[richard]

Hi Patrick,

Yes, it is a bit of an overkill, but I will be modifying the trailing edge to have wiggles in it's profile (both chord and span wise) - I'm new to panel methods, but according to Martin Hepperle they aren't appropriate for very wavy surfaces, so for now I'm soldiering on with fvm.

I'll give the zero velocity patch a try and see how I go.

Thanks Pat
Richard

[richard]

I modified the two TE cells and it just makes a zero velocity point at the tip of the TE around which is a colorful wagon-wheel of velocity!

I had more of a look at panel methods, and I think I was wrong (or my interpretation of Martin Hepperle's words was wrong). I'll try it and if I run into snags with my wavy profile I will increase the number of panels and go from there.

If anyone else has suggestions for making the FVM potential solution on an airfoil work, let me know, I'm willing to experiment.

Cheers,
Richard.

[mattijs]

You could try to split the mesh along faces which more or less follow the expected streamlines (using createBaffles). You'll have to program a bit to construct a faceSet with all those faces in it.

Or just allow some more physics and run simpleFoam ...

[pbo]

oops,
forgot that the governing equation in potentialFoam is the pressure equation. So the zero-velocity boundary condition at the TE, should be indirectly enforced through a pressure boundary condition. Try to enforce at the TE panels (on top of the zero-velocity bc) a pressure equal to the total pressure in the freestream (roughly 0.5*U_inlet^2 + p_outlet -- rho is missing cause p is normalized by rho in incompressible solvers).

Otherwise, as Mattijs said, try simpleFoam. Use a zero viscosity, and slip bc at the airfoil surface to simulate an inviscid flow, along with a bounded convection scheme (the amount of artificial viscosity produced by these schemes should enable the Kutta condition 'automagically' like in a real viscous flow, and provide a stabilizing effect on the computation -- hopefully)

As regards the panel method, any higher order one (quadratic doublet or linear vortex strengths) should do the trick (with an appropriate panel distribution around the bends, and provided the bends on your profile are more of the rounded types than of the saw-toothed type).

At some point, you should however give in to viscous flow simulations, because the waviness of your airfoil surface may have a significant effect on the boundary layer behavior.

[richard]

Thanks guys,

Yes they'll definitely be the rounded type bends - and yes I'm doing viscous solutions too (actually I'm comparing them).

What I'm aiming for is cheap (potential + integral BL) and expensive (rans) methods of finding the boundary layer properties at the TE - and hopefully have both in OpenFOAM.

I just tried the pressure condition and I get rather high velocities (1290m/s!) around the TE.
I'll try simpleFoam to see if it works, but I would assume the run time would be in the same order of magnitude as the ones with viscosity, which makes the whole exercise useless for me.

Perhaps OpenFOAM would be good at inverting a matrix from a panel method formulation..

[cwang5]

Hi,

I am new to the moving mesh option in OpenFoam, and I'm wondering how to setup the boundary condition for the "tail" motion? I have managed to use splitmesh to generate the tail as internal surface, but I am lost in setting up the boundary condition so that the root of the tail could not move. Can someone help me on this topic? Thanks

[mehulkumar]

Hey..Guys.
I am a new user of OF.
I want to simulate subsonic flow over NACA0012.I am using O topology.I want to learn, to make mesh such that according to angle of attack, boundary should be change.I have not much idea about it.

Pls. guide me.
Give me some reference.

[cwang5]

Originally Posted by kassiotis

For the one interested, I have made a short rapport on the moving mesh subject. Hope that will help someone.

Please, be kind with my (poor) English ! And remind this is only a draft for the moment

<https://perso.crans.org/kassiotis/op...movingmesh.pdf>

CK

Hi, I am wondering how exactly did you enforce the null displacement at the origin of the appendage? Do you specify it in pointMotionU file or else where?

[marico]

Hi mesh movers!

First of all thanks for providing the tools to the responsible humans!
The tetFem solver (cellDecomp.) is quite slow but that doesn't matter. My problem now is the following:

The solver works fine and solves with 300 - 700 iterations; tolerance = 1e-9. Mesh stays fine. UNTIL the INITIAL residual falls to ~ 1e-8, only 10-20 iterations are done and the mesh gets destroyed beautifully...

1. Can this be caused by skew faces (I've got 50 out of 500.000; Max skewness 77) ???

2. Or due to a too high timestep ???

3. Is there a possibility to define a minimum number of iteration like for other solvers ??? Sorry: How to do it?

BTW: I'm using 1.4.1-dev. Thanks a lot!

Marco

[Harbeer]

Hi guys, i am simulating FSI on a zero-thickness membrane. But the mesh motion solver will destroy the boudary layer, please see my pictures. What i have figured out is that, the problem happens due to the laplacian equation is solved on the cell displacement instead of point displacement, and when doing volumn-point interpolation it gives inverted cells on the boundary layer.

From google i read the vortex-based mesh motion is only in the OF-dev, but in standard OF there is only fv based mesh motion. Therefore, i can not use Mr. Jasak's motion solver, although it looks able to solve my problem.

Any ideas?

before inverted:
beforeInverting.png
after inverted:
afterInverting.png

[hua1015]

Dear Foamer,
Now,I want to do FSI with openFOAM to solve fluid and other third FEM open source codes to solve solid.But I am not sure how to invoke FEM solver in openFOAM? Since it seems that you have coupled a FEM code (Feap) and OpenFOAM to solve FSI.
Any suggestions will be appreciated.