Interesting concept. At $150,000,000 for 500MwH = $300/KwH of energy storage. Tesla's Megapack solution is $300/KwH also, giving this a good running chance if they can get costs down with scale.
I like that it can potentially run for decades without chemical degradation, and that it is environmentally friendly since most of the parts (compressors, dryers, generators, tanks) can be recycled down the line with minimal environmental impact.
Things that seem concerning at first glance:
* Maintaining sub -196'C requires a lot of complex refrigeration, monitoring and pressurization equipment, which can be prone to failure (see Hampson–Linde cycle). The expansion factor of liquid air to air is roughly 800x.
* Round-trip energy efficiency: I can't see a system taking electrical input to produce liquid air on the way in and similarly to run a turbine generator on the other end without having substantial losses both ways... I believe Tesla is around 88% on their system.
* Gasses used in refrigeration continue to be an environmental and health concern even with modern compounds.
Hopefully with some cost scaling or proof of longevity it can be a valid solution.
The linde cycle is not used industrially, they are using a reverse brayton cycle. This cycle use air as the refrigerant, and is probably the most widely distributed industrial cycle in the world (because liquified gases do not transport easily).
Obviously, besides batteries, you need other components to transmit, convert, etc electricity, yet the difference is quite large.
I guess question becomes - how much it will cost to maintain such cryogenic storage solution vs replacing lead acid batteries over time.
EDIT: while shopping for a used UPS unit for my server rack, with expectation that the unit itself will come without batteries and I will purchase new batteries separately, I was getting about $125-$150 per kWh for everything, including UPS itself and a set of new batteries.
I am not proposing to use used tech to build such a storage, just interesting to look at prices.
Lead acid has a very poor cycle life and poor round trip efficiency as well. Also, its advertised capacity isn't usually reachable even at very low C rates.
biased/sponsored test but still casually informative test of actual lead-acid performance: https://youtu.be/iy3hga_P5YY
Besides inefficient chemistry remark above, car batteries classified as high-discharge as opposed to deep-cycle. We need the latter for energy storage. If you'd like to assess the cost of lead-acid, check out «marine» batteries instead.
Using Tesla's own numbers for a turnkey installation such as this. Costs will probably come down another $50-80 per kWh with improvements in battery costs.
As for the price per kWh, I guess this system scales sublinearly (as opposed to more or less linearly with LiIon). They can probably increase the number of tanks storing oxygen in the future for way less than $300kWh (assuming they keep the same system wattage).
If you're keeping the same power input/output (50MW in this case) then you can add more tanks and scale out capacity with cost savings on storage, generation or transformation. "A bigger gas tank doesn't make the car go faster."
Scaling it out proportionally would only have economies of bulk purchase and site engineering.
I came to precisely the opposite conclusion. Batteries keep getting cheaper every year, while air pressurization technology does not (that is a very mature technology). So if this air pressurization is only the equal of batteries in year one, then it's already lost.
The point is not about future price evolution, but the marginal price of storing 1h more, if you just need another tank it will be cheaper than having the same energy in a battery, with the same system to store and destore the liquid air. Li-ion is not a good fit for long term storage.
But if you’re only inputting and extracting at 50MW then having more than 600MWh of storage means it takes longer than 24h to fully charge and discharge. If anything 500MWh of storage is already well into diminishing returns as daily storage makes money every day but long term storage has less frequent returns.
If every (normal) day you discharge less than you charge, you can build up a reserve for unusual events. When you get close to your storage limit, you can sell power from out of that reserve during periods of peak usage/price, when your solar array is running at capacity.
Storing for say 1 month means you get 12 payouts per year, storing it for 12 hours means you get 365 payouts per year. If your on average getting say 1c/kWh from the distance in values prices every day then your monthly sale needs to be at 30c/kWh to be worth it. Prices simply don’t rise that high every month making long term storage less valuable than daily storage.
On top of this these things aren’t 100% efficient. Buying at 7c/kWh and selling at 8c/kWh is losing money if these are say 80% efficient. Which means you’re never going to profitably buy and sell throughout the day. Thus the 500MWh capacity will mean some long term storage.
If we are going to go full-renewable we will need storage that works over night, storage for cloudy days and for days without wind. A low wind situation can last a week for example.
So we do need long-term storage and we will have to finance it so that when we need it, we can count on it
First Hydro is already effectively long term storage. It’s fixed across a year, but releasing 5x as much as normal is fine as long as releases are scaled back to account for the discharge.
Next solar and wind don’t correlate and solar output never hits zero on cloudy days. So, excess capacity always directly reduces the need for storage. Working out the ideal amount of excess capacity isn’t trivial, but unless storage costs tank dramatically something like 85+% on worst case days is likely viable. Which could then easily be covered by hydro.
Sometimes hydro is inconveniently on the other side of two national borders, as here, where Paraguay is on the other side of Argentina. There are good reasons to stay on pleasant terms with Argentina and Paraguay, but avoiding blackouts is among the less-good reasons.
While it’s natural to optimize for a single connection, having redundancy also avoids this issue. Paraguay > Argentina > Chile and Paraguay > Bolivia > Chile are both viable options. Countries should have strong grid conditions to every country on their border as it provided both countries redundancy at very low cost. As a side effect all of a North and South America should end up being able to share power with each other, as should all of AfroEurasia.
A large issues is islands, but plenty of people setup off grid houses completely dependent on Solar. It’s simply generally more expensive.
If you knew anything about Bolivian history you would not suggest Paraguay > Bolivia > Chile as a viable option. In any case, both routes depend wholly on the agreeableness of Paraguay.
I am very glad you are not responsible for planning for the availability of power I depend on.
I am saying it’s viable not dependable. Chile shares borders with 3 countries, each of those countries shares borders with another country. So no 2 counties could block Chile.
Now, if your saying Chile can’t trust it’s maintaining good relationships with any of those 3 counties then sure.
The only expense for excess storage, above capital cost of the tankage, is refrigeration to compensate for heat leakage through the insulation.
All the tradeoffs, together, are an optimization problem that is not hard to solve for conditions found. Since the ultimate source of power, solar, is the cheapest ever devised, they start with a huge advantage. It would be foolish to bet against them.
Sure, but it’s an optimization problem up against simply building more solar. That’s the part of the equation most armchair experts are ignoring. Excess solar production is basically guaranteed due to how cheap it is. It’s likely that some long term storage is optional, but at scale it’s likely to be relatively small percentage.
More solar generates power only when the sun is shining.
This apparatus is for use mainly when the sun is not shining, or when it is not shining enough. Sometimes the sun barely penetrates the clouds, and then you need a boost all day. Maybe it would be cheaper to import power from a dam in Paraguay than to keep your tanks full and cold until you need it. But it might be cheaper to buy power only if they know you don't really need to.
Energy economics are more complicated than we can safely generalize about here. The safe bet is that somebody spending $150M has thought it through.
Sure but note they didn’t get more than 10 hours of storage from this system. It’s designed to have close to zero long term storage like every single grid backup system currently in production. Something like 20% reserve capacity is completely rational, but you get there by having redundant daily storage so having one or more of them going online doesn’t break the grid.
Batteries should see drastic declines as well. $300 is already considered a steep price even though it is the all-in price. We are talking about $50 at the cell level for LFP chemistry soon.
And the Tesla Dalhousie guy says the LFP batteries they tested basically have unlimited recharge cycles. Allegedly 200 wh/kg was supposed to be commercialized by CATL for LFP this year, but I haven't heard anything about it so I assume either COVID has delayed it or they were just being too optimistic.
The other grid storage I wonder about was that Army salt water battery that Jeff Dahn said fairly effusive things about a few years ago. I have heard nothing of it since, so I assume it had cycle problems or other fundamental issues.
* Round-trip energy efficiency: I can't see a system taking electrical input to produce liquid air on the way in and similarly to run a turbine generator on the other end without having substantial losses both ways... I believe Tesla is around 88% on their system.
Well you can just power an engine on the temperature gradient. That's going to incur some losses though. This seems to be the downside of storing energy cooling stuff, you have to fight the second law of thermodynamics both ways.
You get to run both a heat engine, on the temperature difference, and a turbine, on the resulting boiled-off gas. So, you get benefit of thermodynamics both ways. Furthermore, the heat engine uses heat pumped out and saved from liquifying the gases.
It is typically a mistake to assume people behind an innovative development, that they are spending $150M on, are idiots.
In the end you're still powering something by letting something cold grow warm, so I'm not seeing how you could avoid the second law of thermodynamics.
I suppose that in theory the cooling step would be where you could recoup your losses a bit by using an efficient heat pump, but getting the coefficient of performance greater than 1 could be pretty difficult for such extreme temperature differences.
Turbochargers run in open-cycle engines that are not bound by Carnot limits. Diesels run on the Rankine cycle, and gasoline engines run on the (similar) Otto cycle. Both of those take in cold air and exhaust hot air, so can be more efficient than an engine that runs a fixed supply of working fluid around the Carnot cycle.
Turbo- and super-charging enables an engine with a given displacement to take in more air, and thus can burn more fuel, than one that must fill cylinders with no more than ambient air pressure; and can thereby deliver more power. It doesn't change the engine's thermodynamic profile.
What do they use for the refrigerant? Refrigerant leaks have some of the highest green house gas potentials of all compounds. Depending on the product > 2000 times global warming potential of CO2.
I would imagine a facility like this would have much better monitoring for refrigerant pressure drops than a typical consumer-grade system, plus a staff onsite that would be able to close valves and rapidly find and repair leaks.
No matter how fast they notice it, a leak will have a huge impact even if just for a second.
Bigger system means bigger pipes or faster transport, either will result in more loss than what a whole consumer grade system can possibly contain.
But like someone said, they most likely wont be using any consumer grade gasses. Big systems use gas that is more dangerous for humans nearby, but has no impact on the ozon.
Despite your disclaimer, I do feel this is too glib. What is the takeaway? Yes, a bad thing happened one time at that facility (and others, at other times). That doesn't have much bearing on what I said. It is still reasonable to suspect that refrigerant leaks will not be a major problem at this facility.
I am saying that in general I would prefer a system that’s safe by design, not by the fact that there are going to be people there who can make very human mistakes and create a problem. It is all relative: is it safer to run a facility like this or to transport a tanker-full of crude oil across waterways? But just saying that well there will be people there to open a valve when bad things happen is naive.
When comparing the risk of refrigerant leaks in cooling systems in general vs heat exchange systems at a highly technical and actively monitored facility, it is relevant that the facility is actively monitored and probably has much better controls than typical refrigerant-using installations. It's not naïve to observe that most refrigerant leaks are likely from systems that are not monitored (i.e. almost all residential systems and probably most commercial ones).
I would like to remind everyone that nearly every single industrial process, used to manufacture common, everyday items necessary for civilization, at one point or another, handles materials, fluids, or gases at very high pressure.
If 'gases under pressure' is a deal breaker for you, I hope you don't use gasoline, or buy any products that are shipped by gas-powered, or diesel-powered vehicles.
The fact that you can name all the major incidents is a testament of the safety of the technology (it really is — see deaths per TWh of various technologies).
Fukushima was hit by an earthquake and tsunami. Most deaths attributed to the Fukushima disaster were people who froze to death because they couldn't afford heating after electricity price hike caused by Fukushima's shutdown.
Keep in mind that nuclear plants have been operating since these incidents all over the world, without further problems. There are over 440 reactors are running right now. Even the Chernobyl power plant kept operating the remaining three reactors until the year 2000.
And even they they only, to my knowledge, shut down the remaining three reactors in exchange for funding from the international community for the New Safe Containment sarcophagus. Nine other similar RBMK reactors have continued to operate (albeit with some safety retrofits) since then without incident.
>> Most deaths attributed to the Fukushima disaster were people who froze to death because they couldn't afford heating after electricity price hike caused by Fukushima's shutdown.
Do you have a source for this? How many froze, and of those which were for economic reasons?
I was in Japan at the time, and the assertion doesn't look right, unless it's comparing tiny numbers: e.g. zero nuclear deaths and two frozen through economics, technically means "most deaths" were from freezing.
Yes, both are small numbers of people. I think it's a very illustrative effect — how few people were directly affected by the disaster, and how second-order effects affecting whole populations are magnified. Similarly how fear of flying after 9/11 caused people to take less-safe-per-mile car instead.
Their models predict excess deaths from economically driven cold which compare marginally unfavourably with deaths from nuclear related dislocation.
Even if the model holds correct, the government's policy to shutdown reactors immediately was likely the right choice. Just in terms of number of expected deaths.
Two factors lead me to think so: aftershock risk and the possibility of systemic failure in the nuclear industry.
To split hairs, Singapore is a one-party police state, but it's not totalitarian.
A totalitarian state directs all of its population's efforts in pursuit of state goals. The USSR during WWII (Or during the civil war, or during the years in between the civil war and WWII) would strongly qualify - and between Stalin's death, and Glasnost, would semi-strongly qualify. North Korea would strongly qualify. Singapore, modern Russia, or modern-day China would not, because most activity that takes place in them is undirected market economics. They have a few directed economic areas, deemed of import to the state, but for the majority of activity taking place in them, the state is not involved significantly closer than it is in western countries.
I mean… you can recycle batteries, too. And the environmental “issue” with batteries is largely a red herring (steel mining and even recycling would look bad if you had a campaign of misinformation against it like batteries do, cherry-picking dramatic industrial-scale images without comparison with what it is replacing).
$300/kWh isn’t anything to really write home about.
Not opposed to it, though. While I don’t think it’s a slam-dunk win against batteries by any stretch, it’s good enough not to be a waste of money if someone builds one (which is probably true for that cement-block gravity energy storage thing).
Yeah, and judging by the prices on ebay, the "market rate" for some batteries is pretty high. It might not stay that way when the higher volume electric cars start to age out.
This press release doesn't say anything about system efficiency. If it was high they'd be promoting that fact. Since it involves refrigeration it's probably quite low. That makes me wonder why they went with this system.
> That makes me wonder why they went with this system.
Cost, presumably. The stated figures would give $300/kWh, which seems to be about where the all-in cost for lithium ion facilities lands, but this would likely have a longer lifespan.
Offhand impression. With thermal storage systems. Round trip efficiency numbers range from really bad at 50%. To really good at 80%. Batteries are probably 80-90 percent efficient.
Notably to me is that cost per mega watt hour varies a lot more than two to one. Which says to me energy storage is economically feasible using a number of different technlogies.
I mean, yes. In general, the problem these systems are trying to solve is "ugh, we have too much electricity right now, why can't we use it later" (in particular, areas with a lot of wind generation often have to stop the turbines spinning). At time of storage, the energy is essentially free (or of negative value, really; in the absence of storage the excess energy is a _problem_) so efficiency of storage is much less important than facility cost per MW.
Yes, I’ve wondered what the issue is with doing things like this, I mean surely there’s a way to do this with molten salt blocks too. What is the efficiency on this and how easy is it to maintain!?
"Liquid Air Energy Storage systems have the potential to be a competitive local and grid scale energy storage technology. They also have the potential to facilitate the penetration of renewable energy technologies. However, there is a clear disconnect between what has been proven in literature, and what has been demonstrated in practice."
I'm very concerned with the credibility of that source based on this alone. Unpublished and/or future-dated articles are to be taken with a pile of salt.
What you are seeing is the online preview, August 2021 is when it appears in print.
This kind of publishing is becoming more and more common, most likely to boost the journals impact factor (average number of citation within 2 years of publication). It's sketchy but doesn't say much about the article, only the journal.
Nothing sketchy about the journal either. The next issue may come out in August due to the publication schedule and article queue. But there's no reason to delay availability of the article after it's addressed all the reviewer and editorial concerns.
I'd think one major issue at this scale would be heat generated when pressurizing, and heat lost (= freezing) when depressurizing, the latter which could cause weakness and damage in any materials affected. But, I'm not a scientist or anything, take this armchair take with a grain of salt - there's other places where they do a lot with compression/decompression at scale, e.g. natural gas storage and transport.
Molten salt is going to be way more efficient on a round-trip basis. According to wikipedia, its something like 70% for molten salt, and 25% for cryogenic.
There are likely other engineering factors involved, but those efficiency figures are pretty damning on the surface.
Also from Wikipedia, the efficiency if 25% if you let the heat of liquefaction and the "cold" of vaporization go to waste, but it can be improved to 70% if you store that heat and re-inject it in the cycle.
Metal tanks. Liquified air is pretty compact. Probably, as the tech matures they will move to burying them, and the heat reservoirs too. For a pilot plant, keeping everything exposed and maintainable is important.
As they get bigger, storing the heat removed in molten salt might prove useful.
The tradeoff of excellent underground insulation vs. accessibility for maintenance is tricky. For the case of molten salt, expectation of corrosion complicates it further.
With any luck, plummeting aerogel insulation cost will make the question moot.
I've pumped scuba tanks and other light pressure stuff - in those applications a fair bit of energy is lost to waste heat. Curious how this is handled in terms of efficiency
The article is a bit light on details, but I'd guess that both heat and cold generation (due to compression and decompression of the air) could be used for a district cooling system.
I like that it can potentially run for decades without chemical degradation, and that it is environmentally friendly since most of the parts (compressors, dryers, generators, tanks) can be recycled down the line with minimal environmental impact.
Things that seem concerning at first glance:
* Maintaining sub -196'C requires a lot of complex refrigeration, monitoring and pressurization equipment, which can be prone to failure (see Hampson–Linde cycle). The expansion factor of liquid air to air is roughly 800x.
* Round-trip energy efficiency: I can't see a system taking electrical input to produce liquid air on the way in and similarly to run a turbine generator on the other end without having substantial losses both ways... I believe Tesla is around 88% on their system.
* Gasses used in refrigeration continue to be an environmental and health concern even with modern compounds.
Hopefully with some cost scaling or proof of longevity it can be a valid solution.
1. https://en.wikipedia.org/wiki/Hampson%E2%80%93Linde_cycle