[pedroxramos]

Hi,

to study the fluctuations of the velocity and the fluctuations of the concentration of a passive scalar, I've inserted this in my controlDict file:
fieldAverage
{
type fieldAverage;
functionObjectLibs ( "libfieldFunctionObjects.so");
[...]
fields
(
U
{
mean on;
prime2Mean on;
base time;
}
s
{
mean on;
prime2Mean on;
base time;
}
);
}
#includeFunc scalarTransport

As you know, this will give me U_prime2mean and s_prime2mean (indicated with arrows)

Now, I want to couple both fluctuations and get uprime_sprime_mean.
Does anyone know how?

[Tobi]

Hi,


maybe you can use the new fields and use a calc function. E.g.

postProcess -func calc ...

It might also work on the fly, by using the calc function directly inside the controlDict but for this, the new generated fields should be registered in the database. Out of the box, I am not sure, if this can be done on the fly.

[pedroxramos]

Originally Posted by Tobi

Hi,


maybe you can use the new fields and use a calc function. E.g.

postProcess -func calc ...

It might also work on the fly, by using the calc function directly inside the controlDict but for this, the new generated fields should be registered in the database. Out of the box, I am not sure, if this can be done on the fly.

Tobias, thanks, but I'm afraid a postProcessing tool won't work because i want the fluctuations throughout time...

[agustinvo]

Hi Pedro,
I faced the same problem when performing LES. I think you want to do something similar, isn't it?


What you would need is to create a functionObject that reads U, UMean and T,TMean and computes
ut=(U-UMean)*(T-TMean)


You can do this, of course, directly in your solver by calling TMean and UMean, defining ut...


In my case I went farther (other turbulent statistics) and I modified a function object in order to generate the fluctuating fields.

[pedroxramos]

Gracias, Agustín!

do you mean I need to write a separate solver (copy-paste the files of an existent one) and adapt two things:
In the fields.H file, all variables are defined. I can add my new variable there;
In the solver, in the while loop of the time, after the flow equations are solved, add the operation for you new variable. For example: newVariable = c * U;

is this your suggestion?

[agustinvo]

De nada Pedro


Yes, it is. Remember that you must include not only ut, but also UMean and TMean in your new solver. You can run it as you prefer, but please wait until you have a converged UMean and TMean!

[pedroxramos]

Yes, ok, but I already get that from here:

U
{
mean on;
prime2Mean on;\leftarrow
base time;
}

no?

[agustinvo]

Effectively. You must run first a simulation averaging U and T (with your usual solver), and once UMean and TMean have converged you can use the second solver.

[pedroxramos]

aaaaah, I get it now, thanks a lot!!!

[pedroxramos]

Originally Posted by agustinvo

Hi Pedro,
I faced the same problem when performing LES. I think you want to do something similar, isn't it?


What you would need is to create a functionObject that reads U, UMean and T,TMean and computes
ut=(U-UMean)*(T-TMean)


You can do this, of course, directly in your solver by calling TMean and UMean, defining ut...


In my case I went farther (other turbulent statistics) and I modified a function object in order to generate the fluctuating fields.

Agustín, thanks a lot. I also want to do a quadrant analysis with the uRMS, so I also need to sabe the fluctuating fields. Would you mind sharing your function object?

[pedroxramos]

Originally Posted by pedroxramos

Tobias, thanks, but I'm afraid a postProcessing tool won't work because i want the fluctuations throughout time...

hi again, I'm still having some problems with this, would you mind sharing your scripts?