The mesh is a 2D structured one. In fact it is a 2D problem but we have added a third dimension (with only one cell) because Star CD works only with 3D meshes.
It is a rectangular domain of 10 x 1 meters (length x width) with 1000 cells.
On the following picture, you can see the mesh in 2D and the 3D visualisation with only one cell along the z axis.
To improve the visualisation and the calculation of important values, we had to refine the mesh in y direction. In the refined meshes, there are 20 or 40 cells following the test case. The construction of a third dimension creates two new boundaries : plans (x,y) which are directly imposed as symmetry conditions by the code.
By default, Star CD considers boundaries which have not been defined as walls, so we just needed to define the inlet and the outlet boundaries.
The velocity is imposed at inlet (U, V, W) = (0.0001, 0, 0) and the flux is fixed at outlet.

Different boundary conditions : inlet in red with velocity profile, outlet in green and symmetry in blue.

We tried first to run the code with a Newtonian fluid to verify if we had the Poiseuille flow in a laminar case.
The Reynolds number is :

where Umoy is the initial value of inlet profile and  Uaxe=1.5 Umoy in theory :

To verify that the simulation of a Poiseuille flow is good, two properties must be verified : the parabolic velocity profile and the pressure loss that must be linear and equal to a theoritical value calculated as following :
The results were good and we have this type of simulations :

We verify that we have a parabolic profile in y and that the pressure loss along the x-direction is equal to the theoritical one.
Considering the number of cells in the y direction, it is quite normal not to observe a "perfect" parabolic profile.
Indeed the axis velocity is almost equal to its theoritical value : 1.48E-3 instead of 1.5E-3.