[FlyingCat]

Hello,



I am trying to simulate a 3D fluid-structure interaction problem.



The problem corresponds to an airfoil (chord=8 cm) in water (1 m/s).


The mesh consists in 15 M of cells. The boundary layer is meshed with approximately 20 cells (I cannot do better with cfMesh and SnappyHexMesh).



I ran the fluid simulation (without interaction with the airfoil) and had consistent results (until 1 sec, then I stopped).



Now, I would like to run the case with the deformation of the solid. I am using preCICE and CalculiX as the interface coupling and the solid software code. They work perfectly and the question does not concern them.


With a CFL of 0.3, with or without the boundary layer mesh, the fluid-structure simulation diverges. After 15 time steps (dt=1e-5), the velocity increases up to 3 m/s below the airfoil, at the first boundary cells whereas the displacement is smaller than 0.02 mm.


I tried smaller time steps but I cannot use smaller than 1e-6 seconds due to the calculation time. And the problem remains even if the velocity diverges further (30 times steps for dt=5e-6 s and no boundary layer mesh).


So my question is, is it possible to run a 3D simulation of an incompressible fluid, which is based on a fine mesh, with mesh motion?



Does this mean that I must either decrease the time step or coarsen the mesh?



Sorry, I cannot share any picture due to containment. Thank you