Here is a copy of another answer regarding alternator design for windmills on another site. I'm too slack to re-write it, so I present it in full for your perusal. It may answer some more questions you may have
.... start waffle
I will approach this a bit differently.
We can measure the voltage from the power point, and the voltage meter will say 240vac... thats fairly obvious for a 240v system this country runs. (Aust)
If we look at the oscilloscope, we can see that it is not a 240v pulse at all, but instead rises from zero up to 320v and back to zero, but the meter said 240vac??
If we look at a square wave 240vac with an ac meter, it will say 240vac. The oscilloscope will see that we start at zero, rise instantly to 240v and then drop to zero instantly again at the end of the time span (note the time for each wave covers 1/100th second for both/all of this discussion.)
The meter is correct in both cases, even though one had a peak voltage of 320v, and the other was 240 peak.
It is the area under the curve that was the same in both instances, even though the peaks were different.
Doubling the diameter and so doubling the speed with the same magnets and coils that the smaller diameter one had, carries the same lesson.
The skinny high peaks of the big diameter and the smaller fatter peaks of the smaller diameter both have the same area under their respective wave form curves..
So we can say there is no difference.
In practice there is a difference, and you may recall somewhere recently, Flux said there was a diameter that was ideal, and others that weren't.
space is one reason, overlapping flux is another , and both these should be better in the larger diameter... and they are..... but......
Think of the skinny high peak waveform, where the peak was higher the ohms law comes to kick you in the butt.... If your coil R is the same in both instances, but your peak EMF is higher, then Voltage loss in the same coil is higher.... ie I=E/R, so if E is higher, then I is higher for the same rpms with the larger stator.... worse the losses in that coil will be I^2xR...so that makes it worse still.
So now we can say that the losses in the bigger diameter machine will necessarily be higher than the lower voltage spread over a longer angular interval... because the peak currents are lower through the same resistance.... we lose.... but fun while it lasted.
If we fill in the gaps symmetrically to make twice the size machine, then thats a new machine entirely.
"*for the record
this is a theoretical discussion of what is possible, not something that i am claiming as new, better than, worse than, or even desirable or useful. as Flux has stated there really is nothing new in the generator game other than materials used. every possible configuration has been tried, proven and used by the late 1880's including the air core axial machines.
kicking the ball back to center court"
Matching the load is the goal, and we need to match the impedance of the air to the impedance of the battery. This involves a mill with blades, an alternator and a basically fixed impedance battery.
The basic mill tries to match the variable impedance of the air, to the variable impedance of the alternator to the fixed battery.
We can fiddle with blades, blade size, wire, magnets, mill diameter, resistance in the line, black boxes, star delta switching, transformer tapping, high leakage transformer coupling, direct coupling and the list goes on.
The fun thing about mills, is that it HAS to be built with an eye to everything. The moment we try to optomise any part, we alter some other parts.
Folks like Chris optimised the resistance in his alternator. He did this with higher speed, which allowed his resistance to be very very low.... but this makes it virtually unusable for direct coupling, so black box is needed to match back up all the mismatch imposed by the low resistance.... if he had zero resistance, and no black box, it would never go beyond cut in.
We can use higher resistance stators, and get good low wind performance, but the high winds burn them up... unless we can use a black box, or cool it, or star delta it perhaps... of furl early.
There are other tricks that can be used which you would have seen in your bigger alternators. An automatic star/delta machine can be made with the addition of 2 more diodes, so that when the losses in the star configuration rise to around 40%, it will use the extra two diodes (from the star point to the + and - outputs, and this will convert the star to delta with no switching..... but then we hit an impedance mismatch because our stator resistance drops to a third of what it was.
You will have seen this when you open a 100A plus alternator... and find 8 diodes in the diode block, not 6.
So next we open the gap to stop stalling, but this degrades the alternator at the same time, or we could add resistance to the line, we miss out on some power, but take some heat out of the stator, and allow the blades to run better.
We can get out of some of this by upping the voltage and poles, and using a transformer.... this can be quite cunning, as we can tap change with triacs, and we can wind the transformer to emulate a battery charging transformer by making loose coupling with plenty of leakage, this will help match the prop to the air, and shift the heat out of the stator to some extent, and if the mill has an iron core (motor conversion, radial iron core (AWP, F&P etc)), we can mitigate some of the effects of armature reaction, and match the loads much better.
In fact we could emulate a full blown mppt with a jump table,lower leakage transformer and a bunch of taps and triacs if we were keen, and liked rats nests.
What I'm getting at is we can get hung up on alternator design. If we build an mppt (analogue or digital whatever), we can use virtually any old thing and match the load to the air.... then it's simply a matter of how powerful we want..... general rule ... for a fixed rpm, bigger is betterer for more power. More rpm (speed increaser makes them smaller again)..... back to angular velocity I guess for fixed flux.
Things that get lost from time to time.....
Every turn of wire in the coil has it's own vectors with relation to the field changing in it's vicinity. All those vectors add up to the final emf in that coil at any time in the cycle.... in fact at top dead center, the total emf vectors add to zero.
With axials, the legs are not points, but spread across an area... all seeing the flux differently. So magnet size, leg width, hole diameter, coil height, area, magnet spacing all effect the outcome.....and the list goes on.... the only thing we need not worry with is the armature reactance. The synchronous impedance will be pretty close to resistance for our purposes.... and we know we need some resistance so the voltage can rise in the coils so the prop wont be in hard stall from the get go... or mppt.
We should know or be told....... that we cannot design an alternator on it's own. It is part of a symphony. All things must be taken into account if we don't use mppt, but we can build a darn good machine without it to.... but then it has to be part of a team.... not a solo player. Mppt takes the skill out of windmill design.
Personally, as an owner/builder of a few quite powerful axials, and a not so innocent bystander to a radial flux machine, now (hindsight is cool isn't it) I would want an iron cored radial any day of the week, even with (and I WANT) oodles of armature reaction (Flux rolls eyes).... and ferrite magnets (eyes roll again).
Still the best, toughest by far, bullet proof home powering mill I have been in contact with is the AWP HV transformer coupled unit with the rewind we did for it.
That thing is fairly well matched, and burn out proof (thanks to the reaction component)... and the magnets don't degrade either
I just can't get excited with efficiency when I'm not paying the bill for the wind.
Stator efficiency for finding the watts lost for the sake of heating I have to be interested in.
System efficiency pales into insignificance compared to matching the items to the wind.
Paying for fuel makes it a whole different ball game.