Fair enough. Not a bad idea, but one which I belive, when you do the necessary calculations, will prove to be more expensive, and less effective than simply running oversized conductors.
You are quite correct, power companies do this all the time; it is the standard method of moving power around. But this is not without a price, the power companies are always trading off one factor for another to find the most effective (usually cheapest) method of delivering the necessary power.
I would really appreciate some teaching on all of the issues that would go into this calculation. I can make guesses, but it is over my head in terms of experience.
I know transformers have impedance, and that this will result in the secondary voltage dropping as load increases, so the transformers themselves introduce 'voltage drop'. But transformers have taps, so that you can set the no load voltage to be a touch higher, giving the target voltage under load. So with a transformer set you can to some extent 'zero out' your voltage drop. You will still see a dynamic change in voltage with load, but you've now placed that range of voltages _around_ your target voltage, rather than simply below it. I also don't know what sort of impedance to expect for single phase transformers of this size.
Transformer impedance is different from resistive voltage drop because inductive impedance doesn't actually consume real power...so if you are consuming 30kVA with a 5% inductive 'voltage drop', your power costs will be less than if you were consuming 30kVA with a 5% resistive voltage drop. But on the other side of this energy savings a totally unloaded transformer will still consume some power. This means that even when you client turns off _everything_, they will still be consuming (and paying for) electricity. Figuring out the overall energy use over time will not be a trivial calculation.
Curious to see what the 'optimal' design here is.
-Jon
