Interesting article. I think it is important to point out that California is one of the best possible regions in the world for solar + batteries powering the electrical grid. In other regions the capacity factor of solar power drops significantly and changes the conclusions.
Geography is the most important constraint on renewable energy, but it is typically missed in this type of analysis.
I would caution the use of Lazards LCOE data. They are pretty notorious for making optimistic assumptions for renewable energy.
I would also like to see the solar + CCGT (without batteries) as an option in the graphics. CCGT could run at nights, in the winter, and during cloudy days. My guess is that this the most cost-effective option in the American Southwest for the foreseeable future.
I would also add that charging a significantly larger number of EVs at night complicates the transition.
Indeed, you definitely have to be careful when using Lazard LCOE data, here is the note at the bottom of page 8 of their 2024 LCOE+ report. "Other factors would also have a potentially significant effect on the results contained herein, but have not been examined in the scope of this current analysis. These additional factors, among others, may include: implementation and interpretation of the full scope of the IRA; economic policy, transmission queue reform, network upgrades and other transmission matters, congestion, curtailment or other integration-related costs; permitting or other development costs, unless otherwise noted; and costs of complying with various environmental regulations (e.g., carbon emissions offsets or emissions control systems). This analysis is intended to represent a snapshot in time and utilizes a wide, but not exhaustive, sample set of Industry data. As such, we recognize and acknowledge the likelihood of results outside of our ranges. Therefore, this analysis is not a forecasting tool and should not be used as such, given the complexities of our evolving Industry, grid and resource needs. Except as illustratively sensitized herein, this analysis does not consider the intermittent nature of selected renewables energy technologies or the related grid impacts of incremental renewable energy deployment. This analysis also does not address potential social and environmental externalities, including, for example, the social costs and rate consequences for those who cannot afford distributed generation solutions, as well as the long-term residual and societal consequences of various conventional generation technologies that are difficult to measure (e.g., airborne pollutants, greenhouse gases, etc." https://www.lazard.com/media/xemfey0k/lazards-lcoeplus-june-2024-_vf.pdf
Electricity policymakers are optimistic about the flexibility of electric vehicle (EV) charging times, which could help manage grid demand more effectively. In Australia, a majority of EV charging currently occurs during the day to take advantage of lower electricity prices. However, I remain skeptical about the generalizability of this behavior.
The current cohort of EV drivers in Australia is still relatively small and may represent a self-selected sample of the population. These early adopters are likely to have flexible work arrangements and access to rooftop solar panels, which incentivizes daytime charging. As EV adoption becomes more widespread, the charging patterns may shift, potentially leading to different demand profiles on the electricity grid.
To gain a more comprehensive understanding of EV charging behavior, it would be instructive to examine electricity data from Norway, which has one of the highest EV penetration rates in the world. However, it is important to note that Norway's electricity generation is primarily powered by hydroelectricity, rather than solar or wind. This difference in the energy mix could influence charging patterns and grid management strategies.
In 2007 I ran a pilot project with solar + flow batteries that was intended as proof of concept to replace diesel gensets generating electricity in communities throughout Africa.
Two problems appeared: the flow battery technology was immature, and if solar & batteries must replace 100% of the electricity the size of solar array and batteries becomes gi-normous. So the project never progressed beyond the pilot stage.
In March of this year I met someone (Manoj Sinha at Husk Power systems) who made a business of that exact same concept, but with two notable changes:
-Lead acid batteries instead of flow batteries.
-Keep the diesel gensets, which means only 90% of the electricity is provided by the photovoltaic array.
Not to complain too much, but could the graph axes labels be adjusted? Perhaps the graph generator doesn’t allow for it, but rather than 50K to 150K on the X-Axis and 0 mwh to 800000 mwh on the secondary Y-Axis, why not 50GW to 150GW and 0 GWh to 80GWh, respectively? It was not immediately clear what meant what until I read the note above it.
(Also to be pedantic: the unit is MWh not mwh! Small m is milli, small indeed.)
There's probably a better way, but if you go to the "visualize" page in Datawrapper, then hit the annotate button, at the bottom of that page there's an "add text annotation" which you can use to add an axis label.
Hello Brian, I am a big fan of your work and particularly enjoyed your article on the potential and perils of solar power. I have long advocated for a "90 percent campaign" for renewables, and my current residence in South Australia provides a compelling case study for this discussion.
South Australia's state electricity grid boasts the highest penetration of renewables in the world, with approximately 75% of all electricity in 2024 being powered by renewable sources. This is the state where Elon Musk effectively launched the grid-scale battery industry. The state government has set an ambitious goal of achieving 100% renewables by 2030. However, given our connection to the east coast grid, we will likely achieve a "net zero" grid rather than a fully zero grid.
The Eastern states have been particularly stringent about approving new gas projects and pipelines, largely due to pressure from environmentalist groups. If these puritanical tendencies are not addressed, they risk undermining the economics of solar and wind power.
On a related note, it seems unrealistic to assume that the grid will be powered solely by solar and gas. Most grids incorporate a mix of solar and wind, as these sources are complementary. Wind power tends to peak during the evenings, nights, and winter months, providing a balance to solar power. While I believe it is uneconomical to go beyond 90% renewables, this balanced approach can help achieve a more stable and sustainable grid.
Recently, Michael Liebreich had a podcast where his guest made a similar argument, proposing the use of diesel engines (similar in size to those used in marine ships) as a better alternative to open-cycle gas turbines. What do you think of the proposal?
Everything from the current PV and battery price drops to the quantity of PV that China has installed was deemed impossible by skeptics. I’ve updated my priors to assuming that almost every analysis is flawed due to the projected prices being much higher than they will actually be in the future. Looking backwards you’d win every possible bet by making that assumption.
To the critics, adding other energy sources makes the case for solar substantially better. Uncorrelated energy sources like wind and hydro reduce the storage required. Sources like geothermal and nuclear compliment renewables further by providing both baseload power and heat.
It's abundantly clear at this point that solar can cheaply provide 40-50% of energy demand after addressing some straightforward technical challenges. Nuclear, wind, hydro, and gas can make up the slack. Eventually, the gas power can be made green with DAC or replaced with clean sources of baseload power.
Baseload sources like nuclear don't complement solar very well. The cost optimized solutions tend to be one or the other, not a combination. Both are inflexible (but in different ways).
No. In the limit of zero cost storage, the lowest LCoE source wins. And that's solar. Why charge the batteries with expensive nuclear when they could be charged with cheap solar?
I think you mean "40-50% of electricity demand" rather than "40-50% of energy demand," correct? Those are two very different things, and the first is far more realistic than the second. Right now in the US solar is only 1% of energy with wind being a little more.
Plus, there will likely be great variations based on geography. Solar radiance varies greatly by geography. 40-50% of electricity demand is feasible in some geographies, like the American Southwest, but not in much of the rest of the world.
Yes you're right, I mean electricity demand not primary energy per se.
(I'm optimistic we can find alternatives for hydrocarbons in other industries or just eat the cost of carbon capture, but that's off topic)
Different geographies certainly need a different approach to renewables. But I want to note that solar irradiance or capacity factors aren't the primary problem. Arizona and Maine only differ by ~2x in solar resources and overbuilding solar isn't going to break the bank.
The problem is being bottlenecked by low solar during the winter and needing a lot of batteries/panels to address that. Fortunately wind output is higher in the winter and generally higher overall in cold regions of the US.
So yes I expect cold places to lean on more wind/nuclear/gas in general, especially in the winter.
A pretty rosy picture, but entirely ignores the system dynamics to run a power system. There is no mention of system inertia, one of the reason for the Iberian Peninsula collapse. A power system needs ancillary services like reactive support, reserves, dispatchable capacity responsive to AGC, system inertia, the list goes on. NERC has declared an emergency because of the poor ride through performance of Inverter Based Resources and their habit of shutting down when they are most needed. If you read the list of Disturbance Reports included in the just released NERC Level # NERC Alert, the vast majority have occured in connection to the CAISO system. Not ready for Prime Time i believe is the correct description. See; https://substack.com/@kilovar1959/note/c-118954262
I didn't reference it because the author was specifically talking about CAISO, but fair point. I have a post dropping at 5am tomorrow on my Substack about the Iberian Blackout if you want to check it out.
Every single one????? Even if that were true, how many batteries and at what cost? Right now there isn't enough manufacturing capacity to even keep up with the replacement cycle of that many batteries.
The largest number I could find in the article up top is 800000 MWh (0.8 TWh). Just did a quick calculation on ChatGPT, and replacing every gas peaker plant in the US (142GW) with an equivalent 4-hour storage battery would require about 0.6 TWh.
These are short time horizons. I don't thing battery manufacturing is going to end up being the bottleneck you think it is, at least not globally.
Well capacity and production are two very different things. Just because you have the assembly line capacity does not mean you have the required raw material or the staff to meet that production. If there were excess capacity as you claim, then the present long lead times being experienced would not occur.
Now I am fully aware that battery storage will be an important part of the energy mix going forward. Key words are part and mix. It's has become very clear that an interconnection made entirely of Inverter Based Resources is too unstable. Those that want to push that envelope do so at the peril of customers they serve.
I don't really know what to tell you. These consensus estimates take raw materials and labor availability into account, and so far nobody seems to believe that battery capacity is going to be substantially held back due to them, at least in the 2030 timeframe.
There has been some detailed analysis of the supply chain for raw materials, which shows that amazing reductions in requirements can be done once battery recycling is brought online. Similarly, automation of battery manufacturing is already huge and just taking off, so there's an enormous amount of juice to squeeze out of it.
Overall I'm not quite sure what you're arguing here. You seem to have some general distrust of battery technology, and you're leveraging that into non-specific objections that aren't really consistent with current trends or expert industry analysis.
In the real world, massive deployment of storage batteries is coming. It's coming much faster that you think it will. It will change the way grids work. Some people will deploy it more thoughtfully than others, and will get better results immediately. Some people will be unhappy about it or try to stop it, but they'll probably be ineffective in the same way that people were generally unsuccessful at preventing digital technology from becoming embedded in every field of industry.
There are a number of places that you and I disagree;
Your original piece was published after the Iberian Blackout, yet completely ignores the issues it brought to light. Either you timing was poor, ot you believe you know more than the rest of us.
I commented with a link to the NERC Level 3 Alert for IBRs, which you completely ignored. You act as if NERC doesn’t know what they are talking about.
You commented that batteries could fix all the problems, with zero information how, what technologies would be needed, or what policies would need to change. If you are going to make bold claims they need to be more than one line.
It is well reported that there is a long lead time on batteries, yet you claim there is plenty of batteries to be had. It can’t be both, and you offer only your opinion with referenced facts.
California has exceptional solar potential and little winter heat load.
This whole schema breaks down in the Midwest and Northeast with lousy solar capacity and greater winter heating.
The imbalance between seasons requires more batteries and more gas.
As you admit, your calculations don’t account for all the system costs that are imposed by intermittency. The LCOE cost calculations have been increased almost yearly as we learn more about the cost and performance of grids with higher renewable participation.
The 2024 Lazard costs for California Solar with only four hours of battery backup was similar to the disastrous Votgle nuclear plant.
On a national or global scale, wind and solar can be a major contributor, but it’s wishful thinking to promote very large penetrations of wind and solar power as THE solution to climate change.
Eh. California has higher-than-average solar capacity factors, but not that much higher. California is at 27.7% capacity factor, but the US as a whole is at ~23%.
I agree that large-scale electric heating in cold climates would make these economics worse (though I'm not sure by how much).
The average capacity factor is not particularly relevant, its the worst day / week/ month capacity factor that drives how effective PV or PV + storage can be. California has a high floor.
Indeed, Europe is a notoriously bad place for solar energy. Much of it lies above 50°N. And the floor is almost exactly zero — if you have a blocking high in the wrong place, you can have near-zero output on both wind and solar for weeks.
The key thing here is how this 15% is generated over time. The problem is the output going to zero for possibly an entire week (although a few days is more common). Germany has had a lot of wind power for a long time, that is why even in English, the term ‘Dunkelflaute’ is sometimes used to describe those calm cloudy periods.
And electricity consumption in Germany is down since 2010 which isn't a good sign.
Do you think heat batteries can help manage winter peaks? Also doesn't the Midwest have a lot of wind potential which tends to peak during the winter seasons?
Wind power I'm sure works as a substitute for solar. That's what is used in northern Europe for that reason, but even solar is now efficient enough to be of use.
Have you considered grid stability? Does the recent blackout in Spain, which appears possibly to have been caused by frequency instability due to solar's inability to quickly stabilize frequency variations, mean that further infrastructure would be required to maintain grid stability?
I know South Australia has achieved very high levels of renewables coverage, but it is a small state next to two much larger states (Victoria and New South Wales) that get most of their power from conventional sources, so I expect that SA "borrows" its stability from its neighbors.
Even if you don't trust grid-forming inverters, it doesn't mean you need spinning rotors powered by fossil fuels.
Synchronous condensers (essentially a real-world version of the old perpetual motion joke, a generator hooked up to an electric motor) can provide the necessary inertia.
A CSIRO study in 2022 suggested the costs of adding synchronous condensers to provide inertia to a high-renewables grid would be in the order pf 1-2AUD (so maybe 1 USD) per megawatt/hour of wholesale electricity.
Brian, I'm curious if you've evaluated rooftop (home level) vs large scale PV. Rooftop is more expensive, but it has the benefit of being resilient. The transmission infrastructure is fragile, so co-locating generation and consumption (at the 70-80% level) is safer. But it depends on how much more expensive it actually is.
This is an interesting thought experiment but to exclude wind power from the calculation is overly pessimistic.
That said, the January problem is real and is even worse in colder parts of the world. As well as gas (ultimately hydrogen/biofuel/gas + carbon capture) backup, one thing that has been underexplored is demand that you can turn off during the depths of winter. Shayle Kann jokes about a bitcoin mining ship that follows cheap solar power around the world (urgh) but things like hydrogen electrolysers and desalination come to mind.
Regarding using grid batteries, I have pitched this to our utility several times. It has even better return than you have modeled. Because you are effectively running the grid like a hybrid car. Peaking plants are notoriously inefficient, as is spinning reserve. Grid connected batteries at substations make sense for multiple reasons: dispatchable power for demand spikes, local redundant power during outages, efficient generation load management, load leveling for brownouts. The charging doesn’t really care where the power is coming from. The BMS can set weighting from various sources of power and direct usage accordingly (my daughter worked on a scheme for this in Ontario six or seven years ago).
Solar is great in sunny places for sure. At least the helioscope modeling is mature these days and you can reliably predict outputs. But it won’t ever be the full replacement they claim.
I would think that the amount of batteries needed is not independent of actual grid size (but moving electricity these distances requires actually costing the extra transmission). This is because at the extrema, the limiting factor is not the variance of a solar farm but the covariance of all solar farms. But solar farms in Mountain time see the sun 1 hour before the solar farms in Pacific time (on average), so if you have unified grid over two time zones, the night gap is 1 hour shorter (and your peak is lower). But that means that there is a time where you will be generating all of the power in the east for the west, and visa versa.
Similarly other renewables are more valuable the lower their correlation with the sum of the solar sources. If we could guarantee that wind was not very correlated with solar, then even at worse LCOE it might beat out some gas, especially because again, we are already buying the batteries.
If you could find enough uncorrelated wind basins, then wind supplementation would be very valuable.
Also note that this analysis does not include analysis of demand curtailment. If there are new uses for energy given lower prices (and there will be), those might be viable even if you can only get that sometimes. We can actually accept a system that 10% of the time does not do the 10% least important things we do with power, for a 1% total shortfall.
This also does not interface with the interaction of battery costs and round trip efficiency.
The times when neither show up for long enough are rare enough that a bit of hydrogen (or other non-fossil fuel) burned in turbines can cover those times.
Very low capacity factor backup power is an interesting design niche. I've thought that something based on rocket engine technology could be interesting here: store a fuel, liquid oxygen, and water, then burn them together in what amounts to a thrust chamber to produce high pressure steam, which would be expanded through a turbine exhausting to the atmosphere. The capex/W could be lower and discharge efficiency higher than ordinary combustion turbines.
Or skip the water, just run that rocket exhaust straight through a multi-stage https://en.wikipedia.org/wiki/Tesla_turbine for terrific efficiency without needing a lot of fiddly precision machining of fan-blades. If you've got a great big insulated tank for LOX anyway (condensing it with surplus power at peak-supply times), and need more inert working fluid to trade temperature for pressure so as to prevent some of the cheap metal parts from melting, use liquid nitrogen.
That really depends on your climate. In Europe you will have that one or two weeks without any solar and wind basically every winter. In internet lore, Seattle weather is stuff of the legend.
Right, and that's what the hydrogen is for, covering Dunkelflauten.
A combined cycle power plant is maybe $1.3/W; a simple cycle turbine plant maybe half that. These are a factor of 10 or 20 cheaper than a new nuclear plant. So adding a bunch of these to cover these rare outages is cheaper than supplying the grid with nuclear. The very low capacity factor means the cost of the hydrogen itself would be small.
Europe has widespread geology suitable for cheap hydrogen storage, up to millions of gigawatt-hours worth, far more than would be needed.
In many cities, the peak electric load is on sunny days, due to air conditioning. Solar fits well to handle those peaks. 25% of the load can be AC in Phoenix AZ
Two things get neglected when people just look at how PV or PV + batteries will affect the grid:
a) more transmission can help deal with demand peaks as well as "dunkelflaute" events
b) control over load can both flatten peaks and reduce curtailment
This is understandable.
On the first point, it can be hard to allocate the costs of transmission because its effects are diffuse and, potentially, somewhat staggered where it is intended to anticipate future generation or load patterns.
The second point is more complicated. To better control load on the grid, and, frankly, to unlock the full potential of energy storage, it will require establishing communications with millions of devices, where, previously, grid operators could manage the grid just talking to thousands. We have the technology to do this, but have not figured out how to manage the social coordination that would be necessary to make it work, both from regulatory and commercial perspectives.
Yes, a truly distributed grid (with EV responsive charging, using excess power for immersion heaters, etc) can dramatically improve the efficiency here.
I know this problem is a "many-grad-student" one, but I am curious, as mentioned earlier about the distinction between current electricity demand and total energy demand.
A shift to heat pumps will further electricity demand in winter, and electric cars also boost expected power use over time.
Are these results proportionally correct for higher energy draws, or do the conclusions change?
Interesting article. I think it is important to point out that California is one of the best possible regions in the world for solar + batteries powering the electrical grid. In other regions the capacity factor of solar power drops significantly and changes the conclusions.
Geography is the most important constraint on renewable energy, but it is typically missed in this type of analysis.
I would caution the use of Lazards LCOE data. They are pretty notorious for making optimistic assumptions for renewable energy.
I would also like to see the solar + CCGT (without batteries) as an option in the graphics. CCGT could run at nights, in the winter, and during cloudy days. My guess is that this the most cost-effective option in the American Southwest for the foreseeable future.
I would also add that charging a significantly larger number of EVs at night complicates the transition.
I only used Lazard cost data for gas turbine and nuclear prices (and for gas turbines I also checked it against quite a few other sources).
Indeed, you definitely have to be careful when using Lazard LCOE data, here is the note at the bottom of page 8 of their 2024 LCOE+ report. "Other factors would also have a potentially significant effect on the results contained herein, but have not been examined in the scope of this current analysis. These additional factors, among others, may include: implementation and interpretation of the full scope of the IRA; economic policy, transmission queue reform, network upgrades and other transmission matters, congestion, curtailment or other integration-related costs; permitting or other development costs, unless otherwise noted; and costs of complying with various environmental regulations (e.g., carbon emissions offsets or emissions control systems). This analysis is intended to represent a snapshot in time and utilizes a wide, but not exhaustive, sample set of Industry data. As such, we recognize and acknowledge the likelihood of results outside of our ranges. Therefore, this analysis is not a forecasting tool and should not be used as such, given the complexities of our evolving Industry, grid and resource needs. Except as illustratively sensitized herein, this analysis does not consider the intermittent nature of selected renewables energy technologies or the related grid impacts of incremental renewable energy deployment. This analysis also does not address potential social and environmental externalities, including, for example, the social costs and rate consequences for those who cannot afford distributed generation solutions, as well as the long-term residual and societal consequences of various conventional generation technologies that are difficult to measure (e.g., airborne pollutants, greenhouse gases, etc." https://www.lazard.com/media/xemfey0k/lazards-lcoeplus-june-2024-_vf.pdf
Electricity policymakers are optimistic about the flexibility of electric vehicle (EV) charging times, which could help manage grid demand more effectively. In Australia, a majority of EV charging currently occurs during the day to take advantage of lower electricity prices. However, I remain skeptical about the generalizability of this behavior.
The current cohort of EV drivers in Australia is still relatively small and may represent a self-selected sample of the population. These early adopters are likely to have flexible work arrangements and access to rooftop solar panels, which incentivizes daytime charging. As EV adoption becomes more widespread, the charging patterns may shift, potentially leading to different demand profiles on the electricity grid.
To gain a more comprehensive understanding of EV charging behavior, it would be instructive to examine electricity data from Norway, which has one of the highest EV penetration rates in the world. However, it is important to note that Norway's electricity generation is primarily powered by hydroelectricity, rather than solar or wind. This difference in the energy mix could influence charging patterns and grid management strategies.
In 2007 I ran a pilot project with solar + flow batteries that was intended as proof of concept to replace diesel gensets generating electricity in communities throughout Africa.
Two problems appeared: the flow battery technology was immature, and if solar & batteries must replace 100% of the electricity the size of solar array and batteries becomes gi-normous. So the project never progressed beyond the pilot stage.
In March of this year I met someone (Manoj Sinha at Husk Power systems) who made a business of that exact same concept, but with two notable changes:
-Lead acid batteries instead of flow batteries.
-Keep the diesel gensets, which means only 90% of the electricity is provided by the photovoltaic array.
Not to complain too much, but could the graph axes labels be adjusted? Perhaps the graph generator doesn’t allow for it, but rather than 50K to 150K on the X-Axis and 0 mwh to 800000 mwh on the secondary Y-Axis, why not 50GW to 150GW and 0 GWh to 80GWh, respectively? It was not immediately clear what meant what until I read the note above it.
(Also to be pedantic: the unit is MWh not mwh! Small m is milli, small indeed.)
Ok, I figured out a way to do this, will get these updated (and make sure to do it for future graphs).
Unfortunately datawrapper really makes it hard to add labels to axes on line graphs, I'll see if I can find a way to do this though.
There's probably a better way, but if you go to the "visualize" page in Datawrapper, then hit the annotate button, at the bottom of that page there's an "add text annotation" which you can use to add an axis label.
Yes, that does make the graphics very hard to read. I got lost on a few of them.
Hello Brian, I am a big fan of your work and particularly enjoyed your article on the potential and perils of solar power. I have long advocated for a "90 percent campaign" for renewables, and my current residence in South Australia provides a compelling case study for this discussion.
South Australia's state electricity grid boasts the highest penetration of renewables in the world, with approximately 75% of all electricity in 2024 being powered by renewable sources. This is the state where Elon Musk effectively launched the grid-scale battery industry. The state government has set an ambitious goal of achieving 100% renewables by 2030. However, given our connection to the east coast grid, we will likely achieve a "net zero" grid rather than a fully zero grid.
The Eastern states have been particularly stringent about approving new gas projects and pipelines, largely due to pressure from environmentalist groups. If these puritanical tendencies are not addressed, they risk undermining the economics of solar and wind power.
On a related note, it seems unrealistic to assume that the grid will be powered solely by solar and gas. Most grids incorporate a mix of solar and wind, as these sources are complementary. Wind power tends to peak during the evenings, nights, and winter months, providing a balance to solar power. While I believe it is uneconomical to go beyond 90% renewables, this balanced approach can help achieve a more stable and sustainable grid.
Recently, Michael Liebreich had a podcast where his guest made a similar argument, proposing the use of diesel engines (similar in size to those used in marine ships) as a better alternative to open-cycle gas turbines. What do you think of the proposal?
I'm not an expert, but I'm not buying this.
Why would any serious person care about your unsupported opinion?
Did I offer an actual opinion, you jackass? I merely expressed well warranted skepticism.
Everything from the current PV and battery price drops to the quantity of PV that China has installed was deemed impossible by skeptics. I’ve updated my priors to assuming that almost every analysis is flawed due to the projected prices being much higher than they will actually be in the future. Looking backwards you’d win every possible bet by making that assumption.
Wonderful post.
To the critics, adding other energy sources makes the case for solar substantially better. Uncorrelated energy sources like wind and hydro reduce the storage required. Sources like geothermal and nuclear compliment renewables further by providing both baseload power and heat.
It's abundantly clear at this point that solar can cheaply provide 40-50% of energy demand after addressing some straightforward technical challenges. Nuclear, wind, hydro, and gas can make up the slack. Eventually, the gas power can be made green with DAC or replaced with clean sources of baseload power.
Baseload sources like nuclear don't complement solar very well. The cost optimized solutions tend to be one or the other, not a combination. Both are inflexible (but in different ways).
Don't batteries change this? Inflexible baseload nuclear can become more flexible with batteries right?
No. In the limit of zero cost storage, the lowest LCoE source wins. And that's solar. Why charge the batteries with expensive nuclear when they could be charged with cheap solar?
Some designs for nuclear enable load-balancing, for example in France.
I think you mean "40-50% of electricity demand" rather than "40-50% of energy demand," correct? Those are two very different things, and the first is far more realistic than the second. Right now in the US solar is only 1% of energy with wind being a little more.
Plus, there will likely be great variations based on geography. Solar radiance varies greatly by geography. 40-50% of electricity demand is feasible in some geographies, like the American Southwest, but not in much of the rest of the world.
Yes you're right, I mean electricity demand not primary energy per se.
(I'm optimistic we can find alternatives for hydrocarbons in other industries or just eat the cost of carbon capture, but that's off topic)
Different geographies certainly need a different approach to renewables. But I want to note that solar irradiance or capacity factors aren't the primary problem. Arizona and Maine only differ by ~2x in solar resources and overbuilding solar isn't going to break the bank.
The problem is being bottlenecked by low solar during the winter and needing a lot of batteries/panels to address that. Fortunately wind output is higher in the winter and generally higher overall in cold regions of the US.
So yes I expect cold places to lean on more wind/nuclear/gas in general, especially in the winter.
A pretty rosy picture, but entirely ignores the system dynamics to run a power system. There is no mention of system inertia, one of the reason for the Iberian Peninsula collapse. A power system needs ancillary services like reactive support, reserves, dispatchable capacity responsive to AGC, system inertia, the list goes on. NERC has declared an emergency because of the poor ride through performance of Inverter Based Resources and their habit of shutting down when they are most needed. If you read the list of Disturbance Reports included in the just released NERC Level # NERC Alert, the vast majority have occured in connection to the CAISO system. Not ready for Prime Time i believe is the correct description. See; https://substack.com/@kilovar1959/note/c-118954262
What about the Iberian Peninsula collapse?
You mean like how the cause is yet to be determined?
I didn't reference it because the author was specifically talking about CAISO, but fair point. I have a post dropping at 5am tomorrow on my Substack about the Iberian Blackout if you want to check it out.
You mean like how the cause is yet to be determined?
But just about every single one of these problems can be solved once you have lots of battery storage on the grid.
Every single one????? Even if that were true, how many batteries and at what cost? Right now there isn't enough manufacturing capacity to even keep up with the replacement cycle of that many batteries.
Global battery manufacturing capacity (yearly)
2022: 1.57 TWh
2023: 2.2 TWh
2024: 3 TWh
Estimate for 2030: 7.3 TWh
The largest number I could find in the article up top is 800000 MWh (0.8 TWh). Just did a quick calculation on ChatGPT, and replacing every gas peaker plant in the US (142GW) with an equivalent 4-hour storage battery would require about 0.6 TWh.
These are short time horizons. I don't thing battery manufacturing is going to end up being the bottleneck you think it is, at least not globally.
Well capacity and production are two very different things. Just because you have the assembly line capacity does not mean you have the required raw material or the staff to meet that production. If there were excess capacity as you claim, then the present long lead times being experienced would not occur.
Now I am fully aware that battery storage will be an important part of the energy mix going forward. Key words are part and mix. It's has become very clear that an interconnection made entirely of Inverter Based Resources is too unstable. Those that want to push that envelope do so at the peril of customers they serve.
I don't really know what to tell you. These consensus estimates take raw materials and labor availability into account, and so far nobody seems to believe that battery capacity is going to be substantially held back due to them, at least in the 2030 timeframe.
There has been some detailed analysis of the supply chain for raw materials, which shows that amazing reductions in requirements can be done once battery recycling is brought online. Similarly, automation of battery manufacturing is already huge and just taking off, so there's an enormous amount of juice to squeeze out of it.
Overall I'm not quite sure what you're arguing here. You seem to have some general distrust of battery technology, and you're leveraging that into non-specific objections that aren't really consistent with current trends or expert industry analysis.
In the real world, massive deployment of storage batteries is coming. It's coming much faster that you think it will. It will change the way grids work. Some people will deploy it more thoughtfully than others, and will get better results immediately. Some people will be unhappy about it or try to stop it, but they'll probably be ineffective in the same way that people were generally unsuccessful at preventing digital technology from becoming embedded in every field of industry.
There are a number of places that you and I disagree;
Your original piece was published after the Iberian Blackout, yet completely ignores the issues it brought to light. Either you timing was poor, ot you believe you know more than the rest of us.
I commented with a link to the NERC Level 3 Alert for IBRs, which you completely ignored. You act as if NERC doesn’t know what they are talking about.
You commented that batteries could fix all the problems, with zero information how, what technologies would be needed, or what policies would need to change. If you are going to make bold claims they need to be more than one line.
It is well reported that there is a long lead time on batteries, yet you claim there is plenty of batteries to be had. It can’t be both, and you offer only your opinion with referenced facts.
You make bold claims about how inexpensive battery storage is, again without referenced facts, when industry reports place it among the most expensive assets on the power system. https://open.substack.com/pub/energybadboys/p/battery-storage-is-141-times-more?r=23kggy&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false
California has exceptional solar potential and little winter heat load.
This whole schema breaks down in the Midwest and Northeast with lousy solar capacity and greater winter heating.
The imbalance between seasons requires more batteries and more gas.
As you admit, your calculations don’t account for all the system costs that are imposed by intermittency. The LCOE cost calculations have been increased almost yearly as we learn more about the cost and performance of grids with higher renewable participation.
The 2024 Lazard costs for California Solar with only four hours of battery backup was similar to the disastrous Votgle nuclear plant.
On a national or global scale, wind and solar can be a major contributor, but it’s wishful thinking to promote very large penetrations of wind and solar power as THE solution to climate change.
You might curb your utopian idealism.
Eh. California has higher-than-average solar capacity factors, but not that much higher. California is at 27.7% capacity factor, but the US as a whole is at ~23%.
I agree that large-scale electric heating in cold climates would make these economics worse (though I'm not sure by how much).
The average capacity factor is not particularly relevant, its the worst day / week/ month capacity factor that drives how effective PV or PV + storage can be. California has a high floor.
Try with like New York or Germany...
This is a good point.
I would absolutely love to see similar numbers for Germany
Germany is at about 80% renewable generation right now. And remember that Germany has a latitude range that is almost entirely in Canada.
Germany is basically de-industrializing right now due to electricity becoming scarce.
Indeed, Europe is a notoriously bad place for solar energy. Much of it lies above 50°N. And the floor is almost exactly zero — if you have a blocking high in the wrong place, you can have near-zero output on both wind and solar for weeks.
It's so bad that Germany got 15% of its grid electricity from it in 2024, to say nothing of behind-the-meter. Please try to inform yourself better.
The key thing here is how this 15% is generated over time. The problem is the output going to zero for possibly an entire week (although a few days is more common). Germany has had a lot of wind power for a long time, that is why even in English, the term ‘Dunkelflaute’ is sometimes used to describe those calm cloudy periods.
And electricity consumption in Germany is down since 2010 which isn't a good sign.
Do you think heat batteries can help manage winter peaks? Also doesn't the Midwest have a lot of wind potential which tends to peak during the winter seasons?
Heat batteries can help, but storage viability comes down to how often its cycled vs cost. They will end out undersized vs severe events.
On wind, yes it helps but it still leaves gaps. Those gaps still need to be covered.
I.e.
https://bsky.app/profile/ember42.bsky.social/post/3kiwtsndgsn2y
Wind power I'm sure works as a substitute for solar. That's what is used in northern Europe for that reason, but even solar is now efficient enough to be of use.
Not a substitute. It's just an additional source. Wind, hydro, geothermal are all additional renewable sources of electricity.
Thanks -very thoughtful and insightful.
Have you considered grid stability? Does the recent blackout in Spain, which appears possibly to have been caused by frequency instability due to solar's inability to quickly stabilize frequency variations, mean that further infrastructure would be required to maintain grid stability?
I know South Australia has achieved very high levels of renewables coverage, but it is a small state next to two much larger states (Victoria and New South Wales) that get most of their power from conventional sources, so I expect that SA "borrows" its stability from its neighbors.
Even if you don't trust grid-forming inverters, it doesn't mean you need spinning rotors powered by fossil fuels.
Synchronous condensers (essentially a real-world version of the old perpetual motion joke, a generator hooked up to an electric motor) can provide the necessary inertia.
Thanks. I'm sure there are other solutions. My question is - how much does it add to the cost?
Not a lot.
A CSIRO study in 2022 suggested the costs of adding synchronous condensers to provide inertia to a high-renewables grid would be in the order pf 1-2AUD (so maybe 1 USD) per megawatt/hour of wholesale electricity.
https://www.energycouncil.com.au/analysis/csiro-does-the-maths-re-integration/
Brian, I'm curious if you've evaluated rooftop (home level) vs large scale PV. Rooftop is more expensive, but it has the benefit of being resilient. The transmission infrastructure is fragile, so co-locating generation and consumption (at the 70-80% level) is safer. But it depends on how much more expensive it actually is.
This is an interesting thought experiment but to exclude wind power from the calculation is overly pessimistic.
That said, the January problem is real and is even worse in colder parts of the world. As well as gas (ultimately hydrogen/biofuel/gas + carbon capture) backup, one thing that has been underexplored is demand that you can turn off during the depths of winter. Shayle Kann jokes about a bitcoin mining ship that follows cheap solar power around the world (urgh) but things like hydrogen electrolysers and desalination come to mind.
That's what https://www.terraformindustries.com/ is aiming for.
Great stuff as usual.
Regarding using grid batteries, I have pitched this to our utility several times. It has even better return than you have modeled. Because you are effectively running the grid like a hybrid car. Peaking plants are notoriously inefficient, as is spinning reserve. Grid connected batteries at substations make sense for multiple reasons: dispatchable power for demand spikes, local redundant power during outages, efficient generation load management, load leveling for brownouts. The charging doesn’t really care where the power is coming from. The BMS can set weighting from various sources of power and direct usage accordingly (my daughter worked on a scheme for this in Ontario six or seven years ago).
Solar is great in sunny places for sure. At least the helioscope modeling is mature these days and you can reliably predict outputs. But it won’t ever be the full replacement they claim.
Interesting.
The company which is now Fluence first installed batteries as spinning reserve for a power plant in Chile back in 2007.
Next was frequency regulation, PJM was receptive to the pilot project which worked better than they imagined possible.
Etc etc.
Odd, that 😉
Yes it makes perfect sense but many locales aren’t willing to do it
I would think that the amount of batteries needed is not independent of actual grid size (but moving electricity these distances requires actually costing the extra transmission). This is because at the extrema, the limiting factor is not the variance of a solar farm but the covariance of all solar farms. But solar farms in Mountain time see the sun 1 hour before the solar farms in Pacific time (on average), so if you have unified grid over two time zones, the night gap is 1 hour shorter (and your peak is lower). But that means that there is a time where you will be generating all of the power in the east for the west, and visa versa.
Similarly other renewables are more valuable the lower their correlation with the sum of the solar sources. If we could guarantee that wind was not very correlated with solar, then even at worse LCOE it might beat out some gas, especially because again, we are already buying the batteries.
If you could find enough uncorrelated wind basins, then wind supplementation would be very valuable.
Also note that this analysis does not include analysis of demand curtailment. If there are new uses for energy given lower prices (and there will be), those might be viable even if you can only get that sometimes. We can actually accept a system that 10% of the time does not do the 10% least important things we do with power, for a 1% total shortfall.
This also does not interface with the interaction of battery costs and round trip efficiency.
You need a huge amount of transmission to time shift by a time zone though.
And its not good enough for wind to be weakly uncorrelated with solar, as that still leaves periods where neither show up.
The times when neither show up for long enough are rare enough that a bit of hydrogen (or other non-fossil fuel) burned in turbines can cover those times.
Very low capacity factor backup power is an interesting design niche. I've thought that something based on rocket engine technology could be interesting here: store a fuel, liquid oxygen, and water, then burn them together in what amounts to a thrust chamber to produce high pressure steam, which would be expanded through a turbine exhausting to the atmosphere. The capex/W could be lower and discharge efficiency higher than ordinary combustion turbines.
Or skip the water, just run that rocket exhaust straight through a multi-stage https://en.wikipedia.org/wiki/Tesla_turbine for terrific efficiency without needing a lot of fiddly precision machining of fan-blades. If you've got a great big insulated tank for LOX anyway (condensing it with surplus power at peak-supply times), and need more inert working fluid to trade temperature for pressure so as to prevent some of the cheap metal parts from melting, use liquid nitrogen.
That really depends on your climate. In Europe you will have that one or two weeks without any solar and wind basically every winter. In internet lore, Seattle weather is stuff of the legend.
Right, and that's what the hydrogen is for, covering Dunkelflauten.
A combined cycle power plant is maybe $1.3/W; a simple cycle turbine plant maybe half that. These are a factor of 10 or 20 cheaper than a new nuclear plant. So adding a bunch of these to cover these rare outages is cheaper than supplying the grid with nuclear. The very low capacity factor means the cost of the hydrogen itself would be small.
Europe has widespread geology suitable for cheap hydrogen storage, up to millions of gigawatt-hours worth, far more than would be needed.
In many cities, the peak electric load is on sunny days, due to air conditioning. Solar fits well to handle those peaks. 25% of the load can be AC in Phoenix AZ
https://www.eia.gov/consumption/residential/reports/2009/state_briefs/pdf/az.pdf
Two things get neglected when people just look at how PV or PV + batteries will affect the grid:
a) more transmission can help deal with demand peaks as well as "dunkelflaute" events
b) control over load can both flatten peaks and reduce curtailment
This is understandable.
On the first point, it can be hard to allocate the costs of transmission because its effects are diffuse and, potentially, somewhat staggered where it is intended to anticipate future generation or load patterns.
The second point is more complicated. To better control load on the grid, and, frankly, to unlock the full potential of energy storage, it will require establishing communications with millions of devices, where, previously, grid operators could manage the grid just talking to thousands. We have the technology to do this, but have not figured out how to manage the social coordination that would be necessary to make it work, both from regulatory and commercial perspectives.
Yes, a truly distributed grid (with EV responsive charging, using excess power for immersion heaters, etc) can dramatically improve the efficiency here.
I know this problem is a "many-grad-student" one, but I am curious, as mentioned earlier about the distinction between current electricity demand and total energy demand.
A shift to heat pumps will further electricity demand in winter, and electric cars also boost expected power use over time.
Are these results proportionally correct for higher energy draws, or do the conclusions change?