[luckyluke]

This problem is shown as follows:

,
or similar 3D case.

I wonder if the present OpenFOAM1.2 have the solver for such problem. If the FOAM1.2 has not got this feature, how should I do? From whcih exsisted solver should I carete a new one? Give me some ideas, please.

Thanks all. This is not my project, but I am interested in this simulation. The Fluent6 did have this feature.

[luckyluke]

It is should be noted that the mesh topology will change along the motion.

[lr103476]

Fluent is only capable of solving these dynamic mesh problems with first order methods marching in time.

By the way, the mesh you showed is far too coarse and the solution by far not very accurate. The movement of the cells from on timestep to the next restrict the choice of this timestep. So higher order time integration techniques are very welcome for this type of problems.

I am investigating if OpenFOAM can do the job for 2d flapping wings, and later for 3d wings. It is also very dependent on the Reynolds number. I am studying flapping wings at Re=110. Then, the behavior of the second order discretized convective flux at these changing meshes may play an important role.

Goodluck

[hjasak]

Hi Frank,

Just for my info - did you get it to run correctly? Do you have topological changes as well or just mesh motion?

Hrv

[lr103476]

Hi,

I am still working on a proper set-up of space and time discretization techniques using static cylinders at Re=150. Just for validation. The plan is to extend to moving cylinders using only mesh motion at first. Later also topological changes may be implemented. And, if possible, also some body forces, which depends on the complexity of the wing motion in 3d.

When I succeed in simulating flapping wings/cylinders I will post the results.

Could you maybe comment on this? Maybe you've got some tips on the solver settings at Re=150?

Thanks, Frank

[hjasak]

Heya,

For the test above, you don't need much: take icoFoamAutoMotion and the automatic mesh motion solver will do the job nicely. The second-order discretisation on moving meshes has been tested in detail by my colleague and a former student dr. Tukovic (and he IS thorough) - you will find the details in his PhD Thesis (unfortunately not yet translated to English in full, but I'm sure he can answer questions).

In order to use the automatic mesh motion solver all you need to do is to prescribe the motion of the boundary; the rest is done for you automatically. :-)

For topological changes, I would really need to see the mesh - it very much depends on what the domain, mesh and motion looks like. If you'd like to force the one above to very large deformations, I would have some layer addition/removal in a circle/cylinder around the wing. You would probably want to use a combination of automatic mesh motion and topo changes (ask Tomasso about what we did for in-cylinder flows).

Finally, setting up the solver on Re=150 should not be too tricky. Decent (second order in space and time) spatial discretisation will be fine. You may wish to have a go at my bounded second order spatial scheme if you're worried about numerics. Check that PISO is converging... No further thoughts, really.

Please keep me posted,

Hrv