Turbulence model for airship
 

hey guys,

I'm modeling flow over 700 ft long airship thats modeled by a flatten ellipsoid with lenth of 700 ft, height of 100 ft, width of 200 ft.

I have an excellent mesh with good inflation layer and nice surface mesh that represents the curvature of the airship well. I'm modeling a quarter of the airship in cfx.

I need to know what type of turbulence model i have to use. I've experimented with few of them and the result seems to vary for each one. The size of airship might have an impact on which turbulence model will best fit the simulation.

My current model is SST model with automatic wall function and initializing with intensity value of 0.01 and automatic eddy length scale.

Let me know if this is right assumption or what values i should change.

You have several choices:
1) Read a turbulence textbook to assist you choose an appropriate model (you should do this anyway so you know what the various models are about!). A good one is "Turbulence Modelling for CFD" by Wilcox.
2) Try them all and see what lines up with experimental results best. Choose the one which has the worst match (just kidding).
3) Use some rules of thumb - your model is likely to be very high Re number so you will need to use a wall function approach. SST and k-e are the prime contenders here. I would think it unlikely a RSM approach is required as turbulence anisotropy is unlikely to be a problem. If the body has large separations then it will probably need a LES/SAS/DES approach.

thank you, I'm looking into it.

Is there a way to obtain a profile of the boundary layer. I used location line function to view the profile that i wanted above the airship, but when i input that line in chart the lowest point is the first node of the inflation layer. However, in my solver it is using wall function and i cant view that profile in the chart. The profile simply jumps to certain velocity and then places the turbulent layer, it doesnt start from zero which was specified as the no slip condition on wall.

Yes, use a turbulence model which does not require wall functions. SST is the usual choice here, with a y+ of around 1.

? But this is exactly how wall functions work. You don't model all the sublayer stuff, just jump straight to the log layer.