[fedez91]

I am looking at the implementation of fixedGradient BC in OpenFoam 7. In fixedGradientFvPatchField.C (line 144) we have that the value at the boundary is calculated as:

 Field<Type>::operator=
     (
         this->patchInternalField() + gradient_/this->patch().deltaCoeffs()
     );

What I don't understand is the following: If we have a positive gradient, the value at the first cell center should be bigger than the value at the boundary, therefore deltaCoeffs() should be negative. But by definition deltaCoeffs() cannot be negative (https://openfoamwiki.net/index.php/O...n;;deltaCoeffs).

This leads to the second part that I do not fully understand: From surfaceInterpolation.C (line 318), deltaCoeffs is calculated as

nonOrthDeltaCoeffs[facei] = 1.0/max(unitArea & delta, 0.05*mag(delta));

where unitArea points outside of the mesh (5.1.1.2 User manual https://cfd.direct/openfoam/user-gui...sh-description) and delta points from the boundary center to the cell center, and therefore unitArea & delta should always be negative (they point in different directions). Finally

unitArea & delta< 0 < 0.05*mag(delta))

and

deltaCoeffs = 1.0/(0.05*mag(delta)).

There must be something wrong with this reasoning, but I can't find where. Would you please help me

[Gerry Kan]

Dear Federico:

I believe surface normal is defined as positive pointing out of the domain, so deltaCoeffs will be positive.

Hope that helps, Gerry.

[fedez91]

Originally Posted by Gerry Kan

Dear Federico:

I believe surface normal is defined as positive pointing out of the domain, so deltaCoeffs will be positive.

Hope that helps, Gerry.

Thank you Gerry for your answer. To my understanding, delta points from the boundary to the cell center, or is it the other way arround?

vector delta = C[neighbour[facei]] - C[owner[facei]];

If delta points from the cell center to the boundary my second doubt would be cleared. But what about the definition of fixedGradient? The value of the gradient we define in the BC is actually the gradient in the direction pointing outside of the mesh? It looks a little counterintuitive.

[Gerry Kan]

Dear Federico:

Yes, the 'direction' of deltaCoeffs is from the cell center to the patch center. The direction oof the gradient follows the patch normal, i.e., out is positive.

However, I must confess I have always thought (too simplistically) that deltaCoeffs is 'just the distance between the cell center and patch center', but this forum post gave a more accurate description:

mesh.surfaceInterpolation::deltaCoeffs()

Gerry.

P.S. - After some digging, I think I know where you are coming from ...