Hacker News Reader

Comment by kragen | Hacker News Reader

it sounds like a reasonable enough measure, but even in its devil-may-care heyday, nuclear energy never undercut coal's pricing by much. now that solar energy is far cheaper than coal ever was, this seems like they're trying to fight last century's war: a bayonet charge into a machine gun nest

basically a nuclear power plant is a coal power plant which replaces a coal fire with the nuclear island. you can describe it most charitably as a coal power plant where the fuel is free and doesn't make carbon dioxide or pollute the air. the problem is that coal power plants would already be too expensive to compete with solar on a pure capex basis, even if the coal and other operational expenses were free

maybe if someone finds a way to make commercial nuclear power that doesn't involve driving an electromechanical generator with a steam turbine, nuclear could win. and obviously undersea and in parts of alaska nuclear is a better option

(i wrote a comment earlier tonight on the recent history of solar energy costs https://news.ycombinator.com/item?id=40723188 and its parent has a link to an excellent slide deck by fraunhofer ise https://www.ise.fraunhofer.de/content/dam/ise/de/documents/p... providing a lot more detail)

View on HN

If we're aspiring to 1970s standards then sure. Nuclear power isn't about achieving the lofty heights of the 1970s. I'm just going to wave the famous article by Crawford on the economics of nuclear around: https://rootsofprogress.org/devanney-on-the-nuclear-flop [0]

What you can see there is a very impressive learning curve that was well on the way to very cheap electricity. Then it suddenly and sharply inverts in a shape that defies common sense. Then no-one else really builds any.

"Supporting nuclear" isn't supposed to mean splitting the atom at any cost. The point is if we would just stop regulating the industry into oblivion and let costs drop then it'd be a perfectly competitive form of power. We have this lobby that goes in, wrecks the economics of nuclear then uses the wreck they created as justification for why we can't build any more. And in some senses that is fair because if they had the power to destroy past nuclear projects they can do it again. But lets not pretend that these costs are fundamental to nuclear.

Keeping a weather eye on China, if they do their usual trick and start learning how to build nuclear cheaply they could leapfrog everyone else on energy tech in another industry. On the one hand that'd be fantastic to the point where I'd love it. On the other it is so frustrating to be relying on nominal communists to push the state of the art forward on energy technology after listening to my western compatriots whinge about exactly that topic for decades while blocking the obvious solutions. We could have built the future in the west, we never had to have a do-it-in-China policy.

[0] If you read the article, he's paraphrasing some book by someone called Devanney. I'm going to gamble on nobody reading the article so I can get that wrong without being corrected.

it would be a perfectly competitive form of power by 01970s standards, but naval nuclear reactors, chinese nuclear reactors, and russian nuclear reactors are not regulated into oblivion and have not had their economics wrecked by lobbying. nevertheless, they remain more expensive and less economical than even fossil-fuel plants in both warships and china. conceivably capitalism could solve that problem (us defense contractors and chinese power plant builders are pretty tightly state-linked rather than free-market actors) but we don't have strong reasons for believing so

crawford here talks optimistically about building nuclear plants for 250¢ per (nameplate) watt. europe and india are already building solar farms for 60¢ per (peak, nameplate) watt, a cost that is dropping dramatically year by year. he does plot some plants being built that cheaply in the 01960s in the us, in inflation-adjusted 02010 dollars, but for some reason he doesn't comment on that, and he doesn't find anyone able to do anything like that in recent years; the 'building cheaply' he cites in south korea and india is 190¢–250¢ per watt. this suggests that perhaps either the inflation adjustment or the original cost figures are incorrect

solar does of course have a lower capacity factor than nuclear, but there are numerous countries where it's 30% or better, so 60¢ per peak watt is still less than 200¢ per long-term average watt. also 60¢ today is 41¢ in the 02010 dollars crawford is using: https://data.bls.gov/cgi-bin/cpicalc.pl?cost1=60&year1=20240...

i sympathize with crawford's leanings but even the most generous reading of his claims doesn't support the contention that nuclear can be built cheaply enough to compete with solar

None of those watts are dispatchable though. A solar plant can't look ahead a month and bid a block of kWh on any given day with any confidence.

Solar and Wind in "price per watt" are meaningless when storage is not accounted for. In fact a huge chunk of their actual cost of power has to be buying contracts for the backup plants to provide the kWh they want to promise but may not be able to deliver (hence the true cost of a kWh from a solar plant is basically whatever it would cost a gas plant to produce it).

you'd think it would turn out to be a huge chunk, but it turns out to be relatively marginal. also power producers bid on day-ahead markets generally, not month-ahead markets

The quality isn't the same, but price is its own reward. We'll learn how to shift a lot of industrial production around if the price goes low enough.

Watt-for-watt dispatchable is superior, but for half price there are probably a lot of uses that will turn out to be quite flexible. The energy markets don't appear to have quite gotten to the point where they handle that flexibility (prices keep dipping negative, which is unfortunate if you understand what that implies), but it is reasonable to expect that it is coming.

yeah, negative lmps with solar is 100% perverse bureaucratic incentives. solar panels aren't like coal plants (or conventional nuclear) where they overheat if you stop drawing power from them, and take hours to heat back up if you ramp down their burn rate; they're perfectly happy to be left in the sun either open-circuited or (more safely) short-circuited, and can return to supplying grid power in literally nanoseconds

>We'll learn how to shift a lot of industrial production around if the price goes low enough.

Are you going to send everyone home on overcast days? What if production needs to run 24/7? Solar is a whole solution if and only if storage is solved on a massive scale. Even with adequate storage, it still isn't going to work in some parts of the world or during certain seasons.

In the end, I think we need a mix of energy sources and working to make nuclear cost effective is part of a practical low carbon energy future. I just don't understand the confrontational nature of the solar vs. nuclear argument. Solar is absolutely going to be a huge part of our energy future. Supplementation by cost-effective nuclear would be a great complement to solar. Nuclear costs might not ever get low enough, but we'll never know if we don't put serious effort behind the goal.

yes, i completely agree; i'm skeptical that nuclear will make it in the next decades, but certainly it's worth continued research

Just a FYI that article you linked is written by someone that doesn’t actually understand what’s going on they even say: “I‘m a little fuzzy on the economic model, but” And then say things that aren’t true.

The baseline cost of fuel rods alone ends up being ~0.9c/kWh. That already significantly eats into the margins before you consider the ~500 person workforce required to maintain such a complex system 24/7 365. The physical degradation of equipment like pumps, turbines, and lightbulbs costs yet more money. Decommissioning again yet more costs.

People focus on the upfront construction costs because they are extreme, but even if construction costs were the same as a coal power plant nuclear electricity would still be quite expensive. Further, many things like the initial fuel supply get lumped under ‘construction costs’ but are better thought of as something else like operating expenses.

It is an argument based mainly on learning rates. He doesn't need to know any details about the economic model of nuclear; the observations are based on the economic model changing over time.

> The baseline cost of fuel rods alone ends up being ~0.9c/kWh.

I saw a quote from the 4th generation reactor wiki page (it came up while people were talking about China getting started on the build out of next gen of nuclear tech) that mentioned "100–300x energy yield from the same amount of nuclear fuel" as a potential advantage. People will figure out how to make fuel rods cheaper. We haven't really pushed the limits of this stuff.

Besides, even in the current climate 1c/kWh is still a great lower bound to aspire for. Possibly even that would be worth pushing through the political blowback for.

Gen IV doesn’t mean anything specifically. No individual gen IV designs has every single advantage you see in any of those designs. You need to look at a specific designs in isolation as well as what trade offs are involved in that design.

“100-300x energy yield from the same amount of nuclear fuel”

That’s one of those things that sounds good in isolation, but ends up being more expensive. It’s just like CANDU reactors don’t need enrichment and can operate on natural uranium, but it turns out fuel costs drop by 20% when they started enriching the fuel.

Similarly LWR/HWR could make better use of their fuel if they could operate at higher temperatures, but that causes a world of other problems.

Unfortunately nuclear engineers aren’t dumb, they’re doing things for surprisingly solid reasons even if counter intuitive reasons.

i'm not convinced that the fuel is inherently that expensive; plausibly a lot of that cost is unnecessary bureaucracy and antiproliferation costs. i think the killer is your point, 'even if construction costs were the same as a coal power plant'

Again when you dig into things some things are inherently extremely expensive like operating a lot of gas diffusion centrifuges.

You could cut fuel use in half if the thermal efficiency of nuclear increased to that of a combined cycle coal power plant. But having two water loops (or a liquid salt then water etc) inherently reduces the temperatures the second loop reaches compared to the primary loop. On top of this you want to stay well away from material limits for safety concerns. Net result the maximum Carnot efficiency is only like 45% and achieved efficiency is significantly lower than that.

gas diffusion centrifuges are an excellent example; possibly, for example, silex will turn out to be orders of magnitude cheaper. the absolute thermodynamic limit for isotope separation is many orders of magnitude away from what is currently practiced. and of course if you run your reactors on thorium or plutonium you don't need isotope separation

interesting, i hadn't heard of igcc before. thanks!

Yea, I don’t think things are hopeless here. What you mention might drop fuel costs by as much as 0.3c/kWh which would be a major step forward, but nuclear needs several such wins and it needs them quickly.

People aren’t going to ramp up construction until the economics change significantly and then you’re looking 6+ years out before anything comes online. Meanwhile investors are thinking what will renewables look like in 6-60 years from now? which makes such projects hard to justify without massive subsidies or at minimum locked in pricing for decades.

60 years is way beyond the prediction threshold. in https://dernocua.github.io/notes/backward-exponential.html i calculated that we can produce 16 earth-surfaces of human living space in less than 30 years, using 0.3% of the mass of the moon

60 years ago was 01962, when moore's law hadn't yet been noted, most people were farmers, there was no energy crisis, global warming was a forgotten 19th-century speculation, nuclear energy was hoped to make electricity too cheap to meter, communism was widely believed to be more efficient than capitalism (and produce higher living standards, if you discounted the gulag), mass imprisonment was a defining characteristic of russia rather than the usa, global thermonuclear war was widely considered survivable, overpopulation was driving inevitably toward widespread famine because the green revolution hadn't happened yet, and electric generation capacity in the usa was starting its last pre-energy-crisis doubling

The world doesn’t actually change that quickly. A great deal of energy infrastructure designed or even built well before 1962 is still in use. The hover dam for example was constructed between 1931 and 1936 for ~800 million in today’s money. It’s paid that off over time, but time horizons get really long. People had been evaluating that site for ~30 years before construction began and the project was authorized all the way back in 1928.

With nuclear you’re essentially betting that the plant will have an operating surplus long enough to both pay down the initial construction costs plus interest and be able to set money aside for decommissioning. If you predict the nuclear plant will be shut down 30 years after construction finishes that’s a deal killer. Further while people talk about construction timelines but projects start well before construction begins because you need to convince someone to pay for it. So you’re essentially looking out 40 years before the possibility of profit exists.

the vast majority of energy infrastructure in china was built in the last 20 years. actually that was true in the usa in 01962 and 01972 as well. the world is changing much faster now than it was in 01936, when all we had to deal with was the rise of fascism, the end of the millennia-long gold standard, history's deepest recession, communist revolutions around the world, radical overhaul of the economic system in the united states, and the rapid adoption of automobiles and electricity. you probably think i'm being sarcastic but i'm not

consider that 20 years ago photovoltaic was too expensive to be a viable alternative to nuclear and fossil fuels, youtube didn't exist (and your isp would often shut down your account if you posted a video on your web page), russia and china were friendly to the usa, friendster and orkut were the hot social networks (because you had to be at harvard to get a facebook account), javascript was too slow for most apps so demanding web apps used flash, gmail was only available inside google (so email was still really distributed), most people in the usa didn't have cellphones, cameraphones were new, the iphone hadn't come out, consequently most people who owned computers had root, orrin hatch was trying to ban the ipod, fast company demanded that you fax them a permission form before linking to their website, most people thought the nsa's worldwide dragnet surveillance was a myth, bitcoin didn't exist and was considered probably impossible, ycombinator didn't exist, neural networks were an interesting failed approach in ai history books, it was the year of desktop linux for the seventh year in a row (but there was no android), the only global pandemic in living memory was aids, the leading semiconductor fabs were in the usa and israel rather than taiwan, one laptop per child hadn't yet made it credible that internet-access computers would be cheap enough even for children in poor countries (who now mostly have one in their pockets), nasa still monopolized crewed spaceflight in the usa (if we don't count spaceshipone; carmack's x-prize competitor went 131 feet high), microphones and cameras in most people's homes reporting to secretive overseas data centers were dystopian fiction, the geforce 6800 yielded 15.6 gigaflops, and led illumination was new.

> the world is changing much faster now than it was in 01936

Hardly.

In most ways things were advancing much more quickly in 1936. Look at say medicine, cars, aircraft, physics, nuclear power, even computers and communication systems, and if anything things are currently slower right now than back then. In 1905 the fastest aircraft did 30 MPH and had troubles traveling 30 miles in bad weather, by 1965 we could fly ~1,000x as fast and that’s ignoring interplanetary probes etc. 1935 sat in the middle of that transition but fly somewhere today and there’s good odds you’ll be in a 30+ year old airplane with some new paint an arguably upgraded interior.

Walk into a Walmart today and ~99% of the stuff sold had direct equivalents 10 years ago. Most of it is minor improvements on stuff available 50 years ago. That really wasn’t the case in 1935.

it is true that if transportation speed is your metric, things are advancing much more slowly; even considering 01876 to 01936 physical movement sped up enormously, and it was widely felt that speed was the defining feature of the age. and progress in aviation, as in many other fields, basically stopped 50 years ago

but i'd say that the differences in computers and nuclear power up to 01936 were all zero since neither of them existed. and i'm not sure you're right about physics and medicine either

unfortunately i can't walk into a walmart (they pulled out of my country, which is in an economic crisis worse than any it's seen since, coincidentally for this discussion, the great depression) but i have two rebuttals here:

1. walmart sells starlink, alexa (echo), airpods, impossible burgers, apple carplay, cheap solar panels, and ring cameras, none of which existed ten years ago, and i think doesn't sell divx anymore. they sold cassette tapes in 02014 and don't anymore. they decided the live lobster tanks were inhumane and took those out too. a lot of locations also stopped selling cigarettes. and they no longer sell fish. or ar-15 and similar rifles. also 02014 was the last year you could buy polaroid film anywhere (though maybe not in walmart)

also, there are lots of products which were hobbyist or luxury niches 10 years ago and are now on sale at walmart. 3-d printers and google chromecast come to mind. oled tvs might be another example. you could argue that ring cameras are really just a webcam, but webcams you bought ten years ago weren't remotely controlled by a surveillance company

2. if you walked into a woolworths in 01936, 99% of the stuff sold there had direct equivalents in 01926. i don't have a woolworths catalog from 01936; the closest thing to hand is https://archive.org/details/1937-sears-christmas-wishbook-ca.... selecting 5 random pages (12, 44, 52, 83, 101, all counted from the beginning of the pdf) and 3 random items from each page we get:

- a 13½-inch doll with jointed arms

- the 17-inch version of the doll

- a 13-inch doll with closing eyes and rayon socks that says 'momma'

- a 10¾-inch-wide chalkboard with interchangeable educational chart cards with a whiteboard side for erasable crayons

- 10 coloring-book-like picture cards that come with a glue stick and colored sand to color them with

- 12 picture cards that come with six colors of sand instead of three

- a radio-controlled model train controlled by voice recognition (though i suspect the radio part may have been just as fake as the voice recognition part, because it says 'transformer included' but doesn't say it requires batteries, and also because it would have required several vacuum tubes to actually use radio control, and it cost less than the vacuum tubes)

- another 'remote-controlled' model train which is fairly clear that there's no radio

- a wind-up train

- a wooden men's suitcase covered in top-grain cowhide leather with a tray inside for toiletries such as a toothbrush and a mirror

- a similar suitcase for women

- another men's suitcase with split cowhide leather on a metal frame

- a 1-pound bag of extra fancy fresh mixed nuts (walnuts, pecans, brazil nuts, almonds, and filberts)

- a 1-pound bag of filberts

- a 5-pound bag of brazil nuts

this obviously isn't enough to conclude that 99% of the stuff in the catalog had direct equivalents in 01927, but i think literally every item in this list (generated by python's systemrandom object) did, which i think justifies the claim that probably at least 94% of it did. even electric model trains date from 1897, and hornby was selling toy versions from 01925

> differences in computers and nuclear power up to 01936 were all zero since neither of them existed.

Computers existed in 1936, they were mostly analog but still surprisingly capable. Unlike today where we’ve basically standardized on only using transistors things where rapidly evolving in serval directions at the same time including the use of vacuum tubes and electromechanical relays etc. Look at the evolution of say 3D graphics card pipelines and sure things are getting faster but we’re well past the peak period of innovation when dramatically different architectures where showing up regularly rather than just more processing power.

Nuclear power isn’t limited to electricity, we used to do things like paint watch dials with phosphors that lit up at night. Radioactive quackery had largely died down by 1936, but that in itself was a significant change as was the understanding of worker safety issues.

> starlink, alexa (echo), airpods, impossible burgers, apple carplay, cheap solar panels, and ring cameras, none of which existed ten years ago

Alexa was released in 2014, and it’s arguable how distinct it is from Siri (2011) etc. You’re basically defining new in terms of brands on that list.

Satellite internet, bluetooth earpiece, vegan burgers, cheap solar panels, internet security cameras were all available 10 years ago. Solar panels have gotten cheaper per watt, but in 2014 you could get a 10kW system for ~30k vs 26k today. Adjusting for inflation it’s a bigger price drop but breakeven times aren’t that different.

> Keeping a weather eye on China, if they do their usual trick and start learning how to build nuclear cheaply

Well, that ship has apparently sailed already:

How innovative is China in nuclear power? (1 day ago | 173 comments) https://news.ycombinator.com/item?id=40703088

this is a great find, thank you, but keep in mind that 81% of the prc's new electrical generation capacity last year was wind and solar; 16% was fossil fuels, 2% was hydroelectric, and ½% nuclear. so while they're certainly going to do their best with nuclear, for the foreseeable future they're moving rapidly to solar

https://climateenergyfinance.org/wp-content/uploads/2024/01/...

Who cares? If solar is cheap people can build solar. That isn't a particularly controversial idea. When coal was cheaper coal plants got built, when gas was cheaper people built gas plants. When solar is cheap we can build solar farms.

The issue is that nuclear has the potential to to shift society's energy costs 2+ orders of magnitude cheaper and we're explicitly retarding its progress by policy. Apparently only because of irrational beliefs.

hmm, so, if i get a 360-watt solar module for 30 dollars and set it up someplace with a 30% capacity factor (like in the north of argentina where it's sunny) it will produce an average of 110 watts, which is 950 kilowatt hours per year. if i demand a 10% return on my investment, i'm getting 950 kilowatt hours for 3 dollars, which is 0.3¢ per kilowatt hour

this is already better than wholesale electrical prices, which average about 4¢ per kilowatt hour, but of course that involves all kinds of auxiliary equipment and batteries and whatnot. so let's assume generously that you're saying that nuclear energy has the potential to produce electricity for a wholesale price of 0.004¢ per kilowatt hour, which would be 2 orders of magnitude cheaper

why would you think that it has that potential? you cited crawford and devanney, but they don't say anything that even suggests that they think that. at the extreme they think that you can get those same 360 watts from nuclear power for 360 dollars. what are you basing your estimate on?

https://en.wikipedia.org/wiki/Energy_density#In_nuclear_reac... - the difference in energy density from coal to uranium alone is absurd, let alone if we have some breakthrough that taps in to a more exotic substance. It isn't that much harder to mine these fuels, the only problem is that we haven't really squeezed down the cost of building those honking great plants and focused on making the whole supply chain cheap. We're talking potentials like 6 orders of magnitude improvement over coal (which is currently competitive enough to run an industrial society).

We should be giving the nuclear industry as many opportunities as possible to achieve breakthroughs, not limiting it.

yes, clearly that is correct when it comes to spacecraft, where what matters is energy density; voyager 1 would not be transmitting to us today if it were powered by kerosene or solar panels.

but, here on earth, operating power plants is generally not an expense in proportion to how much their fuel weighs. only about half of the cost of a coal power plant is the coal; the other half is the plant, with its spinning electromagnets, parsons turbines, deionized supercritical steam circuit, cooling towers, and so on. as long as you're using nuclear power as an alternative way to drive steam through the coal-plant machinery, the best you can do is knock the cost down by a factor of about 2, which is 0.3 orders of magnitude rather than 2+

also, by the way, if you can come up with a heat engine that's cheaper by one or more orders of magnitude, you can compete with photovoltaic power plants using not just nuclear energy but also concentrating solar and geothermal. that would be a huge boon

> but, here on earth...

Well, yeah. If you assume we keep doing everything the same as we do right now, we'll get similar results to what we are getting right now. You've put your finger on the core issue with that paragraph, but I suspect not in the way you intended.

What you're not dealing with is that the likelihood we figure out alternative ways of doing things is extremely high. Or it would be if we didn't have a brain-dead regulatory state in the West that blocks us from doing things differently. We're talking a fuel source where we can fit the energy equivalent of 10 trains of coal into an 2L ice cream container - assuming no improvements from using Uranium. There aren't any laws of physics saying it has to be expensive to turn that into useful energy. We haven't really tested any part of the industrial supply chain to see how low costs can go. And any ideas the academics are having aren't being put in to the practice - or at least they weren't until the best and brightest over in China started getting involved. Hopefully it works out.

Not to mention that, as you point out, nuclear is also a technically superior form of energy in a bunch of specific applications where weight and volume are factors.

i agree!

There's a big handwave at "an average of 110 watts." There will be periods where it generates 0. Does that line up with demand? How will you satisfy demand at night? Solving that is the majority of the costs of solar/wind. The cost of the panels is almost insignificant.

currently the cost of the panels and balance of plant, not batteries, is the majority of the costs of solar power. the panels have historically been about ⅓ and are edging up past ½ of the total. batteries are not insignificant but also not overwhelming. obviously the amount of battery needed is dependent on demand patterns

see https://news.ycombinator.com/item?id=40725765 for me trying to quantify this better

How about we keep in mind that they're doing both, in parallel:

    China intends to build 150 new nuclear reactors between 2020 and 2035, with 27 currently under construction and the average construction timeline for each reactor about seven years, far faster than for most other nations.

with a clear stated long term planned intent for a target mix of solar+nuclear.

yes, but the nuclear ingredient continues to be insignificantly tiny. 10 new nuclear reactors per year is maybe 20 gigawatts per year. they installed 216 gigawatts of solar capacity last year, and that's a number that's rapidly increasing. i think the reason they're building nuclear plants is the same reason so many nuclear plants were built during the cold war: to build nuclear weapons

yes but, they're doing both, in parallel.

Currently coal still accounts for nearly 60% of China's electricity. The vast bulk of global solar production comes from China and the bulk of that production of solar panels is being carried by coal power underpinning the required energy demands.

Go on, say solar again.

Act like I have no idea and that saying solar again will change something about the current state of China's energy supply and demand, their plans for the future and my current understanding of those plans.

> i think the reason they're building nuclear plants is the same reason so many nuclear plants were built during the cold war: to build nuclear weapons

There's also having consistent reliable baseload power for 24|7|365 day industry and personal use which they intend to achieve with a lot of solar and somewhat less nuclear.

solar ;)

Well, 216 GW of solar is in no way comparable to 20GW of nuclear. Capacity factor much higher, much more reliable + no storage needed.

What's more, Chinese nuclear power is sometimes also used to heat homes. https://www.globaltimes.cn/page/202211/1278469.shtml

If you want to make an intellectually honest comparison, this is the reality.

This is kind of scary considering Chinas huge variability in quality assurance. I’m sure they can build with great quality most of the time. But it only takes one bad project to set the whole world back another decade or two when it comes to nuclear energy.

Yeah, I know it’s not rational. Doesn’t matter. Humans just are not all that rational. People will shun nuclear no matter how safe you assure them that your reactor is if there’s one big catastrophe somewhere.

> Supporting nuclear" isn't supposed to mean splitting the atom at any cost. The point is if we would just stop regulating the industry into oblivion

Sounds to me like just that.

If deregulation is the only way you could make nuclear work in the 21st century, then it rightfully is on the way out.

> Keeping a weather eye on China, if they do their usual trick and start learning how to build nuclear cheaply they could leapfrog everyone else on energy tech in another industry.

Oh, China learned to do all kinds of things cheaply. They are famous for their tofu dreg and the catastrophes connected with this meme, which is a result of regulative negligence and corruption. I would question the theory that this model should become an example for the rest of the world and especially in technologies which are as dangerous as nuclear power generation.

But there are actually fantastic news coming from there with much lower risks where we could learn from:

https://cleantechnica.com/2023/02/06/renewables-in-china-tre...

The issue with solar (and wind) is unstable generation profile, and need to overprivision a lot to compensate for worst-case scenario, e.g. cloudy windless December (you can't have batteries this big). Usually the way to do it is solar/wind + natural gas plants, which usually shut off

agreed! you can afford hours of batteries but not days or weeks. after looking for quite a while i found this nrel report on utility-scale battery storage system cost: https://atb.nrel.gov/electricity/2023/utility-scale_battery_...

in theory, flow batteries and fuel cells could allow you to afford days or weeks of batteries, but the incentive to invest in those isn't there until the renewable transition is a lot further along. aluminum-air batteries are sort of like fuel cells that burn aluminum, which you could stockpile a lot of, and i think they could potentially be cheaper per watt than gas turbines. and iron-air batteries are another similar variant which can in fact be recharged

my intuition is that gas turbines have a much higher power density than steam turbines because the temperature difference is greater and you don't need an external boiler. but i don't know if that works out to a lower cost per peak watt. do you have any idea?

Many industrial countries already have lots of gas power plants. If they don’t actually need to run that often and don’t need that much fuel, just run them on hydrogen or biogas instead.

Near zero investments to be made. Just somewhat larger operational costs per MWh delivered.

If you collocate some industry that needs hydrogen (necessary for green steel and fertilisers for instance) then making and storing the hydrogen on-site should be fairly cost efficient.

Running hydrogen in gas turbines has been demonstrated already btw.

right now only about a third of the 18 terawatts or so of world marketed energy consumption is electrical. to electrify the rest of the existing economy, we need to triple electrical generation capacity. since some of that new capacity will be intermittent sources like solar and wind (probably almost all of it) we also need to expand peaker plant or battery capacity, probably a lot, in order to supply essential uses that aren't amenable to demand response; the alternative is blackouts and deindustrialization

if your peaker plants are gas turbines, running them less drives up their capex per megawatt hour delivered, so i'm interested in how much that capex is

i agree that hydrogen is a plausible form of storage and can certainly be used for gas turbines

>now that solar energy is far cheaper than coal ever was

I'm sorry, this is simply not true. Usually such claims are the result of ignoring two basic facts:

First, solar is only online 50% of the time, at most (in most regions, considerably less than 50%). That means that you need at a minimum twice the nameplate capacity for a solar plant than for a coal one.

Second, being offline half the time means that you also need sufficient (very expensive!) storage capacity to cover the half the time that the solar plant is not working at all.

In other words, to replace a 2GW conventional plant you're going to need at least 4GW worth of solar cells, plus 24 megawatt-hours of storage.

If you have a source for reliable figures that take these factors into account, and still show solar being "far cheaper", please provide it.

Edit: oh, and no woo-woo battery or solar cell technology that's not currently in mass production. Your statement was that it's "far cheaper" right now, not using future fantasy batteries or solar cells that may (or may not) be on the market in the future.

You nailed it. What do you do when you have cloud cover for two weeks in a row? It’s not unheard of when some areas have 200+ days of at least some cloud coverage.

Nuclear is essential to avoid using coal and not having regional rolling blackouts due to weather. Solar alone is not realistic anytime soon in all regions of the US.

> You nailed it. What do you do when you have cloud cover for two weeks in a row?

Never mind the volcanic winter that seems to happen every few thousand years.

Nuclear is a horrible complement to cheap intermittent renewables. Running it as a peaker multiples your costs and running it as base loads means selling power for negative prices when the sun is shining.

Solar/wind plus batteries for short term storage and pumped hydro for long term storage is the cheapest way to get zero carbon energy. Pumped hydro is more expensive than fossil peakera so build out of that hasn't happened yet.

do you have a source on the pumped hydro costs? i thought it was cheaper, just impossible most places

Costs are high because gas peakers exist and pumped storage generally doesn't. For rarely used long term storage, capital costs dominate, and already built sites don't incur additional capital costs.

All you need is a hill and some water for pumped storage. Those sites are very plentiful.

https://scholar.google.co.uk/citations?view_op=view_citation...

oh, yeah, i was thinking of places where you already have a reservoir suitable for hydroelectric production, not places where you have to build a dam

How do you scale up to producing far more energy in that approach? What about the efficiency to produce large amounts of energy on large spacecraft and other planets and in the ocean?

are you asking how you scale up to producing far more energy with solar panels? world marketed energy consumption is about 18 terawatts, total terrestrial insolation is about 128000 terawatts, and current mainstream panels are about 23% efficient, so if you put solar panels on 50% of the earth's surface, you get 15000 terawatts, which is almost 1000 times more than the humans are using now. on the bottom of the ocean you probably need a different approach, maybe nuclear, or egs geothermal, or maybe running a cable up to the surface, or periodically receiving shipments of thermite in a submarine. some other planets will have no trouble with solar panels; others will need nuclear reactors

Getting only 1,000 times more than humans are using now but requiring 50% of the earth's surface seems like an awful deal. Not only do you need much more of the earth's surface, taking away from trees and habitats and other uses, but you need to significantly increase mining activities to produce the panels and their associated infrastructure. Whereas to get 15,000 additional terawawtts from more nuclear reactors, you could do that with 1,200 - 1,900 additional nuclear reactors occupying just the size of Rhode Island.

But we wouldn't need 50%, we'd need .05%

That depends on how much more efficient solar panels get and what future energy requirements are, but in any case it'll be worse to scale energy creation with solar than with nuclear.

We'll never substantially increase our energy demand if we don't reduce our energy cost. And since nuclear is horrendously expensive, it won't scale.

i would instead say that as long as nuclear energy is horrendously expensive, it won't scale. but nuclear energy is not inherently horrendously expensive; it's just that human technology is very primitive still

mostly you'd be floating them on the oceans (ideally the currently-nutrient-depleted parts of the oceans that aren't teeming with algae), but yeah, that's roughly the limit for solar, and as you're approaching that limit, you need to be thinking about space-based solar power, nuclear power, geothermal power, etc. maybe when you're at 64× of current energy consumption, say

since 02000, total solar installed capacity has gone from a gigawatt to 1.6 terawatts (see https://www.ise.fraunhofer.de/content/dam/ise/de/documents/p... and https://en.wikipedia.org/wiki/Swanson%27s_law#/media/File:19...), which is roughly 10 doublings, one doubling every 2.3 years. but that's peak capacity, and 1.6 terawatts peak is only 220 gigawatts of actual production at a presumed capacity factor of 21% (see https://en.wikipedia.org/wiki/Solar_power_by_country). that's 6 doublings away from world marketed energy consumption; adding the other 6 doublings to get to 64× gives you 12 doublings, and thus about 28 years before this starts to be a concern

probably we should think of this as a lower bound, though; adoption is likely to slow down as solar moves into application areas that are not already electrified or indeed yet done at all by humans, and the last 24 years have been, historically speaking, unusually peaceful

no place has as high a capacity factor for solar pv as 50%; the world average is 14% (i.e., 14 watts average output per 100 watts nameplate output) and the country with the highest is egypt at 35%

(this doesn't mean that you get zero power for 65% or 86% of the day. it means that you get lower than max power at all times except noon. to a great extent you can compensate by that by just installing more panels, but at night you need a better strategy)

but the capacity factor is irrelevant to the fact that solar energy on the power grid generally sells for about half the price that coal-generated electricity sells for, on the same grid, when storage capacity is insufficient. in fact, prices used to go negative at night (to avoid shutting down slow-ramping baseload plants) and now they go negative in the day

https://pv-magazine-usa.com/2020/05/28/record-low-solar-ppas... is an article from four years ago giving some specific prices: a solar ppa had just been signed for 15 dollars per megawatt hour, while the cost of production with coal at the san juan generating station was 44.90 dollars per megawatt hour, even though it was built right on top of a coal mine to save on shipping costs. that's why the san juan generating station has been decommissioned. if you look, you'll find stories like this all over the place, and solar panels now cost half of what they did when that story was written

now, it's true that a ppa that includes battery storage will be more expensive than the 1.5¢ per kilowatt-hour ppa in that article. (https://emp.lbl.gov/pv-ppa-prices has a queryable database of all the ppas signed in the usa, although the usa is less reliable as an indicator of true costs because of how its prices are artificially inflated by protectionist tariffs.) how much more expensive depends on how much utility-scale storage is needed; you suggest 12 hours, but a much more typical number in practice is 3–4 hours, partly because there are still coal plants and partly because electrical demand drops a lot at night

also, you are incorrectly assuming that 'conventional' plants have a capacity factor of 100%, when a more typical capacity factor for a coal plant is 60%

so let's consider a kind of worst case: replacing your suggested 2 gigawatts (nameplate) of coal plants in the usa, where construction costs are ridiculously inflated. before we swung the wrecking ball, those coal plants were generating 1.2 gigawatts (real) of power (10.5 billion kilowatt hours per year), so we need 1.2 gigawatts (real) of solar panels. in the usa the average capacity factor is 21% (the article i linked above is from an area with more sun than average) so that's 5.7 gigawatts peak. typical costs for utility-scale fixed-tilt solar plants in q1 02024 were 98¢ per peak watt https://www.seia.org/research-resources/solar-market-insight... including costs like permitting, design and engineering, etc. so that's 5.6 billion dollars. utility bond yields in the usa are currently at 4.35%, and often these things are amortized over 25 years. i think the amortization calculation is that you have to pay 372 million dollars a year, which works out to 3.5¢ per kilowatt hour or 35 dollars per megawatt hour. definitely too cheap for coal to compete with, but still a lot more expensive than the price that ppa came in at

so, suppose we need 4 hours of storage for our 1.2 gigawatts. that's 4800 megawatt hours. six months ago lithium-ion batteries have fallen precipitously to 139 dollars per kilowatt hour https://about.bnef.com/blog/lithium-ion-battery-pack-prices-... so we need to spend another 670 million dollars on the batteries, which adds about 12% to the cost of the project. except that in real life you need more than just a pile of batteries, you need to pour concrete and run wires and connect inverters and so on. and the batteries won't last 25 years, maybe 8, so you have to amortize this capex over a much shorter period. but it should be clear that this is not a crushing cost that dwarfs the cost of the solar farm

(theoretically lead-acid might be cheaper by a factor of 1.5 or 2 or so, but lithium-ion's advantages seem to have driven it out of the utility-scale market)

aha, here we go. https://atb.nrel.gov/electricity/2023/utility-scale_battery_... says that a 4-hour 60-megawatt lithium-ion battery system costs 446 dollars per kilowatt hour and has 240 megawatt hours of storage. so our required 4800 megawatt hours cost 2 billion dollars. that's about three times the cost estimate above for just the batteries, but that's an estimate from before the batteries dropped in cost by half, so 1.3 billion dollars is a better estimate. this plus the 5.6 billion dollars for the solar plant gives us a total up-front cost of 6.9 billion dollars

since it's getting late, i'll just link you to this four-year-old profile of a ppa for 20 dollars a megawatt hour for generation plus 20 dollars a megawatt hour for storage https://www.energy-storage.news/battery-storage-at-us20-mwh-... and this lawrence radiation lab brief https://eta-publications.lbl.gov/sites/default/files/utility...

the main reason for the difference seems to be how sunny the location is; the technical brief explains:

> Aided by the ITC, most recent PPAs in our sample are priced around $20/MWh (on a levelized basis, expressed in real 2021 dollars, and including bundled energy, capacity, and RECs) for plants located in the West, and $30-$40/MWh for plants elsewhere in the continental United States.

the itc is a subsidy, so the real cost is a bit higher (due to the tariffs)

> when a more typical capacity factor for a coal plant is 60%

This is, of course, patent nonsense. "Not producing maximum power because it isn't needed at the moment" is an entirely different thing from "not producing maximum power because you can't".

A coal or nuclear plant that's producing less than its maximum output because the power isn't needed at the moment can be ramped up if needed (not super quickly, hence the need for peaker plants, but it can be done).

A solar plant that's producing less than its maximum output because it's night, or because it's cloudy, cannot.

> https://pv-magazine-usa.com/2020/05/28/record-low-solar-ppas... is an article from four years ago giving some specific prices: a solar ppa had just been signed for 15 dollars per megawatt hour, while the cost of production with coal at the san juan generating station

1) No storage costs are mentioned. 2) Most places are not New Mexico. 3) I'd bet money that there's some heavy government subsidization involved here. Ah, yes: "EPE will also receive the associated renewable energy credits (“RECs”) bundled with the purchased energy."

> you suggest 12 hours, but a much more typical number in practice is 3–4 hours, partly because there are still coal plants and partly because electrical demand drops a lot at night

Wait: you're claiming that the nighttime zero production from solar plants doesn't matter because there are still coal plants?

I'm sorry, that is a benefit of the coal plants, not the solar plants.

What happens in your scenario when there aren't any more coal or nuclear plants? Again, that might work in New Mexico, but in most regions people would prefer not to freeze in the dark.

for someone who started out demanding that people 'please provide' a 'source for reliable figures', your comment is astoundingly devoid of any sources or indeed factual information. you are very much not living up to the standards of debate i was hoping for; please try to do better

> "Not producing maximum power because it isn't needed at the moment" is an entirely different thing from "not producing maximum power because you can't".

your replacement solar plant also doesn't have to produce the un-needed power, so you have to take the capacity factor into account when you're calculating the replacement size

> No storage costs are mentioned [in the article about the palo verde trading hub prices]

yes, that's why other parts of my comment explore storage costs in great detail

> Most places are not New Mexico

most places that people live are sunnier than new mexico or close enough to someplace that is

> there's some heavy government subsidization involved here

yes, you may note that the comment that you're replying to discusses those subsidies and actually names another one and computes what the unsubsidized cost would be for your suggested 2/4/1.2 gigawatts. as i mentioned, there's also some heavy government taxation involved here; solar panels in the usa cost twice what they cost in the rest of the world. also, as i mentioned, the price of solar panels has dropped by half since the article was written

> you're claiming ... because there are still coal plants

technically what i claimed was that the typical number in practice was 3–4 hours not 'because there are still coal plants' but 'partly because there are still coal plants and partly because electrical demand drops a lot at night'. probably i should also mention wind power, nuclear plants, hydroelectric power, gas peakers, and distributed storage such as car batteries, all of which currently play a role in compensating for the intermittency of solar

the current average is about 3 hours; that will presumably increase as solar becomes a larger fraction of the grid. to estimate how much it increases, we need to start by understanding the current situation, but we also need to predict how extensive demand response will be. you posit that the storage requirement will increase to 12 hours, but that seems unlikely to me. if it did increase to 12 hours, that would make the battery system cost as much as the solar farm. unless batteries somehow got cheaper

what's at issue is not 'freezing in the dark'. people generally do want it to be dark most of the night, because they're asleep, so not being in the dark only requires storing about 40 watt-hours per person, which is a single usb power bank. and you can reliably avoid freezing with a so-called 'sand battery'; 12 hours of 6000 watt heating can be provided by a tonne of sand heated up to 250° when the sun is shining. that's about fifty bucks of sand and a few meters of nichrome wire. if you live in a normal sized house with maybe some insulation you need a lot less than that

so relax, you're not going to freeze in the dark

a mean man scared you with a scary story, but it's not real

rather, what's at issue is industrial process plants like blast furnaces and haber–bosch nitrogen fixation, which are traditionally pretty intolerant of being shut down; it takes a long time to bring them back to steady state after a perturbation. in some cases there are alternative batch processes that compete with continuous-flow processes, which have conventionally been economically uncompetitive, though there are exceptions such as electric arc furnaces. in other cases, it may be possible to design continuous-flow process plants in such a way that they can ramp up and down efficiently, so that they can take advantage of the unprecedented abundance of free energy during the daytime

I’ll never get behind solar as anything but supplemental or “roof top solar” until we have grid scale batteries that can power more than a few hours. I’m talking about days, because weather events happen. which is what we need to call it “a replacement” unless you live in an area like Hawaii or San Diego that has nice weather year round.

> untik we have grid scale batteries that can power more than a few hours

In the West we are doing this with gas. We are deploying trillions of dollars into new gas infrastructure with 15+ year payback periods and 40+ year lifetimes.

We don’t need to wait for batteries. The peaker power and minor storage is already here.

Couldn't this also be stated as solar being the bread and butter, while other sources are the supplement for when the sun isn't shining?

Solar panels are getting cheaper by the day.

Solar power is not.

Because between the panel and the lightbulb lies a battery.

Do you think batteries are getting more expensive over time?

I think the assumption they would get cheaper given a massive change in their adoption patterns is unfounded. We know that lithium prices have gone up from the rise in EVs, and this has in part driven Tesla to switch to LiFePO4's rather then Li-Ion packs in some of their cars. Same effect with coltan. (although lithium is declining currently because EV sales in China have slowed).

What would be the effect of the demand for terawatt-hour levels of batteries? Even if lithium is replaced with sodium in the application, the manufacturing capacity isn't there and won't be for quite some time - and the higher-value applications will get first take (i.e. EVs).

Even if the lithium demand was saturated battery wise, at the steady state point the cost can then simply stabilize around the refurbisment/recycling cost of lithium/sodium/whatever from batteries.

Sigmoidal adoption curves look exponential at the start, linear in the middle,and then become asymptotically flat.

> although lithium is declining currently because EV sales in China have slowed

have they?

'one in three' of all new vehicle registrations in china were electric last year https://www.iea.org/reports/global-ev-outlook-2024/trends-in... or 37% https://autovista24.autovistagroup.com/news/ev-registrations... and the absolute number seems to be doubling this year, though they only expect a 45% market share

these figures do include both hybrids and pure battery models, though

in general the 'experience curve' says that things get exponentially cheaper as you produce exponentially more of them

The experience curve is asymptotic[1].

What do we get if we look at say, FLOPS per US $ over time? This[2]. There is no reason to think batteries are any different, the only margin you have is to argue over which part of the curve you think we're in.

A sigmoid curve[3].

[1] https://www.extension.iastate.edu/agdm/wholefarm/html/c5-208...

[2] https://www.researchgate.net/figure/Number-of-flops-per-Unit...

[3] https://en.wikipedia.org/wiki/Sigmoid_function

That FLOPS graph is an exponential scale... so that straight line "after the sigmoid" is an exponential growth curve. I'm not sure it makes the point you're trying to make.

Yes, and no.

Yes, a lot of people who install residential solar have a battery, and "common wisdom" tends to size the battery up if the solar install is larger.

But nothing stops the battery being really small, or indeed (like commercial solar generation) there being no battery at all.

Yes, of course, solar is not 24 hours. But having cheap energy "only" half the day (1) is a major plus. And as energy cost fluctuates, so do habits. For example since we got solar we tend to use the dishwasher in the mid morning instead of at night.

(1) actual solar availability depends on the width of your country. For example solar plants in say Arizona happily generate good power while the East coast is dark, and panels in say Florida would cater for mornings in California.

And that's before we factor in wind, which is commonly stronger in the late afternoon into the evening.

you may want to \ your *s

never mind, you fixed it

> And as energy cost fluctuates, so do habits

You're advocating for reverting back to pre electrified times, this stuff is straight from the 19th century. Using energy is good, it is progress.

Not at all.

I'm saying that people adapt very quickly when costs are involved.

I'm not saying we sit in the dark at night. I'm saying that we time-shift some energy consumption because some of the day energy is free, and some of the day we pay for it.

The pool pump runs during the day. As does the hot water cylinder. The electrical cost for these us now zero (on most days.) Our night-time electricity usage has gone down because it's more expensive.

When energy costs the same all 24 hours then it's not something I take into consideration. When it's different habits change. I don't use less energy, if anything I use more. But I factor time-of-day into -when- I use it.

Not everything can be shifted. Our peak usage is still in the morning and evening. But scheduled things have moved from night to day.

I can run 10 old 100 watt light bulbs for an hour for 14 cents. Or 100 led light bulbs. I can run my hot tub full blast for an hour for $1.40. A head of lettuce is $4. A bag of coffee is $18. Five chicken breast is $20. A tank of gas for my wife’s car is $80, and for mine is over $200.

The cost of electricity for the loads I can control, such as those other heating or cooling living space, would have to 10x for it to be worth worrying about the cost, which would mean a $1400/MWH. The ceiling on the wholesale market price in ontario, last time I checked, was $2000/MWH.

I can see the price of electricity rising quite a bit across the board, but I don’t think people are going to inconvenience themselves at all to respond to it.

> I can see the price of electricity rising quite a bit across the board, but I don’t think people are going to inconvenience themselves at all to respond to it.

People already do that, and have done so for decades. Many places have different electricity costs for defined peak and off-peak times, and many people absolutely do move their heavy electricity uses outside the peak times when they can.

Naturally ymmv depending on your energy costs. And indeed other costs. ($4 for a lettuce sounds high, we pay pennies for that here.) It'll also vary depending on your overall income.

I would gracefully suggest that your life-style might not necessarily reflect the life style of the general public?

I think my grocery prices are in line with everyone else in British Columbia

it sounds like british columbia grocery prices are insane. are those us dollars? five chicken breasts here is about $3000, which is about 2½ us dollars

Those are CAD

aha, thanks. xe tells me those are 27% smaller than us dollars, so those prices are respectively 1 dollar, 3 dollars, 13 dollars, 15 dollars, 58 dollars, and 150 dollars, speaking in us dollars

or, using today's mid-market rates from https://preciodolarblue.com.ar/, $1300, $3800, $17000, $19000, $75000, and $190000. $19000 is sure a lot more than i'd pay for five chicken breasts

as a clarification from further down the thread, since you said those are canadian dollars, those prices are respectively 1 dollar, 3 dollars, 13 dollars, 15 dollars, 58 dollars, and 150 dollars, speaking in us dollars

me, i pay $3000 for five chicken breasts, which is about 2.3 us dollars

I am baffled by this argument. What point are you trying to make? The energy is still being used, just at slightly different times.

> What point are you trying to make?

What's so hard to understand? Children can't do schoolwork after dusk without electricity. Society doesn't function after dusk without massive electricity. Let's do our work while the sun shines brightly is 19th century level behavior.

You're making assumptions here though that children having to do schoolwork after dark is a fundamentally good thing. The same goes for the assumption that society can't function after dusk - what did society do before the lightbulb was invented?

Taking modern norms as a basis for why we need the tech that allows the modern norms is a logical loop. People did learn things before electricity, and electricity didn't predate society.

if you read william kamkwamba's autobiography, you will gain a major appreciation for what a huge improvement functioning after dusk can be. his 12-watt windmill revolutionized his family's life, for the better, even before it made him world-famous

I have little doubt that going from no electricity to 12 watts can be a huge leap, but I'm not familiar with William Kamkwamba. Help me out here, what goals or metrics did he point to when showing that his family's life was better?

read the book

I may. In the meantime is it really so hard to give highlights or an example that at least better explains the point you were raising? Just saying one person felt their life drastixally improved with one change they made isn't really helpful at all, you can find countless anecdotes to make whatever argument you want.

you can do your schoolwork with a 1-watt led, a 10-watt smartphone, or a 30-watt big-screen backlit laptop. if your nighttime electricity costs are a ridiculous 40¢ per kilowatt hour, abstaining from 30 watts 4 hours a night will save you 43 kilowatt hours per year, almost 18 dollars. even in like lesotho those aren't the kinds of loads you'd have a strong economic incentive to shift to the daytime

it's more things like baking dinner in an electric oven, demolishing concrete with an electric hammer, welding with an arc welder, heating the water in your hot-water heater

Yeah, most of the load I cannot move so cost has no effect on my usage it just means I have less disposable income. 75% percent of my bill is network cost so power source cost make little difference to me.

if solar panels on your roof generate the energy, you don't need the network

Well that's where I am heading. but its a failure of society when each person must provide their own utilities.

probably in some places people will figure out how to provide a cheaper network, even if society fails where you live. hopefully you can move

Dude all he suggested is running your dishwasher while at work instead of at night, and you're acting like he wants to end modern life as we know it. Chill.

i agree that using less energy would be bad; we sure as hell aren't going to achieve atmospheric carbon capture that way, much less terraforming mars. but i think using less energy is not what's being advocated; rather, presuming that consumption patterns will remain unchanged in the face of a new incentive structure will result in an unrealistically pessimistic assessment of the required battery storage. having energy that's literally free during the day is likely to result in using more energy, not less

commercial building hvac systems have been responding to these incentives for decades: freeze water with chillers at night when electricity was cheap (or free, or actually negative cost), then circulate coolant through the ice during the day to get cold coolant to cool air through heat exchangers and thus air-condition your office. you could imagine freezers and refrigerators that worked the same way, storing energy when the sun is up to keep your food cold when the sun goes down

the popular evacuated-tube solar hot-water heater of course only heats the water during the day, storing the hot water for nighttime in an insulated tank, usually supplemented with an electric heating element for the rare occasion that the stored heat is insufficient. and there are already places where electric hot-water heaters respond to commands to preheat water at off-peak hours. these so-called 'sensible heat storage' devices are much bulkier and leakier than the phase-change type from the previous paragraph, but they can be very simple indeed

going beyond phase-change energy storage, tces energy storage uses phenomena like the enthalpy of hydration of the muriate of lime. this provides thermal energy storage that's another order of magnitude more compact; it can be used for heat and dehumidification as well as cooling, and some storage media such as lye can even get hot enough to be used for cooking

this stuff is straight from the 22nd century

How do you define progress?

The biggest challenge I see in similar discussions is that we don't have a shared understanding of what progress because we don't define the goals we're moving towards.

it's the opposite of congress

>How do you define progress?

To some people, progress is not being made unless useless middlemen are raking in the bucks and getting more out of the energy you are using than you are.

I think that's an incredibly uncharitable interpretation of what GP said.

We already change our behavior around electricity costs and demand: for example, a couple years ago PG&E (California) forced everyone to switch to a time-of-use rate plan where electricity costs go up quite a bit between 4pm and 9pm. You better believe I avoid doing things like laundry or running the dishwasher then, and opt to run them earlier in the day or later at night.

If I had solar (without battery storage), I'd absolutely be running my heavier electric load when the sun's out instead of at night when I have to pay for power. Sure, the ideal would be everything is similarly cheap no matter what time of day, and maybe we'll get there in 75 years or so, but until then, I'm fine continuing to do some time-of-use optimization here and there.

keep in mind though that pg&e is kind of a pathological case

That may be true for the US, but lots of the world has variable energy pricing already.

not because of pricing variability, but because pg&e prices are just unbelievably high. would you believe 39¢ per kilowatt hour?

>Using energy is good, it is progress.

Roger.

Using less energy is even better, it is further progress.

not my downvote btw

Very few people have batteries with their solar in my neighborhood.

>Very few people have batteries with their solar in my neighborhood.

This is common. It is of course due to the cost of battery systems, particularly if the solar system was installed several years ago. I just bought a house that has solar and not only does it not have battery storage, but the solar cannot power the house when grid power is down. Seems insane to me to build a system that can't bootstrap from the solar and run things on an as available basis during outages but apparently this is typical.

I'm still learning, but I also think that I am not getting great value for the power I am supplying to the grid during the day. Add to that the extra cost of power during peak evening hours when I don't have solar and a battery to time shift my solar for my own use during the evening seems like a win. It still helps the grid since I am pulling less during peak evening hours.

solar power is getting cheaper by the day; https://www.pv-magazine.com/2024/04/19/europes-renewable-ppa... says:

> Energy consultancy LevelTen says that solar power purchase agreement (PPA) prices fell 5.9% in the first quarter of 2024, with decreases recorded in all analyzed countries except for Romania. It attributes the decline to lower wholesale electricity prices and a fall in solar module prices

however, they haven't yet recovered to the 02020 levels, presumably because the astounding precipitous fall in module prices over the last year hasn't been priced in yet

Battery prices have dropped 50% in the last 12 months.

https://www.pv-magazine.com/2024/03/07/battery-prices-collap...

That’s great.

Grid-scale battery storage is still a terrible idea when you can simply build a nuke station that does the same job with 0.01% of the waste material and space used.

it seems implausible that a nuclear power plant occupies 0.01% of the space of a battery bank of the same peak power capacity; if anything, i'd suspect the discrepancy is closer to the other way around (edit: it is, see the comment below where i do the math, and please stop posting thoughtless bullshit)

if you were talking about the solar farm, then yeah, i'd agree

If that seems implausible you might want to replause your seemometer. A nuclear plant has much higher energy density than any battery.

well, nuclear fuel rods certainly have much higher energy density than any battery. but they're only a small piece of the power plant, and i wasn't talking about the energy density (joules per liter) but the power density (watts per liter or, since we're talking about land use, watts per square meter)

a 10c 2000 milliamp hour 18650 cell produces 20 amps at nominally 3.7 volts, which is 74 watts. if it occupies 18 mm × 18 mm × 65 mm that's 3500 watts per liter. tepco's kashiwazaki-kariwa nuclear power plant is 8 gigawatts, which is 2300 cubic meters of 18650s. if you stack them two meters high, that's 1100 square meters, an area 34 meters square. maybe you have to double that so you can drive a forklift in between the racks of batteries, plus you need some space for things like fireproof bulkheads, wiring, and inverters, but we're talking about maybe 50 meters by 50 meters, 0.0025km². kashiwazaki-kariwa is 4.2km², 1600 times bigger. (also, it's more than two meters tall.) the hypothetical 8-gigawatt battery bank uses 0.06% of the land area. even if you scale up the discharge time from 6 minutes to the usual 4 hours, it's still only 2.4% of the area of the nuclear plant

you can get better batteries than that, too. and if you're worried about land consumption, as you have to be in japan, you can put your batteries in a building with multiple floors

Nuclear has the opposite but related problem of solar; whereas you often generate excess solar and need to store it somewhere during the day, nuclear energy often generates too much during low demand times and has excess energy to store as well. And you can’t easily scale down nuclear during lower demand hours.

One of the largest pumped hydro stations was originally built to store excess energy from a nuclear power plant. https://en.wikipedia.org/wiki/Ludington_Pumped_Storage_Power...

Similar time frame. Slightly smaller capacity, Northfield Mountain was brought online to store power from the Vermont Yankee nuclear power plant. Interestingly the pumped hydro facility is still functioning as a grid battery, even after the nuclear plant was decommissioned for no longer being cost effective.

https://en.wikipedia.org/wiki/Northfield_Mountain_(hydroelec...

I think "simply" is doing a lot of work there. Work that completely overwhelms it.

Building a new nuclear plant is a hugely capital intensive project, not to mention the massive amount of time it takes to complete it and get it fully online.

Places like France (that feed a huge percentage of their demand with nuclear) don't somehow magically have super cheap energy. Nuclear plans cost a lot to build, and a lot to operate.

Those huge capital costs are certainly limiting China's expansion into nuclear:

    China intends to build 150 new nuclear reactors between 2020 and 2035, with 27 currently under construction and the average construction timeline for each reactor about seven years

~ https://itif.org/publications/2024/06/17/how-innovative-is-c...

If they were cheaper they'd likely build 300 in that time.

or 1500

woa there, they've only got three of these:

https://www.world-nuclear-news.org/Articles/Excavation-of-Ch...

:-)

right, you could imagine a much higher level of investment. that waste isolation pilot plant is 400 million dollars over 7 years, 50 million dollars a year. the 217 gigawatts of solar power the prc installed last year represent an investment of probably 88 billion dollars last year, assuming 40¢ per peak watt. so the prc spent more money building solar plants in the last five hours than they will spend on that nuclear waste facility in this entire year

and you say they only have three of them?

Either global warming is a serious catastrophe or it isn't.

If it isn't, this is all moot and we should burn cheap, clean, abundant natural gas.

If it is, surely it isn't a problem to shell out more money to get green baseload generations from nuclear? We need to meet base demand around the clock, and nuclear is the least harmful way to do it.

Or we invest our money in the technology with the largest bang for the buck: renewables.

Then follow the research to handle the resiliency, still at vastly lower costs than anything nuclear.

https://ieeexplore.ieee.org/document/9837910

"Simply" building a nuke station sounds like an oxymoron.

Plus 0.01% is quite small but if it's got U235 it will still be more deadly millions of years later than thousands of percent of other deadly toxic highly energetic materials such as pure TNT, in the time the uranium half-life has barely elapsed. And due to the nature of radioactive half-life, the second half of the uranium will not lose its radiation until many millions of years more than the first half required.

Edit: not my downvote btw

Batteries are getting cheaper by the day too. Price per kWh halved in China in the last 2 years according to BloombergNEF (https://x.com/colinmckerrache/status/1800901575214514475).

>Solar power is not.

Often there is no battery, but regardless when solar cost goes up it could be due to expensive contractors in-between the panels and the installation & maintenance, before the lightbulb even gets a chance to see a single watt.

a major reason in the usa is the tariff regime that has basically completely shut out chinese solar panels from the market; this makes solar panels in the us cost about twice as much as in the rest of the world

Yeah solar is inevitably going to dominate the grid mix over the next couple decades.

https://www.eia.gov/todayinenergy/detail.php?id=61424

Even if you discount by capacity factor it's the #1 new energy source being installed in the US, and similar or better anywhere with enough sun.

I've done some armchair analysis that says that solar plus battery is cheaper than basically any other kind of new plant, obviously the biggest gains come first in peaking and adaptive load, but once you factor in fuel costs it even beats out natural gas now I think.

You do realize solar isn't viable in norther parts, due to peak demand being during cold winters and just less sun in general? Solar is great in the south where peak demand is during the warm sunny summers.

Some like a transmission issue, not an electricity generation issue.

Some transmission issues are solved by local generation

How would transmission solve the massive electricity demand during cold winter nights? Even in more southern areas generation during winter is lower, then pay the massive loss of transmitting that so far north and it is no longer viable.

hvdc transmission doesn't have massive losses from transmitting long distances, so you could literally transmit power from the tropics or even the other hemisphere. but overprovisioning solar locally is probably cheaper

Are there hvdc transmission lines already spanning those kinds of distances? How much raw material and upkeep does that require? If over-provisioning solar, how much more surface area would be needed to have the solar? Seems like the additional costs would pile up to get to current energy needs being met by just solar.

you'll probably be interested in reading https://en.wikipedia.org/wiki/High-voltage_direct_current. it says typical losses are 3.5% per 1000km, and typical voltages are 100–800 kilovolts, but china built a 12-gigawatt 1100-kilovolt link over 3300km in 02019

unfortunately i don't have a good handle on how efficient that link is. the numbers above suggest it would be about 89% efficient, but those are for lower-voltage systems

in general you expect the resistive losses in the cables, at a given diameter and power, to scale with the inverse square of the voltage, so an 1100-kilovolt link should have 47% less resistive losses than an 800-kilovolt link at the same power level running over the same cables. however, corona-discharge losses increase at higher voltages rather than decreasing, and i don't know which one is dominant. so i don't know if that link is closer to 89% efficient or closer to 95% efficient

the distance from the north or south pole to the equator is of course 10000 km, but people don't build very near the poles. for example, from svalbard to algiers is 4700 km. so, yeah, there are hvdc transmission lines already spanning those kinds of distances, but not quite that far yet

transmission lines require utterly insignificant quantities of raw material, but where they cut through forested terrain, they do require upkeep. in the meeting of a tree and a megavolt, neither one comes out unscathed

for some calculations on how much you need to over-provision solar, see my earlier comment at https://news.ycombinator.com/item?id=40724349 considering a counterfactual where somehow norway had to use solar. 6.1 average kilowatts per norwegian at a capacity factor of 6% works out to 102 kilowatts peak per norwegian. if you're using low-cost 16%-efficient solar panels (as i assumed you would in my cost estimate, even though mostly people spring for the more efficient 'mainstream' ones) that would require 635 kilowatts of sunlight per norwegian, which is 635 square meters per norwegian (we rate solar panels on the assumption that sunlight is 1000 watts per square meter). 635 square meters is 25 meters square, .000635 square kilometers per norwegian. multiplying that by 5.5 million norwegians gives you the truly immense area of 3500 square kilometers of solar panels

but wait! that's not all! you can't just lay the panels out flat on the ground and expect them to get a 6% capacity factor. you have to angle them toward the equator and spread them out so they don't shade each other. oslo is at 60° north, so your panels need to be angled at 60° from the horizontal, toward the south. maybe a bit more if you want to increase power production in winter, say 69°. so you have to space them out by 1/cos 69° ≈ 2.8. so actually you need 9800 square kilometers for your €340 billion of solar panels. how much space will you have left?

a lot, it turns out. norway is 385000 square kilometers, so this is still just 2.5% of the country

so norway, despite being the #2 highest user of energy per capita in the world after canada, and having cities that are literally inside the arctic circle, could switch to all solar. it's totally feasible. fortunately they have plenty of hydropower and wind so they don't need to

It's definitely a generation issue. When there's no sun(overcast, nighttime) there's no energy. This doesn't even factor in solars quick deterioration from peak performance and the cost of the panels and environment damage from producing them.

Yeah, I think people looking at just the upfront cost are not really acknowledging or addressing that solar panel installations degrade faster than a nuclear reactor. They also take up more space and as you mentioned will likely end up causing more damage to produce at planetary scale. They are definitely great when coupled with batteries for many use cases including decentralizing aspects of the grid for residential and smaller scale usages, but the raw performance of nuclear is impressive and exciting. So little inputs needed for how much you get, and for how long too. There are old reactors still producing after over 50 years…that is mind-boggling.

Who knows how far the tech can be pushed with modern advancements and less blockers on developing the technology further. It should be in the toolbox as part of a strategy for renewable energy needs on Earth and beyond.

nuclear energy is very exciting and absolutely crucial for space exploration, but not economically competitive with solar in the foreseeable future on earth's surface

there are also solar panels still producing power after 50 years; they do degrade a little, especially in the first ten years, but the 20–30 year panel lifetimes you see published are more of a warranty and accounting issue than anything else. (of course some panels crack or yellow within a year or two)

it's true that solar farms take up a lot of space, but even in high-density countries like japan there is room for them. singapore might have a problem tho

What would surface area needs be like for over-provisioning needs in the US? What if we want to scale energy production by 2x or 10x or 100x for advanced industrial and commercial usage needs in the future? I think then the solar panel approach becomes limited on earth.

You’re right that the panels don’t degrade a ton. I read online that after 20-30 years they might drop 15% efficiency. For residential usage that might be okay, but it does mean needing upkeep and worrying about baseline potential dropping, which in some climates could be bad.

I do think a combination of the technologies is best, since scaling up energy production will be simpler and easier and more resource-friendly with nuclear than with more solar. Moving up the Kardashev scale will require capturing all the energy that can be captured from all sources so why let any go to waste. :)

yes, as you approach kardashev type 1 you will definitely want to start harvesting sunlight from von neumann probes on solar orbit

including transportation, natural gas, etc., but not including foods like corn and canola, the usa uses 100 quads per year, or 3.3 terawatts in si units. its average utility-scale solar power capacity factor is 21%, so you'd need 15.7 terawatts peak of solar farms to supply that, before scaling up by 2× or 10× or 100×. 15.7 terawatts of 24% efficient solar panels would require 65 terawatts of sunlight, which is to say, 65 billion square meters or 65536 square kilometers (to pick a round number). this is of course 256², so, like the entire spectrum of mainstream political opinion in the usa, it would all fit between houston and austin. you could drive around it in a day

well, not quite; that's 29° latitude, so you need to space your panels apart by a factor of 1/cos 29°, about 14%, so they don't shade each other. also in texas, unlike any other phenomenon known to humanity, it would be a bit smaller, because texas has a 25% capacity factor; the reason the usa has an overall lower solar capacity factor of 21% is that some solar farms are in suboptimal places like maine (10%) so the power doesn't require long-distance transmission

so right now it's really tough for nuclear to compete with solar on earth

If you want to scale to 100x power you're going to have to rely on solar power even more than we already do. You seem to be awfully attached to the idea that the reactor must be located on earth. Solar power is fusion power with the reactor being located in space. It is very unlikely that humanity can build a bigger reactor.

humanity, defined loosely, can definitely build a bigger reactor than the sun. the milky way is a trillion times bigger than the sun, and mostly made up of stars (as opposed to large black holes, which are probably effectively inaccessible), so there's plenty of material available

already-existing natural blue hypergiants can reach energy outputs several million times that of the sun, in large part because they're on the order of 100 times bigger, usually limited only by the eddington mass limit. bat99-98 is estimated at 226 solar masses. so designed artificial stars can clearly reach that size, and conceivably, with a better understanding of plasma dynamics, they could be stabilized. in fact, we already know† how to build an even larger star: if you build a star of very low metallicity (similar to natural population-iii stars, of which possibly none survive today), its eddington mass limit is much higher, around 1000 solar masses

more likely, though, the humans will instead build a larger number of smaller, safer reactors. microscopic black holes can convert mass into hawking radiation at manageable photon energies and useful power levels. the necessary experimentation poses no risk of creating a large black hole (the density of matter necessary to grow small black holes to macroscopic proportion doesn't exist outside of the cores of stars, and the necessary quantity of matter at those densities is also literally astronomical) but will surely involve many explosions as starving black holes explode in a final tantrum of high-energy gamma rays, and of course must be carried out in free fall to prevent your nascent black hole from simply falling between the atoms of your laboratory floor before exploding deep inside your chosen planet

constructing larger reactors, by contrast, does pose a risk of producing phenomena such as disappointing white dwarfs, neutron stars, and black holes, or worse, supernovae, rather than a useful power source

if we believe dyson's calculations, though, a much more worthwhile thing to do is to figure out how to slow down our entropy production enough to preserve life into the cold, dark post-stellar era

______

† i mean we know in scientific terms what the structure of such an artificial star would be, where to find the materials, and what would be required to bring them together in the right way. it's fairly simple, actually. the only difficult part is getting a large enough budget to build the necessary fleet of spacecraft to harvest 10³³ kg of hydrogen and helium, about a billionth of the milky way, and bring it together over a distance of several light years; plausibly you need on the order of 10³⁵ spacecraft, about 120 doublings of a von neumann probe

In that transmission is very expensive and inefficient in ways that are challenging to improve because of physics? Transmission isn't getting a lot better very quickly, and certainly not by waving a magic wand.

transmission is in fact getting a lot better very quickly, becoming much less inefficient, because people rose to the challenge; i took some notes about recent hvdc systems in https://news.ycombinator.com/item?id=40725189

i edited that into my comment while you were writing yours, yeah. norway and finland have capacity factors of 6% for solar (compared to 30% here and 21% in the usa) and north of the arctic circle you'd need enough batteries to power you all winter long (though if i did my calculations correctly, it turns out that total yearly insolation rises after you cross the arctic circle)

this is why scandinavian countries are using a lot more hydro and wind than solar. wind and especially hydro are cheaper even than solar, but wind and especially hydro are much more limited resources

still, consider a counterfactual where somehow norway had to satisfy all its energy demand from solar. they burn 78.8 million barrels of oil per year, which is 15.3 gigawatts. they consume 124.29 billion kilowatt hours of electricity per year, which is 14.18 gigawatts (already over 95% hydro and wind, with no significant nuclear or solar component). they burn 3.98 billion cubic meters of natural gas per year, which at .0364 megajoules per liter is 4.59 gigawatts. all of these together (assuming no overlap) are 34.1 gigawatts, and over 5.55 million norwegians, that's 6.1 kilowatts per norwegian. (if that sounds like a lot more than your house uses, that's probably more because most of it is used by transport and heavy industry than because norwegians have to heat their houses more.)

34 gigawatts divided by a 6% capacity factor is 570 gigawatts of solar cell nameplate capacity you'd need. at 02021 german costs of €0.60 per (nameplate, peak) watt of utility-scale solar, (including modules, inverters, permitting, inspection, customer acquisition, etc.; see slide 48 of the fraunhofer deck linked above) this would cost €340 billion, about 8 months of norway's gdp. if you could only spend 5% of their gdp on the transition, you could get it done in about 15 years, maybe 25 if you need batteries

of course it's unnecessary because norway is already renewables-powered! but my point is that solar cells are now so cheap that they're a viable power source even in ridiculously polar countries. incidentally, they now cost half what they did in 02021, so the price would be a lot less now

antarctica is much more interesting than the arctic, though, since there's land there

Just curious where "here" is. South Africa?

sorry, argentina

Kind of curious why you’re worried about Antarctica?

oh, i'm not worried about it; i'm excited about it. but solar power is probably not a good way to power cities in antarctica

Where solar isn't adequate wind is often strong.

Per capita energy (substitution method) from fossil fuel:

- 2003: France 55% ; Denmark 91%

France was betting on nuclear, Denmark bet on wind.

- 2022: France 54% : Denmark 57%

https://ourworldindata.org/grapher/per-capita-energy-source-...

I'd agree. Clearly solar is sun-dependent, and other options may be more suited to other locations.

That said, electricity is one form of energy that is really cheap to distribute (once you are close to "the grid".) Which means that moving electricity from sunny places to darker corners is very practical.

There are some economics in play. Remote populations (like on islands, or in Alaska) are unlikely to be connected anytime soon. These populations are generally dependent on coal.

Nuclear might be appealing, but refueling nuclear takes time. So if it's your only electrical source, you need two reactors, each capable of supplying enough power. And of course the expertise to run it (expertise possibly harder to find in far-north Alaska.)

Ultimately diesel is likely to remain the fuel of choice for small communities because engines are cheap, fuel is easily transported, and expertise is local.

I think you're underestimating the costs of transporting power long distances.

There are losses, sure. And yes it requires infrastructure to be built.

Unlike an oil pipeline there's no possibility of spilling, or blocking wildlife.

And the costs are tiny compared to trucking oil or coal around.

Refueling is why you build more capacity than you need. It’s not like you have to “refuel” every night. It’s also not 1 for 1 (1 running, 1 “waiting”)

It kinda is though because refueling is a long job. Typically in the order of weeks. I'm not sure most communities would enjoy running on no power, or limited power for some number of weeks every couple years.

And sure, maybe "fast refueling " gets built into new designs. A day or two is likely the longest you can tolerate though.

Obviously that's before we consider other concerns of being dependent on a single source. I'm not sure living in remote Alaska, being dependent on a single complex machine for power is an ideal situation.

incidentally i don't think it's true that engines are cheap; diesel-engine generators seem to be on the order of 100¢ per watt, while pv modules are 8¢ per watt and falling. at least, at the few-kilowatts level where i've been able to find prices. maybe wärtsilä's multi-megawatt diesel power plants are cheaper per watt but they don't seem to publish a price list

maybe you just mean that diesels and especially otto engines scale down to smaller scales than nuclear or coal power plants?

Even in the southwest this is a problem - peak output from residential rooftop solar is mid-day but peak demand is early evening when people are getting home, cooking dinner, charging their EVs, etc. Without some major changes to residential storage or large grid-attached storage systems, solar alone isn't going to cut it.

Don't get confused by peak net-demand i.e. demand that is left after you subtract solar.

The newly cheap midday power should be driving demand towards that time period but it won't show up on net demand charts, which will show the famous duck curve belly at that time. That doesn't mean people aren't using energy at that time.

In fact the main way to detect this demand shift in a net demand graph is to see that the evening peaks are dropping year over year, even as they seem relatively larger compared with net demand when solar is working.

But anyway, in California the massive deployment of batteries in the last few years has already solved this problem. Compared with peaking gas plants that only run once a day for a few hours batteries are both cheap and clean.

The bulk of humanity lives between 45 north latitude and 15 south latitude.

http://www.statsmapsnpix.com/2021/11/world-population-by-lat...

> You do realize solar isn't viable in norther parts

I lived in the Yukon for 4 years, would drive many hours due south to get to the capital of Alaska.

I met dozens of people that had off-grid houses powered entirely by solar. Remember the sun is up for 20+ hours a day in summer. In winter they had to be careful, but the had full houses with washing machines, fridges, etc. etc.

This was in 2015 when people had lead acid batteries and only a few kW of panels. I bet it's much more common now.

Reply on news.ycombinator.com