[recap2000]

Hi,


With ansys 18, fluid flow(cfx), I am doing two simulations for heat transfer as a result of fluid flow. The two simulations involve one cupe geometry and one hexagon geometry. Cube and hexagon has air inside of them. And on the faces I cut some piece and inserting ceramic. And there is one more part inside of this ceramic. Now this ceramic doesn't have any interface with the inside air because of this part inside of the ceramic. For the cube geometry I am seeing that there is only solid solid interface for the ceramic and for the hexagon there is fluid solid interface. And cube geometry simulation results are as expected and symmetric. But for hexagon it is very unsymmetric and wrong in my opinion. I am connecting this to the supposedly non happening fluid solid interface on the hexagon geometry. Where can I be wrong? I couldn't find the mistake. I checked the geometry and rebuild it. It seems fine. Can you help?

[ghorrocks]

I am not sure exactly what you are asking, but it appears your interfaces are not set up correctly. The default generated interfaces should be checked in CFX-Pre to make sure they have the right surfaces in them. You can edit the surface list to make sure it is correct.

[recap2000]

Hi,


Thank you for your answer. Let me explain more on the problem I have got. I have pentagon and hexagon structures. Both of them is problematic. I will post pictures of pentagon. You can see pentagon structure in the below pictueScreenshot from 2019-11-15 15-22-22.jpg The empty faces in the pentagon is filled with ceramic pieces that exactly fit. That is the ceramic completes the pentagon wall. Ceramic's inside face is boundary to aliminium heat sinks. There are five of them. Aside from these heat sinks, inside of the pentagon is filled with air completely. Now the problem is that air is not supposed to touch inside of ceramics. I checked the CAD file and this is the case. But, still in CFX-Pre, I see air-ceramic interface, that is to say fluid-solid interface on ceramics. You can see this in the second picture.



In the second picture there are only ceramic and its interfacesScreenshot from 2019-11-15 15-43-21.jpg. Green interface is solid-solid, which is as expected. But blue interfaces are fluid-solid. According to my symmetry there must not be any fluid-solid interface. I am assuming that there is something wrong with the way I mesh this geometry. I might be doing something wrong in the mesh settings. I tried to change size and adaptive options. But I cant find what could cause this in CFX. Do you have any suggestion to fix the problematic interface?

[ghorrocks]

As I said in my previous post, if the default interfaces are wrong then edit them to correct them. Add or remove surfaces until they are correct.

[recap2000]

Hi,


I tried what you said. But I cannot edit the default interface in anyways with (right click) edit or edit in command line options. When I delete the fluid solid interface a new interface with the same properties is created automatically and this is useless. What is the correct way to edit default interfaces under domains?

[Opaque]

How many mesh volumes do you have in your setup?

How many mesh volumes are listed under the solid domain? 1 or more?

[recap2000]

I have 33931 mesh elements and more than one element for solid domains.

[recap2000]

I solved the problem. In case someone else encounters this type of problem you can try this: In the meshing windows choose body selection tool and ctrl+click on the bodies with wrong contacts. Right click and click on show contacts of the selected bodies. If you are sure that these contacts should not exist then right click and delete those contacts on the left menu. Then mesh and in the default interfaces you wont see wrong interfaces.

[jones007]

I have a different but related problem. I have a multi-block domain created in SolidWorks and imported into the Ansys mesher. Most of the connections that it automatically generated seemed correct, but a few had extra faces. Those were easily fixed. If I manually look through them, they all seem correct, although sometimes they indicate faces are in contact with each other when they only share an edge. I left these, but I'm not sure if I should.

The real problem shows up when I run the case. All but one of the fluid-fluid interfaces seem OK, but on one, fluid does not appear to be able to traverse the interface. Streamlines appear to artificially deflect such that they are parallel to the interface, even though they should pass through it.

in CFX-pre I have been digging through the mesh volume and surface database, and was able to add a missing to one of the volumes - the face where the error occurred, but re-running the solution, it appears to have the same problem, so I think I am either trying to fix the problem in the wrong place, or the fix I am trying is superficial.

If anyone has run into a similar problem or knows of a thread where it is perhaps already addressed, the help would be much appreciated.

[ghorrocks]

Note that when you add faces to an interface you have to do it at the interface object level, not the interface boundary in the domain level.

You should be able to manually add a few faces easily to an interface to fix it up. Do you know why you are finding that difficult?

In the output file and the results file there is a variable "non-overlap fraction". This shows which bits of the interface do no have a matching pair. This can be useful to debug the interfaces.

Another point: In ANSYS Mesh you can define named selections which in some cases can make defining interfaces easier. But this does not work well when adjacent domains share a common face as the named selection only gets put on one side.

Final point : In cases where getting the faces correct is really tricky I import the mesh into ICEM and sort it out there. In ICEM you can edit everything, group surfaces together and split them up; even down to individual element faces.

[jones007]

Thanks for the quick response. I suppose I'm a bit of a newbie on multi-block meshes. In all my earlier attempts, the mesher seemed to create all of the face-to-face connections correctly, and I really didn't have to fiddle with anything to get it working right. This is the first case where things didn't automatically get set up right leading to clearly non-physical results.

The interface that is failing is a spot where a large block has a cylindrical cavity cut in it, and then two cylinders are fit into the cavity, where the outer face of both cylinders share the same face of the cylindrical cavity. The interface between one of the cylinders works and the other does not.

I looked at the non-overlapping area fractions. On one side it is around 6e-6, and on the other about 0.375, so probably reflecting the problem.

Update: I just tried splitting the outer block into 2 parts such that it matched the 2 inner parts. Now the non-overlapping areas are both down around 1e-7. Looks like the issue was have faces on two blocks sharing a common face on a second block.