The range of wind EROEI I found during my literature review for the table on page 61 of the Offshore Valuation is staggeringly large. Much larger in fact than any of the other technologies included. It demands consideration of why this might be.
There are two (or three) categories of answer to this question. Firstly there are issues with the EROEI methodology as a whole. Secondly, there are issues with the scale- and site-dependency of wind turbines. And thirdly there’s intentional bias in one direction or another. This bias is mainly expressed through exploiting one or more of the first two sources of error which is why I hesitate to include it as a source in its own right.
Imperfect / non-standard methodology
There are a number of ways of calculating an EROEI. Embodied energy (energy invested) can be counted using process analysis (PA) which looks at all the individual energy using processes and adds up all the embodied energy of materials used. Alternatively it can be estimated from input-output (I-O) tables which give figures for how much money spent in one industry gets passed through as spend in another industry. The energy consumption of these industry sectors is known (for some countries) so the overall energy spend can be estimated.
“Studies using the input–output analysis have an average EROI of 12 while those using process analysis an average EROI of 24. Process analysis typically involves a greater degree of subjective decisions by the analyst in regard to system boundaries, and may be prone to the exclusion of certain indirect costs compared to input–output analysis.” (Kubiszewski et al, 2010)
PA is more appropriate for products where the specifics are known. But downstream spend should also be included, making a hybrid methodology with the benefit of both method.
For the energy return, the problems are different. It is hard to set a figure for down-time. How often will, the turbine be out of action? How long will the turbine last? What capacity factor will it achieve? As Kubiszewski et al say:
Inclusion or non-inclusion of surrounding costs
The classic example is “do you count the energy in the builder’s breakfast?” – but taking a step back from that, do you count the energy in the construction boss’s annual skiing trip to the Alps? All of this money spent has inevitable energy consumption associated with it.
So by that measure, does “value added” contribute to lowering the EROEI? If a company makes less profit and so pays its workers lower wages, does their lower spend mean that the EROEI of the less profitable turbine is lower? And is the EROEI of a turbine made and used in China higher than one made and used in the UK or the USA?
There has been some debate over a study which showed much higher EROEI for turbines in Brazil than for those in the USA. It would be interesting to explore whether this economic aspect was one of the underlying reasons, although two of the authors of the wind meta-analysis paper suggest it has more to do with the electricity used in manufacture (Brazil uses a lot of hydro while Germany uses a lot of coal).
Recycling / recyclability
One of the difficult decisions when setting boundaries is how do you account for the energy saved by using recycled materials or by ensuring the materials are recycled at the end of life? There is a useful appendix to Bath University’s Inventory of Carbon and Energy (known as the ICE database) which discusses this issue. What it really comes down to is that it is important to be clear how this has been accounted for.
However, putting aside these thorny methodological questions, there are other issues which are specific to the EROEI of wind energy. Let’s take a look at these.
Size of turbines
The biggest explanatory variable for the variation of EROEI of wind turbines is the size of the turbine. Bigger is better as I’ve mentioned before. The graph below shows the correlation between rated power and EROEI for the operational turbines in the Kubiszewski paper.
This is a pretty good fit and shows the expected diminishing returns to increasing size (energy output increases with the square of the rotor diameter – proportional to the area of the rotor – while material requirements increase cubically – proportional to the volume of the whole turbine). This means that there should be a theoretical point where increasing turbine size no longer improves the EROEI, although there’s no sign of that yet.
The capacity factor of a wind turbine is a measure of how much energy it produces compared with how much it could produce if it was producing at its rated capacity 24 hours a day, 365 days a year. Obviously the wind is a little less reliable than that. This capacity factor can vary hugely from just a few percent for a badly sited small turbine (Location, Location, Location) to upwards of 30% for an offshore turbine.
This is a more contentious issue when calculating the EROEI of a theoretical turbine rather than an operational one. For an operational turbine the capacity factor is generally known, while for a theoretical one an assumption has to be made, with the potential for inaccuracy (and bias) that that introduces.
The fact that capacity factors show such a large range is another one of the reasons that wind EROEIs are so widely distributed. The reason the influence is not as large as that of power as it only has a linear impact on the energy return.
Another variable/assumption with a linear impact on energy return is lifetime. If you assume an additional 5 years of productive life then there are more years of profits to pay off the initial investment. The studies in the meta-analysis most of this post is drawn from use life spans of between 15 and 30 years with the overwhelming majority using 20 years as the chosen service life.
Bias can be found even when you quote the correct numbers like when anti-wind campaigners talk about wind turbines only producing 30% of their power output when that is exactly the capacity factor you’d expect from a well-sized turbine.
It is clear that a lot of people and groups have an agenda when it comes to wind energy. How do you recognise and avoid bias?
One way of trying to avoid biases, whether unintended biases due to poor experimental design or to more nefarious reasons, is to conduct a meta-analysis of a number of assessments. This is common practice in the world of medicine where the efficacy of a drug or other intervention may be close to the level of statistical significance. Combining a large number of previous analyses allows any small biases to average out – assuming they are normally distributed.
This is why the work I did for the Offshore Valuation was a literature review. Given the time and resources available, a meta-analysis of all the studies I could find was the best way to get a feel for the balance of opinion on the EROEI of a large number of technologies.
A consistent set of boundaries and assumptions should be applied to both the energy return and energy invested parts of the equation. EROEI assessments carried out on the same basis are more likely to give comparable answers.
This is a much harder prospect as boundaries are inherently fuzzy. There are emerging standards for life-cycle analysis (LCA) in Europe and these ought to lead to more easily comparable EROEI calculations but there will inevitably be some omissions.
Be aware that EROEI works best as a relative measure where you compare a number of technologies on the same basis using the same assumptions to see how they perform against each other.
Standardisation can only improve things so much. There will always be room for creative accounting and boundary setting, as well as for making overly pessimistic or optimistic assumptions. It should be abundantly clear that it is very easy to misunderstand / misrepresent the results of study if the assumptions aren’t made clear.
So the number one question to ask when you see an EROEI quoted for a technology is who is quoting it? The source is always important, and even more so for a figure which is as prone to bias as EROEI. Follow up references and drill down into the assumptions.