CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Main Diffenrences between commercian packages ?? (https://www.cfd-online.com/Forums/main/12400-main-diffenrences-between-commercian-packages.html)

Abou October 13, 2006 05:33

Main Diffenrences between commercian packages ??
 
Hi all, I'm doing a research on the 3 main commercial CFD packages : Fluent, CFX and Star CD...

Could you help me ? I'm looking for the pro and cons of these codes, especially for the Aerospace Engineering... I already know that Fluent is not accurate for Highh speeds, but that's all I know.

Many thanks !!

Ben October 13, 2006 07:39

Re: Main Diffenrences between commercian packages
 
Fluent big, part of the Ansys axis of evil, easy to use but getting on a coupled of segregated solver, t-grid is good, gambit is some form of toture....

CFX big, part of the Ansys axis of evil (again) no one quite knows how long it will be around for now they own Fluent, good for turbomachinery, coupled solver only, vertex based = massive memory usage and waiting all year for tranisent to run, easy to use so I hear, meshing is good but only really tet unless you are willing to pay more cash

STAR-CD Older but most flexible of the codes, not that easy to use but contains loads and loads models, best for combustion, owns another code STAR-CCM+ developed by a load of fluent guys, brand new easy to use great to look at very fluent like, choice of solvers, as good for aero as fluent but other physics capabilites are limited at the mo. both codes polyhedral

And thats about it not sure about the other codes


opaque October 13, 2006 08:15

Re: Main Diffenrences between commercian packages
 
Dear Ben,

Have you tried the Fluent 6.3beta? There is a polyhedral solver there and a new pressure based coupled solver..

It is supposed to converge faster than their pressure based segregated solver.. Of course, the memory requirements are larger but who would care when you could get the solution in a lot less wall clock time...

Good luck,

Opaque


Ben October 13, 2006 08:46

Re: Main Diffenrences between commercian packages
 
Yeah I heard about the polyhedrals and the pressure based coupled but I was under the impression that 6.2 and before had a coupled solver already, so what is the difference between this new pressure based coupled and the old coupled?

opaque October 13, 2006 13:19

Re: Main Diffenrences between commercian packages
 
The solver type matrix will be something like

-------------------------------------------

| Segregated | Coupled |

-------------------------------------------

Version | 6.2 | 6.3 | 6.2 | 6.3 |

-------------------------------------------

Density | Yes | Yes | Yes | Yes |

-------------------------------------------

Pressure | Yes | Yes | NO | Yes |

-------------------------------------------


Ben October 13, 2006 14:05

Re: Main Diffenrences between commercian packages
 
cheers for that but i am interested in how the pressure based solver works, how it differs from the density based and why it is "better"

zxaar October 15, 2006 19:34

Re: Main Diffenrences between commercian packages
 
gambit is some form of toture.... he he he ... true to some extent.

CFX big, part of the Ansys axis of evil (again) no one quite knows how long it will be around for now they own Fluent, good for turbomachinery, coupled solver only, vertex based = massive memory usage and waiting all year for tranisent to run

i am expriencing this. Just waiting waiting and waiting ..

Abou October 16, 2006 03:15

Re: Main Diffenrences between commercian packages
 
Hi guys !

Thanks for your answers... Do you know why CFX is so long ??

many thanks !

Phil October 16, 2006 08:36

Re: Main Diffenrences between commercian packages
 
Be careful when comparing transient run times - there are many factors that can affect this (convergence level and/or number of inner loops done within the timestep, etc). Best practice may be different with a coupled solver than with a segregated solver. For instance, with a coupled solver, you should be able to get away with fewer inner loops per timestep. (I have seen LES results with CFX where only one inner loop per timestep was necessary, even though with the default settings it would use 3 or 4.) You should optimize these things on your case before starting a long transient run.


All times are GMT -4. The time now is 07:48.