I solve standard incompressible Navie-Stokes Equations on enough rough grid and where Re=10^5. Has the solution any physical sense? Or I necessarily should use any model of turbulence? I.e. may be laminar flow at Re=10^5? Thanks.

It depends on what you are tryng to simulate. Under certain condition you could have a laminar flows at such Reynolds number... But If you are in a standard case ( for example classical channel flows with implicit roughness ) you Have to use a model for turbulence ( k epsilon , Large Eddy Simulation ) or a DNS ( Direct Numerical Simulation ) which will resolve the fine structures ( and not modelise them ).

Sorry, forget the DNS at such a Reynolds Number...

Hi Alf,

What does your geometry look like?

If you have boundary-layer flow, using Rex, then perhaps the flow is 'just' laminar. If you have any boundary constraints then your Reynolds will drop. In that case you would be surley be missing something at Re~1e5.

Bejan tabulates & graphs this nicely.

What discretisation scheme are you using? What convection-stabilisation scheme do you use? Are you using a steady-solver, or transient?

diaw...

Some approaches like MILES, make use of special numerical schemes and do not use turbulence models