[mattijs]

You can make sure by setting startTime=latestTime in the controlDict and running couplePatches. It will tell you that the ordering is correct or not (and attempt to correct the ordering and write a mesh). It will only work correctly if the two halves of the cyclic are not connected.

OpenFoam does not really make a difference between the rotational and translational; it works out a transformation tensor from the normals on the two patches.

[duderino]

Hi Mattjis,

I tried the tool couplePatch with lots of error messages, like:
Please use topological matching or adapt the featureCos() setting
Continuing with incorrect face ordering from now on!
cyclicPolyPatch::order : Found 0 on patch cyc_ring where the cos of the angle between two connected faces was less than 0.9
cyclicPolyPatch::order : Writing zone 0 face centres to OBJ file "cyc_ring_zone_0.obj"
cyclicPolyPatch::order : Writing zone 1 face centres to OBJ file "cyc_ring_zone_1.obj"
cyclicPolyPatch::order : Number of faces per zone:2{30}


I think OpenFoam does not know whether the angle between the patches is 0 deg or 180 deg. Can I somehow define this angle (with featureCos() setting?)?
You also said :"It will only work correctly if the two halves of the cyclic are not connected."
What do you mean by It? couplePatches? Are in a rotational case the two halves of cyclic not always connected ?

Thank you!

[mattijs]

Correct. It tries to divide your single patch into two halves base either on
- connectivity: a translational patch will have two disconnected regions
or on
- angle between connected faces.

In your case a) does not work since I assume it is one single region and b) does not work since the angle is 0 everywhere.

Either make your case less than 180 degrees or split the patch into two regions (by introducing a tiny (or even 0) gap between the two regions)

[duderino]

The problem is I can't make my angle less than 180 deg. If I do so my "symmetry" is broken.

But If I split the pach into two regions with a tiny gap, the pach won't be connected and be treaded as translational pacht (as I understand it). This also not correct in my case!

Any suggestions? Can't I predefine an angle?

Thanks!

[mattijs]

The patch will not be treated as a translational patch. All cyclics are treated the same, all scalars are exchanged as is and all vectors are rotated by the rotation vector (derived from the normals).

[duderino]

How could I use a cyclic boundary for an annular gap? Here are the two regions also disconnected, but a translational patch would produce the wrong results!

[duderino]

In my case there are more than 1 one cyclic patch. For some the regions are connected (->causes the problem that OpenFoam doesn't know if the angle is 0 deg or 180 deg) and some are disconnected. In my case all scallars AND all vectors are exchanged as is (there is no rotation at all).
So I think OpenFoam calculates the wrong rotation vector.

My qyestion is:
Can this vector be influence by user input?

Thanks Mattijs!

[mattijs]

No.

But you have full source code ;-)

Look for calcTransformTensors in coupledPolyPatch.C and put some Info statements in there to figure out what it is doing.

[duderino]

Ok I will have a look, allthough I am a C++ beginner.

Thanks anyway!

[duderino]

It is realy hard for a beginner to understand the code;-(.

1) Am I right that the default matching method is geometric?
2) Could I use topological matching for my problem?
3) What means featureCos()setting? And how to adapt it?

[mattijs]

1) yes
2) no. since the starting ordering is invalid
3) this is the (cosine of the) angle between two neighbouring faces. If you have e.g. a 45 degree wedge like geometry with a cyclic it uses this angle to split the cyclic patch into two regions.

To change: add to couplePatches.C

after

coupledPolyPatch::setGeometricMatch(true);

the line

cyclicPolyPatch::setFeatureCos(ddd)

and add a #include "cyclicPolyPatch.H" to the top of the file.

(ddd is the cosine of the angle where you want to split the region. Note that this does not work for a 180 degree cyclic where the faces are perfectly aligned)

(And yes it is hard stuff. There are still problems with it)

[duderino]

I have to pass. This is to complex for me. Is there somebodey in the OpenFoam communiy, who solved the problem with the 180 degree cyclic boundary - Andrew maybe or Mattijs. Or maybe someone is planning to implement it?

Best regards!

[duderino]

I have to pass. This is to complex for me. Is there somebody in the OpenFoam communiy, who solved the problem with the 180 degree cyclic boundary - Andrew maybe or Mattijs. Or maybe someone is planning to implement it?

Best regards!

[mattijs]

Hi Duderino,

I never had the need for 180 degree cyclics. As said part of the problem is that there is no gap between the two halves.

Hello

could someone tell me what is doing cyclic?

best regards

Emo