This is an open message to everybody. I was wondering if anybody could advise me on the best CFD package about. I work for a small aerospace research company who work with complex airflows on a daily basis. I was hoping that someone with more experience than myself could advise me what would be the best software to use for a person who has minor experience using Ansys CFD.

This question (or similar ones) is one of the more popular ones on this forum. The usual advice is for you to go into the archieves and read the previous answers. That will take a little time.

After you've done that, you may be able to restrict your question a bit. It will matter if

1. your colleagues are doing CFD (if so, with home-grown or 'adopted' codes from DOE or NASA or academia for instance),

2. if you have the experience to modify existing codes (including commercial codes),

3. what your time span is - if an answer is needed in a few days and you're just starting, a commercial shop may be about the only way - which one? it depends? can you afford to do that?

And on and on. It's an important question each time it's asked. But there're a lot of potential answers!

Hi,

I my view none of 3 major i.e. Fluent, StarCD or CFX are suitable for Aerospace Application. If you are from USA, then better go for NASA codes like WIND, TETUSS etc. Out of three major CFX is the best suited for Aerospace Application if you still prefer comercial solvers, there are many specialized software companies for aerospace applications, have a look at software section for further details.

Regards

A.S.

Hi Shane,

good question....well I would guess that it depends on what problems you wish to solve, what your budget is, timelines, experience, etc.

In my experience the specialized NASA codes are going to be better for a specific application but you will not get much support nor versatility for other "off-design...code wise" applications and you really have to know what you are doing.

If you have short time-lines, a wide range of projects, and less experience....you probably will have to turn to commercial codes!

In terms of commercial solvers the most sucessful in a broad range of applications is Fluent. CFX is very strong in turbomachinery applications. STAR is very strong in automotive. CFD++ for external aero. Look and see what other companies in the same field are using....there probably is a good reason!

Take a serious look at what meshing tools are available too ...as this often is the bulk of project time. Make sure it works for what you really want to do.......try it out since the what adverts and salesmen say is usually superficial and over-optimistic.

Let us know what you find...............Bak_Flow

Start with Fluent.

Hello Shane,

If you want to go the commercial route, come up with a sample problem that you would be interested in from a day to day basis. Then have a competition among the various code providers to see who is best. You will get an idea of what it is like to work with them, if the code works for your problem, and see how they utilize their code for it. If a commercial vendor shys away from such a challenge, then maybe they are not the best solution.

Good Luck,

Erich

Can you tell us what is your typical problem? Luca

Thank you all for your input. My problem relies on a very complex airflow system. Air is sucked in via an axial or centrifugal fan. The fan is located in a duct and the air is nozzled out at right angles to the inflow just after the fan. The air is expelled over an expanding aerofoil section. I know this is a complex airflow as i have had consultancy with various companies specialising in CFD and a number of university Professors. They have all given advice on potential packages to use but would like to get the opinion of others.

You can choose between star-cd and fluent. After few years, with fluent you will be effective but it seems you will do about the same thing all the time a bit like the generic right foot kick of David Beckham. With star-cd you will have to learn much more things but be able to do what nobody else can do more like Zinedine Zadane.

Shane,

I am a CFD consultant in the US, and I have used all the NASA codes as well as many of the commercial codes. As you know there is more to analyzing fluid dynamics than just running a CFD solver. In terms of cost, the modeling/ grid generation phase is far more expensive (especially for complex configurations) due to the intensive labor requirements.

I have now settled on STAR-CCM+ from CD-adapco. They started developing this code from scratch a few years ago with a focus on ease-of-use. From a single user interface you can generate the mesh, run/monitor the solution, and post-process. It's very nice. It has the capability to import any mesh topology (in case you need to run legacy models), but it is the first code to support polyhedral meshes. Polyhedral meshes have many advantages but the primary are: 1) easy to generate complex models, and 2) faster solutions with same or better accuracy.

Good Luck, David

The best CFD programme is the one you like and find it easy (for you) to learn and generate the mesh. All CFD programmes that are now available in the market and those that will be written in the future Solve the Navier Stokes equations, the conservation of matter and the Energy equation. To the best of my knowledge, all are based on a finite volume discreatization and etc. The only exception is the one written by Exa as it is based on Statistical Thermodynamics principles (Boltzsman Equation) So where is the difference, The Graphical User Interface and how easy you can post process the output of the solver. Either you follow the recommendation given by Erich above or ask the CFD vendors to install their programmes with a free evaluation license (Normally 30 days), and see by yourself which one suits your needs etc

I'll correct a bit of misinformation here. Polyhedral meshes are not new with Star. The control volumes constructed by CVFEM codes (like CFX) are also polyhedral in shape. The difference is that they start with a 'standard' mesh (hex/tet/etc) and implicitly calculate the 'dual mesh' within the solver itself.

It is not the same approach.

We have also found, that meshes which are ok in star were not usable for cfx.

if approach is concerned then, the approach of starCCM+ and fluent 6 are exactly same (for Navier-Stokes, you can consult the manuals for both and compare), internally fluent also treats them as polyhedra and i do not think starCCM+ is first to do it. Yes, as far as mesh import and generation is concerned starCCM+ could be said to support polyh. meshes where as fluent could not be.

Dear zxaar, what Fluent versione are you referring? I've never seen poliedhra cells in Fluent (up to 6.2 version). Fluent Inc. said they're working on polihedra cells and this will be available from 6.3. Luca

A tet or hex cell is just a special form of a polyeder.

A face based mesh description is neccesary for using polyhedral cells. This was to my knowledge first invented by Henry Weller (OpenFOAM) and then used by the Fluent.

Later some of the core developers (Wayne Smith ...) left Fluent and joined CD-Adapco to develope Star-CCM+. Thats why the both codes have "some" similarity

So Fluent is in theorie capable of handling polys. They just dont have a suitable mesh generator for polys and the code probably needs some adapting for the use of arbitrary polys with no restriction of the number of faces.

The codes which can handle arbitrary polys at the moment are OpenFOAM, Comet and Star-CCM+ (and Star4). But the only usable mesh generator comes from CD-Adpaco.

Hi Shane,

Firstly, my understanding of the polyherda question is that Star produces a visual polyhedra mesh. CFX and Fluent do not produce the visual representation, however the methods behind the solution are the same. So big wow - you don't get to see a polyhedra mesh !! If you need clarification just check with a Rep from CFX or Fluent and they will show you the documentation to go with this.

Comercial codes and CFD practitioners can be very biased. I know of a lot of users that sware by their particular code. This is because they have developed the skills and knowledge to get their code to perform the way they want it to. The mistake is to then think that this makes it better than the other codes. If say you put someoen who has used Star of the past 10 years in fromt of say CFX and told them to model a basic test case then they would struggle. This goes for all the different codes. So beware the passion of a user who screams one code is better than another.

I would setup a simple test case, using say your inhouse CAD package. Advise the different CFD solver companies that you use CAD package X and want them to come round and setup and run the test case. Make it relatively simple - you don't want them to be meshing for hours or having to tidy up the geometry. Then take see which system you prefere. Which seems more intuative. Also check the stability of the simulation, did it converge quickly ? And most importantly look at the results, how good are they and how simple was it to extract the information that you would need as an Engineer.

Good luck, its the most difficult question to ask / answer on this subject !!

Hi David: I wanted to ask you about the "NASA codes" - I've played around with VULCAN and considered Kiva. I am using PHOENICS (since the CHEMKIN-II solver's been in it for many years) - and I was wondering if you could recommend any of the particular NASA codes for comparison runs with my PHOENICS runs - for usable, accurate, sub- and hypersonic combustion. (combusting flow is my main interest)

Thanks! PattiMichelle

yes you are correct, fluent is restricted from point of view of mesh.