[splif]

Hello

I new in OpenFoam user and I try to get the bulk temperautre about a area.

the theory: my problem

tbulk = integration(U*T*dA) / integration(U*dA)

the nummerical:

I'm not really sure how to sovle this problem.
Postprocessing or ?
I think that you need a summation about the area,like:
sum = sum + T(n)*U(n)*dy (Fortran 2D)
How can i fixed this problem?
Have somebody a idea?

Thanks
valentin

[henrik]

Dear Valentin,

There are two issues. How to define the area and how to parallelise.

I would also advise you to use the flux rather than the velocity since it is guaranteed to be conservative.

Neglecting parallelisation issues and assuming that you want to work with patch (iP) the following will the job:

heatFlux = 
    sum(T.boundaryField()[iP]*phi.boundaryField()[iP])/sum(phi.boundaryField()[iP]);

Henrik

[splif]

Hi Henrik,

sorry for this Question. But i'm not in used to work with openFoam.
what means [ip] ? It is for direction, like [x].

I'm not sure if i get you wrong:
Your equation mean:

q=sum(t*flux)/sum(flux)=sum(t*rho*U)/sum((rho*U)) ?
Is there no multiply wirh dy and dx?

Thanks
Valentin

[gschaider]

Originally Posted by splif

sorry for this Question. But i'm not in used to work with openFoam.
what means [ip] ? It is for direction, like [x].

I'm not sure if i get you wrong:
Your equation mean:

q=sum(t*flux)/sum(flux)=sum(t*rho*U)/sum((rho*U)) ?
Is there no multiply wirh dy and dx?

No. The is all included in phi (that is the flux Henrik was talking about). For the definition of phi look elsewhere (it's been discussed zillions of times)

Bernhard

[henrik]

Dear Valentin,

thanks for your private message (in German). I hope you don't mind if I repeat what I understand is what you are trying to do.

Valentin is seeking to evaluate the local Nusselt number and needs the bulk temperature to do so. The local Nusselt number would be per wall face (additional averaging may apply) and the bulk temperature is a function of the axial position in the pipe (x-coordinate in his case).

The problem is now to evaluate the bulk temperature for a given axial position.

Is this correct?

Henrik

[splif]

Hello Hendriks,

that's right. I search for a summation (lilke (sum(sum( T.yz*U.yz*dy)dz)each Cells) about an area (yz).And every summation should go every cells in x-> direction.
Perhaps somebody has an idea.

Thanks
Valentin

[henrik]

Dear Valentin,

Okay. I would try the following. Create a lookup table for T_bulk as a function of x. To do so, you need a function that maps x into an index.

scalarField vol(nCellsx, 0.0);
scalarField Tbulk(nCellsx, 0.0);
forAll(T, cellI)
{
    if ( inBulkRegion(mesh.C()[cellI]) )
    {
         label index = floor(mesh.C()[cellI].x()/length*nCellsx);
         vol[index] += mesh.V()[cellI];
         Tbulkl[index] += T[cellI]*mesh.V()[cellI];
    }
}

Tbulk /= vol;

Then walk over the wall patch to calculate the local Nusselt number and use the same index function to look up Tbulk (but calculated with the face center's x-coordinate).

This is by no means elegant, it will not parallelise easily and there are better ways of doing this. However, this will get you a long way.

Henrik

[splif]

Hello Hendrik,

thanks a lot.Have a nice evening (in German).

Bye
Valentin