Learning Forum

[BiomedModeller]

Hi  

[Rob]

It could be the defeaturing trying to close off the angle. If you put a section plane along the tube axis (show full cells) how does it look? 

[BiomedModeller]

Thanks for the prompt reply!
Unfortunately, it is a part of huge non-symmetric shape and I have hard time defining cut planes with standard normals. Does this work for you?

[BiomedModeller]

[BiomedModeller]

I think it will become even more problematic when I add the prism layer.

[BiomedModeller]

Sorry to spam a lot, I am just trying to provide enough info to shorten the convergence time to the ultimate solution.
To clarify further, I am using ICEM, not the ansys mesher. Path independent, all Tri  is set. I could not find the defeaturing tolerance in ICEM. Could you direct me towards that to test modulating it?

[Rob]

Not a problem and those are the images I was after.  I don't know how to use ICEM CFD: I last used it's predecessor (Geomesh) about 20 years ago! 

Not sure what's causing the problem to the right of your images: the mesh should be good where the angle between the bodies is large. Where you see the pinch in the bottom image you may need to trim the model slightly to blunt the sharp angle.

Given you're using tet's I'd switch to ANSYS (Workbench) Meshing as you'll have a few other tools to clear up the mesh. 

[BiomedModeller]

Thanks a lot rwoolhou!
Modifying the geometry in this specific case might be straight forward to avoid the sharp angle. However, I need to fix the same issue for a more  complex patient-specific geometry, wherein by no means I can manually fix the entire artery.

Is there any setting in ICEM I can manipulate to get it? Could you please forward this post to any colleague who is expert in ICEM?

 

[Keyur Kanade]

you may want to check fluent meshing for this. 

[Keyur Kanade]

[BiomedModeller]

I fixed the problem with the combination of Global sizing, curvature/proximity, and edge criteria.

The problem now is it takes hours to mesh, and it freezes and crashes on the full model.

Anyone can help me with how I can run ICEM on cluster? The problem is X11 Forwarding.

Thanks

[BiomedModeller]

This prism layer problem persists. Any help from experts please?

 

I think it will become even more problematic when I add the prism layer.

[Rob]

Providing there's little/no flow in that region I'd leave it as is. If the cells are skew have a look at the cell zone separation options. 

[BiomedModeller]

Thanks for the reply. CFX works with that low quality mesh, but Fluent gives an error. I will try to divide the zone into two and only put the prism outside this vicious location

[Rob]

It's more that CFX uses a node spacing rather then cell shape: it's why CFX can struggle with high aspect ratio cells that are OK in Fluent. 

[BiomedModeller]

Dear Rwoolhou, 
The solution of cell zone separation might work for simple geometries (Tried it, worked), but impossible for the real application (See 8th post from the top).

Is there a chance to upload the geometry somewhere private to get your (or other colleagues') comment on it? I am not allowed to publicly share the case.

May I also add another moderator to this talk? I would be pleased to involve Peteroznewman to this loop please.

Thanks

 

[Rob]

Feel free to ask Peter: it's a forum so I think you just have to hope he looks in. 

There isn't a private part to the Community. As you're the ASC you can see if you can log a support ticket with us directly: the local team can then have a look. 

[BiomedModeller]

Thanks a lot for the reply. I will.

How to ask Peter? I know no way to add him to this conversation or send him this post as a message.

[peteroznewman]

Hello,

Do as rwoolhou suggests and see if you can log a support ticket.

I am also willing to look at your mesh, but I don't use ICEM, I am good at ANSYS meshing. If you reply with your email ID, I will send you an email that you can reply privately with a link to the model. I will remove your email ID from the post as soon as I read it. I will be checking this site for the next 10 hours.

[peteroznewman]

Got it. Sent you an email.

[BiomedModeller]

Thanks!
Attaching the files now

 

[BiomedModeller]

you may want to check fluent meshing for this. 

 

 

Dear KKanade,

Per your suggestion I started using Fluent meshing. The problem I am facing now is that the mesher is not respecting the curves, although they are defined.

See this image:

 

Any suggestion?

Thanks

[Rob]

That looks like the surfaces have collapsed onto each other. If the surfaces still exist (check visually) you may need to wrap them. Difficult to say without the files, and that's not something I can do via the Community (we do keep asking but it's export law related). 

[BiomedModeller]

Thanks for the reply.

They have not collapsed and the surfaces do exist.

Wrapping has already been done. 

Here I suspect the problem is the tolerance of the curves. I did not have this issue (for curves) in ICEM. But as we couldn't fix the mesh near where two surfaces were tangent and very close, we switched to Fluent meshing per the suggestion of "KKenade". Now we have solved those meshing problems and, as you can see from the above image, the problem of mesh fidelity to the curves and low resolution (tolerance issue?) problems for curves arise.

Regarding the communication problem of file sharing, I cannot agree more. It really hurts both ends of this communications considerably reducing the usefulness of these correspondences.