I believe Erik’s statement that pumped hydro accounts for 90% of grid storage might be based on out of date data, as battery storage has grown dramatically over the last four years. A May 2024 article on “RenewEconomy” website titled “Battery storage is about to overtake global capacity of pumped hydro” indicates that on a global basis pumped hydro made up 90% of grid storage in 2020, but by 2023 it was down to 68% and expected to end 2024 at 56% and that battery storage would surpass pumped hydro in 2025.
For the US, EIA shows that utility-scale battery storage increased from 1 GW of capacity in 2019 to 20.7 GW in July 2024 (https://www.eia.gov/todayinenergy/detail.php?id=63025). Since pumped storage hydropower (PSH) capacity stayed approximately flat during that period at ~22GW, battery storage is gaining capacity share very rapidly.
However, it is worth noting that battery storage capacity in energy (GWh) rather than power (GW) units was 42.5 GWh at the end of 2023. For PSH, it was estimated to be 553 GWh (https://www.energy.gov/sites/prod/files/2021/01/f82/us-hydropower-market-report-full-2021.pdf). Comparing them in GWh brings home the point of the much longer storage duration of the typical PSH plant versus the typical battery.
Great to hear there are companies working on this.
I wonder if funds could be made available for the purpose of remediation of Mountaintop removal coal mines, turning them into pumped hydro? Instead of fighting to hold some deadbeat accountable for their mess, try to turn it into a community asset.
I've heard tales of pumped hydro historically getting built in order to buffer inflexible nuclear.
Daily and weekly power storage is believably economic both from pumped storage and batteries, but I have yet to read of economically feasible seasonal storage that could account for the power-demand swings from summer to winter and back. What options are there? Iron-oxygen batteries sited in former automobile wrecking lots? Pumped storage with huge upstream lakes that include recreational vacation cabins?
A technology commonly cited for seasonal storage is hydrogen. While it's not very efficient to convert electricity to hydrogen and then back to electricity, this is less important for seasonal storage. (One way to make sense of this is that the hydrogen tank can be filled with cheap or free (curtailed) power over many days.) Iron-air batteries as you mention are another tech that target seasonal storage.
I think on paper flow batteries with very large electrolyte tanks supposedly make sense for very long term storage, the problem is nobody has solved the right combinations of chemistry and manufacturing processes to actually make them cheap to build. (Like if you could magick one into being it would have cheap OpEx, but the CapEx is prohibitive.)
Heat pump geothermal does offer low cost inter-seasonal energy storage. Check out prior Volts interviews. It doesn't make electricity but it greatly reduces demand for energy for heating and cooling.
Yes, I remember that interview and fully agree that geothermal seasonal storage is an excellent option for heating and cooling of buildings and potentially industrial heat as well. We will need some options for electric power too, and those are not so clear to me.
Ehhh. Geothermal is great for how it offers a source of heat differential year round. I don’t know that I’d say it’s as effective for _storage_. It’s not like you can take excess solar in the summer, and pump superheated water down to heat up the earth, and then recover that heat months later. There are serious long term thermal storage ideas (like molten salt, and the box-o’-rocks approach where they line the box with IR-PV) which probably work on a time scale of days out to a week or two. I’m not sure whether any of those is plausible for season-to-season storage, though.
That is incorrect; such systems can and do give true inter-seasonal energy storage, where heat derived from summer cooling of buildings (and sometimes solar energy or waste heat or other sources adding heat on purpose) adds to the thermal mass, to be used as heat during winter. Larger, "busy" buildings with borehole heat pumps can add more heat over a full year to the underground thermal mass than they draw back for winter heating and often require some cooling towers to keep it balanced.
For cold climates the amount of energy for heating is large; reducing the need for direct energy use greatly reduces demands for energy.
Huh. How efficient is that? While thermal conductivity down there isn't that high, it's still not _zero_, I'd think over months, added heat would diffuse away into the background, and the immediately-accessible mass would move back closer to the ambient temp.
I am no expert in the thermodynamics but practical experience shows they do work very efficiently, superior in energy in vs energy out to almost any other options. A previous Volts interview featured thermal heating networks based on geothermal heat pumps that had COP"s (thermal efficiency) of above 5, that is 5 times more energy utilized than was used by the system. https://www.volts.wtf/p/thermal-energy-networks-are-the-next
My limited understanding is that whilst there will be losses at the peripheries earth and rock effectively act as an insulator over longer distances and the larger the heat mass the more efficient. With large fields of boreholes the volume and mass is much larger compared to the "surface area" (where leakage occurs) than for small thermal masses. As the volume around the working mass warms the rate of loss at the peripheries drop off and the system works more efficiently.
Yes I listened to that piece. But the reason these networks get high efficiency (units of energy expended, per unit of energy applied to the environment) is precisely that the earth stays at an extremely steady temperature regardless of season and regardless of how much energy you pump in or out (at least at the scale plausibly used for HVAC). These heat pump systems expend a small amount of energy, to _move_ a larger amount of energy. For this specific purpose, the fact that your "reservoir" can provide or soak up a nearly unlimited amount of energy is a feature. You can keep heating the homes and businesses on the network, by sucking heat out of the ground, and the ground doesn't get notably colder. And equally, you can dump heat into the ground during the summer, and the ground doesn't get notably warmer. I don't believe it's meaningfully the case that you're "recovering" the same heat in winter that you dumped in during the summer. The rock around the deep wells gives you an enormous thermal mass, and diffuses heat reasonably-efficiently over the course of days. If it didn't, it _wouldn't_ work as the heat sink / heat source for these HVAC systems. But the same property would seem to make it a poor heat sink / heat source for long-term storage.
If you run 100 GWh of energy through resistive heaters, apply that heat to water, and run that water down into the ground, you will at that time raise the local temperature some amount. But by six months later, I feel fairly certain, most of that energy will have diffused through the ground out to distances of _miles_, and will be effectively unrecoverable. I could be persuaded otherwise, but I'd want to see actual data. The thermal network piece you're talking about absolutely _does not_ talk about trying to store and recover significant amounts of energy.
There's also the separate technology of deep-geothermal (exploiting fracking techniques to get down deep enough that you can get geothermal heat without sitting on top of a shallow volcanic resource). But that too is reliant on the fact that the temperature at the depths you're drilling to is more-or-less constant, with heat diffusing from other parts of the deeper hotter crust, into the part you're extracting heat from, faster than you can extract that heat. If you tried to inject extra heat to store it, it would be lost to distant parts of that same layer of crust by six months later.
A key aspect is that the economics of these systems work best when demand is high and the requirement is immediate but cannot be exactly determined. For example, half-time during a popular sports event. Being able to spin up turbines to maximum load in 30 seconds is crucial.
It's also worth remembering this pumped hydro facility took 10 years to build and was at the time the UK's largest civil construction project. Battery tech may undergo further transformation to supersede pumped hydro over such an extended time period.
I would sure like to see the economics of storage systems expresses in Tonnes Avoided Carbon (TAC).
Every lithium battery plant that I know of (and I am ignorant, so please correct) is located right next to a a big power station: the Tesla facility east of Reno, the smoking ruin of the Moss Landing, the proposed Morro Bay factory.
These batteries are energy-intensive, and therefore carbon intensive. TAC is the better unit of measure for these computations than is Dollars, no matter how they are adjusted for inflation.
"Median cradle-to-gate carbon footprint of lithium-ion batteries
between 48 and 120 kg CO2e kWh−1."
Coal generation is almost 1 kg CO2e/kWh, gas around 0.5.
Split the difference and the battery embodies 90 kg/kwh, divided by 0.75 kg/kWh avoided fossil generation emissions, divided by 80% depth of discharge, seems to yield about 130 battery cycles for a battery to offset its embodied emissions if the cycled variable renewable generation would have otherwise been curtailed and fossil generation used instead. If the battery is cycled half the days, that seems like about 8 months emissions payback. Not too bad IMHO.
In any case for any storage, duty cycle or similar capacity factor is important to payback of emissions, energy, and $ invested. Kinda like if you drive your pickup 1000 miles/year, you shouldn't buy a Rivian to save the planet. Instead contribute to BEV buses for you school district or transit authority which are driven 40,000+ miles per year.
The existing PHES reservoirs in CO don't need plastic or concrete or asphalt linings. Many reservoirs don't. The lithium batteries I see are located at PV power plants. Even if they located at an old powerplant to take advantage of the grid, the ones in CA are clearly packing in the PV during mid day and releasing it to avoid evening fossil generation. CAISO has oodles of data on that.
As far as Moss Landing, seems like a good case for LFP instead of NMC, and hey Thomas Edison almost burned down the Vanderbilts' house. And did burn down a bunch of other stuff.
I am a simple-minded, or at least simplicity-minded, economist, who finds the simple path to survival to be found in treating terrestrial carbon, that would otherwise be in the atmosphere, like the ultimate scarcity good---the basis for a carbon economy in which *every* item is priced in Tonnes Avoided CO2 (TACs).
Our production cost models (I am an expert on the PROSYM family, which shows how old I am) we minimize Dollars or other nominal currency. It is not so hard to change terms to TACs. I suspect that LADWP is using their house interface to do stochastic structural modeling of the WECC, for example to build the business case for that big wind--and-transmission project they are doing with Anschultz
My claim to membership in the energy elder club is my proficiency at text-mode DOE-2 building energy modeling.
I'm kinda simplicity-minded myself. I'm a big fan of understanding when calculations result in "precision without accuracy," hence my two significant figures and wide input ranges for my little calc of battery GHG "payback."
For something that can be presented as numerically precise as TACs by switching from gas to electricity, or deploying VRE and storage, widely varying conclusions can be drawn from using different methods for looking at carbon emitted in future years by electric supply at any given location. Annual averages, short run marginal or average hourly, long term marginal or average hourly. All will give different answers, and depend on VRE deployment rate assumptions. Then, how, and how much, do we attribute methane leak & vent to gas use in either remaining powerplants or heating systems?
I think we would agree that as the time period in long-term energy storage for VREs gets longer, the number of cycles goes down and the embodied carbon and dollar cost goes up for each unit of stored energy, to the point some solutions may actually emit more hidden embodied carbon than they save.
I'm usually dubious (but not totally cynical) about forest or agriculture sequestration and offsets. But if we electrify most everything and generate only 10% of the that electricity with fossil fuels, using those bio GHG sinks and GHG avoidances to offset small remaining fossil GHGs might be "better" than Herculean efforts at long duration storage of VRE.
Offsets were (note tense) a successful industrial policy action, that gets things moving in the right direction.
In the carbon economy, the basic unit of value is TAC. Carbon firms maximize TAC, and all the values of conventional microeconomics can (and, in our crisis, must) be expressed in those terms. Like wages, ferinstance.
Go browse my Substack and ask me questions there. Or here...I don't want to poach readers from this great site,
"I have found my people" is something I have been looking forward to thinking. I think I may be right!
I want to know whether batteries are really better (in TAC terms) thatn the locally-produced biofuels for hauling the surplus biomass that forest recovery requires. From the perspective of the pathologically overgrown forest, which wants to sequester carbon not send it all up into the atmosphere at once, the biomass facility helps with its waste product; the challenge is to get the highest value work out of the waste biomass. The main value is in the forests (with all the birds and bees and nematodes riding along for free in the carbon economy).
Of course the concrete lining for the closed-loop reservoirs is also carbon-intensive. and just like everything else in the carbon economy must be priced accordingly.
Asphalatized concrete is a new term to me, but asphalt, in the grand, carbon accounting, is a carbon store, so the (durable) tar in the lining earns whatever value the carbon-economy delivers to carbon stores (such as, for example, the carbon capture systems they are bolting onto the Icelandic volcanoes).
The carbon economy is loaded with opportunities for innovation. The forests sector (which delivers liquidity and therefore facilitates credit) has a natural planning horizon that is measured in centuries, rather than the decades we are used to in the fossil economy. Long-horizon projects, like closed-loop pumped storage, are supported by the (base loaded generation of) the TAC-positive forests industries.
Do you have recommendations for references that compare alternate energy storage systems? Such a reference would evaluate costs (initial capital and operating), efficiency, and environmental risk factors that would affect applicability at a specific site.
This is the first I've heard of the Goldendale project here in my home state! Most of the renewable energy headlines here have been about the Horse Heaven Wind Farm and all the legal disputes around it.
I get a kick out of the Goldendale project being able to produce 1.21 Gigawatts!
Just finished listening to this. The bit at the end where he describes 60 acres as not being all that much land for the amount of power / energy you get kind of made me boggle. Like yes rural land is cheap and you can tie into transmission to send the power to Seattle or wherever. But that’s still I want to say an order of magnitude more land than a comparable PV+storage site. Am I wrong? I could go look up the actual areas of sites Tesla has commissioned recently. I don’t have a strong gut sense of how big an acre is, but like Hornsdale is >100 MW and I’m not sure it’s much more than _one_ acre. You can walk across it in a matter of minutes. Moss Landing (currently shut down because the nearby Vistra system is on fire) is like 180 MW, and similarly you can walk across the whole thing in minutes. 60 acres sounds like it ought to take more like 30-40 minutes to walk across, if it’s a square.
I had exactly the opposite reaction. An acre is 43,560'sq or 209x209 feet which could be walked in seconds. The full 60 acres would be @ 7.9x7.9 acres or @ 1650 x1650 feet; a little more than a third of a mile or @ 5.5 football fields. Takes 15-20 minutes to walk a mile so the full acreage would take 5 -7 minutes to walk across. Wouldn't the closed loop system have reservoirs top and bottom and a generation facility? Figure 2-3 acres? so @ 14 acres per reservoir or 3.75 x 3.75 acres ea. or @ 784 feet per side? This coverage to supply the entire city of Seattle for 12 hours? Seems remarkably tiny.
Well, I went and looked with GMaps' measuring tool, and the main Hornsdale battery facility is about 400' by 150'. So I was right, it's not much more than a single acre.
The quote in the text is "So yeah, 1200 megawatts, 12 hours of storage, two 60-acre ponds".
So about 5x as land intensive in terms of power per unit area. (Actually more than that -- the PowerPack 2 system at Hornsdale would be significantly less energy- and power- dense than the current-generation Megapack 2XL.) Less of a difference on energy -- Hornsdale is around 200 MWh on one acre... 1200 MW * 12 hrs / 120 acres, that's around 120 MWh / acre. Still less, but not by a huge multiple.
well again I'm not technically trained but just looking at what appears to be the site at Moss Landing ,CA on a Google maps aerial it appears to be as much as @ 68 acres total including the transmission line and etc. Compared with 60 acres mostly comprised of reservoirs full of water Vs. a lithium battery backup/storage site that burned and spread toxins over a miles wide, possibly thousands of acres swath of fantastic agricultural land. Externalized costs.
The site that burned was the Vistra system which was inside the old power plant, not the Tesla system that's on open ground. AFAIK there has never been a large-scale fire with a Tesla system. A _single_ Megapack burned at the Victoria Very Big Battery site. And at this point Tesla is converting to the Lithium Ferrophosphate chemistry for stationary batteries, because you don't need the additional energy and power density you get from the nickel-cobalt chemistries, when doing a stationary site. And while it is in theory possible to get an LFP to burn, it's pretty difficult.
The vast majority of the Moss Landing site is stuff that was already there long before the batteries, because of the gas plant. Also, the Tesla Moss Landing batteries are something like 1300 MW, so of course they take up more room than the 150 MW at Hornsdale.
As far as the area, unless you think the various transformers and transmission lines are going to be sitting on top of the water, that isn't going to be a difference between the two options, you'll still have that stuff sitting next to the water driven turbines, to convert the voltage and get it onto the grid.
I'm familiar with the Cabin Creek closed-loop system in Colorado (did a fatality investigation there with the CSB). As a storage system it seemed just about ideal. Modest footprint, significant capacity for peaking power in the Denver area. Where are the other closed-loop systems currently operating in the US?
It sounds like their Kentucky project might actually get built.
But I think you downplay the "environmental/progressive/NIMBY/tribal" opposition to the projects in the NW. I hadn't heard much about the Swan Lake, but a very quick Google Search shot back an Oregon Public Broadcasting post (https://www.opb.org/article/2024/04/09/klamath-falls-oregon-pumped-water-energy-storage-project/) calling it "controversial, and that it required a $40M payout to local tribes.
There have been many stories about the Goldendale PHES. Even though the upper and lower reservoirs are only 60 acres each or whatever. A recent example being
Gotta love, "An East Coast company backed by European investors..." Settler-colonialism all over again!
High Country News has a NW tribal reporter who has been all over it (and seemingly all solar or wind in SE WA), and I see "Columbia Riverkeeper" is quite wound up.
I sure like the combination of wind, water and solar but natural energy does disturb nature and its lowish density and location right on the planets surface means that disturbance is quite obvious. OK with me, but disturbing to many.
Good stuff. I struggle, however, to put numbers into perspective. Worse, what are they? In particular, when you say 1.2 gigawatts, is that peak output? Or shorthand for gigawatt-hours. Given this whole article is about storage capacity, I was surprised to not see a single reference (did I miss one?) to (x)watt-hours. You do talk about hours of capacity, so does that mean 1.2 gigawatts for 12 hours?
While I'm here, I would also like a short post (or link to one you may have done?) on what's considered a "typical" household. As in, "could power a city like Seattle" etc. According to Sense, I'm on the high side of their typical customer at 4 kw average use for the last 30 days. Electric cars, heat pumps, and a family that likes to keep the whole house warm. In summer, a quarter of that. 3MW-hour+/month winter, 1MW-hour/month summer.
Sorry for the rant, but I really try to understand the magnitude of things, and it just drives me nuts...
The units of power (watts) and energy (watt-hours) can certainly be confusing!
1 watt-hour is the energy produced by 1 watt for 1 hour. Prefixes killa, milla, giga are are of course 1,000 times more each. So yeah, if you say 1.2 gigawatts (GW) that's the power - and you'd assume it's the power at a peak steady-state. You'll sometimes hear "Rated Power" which means the expected normal steady power. If a pumped hydro facility has 2.4 GWh of energy with a rated power of 1.2GW, you'd expect it to run for two hours before it's empty. (1.2 GW x 2 hours = 2.4 GWh). As for the "typical household" - it's just taking an average.
We often talk about batteries as a power and duration. So, a 2 MW, four-hour system would store 8 MWh. (Tesla’s products generally come in a 4hr, 2hr, and sometimes a power-boost / peak-power version that’s something like 1.7hr, I forget the exact number.)
Well, if they're remotely competent, they're just quoting the power (which is often what matters in relation to substituting for something like a gas peaker plant).
The thing that's a bit confusing about the use of numbers of hours in this interview is that there's a bit of slippage between the duration _for which energy can be stored_ (without significant loss of energy, and without driving the levelized cost of the system too high because the investment is getting amortized across fewer events), and the "duration of the battery" in the sense of how long it will run at its nameplate output, to go from End of Charge (or 100% State of Energy) to End of Discharge (0% State of Energy).
And actually, while I think it's correct that if you want to time-shift energy use by a day or more, pumped storage is probably going to be more efficient, lithium-ion batteries certainly _can_ store energy on the scale of a few days. Depending on the details of your system, the idle losses could be as low as 2% over a week if you're being as conservative as possible (putting the system into a dormant mode until you really need it), up to as high as something like 10% if your system has high idle loads because it's keeping the bus up and the AC system prepped for instant demand, and it's dealing with thermal loads because of very hot or cold weather. And then you'd lose perhaps 10% to the general entropic issues that are usually described in "round trip efficiency" estimates for batteries. (You also would've lost that if you were just continuously cycling, starting to discharge the moment the battery got full).
But, like, if you can take X amount of energy on a high production day, and get back 80% of that on a low day, that's not peanuts. Pumped storage can beat it, but it's not available everywhere. The main reason battery sites typically don't do that is just that there are opportunities for energy arbitrage every day, hour to hour, and taking advantage of them generally makes more money over time.
Batteries are definitely not up to the job of moving energy across time spans of _months_ yet -- but nothing else is either. Whoever figures out a cost-effective way to store excess solar in summer, to discharge in winter, is going to make a trillion dollars.
I believe Erik’s statement that pumped hydro accounts for 90% of grid storage might be based on out of date data, as battery storage has grown dramatically over the last four years. A May 2024 article on “RenewEconomy” website titled “Battery storage is about to overtake global capacity of pumped hydro” indicates that on a global basis pumped hydro made up 90% of grid storage in 2020, but by 2023 it was down to 68% and expected to end 2024 at 56% and that battery storage would surpass pumped hydro in 2025.
For the US, EIA shows that utility-scale battery storage increased from 1 GW of capacity in 2019 to 20.7 GW in July 2024 (https://www.eia.gov/todayinenergy/detail.php?id=63025). Since pumped storage hydropower (PSH) capacity stayed approximately flat during that period at ~22GW, battery storage is gaining capacity share very rapidly.
However, it is worth noting that battery storage capacity in energy (GWh) rather than power (GW) units was 42.5 GWh at the end of 2023. For PSH, it was estimated to be 553 GWh (https://www.energy.gov/sites/prod/files/2021/01/f82/us-hydropower-market-report-full-2021.pdf). Comparing them in GWh brings home the point of the much longer storage duration of the typical PSH plant versus the typical battery.
Just want to add that China's largest pumped storage has gone into operation just last year with 3.6 GW installed capacity, whichi is also the largest of its kind in the world. And according to your disucssion, this one is a closed-looped pump storage https://www.waterpowermagazine.com/news/china-completes-worlds-largest-pumped-storage-hydropower-plant/
Great to hear there are companies working on this.
I wonder if funds could be made available for the purpose of remediation of Mountaintop removal coal mines, turning them into pumped hydro? Instead of fighting to hold some deadbeat accountable for their mess, try to turn it into a community asset.
I've heard tales of pumped hydro historically getting built in order to buffer inflexible nuclear.
Something I played with, is the Pumped Hydro Energy Storage Atlases from Australian National University: https://re100.eng.anu.edu.au/pumped_hydro_atlas/
Daily and weekly power storage is believably economic both from pumped storage and batteries, but I have yet to read of economically feasible seasonal storage that could account for the power-demand swings from summer to winter and back. What options are there? Iron-oxygen batteries sited in former automobile wrecking lots? Pumped storage with huge upstream lakes that include recreational vacation cabins?
A technology commonly cited for seasonal storage is hydrogen. While it's not very efficient to convert electricity to hydrogen and then back to electricity, this is less important for seasonal storage. (One way to make sense of this is that the hydrogen tank can be filled with cheap or free (curtailed) power over many days.) Iron-air batteries as you mention are another tech that target seasonal storage.
If you want to dig into the research, this is an outstanding source: https://www.goodreads.com/book/show/126223558-monetizing-energy-storage
I think on paper flow batteries with very large electrolyte tanks supposedly make sense for very long term storage, the problem is nobody has solved the right combinations of chemistry and manufacturing processes to actually make them cheap to build. (Like if you could magick one into being it would have cheap OpEx, but the CapEx is prohibitive.)
Heat pump geothermal does offer low cost inter-seasonal energy storage. Check out prior Volts interviews. It doesn't make electricity but it greatly reduces demand for energy for heating and cooling.
Yes, I remember that interview and fully agree that geothermal seasonal storage is an excellent option for heating and cooling of buildings and potentially industrial heat as well. We will need some options for electric power too, and those are not so clear to me.
Ehhh. Geothermal is great for how it offers a source of heat differential year round. I don’t know that I’d say it’s as effective for _storage_. It’s not like you can take excess solar in the summer, and pump superheated water down to heat up the earth, and then recover that heat months later. There are serious long term thermal storage ideas (like molten salt, and the box-o’-rocks approach where they line the box with IR-PV) which probably work on a time scale of days out to a week or two. I’m not sure whether any of those is plausible for season-to-season storage, though.
That is incorrect; such systems can and do give true inter-seasonal energy storage, where heat derived from summer cooling of buildings (and sometimes solar energy or waste heat or other sources adding heat on purpose) adds to the thermal mass, to be used as heat during winter. Larger, "busy" buildings with borehole heat pumps can add more heat over a full year to the underground thermal mass than they draw back for winter heating and often require some cooling towers to keep it balanced.
For cold climates the amount of energy for heating is large; reducing the need for direct energy use greatly reduces demands for energy.
Huh. How efficient is that? While thermal conductivity down there isn't that high, it's still not _zero_, I'd think over months, added heat would diffuse away into the background, and the immediately-accessible mass would move back closer to the ambient temp.
I am no expert in the thermodynamics but practical experience shows they do work very efficiently, superior in energy in vs energy out to almost any other options. A previous Volts interview featured thermal heating networks based on geothermal heat pumps that had COP"s (thermal efficiency) of above 5, that is 5 times more energy utilized than was used by the system. https://www.volts.wtf/p/thermal-energy-networks-are-the-next
My limited understanding is that whilst there will be losses at the peripheries earth and rock effectively act as an insulator over longer distances and the larger the heat mass the more efficient. With large fields of boreholes the volume and mass is much larger compared to the "surface area" (where leakage occurs) than for small thermal masses. As the volume around the working mass warms the rate of loss at the peripheries drop off and the system works more efficiently.
Yes I listened to that piece. But the reason these networks get high efficiency (units of energy expended, per unit of energy applied to the environment) is precisely that the earth stays at an extremely steady temperature regardless of season and regardless of how much energy you pump in or out (at least at the scale plausibly used for HVAC). These heat pump systems expend a small amount of energy, to _move_ a larger amount of energy. For this specific purpose, the fact that your "reservoir" can provide or soak up a nearly unlimited amount of energy is a feature. You can keep heating the homes and businesses on the network, by sucking heat out of the ground, and the ground doesn't get notably colder. And equally, you can dump heat into the ground during the summer, and the ground doesn't get notably warmer. I don't believe it's meaningfully the case that you're "recovering" the same heat in winter that you dumped in during the summer. The rock around the deep wells gives you an enormous thermal mass, and diffuses heat reasonably-efficiently over the course of days. If it didn't, it _wouldn't_ work as the heat sink / heat source for these HVAC systems. But the same property would seem to make it a poor heat sink / heat source for long-term storage.
If you run 100 GWh of energy through resistive heaters, apply that heat to water, and run that water down into the ground, you will at that time raise the local temperature some amount. But by six months later, I feel fairly certain, most of that energy will have diffused through the ground out to distances of _miles_, and will be effectively unrecoverable. I could be persuaded otherwise, but I'd want to see actual data. The thermal network piece you're talking about absolutely _does not_ talk about trying to store and recover significant amounts of energy.
There's also the separate technology of deep-geothermal (exploiting fracking techniques to get down deep enough that you can get geothermal heat without sitting on top of a shallow volcanic resource). But that too is reliant on the fact that the temperature at the depths you're drilling to is more-or-less constant, with heat diffusing from other parts of the deeper hotter crust, into the part you're extracting heat from, faster than you can extract that heat. If you tried to inject extra heat to store it, it would be lost to distant parts of that same layer of crust by six months later.
Unfortunately, the episode didn't mention the just-planned storage project in Ontario:
https://www.nationalobserver.com/2025/01/24/news/ontario-invests-2-billion-upgrade-hydro-stations
...which will run a full gigawatt for 11 hours!
It wouldn't perhaps harm to refer to long-standing successful examples outside of America, such as 'Electric Mountain' in Wales, UK, which has operated for almost 40 years. https://en.wikipedia.org/wiki/Dinorwig_Power_Station?wprov=sfla1
A key aspect is that the economics of these systems work best when demand is high and the requirement is immediate but cannot be exactly determined. For example, half-time during a popular sports event. Being able to spin up turbines to maximum load in 30 seconds is crucial.
It's also worth remembering this pumped hydro facility took 10 years to build and was at the time the UK's largest civil construction project. Battery tech may undergo further transformation to supersede pumped hydro over such an extended time period.
I would sure like to see the economics of storage systems expresses in Tonnes Avoided Carbon (TAC).
Every lithium battery plant that I know of (and I am ignorant, so please correct) is located right next to a a big power station: the Tesla facility east of Reno, the smoking ruin of the Moss Landing, the proposed Morro Bay factory.
These batteries are energy-intensive, and therefore carbon intensive. TAC is the better unit of measure for these computations than is Dollars, no matter how they are adjusted for inflation.
There's a great tool for this called CRANE. https://cranetool.org/. It lets you estimate exactly this for a pumped hydro facility!
Google "lithium batteries carbon footprint per kwh"
and get "The dependency of global lithium-ion battery emissions on production location and material sources" 2024
https://www.sciencedirect.com/science/article/pii/S0959652624011739
"Median cradle-to-gate carbon footprint of lithium-ion batteries
between 48 and 120 kg CO2e kWh−1."
Coal generation is almost 1 kg CO2e/kWh, gas around 0.5.
Split the difference and the battery embodies 90 kg/kwh, divided by 0.75 kg/kWh avoided fossil generation emissions, divided by 80% depth of discharge, seems to yield about 130 battery cycles for a battery to offset its embodied emissions if the cycled variable renewable generation would have otherwise been curtailed and fossil generation used instead. If the battery is cycled half the days, that seems like about 8 months emissions payback. Not too bad IMHO.
In any case for any storage, duty cycle or similar capacity factor is important to payback of emissions, energy, and $ invested. Kinda like if you drive your pickup 1000 miles/year, you shouldn't buy a Rivian to save the planet. Instead contribute to BEV buses for you school district or transit authority which are driven 40,000+ miles per year.
The existing PHES reservoirs in CO don't need plastic or concrete or asphalt linings. Many reservoirs don't. The lithium batteries I see are located at PV power plants. Even if they located at an old powerplant to take advantage of the grid, the ones in CA are clearly packing in the PV during mid day and releasing it to avoid evening fossil generation. CAISO has oodles of data on that.
As far as Moss Landing, seems like a good case for LFP instead of NMC, and hey Thomas Edison almost burned down the Vanderbilts' house. And did burn down a bunch of other stuff.
Thank you! There is a lot to digest here.
I am a simple-minded, or at least simplicity-minded, economist, who finds the simple path to survival to be found in treating terrestrial carbon, that would otherwise be in the atmosphere, like the ultimate scarcity good---the basis for a carbon economy in which *every* item is priced in Tonnes Avoided CO2 (TACs).
Our production cost models (I am an expert on the PROSYM family, which shows how old I am) we minimize Dollars or other nominal currency. It is not so hard to change terms to TACs. I suspect that LADWP is using their house interface to do stochastic structural modeling of the WECC, for example to build the business case for that big wind--and-transmission project they are doing with Anschultz
My claim to membership in the energy elder club is my proficiency at text-mode DOE-2 building energy modeling.
I'm kinda simplicity-minded myself. I'm a big fan of understanding when calculations result in "precision without accuracy," hence my two significant figures and wide input ranges for my little calc of battery GHG "payback."
For something that can be presented as numerically precise as TACs by switching from gas to electricity, or deploying VRE and storage, widely varying conclusions can be drawn from using different methods for looking at carbon emitted in future years by electric supply at any given location. Annual averages, short run marginal or average hourly, long term marginal or average hourly. All will give different answers, and depend on VRE deployment rate assumptions. Then, how, and how much, do we attribute methane leak & vent to gas use in either remaining powerplants or heating systems?
I think we would agree that as the time period in long-term energy storage for VREs gets longer, the number of cycles goes down and the embodied carbon and dollar cost goes up for each unit of stored energy, to the point some solutions may actually emit more hidden embodied carbon than they save.
I'm usually dubious (but not totally cynical) about forest or agriculture sequestration and offsets. But if we electrify most everything and generate only 10% of the that electricity with fossil fuels, using those bio GHG sinks and GHG avoidances to offset small remaining fossil GHGs might be "better" than Herculean efforts at long duration storage of VRE.
We need to talk forests and the carbon economy.
Offsets were (note tense) a successful industrial policy action, that gets things moving in the right direction.
In the carbon economy, the basic unit of value is TAC. Carbon firms maximize TAC, and all the values of conventional microeconomics can (and, in our crisis, must) be expressed in those terms. Like wages, ferinstance.
Go browse my Substack and ask me questions there. Or here...I don't want to poach readers from this great site,
"I have found my people" is something I have been looking forward to thinking. I think I may be right!
I want to know whether batteries are really better (in TAC terms) thatn the locally-produced biofuels for hauling the surplus biomass that forest recovery requires. From the perspective of the pathologically overgrown forest, which wants to sequester carbon not send it all up into the atmosphere at once, the biomass facility helps with its waste product; the challenge is to get the highest value work out of the waste biomass. The main value is in the forests (with all the birds and bees and nematodes riding along for free in the carbon economy).
Of course the concrete lining for the closed-loop reservoirs is also carbon-intensive. and just like everything else in the carbon economy must be priced accordingly.
Asphalatized concrete is a new term to me, but asphalt, in the grand, carbon accounting, is a carbon store, so the (durable) tar in the lining earns whatever value the carbon-economy delivers to carbon stores (such as, for example, the carbon capture systems they are bolting onto the Icelandic volcanoes).
The carbon economy is loaded with opportunities for innovation. The forests sector (which delivers liquidity and therefore facilitates credit) has a natural planning horizon that is measured in centuries, rather than the decades we are used to in the fossil economy. Long-horizon projects, like closed-loop pumped storage, are supported by the (base loaded generation of) the TAC-positive forests industries.
Do you have recommendations for references that compare alternate energy storage systems? Such a reference would evaluate costs (initial capital and operating), efficiency, and environmental risk factors that would affect applicability at a specific site.
Keep up the good work.
This is the first I've heard of the Goldendale project here in my home state! Most of the renewable energy headlines here have been about the Horse Heaven Wind Farm and all the legal disputes around it.
I get a kick out of the Goldendale project being able to produce 1.21 Gigawatts!
Just finished listening to this. The bit at the end where he describes 60 acres as not being all that much land for the amount of power / energy you get kind of made me boggle. Like yes rural land is cheap and you can tie into transmission to send the power to Seattle or wherever. But that’s still I want to say an order of magnitude more land than a comparable PV+storage site. Am I wrong? I could go look up the actual areas of sites Tesla has commissioned recently. I don’t have a strong gut sense of how big an acre is, but like Hornsdale is >100 MW and I’m not sure it’s much more than _one_ acre. You can walk across it in a matter of minutes. Moss Landing (currently shut down because the nearby Vistra system is on fire) is like 180 MW, and similarly you can walk across the whole thing in minutes. 60 acres sounds like it ought to take more like 30-40 minutes to walk across, if it’s a square.
I had exactly the opposite reaction. An acre is 43,560'sq or 209x209 feet which could be walked in seconds. The full 60 acres would be @ 7.9x7.9 acres or @ 1650 x1650 feet; a little more than a third of a mile or @ 5.5 football fields. Takes 15-20 minutes to walk a mile so the full acreage would take 5 -7 minutes to walk across. Wouldn't the closed loop system have reservoirs top and bottom and a generation facility? Figure 2-3 acres? so @ 14 acres per reservoir or 3.75 x 3.75 acres ea. or @ 784 feet per side? This coverage to supply the entire city of Seattle for 12 hours? Seems remarkably tiny.
Well, I went and looked with GMaps' measuring tool, and the main Hornsdale battery facility is about 400' by 150'. So I was right, it's not much more than a single acre.
https://maps.app.goo.gl/yuM6EChns1LuEa759
The quote in the text is "So yeah, 1200 megawatts, 12 hours of storage, two 60-acre ponds".
So about 5x as land intensive in terms of power per unit area. (Actually more than that -- the PowerPack 2 system at Hornsdale would be significantly less energy- and power- dense than the current-generation Megapack 2XL.) Less of a difference on energy -- Hornsdale is around 200 MWh on one acre... 1200 MW * 12 hrs / 120 acres, that's around 120 MWh / acre. Still less, but not by a huge multiple.
well again I'm not technically trained but just looking at what appears to be the site at Moss Landing ,CA on a Google maps aerial it appears to be as much as @ 68 acres total including the transmission line and etc. Compared with 60 acres mostly comprised of reservoirs full of water Vs. a lithium battery backup/storage site that burned and spread toxins over a miles wide, possibly thousands of acres swath of fantastic agricultural land. Externalized costs.
The site that burned was the Vistra system which was inside the old power plant, not the Tesla system that's on open ground. AFAIK there has never been a large-scale fire with a Tesla system. A _single_ Megapack burned at the Victoria Very Big Battery site. And at this point Tesla is converting to the Lithium Ferrophosphate chemistry for stationary batteries, because you don't need the additional energy and power density you get from the nickel-cobalt chemistries, when doing a stationary site. And while it is in theory possible to get an LFP to burn, it's pretty difficult.
The vast majority of the Moss Landing site is stuff that was already there long before the batteries, because of the gas plant. Also, the Tesla Moss Landing batteries are something like 1300 MW, so of course they take up more room than the 150 MW at Hornsdale.
As far as the area, unless you think the various transformers and transmission lines are going to be sitting on top of the water, that isn't going to be a difference between the two options, you'll still have that stuff sitting next to the water driven turbines, to convert the voltage and get it onto the grid.
That's an amazing solution. I didn't know anything about this, thanks for sharing.
You are probably aware of the Michigan plant in Ludington, but here is a link to it.
https://en.wikipedia.org/wiki/Ludington_Pumped_Storage_Power_Plant
I'm familiar with the Cabin Creek closed-loop system in Colorado (did a fatality investigation there with the CSB). As a storage system it seemed just about ideal. Modest footprint, significant capacity for peaking power in the Denver area. Where are the other closed-loop systems currently operating in the US?
It sounds like their Kentucky project might actually get built.
But I think you downplay the "environmental/progressive/NIMBY/tribal" opposition to the projects in the NW. I hadn't heard much about the Swan Lake, but a very quick Google Search shot back an Oregon Public Broadcasting post (https://www.opb.org/article/2024/04/09/klamath-falls-oregon-pumped-water-energy-storage-project/) calling it "controversial, and that it required a $40M payout to local tribes.
There have been many stories about the Goldendale PHES. Even though the upper and lower reservoirs are only 60 acres each or whatever. A recent example being
https://www.yakimaherald.com/opinion/guest-commentary-goldendale-energy-project-comes-at-the-expense-of-tribes-nature/article_e8c39852-4bd6-11ef-86b2-533cae58bd45.html
Gotta love, "An East Coast company backed by European investors..." Settler-colonialism all over again!
High Country News has a NW tribal reporter who has been all over it (and seemingly all solar or wind in SE WA), and I see "Columbia Riverkeeper" is quite wound up.
I sure like the combination of wind, water and solar but natural energy does disturb nature and its lowish density and location right on the planets surface means that disturbance is quite obvious. OK with me, but disturbing to many.
Good stuff. I struggle, however, to put numbers into perspective. Worse, what are they? In particular, when you say 1.2 gigawatts, is that peak output? Or shorthand for gigawatt-hours. Given this whole article is about storage capacity, I was surprised to not see a single reference (did I miss one?) to (x)watt-hours. You do talk about hours of capacity, so does that mean 1.2 gigawatts for 12 hours?
While I'm here, I would also like a short post (or link to one you may have done?) on what's considered a "typical" household. As in, "could power a city like Seattle" etc. According to Sense, I'm on the high side of their typical customer at 4 kw average use for the last 30 days. Electric cars, heat pumps, and a family that likes to keep the whole house warm. In summer, a quarter of that. 3MW-hour+/month winter, 1MW-hour/month summer.
Sorry for the rant, but I really try to understand the magnitude of things, and it just drives me nuts...
The units of power (watts) and energy (watt-hours) can certainly be confusing!
1 watt-hour is the energy produced by 1 watt for 1 hour. Prefixes killa, milla, giga are are of course 1,000 times more each. So yeah, if you say 1.2 gigawatts (GW) that's the power - and you'd assume it's the power at a peak steady-state. You'll sometimes hear "Rated Power" which means the expected normal steady power. If a pumped hydro facility has 2.4 GWh of energy with a rated power of 1.2GW, you'd expect it to run for two hours before it's empty. (1.2 GW x 2 hours = 2.4 GWh). As for the "typical household" - it's just taking an average.
We often talk about batteries as a power and duration. So, a 2 MW, four-hour system would store 8 MWh. (Tesla’s products generally come in a 4hr, 2hr, and sometimes a power-boost / peak-power version that’s something like 1.7hr, I forget the exact number.)
Right, so if they just say 1.2 gigawatt, what are they talking about?
Well, if they're remotely competent, they're just quoting the power (which is often what matters in relation to substituting for something like a gas peaker plant).
The thing that's a bit confusing about the use of numbers of hours in this interview is that there's a bit of slippage between the duration _for which energy can be stored_ (without significant loss of energy, and without driving the levelized cost of the system too high because the investment is getting amortized across fewer events), and the "duration of the battery" in the sense of how long it will run at its nameplate output, to go from End of Charge (or 100% State of Energy) to End of Discharge (0% State of Energy).
And actually, while I think it's correct that if you want to time-shift energy use by a day or more, pumped storage is probably going to be more efficient, lithium-ion batteries certainly _can_ store energy on the scale of a few days. Depending on the details of your system, the idle losses could be as low as 2% over a week if you're being as conservative as possible (putting the system into a dormant mode until you really need it), up to as high as something like 10% if your system has high idle loads because it's keeping the bus up and the AC system prepped for instant demand, and it's dealing with thermal loads because of very hot or cold weather. And then you'd lose perhaps 10% to the general entropic issues that are usually described in "round trip efficiency" estimates for batteries. (You also would've lost that if you were just continuously cycling, starting to discharge the moment the battery got full).
But, like, if you can take X amount of energy on a high production day, and get back 80% of that on a low day, that's not peanuts. Pumped storage can beat it, but it's not available everywhere. The main reason battery sites typically don't do that is just that there are opportunities for energy arbitrage every day, hour to hour, and taking advantage of them generally makes more money over time.
Batteries are definitely not up to the job of moving energy across time spans of _months_ yet -- but nothing else is either. Whoever figures out a cost-effective way to store excess solar in summer, to discharge in winter, is going to make a trillion dollars.
Good article. I see a real future with this enabling more renewable penetration.