[micheletuttafesta]

Hi
I am starting my PhD in CFD and I'm currently trying to implement
a simple 1D euler solver.
I use a straigthforward upwind scheme to avoid spurious oscillations
and it works fine in a free expansion test case.
The problem arise when I try to insert a reflective wall
at the end of the shock tube.
Immediately after the reflection strong oscillations appear
in front of the backward moving wave.
I suspect they are caused by incorrect numerical boudary condition,
in my case i impose zero velocity at the right last cell boundary and
zero pressure gradient at the same cell boundary.
See the attached figure near the critical time of reflection:
the red line is rho*u, green line is the pressure
I would like to know if there are better ways to implement
reflection or if it is normal to see such oscillations.
Thank you in advance.
Michele

[ares]

I suggest you to read characteristic boundary conitions. Poinsot and Lele, Lodato, Yoo are some auhors. If you do a google research with characteristic boundary conditions entry you may find a lot of papers.
Physically characteristic boundary conditions are more realistic then imposed values or gradients.
You can see the difference if you do some runs with imposed and characteristic boundary conditions.

Emre