Just one note.
If the US can build 310 divisions in WWI at any time something must be wrong with the manpower. The problem could stem from the fact that real US divisions were much larger then contemporary European ones (4-7 regiments per division iirc, vs. 3-5 in Europe, not to speak of non combat troops) (except theoretical Belgian and Bulgarian strength iirc). I don't think you have given the US a separate unit model (unique techs) to reflect their huge divisions.
If the US division sizes are indeed the reason, there are two relatively simple fixes I can think of. Consider each historical US division as 2 TGW divisions (a great abstraction) and reflect that in the at start forces as well as the unit name file. Alternatively reduce US provincial manpower by a percentage to be determined (I believe an average TGW division uses some 18 manpower, a historic US would probably be 24-30 manpower, so reduce provincial manpower by a bit more then 1/4th).
In WWII the US had serious trouble to raise the planned ammount of divisions and ended up scrapping the last half or so of them. Iirc something similar happened in WWI (canibalizing late forming divisions to bring those on the front up to strength). So even the historic divisions should be hard to achieve.
Marc aka Caran...
P.S.: For the rest I agree with peacetime limitations to manpower to avoid huge build ups. Possibly giving a few pre-war increases to simulate growing readiness for war.