Proper orthogonal decomposition of turbulent flow with k-e
 

Hi. Is anyone familar with performing a proper orthogonal decomposition/Galerkin projection of the k-e turbulence model? I'm doing a practice run of the method using cavity convection (buoyancy-driven flow) at the moment. The laminar version works well, but not at all with the turbulent counterpart. What are the caveats involved in projecting the k-e model? For example, I'm not sure how the wall function comes in play into this. Do I even need to consider it in the projection since the wall function only affects a couple of nodes at the boundary? I would be interested in getting anyone's insight on the difficulties associated with reducing the k-e model. Thanks!