Elon Musk Should Build Pumped Hydro With Tesla Energy, The Boring Co., & Coal Miners

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!

Recently, headlines were made when an Australian team published a study finding that there was more than enough pumped hydro storage resource locations to satisfy all future needs for storage in a 100% renewable grid. As the global resource map site hosted at the Australia National Museum says:

“We found about 616,000 potentially feasible PHES sites with storage potential of about 23 million Gigawatt-hours (GWh) by using geographic information system (GIS) analysis. This is about one hundred times greater than required to support a 100% global renewable electricity system.”

In other words, we only need to use about 1% of the global pumped hydro resource locations to satisfy our needs. Furthermore, pumped hydro can store energy for weeks and the round trip from electricity to elevated water to electricity is 80% to 90% efficient. It’s already by far the largest form of utility-scale storage in the world, with more than 160 GW of rated capacity of pumped hydro in operation as of the end of 2016.

NREL published a report on the value of pumped hydro in 2018 that’s worth quoting a couple of bits from:

“PSH is a highly flexible, low-marginal-cost, and fast-acting generation asset, and in the market simulations, it was shown to reduce system-wide operational costs in both the day-ahead and real-time markets”

“In all market simulations, the addition of PSH significantly reduced the annual operating costs for the test system. Cost savings ranged from 1.2% to 2.8% in the day-ahead market simulations and between 3.9% to 10% in the real-time simulations.”

Sounds like we have a winner. It’s a rock solid technology, first deployed in the 1890s. It may be dull, but it works, it’s simple, and it’s effective. It saves money on the grid. And as the Australian study showed, there’s absurdly more of the resource than we possibly need.

So why are there regular claims that pumped hydro won’t suffice?

I published The Short List Of Climate Actions That Will Work, and pointed to pumped hydro as a primary storage mechanism that needed to be developed. Comments on that piece included derogatory comments about pumped hydro. Similarly, in my assessment of the climate action plans of the leading Democratic candidates, none call for pumped hydro, but for more R&D into storage, with the intent to reduce costs below battery storage. Since pumped hydro is already below battery storage costs, it’s a head-scratching situation.

Recently, a South African representative of an innovative company in the space reached out to me based on my piece, Joi Scientific’s Perpetual Hydrogen Illusion Comes Tumbling Down. As with many of my clients, the company wanted an independent read on its technology to understand how it fits and whether they are missing something in their analysis.

Deep shaft gravity hydro storage
Image used with permission of Gravity Power

The premise of their innovation is straightforward. If you use a big rock or concrete plug in a shaft full of liquid, you can use pumped hydro technology to move the plug up and down. Imagine a missile silo that’s generating electricity, instead of hiding an ICBM. This allows a much bigger weight with fewer mechanical construction challenges. I did a little due diligence to see if there was any way that I could add value. I reviewed their material, their patent, asked a few questions, and realized that their solution was solid.

The discussion triggered me to think about the Boring Company and Tesla Energy, and the characteristics of Tesla’s Powerpack battery solution. It’s an excellent same-day solution for fast response energy, but the characteristics that make it good for duck curves with solar don’t make it good for longer lasting storage. It’s getting cheaper, but it’s still not a cheap form of larger scale storage. When I did an assessment of the viability of a large-scale, carbon-neutral, secure greenhouse in Canada for a client this year, the Powerpack component of the solution was the most expensive portion if the solution had to remain off grid, and that solution included a 100,000 sq ft high-tech greenhouse, 3 MW of LEDs, a few acres of solar panels and a very big ground-source heat pump, all of which are capital intensive components by themselves. Powerpack’s 4-hour efficiency is 85%, but it’s not so good at 72 hours.

I asked Jim Fiske, the founder and CEO of Gravity Power and the person whose name the patents are in, about the Tesla connection. They had had conversation with Musk about this, but among other things, the shaft diameter that the Boring Company drills are too small for economically viable models.

“The net result of all these considerations is that very large Gravity Power Plants (multiple gigawatt-hours) are extremely cost effective, while very small ones are generally not cost competitive. When I say “cost effective,” I mean the levelized cost of GPP storage is roughly five times lower than li-ion battery plants.”

Between the conversations and the publications, I decided it was time to go deeper on the global study that had come out of Australia recently. Among other things, I was curious to see if it was another example of machine learning in the climate solution space, something I’m digging into as a series leading to a formal CleanTechnica report. The peer-reviewed paper on the global study isn’t out yet, but it’s an extension of an Australia-specific study published in 2018. That study is Geographic information system algorithms to locate prospective sites for pumped hydro energy storage by Lu, Stocks, et al., in the journal Applied Energy.

I realized quickly that there was a disconnect between how most people think of pumped hydro and what the study was saying, a disconnect that might be leading to dismissal of pumped hydro as a large-scale solution.

Outside of interesting innovations such as Gravity Power, there are three types of pumped hydro. One of them has characteristics that mean that it can take 15 years to gain approvals and build. Two of them don’t share those characteristics and are much faster to approve and build. The study focused specifically on the latter two, meaning that the 100x more resource than required is specifically for easier to site and faster to develop resource. That’s a very good news story.

Let’s look at the first type of pumped hydro, open-loop, to gain an understanding of the challenges. That form of pumped hydro is continuously connected to a naturally flowing water feature. That means building a dam, creating a large reservoir, and diverting water that flowed through the environment to power generation. That has a large number of environmental impacts, and is also highly subject to strong pushback from the public downstream and upstream of the facility, who typically and reasonably like the water flowing the way it is and the land unsubmerged. The large majority of the 160 GW of pumped hydro storage that exists today is open-loop.

That’s the form that takes 15 years to build, if it manages to get built at all.

However, the authors of the study focused on two types of closed-loop pumped hydro, dry-gully and turkey-nest.

Diagram of dry-gully and turkey's-nest pumped hydro sites
Image used with permission of RE100 Group, Australian National University

The top diagram shows a dry-gully pumped hydro siting. As the name suggests, this is a land feature that is suitable for damming, but one that has no water running through it. As such, upper and lower reservoirs can be created and filled without impeding water flow, damaging streams or damaging habitat and wildlife that depend on the flowing water.

The bottom diagram shows the turkey’s-nest pumped hydro cross section. It’s called that because turkeys make their nests on the ground, building up the sides. The turkey’s-nest option is suitable for flatter land, where a dry gully doesn’t exist. Flatter doesn’t mean flat as a pancake of course, but one with a gentler decline from a higher elevation to a lower elevation, so that two turkey’s nest reservoirs can be excavated, the earth used to build up the walls and connected with a bored tunnel for two-way flow of water.

The study’s authors refer to this type of pumped hydro as short-term off river energy storage (STORES), and their modeling is based on this. They explicitly looked for resources with excellent characteristics for rapid development.

“STORES is located away from rivers and has little impacts on the environment and natural landscape due to: (1) no interaction with the ecosystem of main stem rivers, (2) no conflicts or competition with nature reserves and intensive land uses and, (3) medium-sized reservoirs located within close proximity to electricity infrastructure and renewable energy resources.”

Being in British Columbia, I’m familiar with reservoir sizes for hydroelectric sites, and know that they are very big indeed. The 800 MW capacity Site C Dam that’s being developed in northeastern BC on the Peace River near the border with Alberta, for example, will have a reservoir that’s 93 square kilometers, or 36 square miles. That’s an area that would cover most of San Francisco, and one that’s quite a bit bigger than Manhattan for perspective. It’s a fifth the size of Lake Tahoe.

The reservoir needs to be that big to enable a sufficient head — the vertical distance between intake and discharge  — for effective generation. When most people think of hydroelectric, that’s what they think of, huge reservoirs that inundate a lot of land that often had people, culturally significant elements, or agriculture on it, lots of concrete, and a downstream that’s radically altered. But that’s not what STORES is.

“PHES system with twin 100 hectares (ha), 1 gigalitre (GL) reservoirs separated by a height difference of 500 m is able to contribute 1 gigawatt-hour (GWh) of storage capacity (assuming an usable fraction of 85% and an efficiency of 90%), or 200 MW of power with 5 hours of storage to the electricity system – equivalent to a large gas-fired power plant.”

They looked for very high-head sites, where the vertical head makes a big difference for the amount of energy that can be stored. After all, it’s a gravity system, and the higher the head, the higher the potential energy of water. It takes more energy to lift a kilogram 10 meters than 1 meter, and you get more energy back.

Upper and lower reservoirs of Italian closed loop pumped hydro system
Image used with permission of RE100 Group, Australian National University

This figure from the study is of the upper and lower reservoirs of the Presenzano Hydroelectric Plant, Italy.

The minimum head that they looked for in the study was 300 meters. Site C, for comparison, has a 50-meter head. That means that the reservoirs can be a lot smaller. The example above, at 100 hectares for each reservoir, is only a square kilometer or about 0.4 square miles. That’s a tenth of a percent of the size of the Site C Dam reservoir. That’s less than a third the size of Central Park in NYC or a quarter of the size of Golden Gate Park in San Francisco. The dam walls were modeled at a maximum of 40 meters (130 ft) for the dry-gully sites and 20 meters (65 ft) for the turkey’s nest sites. These aren’t trivial structures, but for comparison, the Oroville Dam in California is 234 meters (770 ft). By hydroelectric standards, they are modest.

So these are small reservoirs that don’t block rivers or streams, that are sited away from nature reserves and parks, that are sited near transmission lines, that are sited near high renewable energy resource areas and are capable of providing GWh capacity storage. The round trip efficiency is 80% to 90%, and storage can be for days or weeks, although typically its most economical for next day grid balancing per the NREL study. This contrasts to Tesla’s Powerpack which is currently very effective at in-day balancing, soaking up mid-day solar for end of day peaks.

There are two observations. The first is that siting approval for sites like this should be a lot faster and less controversial than for hydroelectric dams in general. Among other things, I wondered if the global siting study was able to have access to detailed data on sensitive or preserved natural areas of the same quality as the Australian study. I reached out to one of the study’s primary authors, Matt Stocks, and he told me:

“We use the World Protected Area Database for environmental exclusions. We haven’t had any issues with this brought to our attention. The land use is more difficult.  Our only land use exclusion is regions of high urban density.  It is not perfect with a number of smaller towns around he world inundated.”

I was reassured that the environmental approvals would not be challenged outside of Australia. The smaller town challenge should be trivial to resolve given that there are 100x more sites than required, so as the resource database is assessed by countries, they can eliminate sites with towns using country data. There’s also a proviso in the online mapping resource that geological, tectonic, and engineering work still needed to be done to validate each potential site. Many won’t be viable for reasons of slope stability and the like.

I was also curious if he had a perspective on whether siting approval was faster with STORES.

​”The individual states in Australia manage the main approvals and have regulatory responsibility for the electricity system.  NSW has released a pumped hydro road map and SA are supporting a number of proposals there.  The approval processes for wind and solar have been significantly streamlined through the states formalising the approval process and this appears to be emerging in Australia for pumped hydro.

In other words, yes. Standardizing on STORES gets us on track and allows streamlined approvals. Since we need a lot of movement by 2030, this is excellent news and something that’s replicable in every country.

Is it enough for the US? Well, what are the requirements? Per ANU:

“An approximate guide to storage requirements for 100% renewable electricity, based on analysis for Australia, is 1 Gigawatt (GW) of power per million people with 20 hours of storage, which amounts to 20 GWh per million people. This is for a strongly-connected large-area grid (1 million km2) with good wind and solar resources in a high-energy-use country.”

And what is the resource size for the United States? The study shows that the United States has about 4,500 GWh of potential STORES sites per million people, over 200x what is expected to be required.

Map of USA with potential pumped hydro resource sites
Image used with permission of RE100 Group, Australian National University

And those stores are close to major population centers for the most part. If some of the flatter states would like pumped storage, Gravity Power is happy to oblige. They are excellent for flat land siting and have even lower environmental impact concerns than STORES.

What else is true about a lot of those locations?

Map of US coal regions
Image courtesy USGS

There are a lot of coal workers in those regions who know how to work rock. Building pumped hydro is strongly aligned to their technical and engineering skill set. And there are 60,000 or more of them who want good work, and would prefer it be not too far from where their families are. I keep suggesting that Democratic candidates should make this a campaign plank, but I haven’t seen any uptake yet. I’ve asked this question of Ike Kirby, PhD and Kamala Harris’ environmental policy advisor, but haven’t had a response yet.

But back to Elon Musk. As I pointed out earlier, the Boring Company isn’t well suited for Gravity Power’s large-diameter shaft requirement. But what about closed-loop pumped hydro as identified by the STORES study? Pumped hydro requires tunnels, not a huge shaft. The tunnels range from 4.5 to 8 meters per existing sites and a Springer study.

What does the Boring Company do? It bores 4.3-meter finished shafts. What does Tesla Energy do? It does energy storage. Start at the bottom, point the Line-Storm upward to the upper reservoir, start it up and use the resulting tunnel rock and soil in the earthen bulwarks of the reservoirs. What else is good for closed-loop pumped hydro? Covering material of some sort to reduce evaporation so that you don’t have to top them up that often, which strikes me as an excellent use case for Tesla’s commercial solar panels, floating on the placid waters of the upper and lower reservoirs. Seems like a no-brainer. So I reached out to Elon.

No response yet, but it’s only been a few hours. If he does respond, I’m sure that will make for an interesting discussion.

Closing off, when I started reading the detailed study, I was curious to see if the approach used took advantage of machine learning or not. And it doesn’t. It’s a geographical algorithmic search process that first excludes a bunch of areas first, then does specific calculations about potentially viable spots. It’s computationally intensive, but using classic techniques, not neural-net techniques. This isn’t to say it doesn’t embody a good deal of insight and innovation, just not that specific type. I asked Matt Stocks why machine learning hadn’t been used for this particular solution.

“Machine learning works really well for large data sets where the machine learning algorithms can learn from one set of data and extrapolate to another. I don’t think there are sufficient examples of closed loop schemes to be able to train the algorithms. And once there is a reservoir there, we can’t see what the land underneath it looks like anymore since the elevation measurements will refer to the water level instead of the ground.”

This aligns well when compared to the CoastalDEM machine learning effort I covered recently which found much larger coastal risk of extreme water levels than had previously been understood. That study had both a global NASA SRTM dataset equivalent to what Stocks et al. used, but they also had a high-quality set of lidar data readings for much of the United States and Australia’s coastline to train it with. The number of existing dry-gully and turkey’s nest pumped hydro sites is small and there’s no equivalent to the lidar data set to correct the elevation and find similar features elsewhere.

So there we have it. Pumped hydro is a highly viable storage technology, it overlaps nicely with the characteristics of Tesla’s existing battery technology, the Boring Company has high-speed tunneling equipment suitable for penstocks, and there are a lot of excellent coal miners who could be repurposed close to home in the United States. Seems like a winner to me indeed.


Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.

Latest CleanTechnica TV Video


Advertisement
 
CleanTechnica uses affiliate links. See our policy here.

Michael Barnard

is a climate futurist, strategist and author. He spends his time projecting scenarios for decarbonization 40-80 years into the future. He assists multi-billion dollar investment funds and firms, executives, Boards and startups to pick wisely today. He is founder and Chief Strategist of TFIE Strategy Inc and a member of the Advisory Board of electric aviation startup FLIMAX. He hosts the Redefining Energy - Tech podcast (https://shorturl.at/tuEF5) , a part of the award-winning Redefining Energy team.

Michael Barnard has 698 posts and counting. See all posts by Michael Barnard