[Greg Givogue]

Hi,

I've been trying to model an aircraft external tank using simpleFoam. When the flow is axial (x-direction only) simpleFoam convergences <1e-6 on all initial residuals and the forces compare well with experimental results. When I change the flow so that it is non-axial (10 deg of angle of attack and 15 deg of side slip), simpleFoam does not converge <1.4 e-5 and the drag values are 3 or 4 times greater than expected. The only thing that I change between the axial and non-axial conditions are the domain patches (rectangular domain with 6 patches). For the axial case I have 1 inlet patch and 1 outlet patch. For the non-axial case I have 3 inlet patches and 2 outlet patches. I have set the patch that the pod pylon rests against as type slip.

Geometry - 35cm diameter, 218 cm long, with conical tips
Mesh - 6M cells, y+=3, growth rate from body of 1.15 (30 layers prisms the rest tets), wake refinement, checkMesh ok
Flow - Re=1.77e7, V=121.6m/s, alpha=10 deg, Beta=15 deg
Turbulence - Komega SST

I have read several of the posts on this subject but I have not been able to find the solution. I'm hoping someone with more experience can point me in the right direction.

Thanks in advance - I appreciate your help!
Greg

[Greg Givogue]

Just corrected the KOmegaSST bug in 1.7.1.

Any suggestions on the the convergence problem?

[FelixL]

Hello, Greg,


regarding your convergence problems: since this is a blunt body at a high angle of attack and side slip I would expect separation around the body. This separation most probably leads to periodic vortex-shedding behind the body, which clearly is a transient phenomena.

Since the SIMPLE algorithm solves the incompressible, steady NS equations, transient flow features can not be captured. Sometimes vortex shedding can be identified by oscillating residuals but usually simpleFoam simply doesn't converge when seeking for a steady solution on an actually transient configuration.


Greetings,
Felix.

[Greg Givogue]

Thanks Felix for responding so quickly. I had a feeling that this was the problem and I was in the process of setting up a run with pisoFoam. I'm a CFD rookie and I've never used pisoFoam so I was a bit hesitant going down that path. Before I get completely off track - can you take a look at my fvSchemes and fvSolution files that I'll be using in pisoFoam? Most of the settings I have chosen comes from the tutorials in the pisoFoam folder and my old set-up in simpleFoam. What is the best solver to use fvSolution and how about the other settings? How about divSchemes?

Thanks so much and I really appreciate your help!

[Greg Givogue]

Actually I lied, I'm having issues with wmake and recompiling the solvers after updating the KOmegaSST.C file. I have started a new thread for this... http://www.cfd-online.com/Forums/ope...implefoam.html

Perhaps someone here will be able to help me with this too... Thanks

Problem solved by playing around with permissions...

[Greg Givogue]

Hi Everyone!

Ok, I have re-ran the problem by first initializing with simpleFoam (1000 iterations) and then running pisoFoam. I get the same problem as before - p fails to converge <10^-5 and k converges and diverges.

I have attached System files for pisoFoam along with log.pisoFoam files (spilt in 2 for uploading). I am completely stuck now...

Thanks!
Greg

[Greg Givogue]

Another problem fixed - my y+ was way to low for using wall functions - should be more like y+>30. Still hoping for suggestions...