Encapsulated Pumped Storage, Part 17: Large-Scale Potential

Series Introduction

The Capitan site, with 10 GWh of storage, was one of the first locations I identified as a promising place for EPS. Since then, I’ve found a number of others. The best EPS sites will, above all, have these characteristics:

  1. Very large energy storage potential, of course. (This means large areas for both upper and lower storage fields.)

2. Proximity to major population centers, where the storage is most needed. This reduces reliance on long-distance transmission lines, which are expensive, hard to get approved and constructed, and waste electrical energy in proportion to their length. (“Proximity” is a relative term; Los Angeles has been getting a lot of electrical power from Hoover Dam, 266 miles away, since the 1930s, and some of its power supply now comes from as far away as Utah.)

3. The land must be acquired, either through actual ownership, or as a long-term right to use it for an EPS facility. The land should be as “useless” as possible; that is, the fewer residents and the less economic, ecological, or recreational value it has, the better. In other words, hardly anyone should be upset at the idea of it being temporarily (in the grand scheme of things) covered with bags of water. (Of course, this will in some cases conflict with #2, proximity to population centers, since empty land close to cities tends to be valued for one reason or another, if nothing more than as a place to escape from urban living for a while.) Obstacles to acquiring the right to build EPS plants also include jurisdictional issues—land in designated wilderness areas, or military reservations, for example, might be impossible to get.

Other factors that are important, but less so than the big 3, include how easy or difficult the terrain is to work with, easy availability of water, proximity to highways for bringing in materials, climate and weather issues, seismic stability, and so on.

Just by looking around the western United States in Google Earth, without aid of a site-finding algorithm (which is still under construction), I’ve found about 16 promising sites where an EPS facility of meaningful size could be built. I’ve been advised not to reveal the locations (Capitan aside) yet, and I see the wisdom of that, because each one of them—no matter how barren—is bound to be someone’s favorite place on earth, and there’s no point making anyone angry so early in the process. So I’ve classified them by the approximate amount of energy they can store, as well as the distance to the nearest big city. The term “Southwest,” which I sometimes use, is technically too restrictive: states where sites have been found include California, Arizona, New Mexico, Colorado, Utah, Nevada, Wyoming, Oregon, and Hawaii.

Those sites add up to a total potential storage of 2,808 GWh, or 2.8 terawatt-hours. Raccoon Mountain, in the comparison section, is the world’s largest energy storage facility in terms of energy capacity (Bath County is the largest by power). The identified EPS capacity is 78 times as much as Raccoon Mountain. How can EPS possibly have so much storage potential?

Traditional pumped storage relies on a very unusual set of conditions. It’s not enough to find two locations at a large elevation difference, which are also a relatively short horizontal distance apart. In addition, either a large body of water must already exist at the higher elevation, or it must be possible to create one artificially using earth-moving and damming techniques. There are few places where a concave shape of large volume, which is what a reservoir is, can be constructed at a high elevation, at an acceptable cost. Traditional pumped storage sites also are typically in well-watered regions, where most land is economically and/or ecologically valuable. Finally, the head available is usually moderate, meaning the water storage volume must be very large in order to store enough energy to justify the project. All these constraints mean that each pumped storage site is a unique gem, and most of the good ones have already been exploited.

EPS sites have much simpler requirements. The two areas at large vertical, but moderate horizontal distance are still required, but they only have to be somewhat flat. It’s not just OK, but actually preferable that the land be waterless, barren, and of little value. Places like this do exist in the American West, and furthermore, some of them are quite large.

I should mention that every site I’ve identified appears, on Google Earth and various maps, to be completely uninhabited. It’s hard for many people to imagine how utterly empty much of the American West is, in comparison with the rest of the country. The cities and towns are tiny islands in vast seas of uninhabited miles. The highways seem empty enough, but in many places, if you park at the side of one of those highways and walk a few miles from it, you might as well be on the Moon as far as human impact is concerned.

Building out one of these large EPS sites to its full potential, all at once, would be very expensive. For example, a site with 50 million square meters (19 square miles) of usable storage at both top and bottom, 1 kilometer of head, and 333 GWh of storage potential, would cost—based on our preliminary estimates—about twenty billion dollars. But the modularity of EPS means that you don’t need to build the whole system, in any respect, all at once. You can start small and keep adding capacity incrementally. One of these large sites is somewhat akin to a large ore deposit—you wouldn’t expect to mine out every last bit of ore before you sell the first truckload. The same is true of solar and wind farms; once you’ve identified a parcel of land with the right characteristics and obtained the right to use it, you can build it out at your convenience.

Suppose, as a thought experiment, we have a mega-site with 333 GWh storage, and 24 hours’ duration, so peak generating power is about 14 GW. If we specify extremely large turbines and pumps, each having 1 gigawatt of power1, then the system would have 14 turbines and 14 pumps, with an appropriate number of penstocks to serve them. This means that an EPS “mega-site” of that size would really consist of a number of more or less independent facilities, built at different times and each having its own particular head, penstock length, and so on. This would be advantageous in allowing each facility to take best advantage of its particular terrain. Cost savings would still be obtained by unifying control and monitoring systems, security, grid connection, and other aspects of the mega-site. The mega-site would also be able to take advantage of technological and cost improvements occurring during the multi-year buildout process.

2.8 Terawatt-hours is an enormous amount of storage—quite possibly more than exists in the world today (it’s hard to be sure because of the reporting issues discussed earlier2). One way to put it into context is to imagine building that much storage in the form of Lithium-ion batteries. Focusing on just one of the raw materials, how much lithium (as a pure element) would that take?

This article says “It is estimated that there’s about 63 kg of lithium in a 70 kWh Tesla Model S battery pack.” So, 0.9 kg/kWh. 2.8 TWh of energy storage would require 2.5 billion kg, or 2.5 million tons of elemental lithium. (This is a one-time investment, if you assume you have very efficient recycling to reclaim the lithium when, after some years, the storage capacity of the batteries has diminished to the point that new ones are needed.)

Is that a lot of lithium? This says that 2018 Li production worldwide was 95,000 tons. At that rate, it would take 26 years to mine the 2.5 million tons of lithium we would need for 2.8 TWh of battery storage. Granted, lithium production will probably increase significantly in the future—but so will our need for lithium for other uses besides long-duration grid storage. By developing EPS in the American West and other parts of the world where conditions are suitable for it, we can spare at least 2.5 million tons of lithium for things like vehicle batteries.

Finally, in a previous post I made a rule-of-thumb estimate that California might need 490 GWh of storage in order to supply 90% of its energy needs from carbon-free sources. The EPS capacity I’ve identified is almost six times that much, and a considerable fraction of it is in California. So I’m confident that EPS can supply all of California’s long-term grid storage needs, and team up with wind and solar to make its grid 100% carbon-free—and that it can do the same for a number of other Western states as well.

Previous: Encapsulated Pumped Storage, Part 16: Lessons From Modeling

Next: Encapsulated Pumped Storage, Part 18: Summary, So Far


  1. The Three Gorges Dam in China has 32 turbines, each producing 700 MW.
  2. This EIA page has a footnote saying: “Update 20 September 2019: A previous version of this article stated that the current storage volume of PSH plants is estimated at 1500 GWh. This has been updated to 9000 GWh based on IHA (International Hydropower Association) (2018), The world’s water battery: Pumped hydropower storage and the clean energy transition, IHA, London”. The earlier estimate of 1.5 TWh seems in line with other estimates as well as with the sizes of the very largest pumped storage plants, so I looked up the IHA article cited to see where they got 9 TWh. That article states, “Current open-loop systems can also go beyond 100 GWh in energy stored such as the Vilarinho das Furnas project in Portugal.” That claim is dubious given that it would be at least three times the size of Raccoon Mountain, which is widely cited as the world’s biggest by energy. I believe the key is that this is an “open-loop system” per the quote above, meaning that it’s a hydroelectric generating station which also incorporates a pumped storage component of unknown size. The upper reservoir is replenished by precipitation captured from a large area. Citing the energy stored by such a system as if it were all pumped storage is misleading. I suspect the 9 TWh of existing pumped-hydro cited by IHA, which is out of line with any other number I can find, is based on tricks like this. I put more credence in the 1.5 TWh estimate originally in the EIA article.

Leave a Reply

Your email address will not be published. Required fields are marked *