CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Main CFD Forum

1 billions cells: useless?

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   November 27, 2008, 10:41
Default Re: 1 billions cells: useless?
  #61
Balduin Bankerotti
Guest
 
Posts: n/a
Google for "Wayne Smith FLUENT CCM+".

Fluent 6 was delayed by some years because a lot of developers left Fluent and joined CD-Adapco and started developing CCM+.

So ccm+ is a much newer solver than fluent developed with the combined knowledge of fluent and cd-adapco.

And is also has a coupled solver and AMG and ... Try to inform yourself before posting rubbish.

  Reply With Quote

Old   November 27, 2008, 18:27
Default Re: 1 billions cells: useless?
  #62
fluent-user
Guest
 
Posts: n/a
especially since ANSYS try to hammer it into workbench like a square peg into a round hole.

This part i agree with you, i also feel after ansys has taken over, their workbench mania has done more harm than good.

In fact fluent 6.3 versions and higher are slower than 6.2 versions.

with this rate they might end up in dump.
  Reply With Quote

Old   November 27, 2008, 20:06
Default Re: 1 billions cells: useless?
  #63
fluent-user
Guest
 
Posts: n/a
So ccm+ is a much newer solver than fluent developed with the combined knowledge of fluent and cd-adapco.

ccm+ newer solver and thats all. It does not mean that they have added new terms to navier stokes equations or invented new ways to descretizise it. In fact both solver has almost same implementations. But fluent provide you with little bit more options where it matters. One such example is interpolation of pressure to control volume faces, fluent gives you 3-4 options for it where as i do not see where i can even select that in ccm+. (based on documentation of ccm+ version 2.10.013 ).

And is also has a coupled solver and AMG and ... Try to inform yourself before posting rubbish.

Yaa i must admit it was my mistake to say coupled solver was not provided by ccm+. I shall rewrite it pressure based coupled solver is not available with ccm+.

Further, now how about you reread my posting, where did i say that ccm+ does not provide AMG. I said it does not provide geometric or full multigrid. (how about you informed yourself for a change this time).

Further just by saying that both provide AMG it does not mean they are same. Here also Fluent provides you with more, for example smoothing part of AMG in ccm+ has gauss siedel and jacobi. Where as you could use ILU based smoother in Fluent (other than Gauss Seidel).

Further you could use BiCGStab based multigrid (AMG), I do not see that in ccm+. And off course RCM was another useful method that is absent in ccm+ (as I already mentioned).

So new does not mean better and more. New may mean that they have some catching up to do.

At the moment Fluent has more there is no doubt about it.

  Reply With Quote

Old   November 27, 2008, 20:12
Default Re: 1 billions cells: useless?
  #64
fluent-user
Guest
 
Posts: n/a
typo

it should read RPM not RCM
  Reply With Quote

Old   November 27, 2008, 20:13
Default Re: 1 billions cells: useless?
  #65
fluent-user
Guest
 
Posts: n/a
i forgot to add one more thing about AMG,

Fluent gives you option of using aggregative or selective AMG methods.

(I think ccm+ only has aggregative , not sure about it though).
  Reply With Quote

Old   November 28, 2008, 07:57
Default Re: 1 billions cells: useless?
  #66
Jeff Vader
Guest
 
Posts: n/a
OK this is getting tedious.

What you are saying is that: 1. ccm+ has a (much)better GUI 2. ccm+ is a (much)newer solver but 3. Fluent is better because it has four different options for pressure extrapolation and four different multigrid implementations.

Does it not occur to you, that having so many options, is a bad thing?

At the end of the day, what matters is producing accurate repeatable results. If, in order to get a reasonable solution, you have to resort to "voodo magic" - like changing your pressure extrapolation scheme - it's a fairly good indication that your results are solver setting dependent and neither accurate or repeatable.

This kind of manually tuned CFD engineering is all but dead, replaced by process oriented simulation, in which engineers are able to explore a design space through simulating multiple design configurations: something that you can only do with a modern GUI (see 1 above).

  Reply With Quote

Old   November 28, 2008, 08:49
Default Re: 1 billions cells: useless?
  #67
fluent-user
Guest
 
Posts: n/a
Does it not occur to you, that having so many options, is a bad thing?

This is really interesting comment.

Why do you think it is a bad thing, some implementions work well in some situtations. Having more options means more versatile solver.

To appreciate what fluent provides one need to understand the benefits of those solvers.

Do you know that RPM method could be used for linear systems those could not be solved by normal AMG methods.

This thread is getting long but the things i mentioned they have their benefits and i can not write everything here.

And yaa i also forgot to mention all the fractional step based non iterative solvers Fluent provide us. (not there in ccm+)

:-D

  Reply With Quote

Old   November 28, 2008, 09:44
Default Re: 1 billions cells: useless?
  #68
fluent-user
Guest
 
Posts: n/a
this thread has not only become but also , as you put it, has become tedious.

Jasak (openFOAM) and Jemcov, A published some works. Have a look :

http://www.h.jasak.dial.pipex.com/Hr...blications.bib

Probably Jasak can himself enlighten you guys why BiCGStab method precondtioned with AMG has advantages.

Or why RPM method is needed.

I think this ccm+ Fluent debate will go on, so i shall better shut up now.

  Reply With Quote

Old   November 28, 2008, 10:25
Default Re: 1 billions cells: useless?
  #69
Andrew
Guest
 
Posts: n/a
Can we have this thread stopped, or at least can a mod remind people that this forum is not a place for commerical CFD codes to be plugged.....
  Reply With Quote

Old   November 28, 2008, 10:50
Default Re: 1 billions cells: useless?
  #70
Abe Lincoln
Guest
 
Posts: n/a
You can't just stop debate about code choice commercial or otherwise. I find it very hypocritical the clear and blatant bias there is to openfoam on these boards, if a vendor posts on these boards then they are immediately shouted at for free advertising but it is fine for the fathers of foam to plug their code, say how great it is and how it can do everything you need. Bear in mind these are the same people who are profiting from Foam, not through software sales but through support and training.

If this is not a place for commercial codes to be plugged neither should it be a place for openfoam to be praised 24 hours a day.
  Reply With Quote

Old   November 28, 2008, 13:18
Default Re: 1 billions cells: useless?
  #71
pc
Guest
 
Posts: n/a
Have to agree with you on that one. Enough about openfoam already.

And yes, I consider 1 billion cells useless. Nice for the sake of breaking records, but beyond that, BFD.
  Reply With Quote

Old   November 28, 2008, 15:24
Default Re: 1 billions cells: useless?
  #72
underGroundMan
Guest
 
Posts: n/a
Open foam is useless piece of shit. If you want to get good results and get them in time, then you have to use commercial code. 1 billion cells with present technology is no big deal, but Sarah thought she performed a miracle. Bless her, she needs to learn little bit more about CFD.
  Reply With Quote

Old   November 28, 2008, 18:31
Default Re: 1 billions cells: useless?
  #73
fluent-user
Guest
 
Posts: n/a
Abe, I agree with you sentiments completely.

But let me explain why I have posted about Fluent vs ccm+.

Over the last year or two, i have been listening to people on this board, who always come out and say something like - "ccm+ is new hence better". "ccm+ gui is shining hence better"

Nobody ever tells whats so great about these new solvers?

Only comments i will here is that since everyone says they are better , they shall be better.

I read manuals of CFX , CCM+ and Fluent very closely. In fact i have all the manuals of Fluent from version 5.2 and tkae lots of interest in how the solver was implemented.

Other than that i have not only read their solver part, i have tried their ways in my unstructured grid based segregated solver code. (Both Fluent's way and ccm+ way).

I found that Fluent's way is much stable. (acurate or not is another debate, but it is more stable).

What i think is people who says ccm+ is better or openFOAM is better are the people who only learn software by its GUI. The person who understands underlying maths, knows one can make other solvers behave very similar if not same.

(It means if i had to work with ccm+ i could make it work exactly the same as Fluent would have for me).

Your point about openFOAM is very valid. I do thing it is praised too much on cfd-online. But i do not use it.

I mentioned the links about jasak, because he with Jemcov (from Fluent) published some work on next generation of solvers. Where my interest lies.

(In fact I have recently come across a phd thesis, whose multigrid implementation was much better than what Jemcov and jasak did. I am spending time with it, by trying to implement it myself in my code.

  Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Import netgen mesh to OpenFOAM hsieh Open Source Meshers: Gmsh, Netgen, CGNS, ... 32 September 13, 2011 05:50
snappyHexMesh won't work - zeros everywhere! sc298 OpenFOAM Native Meshers: snappyHexMesh and Others 2 March 27, 2011 21:11
snappyHexMesh aborting Tobi OpenFOAM Native Meshers: snappyHexMesh and Others 0 November 10, 2010 04:23
external flow with snappyHexMesh chelvistero OpenFOAM 11 January 15, 2010 20:43
physical boundary error!! kris CD-adapco 2 August 3, 2005 00:32


All times are GMT -4. The time now is 10:17.