artifical dissipation on unstructured meshes

Frederic

Hi all,

I'm trying to implement an artifical dissipation scheme on a 3D tetrahedral mesh. (I want to be able to use central differencing instead of Roe's scheme, which has some implicit dissipation ). Two things: - is there any known example in 3D? Most of the literature I found so far was related to 2D unstructured meshes. - also, most cases precondition the artifical dissipation flux term with a coefficient containing the average eigenvalue at the interface between two neighboring cells (also to keep the dimensions consistant). What is the meaning here of using the local eigenvalues? I think I understood that the goal is to yield more dissipation where the distance between two points is great in the direction normal to the face between those two points (Roe does that all alone). Thanks for anyone's input.

Fred.