The Fossil Fuel Energy Cliff and the Great Oversimplification

Recent claims that renewables represent an inevitable energy descent are derived from a new form of 'energy blindness' that ignores the best EROI systems science.

  • Nafeez M Ahmed
19 min read
Nafeez M Ahmed

Are we moving into a low energy future? Is this what ‘the science’ really says beyond doubt? Is it even the right question? This debate was thrust into the spotlight recently when Dr Nate Hagens who runs The Great Simplification podcast released a video on “EROI” – the eponymous concept of Energy Return On Investment (EROI) – criticising the idea that renewable energy is now capable of generating greater energy returns than fossil fuels.

In his video, Hagens provided a fascinating and succinct overview of the science behind EROI, a ratio pioneered by State University of New York systems ecologist Charles Hall – whom I worked with on my last book – which measures how much energy is used to get energy out from any particular resource or technology.

Hagens closed his video with the assertion that scientists who believe renewables now have a higher EROI than fossil fuels are not only incorrect, but “ideological” and pursuing research which is “politically-motivated”. He then criticised the field of EROI as a discipline of study, suggesting that it has become so complicated and fraught with disagreements that the concept in general is no longer useful.

These claims don’t fit together very well. Hagens at once put forward his own specific views of the ‘right’ way to do EROI calculations – this he equated with ‘good science’. He then attempted to ‘debunk’ a recent scientific paper led by long-time EROI expert David Murphy which found that the EROI of solar, wind and batteries was higher than oil, before dismissing EROI in general as no longer a useful scientific concept.

What Hagens really appeared to be saying is that his interpretation of EROI in which renewables represent a decrease in net energy compared to fossil fuels is the correct and scientific one; with all others representing unscientific, ideological and politically-motivated theories which are not useful. Of course, anyone can claim that research they disagree with is based on politically-motivated ideology. That’s not a helpful way of conducting a scientific discussion.

The question, then, is whether his interpretation really is the correct and scientific one. Hagens’ video criticism was quickly followed by a blog post by US oil industry veteran Art Berman, who elaborated on Hagens’ criticisms, arguing that Murphy et. al’s claims went against decades of established EROI research.

So what’s the answer?


Hagens video is based almost entirely on his own PhD thesis from about 15 years ago, and ignores advancements in the field of which he seems largely unfamiliar.

As far as I can tell, following his PhD Hagens did not make any major contributions to the field of EROI. In contrast, David Murphy – currently an associate professor in environmental studies at St Lawrence University – has published numerous rigorous and well-received scientific papers on EROI which significantly advanced the field. A quick search via Google Scholar reveals some two dozen scientific papers on EROI authored and co-authored by Murphy over the last decade, many of which directly address the concerns that Hagens describes in his video. Hagens has never criticised this body of work.

The problem as I see it is not so much that Hagens is ‘wrong’. It’s that his approach to EROI, as exemplified in his PhD thesis, is well over a decade old. The technical findings put forward by Murphy et. al – and many other EROI scholars – fundamentally undermine the underlying assumptions behind Hagens’ arguably outdated research, which underpins his central thesis today that human societies are inevitably heading for a ‘great simplification’ defined primarily by a transition to lower net energy.

In the field of EROI, 15 years is a long time to be out of the game. Data is changing, and theoretical frameworks are evolving. And despite claiming that EROI analysis is now worse, not better, he does little to properly substantiate this claim, showing very little knowledge of the new scientific literature on EROI and focusing almost entirely on explaining his own PhD research.


In his video, Hagens explains why there has been a proliferation of EROI values in the scientific literature. He points out that the key challenge is figuring out the pesky boundary assumptions – in other words, where exactly do we measure the energy inputs, and where exactly do we measure the energy outputs.

In the recent paper published in Sustainability, energy scientists David Murphy, Marco Raugei, Michael Carbajeles-Dale and Branda Estrada found that previous EROI studies comparing fossil fuels with renewables have in effect compared apples and oranges. They were looking at two very different energy systems using the same categories, or applying categories inconsistently across different cases. This has resulted in a wide range of EROI values depending on the assumptions you use about where energy inputs and outputs should be measured.

According to Hagens, however, Murphy et. al’s effort to provide a more standardised analysis based on more accurate boundary assumptions is wrong. That’s because, he claims, Murphy et. al are in effect double-counting the energy inputs for oil. As he explains here:

The issue w Murphy paper is 'chaining' and where to include the energy (from oil/gas) that is put back into refining the oil. Its akin to using in situ heat to process tar sands, or burning the bagasse to make sugar cane ethanol.

Hagens says that the energy from oil used to refine oil is part of the overall energy output – which is therefore much higher. Murphy et. al argue in contrast that the energy being used to refine the oil is still part of the energy input invested to make the oil ready to actually be used. EROI, therefore, is really a measure of the useful work available for the economy. Murphy’s approach is essentially to standardise the approach by calculating how much energy is produced at the ‘point of use’ – that is, the point when it is actually producing useful work for society.

The energy used to refine the oil, even if produced in situ, is still lost in the refining process before it becomes usable by society and the economy, and therefore should not be classified as an energy output. Hagens disagreed, but didn’t really explain why we should think the efficiency loss of energy in the refining process should be ignored.

A study published in 2022 by the Paris-based climate technology think-tank, Zenon Research, came to similar conclusions as Murphy et. al.. They argue that EROI should be measured at point of delivery, rather than point of extraction:

".... most EROIs for fossil fuels were directly calculated at the point of extraction. Such practice limits the usefulness of the EROIp,POD [where POD stands for 'point of delivery' - similar to Murphy et. al's 'point of use'] for comparison between energy resources as the net energy delivered to society is the crucial indicator. Comparing different energy vectors at this point of extraction is not relevant as they do not offer the same final service or must go through different processes between extraction and use. Comparisons are valid when all subsequent processes to obtain a usable energy vector are the same. A typical example would be the supply chain of oil: once extracted, the resource must undergo refinery. Required energy investments for refining of oil derived products are considerable and heavily reduce the maximum attainable EROI."

Ignoring technological learning effects

Hagens goes on to argue that it’s wrong to conclude that the EROI of renewables is continuously improving. His killer stat for this is simple – he refers to the fact that the cost of producing wind turbines rose some 30% or so in recent years due to oil price hikes. This he says is because oil is the master resource.

The assumption is that if oil becomes more expensive, then the cost of producing everything in the system increases. Therefore, the EROI of renewables cannot improve but must decline.

Yet Hagens’ analysis is fundamentally flawed. He ignores the fact that the EROI of solar and wind technologies has already been exponentially improving over the last 50 years, as their costs of production have continued to plummet year on year. That’s because they are disruptive technologies. That 99% cost reduction implies that over time the energy needed to produce a solar panel or wind turbine has reduced due to increasing efficiencies related to materials and manufacturing that permit such prolonged cost reductions.

In the same period, we’ve also seen that these technologies are getting better at capturing and generating energy. These two factors demonstrate clearly that their EROI has improved exponentially since 1975.

Source: US Department of Energy

Disruptions scale due to a variety of factors that drive a real technological learning effect in which the efficiency, costs and performance of the technology improves exponentially. The more the technology is deployed, in other words, the more it improves in a feedback loop driven by R&D, investment, and other factors.

It’s worth noting here that this improvement in EROI of solar and wind continued during, through and after several major oil price spikes, including the peak and plateauing of conventional oil production in 2005, which as many studies have shown involved an increasing shift to more expensive forms of unconventional oil resulting in an overall decline in the EROI of oil. Now despite that overall decline, which included major oil price hikes after 2005, the cost of solar and wind continued to drop overall, and its EROI therefore continued to improve.

So Hagens’ suggestion that a recent price surge proves the EROI of renewables is bound to decline and won’t continue improving is unfounded, and nothing to do with systems science. In reality, due to these economic dynamics and the technological learning effect, as renewables scale up out to 2030, their costs will continue to decline. A tripling of capacity would lead to a further 50% drop in costs.

But I take from Hagens a valuable point. Right now we’re still overwhelmingly dependent on fossil fuels, which we need to use to build out the new post-carbon system.

If we don’t transform the energy system fast enough, we could be locked into the declining EROI of the fossil fuel system in such a way that we cannot breakthrough to a new renewable energy system. One recent study which I reported on for Byline Times found that the declining EROI of oil could make the continued functioning of the industry economically unviable in the 2030s. The study authors warned that this would begin to fatally undermine efforts to invest in building out the new system. That’s why accelerating the energy transformation before 2030 is urgently necessary.

Source: Louis Delannoy et. al, Applied Energy, 2021

Oil expert weighs in

Shortly after Hagens released his EROI video, petroleum geologist Art Berman published a post also claiming that Murphy et. al’s paper is fundamentally incorrect. He concluded that if Murphy and his co-authors were right, then decades of EROI research showing extremely high values for fossil fuels would be wrong. He repeats the same argument as Hagens, and then uses it to offer a new calculation:

Nearly 9% of the total post-extraction costs for oil are for refining. Yet most of the energy for refining comes from the crude oil and refined products used in the refinery. It is, in effect, co-generated. That doesn’t negate the energy investment needed to operate the refinery but it is not a cost to society as indicated in the table… I divided their 8.9% for refining investment by 3 to account for the co-generation described above (it is probably much lower). The resulting oil EROI is 18. That completely removes the good news from Ahmed’s and Bardi’s proclamations of ‘mission accomplished’ and restores oil EROI to the consensus range for the last two decades.

The key error in this argument is where Berman says: “That doesn’t negate the energy investment needed to operate the refinery but it is not a cost to society as indicated in the table”.

But that is incorrect. The term ‘cost to society’ pertains precisely to energy invested which is not available for use by society. While the energy used to refine the crude oil is co-generated, it is still an input into the refinement process before the oil becomes available for actual work in society at the ‘final energy’ stage. In other words, the energy is being used to refine the oil and is therefore not available for society in any case.

What Berman and Hagens are effectively trying to do is classify the energy used to refine oil as an ‘energy output’ that represents useful work for society outside of the energy system. But this classification doesn’t make sense when we consider that it represents work that is specifically related to making the energy usable for society in the first place - because the oil must be refined and processed before it can actually be converted into usable energy for society.

Berman further questions that if EROI for fossil fuels was much lower, how could it have been so profitable?

As earth system scientist Ugo Bardi has pointed out, the profitability of an industry depends on numerous factors outside the energy system related to credit, markets, economic policy, investment, currency values and beyond. But in addition to that, the bottom line is that Murphy et. al’s research suggests that if oil has been profitable with EROI much lower than previously believed, then previous assumptions about economic prosperity requiring much higher EROI levels are questionable.

Because of the huge efficiency losses of converting energy from oil into useable forms (between 50 and 70% of energy is lost converting primary energy to final energy), as renewables avoid those losses they can produce about 50% less energy to meet demand. This means that the presumed minimum EROI to sustain a viable civilisation derived from fossil fuels could be much lower in a more efficient system.

As Marco Raugei points out, the shift to renewables and electrification “may open the door to achieving the required services with much lower demand for primary energy, which in turn entails that a significantly lower EROI than previously assumed may suffice”.

Berman also argues that Murphy et. al’s paper suffers from a lack of transparency as it is based on the ecoinvent database, which is behind an expensive paywall. While it’s always regrettable to have data behind a paywall, that doesn’t mean it isn’t transparent. Sadly, most quality scientific work is published behind paywalls. In contrast with Berman’s claim, the ecoinvent database is widely recognised as one of the world’s most reliable resource for life cycle analysis data, and is widely cited in the scientific literature. Murphy et. al’s use of this database is not a signal of its weakness, but its strength.

Measuring consistently

Murphy et. al’s research is also consistent with the work of other scientific teams. For instance, another leading EROI expert Paul Brockway similarly found that when applying more standardised and consistent boundary assumptions, the EROI of fossil fuels has clearly fallen so significantly that it is probably below the EROI of renewables, which is – unlike oil, gas and coal - improving over time. Brockway et. al found that the EROI of fossil fuels was nowhere near the optimistic 25:1 assumed by many previous studies measuring EROI from the well-head.

… such ratios are measured at the primary energy stage, but should be estimated instead at the final energy stage (e.g. electricity, petrol) where energy enters the economy. Here, we calculate global time-series (1995-2011) energy-return-on-investment ratios for fossil fuels at both primary and final energy stages. We concur with common primary-stage estimates but find very low ratios at the final stage: around 6:1, and declining. This implies fossil fuel energy-return-on-investment ratios may be much nearer to those of renewables and could decline precipitously in the near future.

In short, the main problem with Hagens’ and Berman’s approach is that they want to measure the energy output of fossil fuels at the ‘well mouth’ or point of extraction - the primary energy stage. Yet when comparing them to renewables they are still measuring the latter's energy output at the final energy stage or point of delivery/use. By measuring fossil fuels differently, they grant it an artificial advantage that systematically ignores how much energy is lost in the process of making fossil fuels useable through processes like refining.

As Adam Brandt and other EROI experts have shown, the approach recommended by Hagens and Berman is a different metric within the wider family of Energy Return Ratios (ERR) which is related more to energy extraction.  

When measured more consistently at the point of delivery, recent EROI studies consistently show across the board that the EROI of solar and wind is not only improving but already surpassing fossil fuels. The study led by Kevin Pahud of the Swiss Federal Institute of Technology for Zenon Research points out that:

Between 1990 and 2015, wind turbine EROI increased from 12 to 23, an increase of 92%.

In the same period, EROI for solar PV technologies increased from approximately 1 to around 9 between 1990 and 2015, an increase of 800%.

Based on the historic pattern of decreasing energy investment costs and increasing energy outputs, these will continuously increase through to 2040 to an average of approximately 15 and 28 for solar and wind respectively.

Pahud et. al acknowledge that along the way, a system-wide EROI would likely experience a decline due to the need to continually invest in resource allocation to build out the system. Societal EROI might during this period be lower than these quite high unit-level EROI values, but as he then further points out, that is unlikely to pose an obstacle given that minimum EROI for civilisation figures are probably overestimated as they are derived from fossil fuel EROI figures based on overinflated point of extraction estimates. Their conclusion is:

Looking at inefficiencies of the current system, most notably due to Carnot thermal limits, there is room for consequent improvements. Solving these efficiency issues would drastically reduce final energy demand. The EROI indicator might thus be biased as an assessment of minimum required EROI, and negatively influence renewable potential... Correcting for inconsistencies of this domain, the premise of uncompetitive renewables against fossil fuels offering less energy to societies, from a net-energy perspective, seems unfounded.


Berman closes his post by pointing out:

Murphy et al do not address the effect of intermittency on the net energy supply that society needs to function. They tell us that technology will find a way and perhaps it will but that does not change the present upon which their study is based.

There are now a large number of credible studies which show how the intermittency of renewable energy sources can be addressed by optimising the build-out in different local and regional contexts.

One approach that has been identified in multiple studies which I first learned about at RethinkX is supersizing the generating capacity to at least three times existing demand. This approach reduces storage requirements by as much as 90%, which means that the most expensive component of the system – batteries – can be minimised. This reduces the costs of the build-out significantly, by as much as five to six times cheaper.

I’ve referred to some of these studies before. Finnish energy firm Wartsila found that overbuilding solar and wind by four times peak load requires no seasonal storage, and needs only 4-10 days of multi-day storage capacity, making it the least cost system. Marc Perez at Columbia University found that building out overcapacity of solar and wind by a factor of three times bigger than demand not only dramatically reduces the need for battery storage, but lowers the cost of electricity by as much as 75%, while eliminating intermittency challenges. Perez’s team did another study focusing on the state of Minnesota which showed that this approach would reduce the battery input for seasonal storage by as much as 90%. RethinkX’s modelling of California, Texas and New England – covering regions that represent the climate of most populated regions in the world – also demonstrated this dramatic reduction in battery storage by supersizing solar and wind generation by 3-5 times demand, depending on the region. Such a system requires between 36 and 90 hours of storage a year, which eliminates demand gaps.

Storage can be further ameliorated through many other methods. Vehicle-to-grid technology can reduce stationary battery storage requirements significantly. One study found that if Australia’s entire car fleet was replaced with electric vehicles, their batteries would supply more than double the country’s stationary storage requirements. A combination of smart cross-sector integration of energy systems can also greatly reduce the need for battery storage. For instance, Europe can meet significant demand for home heating from waste heat via heat pumps integrated within a smart energy system that can derive heat from various sources such as wastewater treatment, waste incineration, and so on. Upscaling other forms of storage such as pumped hydro storage, hydrogen produced via electrolysis, compressed air storage, and so on, will also provide pathways to reduce battery inputs. Finally, trans-border global electricity grid interconnection alone can reduce storage requirements by 50%. All these provide a range of pathways to eliminate intermittency.


Both Hagens and Berman conclude their EROI explorations by accusing their critics of being wedded to ideology, or being too stupid to understand EROI. But one could reasonably conclude that they both appear to be ideologically committed to the vision of an inevitable ‘great simplification’ of society which hinges on the idea that we’re moving into a lower net energy system.

As I’ve shown here, there is a great deal of robust scientific research which suggests their claims are oversimplistic at best. Because if the EROI calculation methods of Hagens and Berman are not reliable, then the ‘great simplification’ as the only possible scenario to which civilisation is now heading also becomes questionable. Yet neither Hagens nor Berman appear to be willing to question the vision of the future to which they appear committed.

Berman concludes: “I favor a future society that is based largely on renewable energy. That society will look very different that [sic] what we know today. Substituting renewables for fossil fuels is not a solution without greatly curtailing our total energy consumption. That’s what the physics indicates will happen in a renewable future. I suggest that we stop trying to make renewables look like something that are not and cannot be, and just learn to live with them as they are.”

I completely agree with Berman that future societies based on renewables will indeed look very different. And given that renewables produce electricity without up to 70% waste heat losses associated with converting primary energy from fossil fuels into useful energy, they will certainly entail a dramatic reduction in total primary energy consumption. That's why I argue that those who believe we need a dramatic downsizing of our energy consumption should be cheering on the renewable energy transformation, because that is precisely what it entails.

Yet that need not result in final energy scarcity, given that the system can be designed to produce considerable excess electricity virtually for free.

This is a phase-transition representing the emergence of a completely new energy system with novel properties and dynamics totally unlike incumbent fossil fuel energy systems. The emerging energy system will, therefore, have a number of really interesting features that are quite different to the incumbent system, and which cannot easily be captured in conventional EROI calculations designed to understand fossil fuels:

  1. Once built, a solar, wind and battery system will produce energy continuously at zero marginal costs. Unlike a fossil fuel plant, once built it will not, for the most part, require continued material or energy inputs to maintain the energy output. In effect, after the start of operations, renewables will generate energy continuously without limits virtually for free.
  2. The time period over which a renewable energy unit will generate zero marginal cost energy will be several decades. In fact, it is likely to extend well over 30 years. A five year study by the US Department of Energy’s Sandia Labs of 13 out of 23 tested solar PV module types found that they have effective lifetimes exceeding 30 years because they will still operate at 80% efficiency after those three decades.
  3. Robust modelling shows that the optimal design of renewable energy systems will be capable of generating as much as three to five times more energy than existing demand. As I’ve shown elsewhere, supersizing generating capacity in this way need pose no problems with respect to materials.
  4. This excess electricity will be crucial because we will need it to help electrify industries – including mining, manufacturing and recycling for to build new renewable energy technologies. We can also use that excess to generate clean e-fuels (viable pilot projects have now proven how this can work; and recent innovations on track to become commercially viable around 2026).
  5. The necessity of a viable circular economy in which raw materials and minerals are stringently recycled in this system is clear because despite the longer lifetimes of these new energy systems, they will still ultimately need to be replaced. A significant portion of the excess energy generated by solar, wind and battery systems will therefore need to be devoted to both the circular economy and the manufacturing of new energy units. In an optimally designed system, this looks entirely feasible.
  6. The entire energy system will move from one based on centralised control to distributed generating capacity. While incumbent forces of centralisation will attempt to slow this down, the new system will work better as a decentralised network. Society will move from a division between energy producers and consumers, to energy consumers also becoming producers; and indeed, the largest energy consumers will be incentivised to reduce their costs by investing in becoming energy producers. This will not only drive the transformation of electricity grids, but also drive a wider transformation of the economy as the traditional divisions between labour and capital, producers and consumers, become blurred.

So there are good reasons to conclude that the energy produced in a global renewable system, if we build it out fast enough, will allow us to avoid a net energy cliff from the declining EROI of fossil fuels, and will be enough to sustain a thriving high-technology civilisation within planetary boundaries. But this is a big if - and, I should add, if we fail, that certainly doesn't rule out a 'great simplification' scenario.

Either way, this will not be the same system that exists today. It will have very different dynamics, and at every step of the way societies will need to make the right decisions to maximise and distribute its benefits. It will immediately lead to a reduction in material flows, and it will require those flows to become circular and regenerative in a way that has never been achieved before. The system will need to be managed in a way that is participatory, distributed and networked, with energy flowing and being shared across localities, regions and borders. It will need to be designed to respect planetary boundaries, requiring earth-centric values that recognise the embeddedness of the entire economy and energy system in the natural world.

In some ways, this debate serves as a distraction. The clean energy disruption is underway. The fossil fuel system is in decline. We need to figure out how to best optimise the emerging system, and how best to organise our societies around it. And given the speed with which the fossil fuel system is moving into economic obsolescence as dangerous climate change intensifies, we need to accelerate the deployment of the new energy system as fast as possible to mitigate the damage while avoiding the risk of breakdown and collapse.

But this need not be such a polarised debate. We can’t sit back and let technology ‘do all the work’ to save us. Tools are only as good as their makers. The way in which we deployed industrial-era technologies in such a way as to destroy our very planetary life-support systems shows that as we deploy these new tools, we need new and better ways of not only making tools, but deploying them in the world. That requires rethinking our values, our commitments, and our understanding of the world in a way that transcends the conventional obsessions of today's homo-economicus. To the extent that these debates get us to that point, they are valuable.

If you appreciated this piece, you can keep this free newsletter alive and thriving by joining our community as a Supporter for the price of a cup of coffee a month.

This article was edited on 6 June 2023 to add a note on why the 50-year cost reduction of solar and wind implies a reduction in the energy inputs per unit; and to make clear that the 'great simplification' remains a possible scenario.

Member discussion