Hey Nina,
So I've downloaded and looked at your file. It seems like you're more or less on the right track. From what I understand, the radiating patch should be approximately half a wavelength long and the substrate is typically a quarter wavelength thick. Because you have both free space and the material wavelength, however, you must find an "effective wavelength" for the two. From what I've seen, there are analytic solutions to this for long microstrips, but I haven't found any "effective wavelength" or "effective permeability" formulas for a radiating patch. I, personally, would just use the wavelength in the material. You're using Duroid 5880, which has a permittivity of 2.2EpsilonNot - giving an index of refraction of about 1.48.
Say, for example, that you wanted to radiate at 10 GHz. The freespace wavelength at that frequency is 3 cm. The wavelength in the substrate, however, is approximately 2.02 cm. So the patch should be approximately 1 cm x 1 cm. The thickness of your substrate should be approximately 0.5 cm. Also, you should have an infinite ground plane behind the substrate to reflect radiation that goes "backwards". It should be directly behind the substrate. Your solution should be of type Driven Terminal. The terminal should contact two conductors. One conductor should be the reference and should touch the infinite ground plane. The other should contact the patch through the microstrip you've drawn. This is the actual excitation.
I hope that helps a bit.