CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Pre-processing for Fluent (http://www.cfd-online.com/Forums/main/3076-pre-processing-fluent.html)

Althea January 29, 2001 12:23

Pre-processing for Fluent
 
As readers of the Fluent discussion forum will know, I have been experiencing some difficulties with Gambit. I haven't given up on it (yet) but because some of the problems have impacted major projects I have worked on, I am looking around for potential alternatives. I will need to assess several alternatives to be able to justify any change.

Is there anyone out there who can make any recommendations? I use Fluent 5 and the more complex geometries require hybrid meshes (structured in some regions, unstructured in others). Complex geometries come from Catia (which causes a lot of problems in Gambit). They are not simple shapes and have often been created using Catia surfacing. I also need to be able to include boundary layer cells.

Software with a UK contact is essential.

All suggestions, comments and recommendations will be gratefully accepted.

Best wishes to you all. Althea

sylvain January 29, 2001 12:42

Re: Pre-processing for Fluent
 
Did you look to TGRID. It's older than Gambit, but it's very usefull for what you want to do. Ask your fluent's commercial.

Regards,

Sylvain

John C. Chien January 29, 2001 13:10

Ask support engineer in Fluent Forum PLEASE
 

John C. Chien (Alter Ego) January 29, 2001 15:38

Re: Relevent question Chien. stop whining PLEASE
 

Raza Mirza January 29, 2001 18:07

Re: Pre-processing for Fluent
 
Althea,

As I said in response to your posting on the Fluent forum, you might want to look at ANSA and TGrid combination. ANSA is an excellent tool for surface topology cleanup and surface meshing. TGrid is excellent for tet meshing and might do a reasonable job for hybrid (prism + tets) meshing. I have been using this combination with great success. I also have some 'external' structured mesh that I use in combination with the unstructured mesh. I use non-conformal interface to connect the two meshes.

John C. Chien January 30, 2001 00:39

YOU ARE RIGHT, BUT I DON'T HAVE ANSWERS
 
(1). I think, only the support engineer can provide a meaningful answer to the complex geometry problem. (2). I had used the code a few years back on a complex problem , and I ended up with over 500 surface patches to get the mesh created with tgrid. (3). Recently, using another top of the line mesh generation code, with the help from the support engineer of the mesh code, I was able to get the result using 80 blocks. (4). So, based on my experience in complex geometry, there is still a chance to get the solution, if one knows how to talk to the support engineer. (5). And depending on the geometry, the mesh and the solver, it is likely that the selection of the codes or combination of the codes is a function of the problem at hands. You may have to move from Catia to Pro/Engineer, because it tends to be better supported by the mesh generation codes. One can also try the top of the line geometry and mesh generation code like ICEMCFD, but it is a very tough code and requires daily support from the vendor's support engineer. (6). Even with special mesh generation code, I used to try 20 to 50 times before I can get one successful mesh. (7). So, before one start, "SIMPLIFY YOUR GEOMETRY IN THE FIRST PLACE".

Anders Jönson January 30, 2001 10:49

Re: Pre-processing for Fluent
 
Hi,

I totally agree with Mr Mirza, ANSA is probably the best tool to clean "dirty" cad. It really was a breakthrough for surface meshing and cleaning cad. We also use Tgrid to generate our tetmesh, however I am not to sure how good the prism generation is for complex surfaces.

P.S To John C. Chien, you are not forced to answer all messages! I really think Altheas question was relevant!

Anders

John C. Chien January 30, 2001 15:26

Re: Pre-processing for Fluent
 
(1). Are we (the main discussion forum) being invaded by the Fluent, because of the failure of the Fluent Forum, the failure of GAMBIT, the failure of Fluent? (2). Are we going to discuss another code here to clean up the geometry for the GAMBIT, and Fluent here?(3). The question was perfectly all right, but it should not be posted here, because most readers here are not users of Fluent codes. I would be very happy to talk about the geometry and mesh generation here, if he is interested in writing his code. Apparently, he is not following the rule of using a commercial code. We have discussed this many times here already. Pick the code, only when the vendor can show you the satisfactory result to your problem. If he had done so, he would not be in this situation looking for alternatives. And any alternatives also must be compatible with the Fluent code in order to have a solution. (4). So, it is obvious that the right place is the Fluent Forum. Not here.

Anders Jönson January 31, 2001 03:37

Re: Pre-processing for Fluent
 
John,

I really think you are wrong now, not all posted questions are for you! I think this is a really interesting question about general preproccesing tools. It really does not matter which solver that are used. I think your comment "Pick the code, only when the vendor can show you the satisfactory result to your problem" is stupid. This is only valid if you run simple cases and know that you will continue to run simple cases. If your everyday work involves complex geometries, many dirty cadsurfaces etc, you do your best to try and simulate it. If you succed, then you immediatly try to shorten the lead time and that involves checking out other tools for pre solver and post. So even if Altha is using Fluent, there is many post processors that can shorten the lead time so the question is very relevant. We do use ANSA both for Fluent and Star-Cd cases! Another interesting comment, in your second message you agree that it is a relevant question for the main forum but in your last message you start complaining again!

Anders

P.S To my friends in Switzerland: No phonecalls this time :)

George Bergantz January 31, 2001 04:33

better for Fluent users if in own forum
 
I support John here, but my take is different- it's simply better to keep FLUENT specific issues in the FLUENT forum so that it can be located easily later, and perhaps a wider auidence can share, address and explore their experiences from a ever more FLUENT-centered perspective which has merit.

So my dear colleagues, I urge you, when possible, to post in the respective forums for such things. This site is one of the best on the net, and it's becasue we all try to get along and respect each other and the structure of the site.

Jonas Larsson January 31, 2001 05:01

Re: better for Fluent users if in own forum
 
... and I support Anders and Althea, for the simple reason that this is not a Fluent specific issue. Ansa is a general CAD-repair and meshing tool, which can be used to create grids for many different CFD codes. I also happen to know Anders and I know that the group he works in has a vast experience of dealing with very complex CAD-geometries, so his advice on pre-processors are certainly of general interest and not just of interest for Fluent users. If there was a "Commercial Codes" forum that would be the place to post these general code-questions, but there is no such forum. Perhaps we need such a forum to avoid these kind of discussions? Clearly Althea is looking for an alternative to Fluent-software and then the main forum is the right place to ask. Althea has also previsouly posted all her gambit-specific questions and problems to the Fluent forum, just as appropriate. No need to complain this time IMHO.

John C. Chien January 31, 2001 06:03

Re: better for Fluent users if in own forum
 
(1). I think,it is perfectly all right to discuss the code ANZA in the Fluent Forum, or other Forum, if it can improve the modeling or meshing for the Fluent solver or other codes. (2). The original question was clearly related to the Fluent family of software. It was not related to the ANSA software. (3). If the users of Fluent solver, using ANZA code to improve the CAD and mesh, did not want to answer his question in the Fluent Forum, then I guess, it will overflow to the main forum. It seems to me that users of Fluent codes do not wish to share their experience with other Fluent users in the Fluent Forum. (4). It is really not an important issue at all in terms of where to post the message, as long as his problem can be solved by some other means. (There was only one forum at the begining anyway.)

Saïf-Deen Akanni January 31, 2001 06:32

Re: Pre-processing for Fluent
 
I agree with the sentiments of the message from Anders Jönson. Problems in specific codes can obviously be addressed by third party software. A solution that is appropriate for Fluent can be also suitable for other CFD solvers.

Let's not be so obsessive about what is posted and what isn't.

Oh and Anders I agree. ANSA rocks!


Althea January 31, 2001 07:31

Re: better for Fluent users if in own forum
 
Dear all,

I am amazed to have caused such controversy.

Just to clarify and summarise.

I did indeed initially post to the Fluent forum and we have had several discussions on this issue there. However code vendors in competition with Fluent are often reluctant to post to the Fluent forum (which is fair enough) and it was suggested to me that it might be helpful to widen the discussion and get a broader range of responses by mailing the main forum (just like the old days when there was only one forum).

So far I have been in touch with ICEM, Gridgen and CFX (CFD-build). I hope to fully assess all of them for my problematic geometries.

So far I have established the following: ICEM runs within Catia to generate the mesh. This gets rid of the problems of importing geometry into another package, but means I will have to learn to use Catia. Gridgen can only import IGES, however, the next release (due in the summer I think) will have a direct Catia import capability. It sticks it's mesh over the original geometry allowing small features or gaps to be ignored when desired. CFX-build can import Catia files directly. All three can write out mesh files which Fluent 5 will read. I don't know much more yet, but am continuing my investigations. I am (as always) interested to hear of other peoples perspectives on this subject.

Thank you to everyone who has contributed. Discussions of this sort can be useful in understanding other peoples difficulties. Suggestions like sticking with real geometry or not using surfacing in Catia are very good suggestions, but sometimes they cannot be followed. From a CAD importers perspective it would be nice if everything was created using simple flat sided shapes, but as a fluid dynamacist I would be constantly frustrated.

If you are interested in further discussions on Gambit or Fluent, then do please look in on the Fluent forum.

Best wishes to you all

Althea

PS. For the sake of John C. Chien. I appreciate that you are trying to be sexless in your comments, but I am female, so would prefer to be referred to as she as opposed to he :)

Althea January 31, 2001 07:36

Re: Pre-processing for Fluent
 
Thanks and yes. I am using Tgrid alongside Gambit. However there are some things which cannot be done in Tgrid and which don't work in the current version of Gambit either.

A good point though and the combination of both is very useful.

Best wishes

Althea

Althea January 31, 2001 07:46

Re: YOU ARE RIGHT, BUT I DON'T HAVE ANSWERS
 
Thanks John, please don't get too stressed by my trespassing into the main forum.

I have been (and often still am) in close contact with my Fluent support engineers. They are very supportive and helpful. There are also several of them which come up with clever ideas to get round the problems and spend considerable time trying to help me make progress.

I agree that software support is there to be used and should be the first port of call with any software problem (i.e. not the discussion forums).

However the new releases of Gambit (which overcome some more of my difficulties each time) take time to get released. The information I am seeking here (on alternatives to Gambit) I do not expect to be able to get from the Fluent support engineers!

I would like to be able to change the CAD package I get my geometry from, however the company I work for has hundreds of Catia licenses and users and a very small number of Fluent users. We have been using Catia for a lot longer than we have been using CFD of any sort.

Finally I do get the geometry simplified considerably (where possible) first. Flattening all the curved faces may be a little too radical though :)

All the best Althea

Althea January 31, 2001 07:55

Re: Pre-processing for Fluent
 
Thanks for the (strongly supported) comments on ANSA. I will look into it. It sounds like it has the potential to restore sanity to even the most stressed Gambit user :)

I am not very good with Tgrid (having successfully avoided unstructured grids before the advent of Gambit). However I am getting better and have enquired about some training.

Would I be correct in assuming that structured hexahedral regions cannot be created using the ANSA / Tgrid combination?

Thanks again for your suggestion.

Best wishes.

Althea

sylvain January 31, 2001 12:42

Re: Pre-processing for Fluent
 
Since ANSA has the capability to mesh surface with quads, it is possible to construct prisms from this quads with Tgrid. So it is possible to create hexahedral regions with ANSA/Tgrid (if I understand what you mean by hexahedral regions).

Best regards,

Sylvain


John C. Chien January 31, 2001 16:00

Re: better for Fluent users if in own forum
 
(1). I don't think it's a problem to post a message here, after hearing all of your troubles in trying to get an answer. (2). When posting a message, a little more background information will definitely help the potential readers to help you. This would eliminate a lot of initial "GUESSING" work. (3). When I was studying German language in college, I had a hard time to deal with die-der-den-die, they tend to associate each object with sex. So, I don't know whether your question is "female" or "male" or "sexless". It is likely that most CFD subjects are "female". (4). By the way, recent study shows that "male" and "female" use different part of the brain to do different thing. This could be one of the problem in the forum. "male" tends to provide the background logic, but "female" tends to express the goal. Based on my experience, "female" normally don't follow the logic. So, the best way to do here is, learn as much as possible from the "male"'s point of view, and turn that into "female" side of the brain. (5). Since you mentioned this subject, it is likely that "male" and "female" handle "CFD" in a completely different way. (6). The scientific study has shown that "males" are very good at 3-D geometry and math. On the other hand, "females" depend on their photographic memory of the locations and scenes to create the 3-D world. Your questions seem to show the trend. The CAD and CFD tend to be on the "male" side, which requires a lot of 3-D stuff and math. (7). Anyway, as the "female" user population starts to grow, I think, vendors of commercial codes will incorporate the "female way of thinking" into the codes. (8). It is not because I try to treat the questions and answers in the "sexless" fashion, it is because I don't know how to treat the Internet forum in CFD with additional parameter of "sex". If the book is "sexless", then it is easier for me to treat CFD related subjects as "sexless" subjects. This is a common "male" response. The female answer would be : I like you regardless of what you say.

Raza Mirza January 31, 2001 20:27

Re: Pre-processing for Fluent
 
One way to creat hex regions is how Sylvain has said in his message. Here is my 2 cents worth

1. ANSA is excellent in cleaning up the CAD model 2. ANSA provides excellent control for surface meshing. It has 3 different algorithms for meshing a surface with either all triangles, quads, or a combination of quads and trias (hence what Sylvain has said applies, but is strongly dependent on the prism growth capability of TGrid) 3. ANSA has limited capability to generate hex meshes in some regions. 4. ANSA also has a tet mesher, but I have found TGrid to be better for tet meshing 5. ANSA can output in a variety of popular formats such as NASTRAN, PATRAN. It can also output surface mesh for TGrid

Also, like you we also deal with models coming from CATIA. Often times they are not even cleaned up for CFD analysis. With ANSA we can still do the clean up ourselves, and then once you have done it, incorporating any subsequent changes to a portion only, it can be done very very easily (and believe me this comes in very handy!).

I deal with very complicated geometries (I work in Chrysler and deal with HVAC systems CFD). I have seen presentation from the GM folks of underhood and underbody simulation using ANSA/TGrid and FLUENT. I have friends in Ford who use the same process. Anders is also in automotive industry and uses this process. Hence, in automotive industry, where the geometries are very complicated, this seems to be a successful combination.

One more IMPORTANT thing. I have been using ANSA for over 7 months. I have not had a single "session freeze" or "hangup". It has never died on me. Does not give me any funny error messages. And I have not had to call a single support engineer for help!

It works beautifully.

I am not a salesman for ANSA. Just a very satisfied user!


All times are GMT -4. The time now is 12:11.