One thing you perhaps need to get your head around is the carbon price that DESNZ adds to gas-fired power. There is an economic case for adding the cost of the externality (harms of global warming), but that isn't what DESNZ do. Instead they add an figure which they call "the target-consistent carbon price". It is nothing to do with the externality, or even with carbon, but is simply an arbitrary value designed to make Net Zero happen regardless of the economics.
Your point about LCOE being an inappropriate metric. It is also noteworthy that DESNZ's 30% capacity factor for gas is being compared to unconstrained wind output. That of course is not what is happening now - we are paying windfarms all the time to switch off. The levelised cost of Seagreen, for example is in the range £86-111 if you pretend it is never curtailed. It's £269-390 if you don't. This is expected to get much worse because we are increasingly adding system imbalance curtailment to the thermal constraints costs.
My advice is to ignore all numbers coming out of DESNZ.
the government aim for just 5% of generation being gas.
This is not a realistic target and we actually need to get new gas generation on line as soon as possible, the current fleet is ageing and nuclear is on a downward capacity trend as they also are near end of life.
Unfortunately global demand for gas generators is increasing and lead times are several years forward.
We need to run gas 100% of the time as that is what keeps the grid in load and demand balance
and even at low output it provides essential technical attributes that wind and solar lack. It is a very poor way to run what can be extremely efficient generators, thus adding a cost to the consumer and increasing maintenance costs as well.
Essentially we are running duplicate generation where gas alone could do the job without the expense of building wind turbines and the extensive and unsightly infrastructure it requires.
The mistake was made over twenty years ago to ditch a planned nuclear expansion and instead go for unsuitable wind. Instead of learning of all the deficiencies years ago we continue with them at our very high cost. Rarely mentioned is that wind and solar output declines steadily with age and the life span is short relative to conventional generators. Impractical, costly, unreliable and at times very unstable with, to my mind no redeeming features in using renewable generation.
"Essentially we are running duplicate generation where gas alone could do the job"
If you build a grid where gas alone does the job then by definition the average capacity factor of the gas fleet will have to be lower as it follows load. Peak load can be double or more average load and the only way to handle that is to overbuild (lower capacity factors) or to add storage of some sort to handle the peaks.
For decades, grids have been supported by gas and diesel peakers that operate profitably with single digit capacity factors in spurts lasting for just a couple/few hours. Grids of tomorrow, with high renewable and battery penetration, will still need low capacity factor gas resources, but their run profiles will be entirely different. Instead of running for hours at a time on 50 or 100 days a year, they'll run 24/7 for a few dozen days a year. Instead of being there to provide capacity to the system, they'll be needed to provide energy.
A fairly typical number is that average demand is around 60% of peak demand, so that limits the average asset utilisation for a single technology grid. For dispatchable sources lowest cost is a mix of cheap baseload and low capital cost if higher fuel cost flex. Hydro can serve both in places like Norway and parts of South America, although Norway still gets years where hydro must be supplemented by net imports, and now risks shortages and high prices providing renewables balancing via interconnectors.
We are now entering levels of renewables penetration which can no longer just piggyback on existing grid assets. The result is a huge buildout of low utilisation transmission and stabilisation and storage that adds gigantic costs that are ignored in LCOE type calculations. Moreover, it is not economic to attempt to store and transmit all the surpluses on windy/sunny days, so we will see increasing levels of curtailment which push up the price required to secure revenues to pay for it all.
This is clearly in the UK the result of contracting out planning to the Grid, whose motivations have been to promote solutions that entail More Grid to maximise their asset base and profits. It's still the same people at NESO, with the same mindset.
More realistic economic appraisal is likely to lead to a revaluation on the Limits to Growth for renewables portfolios. Already we have Germany publicly admitting that its suppression of nuclear was a fundamental strategic mistake.
'Gas Plants' is too general a term. The problem is that that the gas lobby prefers combined cycle gas turbines (CCGTs) which are expensive both in captial terms and to run at low capacity terms. To balance wind and solar we need cheaper gas engines and (single cycle) turbines which aren't used much but which are much cheaper as balancing agents compared to CCGTs.
The continued use of LCOE to justify policy decisions on generation mix is utterly baffling … certainly to any engineer.
Unless that is you are trying to hoodwink your audience by pretending that a simple metric can suffice (in this I suggest it’s up there with the ‘wind and solar are free’ argument beloved by policymakers) or are just plain lazy (like investment banking analysts).
It is only useful if you are comparing ‘like for like’ forms of generation. As you touch on, it makes no sense in this context because solar and wind do not provide the same functionality as gas, they are intermittent, non-dispatchable and generally aren’t capable of providing ancillary network services like frequency and voltage control.
Either you reflect the very different services offered by pricing them accordingly (as usually happens in free markets) or you do a total system cost analysis over their lifecycle. The latter is the usual approach in industry when comparing different technology solutions for investment.
Since we have nothing remotely approaching a ‘free market’ for power the differential pricing option is never going to work. That leaves the total system cost analysis approach as the only credible decision making tool which from a slightly different angle is what economists like Dieter Helm have long argued for.
I think you’ve missed the point. I never argued renewables shouldn’t be part of the mix or that they can’t be competitive against other forms of generation. My point was simply that in determining what that mix should be for any particular system LCOE is a nonsense.
As for prices. For the UK in particular the market is so heavily distorted by subsidy mechanisms and other interventions (and that’s without considering what if any part carbon taxes should play) that any signals stemming from prices are largely meaningless.
The only credible analytical approach I know of is to look at the total cost of adding tranches of each type of generation onto the network.
You can do all the generic modelling you like but whilst it might be a pointer to what the global picture would ideally look like it doesn’t alter the need for each network to be assessed. Every country’s power grid will be different, their economy and demand mix will be different, the environment and thus capacity for wind, solar, hydro etc. will be different … and so will the optimal generation mix.
If you were king and could eliminate all the market distorting subsidies and policies while still doing your best to accurately price the externalities of each generation source (eg pollution, climate, reliability, national security, etc), what do you think the lowest cost generation mix would be given technologies available today?
I have no idea and I don’t think our policymakers know either which is why I’d like to see those scenarios worked through on a total system cost basis.
Have you looked into the work by Jenkins at all (Princeton) or by others on the optimal generation mix to achieve clean electricity goals with the lowest total system costs?
This research routinely finds that a fairly high mix of renewables is part of the solution with the lowest system cost. Yes, adding nuclear and even some gas into that mix is essential to keep overall costs down, but it's not like their total system cost based model optimizes renewable generation out of the picture.
I get that in places like the US, Europe, and even China, polisy considerations (eg subsidies) put their finger on the scale for renewables. However, given that the vast majority of new generation that's getting added all around the world is now renewables, even with no policy support whatsoever, I don't see how you can reasonably argue that markets haven't decided that increasing the role of wind and solar generation is the lowest cost way to go.
The assumptions behind the Capacity Market are that new plant can be built for a financing and depreciation annual charge of £49m/GW - the net CONE. That works out at £11.20/MWh at 50% utilisation and £18.67/MWh at 30%. Those costs are somewhat higher in reality currently, but almost certainly below the new DESNZ assumptions. It will be interesting to see how they adjust CM parameters in the light of their new stance.
I also question whether it is realistic to assume gas utilisation as low as 5% by 2030. NESO never reveal their modelling detail, but we know they don't model 8760 hour years let alone inter year variability in renewables output, and only consider an Average Cold Spell with real detail only for a portion of that. It's Goldilocks forecasting, with big assumptions (~10GW) about demand control that in reality imply power cuts. Reality is we will need much more gas.
Yes - great note and good discussion. As you point out, the government's public argumentation around the results of the auction is flawed. I wonder how much Miliband believes it (or does he just believe it is what he has to say?)
There may be an argument for this level of offshore wind, but we need somebody else to make it in a more balanced way than they did.
Nice to see a post triggered by comments from the last one. It's always interesting trying to disentangle the underlying reality from DESNZ press releases, especially when published on the same day.
Next difficult point on generation strategy will be revised transmission charges; what is needed is high costs for remote generation IE Scotland but that is unlikely to be made public before the May election.
Yes, I think we need to start admitting we are paying more for energy security. Even the argument from pro-renewables advocates that CfDs bring down the wholesale price accepts that overall prices don't move much becauee of CfD top-up payments. My first piece this year highlighted that gas set the price higher than current CfDs around 1% of the time in 2025, around three quarters of the time for the new CfDs to come excluding AR7, and just a third of the time at AR7 levels.
I wrote a bit about the changing role of CfDs in a piece last week (from reducing cost to insurance) and whether we could afford to go merchant on some renewable projects if prices are so high, rather than lock government into the elevated prices today.
I'm unable to contribute to the technical debate, but note as a non-technical bystander that the fact this comment section boasts such high quality commentary is a credit to Sam and his substack.
What I feel able to contribute, which hasn't been addressed by anyone in the comments section, is begging the question what is in fact the correct figure to be attributed to the carbon cost of fossil fuels. On some people's estimates, climate change will cost the human race multiple trillions in human displacement, housing and infrastructure damage and large scale water disruption etc etc. Some say climate change is a fantasy. If you accept the former, 30% seems reasonable. If you accept the latter, 0% seems reasonable. This debate matters as it contributes significantly to the cost of fossil fuel energy generation.
The Government supports the 30% figure, that being the case, wind infrastructure rollout by and large seems competitive even if the metric is substituted for a better one as advocated by those in the comments.
I hate to be simple about this but the Small Nuclear Reactor is surely the solution to this problem and it is time the government put enough pressure on the installers to prove this. Starting in the Anglesey project this could be proved, or disproved in five years. Is there no one out there who is aware that Britain has the highest energy costs in the Western World and wants to do something about it?
While these issues are central to the current energy market debate, we must shift the framework from 'affordability versus energy security' to 'affordability and energy security.'
Furthermore, decarbonization efforts should be evaluated through a full lifecycle assessment. For example, while gas plants emit carbon during operation, we cannot ignore the significant carbon footprint generated during the manufacturing and installation of wind turbines.
To achieve a productive outcome, the debate must evolve to account for the entire supply chain rather than focusing solely on end-use emissions.
"<LCOE> can mislead when you compare technologies. Nuclear runs 24/7 at a high capacity factor, wind is intermittent, and gas can fill the gaps. If you only look at the LCOE, you ignore the big benefits of baseload (nuclear) and dispatchable (gas and coal) forms of generation."
This criticism is correct, but it's incomplete.
LCOE also fails to capture the system costs imposed by the inflexibility of baseload technologies like coal and nuclear. The inflexibility of these plants, sometimes physical and sometimes just economic inflexibility, is a major system cost driver. Their inability to (economically or physically) follow load has to be overcome by blending more flexible and usually more expensive generation into the system.
IMO, inflexibility is very similar to intermittency but it doesn't get talked about nearly as much. Both of these limitations/constraints impose costs on the system and the magnitude of those costs can vary widely depending on the particulars of a given grid.
It soon starts getting really complicated unless you analyse on a whole system basis (including required transmission assets), and accept that demand, economically optimal supply patterns are variables so you can only hope for a fuzzy optimum, preferably fairly flat against real life variations.
Baseload generation has to remain below minimum demand where there is no export or storage (see Wylfa/Dinorwig) alternative. It's arguable that the 5.6GW at Gravelines includes offshore nuclear for GB via IFA1 and Eleclink, and ditto MPP3 at Maasvlakte plus BritNed was offshore coal instead of building Kingsnorth D - and much more costly because of the interconnectors. Of course the French use hydro and gas for balancing as well, and are finding out that adding wind increases their balancing requirement and costs.
Optimisation probably results in a bit less baseload to improve the economics of load following.
Just like with intermittency, the impact of inflexibility on system cost depends on the generation mix and load profiles. France can economically run their grid with 70% nuclear generation, but they do it by using imports/exports for grid balancing. So with their current setup, the cost of nuclear's inflexibility is minimal. OTOH, if everyone in Europe had the same amount of nuclear as France, the cost of inflexibility would show up as soon as nuclear generation exceeded 30-40%.
Same thing with solar generation and the cost of intermittency. On grids with 10-15% solar generation, intermittency barely impacts system costs. In fact, adding solar can significantly reduce system costs at those rates because the marginal generator when solar is producing is often an expensive gas or diesel peaker. This is especially true in summer peaking grids when much of the system's load is from AC units on hot and sunny days. OTOH, once solar generation gets above 20-30% grids see that solar "duck curve" and that's where the cost of inflexibility shows up with curtailment and other adverse system impacts.
All that said, batteries are changing everything in terms of how both intermittency and inflexibility impact total system costs. As battery costs continue to decline and as flexibility on the load side increases, it may turn out that both inflexibility and intermittency won't matter that much overall. We could end up back to a place where LCOE is a decent indicator of a generation source's impact on total system costs.
You are asking the right sorts of questions here.
One thing you perhaps need to get your head around is the carbon price that DESNZ adds to gas-fired power. There is an economic case for adding the cost of the externality (harms of global warming), but that isn't what DESNZ do. Instead they add an figure which they call "the target-consistent carbon price". It is nothing to do with the externality, or even with carbon, but is simply an arbitrary value designed to make Net Zero happen regardless of the economics.
Your point about LCOE being an inappropriate metric. It is also noteworthy that DESNZ's 30% capacity factor for gas is being compared to unconstrained wind output. That of course is not what is happening now - we are paying windfarms all the time to switch off. The levelised cost of Seagreen, for example is in the range £86-111 if you pretend it is never curtailed. It's £269-390 if you don't. This is expected to get much worse because we are increasingly adding system imbalance curtailment to the thermal constraints costs.
My advice is to ignore all numbers coming out of DESNZ.
Sam,
the government aim for just 5% of generation being gas.
This is not a realistic target and we actually need to get new gas generation on line as soon as possible, the current fleet is ageing and nuclear is on a downward capacity trend as they also are near end of life.
Unfortunately global demand for gas generators is increasing and lead times are several years forward.
We need to run gas 100% of the time as that is what keeps the grid in load and demand balance
and even at low output it provides essential technical attributes that wind and solar lack. It is a very poor way to run what can be extremely efficient generators, thus adding a cost to the consumer and increasing maintenance costs as well.
Essentially we are running duplicate generation where gas alone could do the job without the expense of building wind turbines and the extensive and unsightly infrastructure it requires.
The mistake was made over twenty years ago to ditch a planned nuclear expansion and instead go for unsuitable wind. Instead of learning of all the deficiencies years ago we continue with them at our very high cost. Rarely mentioned is that wind and solar output declines steadily with age and the life span is short relative to conventional generators. Impractical, costly, unreliable and at times very unstable with, to my mind no redeeming features in using renewable generation.
"Essentially we are running duplicate generation where gas alone could do the job"
If you build a grid where gas alone does the job then by definition the average capacity factor of the gas fleet will have to be lower as it follows load. Peak load can be double or more average load and the only way to handle that is to overbuild (lower capacity factors) or to add storage of some sort to handle the peaks.
For decades, grids have been supported by gas and diesel peakers that operate profitably with single digit capacity factors in spurts lasting for just a couple/few hours. Grids of tomorrow, with high renewable and battery penetration, will still need low capacity factor gas resources, but their run profiles will be entirely different. Instead of running for hours at a time on 50 or 100 days a year, they'll run 24/7 for a few dozen days a year. Instead of being there to provide capacity to the system, they'll be needed to provide energy.
A fairly typical number is that average demand is around 60% of peak demand, so that limits the average asset utilisation for a single technology grid. For dispatchable sources lowest cost is a mix of cheap baseload and low capital cost if higher fuel cost flex. Hydro can serve both in places like Norway and parts of South America, although Norway still gets years where hydro must be supplemented by net imports, and now risks shortages and high prices providing renewables balancing via interconnectors.
We are now entering levels of renewables penetration which can no longer just piggyback on existing grid assets. The result is a huge buildout of low utilisation transmission and stabilisation and storage that adds gigantic costs that are ignored in LCOE type calculations. Moreover, it is not economic to attempt to store and transmit all the surpluses on windy/sunny days, so we will see increasing levels of curtailment which push up the price required to secure revenues to pay for it all.
This is clearly in the UK the result of contracting out planning to the Grid, whose motivations have been to promote solutions that entail More Grid to maximise their asset base and profits. It's still the same people at NESO, with the same mindset.
More realistic economic appraisal is likely to lead to a revaluation on the Limits to Growth for renewables portfolios. Already we have Germany publicly admitting that its suppression of nuclear was a fundamental strategic mistake.
'Gas Plants' is too general a term. The problem is that that the gas lobby prefers combined cycle gas turbines (CCGTs) which are expensive both in captial terms and to run at low capacity terms. To balance wind and solar we need cheaper gas engines and (single cycle) turbines which aren't used much but which are much cheaper as balancing agents compared to CCGTs.
The continued use of LCOE to justify policy decisions on generation mix is utterly baffling … certainly to any engineer.
Unless that is you are trying to hoodwink your audience by pretending that a simple metric can suffice (in this I suggest it’s up there with the ‘wind and solar are free’ argument beloved by policymakers) or are just plain lazy (like investment banking analysts).
It is only useful if you are comparing ‘like for like’ forms of generation. As you touch on, it makes no sense in this context because solar and wind do not provide the same functionality as gas, they are intermittent, non-dispatchable and generally aren’t capable of providing ancillary network services like frequency and voltage control.
Either you reflect the very different services offered by pricing them accordingly (as usually happens in free markets) or you do a total system cost analysis over their lifecycle. The latter is the usual approach in industry when comparing different technology solutions for investment.
Since we have nothing remotely approaching a ‘free market’ for power the differential pricing option is never going to work. That leaves the total system cost analysis approach as the only credible decision making tool which from a slightly different angle is what economists like Dieter Helm have long argued for.
I think you’ve missed the point. I never argued renewables shouldn’t be part of the mix or that they can’t be competitive against other forms of generation. My point was simply that in determining what that mix should be for any particular system LCOE is a nonsense.
As for prices. For the UK in particular the market is so heavily distorted by subsidy mechanisms and other interventions (and that’s without considering what if any part carbon taxes should play) that any signals stemming from prices are largely meaningless.
The only credible analytical approach I know of is to look at the total cost of adding tranches of each type of generation onto the network.
You can do all the generic modelling you like but whilst it might be a pointer to what the global picture would ideally look like it doesn’t alter the need for each network to be assessed. Every country’s power grid will be different, their economy and demand mix will be different, the environment and thus capacity for wind, solar, hydro etc. will be different … and so will the optimal generation mix.
If you were king and could eliminate all the market distorting subsidies and policies while still doing your best to accurately price the externalities of each generation source (eg pollution, climate, reliability, national security, etc), what do you think the lowest cost generation mix would be given technologies available today?
In the UK?
I have no idea and I don’t think our policymakers know either which is why I’d like to see those scenarios worked through on a total system cost basis.
Have you looked into the work by Jenkins at all (Princeton) or by others on the optimal generation mix to achieve clean electricity goals with the lowest total system costs?
This research routinely finds that a fairly high mix of renewables is part of the solution with the lowest system cost. Yes, adding nuclear and even some gas into that mix is essential to keep overall costs down, but it's not like their total system cost based model optimizes renewable generation out of the picture.
I get that in places like the US, Europe, and even China, polisy considerations (eg subsidies) put their finger on the scale for renewables. However, given that the vast majority of new generation that's getting added all around the world is now renewables, even with no policy support whatsoever, I don't see how you can reasonably argue that markets haven't decided that increasing the role of wind and solar generation is the lowest cost way to go.
The assumptions behind the Capacity Market are that new plant can be built for a financing and depreciation annual charge of £49m/GW - the net CONE. That works out at £11.20/MWh at 50% utilisation and £18.67/MWh at 30%. Those costs are somewhat higher in reality currently, but almost certainly below the new DESNZ assumptions. It will be interesting to see how they adjust CM parameters in the light of their new stance.
I also question whether it is realistic to assume gas utilisation as low as 5% by 2030. NESO never reveal their modelling detail, but we know they don't model 8760 hour years let alone inter year variability in renewables output, and only consider an Average Cold Spell with real detail only for a portion of that. It's Goldilocks forecasting, with big assumptions (~10GW) about demand control that in reality imply power cuts. Reality is we will need much more gas.
Yes - great note and good discussion. As you point out, the government's public argumentation around the results of the auction is flawed. I wonder how much Miliband believes it (or does he just believe it is what he has to say?)
There may be an argument for this level of offshore wind, but we need somebody else to make it in a more balanced way than they did.
Nice to see a post triggered by comments from the last one. It's always interesting trying to disentangle the underlying reality from DESNZ press releases, especially when published on the same day.
Next difficult point on generation strategy will be revised transmission charges; what is needed is high costs for remote generation IE Scotland but that is unlikely to be made public before the May election.
Yes, I think we need to start admitting we are paying more for energy security. Even the argument from pro-renewables advocates that CfDs bring down the wholesale price accepts that overall prices don't move much becauee of CfD top-up payments. My first piece this year highlighted that gas set the price higher than current CfDs around 1% of the time in 2025, around three quarters of the time for the new CfDs to come excluding AR7, and just a third of the time at AR7 levels.
I wrote a bit about the changing role of CfDs in a piece last week (from reducing cost to insurance) and whether we could afford to go merchant on some renewable projects if prices are so high, rather than lock government into the elevated prices today.
https://open.substack.com/pub/esgstuff/p/if-renewable-energy-is-cheap-why
I'm unable to contribute to the technical debate, but note as a non-technical bystander that the fact this comment section boasts such high quality commentary is a credit to Sam and his substack.
What I feel able to contribute, which hasn't been addressed by anyone in the comments section, is begging the question what is in fact the correct figure to be attributed to the carbon cost of fossil fuels. On some people's estimates, climate change will cost the human race multiple trillions in human displacement, housing and infrastructure damage and large scale water disruption etc etc. Some say climate change is a fantasy. If you accept the former, 30% seems reasonable. If you accept the latter, 0% seems reasonable. This debate matters as it contributes significantly to the cost of fossil fuel energy generation.
The Government supports the 30% figure, that being the case, wind infrastructure rollout by and large seems competitive even if the metric is substituted for a better one as advocated by those in the comments.
I hate to be simple about this but the Small Nuclear Reactor is surely the solution to this problem and it is time the government put enough pressure on the installers to prove this. Starting in the Anglesey project this could be proved, or disproved in five years. Is there no one out there who is aware that Britain has the highest energy costs in the Western World and wants to do something about it?
While these issues are central to the current energy market debate, we must shift the framework from 'affordability versus energy security' to 'affordability and energy security.'
Furthermore, decarbonization efforts should be evaluated through a full lifecycle assessment. For example, while gas plants emit carbon during operation, we cannot ignore the significant carbon footprint generated during the manufacturing and installation of wind turbines.
To achieve a productive outcome, the debate must evolve to account for the entire supply chain rather than focusing solely on end-use emissions.
"<LCOE> can mislead when you compare technologies. Nuclear runs 24/7 at a high capacity factor, wind is intermittent, and gas can fill the gaps. If you only look at the LCOE, you ignore the big benefits of baseload (nuclear) and dispatchable (gas and coal) forms of generation."
This criticism is correct, but it's incomplete.
LCOE also fails to capture the system costs imposed by the inflexibility of baseload technologies like coal and nuclear. The inflexibility of these plants, sometimes physical and sometimes just economic inflexibility, is a major system cost driver. Their inability to (economically or physically) follow load has to be overcome by blending more flexible and usually more expensive generation into the system.
IMO, inflexibility is very similar to intermittency but it doesn't get talked about nearly as much. Both of these limitations/constraints impose costs on the system and the magnitude of those costs can vary widely depending on the particulars of a given grid.
It soon starts getting really complicated unless you analyse on a whole system basis (including required transmission assets), and accept that demand, economically optimal supply patterns are variables so you can only hope for a fuzzy optimum, preferably fairly flat against real life variations.
Baseload generation has to remain below minimum demand where there is no export or storage (see Wylfa/Dinorwig) alternative. It's arguable that the 5.6GW at Gravelines includes offshore nuclear for GB via IFA1 and Eleclink, and ditto MPP3 at Maasvlakte plus BritNed was offshore coal instead of building Kingsnorth D - and much more costly because of the interconnectors. Of course the French use hydro and gas for balancing as well, and are finding out that adding wind increases their balancing requirement and costs.
Optimisation probably results in a bit less baseload to improve the economics of load following.
How big of a system cost driver is it?
Just like with intermittency, the impact of inflexibility on system cost depends on the generation mix and load profiles. France can economically run their grid with 70% nuclear generation, but they do it by using imports/exports for grid balancing. So with their current setup, the cost of nuclear's inflexibility is minimal. OTOH, if everyone in Europe had the same amount of nuclear as France, the cost of inflexibility would show up as soon as nuclear generation exceeded 30-40%.
Same thing with solar generation and the cost of intermittency. On grids with 10-15% solar generation, intermittency barely impacts system costs. In fact, adding solar can significantly reduce system costs at those rates because the marginal generator when solar is producing is often an expensive gas or diesel peaker. This is especially true in summer peaking grids when much of the system's load is from AC units on hot and sunny days. OTOH, once solar generation gets above 20-30% grids see that solar "duck curve" and that's where the cost of inflexibility shows up with curtailment and other adverse system impacts.
All that said, batteries are changing everything in terms of how both intermittency and inflexibility impact total system costs. As battery costs continue to decline and as flexibility on the load side increases, it may turn out that both inflexibility and intermittency won't matter that much overall. We could end up back to a place where LCOE is a decent indicator of a generation source's impact on total system costs.
Your point about LCOE being an inappropriate metric. Yes ! ( cf. Szymański, Adam, A brief history of the LCOE definition - An update (July 26, 2021). Available at SSRN: https://ssrn.com/abstract=3893462 or http://dx.doi.org/10.2139/ssrn.3893462)