Hi

does anyone know about an accurate, "proven" N-S solver for impulsively started cylinder?

KSR

For the incompressible flow version check the following (excellent) thesis:

Shankar Subramanian, "A New Mesh-Free Vortex Method," PhD Thesis, The Florida State University, 1996.

The method has been used for up to Re=40,000 with excellent success (DNS - no turbulence models). The standard benchmark at Re=9500 compared very well with a spectral element solution using a much smaller number of computational elements.

Adrin Gharakhani

Dr.Adrin

Thanx very much

KSR

Hi,

Supposing one gets symmetric wake solution, say for R=9500, even when for a long time integration, is it a desirable feature in a code? Selvarajan

Yes, IF one can get such a solution for LONG times! A symmetric wake will be a valid solution and moreover it will be an indication of an (extremely) clean code/methodology. I doubt this will be possible and sooner than later numerical perturbations will creep in, giving rise to the familiar asymmetry.

Incidentally, the above has been verified experimentally as well. That is, one can extend the life of symmetry in "real" life if one has a clean setup.

If one has a clean code/algorithm, then it is a simple matter to introduce perturbations in the field and watch the instabilities grow, but the reverse is not true.

Adrin Gharakhani

Hi,

Thank you very much.

Supposing one introduces disturbances and tries to study the vortex shedding features, why does it take unrealistic time for the first vortex to shed? Or, often you need a large disturbance to disturb the flow. Could you please comment on these?

Regards selvarajan

I don't know what you mean by "unrealistic time". In engineering/science one generall has the luxury of quantifying items; why not take advantage of it!

And, unrealistic compared to what? If compared to experiments, how do you know what the perturbation amplitudeS are in this case?

It makes sense to me that small-amplitude perturbations will take longer to grow than relatively larger ones. But, the question for me is whether your numerical methodology can capture inviscid instabilities accurately or the numerical diffusion is so high that it damps out "small" perturbations.

I'm rusty on stability analysis (database-wise), but I'm sure there are theoretical studies for the cylinder flow case (this should give you a theoretical guideline)

Adrin Gharakhani

Hi Dr. Adrin,

Thank you very much. I will look into theoretical studies on stability of flow past a cylinder.

Selvarajan.