Dear all,

Since we actually have few type of turbulence model to work with, which one actually give a good approximation in term of velocity contour for enclosed turbulent flow with high swirl?

Thanks....

Michelle

The model should consider the effects of streamline curvature since they influence on the dissipation rate of turbulence kinetic energy. Some people has introduced new production terms, in the epsilon equation, that depends on the mean radial and axial velocity components. See Elkotb et al in Flows in I.C.E. ASME 1982 and other modifications which are not 100% OK since swirls are typically anisotropic; The turbulent kinetic dissipation in swirls behaves different as in other cases. I'm using Willems(1996) model, i.e. a hybrid model but I can't tell you how good the results are since I'm testing now.

There is a web-site where most of the papers related with theoretical/cfd and experimetal work on ICE are downloadable. In fact there are 4 pages and the 1st starts here: http://www.herc.musashi-tech.ac.jp/s...d/Html/001.htm

By the way, how high the swirl is? Are you using only RANS, no (V)LES? Structured or unstructural grid? Using init. cond. as u = A*tanh(...) + fluctuations?

/Raul

Dear Michelle,

we got very good results by using LES (Smagorinsky model) in anisotropic cyclone flow. For a glance at uor results, please visit my homepage: www.geocities.com/chicao99br

Francisco.

See my earlier answer to you: a SSG-RSM would improve things serioulsy if you cannot do LES. Otherwise, LES will do a nice job indeed. I have seen this hydrocyclone job done in a CFX-5 demo some time ago, and it was pretty nice for a commercial code standard release.