[bibelka]

Hello!

My question may sound stupid, however it is very important to me.
How to model boundaries in z-direction in infinite cylinder, assuming the flow is turbulent?

Sorry, if something is unclear from my description.

[cfdnewbie]

there are no stupid questions

Just use periodic bcs in the z-direction, that should do the trick. what kind of turbulence modelling are you doing? LES?

[bibelka]

Thanks!
RANS, actually, however hope that periodic will still work =)

[cfdnewbie]

Yes, in a way, RANS is the only method in which periodic BCs work without any limitations!

Good luck!

[bibelka]

Thank you! =)) And what are the disadvanteges or limitations of using RANS? LES is a good alternative or not?

[cfdnewbie]

Originally Posted by bibelka

Thank you! =)) And what are the disadvanteges or limitations of using RANS? LES is a good alternative or not?

that's a very broad question. Just the basics: RANS assumes that your mean flow is steady, and models the lumped effect (dissipation) of turbulence. Also, RANS works not to well for separated flows and complex geometries.

LES resolves the larger vortices of the flow, and models only parts of it, so it is inherently unsteady, but many times more expensive than RANS....


that's just a quick answer, there are many more aspects to this. What in particular interests you?

[swetkyz]

Meaningful usage of LES is very difficult. Meshes must be much finer, and care must be taken in setting up the simulation.

There is a great deal of good information in the following AIAA paper:
"Large-Eddy Simulation: Current Capabilities, Recommended Practices, and Future Research." Georgiadis, N. J. et. al.

You may want to do the RANS simulation, and if you find that you want to resolve the turbulent structures more explicitly, try LES as a deeper exploration into your problem.

Good luck!

-Zachariah