Buoyancy Source Terms in k-e Model
 

I've added the generation of turbulence from buoyancy in my k-e model and the solution becomes unstable for certain classes of problems. My question is: should there be turbulence generation in control volumes near boundaries?