[fshak92]

Hi

Does anybody know why using each of div(U) or div(phi) provides same results?
I'm confused because the first one is volVectorField, and the latter is surfaceScalarField.
(Both of them are computed at the same time step and I'm using this in the Poisson eq.(Laplacian(p')=div(U)) )

Tnx

[Astrodan]

I don't know about the numerics, but as far as I understand a volField is stored in the center of the cell, while a surfaceField is stored on the faces. Thus both methods are just different ways of accessing the "same" data, and should ideally return the same results.

To convert them you can work with fvc::reconstruct (surface -> vol) or fvc::interpolate (vol -> surface).

A little attention should of course be paid where phi includes density (though I've seen it called rhoPhi for those cases), usually in compressible or multiphase problems.

[fshak92]

Originally Posted by Astrodan

I don't know about the numerics, but as far as I understand a volField is stored in the center of the cell, while a surfaceField is stored on the faces. Thus both methods are just different ways of accessing the "same" data, and should ideally return the same results.

To convert them you can work with fvc::reconstruct (surface -> vol) or fvc::interpolate (vol -> surface).

A little attention should of course be paid where phi includes density (though I've seen it called rhoPhi for those cases), usually in compressible or multiphase problems.

Thank you very much for your response.
I've noticed that small differences can be seen in the result of those approaches.
Still I would like to know which one is completely correct if Poisson eq. should be solved? and why?
fvm::Laplacian(p') = fvc::div (U)
or
fvm::Laplacian(p') = fvc::div (phi)

Tnx in advance for any idea.