I found the answer in Darren’s post.
“between points defined as X, Y, Z the velocity will be linearly interpolated, so it is important that good resolution is given close to the 0 velocity edge, furthermore, several ‘poles’ of points are recommended. i.e. if the inlet boundary was in the YZ plane normal to X, and Z low was the floor, a CSV where Z varies for example [0.01, 0.02, 0.04, 0.08, 0.16, 0.32, 0.64, 1.28, 2.56…] and Y varies [-200, -100, 0, 100, 200] giving poles of points at -200m etc in the increasing distance from the ground if the ground was located at Z=0.”
So, taking a look to the csv example posted by Darren next to the quoted part …all the vectors that i define should have starting coordinates related to plane position in global reference, said so, if my inlet XZ plane finds at Y=-800, all my vectors should have Y=-800 on table.
And, to have a more accurate simulation all along the width of the wind tunnel, it may not be enough to create only one “vertical line” of starting points on the plane…in this case, i should create more vectors vertical “groups” changing X coordinate while keeping, of course, Y fixed.
In the end, all the vectors that share the same altitude, will have same Uy value.
Made a test and in fact wind profile look finally realistic, with velocity inlet values increasing with altitude.
Used three interpolation sets of points, at X=300 X=0 X=-300, with Y constant=-800, which is the coordinate of my inlet face in global axis reference.
Hope it will help other construction designers 
