I am trying to model flow over a sphere, where the boundary layer on one side remains laminar and separates early, while on the other side of the sphere the boundary layer is tripped into turbulent, and separates later, creating a pressure differential on the ball.

However, I am having trouble modelling this, for free stream velocities ranging from 20 to 40 m/s, and I have been told that CFD cannot model boundary layer separation very well. Is this true, and why? Any advice? Thanks

Hi,

there are some difficulties modelling the wall near flow in CFD. The standart wall functions, that are often used for turbulence modelling are valid ONLY for equilibration conditions.

So it is difficult, to determine the cell size close to the wall - the yplus values. One guidline is, to use non-equilibration wall functions with VERY fine mesh (yplus about 1).

Also, k-w is a better tubulence modell to model seperation and reattachement. Have a look in the fluent manual "wall-near flow" (or something similar...)

That are the hints I can give.

Ralf