thermal analysis of a vacuum chamber

[mzenker]

Please tell me where are the results of my simulation hidden? Is this case.ep? There is written in the SolverManual that I can have results also in ascii format, I tried "Binary output = true" but don't get any additional files.

It depends on the postprocessor you want to use. *.ep this is the format for ElmerPost, the old Elmer postprocessor. There is a new postprocessor integrated in ElmerGUI, but I have never used it. Generally the recommendation is to use ParaView. I have heard that Salomé has a ParaView-based postprocessor. To use ParaView (or Salomé?), you need to write your results in vtu format. You can either replace

Code: Select all

Post file = case.ep

by

Code: Select all

Post file = case.vtu

in the simulation section, or use the ResultOutput solver (see Models Manual).

HTH,

Matthias

[slav]

Hi,

I understand how to write the results to .vtk file, but to be honest I don't like ParaView. I thought of using Save Scalars option as my problem is similar to this of that guy viewtopic.php?f=7&t=2943&start=0&hilit= ... 01b9bb9b03 - a temperature for all coordinates on the inner surfece of my chamber. My goal is to calculate a black body radiation in the geometrical center of my chamber. So I would like a file with these data to calculate a distance from the center to the given point to know what contribution it has in something like weighted avarage. Unfortunately I have really no idea how to access the SaveScalars solver, I found only the ResultOutput solver.

[mzenker]

Hi,

to access the SaveScalars solver, move the corresponding xml file from bin/edf-extra to bin/edf in your Elmer installation. If you reload an existing project with the new solver set, check the settings carefully as they might get screwed up due to a bug in ElmerGUI. I advise to make a backup copy of your sif file before opening ElmerGUI with the new solver set and compare carefully with the newly generated sif file.

HTH,

Matthias

[slav]

Thanks Matthias, I will never guess to do it this way I see the solver now but I realised that I have to put all the coordinates in SaveCoordinates field. The problem is I don't know these numbers. So I looked into the .ep file and what is nice is that there are names of my boundaries (I want only the inner boundaries). So maybe it will be easier to use for my needs but I don't understand the data structure. I found in ElmerSolver manual (Appendix B) something about ascii files structure but it doesn't seem to fit to .ep file. I tried also the .vtu and .vtk format and visualised it in ParaView but it didn't suits my needs

[raback]

Hi Slav

If you need distance to some point why not create a small spherical BC for that and use the DistanceSolver to get the distance.

If you need radiation heat transfer between bodies there is already a BC for that. However, then you center of box control point should not be infidesimal since the radiation conditions require that you have a proper surface representation with the mesh.

-Peter

[mzenker]

Hi,

I tried also the .vtu and .vtk format and visualised it in ParaView but it didn't suits my needs

I am surprised that there is a thing that ParaView cannot do as it is the most-featured open-source postprocessor AFAIK, and can do almost anything thanks to its filters which are freely programmable.

If you don't like ParaView you might also try gmsh as postprocessor.

Matthias

[slav]

Peter,
where can I find the DistanceSolver? I don't see it neither in Elmer nor in the manuals.

Matthias,
ParaView on my Windows is unstable, I got many errors when I click on some of the icons. But this version is better then the previous one I had which was unexpectadely closing up.

[slav]

Hi,

I managed to deal with the result file - its structure is explained in Solver Manual.

Now I'm trying to use the Diffuse Gray option of Radiation. I don't fill any field in the Radiation Settings section (emissivity is set in Material Properities). The View Factors are computed to a file but I got an error. Can anyone guess what is wrong?

Code: Select all

ViewFactors: View factors computed in time (s):  272.70

ViewFactors: 
ViewFactors: Viewfactors before manipulation:
ViewFactors: 
ViewFactors:          Minimum row sum:   0.37640768991410378
ViewFactors:          Maximum row sum:    2.2678340520812736
ViewFactors: 
ViewFactors: Normalizaing Factors...

ViewFactors: Normalization iteration 1:    2.638E-08

ViewFactors: Normalization iteration 2:    2.475E-11

ViewFactors: Normalization iteration 3:    1.497E-16

ViewFactors: Normalization iteration 4:    5.694E-24

ViewFactors: View factors manipulated in time (s):   17.61
ViewFactors: 
ViewFactors: Viewfactors after manipulation:
ViewFactors: 
ViewFactors:          Minimum row sum:   0.99999997822906139
ViewFactors:          Maximum row sum:    1.0000000227669819
WARNING:: ViewFactors: Rowsum of view factors should not be larger than one!
ViewFactors: 

ViewFactors: *** ALL DONE ***

RadiationFactors: Computing factors...

RadiationFactors: View factors filling (%)               9.4680E+01

 Error occurred in umf4num:  -1.00000000000000000     

[raback]

Yes, you have a direct solver - umfpack - which should never be used with diffuse gray radiation BCs since they spoil the matrix structure and the filling gets too dense for direct solvers. Rather use iterative solver. -Peter

[slav]

I checked - I used iterative solver BiCGStab. After your post I tried also direct solvers and other iterative solvers but I get the same error all the time (even when I changed my computer to a much better one).