How long can an industrial generator continue producing power? I asked this same question over in ElectricalEngineering, and the answers were...less than helpful, let's just say.  Plus it got closed for being off-topic.  I'm hoping that both of those are resolved by asking here, instead...
Given a constant input source and infinitely strong materials, is there any limit to the amount of time (from an electro-dynamic perspective) that a generator can continue to produce electrical energy?
What I'm getting at is this: if you want to build an electrical generator and have it last for, say 10,000 years, is there any reason that the conversion process between the electrical and magnetic fields would fail after say, 5,000 years? Or do all of the failures come to down to engineering and materials-sciences issues (e.g. the shaft wears out or the fins snap or the copper windings corrode or something)
I'm curious as to how long an electrical generator can last without needing to have it's "core" components repaired or replaced. e.g. could we launch a nuclear powered satellite into space with a generator on board that could power the satellite for 100,000 years? Or build a hydroelectric plant where the core would never need to be replaced, as long as it was supplied with sufficient oil / replacement exterior components. Basically, minimum maintenance. Something that could be buried and forgotten about and would just keep working
 A: Since this is intended as a pure physics question and ignoring any material limits (and, for sanity's sake, quantum mechanics) the less-than-useful answer is: essentially infinitely, but eventually a thermal fluctuation will make it fail.
A perfect electric generator is a reversible system: some torque (from, say, a water wheel) turns it, and it generates current. Were the current instead input, it would produce an equal torque. Hence it does not generate entropy. This is far from real generators, where resistance in the coils would make them heat up and there would be energy losses due to fluctuating electromagnetic fields. But the perfect generator is in exactly the same state after having turned one turn, and hence there is no time evolution.
In practice wear on the moving parts will make it fail. But if we ignore such practicalities there is still one fundamental problem: it works at a finite temperature (if not due to its own imperfections, then just the environment). The probability of a thermal fluctuation of energy $E$ is proportional to $\exp(-E/k_B T)$. So if there are $\nu$ fluctuations per second the time until you get one that is larger than what the generator can handle after time $\approx \nu / \exp(-E/k_B T)$. This grows exponentially with $E/T$ and hence is rarely a problem.
