Dear Friends,

I am working on preconditioning my existing compressible solver so that I can simulate incompressible flows with good accuracy.(Of course in the limit as M->0, not M=0 though). I have made use of primitive variable formulation in my solver based on [p u v T] along with Weiss-Smith preconditioner and one of the variants of the AUSM scheme.

I have some doubts regarding the characteristic BCs when preconditioning is applied. Could someone clarify as to the changed characteristic BCs to be made use of ? I am presently making use of freestream BCs at the farfield. Is this right ? My test problem is flow past a cylinder with the farfield at 25 diameters.

Regards and Thanks in advance,

Ganesh

Using freestream bc can hamper convergence since there are lot of fast (relative to flow velocity) acoustic waves at low mach number. They will get reflected from the farfield and can slow down convergence. Also lift might be wrong but may not be an issue for the cylinder case. You can take the preconditioned system and linearize it about the freestream and look at the characteristics of this linear system. Then you can devise boundary conditions based on these characteristics.