126 Comments

I really appreciate your optimism. It’s a rare commodity these days. Unfortunately, it doesn’t seem as contagious as cynicism and outrage. Thanks for the post.

Expand full comment

Genuinely exciting developments! When it comes to energy sources, especially clean ones, I say the more the merrier. Hopefully we’ll get nuclear in on this hot green on green action, and kick clean electrolysis in overdrive.

PS - We’ll see if I regret using my excess solar energy for anything other than green electrolysis. ;-)

Expand full comment
Oct 2, 2022Liked by Noah Smith

Conversion energy efficiency of electrolysis for hydrogen is about 80%. Hydrogen liquefies at a far lower temperature than any other element, close to absolute zero, so liquifaction for liquid storage is not an option. Energy density in the gaseous state is less than that of hydrocarbon gases.

As a footnote, the production of hydrogen by electrolyis also produces oxygen as a byproduct, so if it does catch on, LOX prices will be dirt cheap. A small, but nice, bonus for the economy.

Expand full comment

Helium actually has a lower boiling point (-269 C) than hydrogen (-253 C), and liquid hydrogen is made and used for some things, like the Artemis spacecraft. You're right that it's probably not practical for ordinary power storage, though.

Expand full comment
Oct 2, 2022·edited Oct 2, 2022Liked by Noah Smith

Thanks for writing this. If we can make green hydrogren economical, then it opens a lot of possibilities.

Assuming we can produce it cheaply, the main downside of hydrogen is that it's a gas (and not easily liquefied), which makes it hard to store and transport. One solution is to convert hydrogen into ammonia, which is comparable to propane in terms of how easy it is to liquefy and hence store/transport.

I wonder if we will see ammonia used as a fuel for transportation. There are already ammonia-powered tractors. These are convenient because farmers already have ammonia available for use as a fertilizer.

Battery-powered cars are convenient because we have an electricity distribution system. But if ammonia were available at every gas station, then this advantage may disappear.

Expand full comment

No green Methanol is a far superior fuel to Hydrogen or Ammonia. The Nobel prize winning chemist George Olah described the optimal way to replace fossil fuels would be with the Methanol Economy. He described that in his book: Beyond Oil & Gas, The Methanol Economy.

Expand full comment

Yep, methanol is more convenient than ammonia.

The challenge with methanol (or ethanol) is that it contains carbon, so you need a source of carbon to manufacture it. Other than fossil fuels, the only source is the atmosphere. But the atmosphere is only 0.04% carbon dioxide, so it's a challenge to extract it economically.

In contrast, ammonia is just hydrogen plus nitrogen and the atmosphere is 70% nitrogen.

Expand full comment

Are you kidding me? We are dumping 10MT of carbon into the atmosphere every year, mostly from fossil fuels and biomass. The mistake you are making is thinking that transitioning to a low carbon world will be a chemical fuel based economy. That's not possible. The EROI of a non-fossil chemical fuel based economy would be too low to be physically possible. You would collapse the World economy long before you could achieve that.

Everyone is already talking about an electricity based economy. For transportation moving to BEVs, nuclear powered large shipping, nuclear large scale process heat, heat pumps, nuclear district heating. So there won't be much need for chemical fuels. So there is ample stupidly wasted carbon on idiotic biomass power generation, idiotic energy negative agrofuels like ethanol & biodiesel with a carbon efficiency of <10%. You can easily make methanol with a carbon efficiency of 100%.

You have vast amounts of forest overgrowth, which is increasing with high CO2 in the atmosphere, that just ends up back in the atmosphere in destructive and toxic forest fires. You can harvest that economically and convert it to methanol in tractor trailer sized portable plants. You can use flue gas, large amounts from cement kilns. You can use the huge amounts of carbonaceous waste going to landfills. And you wouldn't use atmospheric CO2, ocean CO2 is 50X that amount and much easier to extract.

So in conclusion there is vastly more recycled carbon or substituted carbon available than for the amount of liquid fuels we need. A non-carbon chemical fuel is just no advantage whatsoever. Although some locally produced hydrogen might be used for process heat.

Expand full comment
Oct 2, 2022Liked by Noah Smith

Looking around for mfg companies for electrolyzers I ran across this blog:

https://cicenergigune.com/en/blog/electrolyzers-manufacturing-industry-everyone-lead

which seems to agree about exponential growth but not about why.

Expand full comment
Oct 2, 2022Liked by Noah Smith

We should not dismiss the enormous potential in using hydrogen for long distance air travel. The fuel consumption of an aircraft is fundamentally a function of its weight and hydrogen is as good as it gets for energy weight density. Infrastructure is only needed at relatively few locations (airports) and cook-off shouldn't matter as much as for say ships.

Expand full comment

The same argument is true for rocket travel but more so by a factor of 10X. Added fuel weight is directly taken off of the 1% or so of lift-off rocket mass that makes it to orbit. So H2 has a giant advantage. And yet Musk & Bezos looked hard at fuels for their new rockets and went with methane instead. There is just so many problems with hydrogen. It has shut the SLS down from launching the past couple months due to hydrogen leaks. It is the leakiest fuel, and easiest to ignite.

Expand full comment

Methane might have been chosen because of the viability of manufacturing it on other planets for example from Human waste.

Expand full comment

That's true for Musk vs using RP-1 as the Merlin engines use. But you can manufacture hydrogen even easier than methane just from water. And Bezos also chose methane, he doesn't care about Mars. Musk also stated NASA made a big mistake using hydrogen in the Shuttle and its progeny the SLS, that's why they are getting delay after delay, screwup after screwup with potentially catastrophic hydrogen leaks.

Expand full comment

Not quite. A rocket will carry its own oxygen, which makes up most of the propellant weight. And propellant is (currently) a small fraction of launch costs, unlike air travel. I think there are also fundamental advantages of H2 fuel cells, such as flying faster and higher than air breathing engines. But who knows, maybe LNG will be the preferred solution for air travel too.

Expand full comment

That's not the way it works. Launch a rocket to orbit fuel is most of the rocket takeoff mass. But the payload weight is only a few percent of takeoff weight. So every pound you save in fuel is an extra pound to orbit which is big bucks. It is nowhere near that significant in aircraft where payload is a large fraction of takeoff weight. It is not a fuel cost issue, it is a mass to orbit issue.

Expand full comment

Right and wrong. Obviously liftoff weight is critically important. However, halfway through your burn, half of that weight is gone. Had that fuel been 1 for 1 replaced with cargo, you would shortly be plunging back to earth.

Expand full comment

Yes that's true since usually 2 stage rockets are used. So that means 1kg less fuel on launch gets 1kg more payload to Stage Separation point, and similarly using 1kg less fuel on the 2nd stage will lift that same extra 1kg to orbit. i.e. the Falcon 9 uses RP-1 fuel for both stages, they could instead have used hydrogen on both stages (same engine) and gained extra payload to orbit, theoretically, until you take into account all the difficulties with hydrogen fuel.

Expand full comment

Getting a reliable H2 supply for airports far from port in developing countries would be tricky, and H2 airplanes would need to be a global solution if you want to avoid a two fleet solution for long distance.

Expand full comment
Oct 2, 2022·edited Oct 2, 2022Liked by Noah Smith

Thanks, this is great overview and since David Roberts has blocked me appreciate update.

In long run, how does hydrogen storage on it's tearning curve for electrolysis and storage compare to "overbuilding" solar with its very impressive learning curve?

That is, it keeps happening where it's so often cheaper to build more solar than needed for most solar abundant times of day/year, to have enough solar at low solar times, or just cheaper to build solar panels facing different directions so they are more efficient in low solar times, sacrificing production in high solar times but that sacrifice "curtailment" still nets out to low cost electricity over whole year.

Seems it will depend on location (north vs south) and in long-term, what the delta of the hydrogen and solar learning curves are

Expand full comment

Hydrogen storage is going to struggle with learning curves. It’s basically NG storage but harder, and has a hard floor of 4x NG storage due to volume differences, and NG storage is VERY mature.

Expand full comment

The learning curve could be in material storage.

Expand full comment
Oct 2, 2022·edited Oct 2, 2022

For applications you dont want a 3000+ psi cylinder or LH2, sure. Long term. Storage, no. We have rough ideas on the bounds of this approach by chemistry.

Expand full comment

I am more talking about storing hydrogen in salts for instance. I saw this today, there is still a lot to learn and improve. https://pubs.acs.org/doi/full/10.1021/acscentsci.2c00723

https://interestingengineering.com/science/salts-solve-problem-hydrogen-storage

Expand full comment

Add conversion of hydrogen into ammonia as well. The conventional Haber-Bosch process is unsuitable to deal with the fluctuating nature of most renewable supplies, as well as challenging economy-of-scale if this process is scaled down from thousands ton-scale to that of distributed plants scaling several dozens ton per day.

The good news is that several renewable-compatible Haber processes are in the pipeline and well within reach of commercial application relatively soon. I think it will also go mainstream soon enough and follow similar learning curve to that of PV, onshore/offshore wind, and electrolysis.

Expand full comment

I'm very skeptical of small chemical plants. They currently are the scale they are for very good reasons, and for ammonia, it is typically not the end product, instead it is an intermediate, which you might as well size the Haber-Bosch plant for the site use.

And ammonia for fuel in all but dire emergencies is bad on so many levels.

Expand full comment
Oct 2, 2022Liked by Noah Smith

We’re seeing these learning curves more and more. I wonder if they should be accompanied by rate of deployment plots since that is their fundamental input? A big question: Is rate of deployment constrained by factors other than cost. Jewell, Cherp et al take a hard look at this https://twitter.com/acherp/status/1417373405960155149

Expand full comment

Storage is the big question. Right now it appears salt dome caverns is the only viable <10$/kWh approach, and they are geographically limited. If we are really lucky and they work well we may get that to ~$1/kWh.

Old NG wells are not suitable, unlike for storing NG (contamination, permeability), and there is a hard floor of 4x the cost of storing NG due to volumetric differences, and NG storage is VERY mature.

LH2 is much harder to make, store, and transport than LNG, so we should not expect that to be significant.

H2 is very necessary for decarbonization, but we should not expect a significant trade in H2. Instead the users will need to be near where it is made and stored. This is bad news for Germany’s industrial base…

In genes rap it is better to store H2 for user, rather than power. That should be an extreme backup, instead we can divert the power to make it for direct power most of the time when needed if the electrolysers are cheap enough.

Expand full comment
Oct 2, 2022·edited Oct 2, 2022

Yes, this is exactly what I was going to comment on. Electrolysis looks at first glance like a great place to dump excess energy, but the sticking point is exactly how we store massive amounts of H2, without leaks, for months at a time, and without having the occasional Hindenburg-style accident. There has been some promising research on creating an affordable metal hydride matrix, so the hydrogen doesn't react with its containment vessel or otherwise just leak out, but my understanding is that so far, the energy cost for getting the H2 back out of the hydride is a non-trivial percentage of the total energy you stored.

And then there's the question of how you convert back. I suspect high-temp fuel-cell membranes are going to be better than turbines. But whatever you use, you just have to build a LOT of them, and that poses the same kind of logistical challenge as any of the other aspects of the energy transformation we need.

I suspect in the near term, the most obvious application of excess renewables is going to be desalination. You could pretty much solve the Western US mega-drought (and same for Australia), if you were willing to pump desalinated water inland to places like Lake Shasta. (And yes, there are some environmental challenges with this as well -- we need to be careful about the temperature and concentration of the brine that's pumped back out, to avoid causing some kind of massive dead zone off the north coast. But if you really have massive amounts of excess energy, you can just accept running the desalination a little less efficiently and getting less water out of each cycle than you could if you optimized.)

Expand full comment

Why would you do that when Green Methanol is vastly superior to H2 as a fuel and the easiest fuel to store. There is no way on Earth that H2 can even come close to competing with green Methanol as a clean energy fuel. Methanol is being suppressed as a fuel, why? Because it really does work unlike the scams agrofuels & H2.

Expand full comment

Green methanol is superior as a transportable fuel. But the process to make it is a lot more expensive than an electolyser. So youvare storing the H2 locally to buffer the facilities that make things like methanol. And steel and ammonia etc. Separate the low CAPEX variable step from the high CAPEX, needs to be consistent high CF step.

Expand full comment

That depends on how you make the methanol. Straight from biomass is going to be less expensive than green H2, especially wind/solar H2 which will be very expensive. Methanol made from seawater CO2 & electrolysized water will be more expensive than just H2 because you are also making H2 plus extracting carbon. Local production of H2 for process applications that is piped to the process may be economical but that will always be a small niche application. Green methanol is far, far more versatile than H2 as well as being vastly easier and safer to handle, transport and utilize.

Expand full comment

IIRC methanol burns significantly cooler than H2 -- like, around 2000 Kelvin at best. H2 can run closer to 3000 Kelvin, if you feed its combustion process a greater proportion of oxygen than what's available in ambient air. So I can imagine methanol having some applications, but it's definitely not going to sub in for the things H2 is listed as being best-suited for in the main post.

The tech I think is really interesting for high quality heat is Heliogen's tech-assisted solar-concentrating reflectors. https://heliogen.com/

It takes up a fair amount of area, and it's subject to variable availability, but it probably will end up being MUCH cheaper than H2, and so at least will beat out H2 for high-quality heat in places like the American Southwest, and Australia.

Expand full comment

A lot simpler to just use high temperature molten salt reactors, which are more economical and general purpose.

Expand full comment

When H2 can be produced anywhere there is water and sunlight, the transportation problems should not be a big deal.

Expand full comment

If the point is to move energy from the summer to the winter, you have to store a LOT of hydrogen, for months at a time. If you produce H2 all summer, by early fall you're going to be sitting on enough H2 to incinerate not just your house, but all of your neighbors' houses too -- especially if each of them is also sitting on a similar quantity.

So no, we can't just produce this stuff anywhere, or at least not if we don't develop a drastically better storage technology. It'll have to be produced at central, high-security facilities.

Expand full comment

You have that problem with any energy storage. It's not like gasoline or kerosene can't catch fire or explode on occasion. Europe is building up its LNG stocks for the upcoming cold winter without Russian supplies and there are enough terrorist blows up LNG storage facility, or at least tries to, techno-thrillers out there.

Expand full comment
Oct 10, 2022·edited Oct 10, 2022

You're not wrong.... but hydrogen is just much more difficult to keep sealed up, compared to a liquid fuel, or even compared to other gas fuels like methane or propane. I don't completely understand the physics -- I think it's an over-simplification to say that it's just that the molecule is smaller and so you need a better seal. But you can find plenty of articles about this issue.

https://www.twi-global.com/technical-knowledge/faqs/what-is-hydrogen-storage

https://www.energy.gov/eere/fuelcells/hydrogen-storage-challenges

Expand full comment

Again that is an argument for methanol, the easiest fuel to store and transport. While being environmentally benign.

Expand full comment

Even accepting that as fact, many shorter term time shifts are very possible.

An obvious example is timeshifting solar energy from the day to the evening, which addresses a real problem.

Expand full comment

In the late 90s, the US national HS policy debate resolution one year was that the US should invest in renewable energy. My team's affirmative case was about hydrogen batteries. At that point, most of what we had to back it up was some basic R&D and conjecture about the cleanliness of the technology.

If only I'd had the capital as a 15-year-old to put my money where my mouth was.

Expand full comment
Oct 2, 2022Liked by Noah Smith

Excellent stuff. Harking back to one of our other (free) postings, is there a learning curve for liberal democracy?

Expand full comment
author

No, because it has to be mass-produced.

Expand full comment

Great read, thanks.

Think the issue with this is:

1. It's only truly "green" if it uses surplus green electricity. In most places, we're decades away from being in that position (just look at Germany today - how much they've spent on renewable energy to date and how far they are from having a green grid). Until we get to that point, "green" hydrogen is just displacement of emissions.

2. Utility solar isn't zero emission. The lifecycle emissions are around 50g/KWh. Once you account for the system loss of using Hydrogen (say, around 50%), that gets you to around 100g/kwh. A Tesla model 3, driving on a highway, consumes about 80kw/h per 200 miles so 4kg of CO2 per 100 miles. It's about twice as efficient as a small European ICE car. It's great, but doesn't get us to net zero (and even less so if you include the emissions linked to producing the car).

3. Would love for someone to actually do the maths on just how much land we'll need to produce enough renewable electricity to cover full electricity needs today + electrify most of the fossil fuel use cases. My understanding is that to get there would require 1/8th of the total land area of the lower 48 states.

Expand full comment

Utility solar is higher emissions than coal. Because the inefficiencies of using solar electricity on a modern gas/coal/hydro/nuclear grid means it wastes as much fuel as it potentially could save. Similarly true of wind.

Expand full comment

This is laughable. Bringing 1GW of additional solar power up today, can allow 1GW of coal to be taken down. That will not always be true. But today, excluding carbon ROI (which is 1.6 years for solar and 8-9 months for wind), utility solar is still zero emissions.

Where are these imagined solar emissions you speak of?

Expand full comment
Oct 7, 2022·edited Oct 7, 2022

The reality is we are not seeing this supposed solar emissions savings in Real World grid integration. In fact the Bentek study showed emissions INCREASED following the big wind builds in Texas and Colorado. The grid integration inefficiencies of adding an intermittent, seasonal, unreliable electricity source pretty much eliminate the theoretical fuel savings (and that's all they save) of solar or wind power.

After over $4 trillion spent worldwide on wind & solar, combustion fuels remain @ 90% of World Primary energy, unchanged before that massive rampup done in the past 10yrs. You have to account for all the huge energy inefficiency losses that intermittent wind/solar cause i.e.: Curtailment. Overbuild. Cycling Inefficiencies induced on the shadowing fossil fuel generation. Economic forcing of low efficiency diesel & OCGT instead of high efficiency supercritical coal, CCGT & hydro. Long distance 3-10X oversized transmission. Extremely low EROI (energy return on invested). High materials inputs, >20X nuclear/gas/coal. Vast losses of productive land, >300X nuclear/gas/coal.. Huge waste recycling energy cost. Creating 2 grids which must run in parallel. Vast embodied energy in battery storage. 70% energy losses of hydrogen backup/storage.

As further evidence, a survey of 68 nations over the past 52 years done by Environmental Progress and duplicated by the New York Times shows conventional hydro was quite successful at decarbonization, nuclear energy was also very successful and both wind and solar show no correlation between grid penetration and decarbonization. In other words wind & solar are not replacing fossil, they are a complete waste of money. They only succeed in increasing energy prices which does reduce emissions only by creating energy poverty.

Expand full comment

I am not sure if you believe this stuff or if you are just selling a story and well aware that it is false. I am not going to go thru your entire post but lets start at the top. FWIW, I am a power semiconductor design engineer.

1) Transmission lines must be oversized by 3x. This is silly. Transmission lines are sized exactly to the size of the energy provided. Yes, the peak is above the average. Your claim is the same as saying that if we shut down a coal plant, then the lines were 2x oversized because the plant was functional for 2 years and offline for 2 years so the average was half.

2) I mentioned this in the previous post. Today this is not the case. Today the U.S. only gets about 10% of our energy from Wind+Solar and you are not cycling anything on and off. You are just modulating it's output by 10%. Some day we will get a more substantial amount of power from bursty sources like wind and solar and there will be losses associated with cycling

3). Carbon ROI on wind and solar is 8-9 months and 1.6 years. This includes, extraction and manufacturing of raw materials, production of the turbines, their transport, erection, operation, maintenance, dismantling and disposal, and the same for their foundation and the transmission grid. https://www.vestas.com/.../pdfs/lca_v90_june_2006.ashx

Apparently, you don't like renewables. But you were arguing carbon cost half your points are about monetary cost. Maybe you just copied and pasted from a previous post. Maybe you are being disingenuous.

If you are genuinely interested rather than just politically active in opposition, then you should be more skeptical of your sources.

Expand full comment

1) No they aren't. Transmission lines including associated switchgear are sized to carry the peak load of the power source. With solar the average output is more like 20% of peak. And the marginal savings of adding aluminum to the transmission lines is very low since the power is low value and you only save 20% of what the same amount of aluminum would save on a baseload power source.

2) No the US gets about 2.6% of its energy from Wind + Solar. 10% of its electricity. That's avg generation over a year. Peak wind + solar can match all demand on some occasions, it commonly is a major part of total generation at certain times of the day and times of the year. And does fluctuate, often severely. You can't match that with high efficiency baseload power sources. You need low efficiency diesel/gas/coal. Germany disconnects its big coal generators from the grid when wind/solar are high, so it doesn't have to count all the emissions of burning the coal to keep the boilers fully pressurized. The maximum efficient grid is one with CCGT, supercritical coal, conventional hydro and nuclear all running 24/7. Anything that interferes with that impairs grid efficiency.

3) You are ignoring the carbon effect on grid efficiency, taking that into account indicates an infinite period of ROI for solar/wind carbon. Vestas uses the most optimistic theoretical values for carbon inputs, not real world values. Weissbach did an full lifecycle analysis of wind buffered with pumped hydro and obtained the real world EROI of 16:1 for the E-66 unbuffered and 3.9:1 buffered and 3.9 for solar PV Germany unbuffered & 1.6:1 for solar/pumped hydro, 3.5 for biomass(corn). Ferroni found a EROI of 0.82:1 for solar PV in Switzerland (not incl buffering). Hall found an EROI of 2.45:1 for solar PV in sunny Spain not including buffering. Weissbach did not include curtailment, that will make those numbers worse. All of those are far below the 14:1 needed to sustain a modern civilization. He found hydro EROI of 49:1, nuclear 75:1 and CANDU nuclear is 120:1, the highest of any energy supply.Wind & solar are not feasible replacements for fossil & in fact are just locking us into a fossil fuel future actually called wind/solar fossil fuel lock-in.

No I like rational, efficient baseload renewables conventional hydro and geothermal. Intermittent, unreliable, seasonal wind & solar do not belong on the modern electrical grid. They can be practical in niche applications like a diesel grid, off-grid homes or remote sites.

Yes I should have posted about efficiency instead of cost, so I changed it. And I'm being scientifically accurate not disingenous. My sources are rock solid, I would not say the same for yours.

Expand full comment

1) Ok, so we agree then. Transmission lines are sized for load. Of course average power is less than half of peak. Half of the time the thing is in the dark. It is still true that this is exactly the same as saying, "We shut down the coal plant. It was up for 2 years and down for 2 years. Thus the transmission lines were oversized 2x."

In any case, we are talking Carbon improvements. The carbon payback period for solar is 1.6 years including operation, maintenance, dismantling and disposal, and the same for their foundation and the transmission grid. https://www.renewableenergyhub.co.uk/main/solar-panels/solar-panels-carbon-analysis/?fbclid=IwAR3VRfzg1-HkTdV-8fENB5duCLByYDPj1_o6N6R9Zp4PiVFjp1h2e-pUvi8

All that peak power is still zero carbon energy. There does obviously exist a level of bustiness where solar is not worthwhile. That solar is growing so quickly and coal is plummeting at the same time, suggests that is not the case. You can complain about it all you want but solar and wind have already won.

2) You do get that Germany and the U.S. are both capitalist countries. That companies don't do things for free. That if German energy companies are disconnecting from the grid, it is to save money. It saves money because keeping a boiler hot and up to temperature takes far less money than spinning a turbine.

If that was not the case, then the coal power companies would have just lowered their prices to keep solar from taking that share of the market. Or are you thinking that they are now just burning the same amount of coal as before but taking annual losses every year?

3) I don't care about these purported studies. There is so many bogus studies paid for by energy companies that they are not worth digging into. Explain where theses grid efficiency losses come from. The solar energy itself is zero carbon. Building the grid is paid for in 1.6 years. Solar doesn't increase grid maintenance. If the energy isn't making it to peoples houses then they wouldn't be decommissioning coal plants.

The rest of your paragraph is predicated on some imaginary 14:1 "needed to sustain a modern civilization." This is just made up garbage. Hydro is great. Nuclear is great (but expensive). Solar and wind are so cheap and effective that nobody can afford to build any of the alternatives.

It is scientifically impossible for a solar cell to have a carbon ROI of oo. Does it take an infinite amount of carbon to produce it? We say solar cells have a fixed lifetime (usually 20-25 years). That is, however, the point at which 10% of them see a marked degradation in performance. Many will continue to function nearly indefinitely.

Anyway, I am done with this conversation. I don't really care that much about the issue. On the other hand, I do really object to the presentation of embarrassingly bad made up science.

Expand full comment

Whilst I agree with you that hydrogen will be an important piece of the puzzle (especially in industrial use), the issues with leakage are very material, as highlighted elsewhere. I work in infrastructure and energy financing, and refitting the natural gas grid is extremely demanding. I'm also skeptical of large scale storage for this same reason; the leakage issues (due to molecular size) both make long term storage more difficult *and* make transport to where it's needed harder.

A note of caution on batteries: the learning curve is happening on the non-commodity part of the cost. However, over time, the commodity cost is becoming an ever larger part of the battery cost, as anodes/cathodes/etc get cheaper and more efficient. Many people are working on non lithium/cobalt batteries, and I wish them the best of luck, but if we don't have a breakthrough there, using the current commodity structure *will* lead to a flatter learning curve (as if 90% of the battery cost is the lithium, making the tech part half as expensive only gets you a 5% savings).

Take the two together, and I think storage is going to be one of the hardest nuts to crack of the transition, and will incentivise a lot of overbuild in the medium term (or continued use of natural gas plants).

Expand full comment

I had a similar concern regarding solar PV -- as the cost of the expensive learning-curve parts of a PV panel drop, the non-learning-curve parts, such as the steel for the frame and the glass covering -- become a greater fraction of the total panel cost. Is it possible to look at the sub-elements of technologies to try to figure out where the learning curve effect might bottom out?

Expand full comment

This really deserves an essay-length treatment, but short answer is that I'm not as concerned about this regarding solar and wind because the raw commodity costs are a small proportion of the total cost, such that if we got to the point where commodity cost overwhelmed the tech cost improvement calculation, these energies would be extremely cheap already.

Calculating the exact input cost is difficult though, because how far back do you go to determine an 'input', so you'll have limited luck finding comparable figures.

Expand full comment

To illustrate: Using London Metal Exchange prices, Lithium costs about USD 80 per kg. Steel is roughly 80 *cents* per kg.

Expand full comment

I would quote Shellenberger: "...Solar and wind energy projects require roughly 300% more copper and 700% more rare earth per unit of energy than fossil fuels. Wind, solar, and batteries require 1,000% more steel, concrete, and glass; 300% more copper; and 4,200%, 2,500%, 1,900%, and 700% more lithium, graphite, nickel, and rare earth, respectively, than fossil fuels, to produce the same amount of energy, according to International Energy Agency and others.....

China’s market share of renewables and EV minerals is already twice OPEC’s share of oil, notes Mills, drawing on data from the IEA and others. The U.S. depends on imports for 100% of 17 renewables and EV-critical minerals; for 28 others, imports account for more than 50% of domestic demand. China already dominates solar and battery production. Minerals are 60%–70% of the cost of producing solar panels and lithium batteries...

Attempting to produce solar panels, batteries, wind turbines, and the materials required to produce them in Western nations will make them prohibitively expensive. China's solar panel labor costs have been very low to free, considering that they have been covered as part of its ongoing genocide against Uyghur Muslims in the Xinjiang province.

Meanwhile, the capital costs of solar, wind, and batteries have been rising since 2017 and will increase much more. Energy today only uses 10%–20% of total global minerals, but IEA says its share must increase to 50-70% for the world to transition to renewables. “In normal times,” writes Mills, “energy typically accounts for just under 10% of the cost of most products and services. Doubling the energy cost will have an inflationary impact on the average final price tag for all products and services.”

https://michaelshellenberger.substack.com/p/end-of-renewables-craze-is-near

Expand full comment

There is one thing that has been missing in all these discussions. My brother tested and built hydrogen systems about forty years ago (production of hydrogen from water -believe it or not - in the Mohave Desert), hydrogen powered aircraft engines, etc.

There is a major problem with all this. Hydrogen is, I think, the smallest molecule in chemistry. And it leaks through any standard valve or pipeline connector. I have yet to hear any discussion of the future of hydrogen mention this problem or provide any information about steps taken to reduce leakage in the proposed systems.

Expand full comment
author

You can use the hydrogen to make methanol from air, that's easier to store.

Expand full comment
Oct 2, 2022·edited Oct 2, 2022Liked by Noah Smith

Extracting CO2 from the air to make liquid fuels will always be expensive as it makes up only 0.04% of air by mass. Seems more likely that we end up using ammonia (NH3) as a liquid fuel since N2 is 78% of air making it far cheaper to extract. By my very rusty high school chemistry there are 6129x more N atoms in air than C atoms (albeit it for only 3/4 the H atoms in the final product.)

Expand full comment
author

Yeah we could do that too

Expand full comment

You don't need to use CO2 from air. You can use stranded or flared gas, flue gas(i.e. from cement plants), seawater CO2 or any carbonaceous waste. Forest overgrowth can be efficiently harvested and converted into methanol in tractor trailer sized plants, greatly reducing severe forest fires with all their toxic smoke, hazards to human & animal life and vast CO2 emissions.

The DOE built a demo IGCC coal power plant that could coproduce methanol for 50 cents/gal. The Luigi Mega-Methanol plants can produce methanol from NG for 6 cents/liter. And the NREL forecasts methanol from biomass large scale production at 50 cents/gal or 13 cents/liter. An optimized methanol spark ignition engine can substitute for a diesel engine at 1.5X torque/liter displacement, 40% more compact, ~10% more efficient with a much wider island of high efficiency than the diesel engine, as well as much lower emissions. And methanol burns at higher efficiency than natural gas in gas turbines. Methanol being the easiest fuel to store, with spills having minimal environmental effect.

Expand full comment
Oct 2, 2022Liked by Noah Smith

One alternative is to create the hydrogen from solar electrolysis and then immediately use the Sabatier process to combine it with CO2 captured from air to create methane on site and then use this “green natural gas” wherever natural gas is used now

Expand full comment

Making green methane would be far superior to hydrogen although you would not opt for CO2 from air, CO2 from seawater is much easier and more efficient, and it is idiocy to do any of that when we are stupidly burning biomass for electricity or using agrofuels, all that biomass could be converted into methanol or methane with 100% carbon efficiency.

Expand full comment

This is why consumers will probably never see hydrogen. It will be used extensively as an intermediate in industrial facilities where they do know how to handle it.

Expand full comment

Did you talk about conversion efficiency loss? Cost viability? Why not be techno optimist about nuclear?

Expand full comment
author

Yes, I did talk about both conversion efficiency loss and cost viability. Major sections of the post were about exactly those things.

Expand full comment

I am an investor in a company that has invented an electrolyser that is 95% efficient and removes significant balance of plant costs. When it hits mass production it will significantly bring down H2 production costs. Electrolysis and fuel cells are still not much more than cottage industries and have great productivity gains ahead of them imho, once the full weight of government support and venture capital is applied. Storage is a bit of a mess rn but there are multiple potential solutions from metal organic frameworks to metal hydrides, just waiting to get funded.

Expand full comment

It was an optimistic piece. I learnt a lot from this article as well. https://doomberg.substack.com/p/the-hitchhikers-guide-to-hydrogen

Expand full comment

Noah, I expect there would be a learning curve from nuclear as well if it got any of the attention that the myriad, mostly-failed-to-date technologies on which there are curves (praying they continue) get. There has been almost nothing significant done in nuclear for 50 years for reasons unrelated to science or technology but closely related to "green" politics (despite the fact that nuclear is more green than many such promoted approaches). So it is likely disingenuous to state it does not have a learning curve. If batteries/wind/solar/whatever were ignored to this extent, I am confident they would not either.

Expand full comment
author

In the podcast interview with David Roberts, Doyne Farmer addresses the puzzle of nuclear's lack of a learning curve, and offers some possible reasons.

Expand full comment
Oct 2, 2022Liked by Noah Smith

The big difference is probably the bespoke nature of nuclear builds vs the mass manufacturing of technology like solar panels and wind turbines. The great hope is for SMRs to take advantage of the latter to benefit from Wright’s law so we shall see.

Expand full comment
author

Yep. Mass production is what typically pushes costs down, and nuclear can't be mass-produced with current technology. Modular reactors are supposed to be the solution to this.

Expand full comment

There is a substantial learning curve with nuclear. Although the big factor is developing a supply chain and experienced workforce and contractors in an industry that has basically been shutdown for 40yrs. Even Modular reactors will not work if governments continue to ignore Nuclear as a climate change mitigation solution. The key to modularity is to mass produce. If nations refuse to build nuclear or only build token amounts then the gain from modularity will not materialize.

The most important thing for the success of nuclear is for nations to realize how badly wind and solar have failed. Then they will come to their senses and go on a rapid nuclear buildout, as was done in the 1970s.

Expand full comment
Oct 2, 2022Liked by Noah Smith

Yeah my impression from the Construction Physics article is that large nuclear installations have a sort of anti-learning curve where we keep learning about new things that could go wrong (earthquakes! Tsunamis! Russian invasions!) and needing to redesign to address them, which adds costs over time. I bet the development of large commercial airplanes suffers from similar issues if you look at the costs over time (the A380 took something like 25 billion). I too hope that smaller reactors will help since we will be able to build more of them and mistakes will be less catastrophic, and they could be another potential source of industrial heat.

Expand full comment

You do get a substantial learning curve with large Nuclear power plants. The key is to do what France and other countries did in the 1970's, just start building them dozens at a time. That ensures supply chains are developed and scale production efficiencies can occur. The South Koreans are very good at that. In the UAE, starting from scratch with zero nuclear expertise, zero trained workforce, construction crews and minimal industrial infrastructure they built for a total cost of $24.4B , 4 South Korean APR-1400's in 8 yrs for 5.6GW or $4.8B/GW-yr output @ 90% or 44 TWh/yr. That's 103% of Australia's current total annual wind & solar production. At $24.4B/$50B = 1/2 the cost of Australia's wind/solar. That's in an Apples (nuclear) to Rotten Oranges (wind/solar) cost comparison.

Expand full comment

There was a paper on the costs of the French nuclear effort published around 2000, and the costs were high and rising as the program progressed and they moved to larger reactors. The difference was that France considered reactors important to national security and the costs and overruns were not revealed until relatively recently. I had long been under the impression that the French had figured this out, but in some ways, it was like the atmospheric railway. The French managed to get it working, but it was more a point of honor than a matter of commercial success.

Expand full comment

> I expect there would be a learning curve from nuclear as well if it got any of the attention that the myriad, mostly-failed-to-date technologies on which there are curves (praying they continue) get. There has been almost nothing significant done in nuclear for 50 years for reasons unrelated to science or technology but closely related to "green" politics

I call BS. After all, nuclear's staunchest advocates have a stock rebuttal to anti-nuclear cries of "But Chernobyl!" — they point out that nuclear reactor designs have improved in the 40ish years since Chernobyl was built. And they point to Generation IV reactor designs, for which we have a stream of publications and design work stretching back decades. That contradicts the idea that "almost nothing significant [has been] done in nuclear for 50 years" (and never mind the apparent suggestion that wind and solar power are "mostly-failed-to-date technologies").

Expand full comment