Simulations done with FLUENT 5 ( Water case)
The grid

The grid used was done with Gambit paying attention to the shape. It was tried to keep the same mesh as the one used for StarCD simulations to enable some comparisons between the different CFD codes. The grid contains approximately 60 cells for the height and 25 cells for the lengh. The geometry of the cells looks like squares more than the StarCD grid.
The interns are defined as walls, the right and left boundaries as symmetries. The injection of water is done above with the velocity of 1m/s and the outflow happens below on the left.
The dimensions are: Height 7cm, Lengh 2cm, Interns thickness 1,4 mm, In and out lengh 5mm

The velocity field

Water is injected with a velocity of 1 m/s.  Calculations are done with a standart k-epsilon model.The velocity field looks like that:

Recirculations appear behind the interns as it was expected and fluid is strongly accelerated in the narrowing as it was also expected. But the maximum velocity reaches 6.5 times the initial velocity. Let see the recirculations:

Strong velocities are encountered near the boundaries.
The Reynolds number at the injection is: Re=UL/Nu = 1 * 0.02 / 10e-6 = 20 000. So the flow is clearly turbulent.

At the end of the flow, a intensity of 244% is reached. In fact results about turbulence are not really precise with a k-epsilon model, but give a first approximation. Obviously turbulence increases in the narrowing, but strangely also at the left of the outflow.