[Jonathan]

Hi everyone,

i was just wondering whether someone could explain why there seems in general for unstructured meshes, no mention / worry about non-orthogonal corrections to the face flux values for convection terms as there are for the diffusion terms .

for example:
in OpenFoam, there are specific correction schemes to deal with mesh non-orthogonality when discretizing the laplacian terms in the N-S equations, but nothing for the convective term.

surely, since both calculate face flux values as part of the FV discretization process, there must be an effect just as much on the convective terms due to mesh non-orthogonality as with the diffusion (laplacian) term.

many thanks, and regards
jonathan

[FMDenaro]

Originally Posted by Jonathan

Hi everyone,

i was just wondering whether someone could explain why there seems in general for unstructured meshes, no mention / worry about non-orthogonal corrections to the face flux values for convection terms as there are for the diffusion terms .

for example:
in OpenFoam, there are specific correction schemes to deal with mesh non-orthogonality when discretizing the laplacian terms in the N-S equations, but nothing for the convective term.

surely, since both calculate face flux values as part of the FV discretization process, there must be an effect just as much on the convective terms due to mesh non-orthogonality as with the diffusion (laplacian) term.

many thanks, and regards
jonathan

I am not sure if I understand your question... however, in a FV formulation the diffusive flux requires the normal component of a gradient on each surface, the convective flux does not.

[Jonathan]

hi Phillipo

Originally Posted by FMDenaro

I am not sure if I understand your question... however, in a FV formulation the diffusive flux requires the normal component of a gradient on each surface, the convective flux does not.

thanks very much for your answer - i think you understand me!

ok, so if i understand you correctly, its only the gradient vector at the face which needs to be parallel to the face normal vector, not the flux itself?

many thanks regards
jon