[MarcoC501]

Dear all,
I am struggling with a problem I have in CFX. I meshed a 5 blade passages with one blade staggered in ICEM, the only problem is that when I go to CFX and run the simulation it automatically puts a wall the the interfaces, while I just dont want that. I tried in some ways but none of them worked, for example:
- deleting the mesh interface in ICEM: I deleted the surface mesh, but still there is a volume mesh that automatically generates those interfaces. If I merge the nodes, it seems that ICEM does not understand what I want
- deleting the primitive surfaces in CFX is not possible, and if I disable the default generation (which puts automatically a wall at the interfaces), I get an error message which says "you did not assign to the interface to any BCs".
The mesh is not periodic, and I set no periodic vertices, at this point I really do not know how to proceed. I would really appreciate if you could give me some hints on that. I can also send pictures if needed.
Thanks a lot in advance

[Opaque]

Would you mind attaching an image of your mesh when imported? and if possible to point out the region with the issue.

Keep in mind you can create a "translational periodicity", or a "general connection" interface. Either of the two should solve your problem. I do not recall ICEM providing a mechanism for CFX to recognize an interface; therefore, it must be created manually.

[MarcoC501]

Hi Opaque,
thanks for your reply. You can see attached the image of the mesh: the problem are the interfaces, the ones that now I tried to make periodic ("Translational periodicity").
With "general connection" it was not working as I wanted, and from the simulation results I can see that a wall is generated at each interface.
Another thing I noticed is that the 1:1 connection does not work and there is overflow also when I do not enable "permit no intersection" in the mesh connection definition. The mesh has been created by merging all the passages. From the solver manager, I can read:
Discretization type = GGI
Intersection type = Direct
Non-overlap area fraction on side 1 = 1.00E+00
Non-overlap area fraction on side 2 = 1.00E+00
which is not right I think. If I enable the "nonoverlap conditions" at the boundaries it works as I expect (GGI with non-overlap area fraction very small) but on the other hand it puts an unwanted wall.

[Opaque]

Here is what I understand from your description,

1 - The passages meshes are not conformal with each other
2 - You merged all the passages into a single mesh in ICEM
3 - You imported the single mesh into CFX-Pre

4 - Because the passages meshes were not conformal, CFX-Pre finds several volume meshes (corresponding with each passage or even more)
5 - Because of (1), and (4), you MUST manually create domain interface to effectively "GGI glue" the meshes; however, because the meshes are not conformal there is no way to guarantee the intersection will find a perfect match, so it is possible there may be hub->shroud differences on the sides of the interface, or axial extent differences as well. Those differences can only be treated as wall.

6 - To model those passages w/o wall you need to check in ICEM that the passages have at least identical axial, radial extents on both side of the interfaces. The mesh may be non-conformal, but the geometry must be conformal.

[MarcoC501]

Yes, I did the procedure you assumed. Anyway, the passages are obtained by rotation of an initial passage, so the geometry should be the exactly the same rotated. For the passage with the staggered blade, I simply changed the blade, but not the rest. For the mesh, the only problem was the interface between the staggered passage and the adjacent ones: I matched the spacing with quite good "visible" results. Of course, there might be some parts which are not conformal.
Because of that, I think that all the passages have have identical axial, radial exents: how could I double check this in ICEM
Thanks a lot

[MarcoC501]

Moreover, if I try to go in ICEM and merge the blocks, the overall mesh is not only bad, but also the interface is kept in the hexa edge even though I merge the blocks (all the other geometry "references to the interface" were canceled by me)

[Opaque]

Unfortunately, I am not an ICEM user.

Not sure what merging the blocks meant here, but if the meshes are not conformal, there is no way the mesh is now a single mesh volume in the CFX sense.

Q: When you import the mesh in CFX-Pre, do you get a single 3D volume or multiple 3D volumes? Careful when you look in the outline. You are looking for 3D primitives, not composites. Very likely there is a 3D composite that represents all the primitive sub-volumes created in ICEM.

[MarcoC501]

This is what I get from cfx by opening the .cfx5 file from ICEM. I tried in many ways to work on the geometry as you said, but every trial leads to the same subdivision. As you can see there are 5 3D primitives corresponding to the blade passages: the problem is that I would like to get rid of the 2D primitives related to the surface of the interfaces

[Opaque]

I think you got what I expected, 5 3D primitives, i.e. 5 mesh volumes: 1 mesh volume/passage

You can then create a single domain, selecting the 5 mesh volumes or the composite for joining them (if it exists).

Then, manually create general connection interfaces between the passages, and a rotational periodic interface between the two ends.

You can try setting the Mesh Connection for each interface to Direct if available. If the software cannot do the Direct interface, it will trigger an error when writing the definition file. In that case, you can fall back to Automatic.

[MarcoC501]

I already tried to do what you suggest, and I got the following error:

No control surfaces have been found for the domain interface |
| |
| Interface1 |
| |
| If you expect this to occur, please set the INTERSECTION CONTROL |
| CCL parameter 'Permit No Intersection' to work around this message.|
| Non-intersecting interfaces are only supported for 'dynamic' |
| interfaces which are re-intersected at every time step |
| (e.g. transient rotor stator interfaces or non-stationary |
| interfaces in moving mesh cases). |
| |
| Possible setup errors leading to this situation are: |
| 1. An incorrect axis of rotation for a rotating domain or domain |
| interface. |
| 2. A frame change model is used in which there is unequal pitch |
| shapes, or circumferential orientation, but the Pitch Change |
| model is set to None. |
| 3. The transient rotor stator model is used when the geometry |
| does not span 360 degrees and the Pitch Change model is set to |
| 'None'. |
| 4. The detected normal gap between interface sides is larger than |
| the allowed tolerance. See the documentation for controlling |
| this tolerance.

If I enable "permit no intersection", it will create walls. I do not think I am in case 1 or 2 or 3. I rotated the same passage of 24 degrees and there is no overlap with the blocking in ICEM

[Opaque]

I think your interface setup is incorrect somehow. From the figures you have shown, there is no way you get that message unless the locations for the domain interfaces are not well-posed.

Is Interface 1, the rotational periodicity one from the ends, or one of the internal ones?

[MarcoC501]

Thanks a lot! I think I solved the problem.
I was doing an incorrect setup as you said, i.e. selecting the 2D primitives corresponding to the sides of each passage in the interface definition.
So for example, I was setting up the interface of passage 1 (interface 1) considering the 2 SIDES of the same passage, while I should have selected just 1 SIDE from a passage and the corresponding interface side of the adjacent passage. Now the simulation works as it should
Thanks a lot again