I'm expanding into using CFD sim software & am looking at CFX & Fluent. Can people who know both tell me the strengths & weaknesses of each?

I notice CFX got bought a couple years back by Ansys, which I have used (but I prefer IDEAS). Anyone who used/uses CFX before & after, any good or bad changes to the product or support because of that?

I know that CFX has more inbuilt models when it comes to Multiphase flows. You should wait and see what others have to say.

I don't have a direct experience on CFX. However all my collegues who uses CFX tells it's more robust and accurate than FLUENT, but it's a little more difficult to be used.

Maybe it's the price to pay for accurate solutions

Regards,

ap

i have used both the softwares, at least for combustion i won't go for fluent, in Fluent we have only two realistic models EDC and flamelet, out of which EDC is real pain in arse to get converged, and i feel as if we almost do not have it.

i am sure those who has tried EDC know what i am talking about.

I would add some consideration about FLUENT.

1) FLUENT is easy to use and has a good interface both for pre-processing (GAMBIT) and for setting the case. Postprocessing is not that good, but it's ok if you consider that it's integrated in the solver graphical interface.

2) FLUENT is lacking from a mathematical point of view. It doesn't allow to choose among all discretization schemes available, even when you enable the specific option via text commands. CDS seem to be considered a tabu (they are hidden for all models except LES) at FLUENT, while everyone knows that the upwind schemes are very diffusive and highly inaccurate in many cases.

Morover, if you simulate multiphase flows, you only have the phase coupled SIMPLE alghorithm for pressure velocity coupling. Nothing else is available.

Always in the Eulerian model, you can only use the first order implicit discretization scheme for time integration. The second order (Crank-Nicholson) is not available.

3) FLUENT can be personalized using UDFs. UDFs are perfect if you want to do simple changes to the standard models, but they are a pain if you need to implement some complex additional model.

This is mainly due to the *lack of documentation*.

The UDF manual is completely *insufficient*, and if you need something you have to contact FLUENT support.

In the UDF manuals you don't find many important informations, many macros are not documented, some examples are wrong (the re-implementation of the Syamlal drag model).

A good UDF manual would provide just a list of the macros required to communicate with the solver and of the DEFINE functions, like in a doxygen documentation.

Now, I listed some of what I consider the main lacking parts of FLUENT.

However, if you want a general purpose CFD code and you're not interested in research applications, FLUENT is probably ok, even if some specific area of modeling may not be covered in the proper way. Just check if it's suitable for your purpose.

Regards,

ap

i completely agree with you that when it comes to documentation fluent is certainly not good thing. but CFX documents are well written.

Good to hear. Quality documentation is essential for a CFD code.

Dear George: If you want to model free surface flows especially using VOF model; the FLOW-3D is the best.\ regards

I have worked with both CFX and fluent and I find that CFX is more robust and much better for bigger simulations. The overall control on different parameters is much better in CFX and its more user friendly compared to the Fluent. The documentation is just too good. Especially for Tascflow. The best part os the post-processing, which is probably the best.

So what combustion model does CFX offer? I have been working in this field for auite some time and mainly using in-house code and FLUENT.

I offer that n selecting a CFD software focus on the code (engine) not post and pre processor. in solving turbulence fluid flow I prefer FLUENT.

for that u can check cfx web site there you can find in detail .... but i strongly feel that it depends upon person to person that what he likes .... some people may feel with fluent at ease ..some may find cfx more apealing ... for me in fluent flamelet is one good model for combustion ..and edc though is good .. but has always troubled me .... and after reading lot of postings here about problems faced by people with edc on fluent ..i have finally made opinion that edc with fluent just sucks.

I've seen and used both so I'll add a few possibly useful points.

* Fluent

- Preprocessing (Gambit and Fluent GUI together) seem like they have a fast, yet somewhat antiquated (based on tcl/tk or something like that), user interface. So, if you don't care about a modern look and feel GUI then it will be fine.

- Their solvers have many, many, many different models. However, their feature matrix is extremely sparse. i.e. model x works with model y but not with model z, etc...

- Many of their boundary condition and source term type models are quite "engineer" oriented. i.e. enter some sort of commonly known engineering data and Fluent automatically translates that into boundary conditions or a source term usable by their solver.

- Flow solver robustness is questionable on complex models (combustion, multiphase, etc..) . You will have to play around alot if that is what you are doing.

- Their post processor is nice in that it is directly integrated with their solvers, but is really limited in functionality. You pretty much have to bail and use Ensight or Fieldview to do anything sophisticated.

- and has many presupplied models that are basically built in "macros" fo where you can easily input a few "model" parameters that are

* CFX

- Preprocessing is maybe a bit slower than Fluent in terms of overall speed but it keeps on improving. CFX has a much more modern GUI based on QT, is fully integrated into ANSYS Workbench now which allows direct connections with CAD, is configurable like Fluent's solver GUI. The workflow in Workbench is really really nice and Fluent has nothing that even comes close to this.

- On the other hand, Meshing with CFX is a bit of a dogs breakfast. There is CFX Mesh and ICEM Tetra/Prism and ICEM Hexa. You may have to learn all of these. They are great meshers but it sucks to have to possibly use all of them.

- The solver is very robust. You will play around much less getting complex models working. That is not to say that you wont play around at all!

- There are less built in, user friendly, engineering specific models in CFX (eg: Heat Exchanger Model, Fan Models, Spray Atomiser Models are a few I can think of).

- In a pure tick box feature war CFX would loose, but really, who cares.

- On the other hand, the feature matrix in CFX is far far far fuller than Fluent's.

- However, that being said, CFX is completely flexible enough that you could type in a couple of CEL expressions and possibly 1D interpolation tables, etc... to give yourself similar functionality on your own.

- Post processing wins hands down in CFX. I'd say other than parallelisation, the CFX post processor is on par with something like Fieldview or Ensight.

Hi have seen and used both codes over many years.

I think commercial code utility is covered by:

- models available - solver accuracy and robustness - adaptability to specific problems

In terms of models:

CFX-5 has models for mainline work and better models for turbomachine applications.

Fluent has more models and validation in a broad range of applications including combustion, radiation, in-cylinder flows, fuel cells, acoustics, heat exchangers, etc.

In terms of Solver accuracy and robustness:

Both codes are about the same in accuracy...but hard to compare on an unstructured mesh as CFX is cell-vertex Fluent is cell centered.

Robustness for steady problems is better in CFX but it does not usually take too much tweaking to get Fluent to converge. Fluent has more options and control for unsteady problems.

In terms of adaptability:

CFX has an easy to use CFX command language to do simple things. Otherwise CFX is very closed and user Fortran is not well documented and renders it unusable.

Fluent UDF's are documented, archived and very flexible.

Basically CFX is a more robust solver but Fluent is a more robust solution.

There are my thoughts

Bak_Flow

Hi,

CFX has a subset of the modern combustion models. There are significant gaps like a good pdf generation method, flamelet is adiabatic, no EDC...sure it is hard to converge ALL complex physics models are.

Fluent simply has more combustion models and they have been working for longer.

Regards,

Bak_Flow

can you explain to me what do you mean by "CFX is cell-vertex", fluent i know is definitely cell centered

In many cases Fluent doesn't converge or gives not accurate results due to it's poor math implementation.

I don't know how CFX is in this field.

Regards, ap

In my opinion UDFs are not well documented at all. Many useful macros are not cited at all, and many are just used in some example without explanation.

If you must implement complex models you often need to contact FLUENT support.

Regards, ap

not only udf, there are lot of things not documented, for example , there is a big problem that in sliding mesh problems after defingin the sliding interface mesh check fails (most of the time, at least with me it happened),

i asked fluent support many times, this time they asked me to execute a command on tui before defining interface to avoid this problem, i did what they said and yes it works fine,

this is no where documented and i bet there are lot of people who are facing similar problems, but fluent shamelessly not willing to put this things to documents,

and this is the reason when Jonas asked about splittign this forum i said to put tip and tricks section so that we can gather all such tips for fluent users who otherwise are on mercy of fluent support.

Hi ap from POLITO. Yes you're completely rigth. Fluent UDF documentation is really poor. I had to look at all header files to learn many more macros and did a lot of trial and error. I'm a little disappointed about this side of Fluent. It's a powerful tool but users cannot use all Fluent power because of this. I don't ever try to talk about scheme programming with Fluent...it's a shame!I had to learn scheme commands looking examples from fluent's site...I asked some help from Fluent support about this...but...it's better not to go on...Luca