How to generate this kind of mesh?
Hi, everyone.
I saw some mesh posted by Simon and I think I need this kind of mesh right now. Can anyone tell me the steps to generate this kind of mesh? https://lh6.googleusercontent.com/-G...ltiZone_F6.jpg Thanks in advance It seems that he built prism layer on hexa elements. I had tried 2d surface blocking and generated hexa surface mesh, but when I tried to generate prism layer, some of hexa mesh turned in to triangles. I tried use Octree to fill the volume first and then generate prism layer, all the hexa elements became tris. I don't know how Simon build this mesh, it looks pretty good |
1. Create 2D blocking for all geometry.
2. 2D to 3D, with options > "Multizone Fill" and O-grid around a fuselage and wing, "Fill type" Advanced. |
not working .
|
Quote:
|
I will put together a tutorial for this as soon as I can (yes, I know I have been promising it for a while), hopefully this quarter. It can actually be done in about 5 minutes, but it takes longer if I need to explain it all. ;^) And for some reason, putting together a 15 minute demo seems to take all day.
Simon |
PSYMN - You are absolute help
That would be the great help to us.
|
Quote:
|
MultiZone meshing
Hello,
If it serves for stressing further more, I have been looking for this MultiZone solution for ages (more than 3 years). I am dying to see the solution! Why MultiZone approach (and maybe other bottom-Up features) is not so well documented?, appears a superb feature. Best regards |
The general guideline would be to tell you how it works...
The first step is to Create a "2D surface Blocking" Multizone starts by looking at your geometry. You need to have topology built because it uses that to understand what surfaces are attached to what. If it finds a surface patch with 4 corners, it can map it. If it finds a patch with more or less corners, it can pave it. The topology info is used to connect all these blocks together correctly. If you set a tolerance or have dormant curves, it can merge patches together. After auto blocking all those patches (just a few seconds), it takes the mesh parameters from the curves (so you should set sizes to take advantage of this). You can then adjust this surface blocking in preparation for the second step. In addition to adjusting the topology to what ever you want, you should make sure that your perimeter is closed (no single edges). The second step is a 2D to 3D "Fill". If you set this to sweep, it looks for rings of mapped blocks which could indicate the sweep direction. It can imprint, etc. to automatically handle multi source and target as well as multi direction sweep. If the model just needs a regular fill... it uses code similar to the hexa dominant fill... But instead of filling a quad dominant mesh with hexas, it it fills the surface blocking with volume blocks. You can run Ogrid to insert a boundary layer and it works just like with regular blocking. If you are doing external aero, then the sweep or Hexa block fill are not likely to give you a good mesh. You should use the unstructured fill. This basically fills the volume with one of the unstructured fill options (like tetra), but then you can insert an OGrid to give a nice hexa boundary layer around the aircraft... All of this is done on the ICEM CFD hexa framework, so you can adjust verticies or edge distributions, etc. You can also change block types, etc. |
Quote:
|
I got stucked in periodic conditions
|
It is working now, but I am getting fully hexa mesh, how to convert the outer blocks into tetra type. I have also tried edit block command.
|
Quote:
How to generate full hexa with this though...? I can't find where to choos... |
1 Attachment(s)
I found this have problem on the tip of the airfoil with sharp trailing edge or similar shapes. When I tried to do 2d to 3d fill, it automatically deletes some surfaces...
https://lh4.googleusercontent.com/-T...28/Capture.PNG Also there's some periodic issue... I set periodic in mesh tab, then 2d surface initialization, Edit block=>periodic vertices, I made all the vertices on the periodic surface periodic. the surface is tri-paved surface. It seems that all the nodes are matched, but the elements on the surface are not, the block faces are not periodic. I think it's because type of block face is "free". How to make free surface periodic? I had similar problem before when I tried to generate surface mesh then use Delaunay under mesh tab. Only octree give respect to periodic set up. The surface mesh generation seem ignored the periodic set up. https://lh5.googleusercontent.com/-h...NKQ/s811/1.PNG https://lh5.googleusercontent.com/-d...LRs/s906/2.PNG https://lh4.googleusercontent.com/-s...fWs/s908/3.PNG |
Quote:
Blocking -> Edit Block -> Convert Block Type That should work ! |
WOW this works!
PSYMN, Thank you so much!
|
Mingersai,
Did you check the periodic faces by ploting them on the faces options?, sometimes the collapsed node have to be marked as "periodic" to make the face periodic (I remember they were marked in red color) cheers! |
Quote:
I had contacted customer support and got negative answer :( the periodic is designed for Octree method and mapped type face blocks They suggested me some tricky ways to deal with free type face, I'm still trying to make it work. My application is to fluent and it seems that fluent could define non-conformal periodic faces. I'll try either way.. cheers~~ |
mingersay,
yep in Fluent there is no problem defining periodic faces that are not conformal. I have use it many times. cheers |
Simon when you will going to post the tutorial for generating this kind of mesh because the beginners like me are still waiting for it. I now it will take your time to prepare tutorial but it will help a lot people. Thanks
|
Hey there,
This is my first post on this forum. I'm a fourth year student in engineering school in France and I have a project this year which consists in studying wingtip vortices with CFD. I modeled several wingtips (with and without winglets) with CATIA V5 but I must say I'm having trouble meshing the geometry. I came across this topic thanks to google and I must say, it looks like a great solution to deal with my problem : hexa structured meshing near the geometry and progressive tetra in the fluid volume. I tried to follow the explanations posted a few weeks ago but there's no way I can get anything satisfactory... I would really appreciate a more accurate tutorial describing the steps to follow in order to achieve this kind of mesh. Thank you so much in advance. :) Antoine |
Salut Antoine,
tu remarquera qu'il n'y a pas d'accent dans mon message. Desole pour cela, mon clavier est actuellement anglais. J'ai depose dans mon fichier public un tuto ICEM CFD sur les aile d'avion. Voici le lien. Let me know quand tu l'aura telecharger pour que j 'efface. J'espere qu'il te viendras en aide. Cliques sur le lien ci-dessous: http://dl.dropbox.com/u/35161486/ICE...body_Prism.pdf |
Quote:
|
Quote:
The method posted by Simon work fine with simple geometries. |
@Diamondx : Merci pour ta réponse, j'ai téléchargé le fichier, tu peux le supprimer. Je lirai le tuto demain et je verrai ce que ça donne. Ca a l'air de ressembler à une méthode que j'ai découvert un peu en tatonant mais le problème que j'avais c'est que je n'arrivais pas à avoir un maillage tetra progressif dans la zone "au large" de la géométrie. Donc maillage bien trop lourd et temps de calculs trop importants. Je vais lire ça en détail demain!
@Mingersai : Thank you for your input ! I already read your messages but I'll try again and see how that goes. Diamondx's method seems quite similar when it comes to the final result though, doesn't it? |
Quote:
|
Deat Simon,
I am still not clear about how to combine the o-grid from 2Dto3D operation and the prism part near the fuselage, could you leave us a detailed introduction? Thanks, Zhenyu:confused: |
I am looking for this mesh to get out of the troubles faced with tetra meshing. I tried this method but without any success. I am explaining my procedure kindly correct me and guide me where required....
First general tetra settings...Global scaling factor to 1, surface mesh method to "Patch Dependent", sizes in "Part Mesh setup" and "surface mesh setup" ofcourse when required,voulme mesh method set to "Delaunay" Then I go to Blocking, "2D Surface Blocking", 2d to 3d with "Multizone", Ogrid around wing and fill type "Advance"....real trouble starts here....should I switch back to Mesh(terta framework) and generate volume mesh using "Generate Mesh" tab or should I stay in "Blocking" and generate mesh using "Premsh"? I tried with "Premesh" after few commands ran across console window no mesh appears not even surface mesh...where I am making mistake? Thanks |
@CFDSeeker
Once in MultiZone, you don't go back to compute mesh. It doesn't use Octree tetra at all... You can generate a premesh and convert that to uns mesh. Best regards, Simon |
So Delunay will automatically get generated while in Premesh?
|
Quote:
|
@Simon @Diamondx
2D Surface blocking should be done for the whole geometry or just for the wing surfaces(I don't have fuselage in my geometry) and by which "Type" fully quad, quad dominant? |
Quote:
you must use inflation method in ICEM for generating boundary layer mesh near wall. Please contact me for more information: fluent.workshop@gmail.com |
Quote:
You can preview all the mesh via "PreMesh", and then convert to unstructured to get somehting you can export to a solver. |
2D Surface blocking should be done for the whole geometry or just for the wing surfaces(I don't have fuselage in my geometry) and by which "Type" fully quad, quad dominant?and by which method?
|
Quote:
But others may choose to generate mapped mesh on the wings, then run octree while keeping that mapped surface mesh, then run prism and then delaunay to replace the octree tetras... |
5 Attachment(s)
I am trying to do this but without any success.
1. First I did surface blocking and I get what i shown in image 1, image 2 shows surface blocking on symmetry plane and image 3 shows mesh in wing tip region. Settings are also shown in the image 1. 2. Then 2d to 3d and using the settings shown in image 4, the shape of O-grid block is awful I tried to edit it using edit edge but the effort was futile 3. Image 5 & 6 shows what I get after "Premesh" Now I don't know where I am getting it wrong, may be I have to do some blocking adjustment I also tried it but ended empty handed or may be I am making some mistake? http://C:%5CUsers%5CNetUser1%5CDeskt...tizone%5C1.jpg |
1 Attachment(s)
image 6 is shown here
|
R13 was the first release of MultiZone... It is much better with more recent versions.
But I would still probably mesh this simple geometry with traditional Hexa (CGrid). If you want, you could convert that block between the tip and the farfield to a swept block. |
Quote:
Quote:
|
All times are GMT -4. The time now is 13:30. |