cfmeshing hell - 30P-30N Validation Case

[techGuy]

So why does this happen ?

There was an issue with the extrusion. I have pushed a fix.

The fix works. Meshes are much cleaner visually as well as statistically. Good work, Oliver.

Code: Select all

17:26:14   ***Max skewness = 4.06910924726898, 1 highly skew faces detected which may impair the quality of the results
  <<Writing 1 skew faces to set skewFaces
17:26:14      Coupled point location match (average 0) OK.
Checking faces in error :
17:26:14      non-orthogonality >  90 degrees                        : 0
17:26:14      faces with face pyramid volume <     0                 : 0
17:26:14      faces with concavity >  90 degrees                     : 0
17:26:14      faces with area <     0 m^2                            : 0
17:26:14      faces with skewness >  50 (internal) or 100 (boundary) : 0
17:26:14      faces with interpolation weights (0..1)  < 0.0001       : 0
17:26:14      faces with volume ratio of neighbour cells < 0.0001     : 0
17:26:14      faces on cells with determinant <     0                : 0

Failed 1 mesh checks.

Good 14 degree AoA mesh.png (396.47 KiB) Viewed 1421 times

[techGuy]

cfmesh works so much better now. Almost zero mesh problems.

I had one surface refinement fail when I tried to set the size ratio to 0.0625. When that didn't work I put a refinement solid around it with a size factor of .25 and then set the surface refinement size factor to .25. Then it worked.

[techGuy]

@oliveroxtoby Is there any chance that surface meshing without boundary layers enable isn't working properly ?

What I'd like is a 5cm layer of small elements around my wing surface. What I am getting is no refinement at all, unless I'm selecting boundary layers. Boundary layers give high aspect ratio cells. I'd like low aspect ratio (almost square) cells.

Shouldn't a surface refinement without boundary layers just put a thickness of smaller cells around the surface ?

Surface Mesh Settings.png (37.11 KiB) Viewed 1378 times

Surface mesh without boundary layer.png (577.88 KiB) Viewed 1378 times

[techGuy]

I tried to simulate a 14 degree AoA by leaving the wing at 0 degrees and tiling the incoming flow up by 14 degrees, but that resulted in really bad residuals and force coefficients. It makes sense that it would because the flow is no longer aligned with the outlet.

That shouldn't be an issue. It should work if you make the left and bottom patches both inlets, and right and top patches both outlets. See the UAV demo case.

I tested this by making the left and bottom faces inlets (via 1 inlet patch) and the right and bottom faces outlets (vai 1 outlet patch). It still didn't work. I agree that it should, but it doesn't. Cl is about 1/4 of what it should be and Cd is negative.

Without walls to contain the flow, the wing heavily influences the flow path. Almost no air goes out the back of the box, most of it goes out the top.

bad flow.png (127.79 KiB) Viewed 1333 times

[oliveroxtoby]

@oliveroxtoby Is there any chance that surface meshing without boundary layers enable isn't working properly ?

What I'd like is a 5cm layer of small elements around my wing surface. What I am getting is no refinement at all, unless I'm selecting boundary layers. Boundary layers give high aspect ratio cells. I'd like low aspect ratio (almost square) cells.

Yes. In the file you uploaded it seems to be working as expected. The refined region around the wing remains even if boundary layers are turned off.

[thschrader]

Here is my attempt:
inlet 90 m/s, rho=1,25 kg/m^3
5% turbulence, k-omega with simpleFoam.

mesh: cfmesh 2D-planar, 4x refinement stage, approx 200000 cells

result: total lift 1600 N/m, total drag 50 N/m, gives
c-lift=1,73, c-drag=0,054

The coefficients depending on the definition of the cord length, I used the total length from tip to tip, 1220 mm.
In the wolf-dynamics tutorial I cant find a definition of the chord. Maybe thats the reason why it matches so good...
Another point is the vmax which reaches 200 m/s. This is out of range for a incompressible solver.
Maybe Hisa is the way...

Test file > 1MB, only the results:

multiwing.JPG (102.2 KiB) Viewed 1247 times

[herbk]

Hi Thomas,
about 325 km/h speed is may be a bit much, but i son't know for wich kind of airplane the profile is made....

It wold be nice to see the Lift if the AoA increases and at wich angle it get to Stall...

[thschrader]

...
It wold be nice to see the Lift if the AoA increases and at wich angle it get to Stall...

Which part of the wing? See pics above, the complete profile is in one sketch.
Rotating only one part is diffucult. Or rebuilding the complete model.

[herbk]

Not only one part, - the complet wing.

Edit:: is there a file with the wing?

[thschrader]

Not only one part, - the complet wing.

Edit:: is there a file with the wing?

At 4 deg AoA I get c-lift=2,17, better convergence as with 0 deg AoA.

But at 8 deg I get meshing errors (?). A remedy is to use a finer surface mesh at the wing,
This ramps up the cell count, cfmesh consumes over 16 GB, I am running out of RAM.

I played with the mesh settings, but I allways get the "max-skewness" error.
The rest is ok.

EDIT: before meshing I had updated the cfdof-wb

Code: Select all

***Max skewness = 7.69853766817373, 4 highly skew faces detected which may impair the quality of the results
  <<Writing 4 skew faces to set skewFaces
    Coupled point location match (average 0) OK.
Checking faces in error :
17:07:30      non-orthogonality >  90 degrees                        : 0
17:07:30      faces with face pyramid volume <     0                 : 0
17:07:30      faces with concavity >  90 degrees                     : 0
    faces with area <     0 m^2                            : 0
17:07:30      faces with skewness >  50 (internal) or 100 (boundary) : 0
17:07:30      faces with interpolation weights (0..1)  < 0.0001       : 0
    faces with volume ratio of neighbour cells < 0.0001     : 0
17:07:30      faces on cells with determinant <     0                : 0

Failed 1 mesh checks.