Hacker News Reader

Comment by ben_w | Hacker News Reader

> to be anti-nuclear is to be anti-climate, and pro-coal.

While this was true even 10 years ago, not so much now. While nuclear is still cheaper than the batteries which would otherwise be the main[0] worldwide alternative to backing up PV/wind, it is close, and probable future development curves for renewables and batteries make it not completely insane to reject nuclear.

Personally I would still support new reactors, firstly because I don’t want to keep all my eggs in one basket if my expectations for future battery factories are falsified, and secondly because I think diverse solutions are likely to be cheaper and faster overall than monoculture solutions.

[0] Scandinavia in general might be amazing for hydro, which is much cheaper than batteries, but IIRC most places aren’t so fortunate.

View on HN

> While this was true even 10 years ago, not so much now. While nuclear is still cheaper than the batteries which would otherwise be the main[0] worldwide alternative to backing up PV/wind, it is close

The cost of nuclear is artificially high because special interests have lobbied hard against nuclear, draining western countries of the experience to build nuclear facilities rapidly and also taking away any opportunities for efficiencies of scale.

Moreover, I don't think you can look at the cost of batteries as-is. If we seriously invest in batteries as a solution for grid-scale power storage, the demand will go through the roof and I doubt supply will be able to keep up (at least not on the 10-50 year timeframes that the climate challenge requires) while we're also trying to electrify transport and other industries.

> Personally I would still support new reactors, firstly because I don’t want to keep all my eggs in one basket if my expectations for future battery factories are falsified, and secondly because I think diverse solutions are likely to be cheaper and faster overall than monoculture solutions.

I strongly agree with this. If climate change is an existential threat, then it's supremely foolish to give up on our only proven clean base load supply in favor of unsubstantiated hope for a miracle breakthrough in renewable storage. We should absolutely pursue both, vigorously.

> The cost of nuclear is artificially high because special interests have lobbied hard against nuclear, draining western countries of the experience to build nuclear facilities rapidly and also taking away any opportunities for efficiencies of scale.

Hinkley Point C is build in cooperation by China and France, the at the moment most active countries in building and maintaining nuclear power plants. It doesn't get more "economy of scale"-ly than this.

Smaller, safer, modular reactors should have been our focus decades ago. We are several generations behind where we ought to be, because we haven't prioritized it. The zero-nuclear agenda, combined with cheap natural gas and a (to me) strange fixation on renewables* has held us back.

*I say strange, because renewables have only just now started to get to make sense at scale, despite the reliance on fossil fuels to provide a base load, when we could have had a clean grid by now without them had we tried.

Modular reactors don’t actually reduce costs in any meaningful way. Construction costs are significant, but surprisingly low percentage of overall costs. For example nuclear needs a large highly trained workforce for decades. Every individual pulling 100k * 50 years is 5 million dollars, and you need a large workforce. Arguably security guards and medical staff are unnecessary, but modular reactors don’t lower those costs.

Fuel is similarly expensive and no in practice reprocessing doesn’t really help. Shutting down for refueling is again expensive when you consider it happens 30 times over the lifetime of a nuclear power plant. Decommissioning is extremely expensive even if it can but off. Even highly subsidized insurance is yet again expensive. Overall construction costs aren’t limited to the reactor you still need radiation shielding, cooling towers or the equivalent, cooling ponds for waste, roads, offices etc.

Don’t get me wrong if someone had come up with a great modular reactor 20 years ago Nuclear might be much better off today, but we’re simply past the point where more R&D is going to help. The only possibly for change is if nighttime electricity costs ended up much higher than current daytime costs. However a lot of demand is being shifted to nighttime because costs are currently cheaper, change that equation and nighttime demand will fall dramatically.

It certainly could get a lot more “economy of scale”-y than a single bespoke reactor project. Indeed, look into small modular reactors—they’re designed to be built at scale in factories (rather than individual bespoke plants). The levels ed cost of energy for these is projected to be well below $100/MWh which compares favorably with fossil fuel sources.

One unsolved problem with those small reactors is security. How do you prevent some terrorists from building dirty bombs with material stolen from a company's basement reactor? How do you trace the vast amount of radioactive material at so many sites?

One solution to it from mid-1990s was sealed reactors that are refueled by sending them back to manufacturer, fitting on single railway car. Combined with being rather hard to extract the material due to coolant used (liquid metal solidifying when shut down or leaking) and various anti-proliferation poisons used, it makes a pretty good solution.

I am dubious. Something that fits on a single railway car can be stolen in its entirety, and then the question is “is it valuable enough someone might just do that?”

It’s the same reason that puts me off houseboats: “thieves stole my house”/“my nuclear reactor has gone missing” should be Onion headlines and Pratchett storylines respectively, not things that actually happen.

individual fuel assemblies are much smaller, and by this logic easier to steal.

There's a reason why logistics for certain materials include security services, and it's much harder to abscond with multiple-tens-of-tons container that can have tracking beacon than a single fuel assembly.

One-off megaprojects are the opposite of economy of scale. Standardized production in large numbers allows you to amortize the costs of development and validation as well as maintain a healthy ecosystem of suppliers and employees who can compete for contracts.

> I doubt [battery] supply will be able to keep up.

Any tech that exists today can be made in almost-limitless quantities given a few years and customers willing to pay.

Eg. Say you want to make 100x the worlds current production of nuts and bolts. You just count up how many nut and bolt factories there are, how many steel works, how many mines, etc, and you multiply that by 99 and build them. Provided you have financing (which you will have if enough people are happy to buy nut and bolt futures ahead of time), they're all parallel projects so can be done independently, almost eliminating schedule risk.

The only time it doesn't hold true is when the need for a product is short term, unpredictable, unexpected, or where someone desires to pay less than market price.

We’re talking about enough batteries for the entire world’s total energy needs for weeks at a time. That means the battery industry needs to increase in size by many times it’s current state, which implies many new factories built and so on, never mind contention over natural resources. Meeting such demand seems likely to take many decades and untold trillions in investment.

You should look at the amount of battery plants that are being built right now. This is just in Europe. Economies of scale at work. https://mobile.twitter.com/ZennRoland/status/135382169441102...

An annual production of 600 GWh worth of batteries, that will be mostly used in vehicles. For reference, annual EU power production is roughly 2778000 GWh, it's a drop in an ocean.

Li-Ions are frankly a terrible solution for grid storage, pumped hydro is the only thing that makes sense on these scales.

The production will likely increase if the demand is there and we're not trying to replace the yearly power production. It has its downsides but it's easier to implement than huge hydro infrastructure projects. We'll see a mix of different technologies I guess, depending on the local possibilities.

We don’t need a year of storage, and any unit can be used at least 1000 times (I forget the exact number).

PV being as cheap as it is, it isn’t totally crazy to suggest so much overproduction that 24h of storage is enough. I expect not optimal, but not wildly so.

That reduces the 2778000 GWh to 7610 GWh, which only needs replacing at most every 3 years.

I’m supporting nuclear (and brought up hydro in the first place) because something might get in the way of scaling up battery production or prevent batteries from getting much cheaper (last I checked there’re more expensive per lifetime unit stored than nuclear is per unit generated), but it definitely isn’t crazy to think it will scale up to the required level and come down in price.

I don’t think you appreciate how small that is compared to what would be required for global grid storage. Using numbers from Wikipedia, the world uses something like 2200 TWh of energy every week. The figure I’ve heard thrown around for necessary grid storage is “weeks”, so we’ll say 1 week of storage is required. That means it would take Europe 3700 years to produce enough batteries to build the batteries we need in the next 50ish years. Or if you prefer, we need to produce an annual 44TWh of batteries a year globally for the next 50 years, which means we need 75-Europes-worth of production capacity. Further still, energy demand is projected to increase as the global population grows and as standards of living increase.

Even worse, scaling anything isn’t linear. It’s likely a lot easier to scale from 1GWh/year to 10GWh/year than it is to scale from 10 to 100. For example, in the first 10x scale, you might be able to pull a certain kind of skilled labor or raw material from other industries, but for the next 10x you now need to increase the global supply of said skilled labor or raw material.

> That means it would take Europe 3700 years to produce enough batteries to build the batteries we need in the next 50ish years.

It would probably help to show the working here:

600 GWh/year times 3700 years is 2220 TWh, but if you really do need them to last a week, 1000 cycles at one per week means they last about 20 years. As 1000 is an underestimate, this seems broadly correct to me given current battery construction.

However:

First, we will occasionally want them to last a week; most of the time you’d only want a day. This gives us longer to build out whatever solutions, more battery factories or reactors or whatever.

Second, it is very plausible the non-linearity of scaling will reduce costs rather than increase them. After all, that is what has happened so far, and as for workforce there will be a lot of coal miners available to switch to copper etc.

I’m not saying this is certain to happen, just that it’s at least plausible.

Third, my personal favourite combo with batteries is intercontinental HVDC — shift time zones or hemispheres and you don’t need to care about night or winter, but even much shorter links will connect you with a desert with a completely different climate and weather. Every 1 GW of capacity is 1 GW * (your desired storage duration in hours) of storage you no longer need (same for every 1 GW of nuclear, obviously). If we need a week of storage as you suggest, a single 1 GW line would save on 168 GWh of batteries.

I’m only expecting us to do HVDC in the order of length of e.g. EU to Sahara and therefore only limiting our storage requirement to about a day of storage, perhaps even as good as just overnight storage, but the longer the storage period we need the better relative value a planetary scale grid looks like.

(Downside: an antipodal ring big enough for world power use about a decade of worldwide aluminium production, or alternatively five decades of copper[0]. Again, this is why I prefer mixed approaches).

[0] One of my older comments did the maths for just Europe and just to the Sahara, so multiply the “3 months” and “about a year” in the linked comment by 10 for antipodal and another 5 for global electrical demand: https://news.ycombinator.com/item?id=28474201

> It would probably help to show the working here:

Fair enough, I was on mobile. According to https://en.wikipedia.org/wiki/World_energy_supply_and_consum..., the world consumed 9,717 Mtoe or 112,717.2 TWh of energy in 2020. Dividing that number by 52 weeks/year gives us 2,167 TWh--the amount of energy we need for a week's worth of storage. 2,167 is about 3612 times the amount of capacity that Europe is slated to produce annually (previously I rounded various figures before doing the math, so my number rounded up to 3700 rather than down to 3600).

But as previously mentioned, this doesn't account for growing energy demands or wear on battery stock or many other pertinent factors.

> First, we will occasionally want them to last a week; most of the time you’d only want a day. This gives us longer to build out whatever solutions, more battery factories or reactors or whatever.

My "week" figure is based on the assumption that we need to be able to manage with near-zero wind/sun for a week. While most of the time you won't need a week, we're not concerned about "most of the time", we're concerned about the worst cases. By the way, I have no idea if a week is accurate. I've just heard "weeks" slung around, so I assume that "one week" is the low end. I would be surprised if we could get this down to a day on average. Mind you, this doesn't mean we need 1 week of storage, but it means we need 1 week of something besides solar/wind.

> Second, it is very plausible the non-linearity of scaling will reduce costs rather than increase them

I agree. The specific dynamics depend on the market and the supply chain, but I'm pessimistic, especially after seeing the havoc covid continues to wreak on supply chains the world over.

> Third, my personal favourite combo with batteries is intercontinental HVDC

This is really interesting. I always been a bit surprised that electricity isn't more fungible from region to region, and wondered why this was never really talked about as a serious mitigation against regional weather patterns. I've been thinking about clean energy as Europe's key to energy independence, but it would be cruelly ironic if Europe came to depend on simply a different region for its energy (and also for Africa, if its sunshine had the effect that oil tends to have on poorer countries).

This is also why we are building HUGE reactors.

The bureaucracy for building a nuclear plant is pretty much the same whether you build a 1 watt plant or a 1 terawatt plant.

Thus, it's more cost-effective to build the biggest m-f:n plant you possibly can.

>Scandinavia in general might be amazing for hydro

I don't know. Here in Scandinavia the price of electricity is forecasted to sky rocket, due to lower output of hydro power.[0] If macro weather pattern are to undergo rapid change hydro power might not be the most reliable.

[0]

If you’re using it as a battery, which is what I’m suggesting here, you don’t need to worry about rain.

They are still fresh water rivers. So you still need some supply and flow.

You can't use it as a battery without large reservoirs of water you can pump into the input reservoir. Norway only has half the necessary infrastructure.

Not necessarily -- you can modulate output. The input reservoir gets refilled naturally if your average outflow matches the average inflow.

That is precisely what Norway does. Prices go up when average inflow is low. The biggest source of inflow is snow, which maps to outflow 0-18 months later.

Forgotten reference? But it would be a good thing since it incentivises using hydro mainly as variable power to support solar/wind and not as base power.

> [0] Scandinavia in general might be amazing for hydro

All the possible rivers are already used, and the plants have killed the fish from the rivers. The effect on wildlife is not that nice.

Re "While this was true even 10 years ago..."

A very remarkable thing is the decline in the cost of renewables just in the last 10 years.

The cost number have completely flipped in that time, not even counting the cost risk from delays, underestimated decommissioning costs for nuclear, the uninsurability of decommissioning costs, and the uninsurability of of nuclear accident risk. Solar and wind are cheapest now, by a multiple.

The parent is correctly factoring in storage costs. If you can't store solar or wind energy, then it can't be used for base load generation.

Even that is not true, since you can always combine electricity from multiple sources to form your base load, and there's no law of nature saying that the base load must come from a single source.

But even more importantly, the need for storage arises at much higher renewable penetration levels than you observe in the current world. It will take multiple decades to get to the point where you need large amounts of storage in the first place.

> Even that is not true, since you can always combine electricity from multiple sources to form your base load, and there's no law of nature saying that the base load must come from a single source.

No one is saying it must come from a single source? Combining electricity from multiple reliable sources is fine, but there will be times when the sun isn't shining and the wind isn't blowing so you need to be able to meet energy demands in those cases. The only clean solution is nuclear (and hydro where available).

> but there will be times when the sun isn't shining and the wind isn't blowing so you need to be able to meet energy demands in those cases. The only clean solution is nuclear (and hydro where available).

That claim is highly debatable. Lots of people claim that the storage necessary to overcome these generation troughs make it infeasible but they seem to completely ignore holistic cost minimization approaches using overgeneration and then conclude that the amount of storage required makes it impossible. Meanwhile it's been calculated that modest overgeneration squashes storage amounts (and costs) in massive ways. People have actually studied these things, you know...

See for example the 2018 article On the economics of electrical storage for variable renewable energy sources by Zerrahn et al.

We can't assume overgeneration will be sustainable since there is profit to be made by idling or scrapping generators to reduce costs.

We can safely assume that. First, if the deployed mixture of generators is the cheapest mixed solution devised, there's money to be wasted "by idling or scrapping generators", since by definition that would increase costs, not reduce them, so it would be an irrational step to make. Second, opportunistic consumers not included in the base load can take advantage of these peaks, to the benefit of both sides (generators AND consumers -- consumers get power for peanuts and generators get some extra money they otherwise wouldn't get), for example for charging BEVs on demand -- and you'll need lots of power for that.

Are you assuming grid-scale storage exists and is deployed widely in your scenario?

It's not like overgeneration is free either, and I'm very skeptical that overgeneration can squeeze blood from a stone on regionally cloudy, still weeks; however, I'll read your source when I have a free moment, thanks for sharing, etc.

Nobody says that overgeneration is free. But for example if your wind power costs are $20/MWh and your nuclear power costs are $80/MWh, even 100% overgeneration capacity for wind is still cheaper compared to the nuclear alternative. And if it achieves the desired effect cheaper than storage, then it will be preferred over storage as well. The optimal ratio of storage and overgeneration needs to be modeled for every grid independently; I believe the article I mentioned models it for the German grid.

In practice, though, I expect that you won't even get energy waste from overgeneration if flexible energy consumers like BEVs or flexible hydrogen electrolyzers are connected to the grid. These will account for a major portion of electricity consumption and it will be possible to move their consumption around very considerably.

I agree with your first paragraph, but I’m still far from optimistic that the cost analysis will work out in favor of renewables in the near term unless perhaps we continue to keep nuclear costs artificially high. In particular, I’m not sure that 100% over provisioning (absent storage) gets us anywhere near the same reliability of nuclear.

It doesn't matter how many solar panels you build, when the sun is down none of them are generating power. Likewise when the wind stops, it doesn't matter how many turbines are in the windfarm. Overgeneration certainly does not squash storage amounts.

It reduces but does not eliminate storage.

You’ll still have some windless nights, but enough overproduction means that you don’t have to worry about mere cloudy days or low-wind periods — One suggestion in another comment was needing weeks of storage, and overproduction can reduce that to a day of storage, perhaps even less.

The solution will inevitably be distributed overstorage rather than overgeneration.

10% of everyones car and home battery will be annexed by the grid when needed.

Combining sources from multiple _unreliable_ sources is actually fine - as long as the failure/intermittency isn't correlated. Defining "fine" as reasonably low variance - i.e. guaranteeing X amount of capacity will be available for Y% of the time.

Nuclear doesn't give you 100% for Y above - there are times when your nuclear reactor will also be unavailable - the average is about 32 days per year in the US. So nuclear has a similar problem - as has all generation types - albeit with different statistical properties.

> Combining sources from multiple _unreliable_ sources is actually fine - as long as the failure/intermittency isn't correlated.

This is true for a sufficiently large number of uncorrelated sources, but all solar plants in a region are correlated and all wind plants in a region are correlated so effectively you have only two renewable sources in a region that are decoupled, which is wholly inadequate considering the relatively low reliability of either source. Maybe over-provisioning would help, but I don't know how much you would have to over-provision to get power on a still night, and anyway the whole point is accounting for costs so if it's not storage then you have to account for the over-provisioning costs.

> Nuclear doesn't give you 100% for Y above - there are times when your nuclear reactor will also be unavailable - the average is about 32 days per year in the US. So nuclear has a similar problem - as has all generation types - albeit with different statistical properties.

These aren't similar problems. With wind and solar power, bad weather shuts down all "plants" in a region. With nuclear, you have maintenance windows on individual plants which can be coordinated with other plants in the region such that base load is maintained.

> different statistical properties

Technically true, but doing a lot of rhetorical labor.

You but your claim was that it was only possible to provide reliable supply by combining _reliable_ sources. This is not the case.

Wind and solar are largely anti-correlated in most regions. Depending on the geography, differences in wind output can vary significantly over distances in as little 100km and offshore and on-shore capacity factors also tend to show divergence. At distances above 1,500km wind is uncorrelated in Europe.

The stochastic failures of nuclear reactors are largely independent but not always - a natural catastrophe like Fukushima can take out a bunch of reactors. Or extreme weather (heat), like last summer's drought took out a bunch of reactors in a few northern European countries.

It's not rhetorical labor to point out that it's simplistic to present nuclear as completely reliable (it's not) and that unreliable sources can contribute to a reliable grid. It's pointing out that the problem is more subtle.

Unlike 10 years ago, we have now concrete examples of grids which operate with very high shares of wind and solar without any decrease in reliability and have developed practical techniques for incorporating such sources into grids.

> Solar and wind are cheapest now, by a multiple.

Absolutely! I’m only suggesting nuclear as a backup for low wind nighttime demand. It’s completely pointless to use nuclear as the main power source precisely because of the cost: https://en.wikipedia.org/wiki/Cost_of_electricity_by_source#...

But to deal with the low-wind nighttime demand, you have the option of storage (hydro and batteries aren’t the only options); you can do demand management (it helps people use less when they sleep); you can do intercontinental HVDC (in principle, but think in terms of a square metre of cross section for 100% of EU demand, so it will take a while to build); and you can do other power types — nuclear (expensive), geothermal (geographically specific in ways I don’t understand), tidal (surprisingly expensive), natural gas (sill better than coal).

I’m in favour of all of the above except natural gas. I’m expecting near 100% wind & PV supported by existing nuclear plants and existing and new storage (and a few 10s of GW intercontinental HVDC) by early 2030s… but my expectations are based on hoping grid storage battery prices get a lot cheaper with less evidence than I’d like to be comfortable, which means I want to support all the solutions rather than just my favourite.

I was not disagreeing

> While nuclear is still cheaper than the batteries which would otherwise be...

And here lies the problem, if we see everything through the lens of economics and affordability we'll never get out of it

Get out of what... economics?

And then what would be the driving force of progress which results in all that wondereful green technology?

Technology as a human concept is about solving problems, and throughout history, it's sometimes been about solving problems caused by previous technology. Unless we suddenly get struck by a movement to get back to an agrarian civilisation and stick there in perpetuity like an ultra strict Amish-like society, there's simply no way to escape the race with ourselves where we need more and more technology to fix the problems caused by the previous technologies.

Yes, eventually we'll fail. Or escape to the stars and sacrifice other planets to our selfish genes. But that's ok, eventually the universe itself will fail, you can only slow down entropy for a tiny bit.

They most likely meant economics as in the numbers game rather than in the original sense.

If there are better ways to invest effort in instead of the bottomless pit which nuclear power plants are, why bother? I’m not anti-nuclear in principle, but building new plants just doesn’t make sense. Takes too long and ties up too much resources. Money is just a proxy for those two things.

Reply on news.ycombinator.com