Hello !

Has anyone out there used the NS characteristic BC strategy described by Poinsot and Lele in their paper published in the Journal of Computational Physics 101, pg 104-129, 1992.

I am currently using that approach for the flow past a backward facing step at the exit. I have chosen to let the pressure at the outlet go to P_infinity. At the corners of the top and bottom walls, however, i still set dp/dy = 0; I believe this approach (setting dp/dy=0) is not compatible with the NSCBC approach - as i observe distinct pressure gradients at the exit.

I would like to know if anyone has attempted to use the NSCBC strategy and if they have any experiences to share that might provide some useful input.

thanks, Ravikanth

Hi Ravikanth

I have used NSCBC, but I think the paper mentions how to deal with the corners.

The solution, is to generate L(i)'s (hope you understand the notation ) for both the directions, i.e you apply the NSCBC along the X direction and generate a set of L(i)'s and apply NSCBC along Y direction and generate the set of L(i)'s and at the corner point it would be something like

d(U)/d(t) + [L(i)'s]x + [L(i)'s)y = 0.

This is cumbursome, but is effective. I have worked on this for supersonic flows.

Note that the paper also suggests (in case of subsonic flows, to relax the imposition of pressure (at corner nodes) so that the apparent inconsistency becomes less evident).

Hope you have followed the other papers on NSCBC by the same authors, published in JCP. There are atleast two others.

Bye

Sundar.