It would be good to clarify that when you say solar provided 4% of US electricity generation in 2023, that was utility-scale only. When we add in rooftop it comes to 5.6%. In 2024 it was 6.9%, including rooftop.
Quantifying solar as a percentage of total US *energy* (not electricity) production is trickier. The number you've quoted (slightly less than 1%) is correct (for 2023) only in a technical sense that I think is misleading to anyone but experts. Our World in Data calls this method of energy accounting "primary energy by the direct method". It treats solar (and wind, hydro, and geothermal) rather unfairly because they're credited only for the electricity they produce, whereas fossil and nuclear energy are credited for their heat production, before the big thermal losses in engines and turbines. If we instead look at "primary energy by the substitution method", where we credit solar as the amount of fossil fuel that would be needed to produce the same electricity, then in 2023 it was more than 2% of the total (https://www.eia.gov/totalenergy/data/monthly/pdf/sec12_32.pdf).
One more minor clarification: The capacity factor numbers quoted here are AC capacity factors, which put the AC capacity in the denominator. But as Jenny Chase keeps pointing out, AC capacity is just the size of the wire that connects the system to the grid. If we put the actual (DC) capacity of the *panels* in the denominator, then PV capacity factors are significantly lower, probably averaging 17 or 18 percent for utility-scale facilities in the US.
There's a less glib explanation on page 14 of her book, Solar Power Finance Without the Jargon (2nd edition). On page 73 she says "AC capacity is normally 10–50% less than DC capacity". Pages 193-194 explain one reason for using smaller inverters/connections, although I think another reason is just limited grid capacity. On-site batteries, of course, add a further complication.
Well, in a world of increasing Electric Vehicle penetration, you have huge amounts of "synthetic storage" if cars feed their batteries "intelligently" (for California, that is at noon, for nuclear France, that is at nigth).
Yeah, it seems inevitable that we will eventually have a system where the owner of an EV can program in the amount of their battery that they're willing to sell to the grid (or equivalently, how high a reserve level they want to keep for when they need to drive), and then have some kind of algorithm that decides whether to charge or discharge at different times of day, based on local Time Of Use pricing, participation in a VPP scheme, or whatever else. I already have the batteries on my wall enrolled in a VPP, and there's no reason the battery in my driveway, which is more than twice as large, shouldn't be contributing as well.
Well, I am skeptical of returning electricity back to the grid. Simply the intelligent choice of charging time reduces fossil fuel use by a large margin.
Typically what you want with batteries in a neighborhood is for them to push power out into the _local distribution grid_ in a manner that reduces the need for the long-distance transmission system to deliver energy to the neighborhood substation where it's stepped down from MV or HV to local LV.
You could do this with "a Megapack (or two, or three) at every substation", if you could figure out the right economics for it. Basically make the grid modular, with a contactor at the substation that actually lets the neighborhood disconnect from the transmission system for a while if there's a problem. A few _big_ batteries are typically much less expensive than lots of smaller batteries scattered around the neighborhood. This lets you get the peak-shaving and time-shifting value from all the PV spread out across the neighborhood (which unlike the batteries, _can't_ be stacked up in one place).
But this still leaves you with the fact that a lot of people will have medium-size batteries (bigger than what we typically install for a home PV system, but smaller than a Megapack) parked in their driveways, and it's silly to leave that capacity totally idle a large part of the time. It still is more efficient to have a smaller neighborhood substation battery, and recruit V2G power to supplement it. Again, it's just a matter of figuring out the payment and control mechanisms so everyone wins.
By managing the demand at the distribution grid level, you drastically reduce the need to upgrade the transmission grid. In the absence of doing this, the cost of deploying upgraded transmission lines will be a substantial part of the cost of the clean energy transition.
Projections suggest that in the absence of the ability to locally peak-shave, about 30% of all transmission capacity would be used on peak-usage days that cover only about 3% of the days in the year. Managing distributed resources to cut our total transmission capacity needs by a third is a good investment. (And that's the total transmission. As a fraction of the additional transmission we need to build, it's _way higher_.)
n.b.: I work for Tesla Energy, on supporting grid-scale storage systems. Before I joined the Industrial Storage team I worked on Residential Storage; I was involved in developing the commissioning protocols for PowerWall 2.
Perhaps, but I would say that the most relevant problem in a renewable heavy grid is the daily gap between non manageable supply (nuclear, hydro, wind and solar) and demand. Simply by recharging in peak sun hours (in solar heavy systems) or in low demand hours, you avoid use of the fossil back up (the marginal cost of electricity)
These seem like two sides of the same coin. You avoid putting power onto the transmission grid, by recharging local-distribution level storage, from excess production (with PV being the canonical example, but other variable producers like wind might actually provide excess at night, and distributed wind is starting to be a thing as well). Then you supply loads during lower-production periods, from that same storage.
To some degree you also can fill the storage by pulling excess production at centralized facilities, across the transmission system, during periods of low loads (as aggregated at the transmission level).
In any case, I see distributed storage as helping resolve this problem. It gives you more flexibility and resilience -- like I said, what we want is a _modular_ grid, that can react gracefully if a given transmission corridor is congested or down.
There's a tendency to think of solar power and other forms of power like fossil fuels or nuclear as substitutes, but it's better to think of them as complements. They aren't competitors; they're all on the same team.
The optimal, lowest cost, lowest carbon energy mix for a country should have a mix of different forms of power so that they can cover each others' weaknesses. Fossil fuels are not verboten even in a zero-carbon future, because the excess energy from overbuilt solar capacity can power carbon capture to more than offset any fossil fuel emissions. The best way to provide power when the sun isn't up may actually be gas, rather than batteries, especially when you consider that battery manufacture has a large carbon footprint.
The goal should be to have the most abundant, reliable and cheap energy portfolio possible, rather than deciding which form of energy should "win" and kill off the rest. We have an essentially unlimited demand for energy, not just to improve human wellbeing, but also to gradually capture all the carbon we emitted since the Industrial Revolution and roll back all the damage. So we should be looking for the best team, not the best player.
Not sure if it's outside his bailiwick, but I would love to see Brian (or someone) do an analysis that would try to calculate what the ideal mix would be.
You are understanding solar from an outsider perspective researching the pro solar propaganda. I worked in the solar industry for 6-8 years... and I still do some power systems designs for folks.
1.) No solar panel factory is powered by solar panels. (when it is... call me, I'll invest.)
2.) Your graphs are misleading they show nameplate capacity and don't speak to actual energy generation, a 100W solar panel only produces 100W at laboratory test conditions. In field it may produce a 100W or 0W, depending not only on sunlight but age of equipment (yes solar panels degrade over time), cell damage (hard to detect), and even dirt. Also the power conversion equipment (inverters) fail pretty frequently also means your solar array has a power rating of 0 Watts when this happens.
3.) Solar is NOT cheap you only show $/W module cost, what about inverters, copper wire, permits, labor, etc. I used to charge about $1000 per panel full installation. Now that has gone up to at least $1500, these prices are for medium size systems 10 to 40 panels. Each panel makes about $50 of electric per year. Inverters fail in about 6 to 12 years so factor a few kilo bucks in to replace those. I'm not even going to go into battery storage system headaches... Sorry bud, the tech ain't where it needs to be, wish this wasn't the case.
Re: 3.) Did you see the pie charts roughly 2/3 through the article, which show the $/kw figure incorporates all of the items that you claim were missed?
Yeah, I guess my point was that doesn't look right.... it's $1000 per panel not per kW, each module is about 250W to 350W so it's closer to $3000 per kW. Again, keep in mind you are buying an equipment rating of 1kW, you got to let the system run for 20 years to see what your cost per kWH is... and sure the Mfg says the product is warrantied for 25-30 years. Maybe the module might last that long, but solar panel companies don't last 30 years, good luck claiming a warranty. All the PV manufactures I bought from in the 2010's are gone now...
Are your numbers for residential installations or utility scale? The numbers in the post are for utility scale, and Brian does note that residential installations are significantly more expensive - so it seems like your numbers and his roughly line up?
I did work for big utilities and residential. Obviously you get a discount buying the equipment in bulk if your are a utility, let's say you get that initial cost down to $1000/kW. But then your O&M is much higher.... you got to install fencing, grounds keeping, security, legal, engineering, travel, accounting, and did I mention legal? But a big utility isn't just doing solar projects, so they have plenty of various overhead and can move solar O&M numbers around to make solar look better if they want. As years go by and more data gets collected on O&M, I think people are going to be surprised how costly it is. All things considered I'd say $3k per kW is good estimate of cost for both utility and residential, so I'd say cost estimates in this post are off by about 3x.
My experience is that solar never paid back—and that’s in Vegas. On my project it might have paid for itself after 20 years, but with assuming perfect efficiency on products that historically had a massive drop off after 15.
The advantage of solar, cost, is not as relevant as one might think given the other attributes of solar.
When investors look at a proposed project, the question they want answered is "when do I get my money back?". The size of the investment is only relevant if it is infeasibly large or small.
So cost does not matter, but revenue does. When the sun shines it shines for everybody, so every solar generator is in competition with every other one, as the duck curve demonstrates. So the market price of generated electricity crashes. When the sun does not shine, solar generates zero revenue.
Since solar only generates income when the price is low, it is unprofitable - the answer to the investors' question is "not for a very long time".
To date that uncomfortable fact has been circumvented with "tricks" like power purchase agreements by large consumers who can afford to spend on burnishing their green credentials, and/or taxpayer subsidies. Those subsidies cannot scale to high market penetrations of solar.
I strongly recommend Brett Christophers' book "The Price is Wrong", which explains all of this better than I can, and also explains several other aspects of renewable energy generation which are likely to limit its penetration.
Important points, and this is someplace where on-site batteries, virtual power plants, and demand-based pricing have a lot to offer. Also not discussed is grid congestion. On the flip side, I suggest that it's reasonable to consider power purchase agreements and REC requirements as government support for less harmful generation. The negative externalities and government subsidies for fossil fuel generation are well-documented, and reversing these requires active effort that sometimes involves supporting competitors to establish mature markets.
1 - you don't consider degradation of the panels: the solar industry uses an annual degradation factor of 1%, but those of us who model and track these projects know that the actual degradation factor is more realistically 2% per year; your analysis needs to take degradation into account
2 - the land area required for these projects is immense: in your single family situation, we need an area of 250 square meters (0.06 acres), which works out to over 10,000 acres required to power the households in metro Atlanta - assuming zero packing distance of panels (an unrealistic assumption). 10,000 acres of rooftops with unobstructed sky view is an immense amount of land to clear and devote to power production (and perhaps warehouse use), just to support the Atlanta metro....
From my work, I realize that the only reason solar is currently at all viable is thanks to government incentives - take away the incentives and solar collapses.
I did run an LCOE analysis that includes panel degradation (using 0.5% NREL estimate), it had a really tiny effect on LCOE, changed values by less than 5% (and in most cases closer to 1%). Reran it just now with a 2% degradation, and it had less impact than you might expect. ~20% LCOE bump for solar-only options but once you start adding storage and serving higher capacities it goes down, around a 5% increase.
This doesn't totally surprise me because for the higher generation fractions, panel capacity is significantly over what's demanded, and is very often wasted because storage is full, so falling panel capacity doesn't end up reducing provided energy all that much.
I haven't thoroughly verified this (and the script was vibe-coded with Claude), but I did look at it.
re: land area, Atlanta is one of the biggest metros in the country, a few square miles of solar panels to serve it (which there is plenty of space for, Georgia gets pretty rural not too far outside of Atlanta) seems fine to me.
I notice now that you are using a 7% discount rate - this is vastly lower than financial modeling of these sorts of projects use. You can argue about capital markets and greed and capitalism, but financial reality is discount rates in the realm of 20-25+%, and that's for established entities building additional projects: rightly or wrongly, the industry still views these projects with substantial long term uncertainty. (Projecting out 20 years for new industries is inherently uncertain!)
Relative to land area, the entire Atlanta metro is 136 square miles (Wiki), so we need a solar farm ~15% the size of the city just to supply the energy needed for household use. And as a US average, household energy use is only ~20% of the total energy budget (my recollection), so we are looking at a solar farm almost as large as the city itself to power the city. With that sort of scale, infrastructure and distribution costs (capital projects & transmission losses) become very substantial - and the loss of that land area likely dwarfs all else. (Land cannot be acquired for free and just imagine the eyesore of 100+ square miles of flat roof tops on big box buildings, all the same height - the Atlanta power park would make the DFW airport zone look positively beautiful by comparison!)
And Atlanta is not even in the top quartile of US cities by number of households!
Recently, in the past few years, hundreds of acres of farmland in Arkansas, Tennessee, and other locales has been converted to solar farms - and the impact of this is becoming significant: beware the unforeseen (but not unforeseeable) consequences of tax incentives!
However, I would expect 7% to be somewhat too low in 2025 simply due to the current level of the risk free rate. Seems totally plausible this was an accurate number in 2020 when that IEA report was written though (as interest rates were lower then). As a more recent data point, Lazard's 2024 report (https://www.lazard.com/research-insights/levelized-cost-of-energyplus/) uses a 8% cost of debt and 12% cost of equity at a 60/40 split which averages to a 9.6% discount rate.
Turning several square miles of forest or farmland to solar farms is going to get a LOT of local resistance. And those are the most common locations for solar farms in the east.
250 sq m is also probably pretty close to the average usable area of a new home in the Atlanta area, and I would expect a lot of the homes do this without stairs. For those houses, their own roof area would suffice!
Interesting. Why, though? It wouldn't work for most current industries, sure.
But I can imagine a number of new ones for which this might work very well indeed - from desalinating large amounts of water over long periods to processing commodities at large scale or indeed produce hydrogen.
All of those activities need to be set up once and require huge amounts of energy but it doesn't matter all too much if you overbuilt the capacity (and doesn't cost that much more either) if you get the energy practically for free in return.
The CAPEX for the offtaker plus the storage cost for the product kill this for *almost* anything, espcially at the very low levels (~30%) for true excess, and the excess is very ‘spiky’ in that the amount is inconsistent as well.
CAPEX for desalination is pretty high, the price have come down yes. Seasonal utilization seems more likely though, like in western/northern europe which has a lot more seasonal variation in solar.
That would be interesting to know about why! I suppose some uses like making hydrogen or aluminum probably don’t turn on and off very well, and others that do, like data centers, may be capital intensive enough that they want to run constantly if you build them.
Still, in a world where daytime electricity is too cheap to meter, I bet someone comes up with a clever new way to use it!
To make one final point, the last issue you’re not considering is the *marginal* economics of solar. Which is really to say, the closer you get to 100% solar coverage, the worse the economics get. Because the photons from the Sun are free, producers are always incentivized to sell power at any cost above zero. In a competitive power producer market, this means that on *most days*, a utility is paying nothing for electrons in the middle of the day, and a lot when the sun goes down.
Basically this is the duck curve issue. And it’s the reason why states like California stopped paying households for the solar energy that they generate, because electrons in the middle of the day in CA aren’t worth very much.
It’s actually worse than that, because if you look at the slope of the duck curve, it gets really steep and thus it is much, much harder for a utility to manage (vs a system that’s fully dispatch-able). Because excess electrons have to be discharged somewhere; and CA actually *pays* neighboring states like AZ exorbitant rates to discharge that power for them just to keep the grid balanced.
It’s a bit hard to tell from the article though if the batteries are being used to provide power to market on a cyclical basis (as Brian is describing in his article) vs providing “ancillary” coverage for when a plant somewhere else on the grid goes down. If you know anywhere that has good data on ERCOT I’d appreciate a link, as I’d love to research this question more.
This question gets to the heart of my point though, which was more that increasing renewable penetration potentially becomes less profitable at the margin. Let me illustrate with an example.
Suppose ERCOT had enough solar/batteries to cover 70% of its demand *on average*. Practically speaking, that would probably be something like 7a to midnight per day on average.
Because the marginal cost to produce a watt from a solar panel (or a charged battery) is zero, rational producers would sell power at any price above $0. So really on an hourly basis a solar/battery operator isn’t making money until 1a to 7a (if you’re smart at least and keep your batteries from discharging when power is too cheap).
But what that does though is drive the capacity factor of the system way down, because it’s only effectively discharging when you can make money in the middle of the night.
So if you’re that same producer thinking about *ADDING* capacity, you’d be foolish to, because your new battery would probably depreciate faster than you are making money, so the IRR of that investment is negative.
All that said, I think it’s great (and probably inevitable) that a large percentage of generation in the US starts to come from solar and batteries. ERCOT is a great model and if more regions ran like that (or we had better transmission systems) I think that process would happen faster.
Zooming way out, my main point I want to make in response to Brian’s article is just that I don’t think we need to “overbuild” solar and batteries, because doing so you reach a point of diminish returns. I think the goal should not be to get 100% coverage or even close to it. Let the market do its thing, determine it’s own mix and if you want the government to subsidize/intervene in anything, it should probably be in electric vehicles because those have a much greater carbon impact anyway.
This is why battery build out has been happening at a fast pace. Once solar gets above a certain threshold, you need batteries to time-shift production or you run into duck curves. California basically doesn't have a duck curve anymore because so much battery capacity has been built.
Texas is now building equivalent amounts of battery storage and it's working well:
As ever, the cost to the environment is completely ignored. It doesn't matter how cheap solar gets, overbuilding simply increases that cost as more and more of the natural world is destroyed to build the solar, and destroyed to deploy it.
Given all we know about the impacts of technology to the environment, why do people continue to ignore those externalized costs?
Well it's not an either-or is it?! I still want to know why so many ignore the externalized costs, not matter what it is (solar, wind, bio-ethanol, etc.). I agree bio-ethanol is also a boondoggle and disaster for the natural world.
First of all, that's not how the massive industrial solar arrays all over the west are installed on public land. They first cut down all the joshua trees, yucca, pinyon, etc. then they bulldoze down the rest, to dirt, including scraping off the delicate desert soil crusts that keep all the soil from blowing away in the wind. They also crush all the endangered desert tortoise, lizards, and many other living beings out there. Then they fence it in. Then they install the panels. Then they have to use copious amounts of water to keep the panels clean because since they destroyed the soil crust and all the plants keeping the soil in place, the dust blows onto the panels and dramatically reduces their efficiency.
This also completely ignores the mining, refining, etc. to manufacture the panels (or whatever other energy technology you're analyzing).
Again, I'm NOT comparing. This isn't about "this or that". This is: Why do people who write articles like this one about "understanding solar energy" completely ignore the external costs to the natural world? It is a question. Why is that always ignored? While economists can conveniently wave these costs away in their models, that does not reflect reality. If we're interested in reality, then that should be part of the analysis and discussion. Yet when I bring it up you keep telling me to ignore because bio-ethanol. What?
Those are all very fair points Elisabeth. I agree that almost all energy projects have land use, biodiversity and pollution impacts and I agree they're almost never enumerated for either fossil fuel or renewable projects in most energy discussions.
Solar in Ireland (where I live) or England is generally on previously intensively cultivated agricultural land, so it's net positive from a biodiversity, emissions and water quality perspective.
When you account for the destruction to the land from mining the materials to build the solar panels, and the pollution from the mining, is it still a net positive? You don't see that mining because it's not in Ireland.
When you account for the coal used to refine silicon, is it still a net positive? You don't see the coal mining and the refineries for silicon in Ireland.
When you account for the huge pollution from refining the many materials to manufacture solar panels, is it still a net positive? You don't see that in Ireland either.
When you account for the pollution to build and run ships (on bunker fuel) to transport solar panel parts, probably from China, is it still a net positive? You don't see the impacts of all that bunker fuel in Ireland, or the steel used to build those ships, or the coal used to refine that steel.
When you account for the lax environmental law in the countries where solar PV is made, and in all likelihood, slave labor (or close to it) in those same countries, is it still a net positive? You don't see that in Ireland either.
When you account for what you do with all that energy from solar, i.e. running business-as-usual, the same business-as-usual that is causing the 6th mass extinction on Earth, is it still a net positive? You do see that in Ireland, one of the most environmentally degraded countries in the world.
I'm just wondering how the balance sheet is calculated on "net positive". Which is why I keep asking the question because very few, and not this article, will answer the question! I don't understand why. If we are to face the predicament we are in -- catastrophic ecological overshoot -- we must be honest about reality.
The externalized costs of coal are *staggering*, and yet suddenly people are clamoring for more nature when it's solar power. Like, where was all the concern 10 years ago when US grid was 50% coal???
Also, almost everyone knows how terrible coal is at this point. However there is a large number of people that think solar is somehow better (despite that coal is required to refine silicon... hmmm...). So yeah, I don't need to say "coal is bad". That's obvious. However, apparently I do need to say "solar is also bad" many many times over.
I don't have a proposal. My posts here simply ask why the externalized costs of solar are not included in "Understanding Solar Energy." Those externalized costs are on the environment, on the natural world. But there was no mention of these costs in the article, and I wondered why (there seldom is in articles of this type).
We've created a society utterly dependent on non-renewable materials and extremely high energy, including on fossil fuels for fertilizer (9/10 calories we eat are fossil fuels in a 8.2 billion person world). There is no proposal. We are in catastrophic ecological overshoot and it will not end well. All species in overshoot collapse; ours will be no different. This is ecology 101.
I'm simply asking why an article titled "Understanding Solar Energy" does not include discussion of the externalized costs.
I agree that externalized costs are real and should be considered. I rarely see them placed in a context of tradeoffs wrt to renewables, though. And here in the US, petro-funded anti-solar "grassroots" activists are now commonplace. Re embedded energy of solar, for reference:
"the embodied carbon of solar in 2020 was around 615 kgCO2/kWp of installed capacity. This is 76% lower than the 2,560 kgCO2/kWp that is commonly referenced. First Solar‘s Global Sustainability Director also recently reported a typical value of 500-600 kgCO2/kWp for monocrystalline silicon.
Our estimate is likely conservative as four of the world’s largest solar manufacturers (Longi, Jinko, First Solar, and Hanwha Q-Cells) have signed up to RE100, committing to 100% renewable electricity supplies for their operations. Other manufacturers are taking a similar approach. Solar powered solar."
This is one of the best pieces I have read on the topic until the end. You gave a great overview of the pros and cons - very balanced and fair. Your analysis of the costs, and 4 days of storage feel about right. (Although this is still macro, and you would really want probably 4-6 days of storage in a no generation case, then need extra arrays to refil the batteries.
The question and problem I have with this kind of analysis - which you point out, but do not go into, is that to have solar, you either need to spend 40c per kw (according to your model, or 10x solar and 50kw battery (which is probably 8-10x a typical house battery), or you need two sources of generation.
You can not say the cost is cheap for 25%, but expensive for 100%, because if you have 25% solar, you need another 100% generation, so you need the second thing - it is mandatory. You need solar and nuclear, or solar and wind and battery or whatever - but you need two.
This is the cost of solar, or the costs should be modelled on 10x typical battery and the huge solar array - which is a true replacement - hence why I think LCOE should never be used - or use your 100% case as stated above.
Great work, I am going to recommend this article to many people I know.
1 Numbers don’t support the fact that there is any sort of replacement or “transition” going on. Rather adding capacity like it has always been. Any thermal plant closures are due to either reduction in local electricity consumption (always due to deindustrialization or increased energy poverty) or over optimistic hypothesis invariably leading to grid fragility.
2 Subsidies don’t make stuff competitive, they skew the market and give the impression of competitive, like smoke and mirrors. If you factor in all economic components in the cost of electricity such as synchronicity, availability, stability, then there is no way intermittent unreliable sources will ever be competitive.
I don't support subsidies I'm just presenting the arguments.
1. Well, subsidies -> lower prices of electricity from solar -> lower margins for thermal plants. If consumption decreases -> also lower margins. And if you've made solar cheaper it will push thermal plats out, even if they might have positive effects for the grid at large.
Apologies if you believe I’m misleading you, that was not the objective.
On 1, if I understand well, by giving subsidies we could price thermal plants out of the market. What happens when all thermal plants are gone? Do we continue with the subsidies, do we let the market freely price energy, other? Apart from the economical aspects, is that approach even moral? In other areas that is called dumping and frowned upon, for good reasons. Either way, I don’t believe centrally mandated economics is the way to go or even feasible past the short term when we all run out of money.
On number 2, my point was that such policies do not lead to future price reductions, on the contrary, there will be an upward cost spiral, as we can see happening in all countries with high renewables penetration, and increased grid fragility.
Actually, points 1 and 2 converge into one as today’s subsidies destroy value and inexorably lead to increased costs and decreased services, for ever.
I recently purchased a book entitled “the price is wrong. Why capitalism won’t save the planet” by Brett Christophers. I haven’t read the entire book yet but the author’s main argument is “while prices of solar have tumbled the golden age of renewables has yet to materialize. The problem is that investment is driven by profit not price and operating solar remains a marginal business everywhere on the state’s financial support. We cannot expect markets and the private sector to solve climate crisis while the profits that are their lifeblood remain unappetizing.”
I have not read the book, but I have read a summary of its arguments and a couple of critiques that think he's confusing past and present as well as discounting the coming wave of thermal and battery storage.
The premise is a useful starting point. So let's consider the US EV tax credit. It's a government intervention that has led to 2024 production of 200GWh of battery capacity. Subsidies drive learning curves. A related and more "mature" example: wind revolution was driven largely by price, but only *after* prices dropped. For reference, in 2024, approx. 10% of power generation was from wind.
1. LCOE is a relatively flawed metric given it does not take into account intermittency (which you mention) as the costs of battery storage and the need for peaker gas plants do not get added to solar (or wind) LCOE metrics. Given solar plants do not function in isolation but within a complex system of the electricity grid.
2. Piggy backing on the below, synchronous generators provide ancillary services/value to the grid such as “grid inertia” that solar (and wind) do not provide. With the higher penetration of these resources, firms are building standalone facilities that mimic inertia, but at the end of the day do not flow into the cost of solar (again on LCOE basis) and a cost consumers will pay.
3. Given utility scale solar projects are typically further away from demand centers it requires higher transmission costs which again do not factor into solar LCOE.
4. Given how capacity markets work in grid systems (PJM) and the need to cover peak demand, solar cannot bid as well as competitive gas generators. With the continued penetration of solar and wind and the retirement of base load gas and coal, we have seen NERC reliability point to higher grid unreliability and have seen capacity auction prices ( ie PJM) sky rocket as a result. Ultimately, a cost paid by the consumer.
Given federal subsidies (mainly PTC), solar can bid negative prices in the electricity market which makes other base load generators uncompetitive in the wholesale market which is where you see early retirement of these assets and higher capacity pricing.
Very good article, just want to point out that “cost” of solar going down is true when you look at them in a vacuum, but not so much when you think of the system of electricity as a whole. There is a reason California has the highest (or one of) penetration of solar, but their electricity costs are the highest in the lower 48….
What do you think of cheap heat batteries improving the economics of solar? They can be used for seasonal energy storage (although the roundtrip efficiency of the system is low due to the heat-to-electricity conversion). Since about half of all energy demand is in the form of heat, you can directly supply the stored heat into industrial use cases.
Brian, you might like to look at New Zealand's electricity supply system.
About three-quarters of NZ's electricity comes from hydro dams, and there is about six months' storage behind the dams. It's widely seen as "not enough". especially in dry years. Two large industrial consumers recently closed their doors, in part because the government was asking them to agree to having their supply curtailed at short notice.
If hydro storage, about the cheapest form there is, is undersupplied, then the prospects for every other form look fairly bleak.
With all due respect, hydro storage is not particularly cheap. It's very reliable wherever it is available, this is why countries like it. But building and maintaining the huge dam and turbine systems is not cheap. And it's an undertaking that has far bigger financing needs and cluster risks than building loads of small-scale battery storage sites across the whole country. These are two very different things, even though both price storage of sorts.
Thanks for that article. I think you missed one important measure. If you'd were a home builder you'd want to maximize the utiltiy of your PV installation. Since Power prices during noon are dropping fast due to the solar peak, you're utility function should contain something like: price(t)*P_PV(t). Vertical South Installations produce more power in Winter when you need it for your heat pump. Vertical East/West Installation produce more power in the evening when power prices are rising again. During noon vertical installation produce a lot less energy but the power at noon will be free in a couple of years. Since PV is growing exponentially, hourly prices during noon are gonna drop even faster. Since your PV installation is planned for 15 or 30 years, you should start installing vertical PV NOW.
If energy will be a bottleneck for industry in the future your case seem to imply that europe will be at a significant disadvantage due to seasonal variation in solar.
Thank you for this thorough overview.
It would be good to clarify that when you say solar provided 4% of US electricity generation in 2023, that was utility-scale only. When we add in rooftop it comes to 5.6%. In 2024 it was 6.9%, including rooftop.
Quantifying solar as a percentage of total US *energy* (not electricity) production is trickier. The number you've quoted (slightly less than 1%) is correct (for 2023) only in a technical sense that I think is misleading to anyone but experts. Our World in Data calls this method of energy accounting "primary energy by the direct method". It treats solar (and wind, hydro, and geothermal) rather unfairly because they're credited only for the electricity they produce, whereas fossil and nuclear energy are credited for their heat production, before the big thermal losses in engines and turbines. If we instead look at "primary energy by the substitution method", where we credit solar as the amount of fossil fuel that would be needed to produce the same electricity, then in 2023 it was more than 2% of the total (https://www.eia.gov/totalenergy/data/monthly/pdf/sec12_32.pdf).
One more minor clarification: The capacity factor numbers quoted here are AC capacity factors, which put the AC capacity in the denominator. But as Jenny Chase keeps pointing out, AC capacity is just the size of the wire that connects the system to the grid. If we put the actual (DC) capacity of the *panels* in the denominator, then PV capacity factors are significantly lower, probably averaging 17 or 18 percent for utility-scale facilities in the US.
Thanks for the clarifications.
This is interesting re: capacity factors, is there somewhere I could read more about this?
Here's the short version from Chase: https://bsky.app/profile/solarchase.bsky.social/post/3l57n5d4zzp2k
There's a less glib explanation on page 14 of her book, Solar Power Finance Without the Jargon (2nd edition). On page 73 she says "AC capacity is normally 10–50% less than DC capacity". Pages 193-194 explain one reason for using smaller inverters/connections, although I think another reason is just limited grid capacity. On-site batteries, of course, add a further complication.
A convenient place to look up and compare DC and AC capacities of US solar farms is the LBL/USGS US PV database, https://energy.usgs.gov/uspvdb/ (map viewer at https://energy.usgs.gov/uspvdb/viewer/).
Well, in a world of increasing Electric Vehicle penetration, you have huge amounts of "synthetic storage" if cars feed their batteries "intelligently" (for California, that is at noon, for nuclear France, that is at nigth).
I wrote this post:
https://forum.effectivealtruism.org/posts/jJap6KhzFe3mgh32M/electric-vehicles-and-renewable-electricity
and probably this book gives the details:
https://link.springer.com/book/10.1007/978-3-031-09079-0
Yeah, it seems inevitable that we will eventually have a system where the owner of an EV can program in the amount of their battery that they're willing to sell to the grid (or equivalently, how high a reserve level they want to keep for when they need to drive), and then have some kind of algorithm that decides whether to charge or discharge at different times of day, based on local Time Of Use pricing, participation in a VPP scheme, or whatever else. I already have the batteries on my wall enrolled in a VPP, and there's no reason the battery in my driveway, which is more than twice as large, shouldn't be contributing as well.
Well, I am skeptical of returning electricity back to the grid. Simply the intelligent choice of charging time reduces fossil fuel use by a large margin.
Typically what you want with batteries in a neighborhood is for them to push power out into the _local distribution grid_ in a manner that reduces the need for the long-distance transmission system to deliver energy to the neighborhood substation where it's stepped down from MV or HV to local LV.
You could do this with "a Megapack (or two, or three) at every substation", if you could figure out the right economics for it. Basically make the grid modular, with a contactor at the substation that actually lets the neighborhood disconnect from the transmission system for a while if there's a problem. A few _big_ batteries are typically much less expensive than lots of smaller batteries scattered around the neighborhood. This lets you get the peak-shaving and time-shifting value from all the PV spread out across the neighborhood (which unlike the batteries, _can't_ be stacked up in one place).
But this still leaves you with the fact that a lot of people will have medium-size batteries (bigger than what we typically install for a home PV system, but smaller than a Megapack) parked in their driveways, and it's silly to leave that capacity totally idle a large part of the time. It still is more efficient to have a smaller neighborhood substation battery, and recruit V2G power to supplement it. Again, it's just a matter of figuring out the payment and control mechanisms so everyone wins.
By managing the demand at the distribution grid level, you drastically reduce the need to upgrade the transmission grid. In the absence of doing this, the cost of deploying upgraded transmission lines will be a substantial part of the cost of the clean energy transition.
Projections suggest that in the absence of the ability to locally peak-shave, about 30% of all transmission capacity would be used on peak-usage days that cover only about 3% of the days in the year. Managing distributed resources to cut our total transmission capacity needs by a third is a good investment. (And that's the total transmission. As a fraction of the additional transmission we need to build, it's _way higher_.)
https://www.volts.wtf/p/rooftop-solar-and-home-batteries
n.b.: I work for Tesla Energy, on supporting grid-scale storage systems. Before I joined the Industrial Storage team I worked on Residential Storage; I was involved in developing the commissioning protocols for PowerWall 2.
Perhaps, but I would say that the most relevant problem in a renewable heavy grid is the daily gap between non manageable supply (nuclear, hydro, wind and solar) and demand. Simply by recharging in peak sun hours (in solar heavy systems) or in low demand hours, you avoid use of the fossil back up (the marginal cost of electricity)
I shall recognize that I came from a system with prívate generation and nationalized distribution, and that probably change the focus of interest.
These seem like two sides of the same coin. You avoid putting power onto the transmission grid, by recharging local-distribution level storage, from excess production (with PV being the canonical example, but other variable producers like wind might actually provide excess at night, and distributed wind is starting to be a thing as well). Then you supply loads during lower-production periods, from that same storage.
To some degree you also can fill the storage by pulling excess production at centralized facilities, across the transmission system, during periods of low loads (as aggregated at the transmission level).
In any case, I see distributed storage as helping resolve this problem. It gives you more flexibility and resilience -- like I said, what we want is a _modular_ grid, that can react gracefully if a given transmission corridor is congested or down.
Would this not require rather substantial upgrading of the existing grids in nearly all locations?
There's a tendency to think of solar power and other forms of power like fossil fuels or nuclear as substitutes, but it's better to think of them as complements. They aren't competitors; they're all on the same team.
The optimal, lowest cost, lowest carbon energy mix for a country should have a mix of different forms of power so that they can cover each others' weaknesses. Fossil fuels are not verboten even in a zero-carbon future, because the excess energy from overbuilt solar capacity can power carbon capture to more than offset any fossil fuel emissions. The best way to provide power when the sun isn't up may actually be gas, rather than batteries, especially when you consider that battery manufacture has a large carbon footprint.
The goal should be to have the most abundant, reliable and cheap energy portfolio possible, rather than deciding which form of energy should "win" and kill off the rest. We have an essentially unlimited demand for energy, not just to improve human wellbeing, but also to gradually capture all the carbon we emitted since the Industrial Revolution and roll back all the damage. So we should be looking for the best team, not the best player.
Great point!
Not sure if it's outside his bailiwick, but I would love to see Brian (or someone) do an analysis that would try to calculate what the ideal mix would be.
We could just tax net CO2 emissions and find out.
You are understanding solar from an outsider perspective researching the pro solar propaganda. I worked in the solar industry for 6-8 years... and I still do some power systems designs for folks.
1.) No solar panel factory is powered by solar panels. (when it is... call me, I'll invest.)
2.) Your graphs are misleading they show nameplate capacity and don't speak to actual energy generation, a 100W solar panel only produces 100W at laboratory test conditions. In field it may produce a 100W or 0W, depending not only on sunlight but age of equipment (yes solar panels degrade over time), cell damage (hard to detect), and even dirt. Also the power conversion equipment (inverters) fail pretty frequently also means your solar array has a power rating of 0 Watts when this happens.
3.) Solar is NOT cheap you only show $/W module cost, what about inverters, copper wire, permits, labor, etc. I used to charge about $1000 per panel full installation. Now that has gone up to at least $1500, these prices are for medium size systems 10 to 40 panels. Each panel makes about $50 of electric per year. Inverters fail in about 6 to 12 years so factor a few kilo bucks in to replace those. I'm not even going to go into battery storage system headaches... Sorry bud, the tech ain't where it needs to be, wish this wasn't the case.
Re: 3.) Did you see the pie charts roughly 2/3 through the article, which show the $/kw figure incorporates all of the items that you claim were missed?
Yeah, I guess my point was that doesn't look right.... it's $1000 per panel not per kW, each module is about 250W to 350W so it's closer to $3000 per kW. Again, keep in mind you are buying an equipment rating of 1kW, you got to let the system run for 20 years to see what your cost per kWH is... and sure the Mfg says the product is warrantied for 25-30 years. Maybe the module might last that long, but solar panel companies don't last 30 years, good luck claiming a warranty. All the PV manufactures I bought from in the 2010's are gone now...
Are your numbers for residential installations or utility scale? The numbers in the post are for utility scale, and Brian does note that residential installations are significantly more expensive - so it seems like your numbers and his roughly line up?
I did work for big utilities and residential. Obviously you get a discount buying the equipment in bulk if your are a utility, let's say you get that initial cost down to $1000/kW. But then your O&M is much higher.... you got to install fencing, grounds keeping, security, legal, engineering, travel, accounting, and did I mention legal? But a big utility isn't just doing solar projects, so they have plenty of various overhead and can move solar O&M numbers around to make solar look better if they want. As years go by and more data gets collected on O&M, I think people are going to be surprised how costly it is. All things considered I'd say $3k per kW is good estimate of cost for both utility and residential, so I'd say cost estimates in this post are off by about 3x.
My experience is that solar never paid back—and that’s in Vegas. On my project it might have paid for itself after 20 years, but with assuming perfect efficiency on products that historically had a massive drop off after 15.
Yep, it's sad... I really wanted solar power to work, but Physics always gets the final say.
The advantage of solar, cost, is not as relevant as one might think given the other attributes of solar.
When investors look at a proposed project, the question they want answered is "when do I get my money back?". The size of the investment is only relevant if it is infeasibly large or small.
So cost does not matter, but revenue does. When the sun shines it shines for everybody, so every solar generator is in competition with every other one, as the duck curve demonstrates. So the market price of generated electricity crashes. When the sun does not shine, solar generates zero revenue.
Since solar only generates income when the price is low, it is unprofitable - the answer to the investors' question is "not for a very long time".
To date that uncomfortable fact has been circumvented with "tricks" like power purchase agreements by large consumers who can afford to spend on burnishing their green credentials, and/or taxpayer subsidies. Those subsidies cannot scale to high market penetrations of solar.
I strongly recommend Brett Christophers' book "The Price is Wrong", which explains all of this better than I can, and also explains several other aspects of renewable energy generation which are likely to limit its penetration.
Important points, and this is someplace where on-site batteries, virtual power plants, and demand-based pricing have a lot to offer. Also not discussed is grid congestion. On the flip side, I suggest that it's reasonable to consider power purchase agreements and REC requirements as government support for less harmful generation. The negative externalities and government subsidies for fossil fuel generation are well-documented, and reversing these requires active effort that sometimes involves supporting competitors to establish mature markets.
Great analysis except for two facts:
1 - you don't consider degradation of the panels: the solar industry uses an annual degradation factor of 1%, but those of us who model and track these projects know that the actual degradation factor is more realistically 2% per year; your analysis needs to take degradation into account
2 - the land area required for these projects is immense: in your single family situation, we need an area of 250 square meters (0.06 acres), which works out to over 10,000 acres required to power the households in metro Atlanta - assuming zero packing distance of panels (an unrealistic assumption). 10,000 acres of rooftops with unobstructed sky view is an immense amount of land to clear and devote to power production (and perhaps warehouse use), just to support the Atlanta metro....
From my work, I realize that the only reason solar is currently at all viable is thanks to government incentives - take away the incentives and solar collapses.
I did run an LCOE analysis that includes panel degradation (using 0.5% NREL estimate), it had a really tiny effect on LCOE, changed values by less than 5% (and in most cases closer to 1%). Reran it just now with a 2% degradation, and it had less impact than you might expect. ~20% LCOE bump for solar-only options but once you start adding storage and serving higher capacities it goes down, around a 5% increase.
This doesn't totally surprise me because for the higher generation fractions, panel capacity is significantly over what's demanded, and is very often wasted because storage is full, so falling panel capacity doesn't end up reducing provided energy all that much.
I haven't thoroughly verified this (and the script was vibe-coded with Claude), but I did look at it.
re: land area, Atlanta is one of the biggest metros in the country, a few square miles of solar panels to serve it (which there is plenty of space for, Georgia gets pretty rural not too far outside of Atlanta) seems fine to me.
I notice now that you are using a 7% discount rate - this is vastly lower than financial modeling of these sorts of projects use. You can argue about capital markets and greed and capitalism, but financial reality is discount rates in the realm of 20-25+%, and that's for established entities building additional projects: rightly or wrongly, the industry still views these projects with substantial long term uncertainty. (Projecting out 20 years for new industries is inherently uncertain!)
Relative to land area, the entire Atlanta metro is 136 square miles (Wiki), so we need a solar farm ~15% the size of the city just to supply the energy needed for household use. And as a US average, household energy use is only ~20% of the total energy budget (my recollection), so we are looking at a solar farm almost as large as the city itself to power the city. With that sort of scale, infrastructure and distribution costs (capital projects & transmission losses) become very substantial - and the loss of that land area likely dwarfs all else. (Land cannot be acquired for free and just imagine the eyesore of 100+ square miles of flat roof tops on big box buildings, all the same height - the Atlanta power park would make the DFW airport zone look positively beautiful by comparison!)
And Atlanta is not even in the top quartile of US cities by number of households!
Recently, in the past few years, hundreds of acres of farmland in Arkansas, Tennessee, and other locales has been converted to solar farms - and the impact of this is becoming significant: beware the unforeseen (but not unforeseeable) consequences of tax incentives!
Virtually every source I looked at lists discount rates in the 3-8% range for LCOE calculations, not 20-25%. For instance, IEA https://www.iea.org/reports/projected-costs-of-generating-electricity-2020
Do you have a source for the 20-25% figures?
20-25% is obviously wrong.
However, I would expect 7% to be somewhat too low in 2025 simply due to the current level of the risk free rate. Seems totally plausible this was an accurate number in 2020 when that IEA report was written though (as interest rates were lower then). As a more recent data point, Lazard's 2024 report (https://www.lazard.com/research-insights/levelized-cost-of-energyplus/) uses a 8% cost of debt and 12% cost of equity at a 60/40 split which averages to a 9.6% discount rate.
> just imagine the eyesore of 100+ square miles of flat roof tops on big box buildings, all the same height
Or... you could put them on the existing rooftops of Atlanta itself? Less transmission losses that way.
Turning several square miles of forest or farmland to solar farms is going to get a LOT of local resistance. And those are the most common locations for solar farms in the east.
"20-25% discount rates" - LOL. Discount rates for utility scale wind and solar are sub-10%.
Doesn't the US have huge government disincentives to solar? Last I heard they charge exorbitant tariffs.
250 sq m is also probably pretty close to the average usable area of a new home in the Atlanta area, and I would expect a lot of the homes do this without stairs. For those houses, their own roof area would suffice!
The demand curves are not fixed. You are assuming that excess energy produced will either go to storage or be wasted.
But industries and uses can spring up when there's cheap energy around.
That in turn would stabilise prices a bit, and make solar more profitable.
I had something in there regarding this in an early draft, but got some expert feedback that in practice we probably won't see much of this.
Interesting. Why, though? It wouldn't work for most current industries, sure.
But I can imagine a number of new ones for which this might work very well indeed - from desalinating large amounts of water over long periods to processing commodities at large scale or indeed produce hydrogen.
All of those activities need to be set up once and require huge amounts of energy but it doesn't matter all too much if you overbuilt the capacity (and doesn't cost that much more either) if you get the energy practically for free in return.
The CAPEX for the offtaker plus the storage cost for the product kill this for *almost* anything, espcially at the very low levels (~30%) for true excess, and the excess is very ‘spiky’ in that the amount is inconsistent as well.
CAPEX for desalination is pretty high, the price have come down yes. Seasonal utilization seems more likely though, like in western/northern europe which has a lot more seasonal variation in solar.
Heat batteries are promising in this regard. https://www.rondo.com/
That would be interesting to know about why! I suppose some uses like making hydrogen or aluminum probably don’t turn on and off very well, and others that do, like data centers, may be capital intensive enough that they want to run constantly if you build them.
Still, in a world where daytime electricity is too cheap to meter, I bet someone comes up with a clever new way to use it!
Someone already has: https://www.terraformindustries.com/
To make one final point, the last issue you’re not considering is the *marginal* economics of solar. Which is really to say, the closer you get to 100% solar coverage, the worse the economics get. Because the photons from the Sun are free, producers are always incentivized to sell power at any cost above zero. In a competitive power producer market, this means that on *most days*, a utility is paying nothing for electrons in the middle of the day, and a lot when the sun goes down.
Basically this is the duck curve issue. And it’s the reason why states like California stopped paying households for the solar energy that they generate, because electrons in the middle of the day in CA aren’t worth very much.
It’s actually worse than that, because if you look at the slope of the duck curve, it gets really steep and thus it is much, much harder for a utility to manage (vs a system that’s fully dispatch-able). Because excess electrons have to be discharged somewhere; and CA actually *pays* neighboring states like AZ exorbitant rates to discharge that power for them just to keep the grid balanced.
Interesting, thanks for the article!
It’s a bit hard to tell from the article though if the batteries are being used to provide power to market on a cyclical basis (as Brian is describing in his article) vs providing “ancillary” coverage for when a plant somewhere else on the grid goes down. If you know anywhere that has good data on ERCOT I’d appreciate a link, as I’d love to research this question more.
This question gets to the heart of my point though, which was more that increasing renewable penetration potentially becomes less profitable at the margin. Let me illustrate with an example.
Suppose ERCOT had enough solar/batteries to cover 70% of its demand *on average*. Practically speaking, that would probably be something like 7a to midnight per day on average.
Because the marginal cost to produce a watt from a solar panel (or a charged battery) is zero, rational producers would sell power at any price above $0. So really on an hourly basis a solar/battery operator isn’t making money until 1a to 7a (if you’re smart at least and keep your batteries from discharging when power is too cheap).
But what that does though is drive the capacity factor of the system way down, because it’s only effectively discharging when you can make money in the middle of the night.
So if you’re that same producer thinking about *ADDING* capacity, you’d be foolish to, because your new battery would probably depreciate faster than you are making money, so the IRR of that investment is negative.
All that said, I think it’s great (and probably inevitable) that a large percentage of generation in the US starts to come from solar and batteries. ERCOT is a great model and if more regions ran like that (or we had better transmission systems) I think that process would happen faster.
Zooming way out, my main point I want to make in response to Brian’s article is just that I don’t think we need to “overbuild” solar and batteries, because doing so you reach a point of diminish returns. I think the goal should not be to get 100% coverage or even close to it. Let the market do its thing, determine it’s own mix and if you want the government to subsidize/intervene in anything, it should probably be in electric vehicles because those have a much greater carbon impact anyway.
“Let the market do its thing…”. That’s the rub, we don’t let the market work. I wonder where renewables and EVs would be without all the subsidies.
Or Boeing? Or GM? The list of state subsidized US manufacturers is looooong.
This is why battery build out has been happening at a fast pace. Once solar gets above a certain threshold, you need batteries to time-shift production or you run into duck curves. California basically doesn't have a duck curve anymore because so much battery capacity has been built.
Texas is now building equivalent amounts of battery storage and it's working well:
https://www.utilitydive.com/news/texas-ercot-storage-deployment-saved-at-least-750m-since-2023-acp/735122/
As ever, the cost to the environment is completely ignored. It doesn't matter how cheap solar gets, overbuilding simply increases that cost as more and more of the natural world is destroyed to build the solar, and destroyed to deploy it.
Given all we know about the impacts of technology to the environment, why do people continue to ignore those externalized costs?
I would encourage you to redirect your ire to bio-ethanol - which consumes 1000x the land per kW of energy as solar.
Well it's not an either-or is it?! I still want to know why so many ignore the externalized costs, not matter what it is (solar, wind, bio-ethanol, etc.). I agree bio-ethanol is also a boondoggle and disaster for the natural world.
There is new research that well managed solar fields (with planted ground cover and grazing) can act as biodiversity havens. They're far better than mono-cropped bio-energy for nature: https://www.theguardian.com/news/2024/mar/01/weatherwatch-how-solar-farms-benefit-bees-and-butterflies
First of all, that's not how the massive industrial solar arrays all over the west are installed on public land. They first cut down all the joshua trees, yucca, pinyon, etc. then they bulldoze down the rest, to dirt, including scraping off the delicate desert soil crusts that keep all the soil from blowing away in the wind. They also crush all the endangered desert tortoise, lizards, and many other living beings out there. Then they fence it in. Then they install the panels. Then they have to use copious amounts of water to keep the panels clean because since they destroyed the soil crust and all the plants keeping the soil in place, the dust blows onto the panels and dramatically reduces their efficiency.
This also completely ignores the mining, refining, etc. to manufacture the panels (or whatever other energy technology you're analyzing).
Again, I'm NOT comparing. This isn't about "this or that". This is: Why do people who write articles like this one about "understanding solar energy" completely ignore the external costs to the natural world? It is a question. Why is that always ignored? While economists can conveniently wave these costs away in their models, that does not reflect reality. If we're interested in reality, then that should be part of the analysis and discussion. Yet when I bring it up you keep telling me to ignore because bio-ethanol. What?
Those are all very fair points Elisabeth. I agree that almost all energy projects have land use, biodiversity and pollution impacts and I agree they're almost never enumerated for either fossil fuel or renewable projects in most energy discussions.
Solar in Ireland (where I live) or England is generally on previously intensively cultivated agricultural land, so it's net positive from a biodiversity, emissions and water quality perspective.
When you account for the destruction to the land from mining the materials to build the solar panels, and the pollution from the mining, is it still a net positive? You don't see that mining because it's not in Ireland.
When you account for the coal used to refine silicon, is it still a net positive? You don't see the coal mining and the refineries for silicon in Ireland.
When you account for the huge pollution from refining the many materials to manufacture solar panels, is it still a net positive? You don't see that in Ireland either.
When you account for the pollution to build and run ships (on bunker fuel) to transport solar panel parts, probably from China, is it still a net positive? You don't see the impacts of all that bunker fuel in Ireland, or the steel used to build those ships, or the coal used to refine that steel.
When you account for the lax environmental law in the countries where solar PV is made, and in all likelihood, slave labor (or close to it) in those same countries, is it still a net positive? You don't see that in Ireland either.
When you account for what you do with all that energy from solar, i.e. running business-as-usual, the same business-as-usual that is causing the 6th mass extinction on Earth, is it still a net positive? You do see that in Ireland, one of the most environmentally degraded countries in the world.
I'm just wondering how the balance sheet is calculated on "net positive". Which is why I keep asking the question because very few, and not this article, will answer the question! I don't understand why. If we are to face the predicament we are in -- catastrophic ecological overshoot -- we must be honest about reality.
How is losing farmland a net positive? It typically just results in clearing land for farming elsewhere to offset the loss.
The externalized costs of coal are *staggering*, and yet suddenly people are clamoring for more nature when it's solar power. Like, where was all the concern 10 years ago when US grid was 50% coal???
"suddenly"? haha. Apparently you have no idea how long some people have been fighting the ongoing ecocide of our planet.
Also, almost everyone knows how terrible coal is at this point. However there is a large number of people that think solar is somehow better (despite that coal is required to refine silicon... hmmm...). So yeah, I don't need to say "coal is bad". That's obvious. However, apparently I do need to say "solar is also bad" many many times over.
I'm confused. What's your proposal? Treadmills?
I don't have a proposal. My posts here simply ask why the externalized costs of solar are not included in "Understanding Solar Energy." Those externalized costs are on the environment, on the natural world. But there was no mention of these costs in the article, and I wondered why (there seldom is in articles of this type).
We've created a society utterly dependent on non-renewable materials and extremely high energy, including on fossil fuels for fertilizer (9/10 calories we eat are fossil fuels in a 8.2 billion person world). There is no proposal. We are in catastrophic ecological overshoot and it will not end well. All species in overshoot collapse; ours will be no different. This is ecology 101.
I'm simply asking why an article titled "Understanding Solar Energy" does not include discussion of the externalized costs.
I agree that externalized costs are real and should be considered. I rarely see them placed in a context of tradeoffs wrt to renewables, though. And here in the US, petro-funded anti-solar "grassroots" activists are now commonplace. Re embedded energy of solar, for reference:
"the embodied carbon of solar in 2020 was around 615 kgCO2/kWp of installed capacity. This is 76% lower than the 2,560 kgCO2/kWp that is commonly referenced. First Solar‘s Global Sustainability Director also recently reported a typical value of 500-600 kgCO2/kWp for monocrystalline silicon.
Our estimate is likely conservative as four of the world’s largest solar manufacturers (Longi, Jinko, First Solar, and Hanwha Q-Cells) have signed up to RE100, committing to 100% renewable electricity supplies for their operations. Other manufacturers are taking a similar approach. Solar powered solar."
https://etude.co.uk/how-we-work/low-embodied-carbon-of-pv/
Hi Brian,
This is one of the best pieces I have read on the topic until the end. You gave a great overview of the pros and cons - very balanced and fair. Your analysis of the costs, and 4 days of storage feel about right. (Although this is still macro, and you would really want probably 4-6 days of storage in a no generation case, then need extra arrays to refil the batteries.
The question and problem I have with this kind of analysis - which you point out, but do not go into, is that to have solar, you either need to spend 40c per kw (according to your model, or 10x solar and 50kw battery (which is probably 8-10x a typical house battery), or you need two sources of generation.
You can not say the cost is cheap for 25%, but expensive for 100%, because if you have 25% solar, you need another 100% generation, so you need the second thing - it is mandatory. You need solar and nuclear, or solar and wind and battery or whatever - but you need two.
This is the cost of solar, or the costs should be modelled on 10x typical battery and the huge solar array - which is a true replacement - hence why I think LCOE should never be used - or use your 100% case as stated above.
Great work, I am going to recommend this article to many people I know.
Cheers
Scott
One thing I don’t understand is if solar is so cheap why does it need subsidies?
I think the argument is:
1. that it speed up replacement of existing generation.
2. makes it competitive earlier on the cost curve even in situations where other generation has other benefits.
1 Numbers don’t support the fact that there is any sort of replacement or “transition” going on. Rather adding capacity like it has always been. Any thermal plant closures are due to either reduction in local electricity consumption (always due to deindustrialization or increased energy poverty) or over optimistic hypothesis invariably leading to grid fragility.
2 Subsidies don’t make stuff competitive, they skew the market and give the impression of competitive, like smoke and mirrors. If you factor in all economic components in the cost of electricity such as synchronicity, availability, stability, then there is no way intermittent unreliable sources will ever be competitive.
I don't support subsidies I'm just presenting the arguments.
1. Well, subsidies -> lower prices of electricity from solar -> lower margins for thermal plants. If consumption decreases -> also lower margins. And if you've made solar cheaper it will push thermal plats out, even if they might have positive effects for the grid at large.
2. You are intentionally misunderstanding me.
Apologies if you believe I’m misleading you, that was not the objective.
On 1, if I understand well, by giving subsidies we could price thermal plants out of the market. What happens when all thermal plants are gone? Do we continue with the subsidies, do we let the market freely price energy, other? Apart from the economical aspects, is that approach even moral? In other areas that is called dumping and frowned upon, for good reasons. Either way, I don’t believe centrally mandated economics is the way to go or even feasible past the short term when we all run out of money.
On number 2, my point was that such policies do not lead to future price reductions, on the contrary, there will be an upward cost spiral, as we can see happening in all countries with high renewables penetration, and increased grid fragility.
Actually, points 1 and 2 converge into one as today’s subsidies destroy value and inexorably lead to increased costs and decreased services, for ever.
I recently purchased a book entitled “the price is wrong. Why capitalism won’t save the planet” by Brett Christophers. I haven’t read the entire book yet but the author’s main argument is “while prices of solar have tumbled the golden age of renewables has yet to materialize. The problem is that investment is driven by profit not price and operating solar remains a marginal business everywhere on the state’s financial support. We cannot expect markets and the private sector to solve climate crisis while the profits that are their lifeblood remain unappetizing.”
That may have been true at some point in the early 2010's, but it is no longer true due to cost decreases.
I strongly encourage you to read the book. Investors do not care about relative cost, only relative profitability.
I have not read the book, but I have read a summary of its arguments and a couple of critiques that think he's confusing past and present as well as discounting the coming wave of thermal and battery storage.
We shall see, I guess. That profitability is being ignored by the LCOE crowd is certainly a valid point.
A carbon tax would help. It would be better than subsidies.
But solar is getting cheaper all by itself right now, too.
The premise is a useful starting point. So let's consider the US EV tax credit. It's a government intervention that has led to 2024 production of 200GWh of battery capacity. Subsidies drive learning curves. A related and more "mature" example: wind revolution was driven largely by price, but only *after* prices dropped. For reference, in 2024, approx. 10% of power generation was from wind.
Great post.
Just would point out a few things to think about:
1. LCOE is a relatively flawed metric given it does not take into account intermittency (which you mention) as the costs of battery storage and the need for peaker gas plants do not get added to solar (or wind) LCOE metrics. Given solar plants do not function in isolation but within a complex system of the electricity grid.
2. Piggy backing on the below, synchronous generators provide ancillary services/value to the grid such as “grid inertia” that solar (and wind) do not provide. With the higher penetration of these resources, firms are building standalone facilities that mimic inertia, but at the end of the day do not flow into the cost of solar (again on LCOE basis) and a cost consumers will pay.
3. Given utility scale solar projects are typically further away from demand centers it requires higher transmission costs which again do not factor into solar LCOE.
4. Given how capacity markets work in grid systems (PJM) and the need to cover peak demand, solar cannot bid as well as competitive gas generators. With the continued penetration of solar and wind and the retirement of base load gas and coal, we have seen NERC reliability point to higher grid unreliability and have seen capacity auction prices ( ie PJM) sky rocket as a result. Ultimately, a cost paid by the consumer.
Given federal subsidies (mainly PTC), solar can bid negative prices in the electricity market which makes other base load generators uncompetitive in the wholesale market which is where you see early retirement of these assets and higher capacity pricing.
Very good article, just want to point out that “cost” of solar going down is true when you look at them in a vacuum, but not so much when you think of the system of electricity as a whole. There is a reason California has the highest (or one of) penetration of solar, but their electricity costs are the highest in the lower 48….
What do you think of cheap heat batteries improving the economics of solar? They can be used for seasonal energy storage (although the roundtrip efficiency of the system is low due to the heat-to-electricity conversion). Since about half of all energy demand is in the form of heat, you can directly supply the stored heat into industrial use cases.
I know someone working on this, broadly bullish on the idea (though I have no particular knowledge or expertise on how promising it is).
Brian, you might like to look at New Zealand's electricity supply system.
About three-quarters of NZ's electricity comes from hydro dams, and there is about six months' storage behind the dams. It's widely seen as "not enough". especially in dry years. Two large industrial consumers recently closed their doors, in part because the government was asking them to agree to having their supply curtailed at short notice.
If hydro storage, about the cheapest form there is, is undersupplied, then the prospects for every other form look fairly bleak.
With all due respect, hydro storage is not particularly cheap. It's very reliable wherever it is available, this is why countries like it. But building and maintaining the huge dam and turbine systems is not cheap. And it's an undertaking that has far bigger financing needs and cluster risks than building loads of small-scale battery storage sites across the whole country. These are two very different things, even though both price storage of sorts.
Thanks for that article. I think you missed one important measure. If you'd were a home builder you'd want to maximize the utiltiy of your PV installation. Since Power prices during noon are dropping fast due to the solar peak, you're utility function should contain something like: price(t)*P_PV(t). Vertical South Installations produce more power in Winter when you need it for your heat pump. Vertical East/West Installation produce more power in the evening when power prices are rising again. During noon vertical installation produce a lot less energy but the power at noon will be free in a couple of years. Since PV is growing exponentially, hourly prices during noon are gonna drop even faster. Since your PV installation is planned for 15 or 30 years, you should start installing vertical PV NOW.
Most Solar people still haven't grasped the idea.
If energy will be a bottleneck for industry in the future your case seem to imply that europe will be at a significant disadvantage due to seasonal variation in solar.