Hi,

i've run a 2d flow past a circular cylinder and the drag/recirculation values are quite good.

i tried to modify it to simulate a rotating/spinning cylinder. i believe i only need to change the boundary velocity and nothing else. is that so?

i tried to run the simulation for different angular velocities at Re=20. the velocity vector plots seem correct compaared to the published one. cd & the viscous lift force are quite good. however, the lift force due to pressure is always about 55-65% of the correct results.

i wonder what's wrong. Did i miss out something? is there any modification required at the post processing stage?

tks in advance

compare the pressure distribution on the surface with published data. also check on the locations of stagnation point and separation point.

If you want to perform a moving grid simulation first you have to decide what is the system of reference in which you are resolving the governing equations: 1) if you choose the absolute frame of reference you have to rotate the grid at each time step ----> moving grids (computationally expensive but simple to implement)

2) if you choose the relative frame of reference you have to introduce a grid velocity (which affects the convective fluxes) a source term which contains the centrifugal effects and a projection of the far-field conditions on the relative frame of reference (much more difficult to implement but more efficient).

I think you are missing something if you only "changed the boundary velocities".....

don't get confused by mar's comments, zonexo. your method is valid and the only reasonable thing to do for your case.

Mar, you're missing the fact that this case of structural motion is very simple, almost trivial: the cylinder is rotating about it's own center axis, such that the wall surface is displaced onto itself. It makes no sense to rotate or even deform the grid, since there's no change in the shape or position of the cylinder. Likewise, it would make no sense to use anything else than the absolute coordinate system. What doesn't work for the general motion or deformation is perfectly fine for this special case: All you need to do is to assign the appropriate boundary velocities, i.e. the vector product btw omega and radius, on the cylinder surface.