Encapsulated Pumped Storage, Part 10: A First Look At (Dollar) Costs

Series Introduction

One of the core claims I’ve made for EPS is that it will be very inexpensive. So far, I haven’t made a case for that. It’s time to scratch the surface of this large and complex subject.

The word “costs” can mean things besides money, like the harm done to future generations by the greenhouse gases we emit now, the costs to health of mine tailings seeping into groundwater, and so on. They’re all important, but for now I mean costs that are measured in hard currency, dollars (or Euros, pesos, or yuan). Whether we talk about capital cost, operating cost, life-cycle cost, or some other category, the fact is no technology will be built at large scale if there is a competing technology that checks most of the same boxes, but costs significantly less (or is believed or expected to).

The Complexity Of Cost

Capital costs: A few that come to mind are:

  • Land, or specific rights to use land, bought from private owners;
  • Standard manufactured items: turbines, pumps, off-the-shelf valves and piping; motor/generators and other electrical components;
  • Items made to spec: bags, custom valves, custom water-conveyance manifolds, supports;
  • Raw materials: plastic and steel pipe, concrete for foundations and retaining walls, coatings, adhesives;
  • Contract work by specialist firms: land clearing, excavation, road building, construction of site buildings (powerhouse, operations center, maintenance shop);
  • Labor not contracted out;
  • Regulatory and funding costs: lobbying, proposal writing, agreements, compliance;
  • Planning and design: electrical, civil, mechanical, manufacturing engineers; geologists, analysts;
  • Supervision.

Financing cost/opportunity cost: In the business world, money doesn’t just sit around. It’s always invested in something that is expected to produce a return. In order for funds to be available to your project, those funds have to be borrowed at interest. The interest cost is part of the total price tag of your system, and the longer it takes from getting the first truckload of money to getting the plant in operation and starting to pay off that debt, the more expensive the system becomes. Obviously, interest rates fluctuate over time, and that will have a large impact on the cost of projects that take a long time to build.

Operating cost: Hydroelectric projects don’t consume fuel or produce waste during operation. But they do need to be monitored, maintained, and repaired.

Decommissioning and retirement cost: We Americans have a long tradition of trying to walk away from these (e.g. all those retired but unsecured oil and gas wells), but that’s becoming less acceptable. At the end of the plant’s life, what needs to be dug up, hauled off, recycled? What toxins or contaminants have accumulated in the soil? Those costs should be included for a fair comparison between different technologies.

Levelized cost: this has become a popular way to boil everything down to a single number. To “levelize” cost means to treat it as a constant cash flow over the life of the project. If a facility lasts, say, 30 years, then supposing that someone else built it for you, took on all the costs (capital, financing, operation, and retirement), and charged you a fixed monthly payment, how much would that payment have to be for them to break even at the end? It’s a useful metric. But for technologies with high up-front costs and low operating costs, I often find it more useful to look at the up-front cost per unit of storage, because there are fewer assumptions baked into it.

Hydro Takes Too Long

There are two main complaints about pumped storage (and hydroelectric generation as well). The #1 complaint, which usually ends the discussion, is that there’s nowhere to build big new plants. The #2 complaint is that it takes too many years to get a new hydro plant into production, resulting in high cost and high risk. EPS directly addresses both of these complaints. It brings a vast expansion of potential sites, and it can speed up the construction of the storage aspect, which is the intrinsically slow part, significantly.

Even better, unlike conventional pumped storage using reservoirs, EPS does not need to go from zero to 100% all at once. A facility can be built, and put into service, gradually. As soon as some part (say 10%) of both the upper and lower storage fields are provisioned and plumbed, and a first penstock (which can be relatively small and surface-mounted) is in place, a small powerhouse can be dug into the lower storage field, and provisioned with off-the-shelf power components which are also sized for the initial startup. At that point, the system can be turned on, and start generating revenue and reducing emissions. While it’s doing that, the storage fields can be expanded, additional penstock(s) installed, and a larger, more permanent powerhouse put in place. All these components of the system can be added onto and brought online as they’re ready. (The smaller temporary components can then be removed, and perhaps reassigned to another startup site.) The system’s power and energy capabilities will ramp up over time. This strategy spreads out the cost, reducing up-front capital requirements.

This also lowers risk. Once the first increment is brought online, the payoff is based on actual performance numbers, as measured from the first day of operation, and the risk to additional capitalization is much reduced. Investors will know that the pace of cash inflow can be adjusted up or down depending on market conditions. This reduction of time and risk will be very appealing to most investors.

Existing Pumped Hydro System Costs

I’ve looked at publicly available information on a number of pumped-hydro systems that are in production today. The key cost metric is capital cost per unit of energy storage capacity: dollars per kilowatt-hour. For many installations, this is impossible to determine, either because no cost is published, or because only the power and not the energy storage is listed. (I hope I’ve already adequately explained why energy, not power, is the most important measure.) In many cases, the energy storage can be determined from other data, such as power times duration1.

I’ve converted quoted prices and currencies as of the year of completion of the project into 2020 US dollars2. Even so, I’ve looked for as many newer sites as possible, because comparing costs from, say, the 1970s with modern costs is problematic—many factors, like labor practices and environmental laws (and litigation) are just too different now. Here are my findings so far:

NameLocationYear CompletedStorage (GWh)Cost ($/kWh)
DinorwigWales19849.1$48.00
Bath CountyVirginia198524$67.00
GoldisthalGermany20048.5$121.00
AvceSolvenia20102.8$75.00
IngulaSouth Africa201721$167.00
KazunogawaJapan202012$183.00

These numbers must be treated with caution for all sorts of reasons, but the range is not too broad for projects of such different types, sizes, and eras. We might say that $50-200 per kWh is a reasonable estimate for conventional, reservoir-based pumped hydro storage.

How does this compare with other energy storage methods? It’s hard to trust cost numbers for technologies that are not yet in grid-scale production, and very few are. CSP (concentrating solar power, which has a built-in storage aspect) has not been faring well lately, and I won’t discuss it here. Lithium-ion battery storage is currently the most-talked-about storage technology by far, and has moved beyond the research stage into production. So for comparison, here are the numbers for the Hornsdale Li-ion battery plant I’ve cited before:

NameLocationYear CompletedStorage (GWh)Cost ($/kWh)
HornsdaleAustralia20170.185$616.00

You may be thinking, “Picking on Hornsdale again? Did you just choose an unusually expensive Li-ion facility to make pumped hydro look good?” No. Hornsdale is a reasonable comparison site: it’s new; and it’s large by battery standards, so it should benefit from economies of scale. But the biggest reason it’s here is that they actually publish both their storage capacity and their overall cost, so cost per kWh can be computed. That’s hard to find. (As I mentioned, the hydro plants in the previous table were cherry-picked for much the same reasons.) The full story for Li-ion, and its current haloed status (where the word “storage” is now accepted shorthand for “lithium-ion batteries”), is much more complicated, and will have to wait.

What This Means For EPS

All the facilities in these tables are currently in production. No EPS facility has been built, so any claims I might make about cost would be premature. (A good cost estimate will require consulting with vendors and contractors, and then computing cost from the bottom up.) But I do think a case can be made that EPS will be cheaper than conventional pumped storage. Technology aside, the fact that it can be built on low-value, vacant land, quickly and incrementally, should give it the edge on cost. The tables above show that pumped storage is significantly cheaper than current Li-ion battery prices for grid storage. In any case, it’s unlikely that anything like Dinorwig or Bath County will be built again, due to the lack of acceptable sites.

By contrast, many sites are available for EPS, comprising terawatt-hours of storage potential—a bold claim given that the U. S. Department of Energy estimates that the total installed electrical energy storage capacity in the United States is currently 250 GWh (which is at least 95% conventional pumped hydro, and barely growing), and perhaps 1,500 to 1,600 GWh in the entire world3.

Project Budgets (added 9/5/20)

Now we can determine how much we can spend on the 10 GWh Capitan project and still win on cost. 10 GWh is 10 million kWh. Let’s say that a “very good” price for EPS is $100/kWh, and an “excellent” price is $75. Then our total budgets would be:

Very good: $1 billion

Excellent: $750 million

Previous: Encapsulated Pumped Storage, Part 9: Transient Valve Operation

Next: Encapsulated Pumped Storage, Part 11: Bags


  1. The Global Energy Observatory has been very useful in doing this research.
  2. I used this site to convert other currencies into USD at a given year, and this site to convert past to present USD.
  3. link

2 thoughts on “Encapsulated Pumped Storage, Part 10: A First Look At (Dollar) Costs

  1. I like the idea that MPS can scale up gradually, give some early returns, and grow almost organically. I also like the idea of the startup package being reusable; a nice eco-aesthetic.

    I wonder if batteries, given the cost, make more sense as a local, and for now, boutique storage. That is, someone with enough money to install solar could also install a Tesla Powerwall, or equivalent. If solar gets appropriate government support, then maybe the masses could get batteries financed into their solar package. I suppose this could help buffer local demands, but probably wouldn’t soon have the ability to respond to the bigger swings in demand that mass storage systems like MPS could handle.

    Speaking of local vs.regional storage: do the various power grids basically act as one big grid? Does the power generated in New Mexico, for example, directly or indirectly benefit New York City?

    1. Li-ion batteries are fantastic. Yes, they’re not cheap. Still, I think we need all of them we can get, for the foreseeable future (which is now what, 5 years? If that?).

      Nothing else currently is any good for cars (sorry, Toyota, but the hydrogen car thing is not going well). And Powerwalls and the like (you can DIY for cheaper) are the only way to buffer rooftop solar. (MPS doesn’t work at household scale.) Uses like that will absorb all the Li-ion batteries we could dream of producing for years to come. And that’s what we should use them for, not grid-scale storage at anything beyond a few minutes’ duration.

      The dirty secrets of the national grid are (1) you lose a LOT of energy to transfer it across a time zone or more (unless you use astronomically expensive HVDC), and (2) it’s politically near-impossible to build new transmission lines because NIMBYs still rule. I’ve been looking for MPS sites as close as possible to major metro areas because of the transmission issues. (Fortunately, I’ve been finding them.)

Leave a Reply

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