"SaveData" "SaveScalars" in parallel

[mark smith]

Hi Elmer users,
Using Elmer Version: 9.0 (Rev: Release, Compiled: 2021-05-08) on windows 10

I'm doing a cfd/thermal problem and would like to save some data on a boundary surface using the save data procedure, in particular the min, max and mean temperature and the temperatures variance over the named boundary surface, the min,max and mean can be obtained using the "Parallel Reduce = Logical True" keywords in the Solver and the following lines

Variable 1 = Temperature
Operator 1 = "boundary min"
Operator 2 = "boundary max"
Operator 3 = "boundary int" ! mean temp is boundary int divided by boundary area
Operator 4 = "area"
Operator 5 = "int variance"

but I get the following WARNING:: SaveScalars: Reduction not implemented in parallel:int variance ;-(

If i don't use the Parallel Reduce, how do I postprocess the data files to extract the variance over the named boundary from the variancies of the partial boundary areas in each partition?

I could do it by postprocessing in paraview but I'd like to automate a series of runs looking for the best design that gives the most uniform temperature.
Best Regards
Mark

[Rich_B]

Hello,

If you have a chance, posting a small working example of your issue will be very helpful. Include the geometry and sif files, and it helps to direct the solver output to a text file and include the log file in an archive.

Thanks, Rich.

[mark smith]

Hi All,
Having reread the solver manual section on Saving Scalar Values to a File, I think that "int variance" works on the whole mesh rather than the boundary I'm interested in, unfortunately "boundary int variance" isn't available as it throws:
WARNING:: SaveScalars: Unknown operator: boundary int variance

So I think I'm going to have to get the temperature variance over the boundary surface by post processing the results outside of Elmer ;-(

Hi Rich_B, I don't have a small working example to post, not sure it would be of much help as it would only report what I already know, but thanks for replying.

Regards
Mark