|MadSci Network: Physics|
There are a couple of things wrong with this idea. The first is that you only use the tiny fraction of water that contains deuterium. By tiny, I mean 0.015% (it's on the standard table of isotopes that you can find at http://www2.bnl.gov/ton/ ) of hydrogen is deuterium. Since you have 2 hydrogen atoms per water molecule, 0.03% of water will be heavy. You'll actually have to re-separate it later after you get rid of the 0.2% of heavy water that results from having an oxygen-18 in it (8 protons, 10 neutrons) and 0.038% with O-17. All told, it's still only about 0.3% of the total water After the heavy water is separated from the normal water, then the normal water can be returned to the source. This will leave the source relatively unchanged. The total mass of deuterium from the 500 l of water is only 165 g. There's another reason, aside from that we're now only talking about 10 million cubic meters of water, or a layer of ocean which is 1 meter deep and 10 square kilometers in area. That may sound like a lot, really, but there are 361,000,000 sq. kilometers of ocean in the world (result of a google search). That means, on average, you're lowering the ocean surface by under 30 nanometers to power the world for everyone for a lifetime. With global warming, we're concerned about raising that ocean level by several meters, or billions of times the change in the surface level. So no, even if we used all the water that was taken in for deuterium extraction was somehow used and even if we ran for 100 lifetimes we would lower the ocean surface by 1 mm. And the CO2 from current power plants will cause enough global warming to counteract any such effect many times over. There are also advances in technology to be considered for the future. Perhaps we will conceive of a better power source than fusion. As soon as people do achieve fusion, they'll want to step up to more advanced versions (CNO cycle fusion and such). As to the tritium, you said you need 30 grams of lithium, which is itself a less common isotope (7.5% of natural lithium). To get the right amount of lithium, however, you actually probably need more like 480 grams to get 1 6Li for every deuterium, meaning you need 6.4 kg of natural lithium. However, it's actually easier to extract tritium from the cooling water of a reactor (deuterium is naturally converted to tritium via neutron capture) than from lithium. It's also environmentally safer, since you don't have to mine lithium that way (mining is one of the most destructive things we do to this planet). After all, you need neutrons to get tritium from lithium anyhow (although this process produces significant heat and can be used to produce energy). Power production is an enormous industry with wide-reaching implications because of its massive scale. You have to think about all of these parts in detail before passing judgement on any one thing about it, as the length of this question response will attest to.
Try the links in the MadSci Library for more information on Physics.