[ThomasHultsch]

Dear FOAMers,

I am implementing a sub-grid scale model for convection dominated scalar transport (of a volume fraction of soil).
I am using the scalarTransportFoam solver with given velocity field.
With the SGS-model I calculate wall-normal gradients of the transported scalar to get realistic diffusive flux on a coarse mesh.
I already wrote a new boundary condition derived from the fixedGradientBC to correct the gradient at the wall and its working pretty well.

Now I also have to correct the gradient at the "next to wall" faces, i. e. at the faces between the wall cells and the 2nd row of cells. Here the standard linear interpolation creates way to high diffusive flux.

I already found out how to get the cellIDs via fvPatch.faceCells() and mesh().cellCells(). But I dont know how I should implement the correction of the gradient or diffusive flux for these cells.
My plan was to use an deferred correction of the diffusive flux, that is calculated explicitly (from old timestep and SGS-model) and added to the source vector (b vector) at field calculation. But I dont know how to achieve this.

Question: What is the best or easiest way to correct a gradient or diffusive flux at faces that are not member of an patch, but internal faces? The cell values should not be changed, because they are needed for advection.

All hints are appreciated.

Thanks in advance

Thomas

[ThomasHultsch]

Hello FOAMers,

a little update from my side:

I am trying to use the fvOptions to correct the flux in the near wall cells.
I use my boundary condition to write a cellSet with the cellIDs of the patch cells and the adjacent cell row.

Now a problem occured in the function that sets the cellSet. The compiler tells me labelUList not having a member named "found". But it has one. I also tried the find() function with find() == -1. Its the same with that.

My code where this happens:

const labelUList& fC = this->patch().faceCells();

forAll(fC, fCi)
{
    const labelList& nC = this->patch().boundaryMesh().mesh().cellCells(fC[fCi]);
    forAll(nC, nCi)
    {
        if (!fC.found(nC[nCi])) corrCellSet.insert((fC[fCi], nc[nCi])));
    }
}

I really dont know, why the compiler complains about the found() function.

Maybe one of you has a idea, where the mistake is.

[ThomasHultsch]

Hi,

I found out, the troubling functions find() and found() where just added in v1712. Since I use 5.0, they are obviously not available.

The correct code is:

if (findIndex(fC, nC[nCi]) == -1) {...}