[ziyasaydam]

Hello everyone,

I wonder if anyone has conducted sailing yacht drag analysis with multiphase option in STAR-CCM. I am comparing experimental results of different size models with my analysis in STAR-CCM.

My major problem in all these analysis is being unable to find the right wall treatment. After finding the right size of cells for the free surface, wake regions and surrounding regions of the hull, keel and rudder -all being the same size-, I keep halving the size of cells on the surfaces of the hull, keel and rudder (to decrease the Y+ values) until the solution becomes grid independent. Note that during this approach, I adjust the time step in order to obtain a Courant Number of around 1-10.

With this approach, different size and different geometry models become grid independent at very different Y+ values. On a 6 meter model case the solution became grid independent at Y+ values of ~500, whereas on a 2.4 meter model as I decrease the Y+ from 400 towards 50, I keep deviating from the experimental results...

Does anyone have an idea about possible reasons for inconsistencies in the results? My expectation was to see that the results get reasonably closer to experimental results as I decrase Y+ towards ~50...

Thanks in advance,

Ziya

[Vinicius]

Hi Ziya,

For any drag calculation your Y+ value must be 1 or below...

Y+ between 30 and 300 are just an aproximation of the behavior of the boundary layer... if want to capture the drag, you must solve the boundary layer using the Y+ around 1...

Also, for VOF models, be sure to refine the mesh in the interface between the fluids..

Hope it helps..

Regards

[ziyasaydam]

Hi Vinicius,

Although it is an approximation, I tend to believe that the wall function approach should give reasonably accurate results with fewer number of cells compared to enhanced wall treatment. Considering the necessity of analysis in full scale as well (assume 30-40 meter long yachts), having a Y+ value of 1 would lead to number of cells impossible to handle within the available resources. Also, I am using Reynolds Stresses Model for turbulence and in some cases only high Y+ treatment is possible.

Therefore, the question becomes: Is there a best practice rule with wall functions in yacht drag analysis with VOF case?

Ziya

[Vinicius]

I don't think you have to use the Reynolds stress model. A RANS k-omega should be enough... Are you using trimmed cells or polyhedral?

[ziyasaydam]

I am using trimmed cells in the model. Most of the work I have seen in this field makes use of RSM due to its suitability for streamline curvature of the flow.

[tobe]

I have done quite a lot of hull modeling, at Fn=1 region, and find that the y+ has little impact, but cell size of the free surface is critical. We also don't use the Reynolds Stresses Model

[ziyasaydam]

If you are solving for drag, which turbulence model are you using?