[Callumcmcl]

I'm working on tutorial 4 in the 2018 FLUENT tutorial guide which constitutes crossflow over a staggered tube bank. I am studying the affects of altering the mass flow rate and don't know if a turbulence model is needed. The intro states that the Reynold's number is 100 so I have selected laminar under the viscous models, is this correct? Also, I believe I have managed to calculate new Reynold's numbers as I adjust the mass flow rate of the crossflow, again these lie within the laminar flow range so is the selection of laminar viscous model adequate for the investigation or should I use a turbulence model as there is eddy formation during the higher mass flow rates even though they are laminar.

Cheers

[LoGaL]

It's fine if you get some eddies.

Also, laminar is not a viscous model, you are simply solving Navier Stokes Equations in a "naked " way without additional equations accounting for turbulence modelling. But the dynamics of the turbulence is already contained in the naked Navier Stokes Equations, the additional equations (usually RANS models) are added for the sake of reducing computational cost, as counterintuitive as it may sound. So if you get some eddies and a very minimal level of turbulence, your "laminar" equations will be able to solve it on the mesh you provided.

Actually, with proper numerical schemes and a super computer they would be able to solve any turbulent flow (that is what DNS is)

[Callumcmcl]

Thanks for the reply, I am fairly new to CFD and still trying to learn as much as possible. I didn’t know the selection of laminar and the unmodified NSE would be sufficient to solve for the turbulence in the model. So as long as the Reynold’s number remains low enough the laminar solver would be able to generate an accurate solution?

[LoGaL]

yes, some very very light turbulence or instability can be solved on a coarse mesh. For higher Reynolds Numbers, for which you are sure it's turbulent, use turbulence models