Acoustics Simulation (from Comsol to Elmer)?
Re: Acoustics Simulation (from Comsol to Elmer)?
Tried it, still not working. Any help would be appreciated.
-
- Posts: 2237
- Joined: 25 Jan 2019, 01:28
- Antispam: Yes
Re: Acoustics Simulation (from Comsol to Elmer)?
You would need to post your mesh and sif files, so they could be reviewed.
Re: Acoustics Simulation (from Comsol to Elmer)?
Ok this is the file: https://mega.nz/#!ZR9yGaSB
In case_test.sif is everything you need
In case_test.sif is everything you need
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Acoustics Simulation (from Comsol to Elmer)?
Hi, the link cannot be opened. It needs a key. -Peter
Re: Acoustics Simulation (from Comsol to Elmer)?
UPDATE: Sorry my bad, this is the correct one: https://mega.nz/#!Ic8BgYzK!XC7-eaezYHlo ... c_LQCO469c
Last edited by panosvar on 07 Nov 2019, 15:37, edited 1 time in total.
Re: Acoustics Simulation (from Comsol to Elmer)?
I tried also with the 2d equivalent (2d catopsis of the model), but results are the same. I'm thinking of 2 issues right now, either:
1) The conversion from Comsol format (.mphtxt) to Elmer isn't accurate, or
2) It's something that I still don't get about how Helmholtz is applied on Elmer
Anyway, below you can find a picture from the model in Paraview, plus the Elmer project files (you can find the solution in 'case_test.sif').
1) The conversion from Comsol format (.mphtxt) to Elmer isn't accurate, or
2) It's something that I still don't get about how Helmholtz is applied on Elmer
Anyway, below you can find a picture from the model in Paraview, plus the Elmer project files (you can find the solution in 'case_test.sif').
- Attachments
-
- brick-6-single-noPLM-filled-new-2d.rar
- (466.35 KiB) Downloaded 238 times
-
- Capture.PNG
- (63.88 KiB) Not downloaded yet
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Acoustics Simulation (from Comsol to Elmer)?
Hi
You had only one major problem in your sif: the frequency was 40 kHz resulting to wavelength of ~9 mm. On the other hand your characteristics mesh size is ~7 cm. Hence you cannot capture the sinusoidal waves. With standard FE you need easily 10 elements for wavelength. This of course limits the wavenumber. Does Comsol use some enriched FEM, or is there some difference in scaling?
-Peter
You had only one major problem in your sif: the frequency was 40 kHz resulting to wavelength of ~9 mm. On the other hand your characteristics mesh size is ~7 cm. Hence you cannot capture the sinusoidal waves. With standard FE you need easily 10 elements for wavelength. This of course limits the wavenumber. Does Comsol use some enriched FEM, or is there some difference in scaling?
-Peter
- Attachments
-
- case.sif
- Modified sif file
- (2.71 KiB) Downloaded 223 times
Re: Acoustics Simulation (from Comsol to Elmer)?
Hi Peter,
Thanks for the reply. This brick only operates as expected for 40kHz, that's why the length of the brick is λ = c / f = 340 / 40000 = 0.0085 m. So, the problem is with the mesh? The fun fact is that this runs pretty accurate in Comsol and I just export this mesh from Comsol to Elmer. How did you check that my characteristics mesh is 7cm? Also, as I can see you ran the Helmholtz solver in both materials, does it need to be like that?
Panos
Thanks for the reply. This brick only operates as expected for 40kHz, that's why the length of the brick is λ = c / f = 340 / 40000 = 0.0085 m. So, the problem is with the mesh? The fun fact is that this runs pretty accurate in Comsol and I just export this mesh from Comsol to Elmer. How did you check that my characteristics mesh is 7cm? Also, as I can see you ran the Helmholtz solver in both materials, does it need to be like that?
Panos
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Acoustics Simulation (from Comsol to Elmer)?
Hi
If you want to describe a sinusoidal wave with polynomial basis functions you need easily around ten elements for each wave.
I checked that the size of your system was ~4 x 8 m^2, and you had 16 k triangles. Just counted in my head that it rougly equates in having 250 rectangles per m^2 and hence 1/16 m edge length. Ball park figure.
If Comsol can solve the exact problem with same dimensions then it could, for example
1) Use some formulation where basis functions are enriches with plane waves
2) Solve different ansatz, I know for optics they have "wave envolope"
Haven't looked at Comsol myself.
-Peter
If you want to describe a sinusoidal wave with polynomial basis functions you need easily around ten elements for each wave.
I checked that the size of your system was ~4 x 8 m^2, and you had 16 k triangles. Just counted in my head that it rougly equates in having 250 rectangles per m^2 and hence 1/16 m edge length. Ball park figure.
If Comsol can solve the exact problem with same dimensions then it could, for example
1) Use some formulation where basis functions are enriches with plane waves
2) Solve different ansatz, I know for optics they have "wave envolope"
Haven't looked at Comsol myself.
-Peter
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Acoustics Simulation (from Comsol to Elmer)?
Hi
To add, whether you want to model the plastic parts is up to you. It seems that the sound propagates there too. Probably less with the higher frequencies.
-Peter
To add, whether you want to model the plastic parts is up to you. It seems that the sound propagates there too. Probably less with the higher frequencies.
-Peter