CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] unknown pre-mesh distortion (http://www.cfd-online.com/Forums/ansys-meshing/120778-unknown-pre-mesh-distortion.html)

treenw July 14, 2013 22:49

unknown pre-mesh distortion
 
3 Attachment(s)
Dear veteran meshers,

After long and repeated meshing work ,i got quite a frustrated results:the mesh over one side of the aircraft is mysteriously distorted ,while the opposite side is fine.The two sides are symmetrical.
I did the association work for curves and faces all over again,but the results are the same.Can you help locate the problem of this issue,I really don't want to redo the blocking and splitting....

treenw July 15, 2013 02:30

problem solved after re-create the nacelle geometry.

Far July 15, 2013 03:29

you are making full hexa?

treenw July 15, 2013 07:52

1 Attachment(s)
Quote:

Originally Posted by Far (Post 439732)
you are making full hexa?

yes,FAR.quite stupid I know ,very grid-inefficient,already got 600 million,barely run on my 16G RAM PC.
Just want to see the whole picture.
Still got the Y-block generated mesh around the dorsal fin with ANGLE critiria like shit ,any suggestions?

diamondx July 15, 2013 11:39

the procedure is time consuming !!! expect bad element quality if you don't simply geometry. you should add o-grid for the engine's nozzle exit...

Far July 15, 2013 12:13

Quote:

Originally Posted by treenw (Post 439793)
yes,FAR.quite stupid I know ,very grid-inefficient,already got 600 million,barely run on my 16G RAM PC.
Just want to see the whole picture.
Still got the Y-block generated mesh around the dorsal fin with ANGLE critiria like shit ,any suggestions?

It will take some time. Look at the meshes by Simon done for drag prediction workshop

cesarcg July 15, 2013 15:55

Quote:

Originally Posted by treenw (Post 439793)
yes,FAR.quite stupid I know ,very grid-inefficient,already got 600 million,barely run on my 16G RAM PC.
Just want to see the whole picture.
Still got the Y-block generated mesh around the dorsal fin with ANGLE critiria like shit ,any suggestions?

Why are those cells so distorted? Won't you use the symmetry boundary condition to reduce the number of cells? I think that the Y-block may be not necessary.

Regards.

treenw July 16, 2013 22:56

2 Attachment(s)
Quote:

Originally Posted by diamondx (Post 439845)
the procedure is time consuming !!! expect bad element quality if you don't simply geometry. you should add o-grid for the engine's nozzle exit...

Hi diamandx,
Thanks for reply,how do you think i should simplify the geometry,for me it's already the basic aerodynamic shape of a aircraft.
And the engine exhaust blocks are already splitted like this:

cesarcg July 16, 2013 23:12

I suggest you to use a symmetry boundary condition in order to reduce the number of cells and to simplify the blocking procedure. Otherwise you will require a huge quantity of memory to be able to load the mesh into fluent or any other CFD code. I also see that you have a Y-block right behind the exhaust of the nozzle, which in my particular point of view is adding unnecessary complexity to the blocking topology.

I guess that diamondx is suggesting to implement an o-grid topology at the face of the exhaust like the one shown in the attached picture. That would help to avoid significantly those distorted cells you have in your recent mesh.

http://laurent.nack.pagesperso-orang...cem/grido1.gif

Regards,
CÚsar

diamondx July 16, 2013 23:25

Another thing, you could have used symmetry for your geometry and then transforming that mesh by mirroring it...
@FAR, have you got any geometry like this, this can be a good subject for a contest... :)

treenw July 17, 2013 21:48

Quote:

Originally Posted by Far (Post 439849)
It will take some time. Look at the meshes by Simon done for drag prediction workshop

great tip!
I'm right now learning the stuff from drag prediction workshop searched from google.
thanks!

treenw July 17, 2013 21:57

Quote:

Originally Posted by cesarcg (Post 439868)
Why are those cells so distorted? Won't you use the symmetry boundary condition to reduce the number of cells? I think that the Y-block may be not necessary.

Regards.

HI Cesarcg,
those cells are distorted due to some geometry association promblem,already smoothed after re-create part.
Didn't use the split-in-half model because i really want to see the unsymmetrical flow pattern expected for later purposes.
And for Y-block,i just don't know any other methods to split a sharp edge wing tip block.

treenw July 17, 2013 22:02

Quote:

Originally Posted by diamondx (Post 440186)
Another thing, you could have used symmetry for your geometry and then transforming that mesh by mirroring it...
@FAR, have you got any geometry like this, this can be a good subject for a contest... :)

good idea,that should save half the trouble i expect.

treenw July 17, 2013 22:30

1 Attachment(s)
Quote:

Originally Posted by cesarcg (Post 440184)
I suggest you to use a symmetry boundary condition in order to reduce the number of cells and to simplify the blocking procedure. Otherwise you will require a huge quantity of memory to be able to load the mesh into fluent or any other CFD code. I also see that you have a Y-block right behind the exhaust of the nozzle, which in my particular point of view is adding unnecessary complexity to the blocking topology.

I guess that diamondx is suggesting to implement an o-grid topology at the face of the exhaust like the one shown in the attached picture. That would help to avoid significantly those distorted cells you have in your recent mesh.

http://laurent.nack.pagesperso-orang...cem/grido1.gif

Regards,
CÚsar


Thank you Cesar,very good suggestion,much appreciated.

That o-grid should eliminate the low quality cells at the end of cylinder-shaped nacelles and fuselage,I suppose?

Also I didn't use symmetry boundary because i had unrealistic confidence of my 8-core 16G ram PC(I used to have a Pentium 4 with 2G ram that can handle 1 million grids),infact 600 million is alomost the limit for now,with first-order and turbulence on.I'll resort to half model if this full hexa approach dosen't work out.

Please shine some light on the blocking strategy for my four sharp edge parts without using Y-block:wings,horizental tail,nacelles,dorsal fin(pointed at the frontmost) .

cesarcg July 19, 2013 01:28

Quote:

Originally Posted by treenw (Post 440438)
Thank you Cesar,very good suggestion,much appreciated.

That o-grid should eliminate the low quality cells at the end of cylinder-shaped nacelles and fuselage,I suppose?

Yes, I think that should improve the quality of the cells. It is very important to avoid cells with very acute angles. Read this post by Far for further inquiries Mesh quality criteria

Quote:

Also I didn't use symmetry boundary because i had unrealistic confidence of my 8-core 16G ram PC(I used to have a Pentium 4 with 2G ram that can handle 1 million grids),infact 600 million is alomost the limit for now,with first-order and turbulence on.I'll resort to half model if this full hexa approach dosen't work out.
I don't know what you mean with having unrealistic confidence of your 8-core PC. First of all, what type of numerical analysis are you pretending to do? I mean, do you want to focus on specific parts of the geometry of the aircraft? If so, that will be helpful to give more importance to the nacelles, or wings, or fuselage, etc.

Quote:

Please shine some light on the blocking strategy for my four sharp edge parts without using Y-block:wings,horizental tail,nacelles,dorsal fin(pointed at the frontmost).
The meshing of the rear part of the aircraft will be very time consuming, if successful, lol. Maybe, you'll have to start thinking how to simplify the geometry as suggested by diamondx, since in my short experience the region where the engine is joined to the fuselage, by means of an extruded-airfoil-like geometry, will be very difficult to mesh. This last assuming that you try to mesh it with hexa.

Regards.


All times are GMT -4. The time now is 22:51.