Learning Forum

[aitor.amatriain]

Hello,

I am simulating a flow around a three-component wing. The flow is compressible (inlet uniform velocity 61.5 m/s), I am considering periodic BCs (spanwise), outler pressure 1 bar, operating pressure 0 bar, and non-reflecting BCs for inlet and outlet.

Although it does not fully converge (expected), the RANS solution seems fine. However, when I switch to unsteady mode with the DDES turbulence model and the BCD scheme for the momentum equation (second order upwind for the rest of the equations, second order for pressure), I get into trouble. After many time steps simulated, the instantaneous vleocity contour is as follows:

As you can see, there are velocity (and pressure) oscillations that are completely unphysical. I would like to confirm the following:

• I have tried other models such as SBES, even URANS, with different BCD scheme boundedness values, and the problem persists (even with URANS). With 1 (maximum possible value) is what you see in the pictures.
• If I use second order upwind scheme for the momentum equation, then this problem related with the oscillations dissappears. But I canot follow this path because DDES needs at least BCD to be useful.
• I have verified that the pressure fluctuations are some order of magnitude higher than those I should get.
• As expected, removing the non-reflecting BCs does not help.
• The maximum Courant number is 0.25.
• The convergence is excelllent. For example, the Continuity equation passes from 1e3 (first iteration) to 1e-4 (last iteration).
• I have tried a coarser and a finer mesh without success.
• The minimum orthogonal quality of the mesh is 0.45. It is a fine mesh, with around 1000 cells along the profile, and 20 elements in the normal direction of the boundary layer, with the wall y plus equal to 1.
• It is a compressible flow with the ideal gas equation (air). If I simulate the case in incompressible, then the problem dissappears.

Could you please provide me some help related to this issue?

 

Thank you

 

[Federico]

Hello Aitor, 

what does your computational domain look like. You mentioned that you are using spanwise periodic BC. Can you show this?

[aitor.amatriain]

Hello Federico,

Thank you for your answer. The top figure is a cut with the plane z = 0. The domain is extruded in both directions (from z=-c/4 to z=c/4, where c is the chord of the main wing element), and a periodic boundary condition is defined in both surfaces (z=-c/4 and z=c/4). I have done this directly in Fluent Meshing, so that the mesh coincides in both surfaces.

Regarding the other extensions, the bottom figure shows them. The inlet is approximately located 10c away from the wing, the outlet 20c away from the wing, and the slip walls 10c away from the wing.

[aitor.amatriain]

Hello Federico,

Do you have a solution for this problem? Or I can assume that this is a problem related to the numerical model implemented in Ansys Fluent?