CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Programming & Development (https://www.cfd-online.com/Forums/openfoam-programming-development/)
-   -   using fieldAverage library to average postprocessing (https://www.cfd-online.com/Forums/openfoam-programming-development/70396-using-fieldaverage-library-average-postprocessing.html)

jiejie April 4, 2013 07:43

Quote:

Originally Posted by Hanzo (Post 418295)
If my calculations are right then UPrime2Mean should exactly be what it says:
U-prime, squared and averaged = <u'^2>. To check this, you can compute u' using U and <U> , calculate u'^2 and average it manually. I did this for a series of data sets and when I compared UPrime2Mean with manually generated <u'^2> they turned out to be the same.

Another hint is the magnitude. In my computations, the biggest components of UMean_X are around 0.97 and UPrime2Mean_XX of 0.0025.
UPrime2Mean cannot be U^2

Hi Hanzo

Thanks for your quick reply. I was not saying UPrime2Mean is U^2. I calculate the UPrime and found it is the same as sqrt(UPrime2Mean). I got confused as eelcovv hinted " you can obtain the <u'^2> by subtracting the Umean squared from the Uprime2mean". I am not trying to offend anyone but find the right meaning of UPrime2Mean.

Thanks

jiejie

Hanzo April 4, 2013 08:47

Quote:

Originally Posted by jiejie (Post 418313)
I am not trying to offend anyone but find the right meaning of UPrime2Mean.

I am sorry if my comment let you feel like this. I did not want to say that you are offending anyone. It's just my raw english :o I am also highly interested in the correct meaning of UPrime2Mean.

Quote:

Originally Posted by jiejie (Post 418313)
Hi Hanzo

Thanks for your quick reply. I was not saying UPrime2Mean is U^2. I calculate the UPrime and found it is the same as sqrt(UPrime2Mean). I got confused as eelcovv hinted " you can obtain the <u'^2> by subtracting the Umean squared from the Uprime2mean".

How did you compute UPrime? Did you take U - UMean and averaged this over several time steps? Or for a single realization only?
Averaging this would give <u'> (should be a value close to zero, right?). And then you compared to sqrt( UPrime2Mean ) (which is related to the rms value sqrt(<u'^2>) according to my understanding).
If you are right then UPrime2Mean gives <u'>^2.

Could you describe a little bit more in detail what you calculated?
I really hope that UPrime2Mean_XX corresponds to <u_x'^2>,
UPrime2Mean_YY to <u_y'^2> and so on. If not I have to recheck quite some of my results :D

jiejie April 4, 2013 08:58

Quote:

Originally Posted by Hanzo (Post 418332)
I am sorry if my comment let you feel like this. I did not want to say that you are offending anyone. It's just my raw english :o I am also highly interested in the correct meaning of UPrime2Mean.

Could you describe a little bit more in detail what you calculated?
I really hope that UPrime2Mean_XX corresponds to <u_x'^2>,
UPrime2Mean_YY to <u_y'^2> and so on. If not I have to recheck quite some of my results :D

Hi Hanzo

I thought I might upset eelcovv as I was confused with his hint :D. Anyway, I actually output the flow field for every single time step and use the postprocessing utility sample to probe the velocity at a few locations. Then, I calculate the the veocity RMS by its definition, which should be the UPrime^2. I did this a while ago, I think the RMS value is very close to the sqrt of UPrime2Mean. That's why I am thinking UPrime2Mean is giving the correct result.

jiejie

eelcovv July 22, 2013 16:26

Hi jiejie,

You are correct, my hint in an ealier post is misleading. The uprime2suare is already correct for U^2.
So if you define U(t)=<U>+u'(t) then the uprime2square constains the in this order:

<u'^2> <uv> <uw> <v'^2> <vw> <w'^2>

no need to subtract anything. Forget my remark about that.

If you want to plot for instance the turbulent kinetic energy k over a line sampled from you uprima2square you can just do

plot 'sampledline_with_UPrime2Square' u 1:(0.5*($2+$5+$7))

cheers

snappyHex November 14, 2013 08:03

Hi eelcovv,

Thanks for the utility, it is really helpful!
However I am not getting the results that I would expect.
I have two time directories (0 and 90) and I want to get the average of the U field. Running the application I get the Umean file but the results are not correct,
i.e. U in 0 : (0.1107 -0.0027 0.0006),
U in 90: (0.0036 -0.1216 -0.0004)

and the average I get is: (0.1017 -0.0126 0.0005).

Could you give me any hint of what it can be wrong?
Thanks!

My controlDict:
Code:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                |                                                |
| \\      /  F ield        | OpenFOAM: The Open Source CFD Toolbox          |
|  \\    /  O peration    | Version:  2.1.1                                |
|  \\  /    A nd          | Web:      www.OpenFOAM.org                      |
|    \\/    M anipulation  |                                                |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version    2.0;
    format      ascii;
    class      dictionary;
    location    "system";
    object      controlDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

application    simpleFoam;

startFrom      startTime;

startTime      0;

stopAt          endTime;

endTime        90;

deltaT          1;

writeControl    timeStep;

writeInterval  90;

purgeWrite      0;

writeFormat    ascii;

writePrecision  7;

writeCompression on;

timeFormat      general;

timePrecision  6;

runTimeModifiable true;
functions
{
    fieldAverage1
    {
        type            fieldAverage;
        functionObjectLibs ( "libfieldFunctionObjects.so" );
        enabled        true;
        cleanRestart        true;
        outputControl  timeStep;
  //      outputControl  outputTime;
        outputInterval  1;
        fields
        (
            U
            {
                mean        on;
                prime2Mean  on;
                base        time;
            }

       
        );
    }
}


Nucleophobe January 28, 2014 15:28

1 Attachment(s)
Quote:

Originally Posted by snappyHex (Post 461955)
Hi eelcovv,

Thanks for the utility, it is really helpful!
However I am not getting the results that I would expect.
I have two time directories (0 and 90) and I want to get the average of the U field. Running the application I get the Umean file but the results are not correct,
i.e. U in 0 : (0.1107 -0.0027 0.0006),
U in 90: (0.0036 -0.1216 -0.0004)

and the average I get is: (0.1017 -0.0126 0.0005).

snappyHex,

I know this is an old thread, but I am trying to understand what's going on with the postAverage utility as well. Did you ever resolve the problem?

I tried to replicate your results, but I instead get
(0.075 -0.04233333333 0.0002666666667)
in 90/UMean. Strange.


I tried creating a single-cell mesh with a uniform velocity of (1 0 0) in timestep '0', and a uniform velocity of (2 0 0) in timestep '1'. Rather than a result of (1.5 0 0), I get (1.333333333 0 0)!

I have attached the case for reference (see README file).

Edit: This seems to be the same problem Yann was having:
http://www.cfd-online.com/Forums/ope...tml#post384964

I think the key lies in the 'uniform/time' files saved in each timestep directory; the information in these files (e.g. deltaT) determines how the averaging is performed! Removing all 'uniform' directories via 'rm -r */uniform' (be careful) changes the results for me. However, it's still not quite right.


For now, I am using ParaView's 'Temporal Statistics' filter and saving the resulting data to VTK (Save Data -> .vtm filetype).

Nucleophobe January 29, 2014 16:40

postAverage 'bug' fix
 
All,

EDIT: the original code for postAverage seems to work fine on OpenFOAM/2.2.0/051613. I was having problems with OpenFOAM/2.1.x/071612. If in doubt, check your setup with the case I posted above or something similar.

I have found a solution, but so far it only works if your data is saved at consistent time intervals (1, 2, 3 etc., not 1, 3, 4, 4.5 5.1...).

The utility was counting values from the first timestep twice when computing the average. So, for instance, the average of '1' and '2' was (1 + 1 + 2)/3 = 1.3333 instead of (1 + 2) / 2 = 1.5. This may or may not be a big deal; if you are averaging a lot of data with a small standard deviation, you won't even notice. However, if you are trying to average only a few time directories, it can make a big difference.

I have fixed the problem by subtracting one from the startTime index 'timeI'. You should also clear or move your '(time)/uniform' directories, or else the '(time)/uniform/time' files will mess up the average. Perhaps changing the functionObject options in 'system/controlDict' would make this unnecessary and allow for averaging data saved at inconsistent time intervals, but I have not tried this yet.

Here is my change to postAverage (see bolded line):
Code:

\*---------------------------------------------------------------------------*/

#include "fvCFD.H"

int main(int argc, char *argv[])
{
    argList::noParallel();
    timeSelector::addOptions();

#  include "setRootCase.H"
#  include "createTime.H"

    instantList timeDirs = timeSelector::select0(runTime, args);
    runTime.setTime(timeDirs[0], 0);
#  include "createMesh.H"

    // Cycle through time directories
    forAll(timeDirs, timeI)
    {
      Info<< "Setting startTime index to: " << timeI - 1 << endl;
      // It is important that we set the startTime  index to 'timeI-1' to avoid counting values in the first time directory twice
      runTime.setTime(timeDirs[timeI], timeI - 1);
      Info<< "Adding fields for time " << runTime.timeName() << endl;

#      include "createFields.H"

      runTime.functionObjects().execute();
    }

    Info<< "\nEnd" << endl;


    return 0;
}

// ************************************************************************* //

This is giving me the results I expect. If someone else knows a better way to fix this problem, please chime in!

Good luck,
-Nuc

Cluap September 2, 2015 04:58

1 Attachment(s)
Hi everybody !

Thanks eelcovv for this very usefull tool !

However I am now trying to use it with a dynamic mesh case and I am facing some difficulties.
In order to make it work with a mesh changing over the time I've added #include "dynamicFvMesh.H" in the top of the postAverage.C file, I've switched #include "createMesh.H" to #include "createDynamicFvMesh.H" and I've added the line mesh.readUpdate(); in the time loop.
(see the file attached)

When I run the tool it seems to do the calculations but at the end it gives me completely absurd results.

Anybody knows how to make it work correctly with a dynamic mesh ?

Thanks in advance

manuc August 8, 2016 04:53

Hello

Were you able to write the data only at last time step

pintu0101 May 19, 2017 08:58

unable to run the code
 
sir,
plz tell me how to run the given code and how to get time averaged pressure or velocity for the whole domain in a dat file.

sitajeje July 9, 2017 06:53

The standard fieldAverage utility in OpenFOAM4.0 with the command "simpleFoam -postProcess" can also do this postAverage job.


All times are GMT -4. The time now is 11:52.