I know that in some commercial finite-element packages the Streamline Upwind Petrov-Galerkin method (SUPG) is used for the convection terms. Cross-stream numerical diffusion is much better avoided with this method. I myself am using a finite-volume package and I wonder if anyone knows the equivalent version of SUPG for finite-volume simulations.

Thanks very much

Bart Venneker

Hi,

The equivalent existed in the early 80s and it was called SUD (Streamwise Upwind differencing of Skew Upwind differencing). Unfortunately, it suffers from unboundedness and it is too diffusive, so it has been dropped (from the mainstream) quite a long time ago.

Hrv

I have testes some Skew Upwind Schemes myself (SUDS, FMSUD, Cupid, STOIC, SUWD) and I noted the unboundness near discontinuities. That's why I find it curious that in finite-element these problems are apparently solved, whereas in finite-volume (or difference) --- which is usually in advance of FE --- this line of reasoning has been rejected.

Thanks for your response