[kirubakaran]

Just (potential energy) * (efficiency)

I am going to assume 100% efficiency for simplicity. [1]

mgh Joule = 20 Watt * 4 hour * 3600 s/hour

(Joule = kg * m^2 / s^2 = W * s)

Lets say you pump water to a height of 10 meters.

m = 20 * 4 * 3600 / (9.8 * 10)

m = 2938.77 ~= 3000 kg

So you need about 3000 liters (1L = 1kg for pure water) of water at a 10m high tank to power a 20W bulb for 4 hours. This about 800 US gallons. Considering other inefficiencies, you'll need more.

[1] http://www.jcmiras.net/jcm/item/93/ (first google result for efficiency of a generator) says typical efficiency of a water generator is 0.95.

[HeyLaughingBoy]

And compare that to the battery solution: 20W light at 12V: 1.7amps continuous 4 hours x 1.7A= 7 Amp-hours

That's about the capacity of a small 12V lead-acid battery that you can carry in one hand. Weighs about 4 lbs. I use one to provide LED lighting in my chicken coop because it's too far from the house to run AC power out there.

No contest: in the absence of extenuating factors, a cheap battery is better than the "innovative" (?) water storage method for such a small amount of energy.

[bulanga]

thanks that's very helpful indeed.

[sho]

In other words, you would need such a ridiculously large tower that the idea is not really viable. 3 tons of water for 80 watt hours? That's 37kg of water per watt hour. I use more like 4 kilowatt hours in a night; that's almost 150 tons! You're starting to talk about a serious, and seriously expensive, engineering project.

Maybe it would be viable if you lived next to a convenient cliff or other ready structure to simply place a big tank on top of with a pipe hanging down but if you actually have to build the structure yourself it sounds like a non starter. And that's without even considering the other engineering hassles you'd have, like corrosion, filtering seawater, maintenance, etc.