Future Nuclear Renewables

The Catch-22 of Energy Storage

Pick up a research paper on battery technology, fuel cells, energy storage technologies or any of the advanced materials science used in these fields, and you will likely find somewhere in the introductory paragraphs a throwaway line about its application to the storage of renewable energy.  Energy storage makes sense for enabling a transition away from fossil fuels to more intermittent sources like wind and solar, and the storage problem presents a meaningful challenge for chemists and materials scientists… Or does it?

Guest Post by John Morgan. John is Chief Scientist at a Sydney startup developing smart grid and grid scale energy storage technologies.  He is Adjunct Professor in the School of Electrical and Computer Engineering at RMIT, holds a PhD in Physical Chemistry, and is an experienced industrial R&D leader.  You can follow John on twitter at @JohnDPMorganFirst published in Chemistry in Australia.

Several recent analyses of the inputs to our energy systems indicate that, against expectations, energy storage cannot solve the problem of intermittency of wind or solar power.  Not for reasons of technical performance, cost, or storage capacity, but for something more intractable: there is not enough surplus energy left over after construction of the generators and the storage system to power our present civilization.

The problem is analysed in an important paper by Weißbach et al.1 in terms of energy returned on energy invested, or EROEI – the ratio of the energy produced over the life of a power plant to the energy that was required to build it.  It takes energy to make a power plant – to manufacture its components, mine the fuel, and so on.  The power plant needs to make at least this much energy to break even.  A break-even powerplant has an EROEI of 1.  But such a plant would pointless, as there is no energy surplus to do the useful things we use energy for.

There is a minimum EROEI, greater than 1, that is required for an energy source to be able to run society.  An energy system must produce a surplus large enough to sustain things like food production, hospitals, and universities to train the engineers to build the plant, transport, construction, and all the elements of the civilization in which it is embedded.

For countries like the US and Germany, Weißbach et al. estimate this minimum viable EROEI to be about 7.  An energy source with lower EROEI cannot sustain a society at those levels of complexity, structured along similar lines.  If we are to transform our energy system, in particular to one without climate impacts, we need to pay close attention to the EROEI of the end result.

The EROEI values for various electrical power plants are summarized in the figure.  The fossil fuel power sources we’re most accustomed to have a high EROEI of about 30, well above the minimum requirement.  Wind power at 16, and concentrating solar power (CSP, or solar thermal power) at 19, are lower, but the energy surplus is still sufficient, in principle, to sustain a developed industrial society.  Biomass, and solar photovoltaic (at least in Germany), however, cannot.  With an EROEI of only 3.9 and 3.5 respectively, these power sources cannot support with their energy alone both their own fabrication and the societal services we use energy for in a first world country.

Energy Returned on Invested, from Weißbach et al.,1 with and without energy storage (buffering).  CCGT is closed-cycle gas turbine.  PWR is a Pressurized Water (conventional nuclear) Reactor.  Energy sources must exceed the “economic threshold”, of about 7, to yield the surplus energy required to support an OECD level society.
Energy Returned on Invested, from Weißbach et al.,1 with and without energy storage (buffering).  CCGT is closed-cycle gas turbine.  PWR is a Pressurized Water (conventional nuclear) Reactor.  Energy sources must exceed the “economic threshold”, of about 7, to yield the surplus energy required to support an OECD level society.

These EROEI values are for energy directly delivered (the “unbuffered” values in the figure).  But things change if we need to store energy.  If we were to store energy in, say, batteries, we must invest energy in mining the materials and manufacturing those batteries.  So a larger energy investment is required, and the EROEI consequently drops.

Weißbach et al. calculated the EROEIs assuming pumped hydroelectric energy storage.  This is the least energy intensive storage technology.  The energy input is mostly earthmoving and construction.  It’s a conservative basis for the calculation; chemical storage systems requiring large quantities of refined specialty materials would be much more energy intensive.  Carbajales-Dale et al.2 cite data asserting batteries are about ten times more energy intensive than pumped hydro storage.

Adding storage greatly reduces the EROEI (the “buffered” values in the figure).  Wind “firmed” with storage, with an EROEI of 3.9, joins solar PV and biomass as an unviable energy source.  CSP becomes marginal (EROEI ~9) with pumped storage, so is probably not viable with molten salt thermal storage.  The EROEI of solar PV with pumped hydro storage drops to 1.6, barely above breakeven, and with battery storage is likely in energy deficit.

This is a rather unsettling conclusion if we are looking to renewable energy for a transition to a low carbon energy system: we cannot use energy storage to overcome the variability of solar and wind power.

In particular, we can’t use batteries or chemical energy storage systems, as they would lead to much worse figures than those presented by Weißbach et al.  Hydroelectricity is the only renewable power source that is unambiguously viable.  However, hydroelectric capacity is not readily scaled up as it is restricted by suitable geography, a constraint that also applies to pumped hydro storage.

This particular study does not stand alone.  Closer to home, Springer have just published a monograph, Energy in Australia,3 which contains an extended discussion of energy systems with a particular focus on EROEI analysis, and draws similar conclusions to Weißbach.  Another study by a group at Stanford2 is more optimistic, ruling out storage for most forms of solar, but suggesting it is viable for wind.  However, this viability is judged only on achieving an energy surplus (EROEI>1), not sustaining society (EROEI~7), and excludes the round trip energy losses in storage, finite cycle life, and the energetic cost of replacement of storage.  Were these included, wind would certainly fall below the sustainability threshold.

It’s important to understand the nature of this EROEI limit.  This is not a question of inadequate storage capacity – we can’t just buy or make more storage to make it work.  It’s not a question of energy losses during charge and discharge, or the number of cycles a battery can deliver.  We can’t look to new materials or technological advances, because the limits at the leading edge are those of earthmoving and civil engineering.  The problem can’t be addressed through market support mechanisms, carbon pricing, or cost reductions.  This is a fundamental energetic limit that will likely only shift if we find less materially intensive methods for dam construction.

This is not to say wind and solar have no role to play.  They can expand within a fossil fuel system, reducing overall emissions.  But without storage the amount we can integrate in the grid is greatly limited by the stochastically variable output.  We could, perhaps, build out a generation of solar and wind and storage at high penetration.  But we would be doing so on an endowment of fossil fuel net energy, which is not sustainable.  Without storage, we could smooth out variability by building redundant generator capacity over large distances.  But the additional infrastructure also forces the EROEI down to unviable levels.  The best way to think about wind and solar is that they can reduce the emissions of fossil fuels, but they cannot eliminate them.  They offer mitigation, but not replacement.

Nor is this to say there is no value in energy storage.  Battery systems in electric vehicles clearly offer potential to reduce dependency on, and emissions from, oil (provided the energy is sourced from clean power).  Rooftop solar power combined with four hours of battery storage can usefully timeshift peak electricity demand,3 reducing the need for peaking power plants and grid expansion.  And battery technology advances make possible many of our recently indispensable consumer electronics.  But what storage can’t do is enable significant replacement of fossil fuels by renewable energy.

If we want to cut emissions and replace fossil fuels, it can be done, and the solution is to be found in the upper right of the figure.  France and Ontario, two modern, advanced societies, have all but eliminated fossil fuels from their electricity grids, which they have built from the high EROEI sources of hydroelectricity and nuclear power.  Ontario in particular recently burnt its last tonne of coal, and each jurisdiction uses just a few percent of gas fired power.  This is a proven path to a decarbonized electricity grid.

But the idea that advances in energy storage will enable renewable energy is a chimera – the Catch-22 is that in overcoming intermittency by adding storage, the net energy is reduced below the level required to sustain our present civilization.

BNC Postscript

When this article was published in CiA some readers had difficulty with the idea of a minimum societal EROI.  Why can’t we make do with any positive energy surplus, if we just build more plant?  Hall4 breaks it down with the example of oil:

Think of a society dependent upon one resource: its domestic oil. If the EROI for this oil was 1.1:1 then one could pump the oil out of the ground and look at it. If it were 1.2:1 you could also refine it and look at it, 1.3:1 also distribute it to where you want to use it but all you could do is look at it. Hall et al. 2008 examined the EROI required to actually run a truck and found that if the energy included was enough to build and maintain the truck and the roads and bridges required to use it, one would need at least a 3:1 EROI at the wellhead.

Now if you wanted to put something in the truck, say some grain, and deliver it, that would require an EROI of, say, 5:1 to grow the grain. If you wanted to include depreciation on the oil field worker, the refinery worker, the truck driver and the farmer you would need an EROI of say 7 or 8:1 to support their families. If the children were to be educated you would need perhaps 9 or 10:1, have health care 12:1, have arts in their life maybe 14:1, and so on. Obviously to have a modern civilization one needs not simply surplus energy but lots of it, and that requires either a high EROI or a massive source of moderate EROI fuels.

The point is illustrated in the EROI pyramid.4 (The blue values are published values: the yellow values are increasingly speculative.)

Finally, if you are interested in pumped hydro storage, a previous Brave New Climate article by Peter Lang covers the topic in detail, and the comment stream is an amazing resource on the operational characteristics and limits of this means of energy storage.


  1. Weißbach et al., Energy 52 (2013) 210. Preprint available here.
  2. Carbajales-Dale et al., Energy Environ. Sci. DOI: 10.1039/c3ee42125b
  3. Graham Palmer, Energy in Australia: Peak Oil, Solar Power, and Asia’s Economic Growth; Springer 2014.
  4. Pedro Prieto and Charles Hall, Spain’s Photovoltaic Revolution, Springer 2013.

By Barry Brook

Barry Brook is an ARC Laureate Fellow and Chair of Environmental Sustainability at the University of Tasmania. He researches global change, ecology and energy.

642 replies on “The Catch-22 of Energy Storage”

I’ve been keenly awaiting this article. It should be entirely accessible for anyone who cares enough to want coal replaced and who realises that arithmetic matters.

It occurred to me during reading that it may be difficult to demonstrate chemical storage on even a single wind farm scale. Such batteries would be ideally be situated beside the turbines, but since buying wind energy is prioritised anyway, from where would the impetus come to invest in a capacity that would effectively eat into the megawatts being sold during times of generation? It sure doesn’t sound viable financially, as well as energetically uneconomic.

If there is an actual example in spite of this, I’d love to know about it!


Reblogged this on Colder Air and commented:
“It’s important to understand the nature of this EROEI limit. This is not a question of inadequate storage capacity – we can’t just buy or make more storage to make it work. It’s not a question of energy losses during charge and discharge, or the number of cycles a battery can deliver. We can’t look to new materials or technological advances, because the limits at the leading edge are those of earthmoving and civil engineering. The problem can’t be addressed through market support mechanisms, carbon pricing, or cost reductions. This is a fundamental energetic limit that will likely only shift if we find less materially intensive methods for dam construction.

This is not to say wind and solar have no role to play. They can expand within a fossil fuel system, reducing overall emissions.”


Great post. Thanks John. I remember being blown away by Graham Palmer’s coverage of these concepts and thinking that this is one of those things that is incredibly obvious once somebody has pointed it out, but thousands and of people have studied energy for a really long time and just plain missed it! Or perhaps somebody will find it mentioned in some thermodynamics tract from the 19th century … those early thermodynamics thinkers had some real smarts!


‘Todd’ asks,

is it even possible to build storage on the scale … solar would require?

We won’t know until someone tries. They’ll either try the below, or something better that I haven’t been able to see:

… heat from a source significantly hotter than 2500 K can usefully act on … an ore of iron – very much cheaper and more abundant: magnetite (Ehrensberger et al., 1997; Mohai et al., 2007):

(1-x)/(1-4x) Fe3O4 → (1/2) O2 + 3/(1-4x) Fe(1-x)O(liq)

For x=0, NIST data imply this process has enthalpy change 372.3 kJ/mol, plus another 39.2 kJ/mol that the oxygen would give back in being cooled from 2500 to 298.15 K. Being liquid, the ferrous oxide tends to separate from the oxygen, so they can be cooled without recombining. Losing the 39.2 kJ would be reasonable for a solar power station that focussed a large image of the sun down onto a high-altitude outdoor stream of magnetite, for then the half-mole of oxygen could go directly into the upper air. If such a station annually turned 32.6 billion kg of magnetite into 1.9 billion kg of oxygen and 30.7 billion kg of ferrous oxide, its annual average output could be expressed as 1 GW(FeO).

Where summer is much sunnier than winter, ferrous oxide production rates in winter, spring, summer, and fall might average respectively zero, 1, 2, and 1GW(FeO). By summer’s end, 7.7 billion kg of ferrous oxide, a gigawatt-season’s worth, could accumulate, perhaps as an outdoor conical heap 300 m across the base. If a steady year-round ferrous oxide gigawatt were taken, the iron by winter’s end would be in a slightly larger magnetite pile. Other kinds of gigawatt-season energy reservoir – two billion lead-acid car batteries, a cubic km of water raised 800 m – are larger or more costly or both …

(from my fireproof fuel paper)

Note, in the above we store solar energy before partially converting it to motor fuel or electricity. Another scheme like this involves insulated pits full of heated rocks.

One can find rocks that are unharmed by many temperature swings between, say, 250 Celsius and 750 Celsius in greater quantities than one can find magnetite, but to store the same gigawatt-season, one would have to. Who wants to compute the size of that pit?


“This is not to say wind and solar have no role to play. They can expand within a fossil fuel system, reducing overall emissions.”

Doesn’t this reduce your EROI of both systems? If you are only using the capital equipment part of the time and it still has a useful lifetime, then by timesharing, aren’t you going to get an even worse result. Or at best, the part time use of “renewables” will drag down the EROI of your fossil system?

I also do not understand this:

“Nor is this to say there is no value in energy storage. Battery systems in electric vehicles clearly offer potential to reduce dependency on, and emissions from, oil (provided the energy is sourced from clean power). Rooftop solar power combined with four hours of battery storage can usefully timeshift peak electricity demand,3 reducing the need for peaking power plants and grid expansion.”

You still have the energy cost to build the low EROI solar collectors. How is this a useful timeshifting? You haven’t time shifted. You’ve moved generation from a high EROI source (presumably whatever energy source was used to make the solar collectors) to a low EROI source (solar collectors). Why not just charge the batteries from nuclear sourced electricity during off-peak times. That would be timeshifting. Charging batteries with solar collectors is another example of low EROI unsustainable energy generation.

I think these two paragraphs were an attempt to throw a sop to the “renewables” supporters. That may be good rhetorical style. I’m more inclined to believe that what they need is shock therapy and to be told that their beautiful dream is nothing but an unworkable fantasy in no uncertain terms.


Solar PV without any subsidies can be done at 8 cents per kWH
And it stays at 8 cents for 30 years

It takes about 6 months of run time on a PV to create the energy needed to create the PV.

Nuke dies at 14 cents as per Kewaunee

Coined “solar deniers” LOL


Author of that paper used old data (from 2006/2007) and made numerous flawed conclusions at least for solar PV.

In recent years a lot of progress has been made in solar PV manufacturing. Processes have been optimized, amount of raw material reduced and solar cell efficiency increased. For example: 5 years ago 16g/Wp of silicon was needed, while now industry uses only 5g/Wp (at increased efficiency, which implies less inverters, racks, etc). Using outdated numbers leads to erroneous conclusions.

German Fraunhofer institute uses more up to date data and it says that energy payback is about 2.5 years for northern Europe and less than 1.5 years for southern Europe. Given that lifetime of panels is at least 30 years this gives EROEI of 10-20.

Click to access photovoltaics-report-in-englischer-sprache.pdf

Improvements in EROEI are still being made with no end in sight, at least for the foreseeable future. Concentrated PV has energy payback of less than 1 year so it has EROEI of more than 30, which is more than wind.

Wind is also improving EROEI by going for ever larger and more efficient turbines. The paper calculated wind EROEI based on 1.5MW turbine, but in reality 2-3MW is pretty much standard, with 5-8MW becoming more and more common (at least in northern Europe). More efficient turbines have better EROEI.

I agree that embodied energy of storage should be included. But just as a reminder of what’s feasible: Norway already has about 84TWh of hydro storage capacity (up to now it was only used for managing multi-seasonal water variations). If Germany wanted to go 100% renewable they need 35TWh. This means that already today they can do it with the help of Norway hydro and there’s enough spare capacity for a bunch of other countries to go along.

On the other hand electrochemical storage would still make sense (for transportation if not anything else). Batteries can be deployed locally and do not need transmission and there are certain EROEI advantages to that. But precise implications are very unclear.

The main flaw of the paper and article is that it uses outdated data, does not consider improvements that have been made (not to mention additional ones which are in the pipeline) and wildly extrapolates beyond any reasonable justification.


I’m more inclined to believe that what they need is shock therapy and to be told that their beautiful dream is nothing but an unworkable fantasy in no uncertain terms.

You can tell them, but they would rather believe their fantasies.  You can show them how the dream is just a delusion with facts and figures, but you’ll find that they are either generally or selectively innumerate (their will to believe is stronger than the influence of facts).

About the only thing that works on such people is social sanctions.  They are herd animals by and large, and throwing them out of the group if they express such delusions is something they take seriously.


How were externalities accounted for in these studies? PPM and mercury contamination have real monetary and health effects that are probably exempt from the calculation. Wind has bird and land use issues, but the cost of water & related wildlife impacts for combustion and cooling based systems would increase the O&M quite a bit. This is assuming of course, that the analysis above omitted these costs.

Further, there is a supposition that the storage would be specifically keyed to ephemerals rather than to regional overproduction. Storage allows for integration between baseload and ephemerals if baseload sources feed the storage: particularly in the case of nuclear, where ramp down rates aren’t nearly dynamic enough to deal with energy surges.


I have not looked at the mathematical model, but it seems like the potential impacts of improving energy efficiency, i.e., reducing the amount of energy necessary to do the work of building trucks, building bridges, keeping people alive, etc.- – a/k/a “doing more with less” – – is not accounted for under the framework presented here. It seems like improving efficiency would move the “economic threshold” line downward.

Thus, leaving out the dynamic effect of improving efficiency seems a weakness of the framework. After all, most people (present company excepted) don’t really care about energy; they care about trucks, food, etc., – – i.e., the things we can do with energy. If we can do more things with the same amount of energy then, the useful value of the extracted energy increases and the “economical threshold” should move downward. Right?

I think, therefore, that though this is an important analysis, an improved model would account for improving efficiency.

From what I gather, the institutional environmentalists and their gurus (e.g., Lovins) presume (without good evidence, in my opinion) that really dramatic increases in efficiency can be generally accomplished, and they further presume that when this improvement is combined with improved performance (e.g., increasing capacity factors) for their preferred energy resources, then these factors, together, overcome any potential deficiencies of they type discussed here. I have not seen any analysis showing that, but what I state above seems to reflect the contours of their argument as to why, generally, they say “we can do it all with renewables and efficiency” with a straight face.

I don’t think that is true, but, giving them their due, efficiency of energy use generally is increasing, and the conversion of diffuse energy to useful energy by their preferred resources are also increasing. (correct me if I am wrong on this)

Efficiency has increased and GNP (as a marker for useful “stuff”), does not march in lockstep with the amount of energy inputs, though it may have at another time. Further, efficiency of energy use is quite variable across societies (e.g., compare energy/GNP for European countries v. USA, Australia).

Also, improving energy efficiency across society would seem like it would reduce the energy invested in making the renewable energy facility itself, and thus alter the EROEI. A further complication is that the EROEI is not uniform. A breakthrough (think: the efficiency of energy used for computing) in one area may be very dramatic, while efficiency in another area remains more or less unchanged.

So it seems like these dynamic elements should be addressed in order for a mathematical model like this one to be more useful.

This is not to imply that any of this changes the superiority of nuclear, unless nuclear were, somehow, specifically excluded from the benefits of improving efficiency overall.

Moving the “economical threshold” downward and the energy productivity line of renewables upward may get them over the economic threshold. For those who, foolishly, in my view, want to cling to the notion that nuclear is categorically unacceptable, dynamic factors that might move renewables over the “economic threshold” would be proof that they are “good enough.”


If Solar efficiency is continues to increase and costs continue to decrease, these numbers will change dramatically. ( Batteries are going up in storage by 7% a year (doubling in storage in 7 years). In addition the number of recharge cycles is increasing as well. In the long term, battery cost may go down even more as recycled lithium becomes more widely available. At this instant in time, you are right. Are your right 21 years from now?


Are there any diffusion plants left running in the world today? I don’t think so. So, instead of an EROI of 75, nuclear should have an EROI of 105. The best just got better.


LOL, almost as funny a metric as those who promote nuclear as good because of energy out compared to volume of fuel. I hope you can see the irrelevance


Having a poke around Weißbach there are some astonishing assumptions in there, and little mention of how geography plays a role.

They assume 2000 hours a year for wind, for a CF of near 22%. Most of Australia has averaged closer to double that, at around 40%. That doubles the EROI for wind.

It also assumes 1000 hours for solar, based on German numbers. Well, much of Australia gets about 6 hours of sunlight a day, for almost 2200 a year. So multiply the EROI of solar by 2.2 in Australia.

The logic of the whole idea is flawed in my opinion anyway. Take an extreme example; every building has enough solar on it to cover its needs. So why then does solar have to provide 7x the EROI? Then the amount of storage required for each source is nonsense as well. No mention of efficient onsite use, which will significantly reduce the need for storage, which I think most sources drastically overestimate.

I also think the notion that this information is somehow useful or going to be used by anyone is fanciful at best. There is 3GW of solar in Australia, all on rooftops. Is the assertion that these 1.2 million households should not have bought their own powerstation and should have instead held off to fund a nuclear power plant? Seriously, what good is this information?


The essential problem with that 3GW of solar PV on Australian rooftops is the poor energy efficiency of the existing PV cells, not to mention the threshold insolation value before they even start to produce electrical output. The fact that the installers didn’t have to expend energy on constructing the buildings is only one aspect of the total energy input. They still required manufacture, transport and a substructure to mount them on, not to mention ongoing maintenance, all of which is part of the energy investment. Solar hot water systems have 2-3 times the energy efficiency of PV cells, and the latter need a substantial technology breakthrough before they can be truly said to pay their way.


evecricket, I don’t disagree with your points, but the study notes Germany, and that is the capacity factor I understand they get there. Their prices have gone up, they do have 190GW of capacity for a peak of 80+GW, and they did rapidly cut back feed-in tariff based on reduced capacity corridors (in solar’s case despite share of production being only around 5%).
There is some “good” in the information, even if it provides more a template for analysis than firm universal measures.


Utility scale storage has its uses. New concepts for such are reported. Some are found in
with a quite interesting thermal store written about, with a link, toward the end.

What such units cannot do is store energy for a protracted period. So, for example, if the risk of no wind generation for, say, 7 weeks is sufficiently high some other form of dispatchable reserve generation is required. In some localities this might be standby biomass burners.


Everyone must read this article. And then Weißbach et al. If you are as widely read as I hope there will be a bonfire of pushback. One expected tactic is “But you took away our factor of 3!” As Weißbach wrote:

In particular the so-called “renewable” energies have often been treated in a confusing manner by weighting their output by a factor of 3 (motivated by the “primary energy”) but comparing it with the unweighted output of other energies like nuclear.

Have you a rebuttal that is understandable by the average Sierra Club type?


Googling “nuclear eroi” turns up a number of articles which use the EROI from this paper:
Manfred Lenzen, Life cycle energy and greenhouse gas emissions of nuclear energy: A review (pdf)

It puts the EROI at… 5? Weißbach doesn’t mention this paper directly, but it turns up again and again as a “reliable” figure. Figures of 50+ are often dismissed as “industry” figures.

Sadly I think debating EROI against renewables is likely to be bogged down by which figures for nuclear you decide to pick. There’ll always be somebody who wants to rely on the lower figures, while quietly neglecting to investigate the age of the source data.


The Weißbach work was also analysed here This information is crucial for people to access when we have green political leaders like Parnell proclaiming solar, wind and tomorrow’s batteries are all we need for future prosperity, jobs, education, opportunities and presumably important transitions like vehicle electrification, scaled up desalination, emissions-free fertiliser production etc…

I’d honestly be interested to see a serious critique attempted. Although grid-scale chemical storage is obviously unachievable once one grasps the scales involved, the poor results for diffuse, intermittent renewables are perhaps less intuitive, even shocking for some. There will be denial, when what there should be is serious examination of Weißbach’s methods. Motivated by the desire to identify the swiftest, most realistic and effective path to national decarbonisation, of course.


Geoff Russell:

this is one of those things that is incredibly obvious once
somebody has pointed it out, but thousands and of people have
studied energy for a really long time and just plain missed it!

Yes. Its amazing we’ve travelled so far down this road without it (storage EROI impact) being identified as a problem. I just read a US DOE report from 2007 entitled “Basic Research Needs For Electrical Energy Storage”. Even in 2007, the EROI problem was not on the radar.

The Carbajales-Dale paper is the first one I’ve seen that calls for a research focus on embodied energy reduction. That paper is 2014.

EROI itself is a surprisingly recent idea. Hall first wrote about it in 1981. I assume we’d missed it because we had up till then been dealing with energy sources with very high net energy, and because system level thinking is remarkably uncommon. But this is no longer the case, and the EROI question must be top of mind as we enter a period of energy transitions.


Jeff Walther:

Doesn’t this reduce your EROI of both systems?

Yes it does. You can only expand wind and solar, and stored energy therefrom, within a fossil fuel system until EROI reaches a minimum acceptable threshold. Their greenhouse gas abatement potential is therefore strictly limited. The way Graham Palmer put it, wind and solar can mitigate some emissions from fossil fuels, but cannot replace them.

How is this a useful timeshiftimg?

I think these two paragraphs were an attempt to throw a sop to the “renewables” supporters.

Its useful for the reason I gave in the passage you quote: reducing peak loads. The system EROI is reduced, but the reduction is traded off against building more peak capacity into the grid, which we may want to prioritise. This is not a “sop”, its decent data from a case study of solar PV in Melbourne. The EROI – utility tradeoff is one that’s available to make if we choose. I fully agree that it is a poor second choice to solving the problem with nuclear, or even nuclear with storage.



“They assume 2000 hours a year for wind, for a CF of near 22%. Most of Australia has averaged closer to double that, at around 40%. That doubles the EROI for wind.”

The capacity factor for wind in Australia is not ~40% it’s ~30%. According to BP statistics last year it was 30.1% with 9.2 TWh of electricity generated from 3489MW of installed wind capacity.

That’s certainly higher than what is seen in Germany, but it’s nowhere near double.

You might think new capacity installations would have a higher CF and the average is dragged down by older installations, but looking at previous years this is very unlikely. In 2007 for instance there was 972MW of wind installed producing 2.9 TWh. That’s a CF of 33%. If new wind farms were around 40% an almost quadrupling of capacity in the last 6 years would have seen average CF increase markedly. It hasn’t, they have stayed between 25% and 33% since 2006 with a fair degree of year on year variability.

“It also assumes 1000 hours for solar, based on German numbers. Well, much of Australia gets about 6 hours of sunlight a day, for almost 2200 a year. So multiply the EROI of solar by 2.2 in Australia.”

This is mentioned in the paper:

“For locations in south Europe, the EROIs are about 1.7 times higher due to the higher solar irradiation, but a higher irradiation also speeds up the aging.”

So yes there are regions where the solar isolation is higher, but even if that meant the EROI doubled it would still be below 7 when used with storage, and slightly more when unbuffered.

“The logic of the whole idea is flawed in my opinion anyway. Take an extreme example; every building has enough solar on it to cover its needs. So why then does solar have to provide 7x the EROI? Then the amount of storage required for each source is nonsense as well. No mention of efficient onsite use, which will significantly reduce the need for storage, which I think most sources drastically overestimate.”

You talk of assumptions in the paper, then say that every building has enough harvestable solar resource on it’s surfaces to cover it’s needs? Can the Eureka tower in Melbourne cover all it’s residents electricity needs, let alone it’s energy needs all year round? Can any skyscraper or apartment block? Or even inner city or suburban homes?

I stand to be correct with good evidence that this is indeed the case, but that seems to me to be a really huge assumption that seems unrealistic.

Your are right that efficiency measures are very important, but the energy still needs to be produced from somewhere and use will never be able to be matched up to solar production. A lot of energy is needed in the mornings, that means storage is very likely to difficult to limit if you want each building to be off grid as you seem to be implying.

More importantly you seem to misunderstand the concept or EROI. It doesn’t mean that a possible off grid system needs to supply 7 times the energy used by that building.

It means that the energy used to construct those solar panels – and the energy used in the other stages of its life cycle – has to be seven time less than what is provided to the end consumer.

eg. If the building uses 7MWh per year, the input energy into providing the solar+storage system must be smaller than 1MWh.

“I also think the notion that this information is somehow useful or going to be used by anyone is fanciful at best. There is 3GW of solar in Australia, all on rooftops. Is the assertion that these 1.2 million households should not have bought their own powerstation and should have instead held off to fund a nuclear power plant? Seriously, what good is this information?”

The argument isn’t so much that renewables such as solar and wind are all bad and should not be part of the mix of solutions. More that that they will find it extremely difficult if not impossible to provide all our energy needs because the energy used to provide storage on top of what’s needed to generate the energy itself is too much to leave some left for sustaining society.



They assume 2000 hours a year for wind, for a CF of near 22%.

As quokka1 pointed out the world fleet CF is about 23%. If memory serves, that’s also about the CF of the Chinese wind fleet. The 22% figure is a much more defensible figure than 40%.

It also assumes 1000 hours for solar, based on German numbers. Well, much of Australia gets about 6 hours of sunlight a day, for almost 2200 a year. So multiply the EROI of solar by 2.2 in Australia.

Quoting Weißbach et al.:

in south Europe, the EROIs are about 1.7 times higher due to the higher solar irradiation, but a higher irradiation also speeds up the ageing.

Same applies to Australia. Read the paper again, the authors understand and address geographic variation, and make reasonable choices with sensible justifications for them. This is not to say there aren’t other locales with different profiles in all sorts of parameters, with corresponding local differences to the numbers presented. But these differences do not materially affect the conclusion.

Take an extreme example; every building has enough solar on it to cover its needs. So why then does solar have to provide 7x the EROI?

I’m not sure what to make of this. Every building doesn’t have enough solar on it to cover its own needs, nor could it. The enterprise of civilisation entails many activities beyond the capability of rooftop solar to support. Does it cover, eg., building the building on which its mounted? Mining of the steel, firing the concrete, smelting the glass, feeding the occupants, paving the road to the building, and many, many other things. “Efficient on-site use” is simply not in the ballpark of the scale needed to change the import of the analysis.

I spent quite a bit of ink both in the body of the article and in the postscript trying to explain why a large EROI, greater than 1, and approximately 7 (not “7x”), is required as a minimum. To answer your question I’d urge you to reread the paragraph that begins: “There is a minimum EROEI, greater than 1”, and the Postscript.

I also think the notion that this information is somehow useful or going to be used by anyone is fanciful at best.

Well lets hope otherwise because this is a serious problem with a number of proposed energy transition pathways. Maybe you could start by explaining it to your cohort.

There is 3GW of solar in Australia, all on rooftops. Is the assertion that these 1.2 million households should not have bought their own powerstation and should have instead held off to fund a nuclear power plant?

That power station on their roof was built by coal and oil. A substantial fraction of the energy it then produces in its service life is in effect of fossil origin. When you add the balance, that power station falls well short of the full societal cost of its own production. At the end of its service life, another injection of fossil fuel is required to put another module up on the roof. Yes, looking at the total system, if the intention was to either transition away from fossil fuel dependence, or eliminate greenhouse gas emissions, the nuclear plant would have been the better investment. This can’t be achieved by the same economic arrangements as the individual purchase of solar panels, but a recognition of the relative viability of the two pathways should inform energy policy and climate strategy.

Seriously, what good is this information?

Really? Heaven forfend we make existentially important choices from an informed stance.

This information is of no good whatsoever if you want to keep installing solar panels without introspection.


John, great post. I think there are three main lessons here.

The first is that the your post shows the necessity of taking a systems approach to trying to understanding these things. This is the sort of approach advocated by Ted Trainer, Josh Floyd, and Charles Hall and others.

Arguing, as evcricket does, that capacity factor matters doesn’t detract from the basic problem that storage essentially kills the EROI. I agree with evcricket that trading off overbuilt capacity versus storage can ameliorate the storage issue, but the basic problem is this – for about 5,000 hours a year, there will be zero PV generation in the NEM. The combination of PV and modest storage can provide network support but does not substitute as a high-EROI primary energy source.

Secondly, batteries are an enabler of portable devices that consume electricity (laptops, phones, EV’s etc), but are a severe limitation when they are an essential part of a primary energy supply system. Indeed, even within consumer driven, premium product categories, the battery is the main limiting factor – smart phone developers are hamstrung by needing to balance the need for 8-10 hours out of a single charge versus giving consumers greater phone performance.

And lastly, the redox chemistries of conventional batteries are not amenable to the sort of exponential improvement that we have been accustomed to with solid state electronics – despite a century’s separation, the Tesla Roadster’s lithium battery (120 Wh/kg) possesses a specific energy density only around 6 times that of Edison’s early nickel-iron battery (22-25 Wh/kg) that powered the 1914 Detroit Electric Car – no Moore’s Law here! Nonetheless, near-term lithium batteries will advance this – we could see commercially competitive lithium at 400 Wh/kg in the foreseeable future, but nothing remotely like an exponential Moore’s-type law.


Have you a rebuttal that is understandable by the average Sierra Club type?

Steve, I have had quite a few discussions about this. This is what I put in Energy in Australia (pg 46) , which provides a brief overview –

A complicating factor is that conventional PV LCA analyses are expressed in terms of primary energy, but since fuels have differing quality and usefulness (for example, a joule of electricity is more useful than a joule of heat from coal), there is an argument that the EROI should include some provision to account for the varying usefulness (Murphy et al. 2011). The standard use of primary energy provides a consistent framework for LCAs, but may not always deliver the most meaningful results.

For example, Raugei et al. (2012) argue that the EROI of PV should be include provision for the average electricity thermal generator efficiency (ηgrid) to account for the fact that PV generates electricity directly, rather than via a heat engine as occurs with most generation. Taking a typical grid efficiency of around 0.31 thereby increases the “primary energy equivalent” (EROI,primary energy equivalent) around threefold.

Indeed, in a context in which PV displaces the use of high-cost diesel in remote grids, discussed later in this chapter, the conversion can make a lot of sense. Other examples include end-uses dependent on electricity such as lighting and electronic devices.

However, the validity of such conversion is highly context specific and will not apply in most cases. It assumes high fuel substitutability and conversion efficiency from electricity to other fuels (Murphy et al. 2011) and also ignores the stochasticity of PV. Since electricity only accounts for 18 % of global final consumption of energy International Energy Agency (IEA) 2012, it is not obvious that applying a universal threefold conversion factor is appropriate; indeed, the conversion can also work the other way (Prieto and Hall 2013).

For example, liquid fuels are far more valuable than electricity for transport applications, and the electricity-to-wheels conversion efficiency is usually low; studies typically report an electricity-to-wheels conversion efficiency of no better than 25 % for electricity-to-hydrogen-based transport (see Bossel 2004; Shinnar 2003).

Transport makes up around a third of global primary energy, and in the case of aircraft, shipping, heavy road, mining, and other heavy equipment, liquid fuels appear to be a necessity for the foreseeable future (Smil 2010).


Here is a study of a 2.5ha diary farm that produces electricity from its own residues (10tonnes/ha dry biomass) with an EROEI of 8:1 .

Should notice that
1. The energy produced is a byproduct of farm’s pig and goat production, so its food producing capacity is not affected.
2. Since burnable fuel can be stored for months (biomass in this case) there is no need for aditional batteries or other form of storage to backup low production hours or days of wind or solar plants.

So biomass can provide both capacity for night/low wind times and this capacity is in itself obtained at 8:1 EROEI

The conclusions of a study will be altered alot from choosing premises e.g. by stating 3.5:1 EROEI quoted for biomass in the article it dismisses its real capacity of being both energy source and energy storage.


With an EROEI of only 3.9 and 3.5 respectively, these power sources cannot support with their energy alone both their own fabrication and the societal services we use energy for in a first world country.

The order of these EROIs is switched.


Graham, Thank you very much. I have been working through the Weißbach et al spreadsheet and data. The paper is a model of transparency. Moreover, if the reader prefers different assumptions they can recalculate the EROIs and even generate amended bar charts.

Also notable, Weißbach uses a simple-but-uncommon transparency technique of inserting into the notes columns of the Material Inventories spreadsheet the URL of the data source used (one of which has gone 404, such is life).

I cannot find anything in the Weißbach methodology to object to. Let’s hope that eyes are opened in the back rooms of the NGOs who persist in promoting the “renewables are all we need” policies. And of course they must purchase your book, which should nail the all-renewables coffin closed.

One transparency quibble: if the spreadsheet authors had used Excel named variables it would greatly accelerate the work of auditing what they are doing.


Here is a study of a 2.5ha diary farm … with an EROEI of 8:1

This is an interesting case study with the sort of positive result that’s possible with an integrated farming system. On the one hand, it puts some useful numbers to the inquiry, with the sort of result you would expect intuitively. But it’s also a reminder of how dependant we are on high-EROI sources – it’s all very well for a farm to be self-sufficient, but in Australia, only around 3% of the workforce is employed in agriculture. Every Australian farmer provides food, on average, for about 600 people.

If each farm can produce energy for its own needs, with perhaps some surplus, we could imagine a rural village community living off its own labour, producing surplus food and perhaps a little surplus energy. But what about droughts, bush fires or sickness? What about advanced society with higher education, advanced healthcare and pharmaceuticals, defence, the public sector, law and administration, the arts and sport, or caring for the aged and disabled? This gets back to John’s EROI pyramid and the necessity of high-EROI if we are to enjoy the richness and diversity of modern life without every other person being employed to produce energy.


What battery technology was used? How was this justified? Can the material inputs to build batteries be reduced? What are the storage requirements when 10% to 50% of the grid is fossil (or nuclear for that matter) or when reliability of the system is reduced? Why do different studies come up with wildly different (unbuffered) EROI estimates? How much can demand side management (i.e. make chemical feedstock when supply exceeds demand) reduce the requirements for storage?

For example, if the storage requirements of renewables do not become an issue until 75% penetration (world-wide), then it becomes a significant amount of time until this becomes an issue, by which time it’s likely significantly improved technology will exist. It’s a similar story with nuclear. GEN III is not good enough, but we can still implement a massive amount of GEN III before its limitations become a problem. So, how is relying on Gen IV nuclear (and believe me, I have read essentially every single discussion on this website), which like fusion has always been a few decades away, any more valid than relying on future renewable, storage, and smart-grid technology?

The message of this article therefore should be that further research and development is required in order to have the capability to build a system that is completely 100% carbon free, as the technology does not yet exist. It shouldn’t be about the way we reduce emissions now. Reality does not support that conclusion.

Also France implemented nuclear with gaseous diffusion enrichment technology, which many studies show has a lower (unbuffered) EROI than wind, oil, and coal. ( Murphy, D.J.; Hall, C.A.S. (2010)). That means they decarbonized their grid with a lower EROI source, not a high EROI source as stated, so it supports the opposite conclusion as that presented.

In the real world, we have a government that just got rid of the carbon tax. We still have climate deniers affecting policy. So, I don’t see discussions like this as being terribly relevant to the actions we need to make now.


The problem I see with such studies on the EROI is that they describe the final state of a transition towards an all renewables energy system. A final state we may never reach. I can see many critiques that can be deployed along these lines.

First of all, all wind & solar proponents advocate big energy savings. In essence, this is equivalent to a decrease in the required EROI to sustain our complex civilization.

Second, there is the argument of technological change following which, in the far off future, innovation will allow an increase of the EROI for your favorite energy source or storage system.

Third, and maybe most important, we’re quite far from the final state. Right now, we’re pretty much in the “unbuffered case” everywhere wind turbines and solar panels are built. Storage comes in the form of heaps of wood and coal, gas reservoirs, fuel tanks, uranium rods and dammed lakes. It’s not tommorrow that the need for storage of wind/solar power will be felt.

A more striking proof of the difficulty of implementing a wind/solar + storage system is based on a look at the costs of such a system. Not only are wind and solar expensive, but storage systems are very expensive. You have to invest in a reservoir, the bigger it is the more expensive it gets, and the costs increase probably faster than the size of the reservoir. There is also a facility that will transform the flow of energy into a store of energy. It will be expensive because this facility will never be able to operate 24/7: capacity factors will be low, which is very bad for the costs. Finally, if you must have seasonal storage, the size of the reservoir make it a financial challenge. The value of the energy stored must be financed in some way.


Do any of you think an all nuclear system is possible? Demand roughly halves over night, so either half the plants would need to turn off or they all need to ramp down by half. That’s an incredibly bad way to run a Rankine cycle plant.


@ evcricket,

France is very close though they have some gas and hydro. In future more demand side management could be implemented to reduce the amount of gas needed. For Australia the question becomes: what do we do on a hot summer day?



    "Do any of you think an all nuclear system is possible?  Demand roughly halves over night, so either half the plants would  need to turn off or they all need to ramp down by half. That’s an incredibly bad way to run a Rankine cycle plant."

Possible? Yes.

Desirable? No.

Best possible option? No.

If storage was added to nuclear – it’s not only applicable to solar and wind- it could easily deal with the ups and downs of supply without ramping up or down. Charge storage at night, then use it to supply the extra demand during the day. And with it’s high EROI, nuclear could still power society even with the extra energy storage would require.

But this to me is just as silly as arguing for a 100% renewable system.

We have plenty of options to choose from. Nuclear, hydro large & small, wind, solar, tidal, geothermal, biomass, waste methane, efficiency etc. Even CCS might have a limited role in the medium term. Especially in situations like cement manufacture or maybe even retrofitting coal plants.

They all should have a place and role if we are to succeed in combating climate change and providing energy to all the people on this planet so they can have a good standard of living.

What role they should have should be decided on a case by case analysis that is heavily based on context.

Hydro and nuclear would likely do most of the heavy lifting, but I don’t see why renewables and other forms should not also play a role in some situations and make significant contributions to our energy supply.

Rooftop solar with 4 hours of storage, while it’s EROI might be small, that could be offset by a reduction in peak load during hot days in areas with good insolation.

So is it possible to have 100% nuclear? Probably, but I doubt it makes sense. Other technology will suit some situations better than nuclear will, even if it’s tied with storage.

Arguments that come down to 100% nuclear, 100% CCS, or 100% RE are to me the best bet to make sure we can’t solve the crisis we are facing.

We need every tool we have at hand to do what it can where its best suited. We have been talking about this problem for 30 years, and we haven’t really gotten anywhere except a small reduction in the growth rate of greenhouse gases.

We simply don’t have the luxury of picking one type of energy source based on some sort of tribalistic battle between teams stuck to one set of solutions..

PS. Any response to mine or John’s replies to your first post?

PPS. As an aside I read a paper by Charles Hall – the ecologist who first came up with the EROI concept – that suggests that to provide a foundation to a modern society, the EROI needs to be around 20-30. So 7 is possibly a minimum to support civilisation, but not the modern one we have gotten used to in advanced economies and the developing nations are striving for.

It’s quite interesting and well worth a detailed read.

You can find it here:

Abstract: Abstract
The near- and long-term societal effects of declining EROI are uncertain, but probably adverse. A major obstacle to examining social implications of declining EROI is that we do not have adequate empirical understanding of how EROI is linked, directly or indirectly, to an average citizen′s ability to achieve well-being. To evaluate the possible linkages between societal well-being and net energy availability, we compare these preliminary estimates of energy availability: (1) EROI at a societal level, (2) energy use per capita, (3) multiple regression analyses and (4) a new composite energy index (Lambert Energy Index), to select indicators of quality of life (HDI, percent children under weight, health expenditures, Gender Inequality Index, literacy rate and access to improved water). Our results suggest that energy indices are highly correlated with a higher standard of living. We also find a saturation point at which increases in per capita energy availability (greater than 150 GJ) or EROI (above 20:1) are not associated with further improvement to society.


“Fear of Radiation (unnecessarily hasty evacuation and other measures) has killed 761 and radiation has killed none from Fukushima” as of August 07, 2012

573 certified deaths were due to evacuation-related stress at Fukushima. Zero due to radiation. As of February 4, 2012

Fewer than 100 died from Chernobyl radiation. The Chernobyl reactor was a primitive Generation One machine without a containment building. American reactors have containment buildings that can contain any accident.

A nuclear power plant can not explode like a nuclear bomb. A reactor is nothing like a bomb. I would have to tell you how to make a bomb and how to make a reactor to explain why. The reactor at Chernobyl did not explode like a nuclear bomb because that is not possible.

In the 1960s we recycled spent nuclear fuel. See “Plentiful Energy, The Story of the Integral Fast Reactor” by Charles E. Till and Yoon Il Chang, 2011. Also see:

We get 99.9% of our radiation from natural sources, called Natural Background Radiation. The total radiation in Fukushima is less than our Natural Background here in Illinois, USA.



Offtopic, so, as an aside, no-one’s talking about 100% nuclear. France and Ontario get by with approximately 75%/25% nuclear/hydro. This works economically, and in terms of the power dynamics, and is extremely low emissions.

I’m no fan of hydro. The Australian environment movement was blooded in a battle against hydro, for good reason. The hydro component could be eliminated by coupling molten salt energy storage with nuclear power. Cal Abel ran the numbers on coupling molten salt storage with an Integral Fast Reactor, based on the Andasol engineering.

This power system could provide both baseload and peaking capability in a single plant, at near zero emissions, with viable economics, and without the geographical limitations of hydroelectric power. And with closed fuel cycle fast reactor EROI of ~10^4 (as Edward Greisch mentioned above), storage works from a net energy perspective.

No-one’s developing this, but the technical elements are all there. So yes, an all nuclear system is, in principle if not current practice, possible.


Liquid Air Energy Storage (LAES) has gotten some press recently.  It’s not terribly efficient (50% in the implementation I read about) but it scales really well because it uses no scarce materials.

Dumping overnight excesses of nuclear generation into LAES systems would be one way to productively use it and also meet the daily demand peak without consuming fossil fuel.


Scott: actually read all of

We can’t build more than 2% of the required battery. It just isn’t possible. READ the list of references I gave you before spouting more nonsense.


Rob andrews: About 20 doublings. Times 7 years = 140 years.
1. We will be extinct in 40 years
2. There is no reason to believe improvement will continue that long.

Batteries are OUT. Concentrate on current technology. currently available. Eats nuclear “waste”


Edward Greisch,

This isn’t the comment section of Youtube. If you are going to respond to my post, try actually understanding it. What you have done is link an analysis (that I have already read and considered) which has makes many flawed assumptions and only supports the conclusion which its author made. Not yours.

An actual analysis could use models of renewable energy systems (or weather patterns over a large geographical area), a model of an electrical grid (including demand side management), as well as energy storage facilities. From here the “overbuild” of renewables, the amount of fossil and/or nuclear, and/or energy storage could varied to see how the resultant cost or reliability changes.

Studies like this have been done, many flawed, none as flawed as what you have presented here. Go run a search for “Renewable Energy Penetration” on IEEE Xplore rather than link to more Blog posts. It’s always amazing how little work by actual electrical engineers comes up in these discussions.


So the notion that such a large amount of energy storage is required simply vaporize. In Europe, very high levels of renewable penetration can occur with only transmission line upgrades. Yes, a lot of transmission lines is expensive. Point is, some effort needs to be made into quantifying these costs and comparing with the alternatives.

Also PRISM hasn’t been built nor is it licensed, therefore it’s not “currently available”. Try a decade, from the point it is decided when it will be built. The economics are largely unknown, and that doesn’t count the pyroprocessing plant.


@Edward Greisch:

Gasoline: ~10000 Wh/kg
Lithium batteries: ~200 Wh/kg

That’s a factor of 50, not a million. And considering inefficiencies of combustion engines and efficiency of electrical engines then batteries need to improve by about a factor of 10-15.

First lead acid were about 20-30Wh/kg, today’s best lithium are up to 400Wh/kg. So historically it’s been done. Batteries are certainly not ‘out of the question’, at least for transportation.

But storing primary energy is somewhat different issue. I don’t know how did the authors come to substantially different conclusion than other studies:

This study finds that pumped hydro can store 210 times more energy than it used for construction, lithium batteries 10 times more, and lead acid only 2 times more.

So if EROEI of solar is 10 then pumped hydro storage would reduce it’s EROEI to

k = 10/210,
(10 – k)/(1 + k) = 9.5, instead of 10

That’s assuming 100% of energy is stored, which is unrealistic but it proves that EROEI doesn’t change much. If lithium batteries are used then the difference is bigger:

(10 – k)/(1 + k) = 4.5, instead of 10

So lithium batteries reduce EROEI much more, but given that lithium batteries and solar PV are still improving this is unlikely to be a problem. For example one doubling of efficiency of lithium batteries and using concentrated solar PV (which has EROEI above 30) would give a combined EROEI of above 10.


Good article, but there are quite a few issues:
– There were arts even in very low EROEI societies. Think Greece and Rome or even folk arts 200 years ago. So that pyramid is meaningless.
– No way nuclear has that good EROEI, current plants cost even more than solar or wind (nuclear is an upfront energy sink).


Read Collapse of Complex Societies by Joseph Tainter.
What is important is across all energy producing systems we are moving to a lower EROEI. The improvements in solar and / wind do not negate this due to high upfront costs and the EROEI is not 50-100 like with conventional gas or oil. Not even in the same ballpark.

There is no way that the current civilization can survive an overall constantly falling EROEI ratio. You can argue how long the decline will take, or how fast will it decline, but it is inevitable and unstoppable (except a miracle cold fusion tech or similar).

Fossil fuel advocates often forget that EROEI on fossil fuels is dropping fast. Metal ore grades are dropping fast too. Infrastructure maintenance meanwhile is very high in energy costs.

There will be a collapse in the current economic and social paradigm. In fact, it is already in progress.


Edward Greisch: you are spreading misinformation. Are you working for the nuclear industry by any chance?

According to the majority of recent studies nuclear is THE MOST EXPENSIVE energy source. It is not high EROEI at all.

And surely more people have died due to radiation than you state, because there are no reliable official statistics for this (as nuclear is always state sanctioned, they have an interest in lying about it all the time).


And nuclear as zero emission??? That is a HUGE LIE.

Mining of uranium uses a lot of fossil fuels. Building the plant uses a lot of fossil fuels. Operating personnel is not using cars running on nuclear either.

Nuclear is not zero emission, it is just that the emission is taking place somwhere else. Just like the celebration that western industry now uses less energy. Yes, because we import everything from China where they burn a huge amount of coal.


As I understand electricity market planning, EROEI never is considered. For example, add a wind farm which mostly generates at night when wholesale market clearing prices are low. Suppose the wind farm needs a minimum price of US$0.048/kWh to avoid financial loss. Then adding 80% efficient utility scale storage which requires US$0.02/kWh when generating to pay for capital, interest and operations makes financial sense when the daytime market clearing price exceeds 0.048/0.8 + 0.02 = US$0.06/kWh.

The same applies to a nuclear power plant but I know of none which has an attached storage.

[For simplicity searate transmission costs are not considered.]


Thanks David – I would expect business decisions to be made just as you outlined: by considering the components of ROI (income, expense, financing, …). If EROEI isn’t reflected in the risked-ROI calculation then it isn’t relevant to the investment decision.

Question: would the prices you see be roughly a proxy for EROEI if the market was reasonably “pure”? By that I mean, subsidy-free and externalities incorporated. Say CO2 price of USD $50/ton.



Author of that paper used old data (from 2006/2007) and made numerous flawed conclusions at least for solar PV.

Using outdated numbers leads to erroneous conclusions.

The main flaw of the paper and article is that it uses outdated data, does not consider improvements that have been made (not to mention additional ones which are in the pipeline)

The Carbajales-Dale paper I cited for solar PV is from 2014. To be any more current I’d need a time machine.

That paper comes out of the Stanford Climate and Energy Group, which is highly motivated to find ways in which renewable energy can be made to work. And they canvas the current and in-pipeline advanced solar PV technologies: single-crystal (sc-), multi-crystalline (mc-), amorphous (a-) and ribbon silicon (Si), cadmium telluride (CdTe), and copper indium gallium (di) selenide (CIGS).

Their conclusion is unequivocal:

Since CIGS and sc-Si both run an energy deficit even before the inclusion of storage, they cannot support any level of storage. CdTe, mc-Si and a-Si can afford up to 72 hours of geologic storage [pumped hydro -jm], but fewer hours of either mixed technology or all-battery storage.

And there is a critical qualification to this conclusion: they assume the storage lasts forever – the energy cost of replacing batteries or whatever at end of life is not included in the calculation. They also exclude any round trip losses in storage. (These omissions are deliberate and explicitly acknowledged.)

Further, the energy intensity of battery storage is estimated by those authors to be about ten times higher than pumped hydro. So if pumped hydro storage is marginal, battery storage is right out.

Finally, Carbajales-Dale et al. use a success threshold of, simply, positive net energy, i.e. they’re looking for an EROI >= 1. This is too low – they need to consider a net energy of EROI >=~7 (or perhaps much higher) if solar with storage is to be a source of primary energy in a society sufficiently advanced to make solar panels.

So their conclusion really should be, solar can work with storage only in some advanced forms, and only with pumped hydro, not with batteries, and not at all if you require it to do more in society than just rebuild itself.


Frank Jablonski,

Yes, improving overall energy efficiency of the whole society would presumably reduce the societal EROI value. Different societies are structured differently, are more or less advanced, and operate under different cultural norms. We could, in theory, work a number of axes to support society at lower net energy. Indeed, we will have to.

Beware, however, the grandly-named Khazzoom–Brookes postulate, which posits that:

energy efficiency improvements that are economically justified at the microlevel, lead to higher levels of energy consumption at the macro level.

Whenever we make more net energy available, we tend to find things to do with it, which ultimately grows, not shrinks, societal energy use.


Rob Andrews:

At this instant in time, you are right. Are your right 21 years from now?

Yes, I expect so. Graham Palmer above observes that Tesla’s Roadster battery has only 6x the capacity of Edison’s battery of 100 years ago. Storage technology is not amenable to the same kinds of drivers as the information technologies.

These numbers are also unreasonably favourable. The storage impact on EROI is based on pumped hydro. This is the best case. Battery storage is about 10x worse, according to Carbajales-Dale et al. Any storage technology that requires complex engineering and purified materials, such as batteries, or molten salts, and so on, will probably be somewhere between pumped hydro and batteries.

Pumped hydro is limited to suitable geographic locations, so would be unlikely to play a major role in primary energy supply. If the more scaleable storage technologies were used, even decades down the track actual performance is probably worse than Weißach’s numbers.


Tim Dettrick,

Sadly I think debating EROI against renewables is likely to be bogged down by which figures for nuclear you decide to pick

Irrespective of reported values for nuclear EROI, the key point of the article stands: low EROI energy sources cannot be paired with storage for primary energy supply.


@John Morgan: graph of EROEI in the article is from Weißbach. If you follow paper’s references you find that for wind he uses 1.5MW wind turbine from year 1995 and for solar PV data from 2005/2006 (updated 2007). This strikes me as being outdated.

I’ve looked into Carbajales-Dale paper (it’s open access: ) and it does use much more recent data. It says that energy payback time (EPBT) is:

on-shore wind: 0.34 years
sc-Si solar PV: 2.04 years
mc-Si solar PV: 1.34 years

They assume 25% capacity factor for wind and 11.5% for solar. This gives you EROEI of 58 for wind and 15 and 22 for solar PV (assuming 20 and 30 year lifetime respectively). This is similar to what German Fraunhofer institute says, but significantly different from Weißbach.

Weißbach should be updated with newer data and take into account technology improvements that have taken place.

It’s true that when Carbajales-Dale discuss storage they’re only interested in threshold of EROEI >= 1 and only during growth phase. But the data they provide can be used to calculate what happens if we want EROEI >= 7 in a sustainable steady state economy.

Assuming we want 72h of storage then energy payback time for wind increases from 0.34 years to about 1.1 years (see figure 3 and 4). If lifetime is still 20 years then that gives you EROEI of 18, which is well above 7.

For mc-Si solar PV adding 72h of storage would increase energy payback time from 1.34 years to about 2.4 years (again see figure 3 and 4). If lifetime is 30 years then EROEI is 12.5, which is above 7.

But these are numbers for steady state economy, not growth economy. Growth economy would reduce EROEI significantly because a lot of energy is invested just to grow energy supply.

Growth economy is the situation that Carbajales-Dale is discussing. When he says that “CIGS and sc-Si both run an energy deficit” this is because they’re growing too fast. This implies that on global scale solar PV cannot sustain such extreme growth rate (65% for sc-Si) if it wants to become a significant energy source, but it doesn’t mean that more modest growth rate (say 20%, as in ribbon cells) or that a sustainable steady state economy isn’t possible. The numbers Carbajales-Dale provide show that it is possible.


@blackVoid: (and some others) the “E” in EROI is “Energy” not fossil fuels. You could calculate the EROI of uranium mining assuming that your entire mining fleet of trucks is battery powered or nuclear powered or solar powered (somehow!). The article isn’t about emissions of CO2 per mega joule, it’s about the mega joules in for the mega joules out. Please think long and hard about that before posting responses. When you write “U mining uses lots of fossil fuels” you show that you haven’t quite grasped what the article is about. To repeat (again) the issue being discussed isn’t emissions but EROI.


@ John Morgan

I am aware of the postulate added to the (Jevon’s) paradox.

All the same, if we take GNP as a proxy for “stuff” and the ratio of energy/GNP changes such that energy goes down relative to GNP, then it seems like the EROI necessary to sustain a society in the fashion-to-which-it has-become-accustomed should be reduced.

Of course, this metric of measurement is simplified. It ignores, for example, the degree to which the energy that goes into “stuff” that makes its way, partially or wholly, into the GNP of another nation, is produced offshore. That energy should be imputed back into the energy side of the energy/GNP ratio.



The supplementary material of Carbajales-Dale discusses some of the assumptions –

a) they have converted the embodied energy of storage to primary energy equivalent by dividing the calculate primary energy by 3. The justification for this is that many energetic inputs to electrochemical storage manufacture and deployment are either currently, or could in future be electrified. See my earlier post This is a problematic area and would require an additional thread to draw out the issues.

also, in the main paper –

b) the battery embodied energy only includes the initial battery but not replacements. A typical PV system may require 3 or 4 battery replacements during a 20 to 30 year life. (section 4.2)

c) the conversion and inverter losses are excluded. No provision is made for inverter replacement or additional losses and costs associated with operation

In conclusion, a slightly more rigorous treatment of storage could easily increase the energy debt of storage 10-fold.


I like solar energy because it has about an EROEI of about 10, but realize that is like 2.5 years worth of average use (in the sunny areas) JUST to be able to “make itself” much less power all the processes involved as you explain.
Additionally, what is the energy input for various different types of batteries? Would it be “energy cheaper” to convert the direct electricity to heat an electrode to make liquid fuels such as ammonia, or even by direct electrolysis?
As for nuclear, I wonder about that too, for better load following. Of course, there should be less storage required with the reliability from some kind of meltdown proof nuclear.



Wouldn’t the much greater amount of mass required to build pumped storage incur higher overall energy inputs than higher density storage options? Efficiency of overall battery production can include robotic “workers” which don’t have to expend energy for all human needs and transportation. (I admit, don’t know enough chemistry to prove myself wrong). Recycling may reduce “overall” inputs as well.
Nuclear made synfuels might be the better option, even though far less efficient from tank to wheels.
From an efficiency point of view, it seems that batteries would be best for direct electricity sources like wind and that the cheapest possible molten salts or other medium for the thermal sources such as high temp nuclear. I can’t believe that thermal storage, being of less mass, would require more energy input than the massive scale structures of pumped storage, especially, if they last as long.



The Gemasolar CSP plant with molten salt storage has a nameplate capacity of just of just under 20 MW. The storage is 7,900 tones of 60% potassium nitrate and 40% sodium nitrate.

Click to access Molten_salt_tower_plant_GA_Azcarraga.pdf

That’s about 400 tonnes/MW or 400,000 tonnes/GW and definitely “non-trivial”. I guess it would have to be of higher purity than agricultural grade products.



There are a couple of studies of the energy inputs for batteries and pumped hydro and compressed air energy storage, both by Barnhart, both cited in the Carbajales-Dale paper. They conclude batteries are about 10x more energy intensive than pumped hydro or CAES.

The reason for this is the diversity of materials used in batteries and the very high purity required. Any process that requires highly purified materials will typically have high energy inputs, and batteries require very pure materials, and very advanced materials.

Purification takes energy. You’re fighting the Second Law all the way up the concentration curve. There is a relation shown in a so-called Sherwood plot between the price of a commodity and the extent of purification required to produce it. Its fairly universal behaviour, as shown in these two examples. The price is something of a proxy for the amount of energy required to yield the purified commodity. It follows a log-log law.

Batteries required very pure materials for cathodes, anodes, electrolytes, and the salts and solvent and polymers etc. they’re made from. Some of these aren’t just the product of purification, but may also involve multistep syntheses, each with their own separation and purification steps, and losses along the way. Some of the components are advanced microstructures materials, the product of long material process chains between the final product and the ore.

Pumped hydro in comparison requires concrete and steel production and earthmoving. Energy intensive materials to be sure, but compared to batteries quite short material transformation chains from the ores. And, depending on the geography, you can put quite a lot of water behind a single dam wall.

I don’t know of any studies of molten salt energy inputs. I imagine its somewhere between pumped hydro and batteries. You need a large mass of purified salts in a complex engineered system to stored a rather small amount of energy with large round trip losses. Geoff and quokka note some other limitations above.



Pumped storage is a much better option than thermal storage.

Assume you need to store 1 gigawatt year of energy. That is not a lot in today’s society. Norway has about 10 GWy of stored hydro energy, and we’re just 5 million people.

Click to access FOLDE2013.pdf

1 gigawatt year (1 GWy) is about 8.8 TWh, or 3.2*10^16 joule.

Imagine you were to store this as thermal energy in a huge, insulated block of steel by heating it to just below melting. I’m not saying that steel would be the optimal material, it’s just an example. Assume you can heat it by 1000 degrees C to store the energy. How much steel would you need?

Steel has absorbs 0.45 J per degree per gram, so 1 metric ton absorbs 450 MJ when heated 1000 deg C. You need about 70 million tons of steel. That’s several percent of the yearly world production.

But you stored heat, so you will lose about two thirds of the energy if you need to convert it to electricity. You need 200 million tons of steel, and it’s still a drop in the ocean.

Any attempt to solve the problem with batteries etc will run into the same fundamental issue – there’s simply too much energy that needs storing. If you try to use lead acid batteries you will have exhausted the entire worlds known and estimated lead resources long before you’re done installing a mere week of backup power in the USA.

Alternatively, you could build a dam across a valley and pump about 4 cubic kilometers of water one vertical kilometer up there. Round trip efficiency is about 80 %. This alternative is going to be quite a bit less expensive. The problem is that most poulated areas don’t have access to large valleys 1000 meters above sea level, and building 1000 meter tall structures containing four billion tons of water is not realistic.

Compressed air in underground caverns is another possibility which actually works fairly well, but then you need a huge, airtight cavern. Making it will consume too much energy.


Thanks, everyone for the links!
There sure is a lot of options with the high EROEI of nuclear. I’m still not sure which option is best: Thermal to clean fuels (and then less efficiency), or direct electricity to (energy intensive) batteries, at rather high efficiency.


I remain disappointed with the lack of interest and support for the Integral Fast Reactor. It is a shame that if the reactors in Japan had been IFR’s they would have just shut down with no need for cooling water and they would not have the problem the now have.


[…] Is Solar PV even a source of energy when one considers trying to ‘buffer’ it with storage? Does the energy cost of building the solar PV AND the storage render solar PV a net energy SINK rather than energy source? Or, in other words, do you pour more coal and gas and oil into building solar PV + storage than you get back as ‘clean’ energy? Apparently so! Not only this, but we need a minimum of 12 times the energy return on energy invested (ERoEI) to run the modern world. Solar thermal + storage only gives us 9, and that’s the best performing! Sorry folks. The ERoEI of a renewable grid + storage seems to be too low. Nuclear has an ERoEI of about 75. It’s nuclear or it’s climate change. The science says so. […]


Paraphrasing Doonesbury,

“If some abundant but fickle government money wanted you to, could you call a very effective, very high-ERoEI fossil fuel combustion preventer low-ERoEI?”

“High ERoEI? High ERoEI my foot!”


Too bad no-one wants to rise to my challenge, above, about the insulated rock pit. It’s bone-easy. Just multiply suitable rocks’ volumetric heat capacity, across some reasonable delta ‘T’, by a volume packing fraction, then divide into a gigawatt-season — which could also be accurately called a gigawatt-quarter-year, if you’re not comfortable with using season as a time unit — and you’ve got your pit volume.


Hi all,
my apologies to the moderator for posting this in the ABC Catalyst thread: I don’t know how that happened.

However, given this is where we are discussing the ERoEI of renewables I thought I would raise a quite considerable paper that critiques nuclear ERoEI: a paper that did the rounds a few years back, and seems to have gained some traction in high places. It’s by Australian Professor Manfred Lenzen, who is:

>>Manfred Lenzen is Professor of Sustainability Research at the University of Sydney, and leader of the Integrated Sustainability Analysis research group.

Physicist by training, Prof Lenzen has a background in energy systems and especially renewable energy sources. After emigrating from Germany to Australia in 1995, he took up a postdoctoral position at the University of Sydney in order to work on the vacuum glazing project. In 2001, he and his colleague Christopher Dey founded ISA, a centre for developing leading-edge research and applications for environmental and broader sustainability issues, bringing together expertise in environmental science, economics, technology, and social science. Since then, ISA has been prominent in the media.

Prof Lenzen has led numerous projects providing advice to all levels of government and also numerous businesses. For example, he was commissioned by the Australia Federal Government to undertake research on a first-ever Triple Bottom Line assessment of the Australian economy, and on the life-cycle energy consumption and greenhouse gas emissions of nuclear power. Further, he has created a popular online Environmental Atlas.

Prof Lenzen has (co-)authored more than 100 articles in the international peer-reviewed literature. He is also Editor-in-Chief of the ISI-listed journal Economic Systems Resesarch.”<<>>What an interesting paper! Although it is 6 years old, it succinctly covers a lot of ground and from an Australian perspective, to boot.

The introduction makes clear that only LWR and HWR reactors were considered, so Type IV was not considered. That answers EN’s first question.

Off the top of my head, that suggests that the majority of the upstream fuel emissions due to mining and processing will be eliminated by Type IV, but how much?

Perhaps this subject is worth a review, bringing this paper and those on similar topics together and introducing reliable extension to Type IV and SMR, because EROI will be one battleground where global energy futures are fought, won and lost.<<<
posted 4 hours ago by singletonengineer


What happened? A whole middle chunk of my post disappeared? Some formatting issue in the way I quote with arrows like these? >>> ?

I tracked down the PDF from his 2008 paper
“Life cycle energy and greenhouse gas emissions of nuclear energy: A review”

It’s discussed at:
Carbon Brief

The Conversation

I’d love to see some really careful work analysing his findings, as if nuke’s ERoEI really is 5, then solar thermal’s buffered 9 is the way to go with a grid topped up by the higher ERoEI of ‘raw’ (unbuffered) wind where possible, falling back on solar thermal’s ‘buffering’.

Which forces me to ask: did the paper above analyse mixed sources? What if we have higher ERoEI sources kicking in when they can, and then lower buffered sources taking over when they drop off? Won’t that raise the overall ERoEI of the entire system?


Here is a comment (at the other site) that has links about nuclear
The graphic clearly suggests that the molten salt reactor would reduce material inputs (and thus embodied energy) by many factors over conventional (once through) nuclear.
And here is a link about ESOI, from Standford
The total of stored energy delivered divided by the total of energy inputs is ESOI. To my surprise (and as others have already told me, here, batteries fair VERY low (but still deliver more than they require to make)!


Mike Stasse asserts incredulously:

There’s NO WAY nulear has an ERoEI of 75. Complete pie in the sky. You’d be lucky to get TEN

He cites BNC in the post at his link.  Maybe he’d also take Next Big Future’s take on material inputs for new nuclear.

AP1000 is 42t(steel)/MW(avg)
Concrete is not given, but 1970’s PWRs were 190 m³/MW(avg).

Energy required to make steel is 19 million BTU per ton (about 21 GJ/MT).
Energy required to make concrete is roughly 1.1 MJ/kg, or about 2.64 GJ/m³.

Major material inputs are (882+502)=1384 GJ per MW of average output.  This energy is recovered in (1,384,000 MJ / 1 MJ/sec) = 1,384,000 sec = 16 days.  That’s assuming that the energy to build is counted against energy output; if it’s against raw thermal energy, it’s repaid in about 5 days.

If we assume 16 days, the energy of the materials used in construction is repaid about 23 times per year, or 912 times over the initial 40-year licensing period.  It’s pretty obvious that 75:1 is pessimistic, and Stasse’s assertion of 10:1 is delusional.

Given that the energy overhead of centrifuge enrichment is about 0.1%, the EROI of nuclear power plants must be well in excess of 100.

The big problem is lawyers; adding lawyers to the process can reduce the EROI of anything to less than 1 in short order.  The solution is to shoot them, which diminishes their billing rate tremendously and practically eliminates actions filed.


The reasoning in the article’s quote from Hall seem unassailable, but quantifying the details does seem surprisingly murky. Lenzen does good work and his estimate of 0.2 Kwh-th for each kwh-e (which implies a thermal EROI of about 15) for nuclear has to be taken seriously. So why is it so different from the Weissback estimate of 75 (the WNA also has a figure in this region but the components are very different). Why the huge range? Somewhere there’s some very different assumptions being made but I haven’t spent the time to sort it out. Perhaps it doesn’t need sorting out. Here’s what we know: First, this dependency of advanced societies on high EROI fuels is a correlation … it’s not an F=ma type law. If you look at the Lambert/Hall paper cited above ( ) you can see that the trends are clear, but they are just that, clear trends but in a cloud of points with plenty a long way from the trend lines. So perhaps advancements in renewables will push them up a little higher in the EROI stakes and perhaps that will be enough. I thought this article was a slam dunk argument against renewables with storage when I first read it, but I’m not quite so sure any more. Nor am I going to waste much sleep over the Lenzen low figure because we know nuclear can power advanced societies because it does … it’s EROI is high enough, whatever it is. We also know that the EROI of fast reactors will be much, much higher.


///Nor am I going to waste much sleep over the Lenzen low figure because we know nuclear can power advanced societies because it does///
If the Lenzen figures are correct, then France cannot be taken as proof that nuclear’s ERoEI is enough because the French don’t use nuclear power to mine uranium. They use oil (diesel, etc). As you know, the high ERoEI of an electrical supply system is especially important if we are considering dumping some serious ERoEI points into replacing liquid transport fuels. EV’s, hydrogen, synfuel, boron – basically whatever the next big thing is – all require a power source so abundant that it can dump quite a few ERoEI points ‘charging’ up the fuel replacement. So, hypothetically, nukes could have an ERoEI of 5 if France is borrowing of oil’s historically high ERoEI.

PS: Engineer Poet’s calculations above are very comforting regarding the power plant, but the uranium supply system still needs analysis.


but the uranium supply system still needs analysis

The freely accessible Weissbach et al. paper provides this:

… The publication by Hoffmeyer et al. [45], used already for coal power plants, turned out to be a good
basis for a nuclear power EROI evaluation as well. It describes, however, too low energy consumptions for Uranium extraction, and the inventories for working chemicals used for that are missing. Here, the mass flows as described in the essay by Leeuwen [47] has been used but the old inventory data mentioned there had to be replaced with modern ones (see attached spreadsheet [11]) …

But if you don’t want to download and read a lot of stuff, here are two shortcuts. (1) Unenriched uranium produces about 180 thermal MWh per kilogram in the Darlington plant near me. Let us suppose an embarrassingly large number of thermal MWh of diesel fuel, say 18, had to be used to get that kgU. Diesel fuel’s energy density is 0.010 MWh per litre, so that would be 1800 litres.

Diesel costs have to be somewhat higher in northern Canada than they are where I am, more than $1 a litre, but let’s go with a dollar a litre. (Does that link work for you, ‘EclipseNow’?)

That gives that the kilogram uranium should cost way more than $1800, if its energy is 10 percent converted diesel fuel energy, because there are several other significant costs. But in fact it’s selling for — multiply the price given at by 2.60 to convert from pound U3O8 to kilogram U — $81.

Shortcut 2: can you find photos of uranium mines in oil-and-gas company PR? You can definitely find ones of wind turbines. If uranium mines are good customers, they’ll be there too.


@Eclipse now

Mining is largely done using electricity.

France gets much of its uranium from open cut mining in Niger. The only thing that needs the oil is large dumpers, which cannot easily be connected to the grid for obvious reasons.

Blasting agents are typically made from natural gas and nitrogen from the air.

The loaders, slurry pipelines, crushers/milling equipment, conveyors and everything else is electric.

The finished product from mining, yellowcake, is so energy dense that it barely matters energy wise if you transport it by boat or by horse and carriage.

Then there is enrichment and fuel fabrication. Centrifuges and what little remains of gaseous diffusion is all-electric.

Sweden gets most of its uranium from open cut too. From , total energy use per kWh from Forsmark NPP (from mining, to operation, to decomission and disposal of waste).

Renewable material resources
Wood 5,0E-04 g (2.25E-6 kWh thermal)

Non-renewable energy resources
Crude oil 4,1E-01 g (5.2E-3 kWh thermal)
Hard coal 8,6E-01 g (7.2E-3 kWh thermal)
Lignite (wet) 4,0E-01 g (2.7E-3 kWh thermal)
Natural gas 3,0E-01 g (4.5E-3 kWh thermal)
Uranium in ore g 2,1E-02 (470 kWh thermal in fast breeder. This is typically not included in EROEI for the same reason coal is not included in coal EROEI)
Peat g 5,4E-02 g (1.9E-4 kWh thermal)

Renewable energy resources
Bio mass (dry) 5,8E-02 g (5.8E-2 kWh thermal)

Potential energy through hydro turbines 2,0E-03 kWh
Solar electricity 8,8E-10 kWh
Wind electricity 6,2E-08 kWh

Electricity use in the power plant 2,0E-02 kWh (2% of electricity generated is used in the plant for its operation and is thus not typically included in EROI, it is instead subtracted from electricity generated).

In parenthesis is my estimate of the energy content. That’s not entirely accurate (e.g. for the crude oil estimate I have used the energy density of diesel fuel).

Total energy use, mixed units (kWh thermal and kWh elecric just added) per kWh nuclear energy: 0,022 kWh

Adjusted by a factor 1/3 for thermal sources, except for NG which is adjusted by a factor 1/2. This represents how much electricity you could have made with this fuel if you had burnt it for electricity instead (note, much of it is ACTUALLY burnt to generate electricity, and then used for mining and refining uranium and so on). Per kWh of nuclear energy: per kWh 0.0094kWh.

So EROEI is about 50 – 100 depending on how you massage the numbers (adjust or don’t for capacity of energy form to do work)

The gas and oil fill unique functions. The coal and other electrical sources can be substituted easily.

“electricity returned on oil and gas invested” (kWh electric / kWh thermal) = 103.


Some other fun facts from the Forsmark EPD:

High level waste volume (reactor core components and spent fuel) per kWh: 6,4 cubic milimetres per kWh, of which 3,3 mg is spent fuel elements of which 2,3 mg is uranium which passed through the reactor unchanged and most of the rest is cladding etc.

Total emitted CO2 is 4,3 g/kWh (se especially diagram on page 24)

Change in land use, when grouped in biotopes: critical, rare, general and “technotopes” (e.g. buildings, concrete, asphalt) per kWh of electricity generated is:

critical: -3,8 mm^2
rare: -2,7 mm^2
general: +2,4 mm^2
technotopes: +5,1 mm^2.

I’d love to see such numbers for wind (including rare earth mining, access roads, foundations, power lines … ).


Eclipse Now writes:

the paper I cited concludes ERoEI of 5 because it also counts the energy cost of mining, milling, and enriching the uranium.

Lenzen cites cites Storm and Smith on the very first page.  Any paper that does that must be considered ispo facto fraudulent at the outset.  It astounds me that all associated with such fraud have not been removed from their academic positions.  They must have powerful protectors… and who’s more powerful than fossil fuel interests?

Let me perform another very simple sanity check on the 5 number (after grlcowan’s fine pass).  Since we’ve already established that the energy cost of the steel and concrete of the plant itself is minuscule, let’s guess what would have to go into the fuel if that’s where the balance of invested energy is going.  Assume:  thermal energy output 45,000 MW-d/ton heavy metal, enriched to 3.5% U-235 with 0.2% tails, and a price of $31/lb of yellowcake.  If your payback is 5:1 you’ve got 9,000 MW-d of energy invested per ton of enriched U, 1422 MW-d per ton raw uranium (15.8% yield, remainder tails), 711 kW-days per pound.

A kilowatt is 3414 BTU/hr, so 711 kW-d is 58.3 million BTU.  Typical bituminous coal yields around 25 million BTU per ton, so about 2.3 tons of coal (or the energy equivalent) would be needed to produce one pound of uranium.  The nominal price of bituminous coal in the USA has been over $50/ton for years.  So Lindzen’s number, which you take without question, is roughly equivalent to saying that someone is investing the energy of well over $100 worth of coal to produce a pound of uranium… and then selling it for $31.

The same energy input from petroleum would cost several times as much (see grlcowan’s envelope-back above).

Nobody who believes such a thing can be called sane.  Nobody who says such a thing should be called anything but a fraud and a liar.  And for Lenzen to still be on the faculty of the University of Sydney, instead of having been drummed out for academic misconduct, shows just how deeply the academic study of “sustainability” is corrupted.


Thanks all: I’m going to have to re-read your posts a few times for it to really sink in, but it sounds like there’s been some seriously dodgy misinformation spreading virally and that depresses me no end. No wonder society is so misinformed about energy. The Storm and Smith work spreads virally through Lenzen’s paper to some significant places, and then the peaknik doomers use it to shout “We told you so!” Shameful.


Yes, thanks indeed E-P, Soylent et al but I’m persuaded more by the $100 for $31 argument than the abuse at the end E-P. I remember now that Peter Lang ran a similar argument (somewhere) … that the price of U was simply inconsistent with a low EROI.


Weißbach et al’s equivocation of the portion of GDP spent on energy with an EROI threshold is without basis, the EROI of a process can be squared by running it twice. For example a process with an EROI of 3.5 run twice has an EROI of 12.25 with a 29% increase in non-energy costs, run it 4 times and EROI is about 150 for 40% cost increase. There is clearly no hard limit to EROI.

For a steady state economy Total cost = One cycle non-energy cost ⨉ EROI ÷ (EROI − 1)


Soylent: “France gets much of its uranium from open cut mining in Niger. The only thing that needs the oil is large dumpers, which cannot easily be connected to the grid for obvious reasons.”

How much of that oil could be replaced with electricity using these two technologies?


@David B Benson. That bountifulenergy blog is a good example of how easy it is for someone to misunderstand these kinds of concepts and produce rubbish numbers.


I love that your response to a paper that disagrees with your world view is to discredit it. It was a report for cabinet, prepared by a reputable institute. But the results put nuclear in a bad light so it gets the short treatment.

Can any of you name a paper you read recently that caused you to doubt your commitment to nuclear power?


EVcricket, that’s psychobabble. I was the one pushing the Lenzen paper here for analysis, OK? This isn’t some Big Lebowski “The Dude abides!” This isn’t about my gut feeling, and just picking which authorities to trust. It’s about what gets through the more objective worldview crunchers of mathematics and engineering and real world science. And lastly, my argument about France’s ‘hypothetically’ low nuclear ERoEI ‘borrowing’ from oil’s higher ERoEI just doesn’t cut the mustard. See above!

See? I was trying to get people to take Lenzen seriously, and read his paper, and jog their worldviews with careful reanalysis. But it seems that E = MC2 really does lead to some very big numbers after all, doesn’t it!? ;-)


@timothy: Think about it a little bit harder and write it down in long hand slowly and carefully.


@evcricket: I remember a story about somebody (I think Fred Hoyle) proving that space travel was impossible because a rocket could never carry enough fuel to power it out of the earth’s gravitational pull. Somewhere there was a wrong assumption … and Lenzen’s result is like that … plainly contradicted by reality. So until I get time to examine it in detail, I’ll provisionally accept the kinds of numbers that make sense … also prepared by qualified people.


Yeah, Lenzen must be wrong because he got bad numbers for nuclear. But Weissbach must be right, despite using decades old numbers for wind and solar, despite that it’s physically impossible that buffered and unbuffered EROEI is the same, despite that his numbers are an outlier, he must be right because he got nice numbers for nuclear. That’s all that matters for nuclear religious adherents.


The critical threshold or EROI for society holds only if the society is homogenous. Any process yielding EROI >1 can sustain our society which requires EROI = 7 if we segment the society. Let there be countries A0 who are primary energy producers who run the process with EROI = E0 + E1 where E0 is the energy used to provide living standard in countries A0. The component E1 will be forcefully or traded away to the countries of level A1. From countries A1 perspective the country A0 behaves as energy source. As long as the amount of energy needed to controll the country A0 is reasonably low compared to E1, the country A1 has the energy source of EROI >7 and can prosper.

This is the way societies have worked in the past — slavery and feudalism — and in the present — capitalism with unbalanced living standards, trade inequalities and human right violations in the developing world. So there is no reason why a present society cannot survive with green energy sources with EROI < 7. Rather the question can we have social equality and green energy sources or we will have green slavery.


Leave a Reply (Markdown is enabled)

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s