CFD Online Discussion Forums - Convergence Issue of Wall Model in Transonic AC Configuration

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- SU2 (https://www.cfd-online.com/Forums/su2/)

- - Convergence Issue of Wall Model in Transonic AC Configuration (https://www.cfd-online.com/Forums/su2/248302-convergence-issue-wall-model-transonic-ac-configuration.html)

Convergence Issue of Wall Model in Transonic AC Configuration

Hello,
I am trying to run a high-transonic jet aircraft configuration using SU2, with standard wall function and SST turbulence model. The aircraft has an inlet at which I apply inlet total conditions boundary condition.
The simulation runs smoothly but cannot reach convergence, especially the energy residual and k and omega residuals, which remain above 1 even after 10000 iterations.
Also, when I check the yplus distribution I see there are very high yplus values (2000+) at the inlet section boundaries, where the wall condition and the inlet BC meet. The solver gives warning that the wall model did not converge in about 300 points. I tried replacing the inlet BC with a wall+wall model BC and the warnings didn't show.

The simulation is set with freestream Mach of 0.95, steady state RANS, JST scheme and venkat limiter and adaptive cfl between 1 to 50.

Do you have any thought about how to impove convergence?

You should make the first cells close to the wall smaller (in the direction normal to the wall) in the regions with large y+, try to get this value close to y+=100 or so. The wall model is pretty robust, but this might be too high to get good results.

There might be other things going on, it's hard to tell right now.

Thank you for your reply.
The first cells near the wall are smalll enough, I created the mesh for yplus=50. As I said, when I replace the inlet BC with a wall BC (zero heatflux) I don't get convergence issues with the wall function and yplus reaches a maximum of 80. I assume the combination of the inlet condition and the wall condition are the cause of the problem.

But if you remove the inlet, how is there flow in your domain?

Also, do you simulate a model aircraft in a wind tunnel? Why is there an inlet connected to a wall in your configuration? Do you have a picture of your geometry, with inlet, outlet, freestream, wall, .. boundary conditions shown?

If you only have high y+-values where the inlet meets the wall, this might be due to either the mesh quality or because your boundary conditions are discontinuous near the wall. If you inspect the solution at this point, do you also see very large changes in velocity, turbulent kinetic energy, turbulence dissipation?

Maybe the inlet is simulating the engine exhaust?

Quote:

Originally Posted by bigfootedrockmidget (Post 845797)

Thank you for your reply. I sumbitted a photo of a generic aircraft (since I cannot share my specific configuration) with the boundary conditions marked. I apply standard wall function to all zero heatflux walls. The mesh was created to match yplus of 50. The farfield wall is set as the outer faces of a large box containing the aircraft. Also, I added the input file I am using. It's an external flow problem of an aircraft in high transonic flow, where I need to apply an inlet condition to the inlet section simulate the effect of air suction into the engine inlet.
When the area marked as "Inlet Wall" has zero heatflux boundary condition, yplus values are small and there are no convergence issues. When I apply inlet total conditions boundary conditions then I get convergence issues and high yplus values in the contour that connects the noslip wall boundary condition area and the inlet boundary condition. Therefore I assume my mesh quality is sufficient and the issue is the inlet boundary condition implementation. Maybe my selection of the proper boundary condition was wrong?
I can't seem to find any abnormal changes of velocity/Mach/pressure contours when inspecting the solution. I am adding tke and turbulence dissipation to the outputs and I will check them as well.

Convergence Issue of Wall Model in Transonic AC Configuration

Quote:

Originally Posted by Orangade (Post 846012)

Thank you for your reply. I sumbitted a photo of a generic aircraft (since I cannot share my specific configuration) with the boundary conditions marked. I apply standard wall function to all zero heatflux walls. The mesh was created to match yplus of 50. The farfield wall is set as the outer faces of a large box containing the aircraft. Also, I added the input file I am using. It's an external flow problem of an aircraft in high transonic flow, where I need to apply an inlet condition to the inlet section simulate the effect of air suction into the engine inlet.

When the area marked as "Inlet Wall" has zero heatflux boundary condition, yplus values are small and there are no convergence issues. When I apply inlet total conditions boundary conditions then I get convergence issues and high yplus values in the contour that connects the noslip wall boundary condition area and the inlet boundary condition. Therefore I assume my mesh quality is sufficient and the issue is the inlet boundary condition implementation. Maybe my selection of the proper boundary condition was wrong?

I can't seem to find any abnormal changes of velocity/Mach/pressure contours when inspecting the solution. I am adding tke and turbulence dissipation to the outputs and I will check them as well.

Thanks for sharing the configuration and providing more info. According to your picture (I understand the canopy like shape is facing the flow), I’d say that CFD-wise the “engine inlet” should be an Outlet BC. Did you try that?