Microstrip Slot Antenna Array

My problem is a 24GHz slot array with 4 elements. Each element is driven (no parasitic ones). The measured gain is only about 0dBi. I think the influence of surface waves is high. My driving network consists of 3 wilkinson dividers.

What are possible modifications?
- Using parasitic elements
- Spacing the elements in a way that the phase of surface wave is identical to the driven wave

In my simulation (ensemble) this problem is not visible. What parameter was forgotten? Is ensemble capable of calculating surface waves?
Has anyone experience in designing microstrip slot antenna arrays? Perhaps a good E-book?

Thanks