As background, I'm trying to model the pressure rise/swirl of fan blades behind a cooling package using multiple frame of reference (rotation local coordinate system) and thus modeling the whole domain in a stationary manner. The geometry itself is a bit complicated so my first try was pure tetrahedral mesh. This didn't to the trick - the pressure rise was not even close to measured data. A affordable way would be to create a hybrid mesh, with prismatic or hexahedral cell in the wall adjacent cell layers, and tetrahedral cells in the rest of the domain. However, as wall function is not very likely to work well under these rotating conditions, maybe a 2-layer model will (although with increased mesh size).

As you all can see, I'm guessing and maybe somebody has had the same problems and solved them. So, any recommendations how to mesh the fan blades and the vicinity, turbulence models, etc, in order to achieve the best possible results.

Cheers / Anders

I would suggest, if you haven't already done so, going back a step and modelling a single blade passage in the rotating frame of reference. Ignoring, for the moment, any upstream or downstream geometry. In this way you can concentrate your computational resources on resolving the most complex flow region. Fan flows are pretty tricky applications to simulate (adverse pressure gradients, lots of separated regions, tip leakage effects, corner vorticies etc etc) so keep the analysis focused.