[sliley]

Hi all,

I'm running a validation case for a high speed inlet into still air - validating the model against experimental data. I'm working in CFX in ANSYS-16. At this point I've run several cases, and get good agreement between experimental data and model if I have the inlet on an adiabatic wall. However, as soon as I try to run it as a free-standing inlet in a free-stream volume the velocity contour starts looking very different both in terms of shape and size.

Turbulence model:
SST (all cases)

Meshes:
Mesh 1 (Case 1 and 2)
mesh 1.jpg

Mesh 2 (Case 3)
mesh 2.jpg

Boundary conditions:
Case 1
End - adiabatic no-slip wall
Start and open sides - opening (entrainment)
Symmetry planes - symmetry
Inlet - constant velocity

Case 2
Start, end and open sides - opening (entrainment)
Symmetry planes - symmetry
Inlet - constant velocity

Case 3
Start, end and open sides - opening (entrainment)
Symmetry planes - symmetry
Inlet - constant velocity

Additional notes:

• The inlet is applied on a flat surface
• I'm not looking to resolve any detailed turbulence, just validate againt experimental data

Results:
Velocity Contour
velocity contours.PNG
Take Case 1 as the data set I want to emulate

Two features I would like to highlight:

• The increased length of the contours in case 2 and 3
• The ballooning feature around the inlet in case 3

Has anyone seen this before? Does anyone have an idea of what has caused it?

I'm particularly curious about the uniformity of the velocity through the start boundary- it stays between 100 and 100.02 ft/s flowing perpendicular to the boundary. That makes me think the BC is doing something funky I didn't account for. That and the fact that there doesn't appear to be any air being drawn in from the the other entrainment boundaries.

Please let me know if you require any additional information.

Cheers,
Simon

[ghorrocks]

Have you done a sensitivity check that your simulation is accurate? This FAQ goes through some basics: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

Your mesh looks pretty coarse so I suspect there are big errors there. That would explain the effects you are seeing.

Isn't the feature you call ballooning around the feature just the flow diverting around the feature? Isn't that what you would expect?

[sliley]

Hi Glen,

Thanks for your response.

I haven't completed a sensitivity analysis on the mesh. However I do not think that the errors are a result of a course mesh - we get good agreement with the experimental results on the same mesh, with the different boundary condition.

I understand that changing the boundary condition has a significant impact on the flow, but I'm curious as to why the change is so large in this case. Especially as I'm not introducing another inlet to the situation.

You may be right that the "ballooning" is a result of the flow diverting around the feature (especially given it's bluff). However, If that were the case wouldn't you expect it to look more like the vertical plane in this image? Not that very large diversion seen in the contour I posted? I may be wrong in this expectation. Granted I would not expect to be drawing such a uniform flow in from the "start" boundary.

I've also edited the original post to clarify why I'm questioning these results. Please see the edit.

[ghorrocks]

However I do not think that the errors are a result of a course mesh

The wider contours and longer mixing length are exactly the sort of thing I expect to see with too coarse a mesh. And too coarse a mesh is the most common problem for inaccurate results.

Likewise your "balloon" is exactly what I would expect to see over a bluff body. It causes the flow to accelerate around it.

[sliley]

Whoops, didn't edit the post in the end. Fixed now. I'm still curious about what's happening at the entrainment boundaries.

Thanks for the help, I'll try a finer mesh when I get an opportunity next week. Do you think that adding a cone to the front of the feature would limit that diversion flow pattern?