|
[Sponsors] |
[ICEM] Meshing adjacent wall geometry and simple ICEM questions |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
June 7, 2012, 07:50 |
Meshing adjacent wall geometry and simple ICEM questions
|
#1 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Since I am now trying to mesh this geometry from scratch in ICEM i hope that creating a new thread is ok.
I am going to have some very simple basic questions i wasn't able to find the answer to and some specific ones concerning my geometry. And this is the idea. The first question i want to ask is whether it is possible to assign equal length to the block edges? Here https://dl.dropbox.com/u/55240438/im...e_vertices.png, for example, to get the best result I have to move vertices in such a way that makes the edges equilateral. Is there really no way to make them equal without calculating the actual length or at least move them simultaneously?! Because when i try to move them together one vertex 'jumps' out from the curve it is associated to. And even Design Modeler allows to assign such simple constraints! It is quite strange that it is so hard to make symmetrical geometry in ICEM. Okay, the second question is more important to me. It is about creating a 'coupled' wall in FLUENT. For the 'shared wall' to work correctly in Ansys Meshing and in FLUENT you have to combine the bodies that share that wall into one 'Part' in Design Modeler. But if I import these objects that were formed into one part into ICEM CFD then i get all mixed and tangled solids and surfaces. But if I import the objects that were not formed into one part then they open okay in ICEM. But this raises the problem - how do I name this 'shared wall' in ICEM so that FLUENT would treat it as a 'coupled wall'? Because the geometry now has two identical but different surfaces. Similar question - in Ansys Meshing or in Desing Modeler it didn't matter if I called a body 'fluid' or 'solid'. FLUENT just interpreted it as 'interior' and I could assign it to be what i wanted. But when I tried doing the same in ICEM, FLUENT only recognized one 'interior' instead of two. May be someone can help me to export the mesh from ICEM to FLUENT correctly for this simple case: picture, geometry, meshed in icem Here is the bayonet geometry just in case https://dl.dropbox.com/u/55240438/im..._exchanger.rar |
|
June 7, 2012, 10:32 |
|
#2 | |
Super Moderator
Ghazlani M. Ali
Join Date: May 2011
Location: Tokyo, Japan
Posts: 1,385
Blog Entries: 23
Rep Power: 29 |
Good morning,
After meshing a geometry yesterday, an expert told me : Quote:
I had made a video when i use these features, you can take a look at it here from 0:05:59 and 0:06:38 : http://www.youtube.com/watch?v=nuiRkZqaEz4 For your case let's say you have a vertex (1,1,1) you want the other one to have the same height y. You will then use move vertices and "set location"-> you will select the first one as a reference, then you check z and y, like that he will only move in the x direction. You can align them all by clicking on align vertices, you select the edge and then the vertices. you can always undo you selected the wrong one. "align vertices in line": selection 2 vertices as a reference, then the third one will be aligned with them.I'll let you play with it, they are very well explained in the help. this section is very important because it's the key to have a good quality mesh. 2-I never got the chance to work with coupled wall but i think they will be set in fluent, not in icem. 3-When i want to use two different interior in my geometry, i named them different like "FLUID_HOT" and "FLUID_COLD" when creating blocks in icem. then fluent will see two interiors, i hope someone can come and give you more details about the second question, i'll try to look at you geometry when i get to the lab. |
||
June 7, 2012, 11:52 |
|
#4 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Thanks for replying.
First, using reference legth/location isn't the same as applying equal length constraint. Imagine a circle with an inscribed triangle. You do not have to know neither the length of the sides nor the position of the vertices of the triangle to know that the sides can be equal. So what I want is to be able to assign those sides as having the same length without knowing the length and then be able to move/rotate them inside that circle however i want without making the lines uneven. Thanks for the video though, it is a good tutorial. The only thing it's lacking is a voice over >> "I never got the chance to work with coupled wall but i think they will be set in fluent, not in icem." It is a key moment in creating a model in Design Modeler - you have make a 'multibody part' from the bodies that share a wall. Only after doing this Ansys Meshing would apply the same mesh to the both sides of the wall and after that FLUENT would understand it is a 'coupled' wall and slit it into a 'wall' and its 'shadow' >>> "When i want to use two different interior in my geometry, i named them different like "FLUID_HOT" and "FLUID_COLD" when creating blocks in icem. then fluent will see two interiors" So it is a matter of naming blocks, not geometry bodies, right? If so, then how should I proceed in the above given example? To create the mesh for the Outer (ring) channel and the inner channel i start from creating one block which i then split with O-grid method where the inner block now represents inner channel and outer block - ring channel. Should i rename the inner block somehow? |
|
June 7, 2012, 12:29 |
|
#5 | |
Senior Member
|
Quote:
http://www.youtube.com/watch?v=Pe6DfdLUFZU&feature=plcp |
||
June 8, 2012, 07:10 |
|
#6 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Thanks, i've watched the video and pictures!
Question: is it necessary to apply o-grid to all the blocks when you just want to apply it to the small cylinder? So I made some initial steps in my geometry. I was very confused at first, but then I found out that it is better to ignore the branch pipes in the beginning. Here's the process (click to enlarge): I was able to obtain the following mesh: But obtaining a mesh like that was never expected to be a problem. The hard part comes now - i need to make inflation, and it needs to go somewhat like this: First, i don't need the refinement of the inner pipes on the larger pipe as i noted on the previous picture. Second, I don't know how to go on with creating inflation that goes round the wall and the circle pipe. I mean i know i have to o-grid it somehow but i am not able to do that yet. Help would be very appreciated! P.S. Oh yes, here's the link to my updated project! : ) |
|
June 8, 2012, 07:14 |
|
#7 | |
Senior Member
|
Quote:
I have a question: How did you make the hybrid mesh? |
||
June 8, 2012, 07:17 |
|
#8 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
The hybrid mesh was made in Ansys Meshing.
And it does have bad elements on the border of tetra and hexa cells which i can't completely get rid of. |
|
June 8, 2012, 07:29 |
|
#10 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Come on, that is a previous mesh that i made almost a week ago. After understanding i cannot improve it further I decided to switch to ICEM and create a new mesh from scratch. That is what I am trying to do now and what I am describing in this thread ) I just showed that picture to give a better understanding of what kind of inflation i am trying to achieve.
>>> "No, not necessary. For my case it was necessary." I don't see how |
|
June 8, 2012, 11:28 |
|
#11 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Okay, so I think I found the right approach to be able to create inflation around the inside pipe and the inside wall.
But I get some strange mix up again when I generate the pre-mesh. I don't even know if it is a problem or the mesh would be okay with this bug left, but it certainly doesn't look okay. I checked the all possible associations but nothing seems to help. The screenshots and the updated project are attached. |
|
June 8, 2012, 17:54 |
|
#12 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
OMG I can't wrap my mind around that inflation going along the wall!
Forget about those 'bugs' and mesh-mixing. The biggest question is: How to make the inflation around that wall inside the bayonet tube?? I believe it should thought of as a sort of half-round airfoil in 3D. I think i have to make some kind of o-grid around it, but how?! Please help! |
|
June 13, 2012, 15:37 |
|
#13 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Ok, i finally made it. The day before yesterday.
Here's how it looked: It wan't perfect yet, but quite good already. Determinant quality was higher than 0.3. And yes, creating o-grid for all the blocks when you need it just for small pipe was indeed necessary. So what i wanted to do before improving the quality further was to try and export the mesh to FLUENT and give it a test try. Mesh converted successfully and I obtained the expected results. Well, when I opened the project file the next day, after nothing at all has changed, guess what i got? It looked the same, but the mesh quality was the lowest. Why? No idea. It's just another reminder that ICEM CFD is one of the worst applications i worked with. That is not to mention that not only you cannot change too much after you made something - only option is to click 'undo', - but even the 'undo' button sometimes makes blocks disappear. Oh, and of course 'undo' doesn't work if you have pressed the 'save' button. So, what i have to do is start from the very beginning. And that's something i had to do very many times with ICEM. |
|
June 13, 2012, 15:40 |
|
#14 | |
Senior Member
|
well done. Great work.
Quote:
|
||
June 14, 2012, 04:40 |
|
#16 |
Senior Member
AB
Join Date: Sep 2009
Location: France
Posts: 323
Rep Power: 22 |
if the blocks disappear, open the index control, and click on reset.
The blocks will come back |
|
June 14, 2012, 04:51 |
|
#17 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
Oh really?!
|
|
June 14, 2012, 04:55 |
|
#18 |
Senior Member
AB
Join Date: Sep 2009
Location: France
Posts: 323
Rep Power: 22 |
it worked for me !!! (ICEM CFD 12.1)
|
|
June 16, 2012, 07:09 |
|
#20 |
New Member
Join Date: May 2012
Location: Moscow
Posts: 27
Rep Power: 14 |
I think it's obvious that index control is not the matter in this case.
I confirmed - the quality of the initial pre-mesh always worsens after I apply boundary conditions to the named surfaces in the output for fluent. So each time i want to export mesh I have to save my work as a separate project so that i can "damage" it. And if i want to make some corrections - i delete this project and go to the previous one. |
|
Tags |
conjugate heat transfer, coupled wall, fluent, icem, inflation |
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
ICEM: Projectile Meshing and Modelling Issues | gman89 | ANSYS | 0 | May 12, 2011 00:01 |
Need help on meshing this simple geometry | nha1g08 | ANSYS Meshing & Geometry | 3 | March 9, 2011 03:13 |
[Other] Geometry and Meshing Questions | Flaky | ANSYS Meshing & Geometry | 6 | October 8, 2010 02:10 |
[Gmsh] Problem meshing an imported geometry | Alicia | OpenFOAM Meshing & Mesh Conversion | 0 | March 30, 2010 05:53 |
Meshing on simple geometry | pukiwawa | FLUENT | 5 | August 31, 2008 13:04 |