Dears, I'm a Fluent user,but now I'm working in a new company who wants to use Star-CD (it seems to be less expencive).

What are the differences between the two codes (pre, post, analysis capability) for industrial applications, according to you?

What are the merits and the defects?

Your opinion is very important. Thanks a lot

Giovanni

Giovanni,

Both codes have their share of merits and problems.

Let's take a crack at this.

1) Preprocessors

Since you have been using Fluent I guess you are a Gambit/Tgrid user. If you are going to use STAR you will have to get familiar with Pro-Am (Samm, Prostar). Personally, I do not care for Gambit. It does simple things very well, but if you want to do anything difficult, you always wind up in Tgrid. Prostar and Samm are inherently more difficult to learn than Gambit, but once you learn how to use them, there really are not any limitations, as to what you can do. That is not to say that you will always be smiling when you are using this software, but with experience you can almost always "get from A to B". For pre-processing I like STAR.

2) Analysis

Fluent is far easier to navigate through in order to get an analysis running. STAR used to have things grouped in several panels (properties, control), which was convenient. As of late, in an attempt to make things "easier" they have developed the Nav Center. This panel is a nested tree nightmare and does anything but make life easier. STAR provides you with far more information at run time as compared to Fluent. Fluent does things "behind the scenes" and shields the user from too much information. Depending on who you are, this can be a good thing or a bad thing. As far as parallel runs go, Fluent has them beat hands down. With several clicks I can have a parallel run going in Fluent. The HPC version of STAR asks way too much of the user. You can interface with STAR via Fortran code and if you are a dinosaur like myself, you will appreciate this. Ultimately, STAR is once again more flexible (perhaps too flexible), but Fluent is more friendly. For analysis, I like Fluent.

3) Post-processing

No question here, Fluent is way in front. You have to do way to much work in STAR to get the plot that you want.

Hope this helps.

Chinor

Well Fluent is in front in postprocessing if you use excel for x-y plots...

Mark

Gambit is very poor for building quality hex meshes and fluent is equally bad at solving them (at least in combustion cases). If a simple change in geometry is required then one has to start building the geometry/mesh again, fluent say there are short cuts using journal files, but I found them almost impossible to get to grips with, especially if the file is over 500 lines long. It has a very limited capability of arbitrary matching of meshes and fine grid emmbedding is an absolute nightmare with hex, as each cell is divided into 2 x 2 x 2 and your mesh will grow at a geometric rate. This is also bad as the rate of change of volume of cells is large when going from a refined mesh to an unrefined one and can provide instabilities. Fluent is probably ok for simple physics type flows that can be easily solved with tet meshes.

User coding is almost impossible to write with fluent as fluent does not allow debugging.

I have used star-cd for 5 years and fluent for 2 years and as previous contributors say, and I agree, fluent is initially easier to use one very quickley reaches its limitations.

In my opinion, once one understands the philosophy behind star-cd i.e. sets, then there isn't a contest.

Cheers

Allan

Hi Allan, your opinion are very interesting, and I would like to ask you some questions.

It is true that Gambit is poor for building quality hex meshes (in particular for boundary layer in complex geometry), so that you have to use TGrid too, but I do not understand the advantages of Star when there is a change in geometry:does the mesh not have to be build again?

In order to obtain good mesh is it important to have PROSTAR and pro*am or the first is sufficient?

About mesh adaption (grid embedding), what is the method of Star?

About the rate of change of volume of cells, do you refer to Gambit meshing or Fluent adaption?

Is it so difficult learning Star? Is a training course necessary or there are sufficient tutorials?

Thank you very much

Giovanni

Hi Giovanni

Gambit is a very poor hexahedra mesh generator. All the meshing of volumes are automatic and it can create highly unstructured hex meshes which one has very little control over, especially as how the cells 'grow' as one moves away from the boundary. Boundary layers are very restrictive, for example, say one wanted a 3:1 refinement then one must have an integer multiplier of 3 cells in the domain (face) and on the edge i.e. 12 - 4. If there were say 13 cells on the edge then this would be impossible, it appears to be not possible to collapse cell faces to edges.

In Gambit it is impossible to copy a meshed block within a model to another location within the same model without it deleting all the mesh and disconnecting all the geometry, it's a complete mess.

If the geometry change is small then in prostar it is possible without too much effort to move/delete/add cells and thus change the mesh. In the area I work in this saves man-days of effort.

I currently use prostar, I do not have any experience of pro-am.

With regards to embedding one can choose how many cells to divide the current cell into in the I, J, K direction, it is not restricted to 2, 2, 2.

For rate of change of volume of cells, in gambit see above, but in fluent after refinement it is always at least 8:1.

Prostar/star is not difficult to learn to use one just needs to think more than with fluent. I learned it by doing the tutorials, there weren't any courses in 1992, but I did attend a training course later when they used the EASY-GUI interface and it was useful.

Hope this answers everything

Cheers

Allan

I am curious that how many pre-processors are available for Star-CD users to create a geometry model and mesh it. And by the way, if a new user tries to learn the pre-processor, which one should be the first to study? Thank you!

Prostar using the commands (instead of clicking with the mouse).

strictly speaking, STAR-CD has one pre/post-precessor PROSTAR. However many mesh generators such as SAMM, ICEM and others can be plugged in.