The economic view of systemic efficiency: energy return on energy investment

The last post looked at what I’ve called the engineering view of systemic efficiency, specifically the concept of available energy, or exergy. I refer to this as systemic because it considers energy conversion processes in relation to their specific operating contexts, in order to understand the useful work that a system can provide. While energy conversion processes serve an infinite array of human purposes, in the proximate or most immediate sense, we carry out energy conversions in order to do work—to effect transformations in our material worlds—and to provide heating (and while technically it’s not necessary to further differentiate it here, to provide illumination also). The systemic view provided by exergy analysis deals directly with the question of how much utility we can derive from an energy conversion process, and so it allows us to think about energy resources and infrastructure in a more concrete way than when we conduct analysis in terms of the nominal heating value of primary sources or fuels, in isolation from the particular situations in which they are used. Differences in energy use situations—different conversion technologies, implemented in different ways, operating in different physical environments—lead to differences in the utility that can be derived from an energy source. In establishing the efficiency of an energy conversion process—the useful energy output from the process divided by the nominal energy input—a focus on conversion systems and their parts (including the particular energy sources involved) only gets us so far. For a comprehensive view of efficiency we need to consider energy conversion processes in terms of all three levels of the basic systems hierarchy of system, sub-systems and supra-system. Exergy analysis provides the means for achieving this.

My reason for identifying this approach to thinking about efficiency as the engineering view relates in part to the scale at which exergy analysis’s systemic approach is most fruitfully applied—namely the plant or equipment scale. In other words, this is most immediately useful at the micro-economic or enterprise level, where we deal with technology components that make up economic units. In macro-economic terms, exergy analysis does have particular value for understanding performance of an economy’s energy sector, and also provides especially valuable insights in relation to transport and manufacturing activities.  Coming to terms, though, with industrial societies—or, as we’ll see, any forms of social organisation for that matter—in physical- or energy-economic terms requires that we look beyond the enterprise and even sectorial levels. That is, we need a basis for thinking holistically about societies and their economic forms that relates energy supply and use at the overall macro-scale. It’s for this purpose that the concept of energy return on investment (EROI) (or energy return on energy investment—EROEI), has started, only relatively recently, to be better appreciated as so important. EROI tells us about the energy available for economic activity other than the supply of energy itself, and it is in this sense that I referred to it in the introductory post on efficiency as, roughly speaking, the economic equivalent of thermodynamic availability.

Charles A.S. Hall first introduced the energy return on investment concept in the early 1970s in the field of ecology, where he applied it to the study of fish migration. It is grounded in the observation that the success of every way of life, at the scale of both individual organisms and their collectives, depends on the relationship between energy supply, and maintenance requirements in energy terms. For a way of life to be successful, it must satisfy the necessary condition of providing an energy yield that exceeds the energy expended on obtaining that yield, by an amount at least sufficient to maintain established social institutions and infrastructure. Ways of life that provide a net energy surplus beyond this maintenance level allow for increase in a population’s potential work output. This can manifest as population expansion; but equally, it can allow for a stable population to enjoy increased low-energy-intensity leisure or cultural pursuits. Ways of life where a ‘maintenance-level’ EROI results from decisions to enjoy greater leisure can be resilient in the face of changing environmental conditions, as the option to work harder to obtain the required or desired energy yield remains open. But where such an EROI is achieved only at the limits of work output and present ingenuity, the associated ways of life are vulnerable to environmental change. If EROI is less than maintenance-level and there are no immediate opportunities to intensify energy harvesting work, an organism or society must consume its capital to survive. For an animal faced with an unusually harsh winter this “capital” is its own fat stores. For a human society, this is physical capital that can no longer be maintained, but that provides a ready source of materials for production processes. In each case the entity—organism or society—is literally consuming itself. Such conditions can’t continue indefinitely—at some point, if complete collapse is to be avoided, EROI must return to at least a maintenance level.

In the simplest terms, EROI provides an index for comparing the net energy provided in different energy supply situations. It compares the quantity of energy provided by a supply process at a specific point in the supply chain, with the energy required by the supply process itself. This is presented as a dimensionless ratio. The point in the supply chain on which the calculation is based varies depending on specific interest. In relation to EROI of fuels, Murphy et al propose a set of general boundaries with corresponding definitions for this.[1]  In some instances, EROI is calculated at the input to a processing system, following extraction of an energy resource (such as crude oil); in others, it may be the input to a post-refining distribution system; it may be the input to non-energy economic sectors; or it may be the point at which the fuel is finally used. At each of these stages, the energy return from a given initial resource input is reduced—each entails its own energy costs that decrease the EROI. As a general rule, to be considered an energy source, an energy storage or transfer medium must provide a positive EROI, i.e. greater than 1:1. Anything less than this can be considered an energy sink—at small scale, there are sometimes legitimate reasons for investing more energy in the provision of a source at point-of-use than the source makes available in terms of useful work or heating, but at the scale of overall economies, such situations if sustained for prolonged periods would entail an overall trajectory of contraction. This is a theme we’ll consider in more detail as the inquiry proceeds.

An EROI greater than 1:1 is not, however, sufficient to ensure an energy source’s economic viability. As Charles Hall points out, for a petroleum source with EROI just over 1:1, “one could pump the oil out of the ground and look at it…and that’s it.”[2] Producing and distributing refined fuels requires a sufficient energy surplus at the wellhead; actually using the fuel—for instance, manufacturing, distributing and maintaining motor vehicles and all of their attendant infrastructure—requires a further surplus; and for ways of life that involve more than just producing and driving around in cars—that is, for worlds of the nature that we actually inhabit and from which we derive the meaning that makes our lives worthwhile—we require a further surplus again. The cultural, institutional, infrastructural and material bases for modern lifestyles—at least in those part of the world where most people are not living under conditions of severe poverty—are enabled by this surplus, and in particular the energy surplus associated with petroleum as our principal transport energy source. Earlier in the petroleum age, the wellhead EROI for US oil production is thought to have peaked at around 100:1. Now it is somewhere in the order of 20:1 for conventional sources.[3] And there are many sources lower than this, EROI for conventional oil typically varying between 10:1 and 20:1 [1], reflecting a host of local determinants related to the oil’s geophysical and political accessibility. For the purpose of providing some social context for these abstract figures, Hall, Balogh and Murphy have preliminarily estimated the minimum EROI required to maintain current rich-world industrial societies in what might be called “survival mode” as 3:1 (calculated at the “mine-mouth” or wellhead for petroleum-based fuels, or farm gate for biofuels).[4] This figure would allow only for the bare minimum economic activity associated with current transport expectations—and would leave very little surplus for non-energy-supply activity, including education, health care and cultural pursuits. They speculate that an EROI of something like 5:1 would likely be necessary to support levels of non-energy-supply-related activity that we in the rich world associate today with functional societies.

Societal and economic implications

Another way of appreciating the significance of EROI for a society’s economic prospects is to consider how it affects the proportion of overall economic activity that must be dedicated to energy supply. Thinking about this requires some care: there are important subtleties that, if overlooked, result in misleading conclusions. The key here is to appreciate that EROI acts as a rough proxy indicator for the scale of an economy’s energy sector relative to the rest of the economic activity enabled by the energy sector. Take for instance a situation in which a key energy source declines in EROI over time from 100:1 to 20:1, as crude oil did in the USA over the course of the 20th century. It may be tempting to consider the decline from 100:1 down to 20:1 as implying only that end-use energy made available by a unit of the primary source declines from 99% to 95%. Instead, we need to think in terms of the 400% increase in energy use associated with economic activity to supply energy from that primary source, and the commensurate increase in infrastructure and institutional scale that this implies. To maintain a given energy surplus from the source, the industry supplying it must increase in scale several times. While technology development and associated efficiency gains will likely keep the proportional increase in the industry’s infrastructure below that of its energy use, energy infrastructure is by its nature disproportionately capital intensive, compared with the overall economic activity that it supports. For instance, in Australia in 2010-11, the energy sector accounted for about 6% of GDP. Over the same period though, over 10% of gross fixed capital formation was expended in the sector, while it employed only 1% of the national workforce.

Approaching this from a slightly different direction, consider the situation where a given society wants to maintain its current activity levels—I’ll assume total work output and heating as a proxy for this—but with energy sources of lower EROI than those with which such activity levels were established as the norm. To do so, two options are available: either i) growth in the energy supply sector must lead growth in the size of the overall economy, with other sectors (at most) growing less rapidly than the energy sector; or ii) the energy supply sector must increase in size at the expense of other economic sectors. The first option is the one that we’ve tended to favour to date. This has been possible because the energy surpluses made available by fossil fuels have exceeded the minimum necessary to support complex modern economies. The difference, as EROI declines, has been compensated in part by increasing end-use energy efficiency. Excess surpluses allow low efficiency energy use to proliferate. In a sense, we invest some of the surplus in technologies, infrastructures, behaviours and so forth that later come to be seen as wasteful; as we’ll see further on in the inquiry, this can result from trading off complexity and efficiency. Higher efficiency is possible, but it can carry costs in terms of institutional and technological complexity. Sometimes, if surpluses allow, we opt for easier—less complex—options in designing problem responses, at greater energy cost. This is possible if energy supply itself presents relatively simple challenges.

But as EROI and the associated energy surplus for principal primary sources continues to decline and as the low-hanging fruit of efficiency improvement is harvested, societies start to converge on the limits of option i). At this point, option ii) may come into play on a wider scale. Here, a society may no longer be able to maintain non-energy related activities of the diversity and scale to which its members are accustomed. Energy supply activity then displaces other “non-productive” activity. At least in principle, it’s possible to imagine a situation where the overall economy is increasing in size due to growth in the energy sector, while other sectors are contracting. It seems more likely, though, that option ii) would involve the expenditure of increasing effort just to maintain a given overall level of activity, with greater and greater resources directed towards energy supply, followed sometime later by the onset of overall contraction. It’s worth noting, though, that by the time net contraction, as measured by macro-scale indicators, is recognised as having set in, a good deal of a society’s cultural capital may already have been significantly compromised. This may be an inevitable consequence of promoting and prioritising growth in crudely defined economic activity, while failing to inquire into what it is that a society’s members actually value.

Interpreting EROI findings

As with the related field of life cycle assessment, EROI analysis is subject to significantly divergent perspectives. This is a consequence of its systems foundations and orientation. The EROI calculated for an energy source is always a system-relative abstraction, dependent on methodological and boundary assumptions (for instance, determination of what should be included as relevant energy inputs and outputs). These issues are well recognised by practitioners within the field—as discussed in several articles for a 2011 special issue of the journal Sustainability, “New studies in EROI”, edited by Doug Hansen and Charles Hall. The article “Order from Chaos: A Preliminary Protocol for Determining the EROI of Fuels” by Murphy, Hall, Dale and Cleveland is particularly noteworthy in this respect.[1] I recommend this, and the special issue more generally (all open access), for anyone wanting to explore EROI analysis in depth; the pioneers and leading exponents of the methodology are all represented amongst the authors. I’ll just focus here on some general considerations that may be helpful in interpreting the findings of EROI analyses for purposes such as thinking about economic transitions and associated strategy and policy.

Methodological issues

It’s perhaps a rather obvious point, but in order to carry out EROI calculations, the analyst needs access to appropriate data on the energy use associated with key production, refining and distribution inputs. This is not such a straightforward matter though. While it’s relatively easy to account for economic inputs in terms of mass—e.g. tonnes of steel, concrete etc—it’s another matter determining the embodied energy associated with the inputs. While such data is now widely available, it is also subject to variation across different production and supply contexts, when the full supply chain is taken into account—as is required for a thorough EROI analysis, if it’s to provide useful data. There’s no doubt that this is an expanding field for which data is improving with time—but it is still the preserve of highly-qualified specialists using somewhat arcane methods that are not immediately accessible or transparent to others with an interest in the findings.

The methodological challenges are well illustrated by the fact that EROI analysis—as with life cycle assessment—employs two distinctly different approaches to addressing its questions: i) “bottom up” process analysis; and ii) “top down” economic input-output analysis:

i)                In process analysis, the production of an input good or service is decomposed into component parts, and the energy associated with the parts is calculated and aggregated along the supply chain. The approach is dependent on the availability of appropriate disaggregated energy data. There are significant costs associated with sourcing data that’s sufficiently comprehensive and accurate to represent the broad range of activities and materials that an analyst might want to take into account, and so these place constraints on the analysis. And because the approach is bottom-up, it can be difficult to understand the significance of omissions: the approach only tells us about what has been included, and by its nature doesn’t offer direct insight into the potential scale of its exclusions.

ii)              In economic input-output analysis, energy intensities are attributed to different sectors of an economy. These are then used to estimate the energy use associated with specific goods and services provided by each sector. This approach relies on scaling in monetary terms: if we know the monetary value of a good or service, then we can arrive at an energy value for it if we know the overall energy intensity per monetary unit for activity associated with the economic sector in question. The accuracy of this approach depends to a significant degree on the extent of variation in energy intensity for each sector. As a general rule, the smaller the number of sectors into which an overall economy is divided, the larger the variation in the energy intensity of its products is likely to be, and so using the sector average for all products may lead to significant errors. The potential for this increases where an EROI analysis relies heavily on a small sub-set of products from a given sector. By decomposing an economy into smaller sectors (and sub-sectors), it is possible to improve the accuracy—but this then comes at the cost of more complicated, and hence more costly, analysis.

Neither the process analysis nor economic input-output approach can lay claim to providing the most reliable or accurate findings under all circumstances—and in fact some sort of hybrid of the two is likely to be required in many cases. As such, there is no universal basis for coming to agreement on whether the methodology adopted in any particular situation is the “right” or even the best one—there is always scope for this to be contested.

A further methodological issue relates to a key Beyond this Brief Anomaly theme: the very abstract nature of energy analysis. When we describe an economic activity in energy terms—attributing to it a value in joules—we are reducing it to a “shadow of its former self”. Working in energy values, rather than quantities of specific energy sources, we lose the ability to discriminate between the different energy sources in terms not only of their consequences and impacts, but also their utility value. Beyond the most superficial level of analysis, a given end-use energy source will require inputs from a range of different sources. So for instance, the provision of automotive gasoline will likely require inputs including automotive gasoline itself, but more significantly, diesel fuel, electricity (itself likely derived from multiple sources) and perhaps natural gas. The gasoline product cannot directly substitute for energy inputs from these other sources, and so it would be misleading to think that a sufficiently high net energy return on investment, on its own, can ensure the viability of a given source. Complex industrial economies rely on a range of different energy sources. While there are some limited circumstances where a fuel with a net negative EROI has important societal value, and so it makes sense to invest more input energy from one source than we get out in another, overall economic viability requires that the viability of each of our principal end-use energy sources is also individually maintained. How, then, do we determine whether an output of a certain number of joules from gasoline justifies, say, an input of a certain number of joules of electricity? To answer this question, we need to make some sort of value or quality correction for energy from different sources. Here is what Murphy et al [1, p. 1896] have to say about this:

One major criticism mounted against EROI research has been that it ignores many of these factors that determine the quality of an energy source. Converting all energy inputs to common energy units using only heat equivalents assumes implicitly that a joule of oil is of the same quality as a joule of coal or a joule of electricity. Since this is clearly not the case, we should account for differences of energy quality within EROI analysis when this is possible.

Actually addressing this is far from a straightforward matter though. Murphy et al propose two approaches for this: i) price-based adjustment and ii) exergy-based adjustment:

i)                Price-based adjustment simply reflects that on a per-unit of energy basis, different energy sources are priced differently. For instance, in Melbourne, Australia where I live, at the time of writing retail customers pay around 7 cents/MJ for electricity, 4 cents/MJ for automotive gasoline and 2 cents/MJ for natural gas. On the basis of this, we can infer that electricity has an economic value a little less than double that of gasoline, which in turn has an economic value about double that of natural gas. These differences reflect real differences in utility, but also suffer from any market distortions and unaccounted-for externalities. Where prices are subject to significant volatility, the adequacy of price-based adjustment becomes especially questionable.

ii)              Exergy-based adjustment involves characterising energy sources in terms of exergy values, rather than nominal heating values. This assumes as a matter of course that it is meaningful to attribute exergy values to fuels outside the particular context in which they are used. Specifically, it involves treating exergy as an intrinisic characteristic of objects or substances, rather than a system-relative conceptual construct. As I discussed in the last post, while this approach to characterising energy sources has gained a significant following, its legitimacy is less well founded.

Beyond the specific merits or otherwise of each approach, that these two ways of correcting for energy quality have such different foundations underscores both the uncertainties, and ultimately, indeterminacies that are part-and-parcel of all attempts to characterise human phenomena in energetic terms. Quality weightings, at the end of the day, call for qualitative methods—they’re dependent on what we humans value and how we value it, and so are not simply an empirical matter. I’ll return to this again as the inquiry unfolds.

Boundary issues

The result of an EROI calculation is entirely dependent on the supply system boundary. Change the boundary, and EROI changes with it. This applies both to inputs and outputs. As discussed earlier, the energy output in which we are interested might be that associated with crude oil at the wellhead, coal at the mine mouth or biomass at the farm gate. On the other hand, we may be interested in the EROI for processed fuels at the refinery fence, or at the point of final consumption, following distribution. Processing and distribution carry their own additional energy costs, and so EROI will decline with the point along the supply chain at which the output is designated. On the other hand, the calculated energy input varies with, metaphorically speaking, how widely we cast the net adjacent to the supply chain—and generally speaking, the broader the boundary, the greater the energy input. This relationship between input and output boundaries is depicted in matrix form in Murphy et al’s article available here (see Table 1 on page 1895).[1] So, for instance, should we include such inputs as the pro-rated proportions of energy used by the financial services and insurance industries in enabling energy supply activities? Or should we only include “direct” energy inputs—those readily attributed to the principal activity of the energy supply sector? The idea of an “energy sector” is, of course, an abstraction for thinking about such questions—it is a sense-making conceptual construct. There is no energy sector outside the context of the encompassing network of economic interactions. The search for “objective truth” in EROI calculations inevitably leads to more comprehensive boundary judgements. But should “objective truth” be the governing criterion for validity here? Might we better served by seeking inter-subjective agreement, based on clearly defined system boundaries? This all depends on the questions with which we start.

In some situations we’re interested in comparing EROI for similar end-use energy services provided by different energy sources, e.g. petroleum fuels and biofuels; in others, we’re interested in comparing EROI for different supply chains for a similar energy source e.g. corn-based ethanol in different geographic regions. In the first situation, more comprehensive analysis is likely to yield better results; in the second, it’s more important that the boundaries are the same for each option under comparison. With both of these questions though, the first step towards a meaningful conversation is to ensure that boundaries for the calculated EROI are clearly specified. In a sense, the accuracy of an EROI calculation can only be established relative to the specified boundaries—this is the critical criterion for validity, rather than “objective truth”. If the boundaries are made clear, then if agreement exists that the calculations based on those boundaries are accurate, the EROI can be considered valid—even if there is still disagreement about whether the analysis is sufficiently comprehensive.

Even so, if EROI analysis is to act as a practical guide in thinking about economy-wide energy source transitions, we do need to be confident that all significant and essential energy inputs are accounted for. Prieto and Hall’s analysis of solar photovoltaic (PV) electricity generation in Spain during the period 2009-2011 provides a valuable case study in this respect.[5] In order to establish the potential for solar PV generation to act as a net energy provider, they have attempted to make the most comprehensive analysis possible both for energy inputs, and for losses associated with outputs—in other words, to establish the “real” energy inputs and outputs (after excluding losses) in practice, at commercial scale. Their methodology and results are available here. This makes for sobering reading: performing a conventional analysis—i.e. one with a relatively narrow boundary, but where accuracy of the inputs is relatively uncontroversial—they arrive at an EROI of 8.3:1; but adding in additional energy costs (and excluding those related to finance and labour) that must clearly be accounted for somewhere, this reduced to 2.7:1. Assuming significant technical improvements brings this back up to 3.5:1; but then making an allowance for finance and labour-related energy inputs previously excluded, they arrive at an end result that is likely to be less than 2:1. The authors acknowledge that recent analysis by Raugei, Fullana-i-Palmer & Fthenakis [6] finds that EROI is higher for newer PV cell technologies, but note that on the whole, this has limited scope to affect the overall result given the importance of balance-of-system energy inputs beyond the generating equipment and its immediate infrastructure.  It’s worth reflecting for a moment on the relationship between Prieto and Hall’s finding, and the figures presented earlier, also from Hall, on the minimum EROI required for a sustainable society. On the basis of their results, the authors conclude that solar PV electricity generation is effectively underpinned by fossil fuels—it is an extender of a fossil fueled industrial economic system, rather than an alternative to this. They note that, so far, there are no solar PV “breeding systems” i.e. systems that use the energy output from solar PV generation to grow solar PV generating capacity.

Given the far-reaching implications, these findings will no doubt be subjected to close critical scrutiny, particularly by those with an interest in the solar PV industry. The key message that I’m hoping is clear in the discussion here is that in making debate in this area coherent, the principal strategy for addressing boundary and methodological issues should involve—following Murphy et al[1]—ensuring clarity about the boundaries and methodologies for any given study.

Comparative EROI for fuels and electricity

I’ll refrain here from attempting to assemble my own set of comparative EROI figures for fuels and electricity from different sources. My reasoning for this follows from the discussion of methodological and boundary issues above: sound interpretation of EROI figures requires that we consider them in the context of their associated methodological and boundary assumptions. Raw numbers on their own can create the illusion of greater certainty than is warranted, and so I think that on the whole, we’re probably better served by preserving the connection with the original context. On the other hand, taking such a conservative approach passes up the opportunity to tell a highly compelling story in which all of humanity has a strong vested interest. With this in mind, climate & energy journalist Mason Inman’s short and accessible article from the April 2013 issue of Scientific American, “The True Costs of Fossil Fuels”, offers a very worthwhile starting point—at least for getting a sense of the broad comparative relationships between different energy sources.[7] The key data from the article is presented in the two tables below. The background on the figures—the original sources and how and why Inman selected them—is available in an accompanying online article, “Behind the Numbers on Energy Return on Investment”.[8] Reading the data compiled in the article in conjunction with the online commentary provides a way of mitigating the potential pitfalls associated with viewing numbers such as these in isolation from their broader contexts.

Indicative EROI for liquid fuels from various sources, from Inman [7]
Primary energy source Global production, 2011 (Mboe/day) EROI (refinery gate)
Conventional oil 69 16
Ethanol from sugar cane 0.4 9
Biodiesel from soy 0.1 5.5
Tar sands 1.6 5
Heavy oil from California 0.3 4
Ethanol from corn 1.0 1.4
Indicative EROI for electricity from various sources, from Inman [7]
Primary energy source (conversion technology)
Global production, 2010 (petawatt-hours) EROI (generation facility boundary)
Hydro 3.5 40+
Wind 0.3 20
Coal 8.7 18
Natural gas 4.8 7
Solar (photovoltaic) 0.03 6
Nuclear 2.8 5

Just by way of illustrating the dependence of such figures on methodological and boundary assumptions, in the table below I’ve presented for comparison figures on the EROI for electricity for a range of sources, including sources roughly equivalent to each of those covered in Inman’s Scientific American article, from an article by Weißbacha, Ruprechta, Hukea, Czerskia, Gottlieba & Husseina published in 2013 in the journal Energy.[15]

Indicative EROI for electricity from various sources, from Weißbacha et al [15]
Primary energy source (conversion technology)
EROI (unbuffered) EROI (buffered)
Hydro electric (medium size) 49 35
Wind (Enercon E-66 turbine, 1.5 MW nominal capacity) 16 3.9
Coal 30 30
Natural gas (combined cycle gas turbine) 28 28
Solar (PV−Germany) 3.9 1.6
Nuclear (PWR) 75 75
Biomass (thermal−corn) 3.5 3.5
Concentrated (solar thermal−desert) 19 9

The “unbuffered” figures are closest in basis to Inman’s. Unbuffered means that no allowance has been made for energy storage to compensate for situations where electricity supply is primarily regulated by the availability of the primary source (sunlight, wind, water) rather than consumer demand. If an intermittent source is unbuffered, then in situations where supply exceeds demand, some electricity may not actually serve any useful purpose—and yet is still counted as part of the energy return. Buffering allows some of the supply excess to be stored until demand increases—but this comes at an energy investment cost, in terms of the storage infrastructure. The buffered EROI figures are therefore lower than the unbuffered figures.

This issue aside, the figures for most sources are similar in magnitude, though the differences clearly depict the dependencies on analysis assumptions and methods. Natural gas appears to present a notable discrepancy, but this is possibly accounted for by the much higher overall efficiency of combined cycle gas turbines compared with the thermal cycle assumed for Inman’s figure.

The major anomaly is in relation to nuclear electricity generation. Clearly the Weißbacha et al’s EROI of 75:1 is fundamentally at odds with the figure of 5:1 reported by Inman. Inman’s figure is taken from Lenzen [16], whose article reports on the findings of a study undertaken for the Department of Prime Minister and Cabinet of the Australian Government. My understanding is that at the time it was carried out, the study was considered to be highly credible—and in fact was commissioned by the Australian Prime Minister at the time in order to address the controversy relating to life cycle energy and GHG emissions of nuclear electricity generation. One significant source of discrepancy between the figures may be in relation to the uranium ore grade that is assumed. Weißbacha et al do not state their ore grade basis; in Lenzen’s study, EROI is found to be extremely sensitive to ore grade—with lower ore grades increasing EROI. Lenzen gives close consideration to this matter, and specifically considers the sensitivity of EROI to ore grade. The EROI figure of 5:1 reflects the relatively low grade but abundant uranium ores characteristic of the Australian resource. There are no doubt other significant methodological and boundary differences that together account for the difference between these studies—my purpose here is not to get to the bottom of this, but rather to illustrate the importance of establishing common foundations if EROI figures are to be compared on anything approaching an equal footing. Clearly, for nuclear electricity generation at least, this is unlikely to be addressed adequately by arguing for the absolute truth of one analysis over another. Instead what we see here is that EROI figures are always system-relative artefacts of socially-situated inquiry processes.

Reflecting on the limits of energy analysis

While EROI has started to gain more attention as an essential indicator of economic viability (see especially Charles Hall and Kent Klitgaard’s Energy and the Wealth of Nations: Understanding the Biophysical Economy [9]), it still lives on the periphery of mainstream economic thinking. Even amongst specialists in energy economics, it is rarely, if ever, mentioned—a quick scan of the International Energy Agency’s website, or the Energy in Australia publication series, gives a sense of its neglect in official data: in short, it doesn’t rate a mention. What makes this especially noteworthy is that EROI analysis offers an important correction to what is perhaps the most significant distortion (introduced earlier in the inquiry in posts here and here) that arises when energy associated with distinctly different primary sources is aggregated on the basis of nominal heating values—as is standard practice in the presentation of official energy data by national and intergovernmental agencies. By correcting production data for EROI, we would have a clearer picture of what is actually available for doing the things that we value, and of how that is changing with time.

Given the potential benefits in terms of information transparency if EROI analysis was embraced by mainstream energy economists, it’s worth reflecting for a moment on some of the barriers to this. It’s easy to cast aspersions by suggesting that economists’ neglect of EROI is a matter of ignorance or wilful neglect; it’s another matter entirely to deal with the challenges that implementing such a scheme would present in practice. There are two very important reasons that energy data—as with all key economic data—is presented as it is. Firstly, the collection and presentation of the data must be affordable—there are significant costs associated with this, which someone has to bear. Data providers, the governments that fund them, and ultimately, the tax payers who fund the governments, often tend to be understandably reluctant to incur greater costs in such areas. Secondly, if the data is to serve its purposes effectively, those who make use of it must be able to readily arrive at shared understanding around its meaning. In important respects, we rely on relatively crude measures for economic performance because more sophisticated measures entail greater institutional complexity, which taxes not only budgets, but our social capacity for coming to common agreement about how things should be measured. This ties back to another central Beyond this Brief Anomaly theme: the role of human knowledge systems not as means of codifying absolute truth, but as our basis for coordinating social action as we negotiate the demands of living well together. With this in mind, we could perhaps make the observation that EROI—as with exergy, and in fact the very concept of energy itself—is another one of the myriad constructs that act as very good servants, but that make poor masters. Making effective use of EROI requires a significant flexibility of thinking—including with respect to the expectations that those of us who advocate for its greater use hold of those who have not yet appreciated both its practical utility and its profound implications for understanding the human situation.

And as Hall himself points out, “While EROI by itself is not enough to judge the virtues or vices of particular fuels or energy sources, it is an extremely important component for such assessments.”[2] In other words, EROI is a necessary but not sufficient criterion of macro-scale economic viability for an energy source. Over the course of the inquiry to date, we’ve looked in some detail at a wide range of the other criteria that also demand attention. Amongst these, there is one in particular with a close relationship to EROI, but that has to date received far less attention—despite it’s implications for large-scale transition in energy sources having perhaps even greater significance than EROI. This is the measure that I dubbed in an earlier post power return on energy invested or PROI. Industrial societies are dependent for their day-to-day viability not on the size of their energy reserves, but on the rate at which useful energy associated with those reserves can be made available to end-users. It is current rates of energy supply that underpin modern ways of life—and it’s the future prospects for supply rates that feed our collective expectations of the changes to which these ways of life might be subject. When we invest energy from today’s principal sources in alternative means of meeting our energy needs, with the expectation that the alternatives will eventually supplant the incumbent sources, our hopes rest not simply on the quantity of energy that the alternatives provide over their lifetimes: they rest also on the ability to pay back the initial investment in a timely manner. Moreover, renewable energy converters such as solar PV systems require a much larger proportion of upfront energy investment than fossil energy sources, before any energy is made available to pay this back. This has critically important implications for sourcing sufficient “investment energy” during periods of rapid roll-out and expansion—as would be required for a wholesale transition in our principal primary energy sources.  While dynamic considerations associated with high front-end investment and variable rate of energy return over the supply infrastructure life cycle have received some attention (see for example Prieto and Hall’s concluding observations [5]; Cleveland’s second of “Ten fundamental principles of net energy” in The Encyclopedia of Earth [10]; Moriarty and Honnery in the International Journal of Hydrogen Energy [11]; and Kessides and Wade in the “New Studies on EROI” special issue of Sustainability [12]), there is still much scope for exploring the implications of this in greater depth. I’ll take this up in a later post.

In considering as we did just now the “necessary but not sufficient” nature of EROI in understanding our energy situation, this is also an appropriate point at which to emphasise the broader limits to energy-based analysis in making sense of where humanity is now, how we’ve arrived here, and where we might be headed. Given the prominence that I attribute at Beyond this Brief Anomaly to energy-related factors in coming to terms with such questions, I’d also like to make clear that I don’t regard this as a matter of energy determinism, or even fundamentalism (although, as indicated in response to a reader’s question very early on, I am treating energy-related factors as fundamental to civilisational prospects, in the sense of establishing basic constraints to opportunity at the physical level). Rather, I see this as a matter of energy contextualism: energy-related factors are contextual for all activity; and we can’t avoid them simply because we don’t like what they imply. I’ll close here with a pair of quotes that inform this outlook, from scholars of energy and history far more learned than me. The first is from Jean-Claude Debeir, Jean-Paul Deléage and Daniel Hémery, in their 1991 book In the Servitude of Power: Energy and Civilisation Through the Ages:

“There is no energy determinism in history, no absolute necessity that could be deduced, at this or that moment of the development or regression of human societies, by the historic state of the sources and forms of energy, in short, there is not energy fatalism. Nevertheless, energy is present all along the chain of causes and effects from which human evolution derives; energy determination, as the history of the last decades vividly demonstrates, is an undeniable constraint.”[13, p. 9]

The second, from Vaclav Smil in his book first published in 1994, Energy in World History:

“it is clear that the kinds of prime movers and levels of energy use do not determine the aspirations and achievements of human societies…Of course, energy conversions are absolutely essential for the survival of all organisms—but their modification and differential utilization is governed by properties intrinsic to the organisms [i.e. rather than explained in terms of their environment].”[14, p. 251]

In the next installment I’ll steer this sequence of posts on efficiency towards its climax, by bringing Beyond this Brief Anomaly‘s systemic approach to bear on one of the more controversial subjects in the sphere of energy and society: the potentially perverse consequences of efficiency improvements associated with rebound and backfire effects.


[1] Murphy, David J., Hall, Charles A. S., Dale, Michael & Cleveland, Cutler (2011). “Order from Chaos: A Preliminary Protocol for Determining the EROI of Fuels”. Sustainability, 3(10), pp. 1888-1907.

[2]  Hall, Charles A. S. (2012). Energy Return on Investment. In T. Butler, D. Lerch & G. Wuerthner (Eds.). The Energy Reader: Overdevelopment and the Delusion of Endless Growth (pp. 62-68), Sausalito: Watershed Media.

[3] Cleveland, Cutler. (2005). “Net energy from the extraction of oil and gas in the United States”, Energy, 30(5), pp. 769-782.

[4] Hall, Charles A. S., Balogh, Stephen & Murphy, David J.R. (2009). “What is the minimum EROI that a sustainable society must have?”. Energies, 2(1), pp. 25-47.

[5] Prieto, Pedro A. & Hall, Charles A. S. (2013). Spain’s Photovoltaic Revolution: The Energy Return on Investment, Hall. New York: Springer

[6] Raugei, Marco, Fullana-i-Palmer, Pere & Fthenakis, Vasilis. (2012). The energy return on energy investment (EROI) of photovoltaics: Methodology and comparisons with fossil fuel life cycles. Energy Policy, 45, 576-582.

[7] Inman, Mason. (2013). The True Costs of Fossil Fuels. Scientific American, April, pp. 58-61.

[8] Inman, Mason. (2013). Behind the Numbers on Energy Return on Investment. Scientific American. Retrieved 21 July 2013, from

[9] Hall, Charles A. S. & Klitgaard, Kent A. (2011). Energy and the wealth of nations: Understanding the biophysical economy. New York: Springer

[10] Cleveland, Cutler (2008). Ten fundamental principles of net energy. The Encyclopedia of Earth. Retrieved 20 July 2013 from

[11] Moriarty, Patrick, & Honnery, Damon (2011). Energy availability problems with rapid deployment of wind-hydrogen systems. International Journal of Hydrogen Energy, 36(5), pp. 3283-3289.

[12] Kessides, Ioannis N. & Wade, David C. (2011). “Deriving an Improved Dynamic EROI to Provide Better Information for Energy Planners”. Sustainability, 3(12), pp. 2339-57.

[13] Debeir, Jean-Claude, Deléage, Jean-Paul  & Hémery, Daniel (1991). In the Servitude of Power: Energy and Civilisation Through the Ages. London: Zed Books.

[14] Smil, Vaclav (1994). Energy in World History. Boulder: Westview Press.

[15] Weißbach, D., Ruprecht, G., Huke, A., Czerski, K., Gottlieb, S. & Hussein, A. (2013). “Energy intensities, EROIs (energy returned on invested), and energy payback times of electricity generating power plants”. Energy, 52, pp. 210–221.

[16] Lenzen, Manfred. (2008). “Life cycle energy and greenhouse gas emissions of nuclear energy: A review”. Energy Conversion and Management, 49(8), pp. 2178-2199.

6 thoughts on “The economic view of systemic efficiency: energy return on energy investment

  1. You are getting there – or have been. Where is the next chapter? I’m looking forward to developments in your thinking. – Oregon Jim

    • Hi Jim,
      Thanks very much for your interest. Family commitments over the past few months have made it all too easy to push BTBA down the priority list–but knowing that there’s some interest in taking the inquiry forward will definitely help there. Hoping to get on to the follow up post to this in the next few weeks…
      All the best,

      • I would like to second that. This is a great resource and a clear overview of the subject that you’re putting together. It would be good to see more of it.

  2. Pingback: EROI and the limits of conventional feasibility assessment—Part 1: The technical potential for renewables | Beyond this Brief Anomaly

  3. Pingback: EROI and the limits of conventional feasibility assessment—Part 2: Stocks, flows and power return on investment | Beyond this Brief Anomaly

  4. Pingback: An integrated view of energy transition: what can we learn? | Beyond this Brief Anomaly

Leave a Reply

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

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

Facebook photo

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

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.