Dec 10, 2023·edited Dec 10, 2023Liked by Hügo Krüger
I clicked on this article because I saw the eye-popping pull quote about the cost of battery storage, but the math doesn’t check out.
The first big problem is that it’s wrong to look at replacing the total installed renewable capacity with battery storage. Power producers know renewables are intermittent and never produce at 100%, so they overbuild renewable capacity. During a Dunkelflaute, you only have to replace the actual deficit, and not the extra overbuilt margin. The graph included in the article conveniently shows the worst case deficit is closer to 60 GW.
The second problem is your figure for how much battery storage costs appear to be way off reality. My best guess is that the report gives a projection of the upfront cost of building battery storage in the future, and not the cost per incident of using it, however I have to admit I am very confused by the cited report. Nonetheless, we can sanity check this by looking at the cost to build battery storage plants. A random one I picked off the top of my head that was built recently is the Australian Hornsdale Power Reserve plant, which Wikipedia says cost $120 million USD for 200 MWh of installed capacity. So, to scale up a plant that can handle a 600 GWh 10 hour Dunkelflaute would cost an upfront cost of $360 billion if built today. The ongoing operational costs are very small compared to this, so I’ll ignore them.
I can’t find for sure how many uses the Australian plant is designed for, but Wikipedia says the battery technology it’s based on, the Powerwall, can deliver 5,000 cycles, so we’ll use that as an estimate. This would bring the per Dunkelflaute, per German cost down to $0.87, rather than the $3,557.69 which caught my attention.
Dec 10, 2023·edited Dec 10, 2023Liked by Hügo Krüger
The distinction is about upfront “cost-to-build” vs the amortized “cost-per-use”. My back of the napkin figure had Hornsdale’s “cost-to-build” at $600/kWh, and it looks like the source[1] you’re pulling from also has it at roughly the same at $527. It’s clear that such a huge plant would cost an enormous sum to build.
My main disagreement is that once the storage plant is built, it’s good for 5,000 uses, over some twenty years, at least based on what we know about similar storage grade battery models.
This is what makes the scheme at all reasonable to consider. The huge plant will only cost $0.87 per German each time it is needed if it is cycle limited, or $18.13 per month if it is limited by lifespan. This is weighed against an income you put at $5,364 per month. Either way, this is not ruinously expensive.
We must compare apples to apples: monthly costs to monthly incomes. Furthermore, $18/month is much more feasible to think about offsetting through arbitrage or lowering through improved efficiency/technology.
This part of the article left me with the impression that battery technology is so ruinously unaffordable that to implement it Germans would have to sacrifice nearly a month’s income (I.e. a Christmas bonus) to afford it. I don’t think that’s supported by the math. I would imagine that we might agree that rolling out this technology will require a large upfront investment to duplicate capabilities offered by already existing, already built coal and gas (… and especially nuclear) plants. It’s debatable whether that is the best course of action. (To be completely transparent, I think it will *eventually* be the best course of action as battery storage prices continue to fall, although I understand why others disagree.) However, it needs to be clear that the financial cost of that investment, even with today’s technology will not break the average German households’ budget. I can understand why some people might say it’s worth paying the $18/month to get this done today.
the point I make above these calculations though is that I would want for voters to feel the direct consequence of these policies, so we can have a more rational response!
LCOE of pumped storage for what it is worth is 20% of the LCOE of batteries (but it takes 10 years to build)
Thank you! I actually almost made an identical mistake once in a school project report, so that’s why I spotted it.
I don’t think we know with extremely high certainty exactly how long the batteries used for grid scale storage will last, but the manufacturer claimed 5,000 cycles or twenty years isn’t a bad guess. The batteries used in phones and laptops last for roughly a thousand cycles. The key things that cause those batteries to degrade are high temperatures and using much more of their capacity per charge cycle (“deep cycling”). Grid scale batteries can greatly mitigate this by having extra weight for overhead capacity and active water cooling. Different chemistries can also help here, but then you stray further from the economies of scale in the rest of the battery supply chain. Since their claim passes a sniff test and given how much the engineers who wrote the specifications have riding on this, I’d be willing to bet that these batteries last for their rated lifetimes.
What I was told when I was actively researching these issues is that the main limitation to pumped hydro build up in Europe and America is that we’re basically using all the easy natural sites for hydroelectric power already, and that expanding beyond that with new dams on the available more marginal sites would also pose a huge political/regulatory hassle. Basically, it’s too much squeeze for not enough juice.
You also have to consider that battery costs have fallen something like 80% in the last ten years. I’m not sure they *actually* will duplicate that performance in the next ten years because progress has been slower lately, but if they do, by the time any pumped hydro is built battery could be cheaper than pumped hydro storage.
I’m not actually sure that consumer electricity prices will rise enough from environmental policy changes to make people notice and demand change. Technical progress on solar and batteries has gotten far enough that even badly designed inefficient environmental policy is lost in the noise of all the other inefficient badly designed policy. My personal power prices are already MUCH higher than they ought to be because of permit denial preventing my electric utility from importing power from a neighbor with excess hydroelectric capacity. Germany could have still used its nuclear plants instead of its coal plants. You get the idea.
This gets me to the real trade off I see in electrical decarbonization: is it worth making a huge investment building this technology now, or should we wait for prices to continue to fall? What will really be the cost of living with and cleaning up the pollution compared to cost of accelerating the retirement of working infrastructure to prevent the pollution? At what point are there cheaper ways to achieve more worthy environmental goals? How do we balance competing goals, like pollution reduction with e.g. environmental degradation due to building more pumped storage? I’ll note that these trade-offs are not made by consumers, but instead made in closed meeting rooms with few or no engineers and often include little consideration of hard numbers. Alas, “Urgently decarbonize as soon as fiscally optimal” is not a catchy campaign slogan.
I agree, and in the context of South AFrica, in the developing world, I am a heretic for saying let's just go for the cheapest option, which at the moment, in our case is to fix the coal fleet.
We are 1% of the world's C02 and if Germany is going to struggle to sell these policies to their own population then you can imagine the upheaval that it will cau for us. Which is why I argue that every country should go at their own pace.
What do you make of my proposal to price electricity as a service? so customers can at least feel directly what these policies cost. I was given the idea by David Friedman, Milton Friedman's son.
I clicked on this article because I saw the eye-popping pull quote about the cost of battery storage, but the math doesn’t check out.
The first big problem is that it’s wrong to look at replacing the total installed renewable capacity with battery storage. Power producers know renewables are intermittent and never produce at 100%, so they overbuild renewable capacity. During a Dunkelflaute, you only have to replace the actual deficit, and not the extra overbuilt margin. The graph included in the article conveniently shows the worst case deficit is closer to 60 GW.
The second problem is your figure for how much battery storage costs appear to be way off reality. My best guess is that the report gives a projection of the upfront cost of building battery storage in the future, and not the cost per incident of using it, however I have to admit I am very confused by the cited report. Nonetheless, we can sanity check this by looking at the cost to build battery storage plants. A random one I picked off the top of my head that was built recently is the Australian Hornsdale Power Reserve plant, which Wikipedia says cost $120 million USD for 200 MWh of installed capacity. So, to scale up a plant that can handle a 600 GWh 10 hour Dunkelflaute would cost an upfront cost of $360 billion if built today. The ongoing operational costs are very small compared to this, so I’ll ignore them.
I can’t find for sure how many uses the Australian plant is designed for, but Wikipedia says the battery technology it’s based on, the Powerwall, can deliver 5,000 cycles, so we’ll use that as an estimate. This would bring the per Dunkelflaute, per German cost down to $0.87, rather than the $3,557.69 which caught my attention.
The distinction is about upfront “cost-to-build” vs the amortized “cost-per-use”. My back of the napkin figure had Hornsdale’s “cost-to-build” at $600/kWh, and it looks like the source[1] you’re pulling from also has it at roughly the same at $527. It’s clear that such a huge plant would cost an enormous sum to build.
My main disagreement is that once the storage plant is built, it’s good for 5,000 uses, over some twenty years, at least based on what we know about similar storage grade battery models.
This is what makes the scheme at all reasonable to consider. The huge plant will only cost $0.87 per German each time it is needed if it is cycle limited, or $18.13 per month if it is limited by lifespan. This is weighed against an income you put at $5,364 per month. Either way, this is not ruinously expensive.
We must compare apples to apples: monthly costs to monthly incomes. Furthermore, $18/month is much more feasible to think about offsetting through arbitrage or lowering through improved efficiency/technology.
This part of the article left me with the impression that battery technology is so ruinously unaffordable that to implement it Germans would have to sacrifice nearly a month’s income (I.e. a Christmas bonus) to afford it. I don’t think that’s supported by the math. I would imagine that we might agree that rolling out this technology will require a large upfront investment to duplicate capabilities offered by already existing, already built coal and gas (… and especially nuclear) plants. It’s debatable whether that is the best course of action. (To be completely transparent, I think it will *eventually* be the best course of action as battery storage prices continue to fall, although I understand why others disagree.) However, it needs to be clear that the financial cost of that investment, even with today’s technology will not break the average German households’ budget. I can understand why some people might say it’s worth paying the $18/month to get this done today.
[1] https://www.nsenergybusiness.com/features/energy-storage-analysing-feasibility-grid-scale-options/
that fair enough, so I had to divide the above numbers by 5000 (the cycle)
thanks for the correction. I will take out that part!
Question though, how many full cycles can a battery really take?
the point I make above these calculations though is that I would want for voters to feel the direct consequence of these policies, so we can have a more rational response!
LCOE of pumped storage for what it is worth is 20% of the LCOE of batteries (but it takes 10 years to build)
Thank you! I actually almost made an identical mistake once in a school project report, so that’s why I spotted it.
I don’t think we know with extremely high certainty exactly how long the batteries used for grid scale storage will last, but the manufacturer claimed 5,000 cycles or twenty years isn’t a bad guess. The batteries used in phones and laptops last for roughly a thousand cycles. The key things that cause those batteries to degrade are high temperatures and using much more of their capacity per charge cycle (“deep cycling”). Grid scale batteries can greatly mitigate this by having extra weight for overhead capacity and active water cooling. Different chemistries can also help here, but then you stray further from the economies of scale in the rest of the battery supply chain. Since their claim passes a sniff test and given how much the engineers who wrote the specifications have riding on this, I’d be willing to bet that these batteries last for their rated lifetimes.
What I was told when I was actively researching these issues is that the main limitation to pumped hydro build up in Europe and America is that we’re basically using all the easy natural sites for hydroelectric power already, and that expanding beyond that with new dams on the available more marginal sites would also pose a huge political/regulatory hassle. Basically, it’s too much squeeze for not enough juice.
You also have to consider that battery costs have fallen something like 80% in the last ten years. I’m not sure they *actually* will duplicate that performance in the next ten years because progress has been slower lately, but if they do, by the time any pumped hydro is built battery could be cheaper than pumped hydro storage.
I’m not actually sure that consumer electricity prices will rise enough from environmental policy changes to make people notice and demand change. Technical progress on solar and batteries has gotten far enough that even badly designed inefficient environmental policy is lost in the noise of all the other inefficient badly designed policy. My personal power prices are already MUCH higher than they ought to be because of permit denial preventing my electric utility from importing power from a neighbor with excess hydroelectric capacity. Germany could have still used its nuclear plants instead of its coal plants. You get the idea.
This gets me to the real trade off I see in electrical decarbonization: is it worth making a huge investment building this technology now, or should we wait for prices to continue to fall? What will really be the cost of living with and cleaning up the pollution compared to cost of accelerating the retirement of working infrastructure to prevent the pollution? At what point are there cheaper ways to achieve more worthy environmental goals? How do we balance competing goals, like pollution reduction with e.g. environmental degradation due to building more pumped storage? I’ll note that these trade-offs are not made by consumers, but instead made in closed meeting rooms with few or no engineers and often include little consideration of hard numbers. Alas, “Urgently decarbonize as soon as fiscally optimal” is not a catchy campaign slogan.
I agree, and in the context of South AFrica, in the developing world, I am a heretic for saying let's just go for the cheapest option, which at the moment, in our case is to fix the coal fleet.
We are 1% of the world's C02 and if Germany is going to struggle to sell these policies to their own population then you can imagine the upheaval that it will cau for us. Which is why I argue that every country should go at their own pace.
What do you make of my proposal to price electricity as a service? so customers can at least feel directly what these policies cost. I was given the idea by David Friedman, Milton Friedman's son.
that article is interesting, thanks for sharing, I added it as a note in the bottom of the post.