does anyone know which method can best handle contact discontinuity?

thanks.

Flux difference splitting (FDS) schemes like those of Roe are better at contacts. Flux vector splitting schemes generally do a poor job but AUSM family is probably as good as a FDS scheme.

You may also wish to look at schemes such as Toro's WAF method and HLLC. Toro's book (has Riemann solvers in the title - do a search on amazon) is a *very* good starting point if you need more info/background.

Scott

HLLC scheme and PPM scheme are also capabile of dealing with contact discontinities, and PPM scheme is more accurate.

thank you all for the info.