Unstructured grid problem - Laminar flow pipe

Roby

In order to perform complex pipe geometry with laminar flow simulations, i've tried to mesh a simple straight duct with a t-grid scheme using the size function in gambit to preserve the boundary layer detection. I set an udf to have a laminar fully developed flow profile at the inlet. I set a size function in gambit to obtain a value of about R/38 for the first cell height as suggested by the fluent manual. I set outflow for the outlet bc. The results i obteined are unespected... e.g. the max velocity plotted on an axial line decreased of about 20% respect to the max inlet value! I've checked the mass balance and it seems to be all right. With structured mesh i obtained good results. Have you got any ideas for non structured grids?!?! I thought it could be a not correct prediction of the boundary layer flow or even some solver parameter i missed...

thanks in advance.