[ander]

Hello!

I'm looking for a modifiable open-source CFD code for a personal project. It should preferably be written in C++. The code needs an N-S solver (preferably implicit for efficiency) that can solve transient problems for flow involving shocks. Modularity and code efficiency are other features that are highly appreciated.

I have looked into OpenFOAM, but its selection of compressible solvers is limited, rhoCentralFoam is the only solver that can capture shocks accurately, and it's only explicit.

I would therefore like to get some pointers to alternative open source C++ CFD software. I have found one called SU2, does anyone have experience with this code?

Thanks in advance!

[LuckyTran]

What's wrong with rhoPimpleFoam?

Btw any software you deal with will have the same issues. Any implicit solver will have the same challenges with resolving shocks compared to a density based solver. The inability of the solver to "capture shocks accurately" is a result of the methods used. So tell me, what methods are you planning to use and i'll find you a solver that does exactly that.

[ander]

My issue with roePimpleFoam is that it is unsuitable for discontinuities as far as I'm aware.

My wish is to use an implicit density based solver that can capture shocks accurately. LU-SGS or GMRES are examples of methods that can be used to solve the linear system resulting from the implicit discretization of this. A dual time step methodology can be used to obtain a transient solution. I'm aware of a couple of user implemented LU-SGS solvers in OpenFOAM, but not sure how well these are validated.

[LuckyTran]

So what you're saying is, even if I were to conjure up an implicit density based solver, you wouldn't use it because I haven't provided the data to validate it even though you (would) have the source code needed to verify that the code does what you say you want it to do?


Also, are you sure rhoCentralFoam already doesn't do what you are asking? Have you looked at it? Because from what I can see, it already can minus your specific choice of linear solver. GMRES was implemented a long time ago, idk anything about LU-SGS. In either case, they're just linear solvers. If you want a specific flux scheme or a specific time-stepping scheme, then there might be some work. But please don't just brush off a solver because "it can't accurately capture shocks" was something you read in some blog.

[ander]

I went with SU2, and it 's honestly much better suited for high speed aerodynamic applications than openfoam. It has a much wider selection of solvers for compressible flows. To your point, trying to modify one of the solvers in openfoam to my needs would present a giant task in itself, which is unnecessary when SU2 already has the desired capabilities.

[LuckyTran]

I mean OpenFOAM had everything you asked for, it's just not the software you wanted. But whatever, it's not like both software have their origins in the same school or anything.

[arjun]

Originally Posted by ander

It has a much wider selection of solvers for compressible flows.

For example compressible multiphase solvers like blastfoam or diffuse interface solvers?

I am curious what you can do with su2 that you can not do with openFOAM apart from may be su2 converges faster.