I am trying to simulate a shock tube problem on either CFX or Star-CD, to study the interactions of the expansion wave with various obstacles. I don't, however, know how to approach setting up such a problem on these packages. I do know that ideally I need an outlet where I can set flow conditions (Temperature, pressure, velocities, etc.) and an inlet that will initially have set conditions but must not reflect any waves passing it. I would appreciate any suggestions on how to achieve this.

I'm not sure but I think that these codes do not have non-reflecting (based on characteristics/Rieman invariants) boundary conditions implemented. It is quite difficult to do this in 3D and make it robust. But anyone can correct me if I'm wrong. I would be interrested in (commercial) codes which do have this capability.

Clifford,

Just some comments:

- The classic shock tube problem has no inlets or outlets. It's just a box.

- At an outlet you can only set one flow condition, otherwise the problem will be overconstrained. I think you mean you want to set the inlet flow conditions instead.

- No "open" (inlet or outlet) boundary condition in the world will perfectly transmit shock waves (i.e. large amplitude acoustics). Even properly implemented non-reflecting conditions only work on weak acoustic waves. See papers on this by Thompson in JCP (mid 80s).

- You can easily setup a "classic" shock tube in CFX-5. Run it transient and use the CEL setup function to create two independent regions with differing pressure/temperature/density at whatever ratios you decide. Make sure you pick a timestep that will adequately resolve the shock propagation speed. i.e. the shock travels no more than a few cells per timestep, otherwise you will get severe diffusive effects.

- Run these calculations in CFX-5 with the High Resolution advection scheme. If you are not satisfied with the "sharpness" you can play with the parameter "Maximum Blend Factor" (at the expense of some non-monotone wiggles). The default is 1, and maximum is 2.

Neale