Swirl flow convergence problem with simpleFoam
I am running a 3d swirl flow case with Openfoam and Fluent.
I am using simpleFoam as solver with kOmegaSST turbulence model. I got a converged solution with ANSYS Fluent 14.0 but the solution with OpenFoam is not converging...
i.e. the residual is oscillating and the flow field doesnot converge to a single solution.
I have tried to summarize my case details.
The geometry has two inlets. A snapshot of the geometry(cut from the middle) is given in the link below.
The Discretization and solver settings used in Openfoam are
The checkMesh log, boundary conditions at 0 are included in BC folder. See attachment...
Openfoam and Fluent solutions are shown in attachments. see snapshots...
Any comments, suggestions to get the converged soltion with OpenFOAM and about modelling swirling flows would be welcomed.
You can view the snapshots in the attachments also.
All your links are not working, you need to use public link if using dropbox.
I think you need to look at the choice of divScheme in the fvSchemes file.
Using upwind could maybe give you a more stable solution but difficult to say since I cant see the files,
Also I think the reason you get bad convergence could be that the problem is physically transient. So the swirling component could change location/direction/magnitude. This is a very common picture if you think its steady but in reality is transient.
The reason why Fluent gives a steady solution could be the choice of scheme and/or relaxation factors and limiters that you aren't aware of.
PS. Increasing the outlet length could also help or setting the right BC on the outlet, but again cant see your files.
links working now
First of all thank you for replying.
I was using dropbox for the first time so didnot know how to make links public.
But I dont know why the some links are not working. I have attached the folder containing
boundary condition files and log files.
My main confusion is how the solution converged with Fluent and not with Openfoam.
If it is a transient flow then it should not converge in Fluent too.
I tried to keep settings almost same in both the cases.
Foam behaves like the literature and to get the same result you need to to have the same solution strategy/setup etc.
You could try these files for fvSchemes and fvSolution, but I doubt the solution will have better convergence. Your BC setup is looking ok.
The steps I would try is to increase the outlet pipe length and next step would be to try the pimpleFoam solver and see the transient behavior of the swirl to see if it moves around etc. And use the simpleFoam data as a starting point for the transient simulation.
Swirl flow convergence
Hello Mr. Nielsen,
I am trying to get a steady state solution with low swirl (may be 50%) of the max swirl velocity because with full swirl the flow is highly transient and it makes no sense to make a steady state solution.
So what i did is
1. First I obtained the steady state solution with zero swirl velocity.
2. Then I increased the swirl velocity step by step (10 %, 20%, 30%... of the max swirl) and tried for the steady state solution with FLUENT and OpenFOAM.
I will try with your discretization and solution settings now for convergence.
I want to carry out a similar exercise of comparing the results from OpenFOAM with that of Fluent. I am assuming that the mesh that you are using here has been was created for Fluent and converted for OpenFoam. Could you please tell me how you went about it?
Thank you in advance!
Converting a Fluent mesh(.msh) file to OpenFoam is straightforward.
Check OpenFoam utilities.(Mesh conversion)
Thank you for quick response and sorry for my late response.
I was able to figure it out and as you said it was quite straight forward.
|All times are GMT -4. The time now is 00:25.|