from physical point of view the roughness height and displacement option for wall BC would be in what dimension?

Both are in metres.

Thanks. And if the geometry scale for the mdl is in cm, a 0.01 value for the height would be 0.01cm? Or still 0.01m?

No, the scale factor when writing the .geom file applies only to vertex coordinates. It doesn't apply to any of the physical models or boundary conditions. Roughness height etc will be still be in metres. Similarly, velocities will always be in m/s, never (cm)/s.

Interesting. The result doesnt seem to show it would was in meters though. Flow near the wall (y < 0.01m) is moving. If roughness height was 0.01m and indeed not 0.01cm, flow would have been stuck in this y<0.01m region.

Would this have s'thing to do with my 3.15 old and 3.15 new. In the new 3.15, there is a unit dimension for roughness but don't think there was in the old 3.15.

No, not yet using 3.22 still...

You shouldn't be in a position to judge what the velocity is at the roughness height or displacement thickness: the roughness is a sub-grid model. If your mesh is fine enough to have near-wall cells the size of the roughness parameters, you shouldn't be using the roughness model. In fact, I think star might ignore the roughness model when this is the case.

Thanks for your inputs. I am obviously confused. Yet, your last statement is confirmed not true. I saw my mis-dimensioned roughness model somehow influenced the result, i.e. the roughness model was not ignored by Star (kw-SST). The details of its influence from the vector plots got me into thinking about the dimensions of roughness & displacement at the beginning.

Well, can't burden u too much with my little problem. Thanks for your comments so far. I'd investigate this further. Cheers!

It's worth mentioning also the roughness model is a means of modifying the wall shear stress. It doesn't directly fix the velocity components. If you were able to look at the velocity at the roughness height (and, as I said, you shouldn't) you wouldn't see u=0.

Thanks again. If u feel there is anything more I should know about implementation of roughness in Star, please give me pointers. Online references for example.