[Larrabee]

Hi, all. I am a new member. Hope that I can learn from everyone here.

I am now conducting a simulation on multistage compressor.

I choose the first one stage as a test, but the solution does not converge. I think the problem might be the interfaces between the stationary region and
the rotation region. However, there is no sign of error on that. I will post some pictures on that.

Based on that I have several questions.
(1) Can two faces with different sizes be created into interfaces?
(2) Can two faces which leave a few distance between them be created into interfaces?
(3) To conduct computation, should I mesh the imported parts as a whole or I can I mesh them independently?

[me3840]

1. Yes, but in general, especially for turbomachinery where the gas path is always continuous, this is not a problem. All the interfaces should be identical.

2. Yes if you make the tolerance high enough, but this isn't recommended. Why are there gaps? Again, for turbomachinery all the interfaces should be perfect, there's no real reason to have anything else.

3. You can do either. Are you intending to do RBM or MRF? If you do RBM then you may as well mesh the independently since during the run the interfaces will become non-conformal anyway.

Are these interfaces mixing plane? Your compressor geometry seems to have blades that are nearly aligned with the flow, which is very strange.

[Larrabee]

Originally Posted by me3840

1. Yes, but in general, especially for turbomachinery where the gas path is always continuous, this is not a problem. All the interfaces should be identical.

2. Yes if you make the tolerance high enough, but this isn't recommended. Why are there gaps? Again, for turbomachinery all the interfaces should be perfect, there's no real reason to have anything else.

3. You can do either. Are you intending to do RBM or MRF? If you do RBM then you may as well mesh the independently since during the run the interfaces will become non-conformal anyway.

Are these interfaces mixing plane? Your compressor geometry seems to have blades that are nearly aligned with the flow, which is very strange.

Thank you, me3840.

I am using MRF for this simulation. A single rotor or stator doesn't have problem with its mesh. But as a whole there is "self-inserting" problem. I don't know why.

The interfaces are in-plane interface.

[me3840]

Do you mean self-intersecting? If so, this means there's probably a problem with your contacts.

Make sure that the different parts you have are imprinted. If you surface repair the imprint description, it should result in 0 pierced faces.

Does your case mesh if you enable "per-part" meshing? If so, then the above is definitely the issue.

Check the output window during initialization. Does it report that your interfaces are conformally matched?

[Larrabee]

Hi, I have changed the modeling method to keep the flow tunnel smooth.
However, the "self-intersecting" problem still exists. Do you have any advice on repairing this problem manually.? Because if I use the automatic repair tool the whole shape will be very weird.

[Larrabee]

Moreover, after I use the surface repair tool to repair the whole model, when I want to generate volumn mesh, it will report that the stator is self-intersecting. However, I check the stator and the whole model,and there is no problem. This confuse me a lot.

[me3840]

Do not use the automatic surface repair tool, it is not designed for this scenario.

Your picture there shows the self-intersections (all the red triangles), I'm not sure how you think there aren't any. It looks like you have an imprinting problem. Have you imprinted the two parts together? That will remove the self-intersections.

[Larrabee]

Hi, thanks a lot!
I use the imprint tool to imprint the parts and now I can get the volumn mesh.