CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Dependency of turbulence model on mesh size (http://www.cfd-online.com/Forums/cfx/24556-dependency-turbulence-model-mesh-size.html)

Kushagra Mittal September 20, 2007 00:14

Dependency of turbulence model on mesh size
 
Hi,

The flow I am trying to simulate is highly swirl in nature. I started with a very coarse mesh without resolving the near wall mesh. I used 4 mm tetra element with T-Grid scheme while the diameter of circular section is 76 mm.

What I find, velocity profiles and other results are exactly same for laminar and turbulent models; respectively. The reynolds number would be around 3*10^5 to 5*10^5.

I can be assumed a beginner in CFD world. Could you please through some light on importance of mesh size and near wall resolution on turbulence models. I have studied about y+ and wall function but I am wondering if turbulent models might not affect simulations AT ALL even with a coarse mesh and high y+. Should not I see some difference in my solutions.

Thanks in advance, KM.

Glenn Horrocks September 20, 2007 18:51

Re: Dependency of turbulence model on mesh size
 
Hi,

Wall functions will not be the problem with your simulation.

If your flow has high swirl then forget about all the two equation turbulence models. This means k-epsilon, k-omega are no good as none of them can handle swirling flows. The only exception is the SST model has a curvature correction option as a beta feature. This may mean the SST model is OK, but I have never used this option so no guarantees.

For swirling flows you really need to use a Reynolds Stress based model or DES/LES. Of course you need to use wall functions with these models but at least they can handle the basic physics of the flow.

Glenn Horrocks


All times are GMT -4. The time now is 11:40.