[kg_059]

Hey everyone,

I am trying to set up a transient simulation that has two fluid domains, one stationary and one rotating. I have connected them via a transient rotor-stator-interface. I have also turned one the non-overlap boundary condition, however I still get flow in the not overlapping areas and the non-overlap fraction in Post indicates, that the non-overlap condition is not working correctly. I have attached a simple sketch of my setup. as well as a picture of the sim result.

I have set the non-overlap condition for both sides of the interface. Could that be the error or is it many a problem that could come from the meshes I use (unstructured).

I have searched through the cfx documentation bit have not found the information I am looking for...

Kind Regards.

[ghorrocks]

Please attach the output file from the simulation. That will help us diagnose the problem.

[kg_059]

Ok, here is the file.

[ghorrocks]

This model does not have two domains, it has three. There is a rotating solid domain as well.

Can you explain what you are trying to do with this model? Why do you need the solid domain?

Also, please post a diagram showing where the three domains are and how you have defined the interface boundaries.

[kg_059]

I have attached a picture of the setup. It is a test-setup to model a wing that can change its aoa via a rotationg mechanism that is conected to a mount (modelled just as wall). the wing is the solid domain but it rotates together with the rotating fluid domain. that works well and does not cause problems. The problem, as far as i can see comes from the interface between the two fluid domains.

[kg_059]

here is a screen-shot of the interface i am talking about. I have turned on non-overlap condition for both sides of the interface.

[Opaque]

Perhaps you need to read the documentation about using the transient rotor-stator interface.

if the domains touching the transient rotor-stator interface have rotational periodicity, the two sides of the interface are replicated in an attempt to have full overlap, that is the point of the sliding mechanism. The mesh faces that after replication are non-overlapping use the settings defined on the NonOverlap conditions.

[kg_059]

Ok, i am not sure if i understand that.
The problem is, that the non-overlap condition does not work, even though the two sides of the interface are clearly not overlapping...
I have turned on the non-overlap condition for both sides of the transient rotor-stator-interface.
The rotation is controlled via a time-depentent function: if(2 [s] < Time && Time < 4 [s], 7 [deg/s], 0 [deg/s])
So there is no periodic rotation.

[kg_059]

I have also tried refining the mesh, because the element size maby was to big, but it dit not fix my problem.

[Opaque]

Originally Posted by kg_059

Ok, i am not sure if i understand that.
The problem is, that the non-overlap condition does not work, even though the two sides of the interface are clearly not overlapping...
I have turned on the non-overlap condition for both sides of the transient rotor-stator-interface.
The rotation is controlled via a time-depentent function: if(2 [s] < Time && Time < 4 [s], 7 [deg/s], 0 [deg/s])
So there is no periodic rotation.

Are you sure the non-overlap condition is NOT working? or it is not giving you what YOU expect? Expectations may be incorrect if the feature is not clearly understood.

Q: Are you using rotational periodic boundary conditions on the domains connected to the domain interface?

[ghorrocks]

The output file shows he is not using perioidicity. I cannot see why the flow is doing what it is in this case.

Can you answer my question from way back: Why do you need the solid domain?

Are you actually looking for the heat transfer in the wing feature?

[Opaque]

Thank you, Glenn.

Now, is the vector plot using Hybrid or Conservative values? Would you mind posting both plots to see the differences between them?

[kg_059]

Originally Posted by ghorrocks

The output file shows he is not using perioidicity. I cannot see why the flow is doing what it is in this case.

Can you answer my question from way back: Why do you need the solid domain?

Are you actually looking for the heat transfer in the wing feature?

No i am not looking for the Heat Transfer so i guess i could scrap the solid domai and just model it as a wall for the fluid Domain…

[kg_059]

Originally Posted by Opaque

Thank you, Glenn.

Now, is the vector plot using Hybrid or Conservative values? Would you mind posting both plots to see the differences between them?

I will Look into that.

[kg_059]

Up until now I have been using the automatic pitch change option for the transient rotor stator interface. I am starting to think, that cfx has a problem with that. But i do not know wich specific pitch angles i should use in that case. I want to rogate the rotating domain by 15° and the static domain should abviously stay still. So 15° and 0° ???!?

[kg_059]

Originally Posted by Opaque

Thank you, Glenn.

Now, is the vector plot using Hybrid or Conservative values? Would you mind posting both plots to see the differences between them?

i barely see a difference

[Opaque]

Originally Posted by kg_059

Up until now I have been using the automatic pitch change option for the transient rotor stator interface. I am starting to think, that cfx has a problem with that. But i do not know wich specific pitch angles i should use in that case. I want to rogate the rotating domain by 15° and the static domain should abviously stay still. So 15° and 0° ???!?

The pitch angle has nothing to do with the rotation. It is about the pitch ratio across the interface. For your case, it should be None since you are not replicating any of the components. The mass flow leaving the stationary domain is exactly the mass flow entering the rotating domain. No need for scaling.

Have you plotted the streamlines starting at the inlet, using the vertex option and an appropriate Reduction Factor?

[kg_059]

Originally Posted by Opaque

The pitch angle has nothing to do with the rotation. It is about the pitch ratio across the interface. For your case, it should be None since you are not replicating any of the components. The mass flow leaving the stationary domain is exactly the mass flow entering the rotating domain. No need for scaling.

Have you plotted the streamlines starting at the inlet, using the vertex option and an appropriate Reduction Factor?

thank you very much for clarifying! :-)

i will set the pitch change to none.

to further isolate my problem I have set up another test case (pictures and out-file attached). I have done a very basic setup with an block shaped stationary domain that has a cutout shaped like half of a cylinder and a rotating domain that fills that cutout. the rotating domain rotates with a set speed. as boundaries I have defined inlet, opening and the transient rotor-stator-interface. everything else is a wall. both sides of the interface again have the non overlap condition turned on, put in post the non-overlap fraction still indicates that the non-overlap condition is not working...

i am clueless as to what I am doing wrong... the setup is very simple now...
maybe it is the pitch change option. I have not set it to none for the new test case but will definitely try that!

ah and before I forget: I have plotted the streamlines from the inlet and from the interface sides. There is definitely still flow coming through the parts of the interface that don't overlap and thus should act like a wall.

Thanks for all your input so far guys!

I really feel like the problem is solvable, it must be. I am probably forgetting something but even after simplifying I still do not find my error :/

[ghorrocks]

I just did a quick test case on this and it worked fine for me:
NonOverlap.jpg

NonOverlap2.png

My test case def file is a super coarse mesh so I can attach it here:
NonOverlap.zip

Have a look and see if you can spot the difference. I have to do some work now so do not have time to

[kg_059]

thank you very very much! i have not looked into your file yet but i assume it is the pitch change setting. i have set it to none in my test case and now the non overlap condition seems to work :-)

i will try to implement it into the more complex test case from wich i started in the thread.

again thank you so much for the help guys!