Hacker News Reader

> "By comparison, the US uses 12,500 TWh of electricity daily"

I was suspicious of your numbers so I did a bunch of math and then realized you're using the European comma rather than a decimal. So 12,500 means 12.5, not 12500.

(EIA.gov says the U.S. used about 3.8 trillion kilowatt hours in 2020 [1]. 3.8 trillion kilowatt hours equals 3.8 billion megawatt hours, equals 3.8 million gigawatt hours, equals 3.8 thousand terawatt hours per year or 10.4 twh per day. If we figure the U.S. has about a hundred million households and divide 3,800 twh by that, we get 38 megawatt hours per household per year. This is a very rough estimate, as it doesn't include industrial/commercial/government users. If we divide by 365*24 to cancel out the time units, we get an average consumption of .004338 MW or 4.338 kilowatts per household. That sounds about right.)

Current battery production is only just barely getting started. China dominates production of LFP cells (which are ideal for grid storage) because of patents which are expiring, so hopefully we'll see more production outside of china in the near future.

LFP cells aren't bottlenecked by nickel or cobalt, and so the main resource constraints I believe are lithium, aluminum, and copper which are all quite a bit cheaper and available in bigger quantities. I think prices are expected to eventually settle somewhere around $80 per kwh of capacity for the cells, and I don't think we're that far off that now. (LFP may eventually be displaced by something else, like lithium sulfur or solid state batteries or something, but I think LFP is probably good enough.)

Maybe lithium or copper will become bottlenecks and prices will rise. Let's say prices do hold at about $80 per kwh. Maybe we'll round up to $120 per kwh to account for pack construction, a building to store the batteries, inverters, chargers, and so on. If the average U.S. adult-aged person uses about 2kw on average, then they need 48kwh of storage for 24 hours. That would be about $5760. If we amortize that over ten years, it's about $48 a month. That's kind of expensive, but it's within the realm of what can be done without assuming any major technological breakthroughs. We probably don't need 24 hours of storage, though, if we have enough renewable energy over-production and backup fossil fuel plants to use in extreme situations.

[1] https://www.eia.gov/energyexplained/electricity/use-of-elect...

LFP is also actually the perfect battery for urban transportation in tropical countries. Their low cost and economics would ensure a big transformation in coming years because few people understand the tremendous impact they have on lifecycle cost (due to high no of cycles).

I personally feel that in 10 years in India it will be cheaper to rent an electric car running on LFP for trips < 300 km along with a driver (cheaper here) than buy your own car.

Just because CATL can manufacture LFP cells at $80/kWh doesn't mean the end-to-end system is available at that price. Home battery storage is nowhere near $80 or even $100 per kWh. It's at $500-$1000/kWh. Tesla Powerwall 2 has the best price per kWh available on the market, at $560/kWh. Inverters able to handle the currents necessary for turning on an electric oven are in the thousands. Installation costs are in the thousands. And then, these batteries do not last forever - they're rated at 3000-5000 cycles

If we imagine a near-future scenario - a battery powered house heated with heat pumps, driving an electric car, during central European foggy winter, which tends to last for about a week at a time. Let's say they use 60kWh/day. We're looking at $100k just for the batteries and installation, even assuming electricity comes free. In ~15-20 years, the batteries reach their recommended cycle life, requiring a choice of another $100k or accepting that what used to be a 3-day storage becomes 2-day

This is not some "extreme situation". It's a completely predictable scenario that recurs with near 100% probability every single year. Then there are places like Ganges river, Bangladesh or Indonesia, home to hundreds of millions of people, where neither solar, nor wind is viable (and land is scarce)

Those prices for home storage units are the price we pay for not having significant domestic LFP production. Maybe making a cheap battery unit is harder that I think but to me it sure looks like there isn't a lot of serious competition in that space, otherwise the units would correspond better to battery costs. (And why would you try to build a company around a product that some Chinese company can make the product at a cheaper cost because they have access to the necessary patents on more favorable terms? Maybe everyone knows that and is staying away, at least until LFP cells become a thing anyone in the world can buy at roughly-equal prices.)

I think it's useful to think in terms of cars. I'm actually doing an EV conversion right now, and I have a motor controller that puts out about 100 kilowatts of 3-phase AC. It came with the motor so I don't have an exact price, but fair market value is probably around $1500. (That's retail cost in quantities of one. Wholesale cost is presumably somewhat less.) It weighs maybe ten pounds or so. That's adequate to power probably about a dozen houses. I don't know if it puts out a sine wave or a square wave. Even if it's the latter, you could imagine a hundred of these things putting out square waves with, say, randomly-varying pulse width and having it average out to a sine wave. There's probably better ways to do that, but anyways the point is that we do have the capacity to switch enormous amounts of power in a very small package for pretty low cost. It's actually kind of amazing.

Maybe if the batteries feed into an HVDC line or they're co-located with a solar plant, then you don't even need to deploy more inverters.

A battery management system for my conversion is a little over a thousand dollars for a setup with about 48 cells. In a big installation you could amortize BMS costs by using bigger cells, or placing them in parallel groups -- the downside being that it might take longer to notice if a cell is going bad. My BMS is made by a company that caters to EV conversions; they're doing low-volume sales. A major utility ordering the equivalent for hundreds of thousands of cells probably can get a nice volume discount.

As far as foggy winters go: renewable energy needs to be traded over a wide geographical area. Purely local generation doesn't really make sense, unless you have some useful purpose for unpredictable amounts of surplus energy. Wind power would need to be a part of it. Even fossil fuels are a reasonable option as long as they're not used very often. Running natural gas plants for a week or two in the dead of winter or in case of grid disruptions seems like a reasonable use of fossil fuels. Indonesia has a lot of land outside of Java. Even Java is pretty sparsely populated outside of cities. Most of it is jungle which we'd like them to keep and farms, but solar doesn't need to take up a huge percentage of available land. Even solar panels on roofs can go a long way.

> Lithium ion battery storage is not cheaper than nuclear, and is not produced at sufficient scale to be viable for grid storage.

It may not be so now, but in 7 years, or in other words, by the time a nuclear power plant commissioned today starts producing power, it definitely will.

Global li-ion manufacturing capacity is poised to triple by 2024:

https://www.luxresearchinc.com/blog/li-ion-manufacturing-in-...

Without a sudden, disruptive change in the cost and rate of deployment of nuclear there doesn't exist a path for it become a significantly larger part of the energy mix.

> Global li-ion manufacturing capacity is poised to triple by 2024

But we need all of that to electrify transportation, so where do you get the ones for the power grid?

Also, what happens to the price of batteries if you get rid of baseload and cause demand for batteries to spike much higher than even the increased amount of battery production capacity?

So 3 years from now, global lithium ion battery production will add up to ~3 hours of USA's electricity consumption? If we stop building EVs, electronics, etc. and dedicate all lithium ion battery production worldwide to grid storage in the US for 8 years we'll have enough for 1 day of electricity storage. Just for the USA. And how much storage we actually need? For 0% fossil fuel usage, some estimates place storage demands as high as 3 weeks [1]. This is still far from viable, even if the predictions hold true.

It's also ignoring that electricity usage is predicted to increase substantially worldwide as countries develop, and that these predictions of lithium ion battery production might not pan out. It'd also severely delay adoption of electric vehicles, as battery capacity is being diverted to grid storage away from EVs.

1. https://pv-magazine-usa.com/2018/03/01/12-hours-energy-stora...

You usually oversize your production in renewable and get extra energy in your peak production. Here in Quebec we have a new tarification and smart water heater, house heating that do peak shaving. Some industries will have access to almost free energy in peak production period and will have incentives to do something valuable with it Charging of EV will certainly play a role too.

Still nuclear can be useful if some country are able to build them fast enough and cheap enough.

The world could redirect a lot of resources into building battery infrastructure. Enough for EVs and Storage.

Your Storage does only need to be for 1-2 days (or more precisely until your renewables come online again to recharge them).

I don't see a massive shift to battery manufacturing any more crazy than building a crazy amount of nuclear power plants.

3 hours of storage is massive, and enable super high penetration of renewables, above 80% of electricity generation.

Even the much vaunted France doesn't have 80% nuclear on their grid.

Plus, energy storage production capacity is growing at an exponential rate. It's doubtful that we could grow our nuclear construction crews at that pace.

It's going to be very hard to nuclear to scale as fast as renewables and storage are, if there's even an economic case to be made for nuclear construction, and somebody finally solves the logistic problems of large construction projects in the modern Western world.

We know we can build storage and renewables, but we don't know how to build nuclear anymore, and none, absolutely none, of the nuclear proponents have any proposals to fix it. The best is an entirely new type of small reactor that has been rejects in the past because of its high per unit cost. Perhaps it will work, but who knows? It's a big risk, whereas storage and renewables are a sure bet.

> 3 hours of storage is massive

It really isn't. For reference, across Europe we're currently having a meteorological phenomenon with bad weather, clouds, little wind, and low temperatures. It has been ongoing since ~october and is projected to continue into the winter. When Texas was hit with terrible weather last year, it lasted a few days. Multi-day storms, which take out solar and wind, aren't unheard of. 3 hours of storage is OK to even out things, but isn't nowhere near close enough to guarantee reliable electricity.

The numbers I've seen say you need ~12 hours of storage to support 80% solar/wind generation composition and 2 weeks to hit 100%. And we're just talking about the US right now. What about the rest of the world?

Personally I think small mass manufacturable fission reactors are the best bet but why are we hung up on either/or? We should aggressively pursue both nuclear and solar/wind generation.

Millions of lives are at stake.

> why are we hung up on either/or?

I agree that we should invest in both. But it is not as easy as waving a magic wand. Energy project development is a complicated dance between local power market regulators, local and federal incentives, private investors, and shovel ready/cost-effective technology. And in the rare occasions when congress opens up the wallet, it is a small pie that everyone is fighting over, so the lobbyists start slinging arrows.

The fact is, nuclear can and will be a really great addition to the energy mix. But there are models out there that show that we can build a carbon free grid without it. The big question mark is on the future cost of storage. The consensus is that those costs are going to come down considerably. If so, then going all in on renewables likely gets us to a carbon free grid faster. But this is also a very US centric take. For example China likely has much easier road to rolling out nuclear (less regulatory hurdles, less local interference/input). Where as in the US, I am sure we can build a big national pro-nuclear movement, but the second someone proposes a real life project, the NIMBY's will come out of the woodwork.

> (less regulatory hurdles, less local interference/input)

This is not the current hurdle when it comes to more nuclear in the US or Europe. Nuclear is very welcome where it is currently under construction, and there's no regulatory reason that these sites have all become construction disasters, it's just bad execution of the reactors.

Even China's attempts to build the French EPR design took twice as long to build as they had estimated initially, and we don't really know how much more in cost.

Nuclear's path to new reactors begins with being able to build on a reliable schedule without exorbitant cost. There are many sites that would welcome more nuclear that would not have NIMBY problems, there's just no one willing to bear all the risk of finding that unicorn contractor that can actually build.

This same problem was also evident during the late 1970s and early 1980s nuclear projects. There was NIMBYism back then affecting projects, but there was also lots of construction malpractice that resulted in big cost overruns. US utilities can not bear the construction risk of a $10B project. Few entities can.

>It may not be so now, but in 7 years ... it definitely will.

Maybe, maybe not. We know nuclear works so why not also make nuclear in case something happens and we can't build enough storage.

Because we simply are unable to do so fast enough. China recently started a massive, largest in the world project to add 150 plants over the next 15 years. That's 10 reactors annually, or an estimated 10GW using the current 49.6GW from 50 units they have already.

Adjusted for capacity factor this is maybe going to keep up with solar deployment, but definitely not with wind capacity growth in the same country. And China has been the world leader in nuclear deployment for a while now.

Renewables win on a "worse is better" basis - yes, they're intermittent, but they're cheap and deploy in a matter of months without too many specialists involved.

You didn't explain why we shouldn't do both. Just because it takes a long time doesn't matter. If we want to ensure we have consistent energy production then nuclear should be built. Even if we can build a massive amount of storage it may not be enough. Global warming is supposed to cause extreme weather. Why not create nuclear which can work regardless of weather?

> 12,500 TWh of electricity daily

You're simplifying out that electric consumption varies during the day.

Excellent point. Peak electricity consumption occurs during the evenings when solar is producing little, if any, electricity. Daily fluctuations of electricity usage put an even greater demand on electricity storage.

That's not correct, peak demand is usually during mid-day, at least in the US:

https://www.eia.gov/todayinenergy/detail.php?id=27212

I'm seeing it peak around the evening in all time zones. This is the link to the actual energy grid visualization tool: https://www.eia.gov/electricity/gridmonitor/expanded-view/el...

Browsing the last day's energy usage yields peaks at:

Eastern: 8pm

Central: 7pm

Mountain 6pm

Pacific: 5pm

Due to AC usage, electricity demand tends to be much higher in the summers than in the winters. And the demand profile in the summers peaks in mid-late afternoon. i.e You are not looking at the highest overall peak of the year, just the hourly peaks from the last few days.