leading edge vortices
 

Hi i am trying to model delta wing in cfx5 for few days. but i am not able to capture the leading edge vortices at all. does anyone has experience in modelling delta wing flow in cfx? I am trying with following method: My domian is 2c upstream,6c downstream and 6c in remaining directions. i am using really fine mesh(tried 25,00,000) elements with clustering around the wing. For intial case using laminar models with AOA around 10-15 and 0.3 mach no. inlet condition is cart vel components, outlet is pressure(subsonic). free sleep at other walls and no slip at the wing.

thank you

Re: leading edge vortices
 

Hi,

What turbulence model? If you are using k-epsilon that might suppress these type of features. Try SST, but you may be forced to use Reynolds Stress models or possibly even LES/DES.

Regards, Glenn

Re: leading edge vortices
 

On a related note, I have found the SST model to give good secondary flow predictions around turbine guide vanes, resolving the horseshoe and inflow vortices reasonably well. The SST predictions to be in good agreement with those from an RSM (BSL) model using the same grid (Y plus around unity) I am not too hot on the maths behind the two models, from my understanding though the SST model does not account for anisotropy in the normal stresses, and as such should be less accurate (in theory)? I was a little surprised by the reults, I was expecting the SST model to perform in a similar manner to the k epsilon in secondary flow prediction?

Re: leading edge vortices
 

Hi,

SST has a correction factor where it tries to account for flow rotation, whereas k-eps has no means of accounting for flow rotation and therefore over predicts the turbulence associated with a rotating flow. This excess turbulence leads to an artificially high viscosity and therefore damps out flow features which are physically present. While the SST curvature correction is better than nothing it is still only a correction factor laid over the top of the turbulence model to attempt to fix an inherent flaw in the basic model.

RSM models theoretically fully account for flow rotation but in practise rarely give additional accuracy due to other considerations such as the limitations of specifying the wall boundary conditions and the fundamental limitations of the RANS approach.

That is why many studies of this type of flow end up using a DES/LES approach.

But for your work, use the simplest model which captures the physics.

Regards, Glenn