How a machine that makes electricity and drinking water from seawater actually works
IPHROS, short for Integrated Pumped Hydro Reverse Osmosis System, is not 3 separate structures bolted together. It is 1 machine that makes electricity and drinking water from seawater, built around a single mountaintop reservoir. Renewable electricity pumps seawater up to the reservoir. Gravity sends the water back down through a hydropower turbine, the same equipment a dam uses to generate electricity from falling water, and at the right height, that same fall already carries enough pressure to drive reverse osmosis, the process that removes salt from seawater by forcing it through a thin membrane, without a second, separate pump.
The idea did not start at Cornell. Haji coauthored the original 2016 concept paper while at the Massachusetts Institute of Technology, with Alexander Slocum and other researchers, and an account published by MIT credits a Kyoto University researcher with first proposing the idea 2 decades earlier.
Cornell University engineers Maha Haji, an assistant professor of mechanical and aerospace engineering, and Matthew Haefner, a doctoral student in systems engineering, published an optimized model of this design in the journal Applied Energy in December 2023. Their optimally designed large system could supply 79.5 million kilowatt hours of electricity and 5.79 million cubic meters of fresh water a day, enough for about 661,000 homes.
The reservoir storage will allow coastal communities to tap into renewable energy for their electric grid and potable water production.
Maha Haji, assistant professor in the Sibley School of Mechanical and Aerospace Engineering, Cornell Engineering. Source 1.
A 16% faster payback, and a tradeoff that costs almost nothing
Haji and Haefner also modeled how the combined system compares with building a hydropower plant and a desalination plant apart.
There's a potential for a 16 percent decrease in the break-even time if you do a combined system.
Maha Haji, assistant professor of engineering at Cornell, an author of the study. Source 2.
This means the combined design could earn back its construction cost sooner than the 2 pieces built separately.
The same 2021 paper behind the design shows that splitting the output of the system evenly between electricity and fresh water, rather than favoring 1, costs very little. At the balanced point, the system gives up 0.0428% of the maximum electricity it could otherwise produce and 1.91% of the maximum fresh water, a gap of roughly 45 times, showing the design can serve both purposes at once for almost no loss in power.
Figures come from the Cornell researchers 2021 peer reviewed paper, comparing a design balanced evenly between the 2 outputs against 2 separate designs that each fully maximize 1 output alone.
Show the numbers
| Electricity output | 0.0428 |
| Fresh water output | 1.91 |
Why seawater costs so much more energy to treat than fresh water
The reason the shared pressure trick matters is energy. Treating ordinary surface water for drinking takes 0.2 to 0.4 kilowatt hours per cubic meter, the 2021 Cornell paper states, while desalinating seawater by reverse osmosis takes 2.5 to 4 kilowatt hours per cubic meter. Dividing those 2 sourced ranges against each other, a MAOWCE calculation neither source states directly, shows seawater desalination uses roughly 6 to 20 times more energy per cubic meter than treating water that is already fresh. That gap is why using the fall of the ocean itself to supply the pressure, instead of running a separate electric pump, is the part of IPHROS that could actually lower cost.
Both figures come from the same 2021 peer reviewed paper by the Cornell researchers behind IPHROS, comparing 2 different water sources, not a disputed range from 1 figure.
Show the numbers
| Ordinary surface water | 0.2 to 0.4 kilowatt hours a cubic meter |
| Seawater desalination | 2.5 to 4 kilowatt hours a cubic meter |
The pumped storage hydropower backdrop IPHROS is proposing to change
None of this is running at scale yet. The 2023 paper is a model, not a built plant. What is running today is ordinary pumped storage hydropower, which stores energy by pumping water uphill when electricity is cheap and releasing it through a turbine later. The United States has 43 pumped storage hydropower plants, and the Department of Energy reports they make up 88% of utility scale energy storage in the United States today, citing its own 2024 Hydropower Market Report. Growing demand for both power and water is what keeps engineers returning to this idea.
We're seeing an increase in droughts around the world and there are regions of the world that naturally just don't have access to clean drinking water.
Matthew Haefner, doctoral student in systems engineering, Cornell. Source 1.