You then turn off half of the motors. The generator is still spinning at the same speed producing the same amount of energy because it does not know the motors are turned off and you need less energy. For that split second the wires are conducting far more energy than what you need. What happens to that excess energy? Where does it go?
Directly into the spark(s) that follow switching such a load.
Power is going to ~zero every cycle, too. It doesn't need to go anywhere at all. A "lucky" switch can open during the zero crossing.
(This is the case for single phase. Three phase always has power going somewhere; you'd need an
extremely lucky contactor to open two lines in the correct phase so that no current is being switched.)
Do we agree the speed of electrons in copper wire is about 1/100 the speed of light? Same would happen in your power supply example if you added 1,000 miles of wire.
Slower than that. But electron drift velocity (which averages
zero in mains, BTW) is completely irrelevant to power transmission -- the contactor opening or closing is communicated at light speed [in the conductors], so the force on the rotor changes almost immediately (10s-100s microseconds).
After that, all the mechanical dynamics and control systems play a role. The instant effect is the generator accelerating incrementally. Its frequency does not change, in that it makes much less than one full rotation with respect to the grid. No, its angle change draws power from neighboring generators (others within the plant, and from the grid at large), and that angle oscillates back and forth, like a torsional spring, until the transient energy is dissipated. Dissipation occurs through losses in the system (generators contain shorting slats*, to reduce harmonic oscillations and dampen these vibrations), transmission losses, and the controller.
*More specifically, synchronous machines do. Shorting slats are simply shorted turns on the armature, which act to slow any change in its magnetic field (which is an electromagnet, so the field is supposed to be static anyway). Note: I haven't studied modern generators per se -- their design may vary. Anyway, even if this isn't the exact solution used, the same effect is required, however it's achieved.
The amount of energy stored in the grid is
extremely small. Only rotating machinery has any inertia at all, and it only accounts for about 10% of the total available power (IIRC). PFC capacitors might account for 30% of that in turn.
When the grid goes down, whether it's a small neighborhood or the entire northeast, it simply disappears, poof. Likewise, when it's restarted, the load is there, ~instantaneously, no windup, no springiness. (In practice, there's a lot of inrush associated with restart. But they can deal with that, as, of course, they must.)
Tim