[odellar]

Is there a way of computing the time averaged flow field from the data produced from a LES (pimpleFoam)?

Or even setting something up before running the LES simulation to instruct it to keep computing the mean flow field whilst solving?

Thanks in advance,
Olie

[amanbearpig]

Hi Olie,

You can use a function in your controlDict to calculate time-averaged values. I've used this to calculate time-averaged values for LES using pimpleFoam myself. Below is the function I include at the bottom of my controlDict file in the LES run directory.

functions
{
  fieldAverage
  {
      type            fieldAverage;
      functionObjectLibs ( "libfieldFunctionObjects.so" );
      enabled         true;
      timeStart       30.0
      outputControl   outputTime;
      outputInterval  100;
 
      fields
      (
          U
          {
              mean        on;
              prime2Mean  on;
              base        time;
          }
 
          p
          {
              mean        on;
              prime2Mean  on;
              base        time;
          }
      );
  }
}

With this you can choose what terms you average (here U and p) as well as when you start averaging and how often you write out the averaged values. Hope this helps!

[odellar]

Thank you - I've just looked in my controlDict file to find that this code was actually already implemented, so now I know what the files UMean, pMean etc are in the time directories!

Two questions, firstly what are UPrime2Mean and pPrime2Mean?

Secondly how do I view the UMean and pMean fields in paraView? It doesn't give me the option to view them..

Thanks again!

Olie

[amanbearpig]

Hi Olie,

UPrime2Mean is the time averaged components of the resolved Reynolds stress tensor - it is the fluctuating velocity components (u') squared (2) and averaged in time (Mean). pPrime2Mean is the corresponding time averaged fluctuating pressure - the fluctuating pressure squared and averaged.

Regarding the viewing of the means in Paraview, I have not had an issue viewing them before, perhaps check and make sure that the time step you are loading into Paraview contains the means (if you don't write out the timeAverage values at every time step they won't be present to view). If you load Paraview, what variables are available to see? And are the means present in the time directory you have loaded into Paraview?

[odellar]

Oh Reynolds stresses - of course, don't know why I didn't spot that!

Arr, I've looked again in paraView and it was because I only ever viewed the first time step (t=0), at which point UPrime and pPrime DON'T exist, but then skipped to the next time step where they DO exist, and can view it fine. Silly me! Perfect.

Thanks a lot!

Olie

[amanbearpig]

Hi Olie,

Glad I could be of help! I've learned alot in OpenFOAM from reading these forums and have been helped by others a great deal, so I'm glad to be able to do the same when I can!

[Howard]

Originally Posted by amanbearpig

Hi Olie,

Glad I could be of help! I've learned alot in OpenFOAM from reading these forums and have been helped by others a great deal, so I'm glad to be able to do the same when I can!

Hi, friend. I would like to ask if I want to calculate the average value of the outlet surface, how to conduct it? or are there resources for these information. Thank you!

[lukasf]

Originally Posted by amanbearpig

      fields
      (
          U
          {
              mean        on;
              prime2Mean  on;
              base        time;
          }

}

Hi,


in my opinion this usage is not correct.


One would first need to calculate a well averaged mean velocity UMean. After that one should active the calculation of prime2Mean.


I tried to do this with OpenFoam v1912 by choosing this setting:

      fields
      (
          U
          {
              mean        on;
              prime2Mean  off;
              base        time;
          }
}

However, if I later set prime2Mean to on OpenFOAM does not read the existing UMean field (which contains a better, well averaged UMean) but instead overwrites the existing UMean field and starts to calculate UMean again.


Has anyone an idea on how to improve this?


My idea was to first determine UMean and then use a second function fieldaverage2 to start using prime2Mean after timestart = 0.005.

    fieldAverage1
    {
        type            fieldAverage;
        libs            (fieldFunctionObjects);
        writeControl    writeTime;
        timeStart       0; //start from the time at which the flow field is properly initialized e.g. 3-5 through flows through the domain. Here I chose timeStart 0 for testing purposes only.

        fields
        (
            U
            {
                mean        on;
                prime2Mean  on;
                base        time;
            }

        );
    }

    fieldAverage2
    {
        type            fieldAverage;
        libs            (fieldFunctionObjects);
        writeControl    writeTime;
        timeStart       0.005; //start determining UPrime2Mean after UMean is properly converged, here e.g. after 0.005s.

        fields
        (
            U
            {
                mean        off;
                prime2Mean  on;
                base        time;
            }

        );
    }

However this error occurs:


--> FOAM FATAL ERROR:
To calculate the prime-squared average, the mean average must also be selected for field U

[Tobermory]

Originally Posted by lukasf

Hi,
in my opinion this usage is not correct.
One would first need to calculate a well averaged mean velocity UMean. After that one should active the calculation of prime2Mean.

No - that's not required thankfully. OF employs a trick that has long been used for LES/DNS solvers, and simply keeps track of the mean and the mean square. From this you can calculate the variance (mean square minus the square of the mean) and the accuracy improves of course as the sample size increases. Why wait for the mean to converge before calculating the mean square? That would just be a waste of computing time.

[Tobermory]

... boy, I hate having to try and write this in Latex, but here goes: