[ecto]

Hi everybody!
I'm trying to simulate a supersonic flow in a duct (see attached image).
It's something like 15 mm in diameter and 20 cm long. It does a 180° bend at the end. To be specific, I'm interested at the resulting mass flow rate at the outlet (see image) with 25 bar at the inlet.
In my calculations the flow should be supersonic with a Mach number near 2.
I did poly mesh with about 1'200'000 cells (surface target 0.5 mm), 15 prism layers with 1st layer 5E-5 mm.
I initialized with velocity values of 800 m/s in the duct, 1bar pressure in the domain, expert initialization Grid sequencing.

What I managed to do is: going almost to convergence in 40'000 iterations (no errors) with the mass flow with settings:
-Constant density
- Coupled flow (No energy) with Co=0.1 and Explicit Relaxation =0.2 (as per Starccm+ guide)
- k-w turbulence all y+

What I cannot do is switching to Ideal Gas and Coupled Energy, because I immediately get the error you can see in the attached image (Overflow on the turbulence). And temperature limited errors. I wanted to switch to these because with that Mach Number it's better to enable compressibility.
I tried also to start with Segregated flow and Temperature, and also with Segregated Isothermal.

I tried also to do a Directed Meshing but I get error of "Loop... missing cad edge" when I want to Automesh the Source Surface.

Does anyone have an idea?
Thanks a lot

[fluid23]

Without seeing your mesh I would have to guess that maybe you haven't refined the shocks sufficiently. Remember, you have extreme gradients at each shock and you must provide sufficient discretization to avoid numerical errors.