[Tomjugj]

I have questions about the outlet boundary conditon in SU2.
For the SWBLI case in V&V (taken from NPARC Alliance Validation Archive)in the SU2 official site[https://su2code.github.io/vandv/swbli/], the flow is supersonic at the outlet, why the cfg file specify the outlet pressure[https://github.com/su2code/SU2/blob/.../config_sa.cfg]? Usually, the flow downstream at supersonic speed is calculated from upstream, but SU2 does not provide such supersonic outlet boundary condition. In Wind-US configuration file, they specify the downstream pressure to be extrapolated, rather than specify the outlet pressure for the Ma=5 SWBLI hypersonic flow.[https://www.grc.nasa.gov/WWW/wind/va...y02/uns_sa.dat]

For similar problems, that is supersonic outflow after shock wave, different from the freestream conditions, how should I set the outlet BC?

[bigfootedrockmidget]

Hi,


The supersonic outlet is automatic. At each point at the outlet, we compute the local Mach number. If it is supersonic, the back pressure is not necessary (and will not be used) since there are no incoming characteristics. If Ma<1, the pressure is used to update the conservative variables. Even though the value given for backpressure might not be in use once the simulation has converged, it might be in use during the iterative solving, so it is best to choose this pressure carefully.

[Tomjugj]

Thank you for your reply; I understand now.

Is it acceptable if there is only one point that is 'not physical'? Does that mean the solution is diverging? In my case, it involves Mach 6 viscous flow over a flat plate. I am using an unsteady Euler explicit solver, and the overall flow appears to be fine, although the shock wave looks blurred. I previously thought that it was due to the boundary conditions, but now I think there might be other reasons.

Do you have any advice or insights on how to address this issue?

[bigfootedrockmidget]

Hi,
The unphysical point might affect the quality of the solution. One point does not sound so bad, but it is a sign that the solver has difficulties trying to find a numerical solution.

The blurred shock can have all sorts of reasons, the mesh might be too coarse or some numerics can play a role, like the order of the scheme.


Best is if you can compare to some measurement or reference simulations and see if you have a good accuracy for the flow conditions you are interested in.