[elden]

Hi,
I'm dealing with external aerodynamics problems with dispersion of gaseous substances.
Simulations are carried out considering URANS and a Scale Adaptive approach for the turbulence model (kOmegaSSTSAS). The tracer gas is transported by means of the scalarTransport FO (i.e. as a passive scalar)

I'd like to postprocess not only the Reynolds Stress Tensor by means of the prime2Mean utility but also the turbulent diffusion flux of concentration : <u'*c'>; <v'*c'>; <w'*c'>
Related to the postProcess of these last three quantities I'm facing two problems:

- The first one is mainly conceptual and it's related to the fluctuating part of the velocity field. Since the turbulence model is the kOmegaSSTSAS, should I compute directly the fluctuations of the velocity field? Or should I consider the assumprtion made in the turbulenceIntensity FO (u'=v'=w') accounting for the fluctuating part as uPrime=sqrt(k*2/3)?
- The second is related to the code needed in order to evaluete the turbulence intensity of a scalar (c') and subsequently the turbulent diffusion fluxes (< u_i '*c' >). My intent was to somehow exploit the FO (fieldAverage.C/H and fieldAverageItem) used for prime2Mean computation but I got lost. I can't find the proper way to do that.

Thanks for reading.
Hoping to find some answers

[elden]

Well,
I think I've menaged to do what I needed. Instead of changing fieldAverageTemplates.C with an additional variable (like, let's say, primeField), I tried to evaluate < u' c' >, < v' c' >, and < w' c' > in the following way:
mean value of U and c thanks the fieldAverage FO --> definition of the new variable U' (3 components vector) * c' as (U - UMean)*(c - cMean) --> mean value of this new variable.
I hope this is a valid approach since the fieldAverage code computes the iterative average.

[pedroxramos]

Originally Posted by elden

Well,
I think I've menaged to do what I needed. Instead of changing fieldAverageTemplates.C with an additional variable (like, let's say, primeField), I tried to evaluate < u' c' >, < v' c' >, and < w' c' > in the following way:
mean value of U and c thanks the fieldAverage FO --> definition of the new variable U' (3 components vector) * c' as (U - UMean)*(c - cMean) --> mean value of this new variable.
I hope this is a valid approach since the fieldAverage code computes the iterative average.

hi,


would you mind sharing your script for this part?

[elden]

Originally Posted by pedroxramos

hi,


would you mind sharing your script for this part?

This is the code. The C2H4 scalarField is the one being transported within the flow field

[pedroxramos]

thanks a lot!

[pedroxramos]

Originally Posted by elden

This is the code. The C2H4 scalarField is the one being transported within the flow field

hi again, are you sure you only need those files? We tried to run it on the pitzDaily (pisoFoam) and it didn't save the new variable...

[pedroxramos]

Originally Posted by elden

This is the code. The C2H4 scalarField is the one being transported within the flow field

is the right side to be put in the createFields.H?

[elden]

In this thread I was talking about the passive transport of a scalar quantity. One way to tackle this task consists into the use of the scalarTransport FO. This FO looks for a file in the 0 folder named as the field name used (in my case C2H4)

[pedroxramos]

Originally Posted by elden

In this thread I was talking about the passive transport of a scalar quantity. One way to tackle this task consists into the use of the scalarTransport FO. This FO looks for a file in the 0 folder named as the field name used (in my case C2H4)

got it now, thanks!

[hhu_lulu]

Originally Posted by elden

This is the code. The C2H4 scalarField is the one being transported within the flow field

Thank you, Elden. Your script is of great help to me.
In my LES case, the Umean and cMean only makes sense when they are statistically converged. So I set a timeStart for this coded FO. I imposed your code in my simulation, trying to calculate time-averaged <u*c> and <u'*c'>. However, the simulation aborted at the half of the calculation, and the error shows:

new cannot satisfy memory request.
This does not necessarily mean you have run out of virtual memory.
It could be due to a stack violation caused by e.g. bad use of pointers or an out of date shared library

Here's part of script calculating <u*c>, the case aborted at runTime ≈ 98s.

fluctuation_flux1
	{
		type			coded;
		libs			("libutilityFunctionObjects.so");
		enabled			true;
		timeStart		74;
		timeEnd			200;
		writeControl	adjustableRunTime;
		writeInterval	200;
		name			flux1;
		
		codeExecute
		#{
			const volVectorField& U = mesh().lookupObject<volVectorField>("U");		
			const volScalarField& s = mesh().lookupObject<volScalarField>("s");		
			
			obr_.store
			(
				new	volVectorField	
				(
					IOobject
					(
						"flux1",
						obr_.time().timeName(),
						obr_,
						IOobject::NO_READ,
						IOobject::AUTO_WRITE
					),
					U*s
				)
			);
			const volVectorField& flux1 = obr_.lookupObject<volVectorField>("flux1");
		#};
	}

fieldAverage1
    {
      type            fieldAverage;
      functionObjectLibs ( "libfieldFunctionObjects.so" );
      resetOnRestart true;
      resetOnOutput false;
	  startTime     75; 		
      endTime       200; 		
      writeControl    adjustableRunTime;
      writeInterval   200;
      
	  fields
      (
            flux1
        {
            mean        on;
            prime2Mean  on;
            base        time;	  
        }
      );
    }

I doubt there's something wrong with pointer stack overflow, sorry for my poor programming.