Here's the basic strategy:
Determine the element currents to produce the desired radiation pattern. A simple strategy will probably work fairly well, and there's no point in optimizing the phase to the nearest 0.1 degree, since you can't build the antenna that closely. There is an intuitive argument that for basically identical radiators (often the case in phased arrays), you want the magnitudes of the element currents equal. NEC does this quite nicely, especially if you use one of the modified versions (I like 4nec2 from Arie) that has current sources (modeled as two port networks that transform voltage to current).
Determine an estimate of the mutual impedances. A good idea is to vary antenna dimensions by small amounts to determine how much variation there is in Mutual Z. Wild variations means difficulty in tuning, and maybe a better physical design is needed. Again, NEC is your friend here, since you can just excite each element in turn, and measure the current in each feed point. You can either work with admittances or impedances.
Calculate feed point voltages to get the desired element currents. This is as simple as multiplying the desired currents by the Z matrix.
Calculate apparent feed point impedances. Divide the feedpoint voltage by the feedpoint current.
Synthesize a network to partially cancel the feed point's reactive component over the band. This is done to reduce the amount of reactive power flowing in the feed lines, which reduces a lot of the loss. Particularly for arrays where the elements aren't resonant at the frequency of interest (a 20m array, working at 40m, for instance), cancelling the reactive power at the feed point goes a long way towards improving efficiency. I'm not so sure about whether it's worth trying to get the magnitude of the impedance to match the transmission line.
Synthesize networks to distribute power among the feed lines with the required phase and magnitude. You'll need to take the feed line length into account, and some mismatch.
Model the entire system and determine actual pattern
Systematically vary parameters to determine sensitivity of pattern to variations in parameters
radio/antenna/phased/design.htm - 1 Dec 2002 - Jim Lux
phased array home antenna home radio home
Jim Lux home
email comments email@example.com