[quarkz]

I have read some articles and papers on vortex shedding. Most simply describe the phenomenon and where and when it occurs, but does not give a physical explanation as to why it occurs.

For example, a simple flow past a cylinder. Why does vortex shedding occurs after a while?

Can someone give a clear physical explanation?

Thanks!

[doubtsincfd]

you must already be aware that it happens because of adverse pressure gradient etc

i had the same question and i believe it is the inherent non uniformity in the inlet flow that triggers the vortex shedding (this might answer why shedding happens first from top or bottom of cylinder)

but what about numerical simulation? what if the inlet flow i specify is uniform, without turbulence and the grid generated is also symmetric about, say, horizontal line passing through center of cylinder? which side of cylinder (top or bottom) will first trigger seperation? do round off errors play a role here? I was digging into this at some point but then left halfway.... thanks for reminding.... i will again work on it.... let me know if you find something useful

[truffaldino]

Originally Posted by quarkz

. Why does vortex shedding occurs after a while?

Can someone give a clear physical explanation?

Thanks!

The simplest explanation in my view reminiscent to that of Kelvin-Helmholtz instabilities when strong velocity shear is present (in BL with velocity inflection point or in recirculation regions): A non-zero curvature perturbation of "dividing" streamline will lead to a slight centrifugal force which in turn leads to a change in pressure thereby amplifying the ripple and stronger and stronger amplification and shedding the vortices.

[otd]

Originally Posted by doubtsincfd

but what about numerical simulation? what if the inlet flow i specify is uniform, without turbulence and the grid generated is also symmetric about, say, horizontal line passing through center of cylinder? which side of cylinder (top or bottom) will first trigger seperation? do round off errors play a role here? I was digging into this at some point but then left halfway.... thanks for reminding.... i will again work on it.... let me know if you find something useful

I had the opportunity to work in the T3 group at LANL for a while in the 70's. This was shortly after J. Fromm and Frank Harlow published a (I think the first) simulation of vortex shedding.

I asked Dr. R. A. Gentry of that group how the shedding was initiated numerically. As best I recall, he answered "I'm sure we perturbed the inlet flow. We couldn't wait around for some random fluctuation to trigger it."

Hope this is useful.

OTD

[lin]

I have the same question. In reality it could be the perturbation of the incoming flow that trigger the vortex shedding. But what cause it in numerical simulation with homogeneous incoming flow and symmetrical configuration? Now the only explanation I could find is round off error when using the floating point calculation.

[otd]

I have the same question. In reality it could be the perturbation of the incoming flow that trigger the vortex shedding. But what cause it in numerical simulation with homogeneous incoming flow and symmetrical configuration? Now the only explanation I could find is round off error when using the floating point calculation.

Read the post just above yours. It refers specifically to numerical simulations. The Fromm and Harlow study was done by deliberately perturbing the inflow initially to break the symmetry. At least that's what one of their colleagues told me.

I suppose round-off or using difference algorithms that break the symmetry imposed by the boundary conditions could trigger shedding in a homogeneous inflow as well.

[derz]

Yes Round-off can trigger it. I have run laminar simulations past cylinders and noticed shedding occur, with 100% uniformity in the flow. The only explanation I and my colleagues agreed on was Round-off error.

But also, if you were to use a turbulence model, applying a level of turbulence at the inlet will speed up the process. The turbulence model you select will also play a role in the shedding of vortices - for example, L.E.S. is much better than just running a transient SST simulation.