See also The EROEI of energy balancing.

I’ve been spending a lot of time thinking about EROEI (energy return on energy investment) lately since Tim Helweg-Larsen at PIRC asked me for help with a report. The report, “The Offshore Valuation” was produced by the Offshore Valuation Group chaired by PIRC and it looks at the value to the UK of the offshore renewable energy resource.

Tim, ever present with the big ideas, asked me to have a look at the EROEI of a whole range of energy technologies. What I did was begin a trawl of the literature hunting high and low for figures on EROEI, EROI, life cycle energy, net energy, life cycle analysis, energy payback time, etc and the result was a table on page 61 of the report listing the average EROEIs I found.

The abstract for the study we delivered is below.

“Energy Return on Energy Investment (EROEI) is calculated as the ratio between energy inputs and energy outputs for an energy generating technology. A process that uses more energy than it produces is by definition unsustainable in the long term. Technologies which require fuel generally have a lower EROEI than those which can extract “free” energy from the environment (wind, waves, tides, sunlight). As part of this study we have a carried out a review of published EROEIs and have found a high EROEI for marine technologies, lower EROEIs for solar and nuclear power, and still lower for coal and gas.

“In addition this study has found a suggestion of increasing EROEI for wind turbines and nuclear power, and a falling EROEI for coal, gas and solar power. Of these, the most significant is the reduction in EROEI of coal. The study examines the reasons for these relationships with the greatest reductions expected to come from carbon capture and storage and the greatest increases expected from increased size of wind turbines and lower energy inputs to nuclear fuel enrichment.

“Finally the study looks at the relative lifecycle efficiencies of grid connection and a number of other balancing methods. Grid connection and HVDC interconnectors are found to have a similar lifecycle efficiency of around 95%. Energy storage technologies have lifecycle efficiencies in the range of 45% to 75%.”

We also created some interesting graphics based on the data. Firstly we adapted the bubble charts idea used by Tom Konrad, the Clean Energy Wonk. His idea is to calculate the energy internal rate of return (EIRR) for technologies. This is the percentage of the energy invested that is returned each year, analogous to IRR for financial investments.

This gave the chart below.

EROEI and EIRR bubble chart

Technology Average EROEI Average EIRR Average lifespan Number found
Coal no CCS 5.5 17% 31 11
Coal w. CCS 1.5 8% 23 2
Solar thermal elec. 9.9 40% 25 7
Gas no CCS 3.5 11% 32 5
Gas w. CCS 2.2 13% 23 2
Nuclear 10.9 36% 29 50
PV 8.3 34% 24 46
Tidal range 115.9 97% 120 1
Tidal stream 14.9 74% 20 2
Wind 25.0 125% 21 108
Wave 12.0 60% 20 2

Edit: references are here.

What’s interesting about this, particularly for the Offshore Valuation Group, is that the offshore resources (and on-shore wind) have high EROEIs and very high EIRRs. There is a clear demarcation between wind, wave, and tidal at the top of the EIRR tree, solar and nuclear in the middle, and coal and gas at the bottom.

Even more interesting (although not very reliable in statistical terms) is the trajectory of historic estimates of the EROEI of some of these technologies. The graph below is projected out to 2020.

EROEI trajectory

This graph is based on some very small data-sets for gas and coal. The justification for including them is that there are sound technological reasons why we can expect the EROEI of gas- and coal-fired generation to fall. Coal- and gas-fired power plants are becoming less efficient. This is primarily due to legislation passed to minimise the environmental impacts. In the past, flue scrubbers were mandated in order to minimise the emissions of sulphur dioxide, a key cause of acid rain. In the future carbon capture and storage is very likely to be required. Both of these impose additional parasitic loads on power plants which make the plant less efficient and so reduce the EROEI.

Additionally, although the peak for coal is not expected for some time, the energy required to mine and transport coal is increasing. Over time high quality black coal will run short and dirtier brown coal will be substituted. This will carry an energy penalty as brown coal is less energy-dense and takes more energy to avoid sulphur dioxide emissions. Peak gas is expected before peak coal. The energy inputs to North Sea oil and gas have been increasing over the years as more advanced oil extraction techniques have been applied. Another major future source of gas in the UK in liquefied natural gas which is chilled, compressed and imported by tanker with a high energy cost.

On the other hand, wind energy is becoming more efficient. With wind energy, bigger is better for energy return. Energy return increases with the square of rotor diameter. If the rotor is twice as big, it produces four times the power. Turbine size has been increasing for many years, which is the key reason for the steep rise in EROEI.

What about PV? The problem with PV is that the improvements in EROEI are incremental, not geometric as with wind (and tidal) turbines. They don’t have the same power law increase in energy return when scaled up. Despite this fact, the error margin in the data and the ongoing development of thin-film PV which has higher EROEI than traditional crystalline PV mean that the true trajectory could well be upwards.

And nuclear. This was the real surprise for me in this analysis. It appears that there are currently two main types of enrichment used, gaseous diffusion and gas centrifuge enrichment. Gaseous diffusion uses more energy and is being replaced by gas centrifuge which is one of the main influences which has led to a higher nuclear EROEI. The peak for fissile materials is not expected for some time, at least at current rates of extraction. Nevertheless, the purest and most convenient supplies will inevitably be used up first causing a downward pressure on EROEI. Some companies are even currently researching producing yellow cake (Uranium ore) from the ash from coal-fired power stations which is, I suppose, a good example of cradle to cradle thinking.

All of this goes to show that for the UK to get over the looming energy gap there is little better in energy terms than to go for the on-shore wind, a suite of offshore renewables, and much as I hate to say it, probably a wedge of nuclear too. Clean coal is a dead end.


Jamie Bull |

Related Posts

Go straight to the tool.The tool has been updated to use client-side javascript and is now much, much faster. The API details in this post are of historical interest only.If you want to know how to design your building you to need to know what the local weather is like. If you want to understand […]

Low-energy bulbsLow-energy light bulbs split the nation down the middle. Half of us believe that these bulbs will save energy, bring down our energy bills, and reduce carbon emissions. The other half believe that there are all sorts of things that haven’t been considered, and that if we look at the whole picture then the […]

The setting of boundaries is hugely important in calculating EROEI. Where do you stop with bringing in other costs? If it’s not on the financial balance sheet can you ignore it? I would say no – if it has an effect, it goes in. That seems to be a limitation of input-output analyses. Where do […]

24 Comments on “EROEI of electricity generation”

  • Dave says:

    Very nice graphics. You should submit them to Telling a very important story, I’ll use them in my presentations (with the relevant references naturally)

  • Jamie Bull says:

    Thanks Dave. I see you have a graphic at Future Energy about various energy storage technologies. I’m currently working on a follow-up piece about the life cycle EROEI of energy balancing (batteries and grids) so you may get another graphic to add to the presentation.

  • Dave says:

    Great! Keep them coming…and feel free to post anything relevant on Future Energy for the audience there.

  • jaggedben says:

    Frankly, the EROEI numbers in your graphs above are … strange.

    First of all, they differ widely from well known estimates of EROEIs such as Charlie Hall’s.

    Can you explain why your numbers are so different from his? Is it because your numbers only apply to the UK, or what? Your assertion that fossil fuels have EROEI’s much lower than, say, solar PV, would be great news if true. But that assertion certainly flies in the face of the conventional wisdom.

    Also, do you really mean to suggest (in your trends graph) that the net energy of gas probably went negative a couple years back? That seems counter intuitive on the face of it (or we’d have stopped using gas) unless perhaps you are talking about the expected EROEI of new plants, as opposed to existing infrastructure.

    Also, was the tidal EROEI study of a real generating plant?

    I could not find any footnotes in “The Offshore Valuation” regarding the EROEI studies you looked at, or your methodology. It seems to me that some additional documentation, and clarifications regarding what you are saying, are needed to support the information you’re putting out here.

    • Jamie Bull says:

      Hi jaggedben,

      Thanks for all your questions. I’ll try and answer those that I can. Yes, I am familiar with Charlie Hall’s numbers, although I haven’t been able to track down any references for the gas or coal numbers. If you look at the figures for electricity producing technologies in his graphs you will find they are in reasonable agreement with those that I found. My theory on his figures for coal and gas are that they are for bringing the fuel to the surface only. There are then the additional energy costs to transport it to where it is used to generate the electricity, the energy penalty due to the efficiency of generation plant, including parasitic loads from flue scrubbers, etc.

      The dates used are the year of the study, so some are projections of future technology such as carbon capture. The likely cause behind natural gas appearing to go negative is due the fact that the most recent studies assume that plants will be capturing carbon. There is a disclaimer next to the trajectory graph stating that they are not particularly convincing in statistical terms due to the low number of data points. The average figures are probably more indicative of the real EROEI of plants operational today.

      The tidal barrage study is of the proposed Severn Barrage so no, not operational. It’s based on figure in a Black and Veatch report for the Sustainable Development Commission available here.

      The other references and calculations are all in a spreadsheet which I’ve been planning on posting for some time but haven’t quite got around to polishing up. Now I seem to be getting a lot more traffic thanks to Energy Bulletin I’ll pull my finger out.

      Edit: References are now in the references post.

      • jaggedben says:

        Thanks for the reply Jamie.

        You’ve made me think about the numbers more closely, and I’m coming around to agreeing with you on coal at least.

        I just did some back of the envelope EROEI calcs on train transportation of coal in the US, assuming 500 gross-ton-miles/gallon for train efficiency and 1500 miles from mine to power plant. Using BTU numbers from the EIA, this gives you about a 47:1 ratio of coal energy to transportation energy required. Clearly when extraction, efficiency, and the cost of building power plants are also factored in, Charlie’s balloon for coal on his graph is unreasonably high, at least for electricity production. (To be fair to him, EROEI for electricity production is probably not what his graph was intended to show. OTOH, that’s about all we use coal for here in the US.)

        Regarding the efficiency factor, something I’ve realized relatively recently is that the habit of using the ‘primary energy’ (btus) of coal, oil and gas to measure ‘how much energy we need’ has been misleading over the years. We need to look at the energy we actually use, and work back from there to compare different energy sources.

  • Roderick Beck says:

    Your figures don’t make sense. The energy required to build a nuclear plant is relatively low compared to what such a plant generates.

    Your source of figures is suspect. The source has an axe to grind.

    The French have already shown you can build a developed country on nuclear power.

    • Jamie Bull says:

      Hi Roderick,

      The references are all in the references post. All the nuclear figures are from a meta-analysis by Lenzen in 2008 (reference 1 on the list).

      I have to say, I don’t understand your problem with the figures at all though. They show a reasonably high, and rising EROEI for nuclear power. In fact doing this piece of analysis has pushed me slightly further into the nuclear corner if anything.

  • Grunergy says:

    We really have to rid ourselves of coal in order to move forward.

  • Dave says:

    Jamie, just another follow-up to this that I thought was relevant (you may be aware of it already, it’s a Charles Hall one) – it’s a paper which suggests that EROEI needs to be at least 3 to avoid the use of fossil fuel ‘subsidies’. It’s talking specifically about biofuels but may be of broader interest.

    I’ve commented on my social forum (

  • Dudley says:

    Thanks for an interesting post Jamie.

    I disagree with your assertion that wind power expands as the square of the size of the device and PV does not. The only reason for this is that the power increases proportionally to the swept area of the turbines blades, and clearly the swept area increases as the square of the diameter.
    However, a PV plant will also increase its power proportionally with the increase in area, and if the PV plant were circular, then it too would increase in power as the square of the size – if the “size” means the diameter of the capturing area (Applies approximately to a square field too if size is the size of one of the sides)


    • Jamie Bull says:

      The point is that the amount of materials increases in line with an increase in area of PV. To a first approximation, the EROEI of PV doesn’t increase with a larger array (ignoring some effects to do with the efficiency of inverters). This is because all of the additional area is made up of the same stuff – silicon, frame, etc. However this doesn’t apply to wind. The EROEI of wind does improve as the swept area increases, since the swept area is mainly made up of spaces between the blades.

  • Scott Troyer says:

    In the table listing EROEI and EIRR values, the EROEI is less for gas using CCS than gas w/o CCS, which would be expected, but the EIRR goes up which does not seem possible. Is this an error and if so what should the correct value be.

    • Jamie Bull says:

      Scott, thanks for your comment. I can see why that looks odd but in fact there’s no error there. The reason is that the average lifespan predicted in the two examples I found of the gas + CCS case is shorter than the gas with no CCS case. For the case with CCS it was 23 years and without it was 32 years.

      So no, no error. But please do bear in mind that with so few data points, and with assumptions having been made, these numbers are only approximations.

  • Maxlyte says:

    With Mid Wales marked out to host many huge turbines do the energy cost calculations factor in all energy used in construction? That is, the concrete bases for turbines and those for the cranes which build them; fuel used to transport to site; fuel used to alter roads to allow access to site including new access roads; fuel used by vehicles delayed and in queues waiting for transport vehicles (Mid Wales looks likely to be one big traffic jam if all the windfarms are built); fuel used to dismantle defunct turbine pads; energy used to create the pylon infrastructure. All the energy used by protesters trying to stop the local area being despoiled?

    • Jamie Bull says:

      If you follow up the references you can see what was included in each study. The main source for the wind turbine figures is a review article by Kubizewski et al (2010). The “Scope as stated” column in the main table tells you what was covered by the study, although you’d probably need to go back to the original papers to understand it exactly.

      Taking the items you mention specifically, if the study is an input/output (I/O) analysis, all costs to the company in all stages in the scope as stated referred to are covered. That will cover fuel to site, fuel used on site, new access roads and pylon infrastructure where they are paid for by the operator. Protesters’ energy is also not included (not least because in most places where the turbines are built it’s not an issue which is protested about) except where it adds cost to the operators. The general principle in I/O analysis is if the operator has to pay for it then it should be included, otherwise it’s not. If the study is based on process analysis (PA) it is likely to omit more stages due to the difficulty of obtaining precise data and so EROEI generally looks better. The majority of data in this review is based on I/O analysis though.

      For extra fuel in traffic jams, that certainly won’t have been included. I’d be interested to see an estimate of what that actually adds per wind turbine. And if it’s very much, I’d also like to see it added to the fuel efficiency calculations for cars towing caravans and tractors transporting bales which seem to be the main thing I’ve been stuck behind in Wales!

    • Rocio says:

      Ken FabosNovember 22, 2010 Does anyone ralley believe society will calmly stand by as we head towards 3b0C, then 4b0C, staring economic and social collapse in the face while we focus on cheap energy as some kind of overriding objective? Going on performance to date, yes.Two full decades of anthropogenic climate change being established science and Australia’s policy focus is fully on maximum expansion of coal and gas extraction and export with some ineffectual climate policies’ to distract and pacify public concerns. Expansion of grid supply by construction of big new coal fired power plants such as in the Hunter Valley and near Lithgow are going ahead and look to me to be intended to prevent the issue of decarbonising our energy supply getting mixed up with the issue of maintaining growth and reliability of supply; we’ll have enough fossil fuel generating capacity that building low emissions capacity will remain optional’ and can be deferred another decade or two. Pressed hard enough by concerned voters leaking to the Greens there might be a switch of emphasis from the LibNatLab middle to push for gas over coal despite gas being unable to deliver the long term emissions reductions needed. Screw the future, Australian’s don’t want their power bills going up and LibNatLab policy makers are sure of it.

Leave a Reply

Your email address will not be published. Required fields are marked *