I want to simultaneously model diffusion of a contaminant in the wall of plastic pipe and the flow of air (and the contaminant) in the actual pipe center. The pipe is thin (ca. 20 mm) and very long, up to hundreds of meters. Obviously there is no need for the same resolution in the length direction, and if I used the same resolution the problem would become prohibitively large.
I could
1) make my equations dimensionless but then I would have to introduce anisotropic diffusivity, and there doesn't seem to be such an option in the modules for PDE or advection/diffusion in the GUI.
2) create a mesh with elements of high aspect ratio, i.e. different size in the two dimensions. But I can't find out how to do this in gmsh.
Does anyone have advice or suggestions? If I don't manage to solve this problem it seems I will not be able to move forward with Elmer.
Thanks!
Simulation with high aspect ratio (long pipe)
-
- Posts: 4
- Joined: 23 Jun 2019, 11:32
- Antispam: Yes
- Location: Sweden
Re: Simulation with high aspect ratio (long pipe)
Hi,
AFAIK anisotropy is possible in Elmer. I have never used it myself. You might search the forum and the Models manual, or wait for an answer from someone else who knows more.
For a mesh with a high aspect ratio, you might ask on the gmsh mailing list (maybe after a search in their archive, I think this has been asked before). Normally you get a response rather quickly.
HTH,
Matthias
AFAIK anisotropy is possible in Elmer. I have never used it myself. You might search the forum and the Models manual, or wait for an answer from someone else who knows more.
For a mesh with a high aspect ratio, you might ask on the gmsh mailing list (maybe after a search in their archive, I think this has been asked before). Normally you get a response rather quickly.
HTH,
Matthias
-
- Posts: 4
- Joined: 23 Jun 2019, 11:32
- Antispam: Yes
- Location: Sweden
Re: Simulation with high aspect ratio (long pipe)
Thank you, Matthias.
-
- Posts: 4
- Joined: 23 Jun 2019, 11:32
- Antispam: Yes
- Location: Sweden
Re: Simulation with high aspect ratio (long pipe)
I found a forum thread about structurally anisotrpic solid materials but nothing about diffusion. So any further help and advice would be appreciates.
Re: Simulation with high aspect ratio (long pipe)
The first source of information about such things is the Models Manual. Have a look at the chapter on Advection/Diffusion. You can also search the manual...
HTH,
Matthias
HTH,
Matthias
-
- Site Admin
- Posts: 4828
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Simulation with high aspect ratio (long pipe)
Hi
What is your Reynolds Number? I.e. is the flow in the pipe turbulent or laminar.
There is no difference whether you deal with the aspect ratio using unisotropic material parameter or huge aspect ratio. They should result to exactly the same linear systems. You should use extruded meshes such that each cross section is exactly the same for best accuracy of convection.
I would be still be worried that numerical diffusion will be difficult to eliminate due to the huge size of the problem.
Perhaps you could only account for transverse diffusion and reduce the dimensionality of the problem, or at perhaps solve the problem in moving coordinate system with explicit periodicity thereby marching the the results one timestep at a time. These will of course depend whether you can make any such approximations.
-Peter
What is your Reynolds Number? I.e. is the flow in the pipe turbulent or laminar.
There is no difference whether you deal with the aspect ratio using unisotropic material parameter or huge aspect ratio. They should result to exactly the same linear systems. You should use extruded meshes such that each cross section is exactly the same for best accuracy of convection.
I would be still be worried that numerical diffusion will be difficult to eliminate due to the huge size of the problem.
Perhaps you could only account for transverse diffusion and reduce the dimensionality of the problem, or at perhaps solve the problem in moving coordinate system with explicit periodicity thereby marching the the results one timestep at a time. These will of course depend whether you can make any such approximations.
-Peter