The planar MoM simulators assume a substrate material that extends to infinity. This is a problem for antennas that fire to the side (horizon).
In your case the substrate is "air", so it should work. However, some other NEC based codes where antenna segments are modelled as wires might be more efficient.
Of course, it is also possible if you want a normal yagi "in air". Just define air (erel=1) as you substrate and use absorbing boundary condition above and below.
Designing the feed for yaggi uda on PCB is very difficult task. Methods of Moments (ADS Momentum) is used for 2D structures for 3D it is better to use HFSS.
dear volker, I see what you mean but wouldn't ads assume an infinitely thin copper? which might cause problems for pattern calculation in the end-fire configuration?