Hi all,

I am simulating a Re=300 flow over a sphere. Acoording to the literature and previous numerical studies, wake at Re=300 is periodic and has shedded vortices..BUT i couldn't get any such periodicity...I read in some other papers..that at low reynolds numbers like..300..1000..flow is generally stable..and to start the vortex shedding..some perturbation has to be given..so I gave a perturbation..like multiplied the streamwise velocity in the upper half of sphere by 1.1 at some (only once) time step. Actually it seemed that it did have vortex shedding but only for some time..it couldnt be sustained and the flow is back to steady again..i am wondering what could be wrong..do i need to refine my grid more..??

regards shailesh

Difficulty in simulating shedding of vortices is I believe a classic problem.

If your numerical scheme (differencing or whatever) has excessive numerical dissipation, it can show up as what appears to be an artificially high viscosity - making the computational Reynolds number much less than the one intended. And, at a lower (computational) Re, the flow is non-shedding. Also, using upwind differencing with a mesh that's too coarse can have the same effect.

Take a look at some of the early (1960's!) papers by P. J. 'Pat' Roache (his book has a chapter on this stuff!) and C. W. (Tony) Hirt. Hirt's J. Comp. Physucs paper on 'Heuristic Stability Theory' shows the technique needed to analyze your algorithm.

It may be that you need only resolve your mesh (add more cells) to get a good simulation.

Good luck!

Jim..thanks a lot!! for the informative comment..i was thinking along the same lines

Simply refine your grid to see if unsteadiness occurs.