My belief is that waste water will be the only long term success for this, but there are plants either being proposed, set up or already existing in deserts that use mined or otherwise created nitrates and phosphates and even (in a minority of cases, I hope) allow water evaporation - in the kind of places where water is likely to evaporate very quickly. After that, there's the task of getting the oil (or lipids, as the algae people call it) out of the cells, for which there are several techniques, all of which use energy - hopefully most of this could come from photovoltaics or something similar, but I expect that there are some that use fossil energy to separate the oil from the rest of the cell.
As for using waste water plants - I expect that there would have to be some re-engineering of current plants, but not necessarily completely new ones built - where there are plants already in place. Where there aren't, it will be a while before anyone sees any profit in building one to make algae oil and then there's the problem of where to get the land to build them on - I don't see a new one being built in Manhattan, for example. Trent's project (omega or something like that?) gets round this for coastal areas (like Manhattan) but there's a lot of places that doesn't apply to.
I don't know how we can expect to filter runoff without causing major changes to river flows and by extension ecosystems, but as it's basically free nutrients, people should definitely be working on it. It's still early days for the whole industry, so most of those setting out now will fail but I expect a few will succeed and they will have a formula for how much oil they can make consisting of elements such as:
1: How much in the way of nutrients that are currently being thrown away does an algae cell need during its lifetime
2: How much of these nutrients are currently thrown away in sewage / runoff / farming / mining waste (I don't know that the latter two happen, but just in case)
3: How much water is freely available to support the maximum amount of algae that 1 and 2 suggest can be grown - you're right to think that it doesn't need to be clean water and there may even be strains that clean water with small amounts of industrial pollutants in
4: How long before each generation of algae produces enough oil to be worth extracting it (about a month, maybe)
5: How easy is the oil to extract, giving
6: how much oil a month can be made
This, combined with how much the oil will sell for will give a clear idea of how much money it's worth investing in a new algae / sewage / whatever plant or re-engineering what's already there. So far, most of these decisions seem to have been made based on what has been done in labs, but with people like Solazyme charging vast amounts for their diesel (admittedly, most of what the US Navy have paid for is further research, but it still looks like thousands of dollars a liter of diesel) you can see that this is still in the research phase and even though it's being treated as the next gold rush by some, it wouldn't be research if we knew where it would end up.