doesn't matter, as long as you keep the same (cumulative) pole area and the same (cumulative) number of turns.
Note that the number of poles isn't needed.
Well, you're right, but.........
If you take a 12 pole 9 coil flat stator three phase unit and double the number of poles but leave the number of coils the same, it takes half the number of turns per coil to get the same voltage. However, that configuration won't work for three phase, nor will it work for IRP because of canceling problems due to the poles of the same polarity passing both coil legs at once.
So let's say this 12 pole 9 coil unit has 50 turns of wire per coil. If you cut everything by one third - 4 poles and 3 coils - but triple the pole area, then it requires 150 turns. But if you drop back to three coils and 4 poles leaving the pole area the same it requires 454 turns of wire to get the same voltage you had with the 12 pole 9 coil.
Getting back to what I said when I first posted on this thread, if you take this same 12 pole 9 coil and increase the number of poles to 16 and the number of coils to 12, it now takes 28 turns to get the same voltage. You increased the number of poles and coils by 1/3, using the same pole area, and it takes only 56% of the number of turns of wire in the coils. You double everything to 24 poles and 18 coils and now it takes only 13 turns of wire to get the same voltage as the 12 pole 9 coil unit with 50 turns.
Refer to what ghurd said about frequency. Low rpm direct-drive turbine generators run at too low of frequency the way it is. Increasing the number of poles and coils in the right relationship increases efficiency of the unit pretty dramatically. Since frequency is also related to rpm, 12 pole direct drive generators are best suited to high-rpm turbines under 10 feet in diameter, although they can be used on turbines up to 13 feet in diameter if you use more pole area with wedge mags or similar. They won't work very well, if at all, on a VAWT unless you use some sort of step-up drive to increase the rpm of the generator enough to get decent efficiency out of it.
--
Chris