|
[Sponsors] |
March 18, 2016, 08:14 |
ICEM-CFD for 3D sphere
|
#1 |
Senior Member
|
Hello everyone!
First of all, I must apologise for asking again about this geometry. Last time I asked, and got some very useful tips, I wasn't able to get any mesh out of ICEM. Due to work related issues, I had to put that aside, but I'm back again with this problem. Basically, I want to mesh a 3D domain, box shaped, around a sphere, using ICEM. Either, using half of a domain (from Z=0 to Z=Zmax) and mirroring it afterwards, or straight 3D from the start (with Z going from Zmin to Zmax). X and Y are from MIN to MAX. So, to start from scratch, avoiding any wrong steps I might have followed in the past, I've come back to the tutorial in youtube. My main problem right now is that, when following ALL steps EXACTLY as shown in the video, I am not getting the same Parts, or the same O-grid around the hemisphere, or other obnoxious results which are quite upseting. could anyone share with me any document, with (baby) step-by-step procedures to mesh a "simple" 3D sphere, please? On top of that, could anyone share information on how to mesh projectile-shaped bodies? meaning, flat vertical rear face, cylindrical body and ellipsoidal (or half-spherical) front part. Thanks! |
|
March 18, 2016, 11:29 |
|
#2 |
Super Moderator
|
||
March 18, 2016, 15:56 |
|
#3 |
Senior Member
|
Hello Sijal,
The tutorial I am trying to replicate can be found at https://m.youtube.com/watch?v=R0F0q9xBK_w My problems start right at the beginning with a different list of Parts in my session than the one shown in the tutorial. I apologize that I cannot show any meshes, because I don't get that far, I cannot even replicate the step for the O-Grid. After trying many, many times, in the last two weeks, and before Christmas, I have given up to follow that tutorial, and that is why I am asking for help if anyone has some guidelines, made by themselves that don't mind sharing. Regards, Carlos |
|
March 18, 2016, 19:58 |
|
#4 |
Super Moderator
|
It is not necessary to have same parts.
Idea is simple, in fact very simple. If it is full sphere, then go for full ogrid, for half sphere half ogrid and for quarter sphere quarter ogrid. Steps are (take a case of hemisphere that is half sphere) 1. define parts name. It may be just random names for the purpose of this tut, say A, B, C etc 2. You must have curve on outer boundary and also on symmetry plane for both outer cube and inside sphere. it is just must for ensure proper anchoring of edges and vertices. 3. create a 3d block 4. Split blocking at top of sphere. Four additional splits at four sides of sphere. (say bottom, top, upstream and downstream). So total splits you have are 5 ! 5. Now change material of block inside sphere to solid 6. Use snap vertices to snap the vertices on the sphere (hemisphere infact ). Remember surfaces should be visible, and preferably only for sphere. 7. now use ogrid command. Select solid block (block inside the sphere), after that select face on bottom side of sphere (flat part of your hemisphere). Tick the option "around" in ogrid panel. Use value of 0.7. Now apply. 8. Set pre mesh parameters and check pre mesh. You are done now. |
|
March 29, 2016, 06:20 |
|
#5 |
Senior Member
|
Thanks Sijal,
Those are the steps they show in the tutorial, though. And, those were the steps I couldn't replicate as I followed the tutorial. I'm going to try with the sphere now. But, I remember I couldn't add points to the sphere's lines back in the day, the same way the add points to the circumference of the hemisphere in the tutorial. Regards, |
|
March 29, 2016, 09:57 |
|
#6 |
Senior Member
|
Dear Sijal,
I have tried the sphere and have arrived to a point more advanced than my last try weeks ago. My problem back then was that, when generating my domain in DesignModeler with enclosure, there were no 'curves' defined around the sphere in ICEM in order to define points and use them to split the domain. Today I have used the Geometry -> Create/Modify Curve -> Surface Parameter option ans was able to create curves around the sphere, and points afterwards. The Isocurve Method was "By Parameter", and selected U and V directions in order to create each curve. After that, I followed the tutorial, and was able to create a preliminary mesh before creating the O-Grid. The attached pictures show how it looks, the mesh is "over-turned" somehow. I thought creating the O-Grid would fix it, but still the same look around the sphere. Any hints? Thanks!! |
|
March 29, 2016, 11:42 |
|
#7 |
Senior Member
|
BTW Far,
Are you the owner of the youtube channel "far cfd", the one where I found the tutorial I mentioned previously? |
|
March 29, 2016, 12:19 |
|
#9 | |
Super Moderator
|
Quote:
|
||
March 29, 2016, 12:21 |
|
#10 |
Senior Member
|
LoL!!! ok, will try that right now.
Great videos by the way! ;-) but, some 3D "simple" geometries could be a good addition :-D C. |
|
March 29, 2016, 21:14 |
|
#11 |
Senior Member
|
Thanks Sijal,
I was trying other approaches and now I cannot replicate my previous steps in order to apply your suggestion. Now, I get a different result when I "Snap Project Vertices". Please, refer to figures. I have attached my files too, abusing of your generosity going crazy over here!!! C. |
|
March 31, 2016, 08:30 |
|
#12 | |
Super Moderator
|
Quote:
|
||
March 31, 2016, 08:54 |
|
#13 |
Senior Member
|
The sphere was generated as Primitive in DesignModeler, with Enclosure.
Once I'm in ICEM, and in order to have curves on the body, I defined points at coordinates on the sphere's surface and defined those curves with three points. Something strange that happens is a warning message I get when I go to "Auto Association" after creating the block "Fluid" (replicating the steps from your tutorial video): *** Auto Association Warning This feature needs geometry connectivity. Do you want to run build topology? *** I have to select 'Yes' in order for something to happen. I guess this warning is related with what you say about not having a full sphere. On the other hand, If I choose to view Surfaces, I can see the entire Sphere. What do you mean with sphere orientation? I have even tried generating the sphere in DesignModeler by revolution of an 180º arc, with the ends orinted in the Z direction, and the sphere looks rotated once I open it in ICEM. |
|
April 1, 2016, 04:03 |
Make it simple
|
#14 |
Super Moderator
|
||
April 1, 2016, 04:42 |
|
#15 |
Senior Member
|
Thanks Far,
The video frame is giving me an unexpected error, I'll try it later as it recommends me to do and let you know. cheers! |
|
April 1, 2016, 05:26 |
|
#17 |
Senior Member
|
Sorry about that! I meant the box where the video is shown. But it works now! I'm on it.
C. |
|
April 1, 2016, 07:16 |
|
#18 |
Senior Member
|
Followed the video and I'm on the way to get what I need
Just for your info, the video cuts when you are about to explain something on Pre-Mesh Smooth. But that is extra. Having a nice sphere, and controlling the cells size and stretching in every block will ne enough to get a good mesh. I understand now that my error was to define points and curves on the geometry as the first step. That was messing up everything when Auto-Associating. I can use now one of the two points created on sphere (when generating topology) to create the fluid block, right? Thank you sooo much!!! |
|
April 1, 2016, 08:37 |
|
#19 | |
Super Moderator
|
Quote:
|
||
April 1, 2016, 09:07 |
|
#20 |
Senior Member
|
Thanks again!
Could you recommend me any reading to understand the concept behind the way ICEM works? Apart from the User's Manual. My geometries will become more complicated (still canonical) in the future and would like to know when to make things in a different way depending on new geometry features, like sharp edges, hollow sections (imagine we drill a cavity on the base of a bullet-like geometry, for example) ... |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
[ICEM] Icem CFD on Linux | mechanicaldesign | ANSYS Meshing & Geometry | 7 | March 11, 2021 19:44 |
Need help icem cfd | kakhtar | ANSYS Meshing & Geometry | 25 | January 31, 2017 01:09 |
Transport mesh from ICEM CFD, to Fluent, to Sysnoise | Wieland | FLUENT | 2 | April 15, 2012 06:28 |
Importing Solidworks part into ICEM CFD | MetalSupremacist | FLUENT | 0 | October 8, 2010 17:46 |
Which is better to develop in-house CFD code or to buy a available CFD package. | Tareq Al-shaalan | Main CFD Forum | 10 | June 12, 1999 23:27 |