curved duct problem
 

I am modelling the laminar flow in the curved duct, My problem is when I use the low Dean number (or low Reynolds number), I have four counter rotating secondary vortices or logically the four counter rotating secondary vortices exist only at higher Dean number (or low Reynolds number). I don't know if the problem reside in the method of resolution or in the criterion convergence (if that, I don't know to change that).

Please give me some help, Thank you.