[SH_P]

Hi,
Is it possible to define roughness for a wall in laminar flow. To be more specific, consider we have a cylinder in a flow with low Reynolds number and the surface of cylinder have a kind of roughness such as some wires on it. I am not sure whether this roughness can trigger the flow to be turbulent or not. Now my questions is :

1. Is it advisable to consider these wires with wall roughness option, instead of creating all of them in geometry and mesh (actually considering them in the mesh is not easy!)?

2. Can a turbulent model, like ones which are used by Fluent, be used in situations like I described : when we are not sure about the flow regime?

Thanks...

[sbaffini]

Dear Soroush,

there are several way this answer (actually the motivations) can be provided.

I start by saying that nowadays Fluent provides also some transitional RANS models which maybe (but i don't know) could do the job. You should check the manual.

Not considering such models, what you usually consider as roughness model is a parameterization of the roughness effect on the classical wall functions, that are fully turbulent. This is not random because it is in this case that such parameterization is known and it is consistent with the fact that it is part of a wall model of a turbulent approach.

You should think about why no wall functions exists for laminar flows to understand why no roughness effect is parameterized in the laminar case. Roughly speaking, the fact is that in a simulation without turbulence model, by definition, you are simulating all the relevant scales of the flow (be aware that laminar does not mean steady or 2D).

Of course you can create some UDF in fluent and use it to specify the shear on the boundary (which is all you can do for a wall b.c. in finite volume) and put some roughness model in it. You will then face the strange fact that on the simulation side (which is laminar) you assume that everything is resolved but on the b.c. side you are saying that not everything is resolved. I think there is some inconsistency.

Consider your question in this way: you want to know if some device is likely to promote transition but you don't want to put it into the simulation... sounds weird.

Besides DNS, LES (still 3D and unsteady) could do the job but you have to provide some minimum geometrical information on the wires (or whatever).