[amluto]

> Suppose that you've got a pipe to the deep sea and a filtration system at the bottom, then a pump on the surface, so that the pipe is mostly filled with air.

That buys you nothing: you would expend exactly the same amount of energy to remove a given volume of permeate from the pipe this way (to keep the pipe from filling with permeate and to get the water to the surface) as you would to pump that volume of permeate through a normal water-filled pipe. In fact, it would be the same pump at the same speed. The only difference would be the pipe arrangement and the pumping system.

[impossiblefork]

Where the pump is located is indeed not critical, it's where the filter is located that is critical.

The filter cannot be on the surface. If we didn't have it at the bottom we would not be able to have flow on the high-pressure side of the pipe that is not through the membrane.

This flow is why this thing has an advantage, and it's because of this flow that the saltwater on the high-pressure side is not much saltier than seawater.

[amluto]

I should perhaps clarify. Filling the pipe with air is unhelpful. The pump (or at least the wet part of the pump) on the surface is actively counterproductive — pumps are much, much, much better at producing high output pressure than at producing suction, and you can’t suck very hard on water anyway until it boils.

Almost all modern “deep well” pumps are at the bottom of the well, and a 50 foot well is “deep” for this purpose.

[impossiblefork]

Ah, yes. I understand now.

So you propose basically pumping into the return pipe from some kind of membrane chamber and making it as on the surface-- just lift the pressure away.

Ah. Yes, then the air pipe I imagined serves no function, and presumably these real machines that are discussed in the article are of the sort you describe.