[Heisenberg White]

Hello everyone,

When using Fluent to simulate the flow around a cylinder with roughness, I found that when the roughness is large(about 0.005 m), there is an underestimation of the drag coefficient Cd. The SST k-omega turbulence model was used. I tried adjusting the height of the first layer of the mesh but did not achieve satisfactory results.

Following my advisor's suggestion, I conducted a simulation of wall-bounded flow to verify whether the velocity profile was reasonable. In my wall-bounded flow simulation, ensuring that the flow field was fully developed both temporally and spatially, I found that when y+ was approximately 50, the velocity profile was not fully logarithmic, and the velocity near the wall was overestimated.
Some sources suggest that the height of the first layer of the mesh (y1) should be greater than the roughness height (ks), so I adjusted my y1. As a result, although the velocity showed a fully logarithmic trend, the dimensionless velocity u_tau was still higher than expected.

My question is:
Why does the wall-bounded flow velocity not match the fully logarithmic velocity profile predicted by the Fluent Theory Guide 4.18.5.(https://ansyshelp.ansys.com/public/a...008.164.html)? Does Fluent employ some other corrections?
Or have you met the same question that the drag coefficient is lower than expected when you set ks > 0?

If you need more details please feel free to ask.
Thanks a lot!