Today's installment will address two of the potential proxies for calculating "true" EROEI (meaning a calculation with no artificial system boundary) for renewable energy sources. Last week's post discussed the price-estimated EROEI theory, and this week I'll discuss two additional potential proxies. Up front, it's important to note that, much like the price-estimated theory, these proxies have significant weaknesses. I don't think either is yet ready for use, but talking through them helps to better define the issues surrounding proxy-calculation of EROEI, and may result in readers figuring out how we CAN make ideas like these work...
Asymptote location model:
The first proxy I'll discuss is--for lack of a better name--the asymptote location model. I'll start by noting that I think this model is still too incomplete to be workable. I include it here because I think it may be fertile ground for someone to develop. Here's the basic concept: under traditional EROEI calculations, there is an artificial system boundary drawn at some point, and the result is an artificially high EROEI (because those energy inputs outside that artificial boundary are not counted). My theory starts with the assumption that, as that system boundary is expanded, the resulting EROEI will approach some theoretical "true" EROEI that lies at an infinite, but uncomputable, system boundary. This is a classic example of an asymptote. If we can plot the degree of system boundary expansion on the Y axis, and the resulting EROEI value on the X axis, then X will approach the true EROEI value as Y approaches infinity (unbounded EROEI calculation). This, in theory, will allow us to fit an equation to a few points (which can be calculated) and deduce the location of the asymptote that represents "true" EROEI without actually needing to perform the impossible calculation of the unbounded EROEI analysis. The problem, of course, is that while it's quite possible to fit that X-value (EROEI) into a meaningful scale, but the same can't quite so easily be said about the Y-value. What does a "5" on the Y-axis (degree of system boundary expansion) mean compared to a "10"? What is the scale? Unless an 8 is double a 4 is double a 2 in some meaningful sense, the equation fit to locate the asymptote will provide a meaningless result. Can we create a meaningful scale for the Y-axis? Maybe. It seems possible to use the number of steps of regression (as in 1=just the energy used at the plant and installation, 2= 1 plus the energy used to create everything in the plant/installation, etc.) as a scale, but this is just speculation. This might be fertile ground for someone looking to develop the field of EROEI analysis, but it's not ready for prime-time at this point. I'm very interested in any ideas readers may have about turning this rough idea into a workable proxy measurement.
Neil Howes sent me a a very interesting calculation for wind energy that used the ratio of worker-years involved in the wind industry to total US worker-years as a means to determine what portion of total US energy consumption was required as input to US-produced wind capacity. His EROEI measurement came out at over 100:1, and I think significantly overestimates the true ratio because, like most EROEI calculations, it artificially limited the system boundary quite severely (for example, it based its worker-year number on a DOE study that estimated the number of wind-energy jobs that may be created for a set amount of production--this didn't include all the supporting industry jobs that would be created in mining, transportation, marketing, finance, training, etc.). Additionally, I think that the brute-force methods for calculating EROEI (input/output and process analysis) necessarily represent an upper bound to the "true" EROEI--they accurately count energy output, and are universally low (to an unknown extent) on their accounting for energy input. As a result, any proxy that estimates higher than the brute-force approach must be reconsidered.
The far more significant inaccuracy is that this methodology assumes a uniformity of the very EROEI it attempts to measure. The measurement is only accurate IF every worker-year can act as a proxy for an equal amount of US annual energy consumption--it can't. Instead, some workers (and their associated industrial/commercial processes) represent far more energy than others. This lays bare the problem with this methodology: it would come back with the same energy input for 1000 worker-years on a 50:1 EROEI oil well as for 1000 worker-years on a 3:1 EROEI solar plant, even where the energy generation capacity of each is the same--the energy input is not necessarily the same. I would argue that the energy input could be seen as the same IF we took a boundary-less approach to attributing worker-years, but then we get back to our overarching accounting problem.
I think that these two proxy-methodologies outlined above both present some potential for development, but neither is yet ready for actual use. They both present novel approaches to the proxy-calculation of EROEI, but seem to me unacceptably ill-defined--both when compared to brute-force EROEI calculations and when compared to the price-estimated theory of proxy-EROEI calculation.
Going forward, I'll look at both solar and wind, and I'll present a survey of traditional EROEI calculations as well as proxy calculations based on the price-estimated model. If readers have any thoughts on other proxy-methodologies to use (or how to make the asymptote or worker-year methods work), please let me know.