Hi

I am using a NS-code which is implicit to solve a turbomachinery application.

I have problems with the initialisation of my testcase which is a linear fade f (rho, rho*c1,rho*c2,...) between GRID-INFLOW and -OUTFLOW conditions. The problems occur in the ghost-cells of the grid inflow. Even though the MACH-level of the initialisation is quite good ( I know this because I know the solution) the Mach-number rises in the ghost-cells (grid inflow) to Ma > 2.

Has anybody an idea how I can fix that? Or why that may happen? I thoguht the "right" velocity-distribution should help to get the exact solution much faster...

Moreover I have heart something about (non-)reflecting boundary-conditions but I have no imagination what it is and when I have to use them? Could you help me to understand it? Do you know any good paper which explains the problem?

Thx

Few References:

Poinsot and Lele (1992) - Journal of Computational Physics

Giovli (1991) - Journal of Computational Physics

Rudy and Strikwerda(1980) - Journal of Computational Physics