[sagarmore]

I am carrying out horizontal axis wind turbine CFD simulation in Fluent to check effect of surface roughness of blades. It’s a steady state simulation using moving reference frame with k-omega SST model and air as working fluid.
Wind velocity=8 m/s
Angular velocity= 52.359 rad/s (500 rpm)
Relative velocity=52.967 m/s
Reynolds number=159790.244
Chord length=0.045 m
Desired y+= 1
Using these conditions first cell height comes to be 5.48048E-06 m (for external aerodynamics based on turbulent flat plate). I want to check torque output for roughness height up to 30 micron. My questions are;
1. Literature suggests that centroid of first cell (half of first cell distance) should lie above roughness height. So how to model roughness height larger than half of initial cell height? Is this requires shifting of law of the wall? If yes please describe.
2. How to specify equivalent sand-grain roughness height?

[LuckyTran]

Originally Posted by sagarmore

1. Literature suggests that centroid of first cell (half of first cell distance) should lie above roughness height. So how to model roughness height larger than half of initial cell height? Is this requires shifting of law of the wall? If yes please describe.
2. How to specify equivalent sand-grain roughness height?

1. The roughness model is based on the rough law of the wall (the rough law of the wall already is a shifted law of the wall model). Implicitly, if your cells are smaller than the roughness height then you are capable of resolving, and should be resolving the roughness features directly rather than relying on some model. Other than this philosophical reason, you can utilize a rough law of the wall model for arbitrary roughness heights (even if it doesn't make sense).

2. The roughness height parameter that you need to input in Fluent is the equivalent sand-grain roughness (it is not the RMS roughness in materials science). Or is your question the other way around (knowing rms, how to compute equivalent sand-grain roughness)?

[sagarmore]

Thanks for your prompt reply.
Does this mean fluent automatically ignores cells whose centroid lies below roughness height and considers cells whose centroid lies above roughness height (thus "virtually" shifting wall)?
If so, then will fluent start measuring y+ from a virtually shifted position of the walll?
I want to use low Reynolds number (k omega SST) model.My worry is about maintaining y+ value as one.

[LuckyTran]

The y+ of the wall adjacent cell is still computed from the wall adjacent cell, cells are not ignored. The virtual wall shift is a simple modification to the calculated y+ of the first cell. Hence "virtual" rather than actual/explicit. I don't mean to argue about semantics but since you are asking how Fluent is going to perform the task, it helps if you use the definition of "virtually shifting the wall" per Fluent's definition and not your own definition. Actually I agree that your definition of virtual wall shift makes more sense.

I believe the virtual wall shift is done for the k-omega SST model but I recommend reading Fluent User Guide 6.3.14.2.9 Law-of-the-Wall Modified for Roughness since different approaches are used for different models.