[lcthompson]

In exploring the sensitivity of solutions obtained from SU2_CFD on a CD nozzle configuration, I'd appreciate some sanity checking on the results I'm seeing. In the attached, the min/max columns are the extrema of the scalars in the flow.vtk file for each run, with the density, energy (hidden), and velocity derived from the conservative_n values per the discussion at http://www.cfd-online.com/Forums/su2...ervative_.html. (With the sub-trans-super transitions in all converging quickly in a well-behaved fashion, the min density/pressure and max mach/velocity values correspond to the outlet state with the other extremes on the inlet side.) Columns U-AA were manually added to check the values obtained directly from the runs.

For the 4 runs extracted here, the inlet/outlet pressure ratio varies from 7 to 10 times the nominal 101kPa ambient atmosphere with the lowest ratio driving the supersonic flow out of the nozzle (i.e. no shocks in diverging area) and completely through the downstream test section of a notional wind tunnel. All other inputs are held constant.

With these ratios, the nozzle geometry forces the solutions to a exiting mach value identical to 5 decimals. Not unexpected given the same exit/throat ratio, though I would not have been surprised either if the code included some effect that slightly varied the result.

Where I am a little uncertain is with the derived velocities also being essentially identical both when simply calculated off the conservative_1/3 values (column S) or using the mach/pressure/rho(/gamma) values (column X). Given the consistent pressure/density ratios across runs (column V), velocity sameness does follow; and it's maybe that consistency that I'm wondering most about. Seems the temperature (and other) effect(s) would have to be the same across runs to yield the same pressure/density ratio in turn driving the same exit velocity. Maybe correct/real, maybe not?

Being new to SU2 and somewhat lightly informed on the implications of gas flow theory, not sure whether this would be in fact a real-world phenomena, a result of SU2 modeling fidelity, or perhaps something in the extraction process that I may be doing wrong. So any help confirming/correcting these observations would be appreciated.