The gist of the system is as follows:
1. let a hanging weight freefall under gravity to spin a generator. (via a pulley connected to a chain drive or gears etc)
2. The power generated by the generator is stored in a battery
3. The weight is pulled up very slowly back to its drop position using a DC motor that uses power from the battery (the DC motor could be actually the generator itself but likely a specially tuned one for slow lifting will be better)
4. Repeat. The system lifting / reset process requires less energy than the energy produced by the drop, so there will slowly be a net buildup of charge in the battery.
Super simple numbers example:
PE of a 100kg mass falling 2m:
PE = (100kg)(9.8m/s2)(2m) = 1960 J
The mass falls at a near g acceleration - I know it won't be exactly g because of the moments of inertia of the generator and generating drive apparatus - but it would be quite close to g. At this acceleration the mass falls for ~0.63s, so the theoretical max power we can hope to extract using the generator is:
P = W / t = 1960 J / 0.63s = ~3000 W
Even if we consider a horrible electric generator with only 50% efficiency (just to be even more conservative and fair, and to capture any losses we've forgotten), we should hope we can get at least 1500 W from the drop.
The "trick" of mine is mentioned in that very slow "reset time" for the weight. That is, use a DC motor to pull the weight up, but give it a long time to do so. In this way, the power used to lift the weight back up to its dropping point is much less than what was produced by letting the weight fall and spin the generator.
Using the same numbers from the example above, if we let the lift time be, say 20 minutes (=1200s) and since we know the W of the lift is the same as the drop PE, then the power needed to lift the weight back to its original drop point is:
P = 1960 J / 1200s = 1.63 W
That means we produced about 1500 W (even with our crappy 50% efficiency generator), and only need ~2 W to reset the system. Assuming you put a nice battery between the generator and the dc motor - you get a net positive storage of energy over time.
Does anybody see any glaring issues with such a system and my analysis? I know when anything in physics appears to be a free lunch, something must be wrong. (Though if you consider these long 'cycle' times, it isn't really a free lunch)
Perhaps I don't understand exactly how DC motors / generators really draw or produce power in the real world - Physics 101 was a long time ago for me :)
The trick to my proposed system, which systems like those in that of the linked article don't do, is my slow reset time. Obviously not feasible for massive applications or load balancing, but a net power storer nonetheless.