Reproducing NAFEMS benchmark test

[cghw01]

Hello,
I am new to Elmer (but not so new to FEM), and after going through a number of ElmerGUI tutorial examples, I decided I wanted to try going all the way from generating a geometry in gmsh to solving the problem in Elmer. I followed the YouTube video here:https://www.youtube.com/watch?v=C42MR4WkJd4 and succeeded in getting the desired result. I then went on to trying to reproduce a predicted result using a NAFEMS benchmark test (https://www.nafems.org/). I have attached a couple of images which describe the benchmark test. It is an axisymmetric (vertical axis) tube with the 3 outer boundaries kept at 293.15 K and the innermost boundary (nearest the axis) kept at zero flux (top and bottom sections) and the middle section with a heat flux of 5x10^5 W/m^2. The thermal conductivity is 52 W/Km. The NAFEMS benchmark gives a temperature prediction at a certain location. I go through the same procedure as was described in the video, but at the moment, all I see is a flat temperature of 293.15 K everywhere. I am probably making a simpleton mistake, but so far it has eluded me.

I have attached the sif file but no more, as I seem to be limited to just 3 attachments.

If anyone can suggest a way forward, it would be appreciated.

Many thanks in advance

Chris Walker

[kevinarden]

nafemsheat.zip

(240.05 KiB) Downloaded 44 times

[Wyatt57]

Hello,

Thank you for the download link, the file will certainly be of great help to me.

[cghw01]

Hello,
Thankyou Kevin for providing a solution.

According to the NAFEMS benchmark, the temperature at location 0.04, 0.04 should be 332.97 K, whereas in my simulation, I am getting a result around 1330 K. The problem could well be to do with an incorrect heat input flux. According to the NAFEMS benchmark test, this is 5x10^5 W/m2. I understand that the input to Elmer should be in W, so I need to multiply the heat flux by the area of the heated region. This area is 2*pi*0.02*0.06 which gives 0.00754 m^2. Multiply this by 5x10^5 and I get 3769.9 W. However, as said, I get the wrong result for the test location. Have I estimated the heat flux incorrectly, or could there be another problem ?

By the way, I gave the wrong cif file previously (cartesian symmetry set and heated boundary condition was set to temperature rather than heat flux), so I attach it again.

(I note that the corrected version that you sent had only one solver, but I seem to need two as I cannot seem to view .vtu files in Paraview, so I generate .vtk files as well)

Again many thanks in advance

Best Wishes
Chris

[kevinarden]

VTU files are generated by default. I use VTU in paraview. If you want VTK or other file formats you have to use the save solver to change the defaults.

W is per area for boundary, or mass if a body force

[Rich_B]

Hello,

Out of curiosity, about this:

as I cannot seem to view .vtu files in Paraview

AFAIK, Elmer vtu files always seem to work with Paraview, do you know why Paraview is giving you trouble? Maybe share what happens when you load an Elmer vtu output file?

Thanks, Rich.

[cghw01]

Hello,
Thanks for the replies.

In Paraview if I select vtk, then under coloring I have a Temperature option (as well as Solid Color, and GeometryIds), but if I select vtu, I have no Temperature option (see attachments). I suspect that means I had some setting wrong in Elmer. In my defence, the video tutorial that I quoted in my first post also had the same problem, so it was suggested to generate vtk rather than vtu.

Note also that the temperature distribution is about right, but the temperature itself is way too hot. This was when I entered about 4000 into the heat flux boundary condition (after correcting for area). If I do not correct for area, then I should enter 5x10^5 in the boundary condition and then the result will be even hotter. Maybe I have made an error elsewhere. When I generated the gmsh file, I assumed the dimensions were in mm (I think I read that somewhere).

Best Wishes
Chris

[kevinarden]

units are SI for dimension SI is meters

[kevinarden]

If you use the result output solver set to vtk format than the data is written to the vtk file instead of the vtu file.
If you have
Post File = case.vtu
set in the simulation section then a vtu file is created by the default in the simulation section.

You should either use the result output solver and specify your format or use the default one in the simulation section, but not both.

[kevinarden]

At 0.04 I get 313 degrees

heat.png

(251.15 KiB) Not downloaded yet

nafemsheat.zip

(112.83 KiB) Downloaded 44 times