[mazhar16823]

Hi,



I have a mesh with 6 prism layers within the boundary layer out of 20 layers as per the definition of boundary layer(where flow velocity becomes 99% of free_strem velocity). The wall Y+ is around 3, considering this do I need to further reduce the near wall thickness or total prism layer thickness.



P.S: I am using Transition SST model, the solution has converged, however there is a little unexpected change in the monitoring variables at some locations i.e. for pressure coefficient, skin coefficient etc. (don't know if this acquires to have more prism mesh refinement). Any suggestions?


Thanks for your time.

[bluebase]

Hi MA,

The Transition SST model in StarCCM needs something like a free-stream wall distance. Determining this quantity without a-priori knowledge/experimental data, is quite difficult to do, in my experience.
Nevertheless, the Transient SST model is a derivate of the SST model, as such you have two options how to resolve the wall: a wall model, or a fine mesh to resolve the boundary layer flow.
If you want to resolve your boundary layer with a wall model, you should not go below y+ of around 30 (as far as i remember, i think there was some info in star's manual). If you want to fully resolve the wall boundary layer than it is generally recommended to have y+ <=1.

With the info of your other posts, i assume you want to simulate a wind turbine. With a wall model you likely get reasonable results as long as you do not have strong detachments. Consider, that wall models where likely defined for specific flow conditions.
For high fidelity/high accuracy simulations, i would not rely on wall models, since fully resolving the boundary layer can (somewhat) improve the correct prediction of the area of flow detachments. Though, do expect differences between experiment and RANS simulations. Even the best RANS models have difficulties with that when you use models for cases for which they weren't calibrated.
If you have strongly circulating flow, you might get good results with the Reynolds-Stress-type models.

For regions out of interest, i would not care to fully resolve them and just rely on wall models, such as the ground level on a hawt,

Best,
Sebastian

[mazhar16823]

Originally Posted by bluebase

Hi MA,

The Transition SST model in StarCCM needs something like a free-stream wall distance. Determining this quantity without a-priori knowledge/experimental data, is quite difficult to do, in my experience.
Nevertheless, the Transient SST model is a derivate of the SST model, as such you have two options how to resolve the wall: a wall model, or a fine mesh to resolve the boundary layer flow.
If you want to resolve your boundary layer with a wall model, you should not go below y+ of around 30 (as far as i remember, i think there was some info in star's manual). If you want to fully resolve the wall boundary layer than it is generally recommended to have y+ <=1.

With the info of your other posts, i assume you want to simulate a wind turbine. With a wall model you likely get reasonable results as long as you do not have strong detachments. Consider, that wall models where likely defined for specific flow conditions.
For high fidelity/high accuracy simulations, i would not rely on wall models, since fully resolving the boundary layer can (somewhat) improve the correct prediction of the area of flow detachments. Though, do expect differences between experiment and RANS simulations. Even the best RANS models have difficulties with that when you use models for cases for which they weren't calibrated.
If you have strongly circulating flow, you might get good results with the Reynolds-Stress-type models.

For regions out of interest, i would not care to fully resolve them and just rely on wall models, such as the ground level on a hawt,

Best,
Sebastian

Thanks for highlighting what Transition SST is about. I prefer "wall-resolved" not "wall modelled" approach since I have already achieved Y+ in a range to resolve viscous sublayer.