[Studi]

Hello everybody,

I have a question regarding the surface an curve mesh setup. It's probably pretty simple, but since I'm new to ICEM, I don't get behind the sense of the setup.
I want to refine the surface mesh on the radial surface, as you can see it's pretty coarse and doesn't map the surface well.
As you can see in the pictures as well, I tried to refine the surface itself with the surface mesh setup and the edges with the surve mesh setup. Regardless whether I'm refining out of the setup menus (options: remesh attached surface/remesh selected surfaces) or the global surface mesh, the mesh stays the same. What am I doing wrong?
Any help is much appreciated!

[Far]

1. respect line elements

2. adapt mesh interior

[Studi]

Originally Posted by Far

1. respect line elements
2. adapt mesh interior

Thank you, Far, but it didn't change anything...
I should have mentioned, that this is only a surface mesh.
A confusing thing for me is the fact, that, if I'm generating a volume mesh, the settings are respected by the mesh (see pic 1). But if I go on and want to change local settings of the volume mesh, e.g. a curve, and ICEM fits the mesh again, the output is 'scrap' and doesn't respect the new curve settings(see pic 2).
Then again, I don't understand, why it's so hard to mesh the shell (surface mesh) first?!

[Studi]

Maybe someone could take a look at the geometry. It was imported via STP. I ask, because during my 'research' I made i simple cuboid within ICEM to play with the curve and surface mesh setups. There it works without any problems.

Regards
Sebastian

[PSYMN]

I usually set fillets as mapped quad surfaces so I get nice rows on anisotropic elements... If I really want triangles, I would convert quad to tri at the end.

[Far]

Simon's suggestion is excellent. However the problem with your case is the sizing. Specially the global parameters.

[Studi]

Originally Posted by PSYMN

I usually set fillets as mapped quad surfaces so I get nice rows on anisotropic elements... If I really want triangles, I would convert quad to tri at the end.

Thanks for the suggestions. I'll try it, but not before next week with a feedback of the results.

Originally Posted by Far

Simon's suggestion is excellent. However the problem with your case is the sizing. Specially the global parameters.

Could you specify you're thoughts a little bit more, please? Did you load the model? I'm thankful for every hint...


Regards
Sebastian

[PSYMN]

I had not looked at your model...

I just downloaded it and tried it. Something strange (not normal behavior) on this simple model. When I get a chance, I will look into why it is not behaving as expected.

[Far]

tried with patch independent and sizing function (min size, refinement etc) and here what i got. I will post link of tin file shortly.

[PSYMN]

Yea, but he should be doing it with Patch dependent quad with mapped surfaces... For some reason, it did not properly use his sizing. Usually, when I see that, the user has just set the ignore size too large, but in my few minutes to look at this, I didn't see that or any other reason for the strange behavior. I'll get back to it this evening if I can.

[PSYMN]

Found the problem...

In addition to the obvious issues like not setting sizes on your end curves, the real killer was that you had set "Curvature and Proximity Refinement" with a min size of 1, so it was ignoring any smaller sizes.

This got me this mesh with size 0.25.
Studi_01.jpg

If I increased the size to 0.5, I got this.
Studi_02.jpg

If I decided to increase the number of nodes around the fillet to 4, I got this.
Studi_03.jpg

If I increased the mesh size on the long curves to 1, I got this.
Studi_04.jpg

And finally, I converted quads to triangles... (assuming that is what you wanted?)

Studi_05.jpg

[PSYMN]

Here is the tetin file... (rename the .txt to .tin)

[PSYMN]

Oh yea, I also did things like turned off the individual surface controls for mesh type, etc. You should start with global settings and only use individual entity controls when they need to deviate from the global...

So you could set quad as the global mesh type, but set all tri on specific surfaces where you didn't want mapped mesh (In this case, I set the size on the opposite side to 1 so the mesh wouldn't try to map triangles)...

Studi_06.jpg

[Studi]

Wow, thank you so much for your help! Both of you!
This is a huge step forward for me in understanding what to do. If we should meet some time, beer is on me!
Thanks again for your time!

[Studi]

Hi again!

There has come up a new question I'd like to ask regarding another region of the part I already introduced earlier.
I attached a picture. Concerning this, I've two questions:
1. Why does the mesh of the blue part reach out to a point of the green part? How to prevent it? As far as I've looked, there is no entity belonging to the blue part.
2. Why is there an offset from the border of the green part?

One general question I've left, too. Before you created a goof surface mesh. Is there a way to build a volume mesh based upon this surface?

Thanks to the specialists!
Sebastian

[mohammad]

Hello everyone,

Can anyone please help me about my question.

http://www.cfd-online.com/Forums/ans...tml#post518634

Best Regards,