Why “100% Renewables” is a Dangerous Chimera

Part 1: The EROI Issue

In August 2014, Barry Brooks uploaded a guest post by John Morgan that became one of the most widely viewed and heavily commented postings ever to The Energy Collective. It was titled "The Catch-22 of Energy Storage", and can be found here.

In 2014, Morgan was an adjunct professor in the School of Electrical and Computer Engineering at RMIT (Royal Melbourne Institute of Technology) and Chief Scientist at a Sydney startup developing smart grid and grid-scale energy storage technologies. The article that Brooks uploaded was originally published in the journal Chemistry in Australia. It draws on a paper by D. Weißbach et. al. titled "Energy intensities, EROIs, and energy payback times of electricity generating power plants". That paper was published in the journal Energy in April of 2013. It’s frequently cited and mostly well regarded for its formal approach to the study of EROI (Energy Return On (Energy) Invested). That’s not to say it’s without controversy; as we’ll see, it runs counter to narratives that are popular and well-established within European Green parties – especially in Weißbach’s native Germany.

It’s worth quoting here the introductory paragraph from Morgan’s article:

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 idea of “enough surplus energy left over .. to power our present civilization” may be novel to many readers. However, it gets to the heart of what analysis of EROI is all about.

EROI threshold

There’s a widespread misconception about EROI, to the effect that any EROI above 1.0 should be sufficient to make an energy resource viable. That’s what we’d expect if we were talking about monetary return on a financial investment, and it isn’t immediately obvious why energy return on energy invested should be different. 

If we could wave a magic wand – or just sign a check – and magically have all the energy returned from a prior investment immediately reinvested to produce more energy, maybe energy ROI wouldn’t be different from monetary ROI. However, money is easy to move around, and debt greases most transactions. That’s not the case for energy. 

Energy invested in an energy resource is gated by labor, time, materials, and capital equipment. All of those things require the support of a complex working society. That society has energy needs. Those energy needs must be satisfied by the returns from investments already in place. Nature doesn’t allow us to borrow energy with a promise to repay it later. So if the returns from the energy investments we’re making aren’t sufficient to support some of those non energy-returning activities, they have to go. Hence, there’s a minimum aggregate energy ROI, substantially greater than one, that has to be achieved from our energy investments in order to keep society running.

How much greater than one must it be? That’s tricky to answer, because it’s a soft number that depends on how we, as a society, expect to live. It also depends on the efficiencies of the mix of technologies we employ. The aggregate energy ROI of our energy resources, for example, might be inadequate to support the level of car ownership and travel that we currently enjoy, but it could still be OK if we relied more on bicycles and electric scooters and stuck closer to home for our food and shopping needs.

Weißbach et. al. figure a minimum EROI threshold of seven for the collection of processes used to power a civilization of the wealth, complexity, and technological development of present OECD nations. That means that production, storage, and distribution of energy must consume no more than one seventh of all the energy consumed in an OECD-level civilization. The rest is needed to sustain activities of the civilization that are not energy-producing.

That conclusion doesn’t say that no civilization could exist if it had to draw on energy resources with an EROI of less than seven. One certainly could exist; it just wouldn’t look like our present civilization. A good portion of the energy-consuming activities that we take for granted would be out of reach. The fraction of GDP associated with production of energy would necessarily be higher, as would the fraction of annual income spent directly or indirectly for energy. Civilizations like that have existed in the past, before the advent of fossil fuels ushered in the industrial revolution and so transformed society. Many believe the world would be better if higher energy costs did compel us to live simply and more frugally in the future.

That’s a complex issue – not to say a can of worms. I don’t want to open that can and devote the rest of this article to sorting worms. But I will say that a society in which frugality is imposed by the low EROI of its energy resources is not one I'd wish upon future generations. I fear the reality would not resemble what advocates for sustainability and simpler living imagine. The larger investment of energy and labor required to eke out a fixed energy payoff would likely expand, not shrink, humanity’s heavy footprint on the world. Simplicity may well be desirable, but it must be a voluntary choice, not imposed by poverty and energy scarcity.

With that as background, let’s take a look at the findings that Morgan reported from Weißbach’s EROI studies. What was it about them that so upset Germany’s anti-nuclear renewable energy warriors? It’s not actually hard to fathom.

Buffered vs unbuffered energy resources

The figure below (figure 3 from Weißbach's paper) summarizes the EROI values found by Weißbach et. al. for a range of energy resources. 

Note that there are two EROI columns for each energy resource: buffered and unbuffered. For firm resources they’re the same – meaning no buffering (i.e, energy storage) is required [1]. For intermittent renewables, buffering is required. It reduces the net EROI for those resources. The energy invested is increased by the amount needed to build the required storage, while the energy returned is reduced by the round-trip efficiency of the storage technology employed. That much is – or should be – obvious and non-controversial. However, there’s controversy about the amount of storage required.

There’s also room for controversy around the specific EROI figures that Weißbach derived. Weißbach et. al. compiled their figures based on selected site-specific EROI studies that had been previously conducted. They limited their selection to studies in which sufficient data was included to quantify real energy inputs in a “bottom up” approach. They rejected studies that were heavily dependent on “top down” estimates of energy costs based on monetary costs of inputs. That warrants some explanation.

It’s common in EROI studies to use monetary costs as a proxy for energy costs for some inputs. The proxy is used when it’s difficult or impossible to determine the actual energy costs for those inputs. Every country has a ratio between GDP and energy consumption that can be found from published statistics. But it’s just an overall average. The actual ratio between monetary cost and energy cost for any particular input can depart widely from that average. That makes monetary cost a low quality proxy for actual energy cost. If the input in question is only a minor contributor to the system being analyzed, then the uncertainty introduced won’t matter greatly. But if energy inputs estimated that way account for a substantial fraction of all energy inputs, the results become too uncertain to have much value. 

For that reason, Weißbach only considered studies that presented sufficient data to permit a credible bottom-up determination of EROI. The data allowed the energy cost estimates to be updated, where appropriate, to reflect advances in technologies. In some cases, where the source studies had considered only the energy cost of fuel delivered to a power plant, the EROI estimates were also revised to include the energy cost and efficiency of the power plant.

While it made technical sense to be selective in the EROI studies he considered, Weißbach’s selections exposed him to charges of “cherry picking”. The high value that he and his team derived for the EROI of nuclear drew perhaps the most heat. Seven years before Weißbach et. al. came out, a paper by anti-nuclear activists Storm van Leeuwen and Phillip Smith had been published. Titled "Nuclear Energy: the Energy Balance", it attempted to show that with the imminent depletion of high grade uranium ores, the energy cost of obtaining uranium from low-grade ores would exceed the energy return from nuclear reactors. The paper was heavily criticized for its methodology and flawed assumptions, but that didn’t keep its “findings” from being widely cited by other anti-nuclear activists. 

Weißbach does reference Smith and Leeuwen’s work, and gives reasons for discounting it in his own studies. The radically different values he derives for the EROI of nuclear power may be a hard lump for Smith and Leeuwen’s fans to swallow. Interestingly, however, one doesn’t seem to hear much about Smith and Leeuwen’s study these days – even from the devout anti-nuclear, pro-renewables crowd. A possible reason may be that, under the assumptions for the cost of mining and materials processing that Smith and Leeuwen employed in their attempt to prove a low EROI / high carbon footprint for nuclear, wind and solar would show up as total non-starters. 

Even most anti-nuclear activists now acknowledge that the materials and energy cost of solar PV and wind energy are an order of magnitude higher, per megawatt-hour of annual output, than they are for nuclear power. Not surprisingly, that difference is reflected in the EROI values that Weißbach cites for nuclear vs. wind and solar PV. 

In the results from Weißbach’s study, renewables fare poorly. The EROI values for combined cycle gas turbines, coal, hydro, and nuclear are all comfortably above the economic threshold of 7. Of the renewables, unbuffered wind and concentrated solar in the desert, at 16 and 19, are reasonably above the threshold. When buffering is factored in, however, the EROI for wind drops to 3.9, well below the economic threshold. The buffered EROI of desert CSP drops by a smaller factor, reflecting the fact that CSP is already partially buffered by storage of heat. But full buffering still drops its EROI to 9, only marginally above the economic threshold.

For solar PV, even the unbuffered output shows an EROI of only 3.9. When buffering is factored in, it drops to a dismal 1.6. However, that’s for rooftop solar in Germany. Germany is a poor location for solar PV. It’s often cloudy, and the high northern latitude of much of the country means short daylight hours and weak sunlight from late fall through winter. The large seasonal variation, in combination with extended periods of “dunkelflaute” (dark and still, when neither wind nor solar are producing much), translate to high storage requirements. That greatly reduces EROI. 

In all cases where buffering was required, Weißbach assumed pumped hydroelectric storage. That seemed appropriate because in terms of megawatt-hours of storage capacity, pumped hydroelectric accounts for the overwhelming share of energy storage capacity in the world. It also has the lowest specific energy investment cost of any storage option. That makes it a conservative choice for calculating the EROI impact of buffering on intermittent resources. 

We’re not done here

Despite the alarmingly low EROI found for intermittent renewables with storage, I don’t think Weißbach’s study can be taken as a “stake through the heart” of the 100% renewables dream for clean energy. I don’t think Morgan, in his article citing Weißbach, thought so either. The message I got from the article was not so much oppositional as cautionary regarding intermittent renewables. It was saying, in effect, “If we’re determined to rely on wind and solar as mainstays for a decarbonized energy economy, then we really need to work on improving their EROI. It’s especially critical that we develop more efficient and affordable options for large scale, long term energy storage”. 

In the seven-plus years since Morgan’s article was reposted here, there has indeed been some progress on those fronts. I don’t think the progress has been nearly as great as wind and solar advocates would like to pretend; it hasn’t been nearly enough to compensate for the 60% drop in EROI that a 2.5x level of wind and solar over-build will lead to. That much over-build will be required to support green hydrogen as the solution for intermittency. Still, viewed strictly as an engineering problem, I don’t think the EROI issue is unsolvable. For starters, I’m sure there are better solutions for intermittency than green hydrogen.

The problem for the “100% renewables” vision (those versions of it that don’t include nuclear power as renewable) is that the EROI issue isn’t just an engineering challenge. When thinking about the future, we need to distinguish three things: (1) what’s possible, in terms of physics and engineering; (2) what’s desirable, in terms of our social and environmental values; and (3) what’s likely, in terms of political, economic, and social dynamics and the institutions we live within. I’ll have more to say about all three in the next installments of a set of postings on why “100% renewables” is a dangerous chimera. But – spoiler alert – while item (1) alone may not be a stake through the heart of the 100% renewables vision, items (1), (2) and (3) in combination will almost certainly prove lethal.

Notes

[1] The different EROI numbers for “unbuffered” and “buffered” hydropower are a “sampling artifact”. Buffering is normally implicit in hydroelectric projects, courtesy of the reservoir behind the hydroelectric dam. It wouldn’t make sense in those cases to buffer the power output separately, because buffering could be achieved more efficiently by regulating flow through the power turbines. However, the EROI number cited for “unbuffered” hydro power  was from a single study of a “run of the river” project. For that project, there was no hydroelectric dam. The EROI number for the “buffered” hydro power was derived separately for more conventional hydro projects that did include dams. In his paper, Weißbach cautions that EROI for hydroelectric projects will vary widely with the particulars of any given project.