Reproducing NAFEMS benchmark test

General discussion about Elmer
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Reproducing NAFEMS benchmark test

Post by Rich_B »

Hello,

Regarding the vtu coloring options, if you load the vtu file, and click on the 'Solid Color' box, does the drop box show a list of variables?

This is documented in 'GetStartedElmer' in the chapter 'Introduction to Paraview'.

Rich.
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Reproducing NAFEMS benchmark test

Post by kevinarden »

If you use the result output solver and ask for a vtk file, Elmer does not write the results to the vtu file that you asked for in the Simulation section, it creates the file but doesn't write any results to it since the user specified vtk format. The intent of the solver is to use one or the other. I don't think the end user has attempted with a vtu only file.
cghw01
Posts: 6
Joined: 17 Oct 2022, 17:46
Antispam: Yes

Re: Reproducing NAFEMS benchmark test

Post by cghw01 »

OK Thanks, I can now see results using .vtu.

I am trying to understand what I did wrong to get the erroneous temperature result. I will report back when I find it !

Chris
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Reproducing NAFEMS benchmark test

Post by kevinarden »

check your mesh dimensions, the units should be meters.
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Reproducing NAFEMS benchmark test

Post by Rich_B »

Hello Kevin,

Just FYI, regarding this:
it creates the file but doesn't write any results to it
Peter fixed this a while ago, so the empty vtu file is no longer created. If someone does both, adding Post File = case.vtu, and adds ResultsOutput solver using the default vtu, then nothing bad happens, it is simply redundant.

This example where both case.vtu and case.vtk are created is a new variation, and seems to work properly, even though not really needed.

Rich.
cghw01
Posts: 6
Joined: 17 Oct 2022, 17:46
Antispam: Yes

Re: Reproducing NAFEMS benchmark test

Post by cghw01 »

OK, One problem was that I was using mm as my dimension in gmsh. This is what was used in the original tutorial video I viewed. Also, I need to have the y axis as the axisymmetric axis and not the x axis.

At point 0.04, 0.04008, I get 332.618 K - so pretty close to the desired value of 332.97 K. I might get closer with a finer mesh, but it will do.

Thanks for the help.

Chris
Attachments
NAFEMS_5.png
(128.08 KiB) Not downloaded yet
NAFEMS_5.zip
(46.46 KiB) Downloaded 39 times
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Reproducing NAFEMS benchmark test

Post by Rich_B »

Hello,

As a follow up, the attached archive includes refining the mesh by changing the characteristic length in the gmsh geo file from 0.01 to 0.005.

An ElmerGUI project is also included, and uses the Save Coordinates function in SaveScalars to output the temperature result at the desired location of 0.04, 0.04. The flag 'Exact Coordinates' is set to true, and then SaveScalars will interpolate the results from the nodal results to the desired location. The results.dat file shows this:
4.000000000000E-002 4.000000000000E-002 0.000000000000E+000 3.329718970521E+002
So the computed temperature of 332.9719 K is close enough to the benchmark target of 332.97 K.

Here are two screenshots of the ElmerGUI project.
project.png
project.png (179.34 KiB) Viewed 590 times
coords.png
coords.png (86.33 KiB) Viewed 590 times
Rich.
Attachments
NAFEMS_5_rev2.zip
(55.87 KiB) Downloaded 36 times
Rich_B
Posts: 421
Joined: 24 Aug 2009, 20:18

Re: Reproducing NAFEMS benchmark test

Post by Rich_B »

Hello,

Attached is another archive, this time with a 3D version of the NAFEMS heat transfer benchmark computed using Elmer. The included readme file details the steps needed to create the mesh using gmsh and to run the simulation with Elmer.

The computed temperature at 0.04, 0.04, 0.0, is 332.958 K, which is fairly close to the benchmark target of 332.97 K.

Rich.
Attachments
NAFEMS_5_3D.zip
(2.2 KiB) Downloaded 38 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Reproducing NAFEMS benchmark test

Post by raback »

Great work Chris and Rick! We of course believe that the Solvers give correct answers but to demonstrate it using industry standard tests is always valuable. Also it helps to identity the usual caveats often related to unit system.

-Peter
Post Reply