The Real EROI Of Photovoltaic Systems: Professor Charles Hall Weighs In
By Ugo Bardi
27 May, 2016
Charles Hall is known for his multiple and important contributions in the field of sustainability, and in particular for having introduced the concept of Energy Return on Energy Investment, EROI or EROEI. He is now emeritus and still active in research; among other things as chief editor of the new Springer journal: "Biophysical Economics and Resource Quality, BERQ. Here, he intervenes in the recent debate on the EROI of photovoltaic systems, sending me this note that I am happy to publish, with some comments of mine at the end.
By Charles Hall
The EROI of our various energy options, and its associated issues, may be the most important issues that will face future civilizations. The present discussion tends to vacillate between people who accept (or advocate) very high EROIs for solar vs people who accept (or advocate) very low such EROIs. I trust only one study, the one I did with Pedro Prieto, who has a great deal of real world experience and data. This study attempted to (conservatively) estimate all the energy used to generate PV electricity in Spain by following all the money spent (per GW) and using physical analysis where possible, and energy intensity of money where necessary. We found that the panels and inverters, which are the only parts measured in most studies, were only about a third of the energy cost of the system. As noted in the responses to Ugo’s last post we estimated an EROI of 2.45:1 in 2008 assuming a lifetime of 25 years and at the juncture with the distribution system. Studies that we think used more or less appropriate boundaries (Palmer, Weissbach) got similar results.
We recognize that subsequent studies to ours would probably have generated higher EROIs because of using panels of lower energy costs or higher efficiency. But there are many ways that it might be lower too. For example Ferroni and Hopkirk, who (despite, perhaps, some issues) have done us a good service by attempting to get actual lifetimes for modules, which were much closer to 18 years than infinity. This agrees with what happened in Spain when, due to post-2008 financial turmoil, manufacturers did not honor their guarantees and legally "disappeared", leaving broken systems unfixed. (And what happened to all those "surplus" Chinese panels that were never used? Should we factor in their energy costs, as we factor in dry holes for oil analysis?) My point is that we need to include empirical, not theoretical, estimates of ALL the energy used to make these systems work.
This is what Prieto and Hall did, imperfectly I am sure, using conservative assumptions of energy costs, many of which now appear too low. Mostly I do not see others doing this, so I mistrust their analyses. I do not know whether Bandhari et al. included only studies using appropriate boundaries, but I would guess that many are for just the panels (and maybe converters), not the whole system required to deliver the electricity. Another way that we were conservative was to not include the (pro-rated) distribution system, as Ferroni and Hopkirk did (i.e. EROIpou, for point of use). It seems to me that we should do this routinely, at least as sensitivity analysis. If you are really analyzing the EROI of solar you need to get the electricity to the factory, the gravel and panels to the installation site etc. etc,
There are at least three reasons that EROI estimates appear much wider than they probably really are:
1) They are often done by advocates one way or another, not by experienced, objective (and peer reviewed) analysts.
2) a common protocol is not followed. Murphy et al. 2011 should be followed or good reasons given for not doing so. They recommend that all investigators generate a "standard EROI (EROIst) so that different studies can be compared, but then suggest that investigators may define in addition other criteria/boundaries as long as they are well defined and the reason for their inclusion given. This protocol is being updated at this time to deal with various concerns.
3) Related to above appropriate boundaries are often not used. For a start "follow the money" as money is a lien on energy. Where there is controversy (e.g. include labor or not, and how) this should be dealt with through sensitivity analysis. Energy quality (e.g. electricity vs fossil) also needs to be considered, as Prieto and Hall did in their final chapter.
The largest problem with EROI studies is that although the concept has been around and even lauded since at least 1977 it has essentially never been supported by legitimate and objective funding sources such as the US National Science Foundation (which however has recognized this as a large failure and is starting a new program on EROI.) As any investigator knows it takes money to do a good job, and this we have not had. Most of the best work has been done on a shoestring or pro bono. This appears to be changing now, especially in Europe, and we hope to see some kind of objective, high-quality Institute/Program in the future. We also need better governmental statistics on energy use and the development of appropriate energy I-O analyses to get a better handle on energy costs. These had been done to high quality in the US 40 years ago but the official Bureau of Census energy use data has degraded, and we have ceased undertaking appropriate energy I-O analyses while the real experts have retired or died.
If these issues can be resolved, which is not too difficult at least in principle, and if the protocols are followed, then I think we will narrow the range of published EROI estimates considerably. In the meantime I have done a fair amount of sensitivity analysis (e.g. Guilford et al 2011; Prieto and Hall 2012) that suggest that at least for the studies I have been involved with the range of uncertainty is well within plus or minus 25 percent (except when using the assumptions of using the energy cost of the full salary of labor or electricity is multiplied by a quality factor of three, in which case the range is two to three). At this time, we do not recommend either of those two factors for general use. This range of uncertainty is much less than the EROI range among the different technologies, as shown in Euan Mearns most recent post.
Guilford, M., C.A.S., Hall, P. O’Conner, and C.J., Cleveland. 2011. A new long term assessment of EROI for U.S. oil and gas: Sustainability: Special Issue on EROI. Pages 1866-1887.
Murphy, D., Hall, C.A.S., Cleveland, C., P. O’Conner. 2011. Order from chaos: A Preliminary Protocol for Determining EROI for Fuels. Sustainability: Special Issue on EROI. 2011. Pages 1888-1907.
Prieto, P., C.A.S. Hall. 2012 Spain’s Photovoltaic Revolution: The energy return on investment. Springer, NY. (about $50)
A comment by Ugo Bardi
This note by professor Hall highlights some elements of the debate and let me comment on it. Basically, I think that there is nothing wrong in the work by Hall and Prieto that arrived at relatively low values of the EROI of PV (note, however, that there is a lot that's wrong in the recent paper by Ferroni and Hopkirk, but that will be addressed elsewhere). The discrepancies are due to different initial assumptions, as Hall correctly states here, and, obviously, different assumptions lead to different numbers.
Then, the question is, what are the "right" assumptions in these estimates? Evidently, it depends on what one is trying to measure. Here lie the problem and the remarkable confusion surrounding the debate. Basically, there are two main possible aims for an EROI calculation: 1) determining whether a technology is an energy source or an energy sink and 2) determining whether a technology can support an industrial civilization similar to ours (maybe including SUVs and plane trips to Hawai'i for middle-class families).
Once this point is clarified, we see that answering these different questions requires different assumptions. For the first question, energy source or sink, the estimate is defined by the life-cycle analysis (LCA) of the plant. For PV, that includes the cycle of all the components of the plant (surely not just the cells!). Within the LCA framework, the result is an EROI of about 11-12 for the most common technologies available today. There is no doubt that a PV plant is an energy source, not a sink.
For the second question, can PV support a civilization, we are dealing with something very different and it is for this purpose that professor Hall defines the "extended EROI" (EROIext). However, how the term "extended" is to be understood is open for discussion. If you think that a civilization should include plane trips to Hawai'i for middle-class people, then the energy required should be factored in the calculation. Without arriving at these extremes, the more elements you add to the energy cost, the lower the final EROI turns out to be and it is not surprising that Hall and Prieto arrive at values between 2-3. These values still make PV an energy source and not a sink, but find it to be hardly able to support plane trips to Hawai'i. But that should have been obvious from the beginning!
There are a few fundamental problems with the concept of "EROIext" that I think make it a scarcely viable idea, but it might become a standard if we all find an agreement on it. The main problem, I believe, is that when we deal with such a thing as the survival of our civilization we move into a very slippery set of questions. One problem is that EROI is not the only parameter that we need to consider, and PV not the only renewable technology available; to say nothing about defining what we mean as "our civilization". So, claiming that PV, alone, cannot support the present civilization may be true, but it is also totally irrelevant. If our civilization has to survive the ongoing crisis it has to go through profound changes that are difficult even to imagine for us. For sure, however, all the renewable technologies able to produce a positive net energy, such as PV, have a role to play in our future.
Ugo Bardi teaches physical chemistry at the University of Florence, in Italy. He is interested in resource depletion, system dynamics modeling, climate science and renewable energy.