OpenFOAM vs code-saturn vs FreeCFD vs ...
 

Which open source Navier Stokes solver do you suggest for external aerodynamic calculation (airplanes (subsonic, transonic and supersonic), cars (maybe also internal cooling flows))?
In case other is selected, can you post the name of the code and some useful link?
Thank you,
Xwang

Ps for the preprocessing I think I'll use Salome.

Openfoam does not seem to have good support for compressible flows, especially with strong shocks. Thats what I have gleaned from browsing through it but I would like to hear from others. I have no experience with code-saturne.

I would have recommended freecfd but their 3-d implementation is not complete; they are trying out a new limiter scheme which does not seem to have been finished for 3-d. But if this situation has changed and the turbulence models are also working, then freecfd would be very good.

Praveen is right. Limiter implementation in Free CFD currently has some problems for 3D. I took a break from developing Free CFD to finish my dissertation. I will soon continue. I actually changed the limiter implementation and fixed the issue already. It is rather trivial. I am now testing and adding a few more features for the next release of Free CFD. Tune in to www.freecfd.com for an announcement.

Turbulence models, on the other hand, are working well.

I suggest you just try the example cases that comes with the code to judge if it is right for you. After all, I am sure the other codes would work for you just as well. It is a matter of personal taste in the end.

What is your opinions after two years?
The majority of polls votes are for openFoam, but the comment are all for freeCFD.
Thank you,
Xwang

My understanding is that since OpenFOAM is not a coupled method it can not handle steady state supersonic flow, i.e. the solution does not converge. This would also mean that OpenFOAM can not handle supersonic pockets. There was an effort, AeroFOAM, to create a coupled solver for such flow, but I gather it is not supported by anyone. In the 2010 OpenFOAM workshop there was a paper on SIMPLE vs. Coupled, http://web.student.chalmers.se/group...SlidesOFW5.pdf

This is possible with rhoCentralFoam, which is the only truly compressible solver available in openfoam.

In the past I have not been able to get rhoCentralFoam to converge to machine zero for transonic (i.e. has supersonic pockets) or low speed supersonic flow. For transonic flow the solution residuals asymptote to a value and for supersonic flow it blows up. When compared to other answers, for example these results http://www.hegedusaero.com/examples/.../Vassberg.html, rhoCentralFoam did not compare well. Note, I did not publish my rhoCentralFoam results on that page. That's not to say rhoCentralFoam doesn't give good results when used by someone who knows the ins and outs. I was just not able to get it to work.

I see you've done some work with NACA 0012. Were you able to converge the results to machine zero? Also, what conditions did you run it with (mach, alpha, reynolds number, turbulence model, and outer boundary conditions) and what were the CL and CD results. If you give me that information I'll run my solver on it and see how it compares.

After digging up information on rhoCentralFoam I'm uncertain it does turbulence modeling. Does rhoCentralFoam include turbulence modeling?

rhoCentralFoam does only global time stepping. So getting convergence for steady state problems will need too many iterations. I have not looked at how well it converges, since it does not seem to calculate residuals. I am trying to implement local time stepping and run it in parallel.

It does not seem to have turbulence models. It should not be difficult to add it though.

I can't remember the details, (I tried different solvers, bcs, etc.), but for rhoCentralFoam I believe I created a code to compare the solutions at two adjacent (n and n+1) time steps to get residuals. The runs were inviscid and the grid was coarse. Anyway, unless I can find my runs, or recreate them, I should not say much.