Learning Forum

[Prezz2077]

I am currently trying to run a simulation involving compressible flow through a converging-diverging nozzle, and into the atmosphere using AIM's "Connect to Fluent" option (meshing is done in AIM, but all results are evaluated in Fluent). My model consists of the nozzle domain with a paraboloid shaped domain attached representing the atmosphere. I have a pressure inlet boundary at the nozzle inlet, no-slip on the nozzle wall, and a pressure outlet of 0 psi (gage) on the surfaces of the paraboloid. I have enabled the energy equation, and am using the k-epsilon model with the standard settings. I have a 5% turbulence intensity at the inlet, and 0% on the outlet. My issue is that whenever I try and run the model, it diverges after about 500 iterations. If I stop it before that, the results that I get do not match the expected results (based on experimental data). Is there something special I need to do to model flow into an external domain in Fluent?

[Raef.Kobeissi]

-You need to make sure that your mesh is convenient for such a problem.
- I advise you to simulate this as a transient case.
- I advise you to use K-Omega SST
-I advise you to use inlet pressure

[Prezz2077]

 Thank you for the response. I followed your suggestions, but the model still diverges. Can you explain how to determine if the mesh is convenient to such a problem?

 

Thanks

[Rob]

Plot the results you have with node values off and post on here. Have you got enough cells across the nozzle to capture the flow field? 

[Prezz2077]

I believe I have enough nodes across the nozzle. I have attached the residuals plot as well as the velocity contours.

[Rob]

Use a plane or isosurface (never display on the fluid or default-interior surface) and zoom in around the jet.  Given your max velocity is of the same order as the speed of light I'd say something went wrong. 

[Prezz2077]

Here is the velocity displayed on a plane bisecting the model:

" alt="" width="744" height="342">

The range for the scale was set to local, as opposed to global. If I run the simulation using only the nozzle as the domain, I get accurate results, which leads me to believe that it is the geometry of the added exterior domain that is causing the divergence, but I am not sure what geometry would work better.

[Raef.Kobeissi]

Why is the exterior looks so large what are your boundary conditions in this problem? Can you attach the case so i can have a look?

[Prezz2077]

The exterior domain is so large compared to the nozzle because I wish to get an accurate picture of the flow after it leaves the nozzle, and how it interacts with the ambient air.

 

The boundary conditions are as follows:

Inlet - Pressure inlet (376000 pa)

Nozzle Wall - Wall, no-slip

Outlet - Pressure outlet (0 pa), turbulent back-flow intensity (0%)

 

The model can be found here: https://www.dropbox.com/s/uzrq8lnoaberu0j/No6_Diffuser_Domain_Fluent.wbpz?dl=0