version 4.0 or 4.1 ? similar things did happen for me in v4.0 and "a", hardly with the new version -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 déc. 2010, 22:46 UTC−5
Hey again, so I'm still having some issue

I was able to get modes by adding a scattering boundary conditions and an incident plane wave, like I said earlier and I got modes like shown in the image
But I can no longer get the same modes by running the same file, but I cant tell what/if anything's changed

Hey again, so I'm still having some issue I was able to get modes by adding a scattering boundary conditions and an incident plane wave, like I said earlier and I got modes like shown in the image But I can no longer get the same modes by running the same file, but I cant tell what/if anything's changed


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 24 déc. 2010, 04:41 UTC−5
Hi

I suspect that is realted to your (original) model build-up and the mesh density, your glass tube is hanging in free space with an air region that is also a cavity (no PML's) and you are looking at very high order modes so I wounder if these are not coupling almost randomly.

Then I do not understand the new model buildup: where is the PML and why remove the air in the middle ?

are you modelling an infintely long glass tube or just a limited one ?, in the first case you can use periodic boundaries to make it "infinite long"

then to remove the "orange" unit issues in your field equation use rather

cos(y[1/um]*pi/1.55)

or

cos( (y*pi/(1.55[um]))[1/m] )

One more thing, are you sure you have a fine enough mesh ? to resolve your wavelength corectly you need at least 7-10 elements per wavelength (also in the glass)


--
Good luck
Ivar
Hi I suspect that is realted to your (original) model build-up and the mesh density, your glass tube is hanging in free space with an air region that is also a cavity (no PML's) and you are looking at very high order modes so I wounder if these are not coupling almost randomly. Then I do not understand the new model buildup: where is the PML and why remove the air in the middle ? are you modelling an infintely long glass tube or just a limited one ?, in the first case you can use periodic boundaries to make it "infinite long" then to remove the "orange" unit issues in your field equation use rather cos(y[1/um]*pi/1.55) or cos( (y*pi/(1.55[um]))[1/m] ) One more thing, are you sure you have a fine enough mesh ? to resolve your wavelength corectly you need at least 7-10 elements per wavelength (also in the glass) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 déc. 2010, 02:06 UTC−5
I've tried various changes to my meshing and mesh density without any luck

The file I posted has a finite glass cylinder (what I want) and a surrounding air domain outside and inside that's defined as a PML, with a scattering boundary condition on the outside with an incident planewave
I feel like I should be able to get modes with these conditions, and at least in one instance of running the model I did. I'm not sure why it won't find modes now or what's changed from before
I've tried various changes to my meshing and mesh density without any luck The file I posted has a finite glass cylinder (what I want) and a surrounding air domain outside and inside that's defined as a PML, with a scattering boundary condition on the outside with an incident planewave I feel like I should be able to get modes with these conditions, and at least in one instance of running the model I did. I'm not sure why it won't find modes now or what's changed from before

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 25 déc. 2010, 05:09 UTC−5
Hi

In your last file I didnt fnd the air "inside", check your domains, then I would have had some air outside before the plm, and I'm not sure to understand your two reflecting surfaces at the end if the cylinder, do you want to have a cavity or ?

Also have at least 7-10 elements per wavelength, otherwise your field is not resolved corretly you are "oversampling"

--
Good luck
Ivar
Hi In your last file I didnt fnd the air "inside", check your domains, then I would have had some air outside before the plm, and I'm not sure to understand your two reflecting surfaces at the end if the cylinder, do you want to have a cavity or ? Also have at least 7-10 elements per wavelength, otherwise your field is not resolved corretly you are "oversampling" -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 25 déc. 2010, 17:49 UTC−5
So I'm trying to model the circulating modes in a cylidrical wall
I've tried setting the top and bottom of the air cavity to the scattering boundary condition but it hasn't yielded anything

I thought defining a hollow cylinder inside a larger air cylinder would leave an air domain in the center of the hollow portion, maybe I was wrong? Regardless since I'm defining the waves by epsilon and k, and I have air with epsilon = 1 and k = 0 it should give me the same results with vacuum in the center anyway

www.megaupload.com/?d=0JXYGUVU
This was a model I ran a couple days ago with the results not cleared, it yielded the sort of results I'm looking/hoping for, but like I said earlier I can't get the same results when running the same model

I've tried increasing the fineness of the mesh as much as possible and still haven't had any better luck

thanks again for all your help so far
So I'm trying to model the circulating modes in a cylidrical wall I've tried setting the top and bottom of the air cavity to the scattering boundary condition but it hasn't yielded anything I thought defining a hollow cylinder inside a larger air cylinder would leave an air domain in the center of the hollow portion, maybe I was wrong? Regardless since I'm defining the waves by epsilon and k, and I have air with epsilon = 1 and k = 0 it should give me the same results with vacuum in the center anyway http://www.megaupload.com/?d=0JXYGUVU This was a model I ran a couple days ago with the results not cleared, it yielded the sort of results I'm looking/hoping for, but like I said earlier I can't get the same results when running the same model I've tried increasing the fineness of the mesh as much as possible and still haven't had any better luck thanks again for all your help so far

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 27 déc. 2010, 23:29 UTC−5
I saw the same issue again, I made a model with a hollow cylinder in an air domain with scattered boundary conditions with an incident e field and pmls set, and I got a bound mode I was interested in. I saved the file but when I rerun it I just get noise as solutions

It's on this megaupload file with the solutions I found left saved (too big to attach)
www.megaupload.com/?d=IJGO3PHY

If any one has any ideas why I can only sometimes get these solutions, or suggestions as to how else to model it, I'd really appreciate it

I saw the same issue again, I made a model with a hollow cylinder in an air domain with scattered boundary conditions with an incident e field and pmls set, and I got a bound mode I was interested in. I saved the file but when I rerun it I just get noise as solutions It's on this megaupload file with the solutions I found left saved (too big to attach) http://www.megaupload.com/?d=IJGO3PHY If any one has any ideas why I can only sometimes get these solutions, or suggestions as to how else to model it, I'd really appreciate it

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 28 déc. 2010, 01:37 UTC−5
Ok, so I think what I was seeing was just an artifact - it appears after I add a PML afer I've solved for modes with a scattered boundary condition without remeshing

Attached is my most recent attempt it's a hollow cylindrical silica shell with an air domain outside and inside, with a pml on the external air domain and scattered external boundary condition - but I still can't solve for any modes
help appreciated
Ok, so I think what I was seeing was just an artifact - it appears after I add a PML afer I've solved for modes with a scattered boundary condition without remeshing Attached is my most recent attempt it's a hollow cylindrical silica shell with an air domain outside and inside, with a pml on the external air domain and scattered external boundary condition - but I still can't solve for any modes help appreciated


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 31 déc. 2010, 04:30 UTC−5
I've still had no luck

Does anyone have any experience modeling 3D photonic cavities without mirrored BCs (i.e., surrounded by air) that could give me some pointers?

I've still had no luck Does anyone have any experience modeling 3D photonic cavities without mirrored BCs (i.e., surrounded by air) that could give me some pointers?

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.