By Paul Lorenzini
In the yin and yang of energy policy debates, we know some can get carried away. Normally we ignore the radical fringe, but sometimes their claims take on a life of their own and need to be addressed. One such charge has found its way as an authoritative reference on Wipikedia, alleging that nuclear power causes more bird kills than wind. There we find a table alleging 0.269 avian deaths per GWh for wind turbines as compared to 0.416 for nuclear power plants. Given all the heat being taken by wind advocates over this issue, one can understand the desire to hit back, but this seemed a bit much. On close checking, it was.
The source is a study by one Benjamin Sovacool. Sovacool first published a report with these claims in 2009 while on the faculty at the Lee Kuan Yew School of Public Policy at the National University of Singapore. Free downloads can be found here and here. In 2012 he published a second version, now being identified as a Visiting Associate Professor at the Vermont Law School and Senior Researcher for Energy Security and Justice at the Institute for Energy and the Environment.
So, how does one arrive at these counter-intuitive conclusions? There are two halves of the analysis, the impacts from nuclear power and the impacts from wind. Take nuclear power first.
The most dominant contribution to Sovacool’s analysis of nuclear power impacts comes from uranium mining and milling operations which he claims “can poison and kill hundreds of birds per facility per year”. In his first report, he supports this by focusing on two “uranium mining” operations “in Wyoming” where he charges that bird deaths are caused by abandoned open pits.
The first is the Canon City Uranium Mine in Colorado (not Wyoming), a mine that operated from 1958 to 1979, and only intermittently since. The owners of the mine were ordered to pay a $40,000 fine when a kerosene spill killed 40 geese in 2008. The spill was a one time occurrence and the operators were required to take steps to prevent further spills. Sovacool assumes the death of 40 geese is a routine occurrence, assumes it happens annually at every operating uranium mine, then based on estimates of the peak uranium production (8.4t of enriched uranium when the mine was operating), and using a conversion of 792 GWh produced per ton of enriched uranium, he concludes the rate associated with the Canon City mine is 0.006 deaths per GWh. Not a very big number, but taking this one incident and leveraging it to represent half of all the uranium mines in the world (as is implied in his averaging of results) is … let’s just say it’s not very good science.
Without belaboring the point, early uranium mining operations have come in for their fair share of criticisms, both with regard to miner health impacts and the handling of mill tailings. But they are strongly regulated today and current operations give a great deal of attention both to health effects and effects on the surrounding ecology. As with any other form of mining, environmental approvals are required prior to operations, compliance with safety and environmental regulations must be maintained during operations, and sites must be rehabilitated at the end of the project. It is simply a fiction to draw from old abandoned projects and apply the data to current operations.
But Sovacool does even worse. His second example does not even involve a uranium mine, it is an old abandoned copper mine in Montana. And it is the data from this example that dominates his conclusions both for the 2009 study and the later 2012 one.
Included in his 2009 report is a Table stating his results are based on “real world operating experience at … two uranium mines”. Referring to the hazards of “open pit uranium mines in Wyoming,” he states “it is not uncommon for these pits to kill 300 birds per year.” The citation is to a US Fish and Wildlife Report where the only bird death claim refers to the death of 300 snow geese in 1995 at the Berkeley Pit mine in Montana. The Berkeley pit is notorious: a one mile by one-half mile wide open pit now filled with water to a depth of 1780 feet that is one of the country’s largest superfund sites. Operated as a copper mine by the Anaconda Copper Company from 1955 to 1982, it has become a tourist attraction.
Is he really claiming this is a uranium mine? The reference in the 2009 paper is admittedly a bit parsed (“it is not uncommon”), but he removes any doubt in his 2012 rewrite where, in reference to open pit uranium mines in Wyoming, he states: “one of these pits killed 300 geese during a single year”. By taking the 300 dead geese, assuming this is an annual event representative of global uranium mining, and then relating that to his estimates of uranium produced by mines in Wyoming and ultimate electricity generated, he arrives at the number 0.45 deaths per GWh.
In his 2009 report, he averages the result from Canon City (0.006/GWh) with the result from the Berkeley Mine (0.45/GWh) to arrive at a mining impact of 0.228 avian deaths per GWh. The 2012 report, however, excludes without explanation the result from Canon City and takes the result from the Berkeley Mine to be the bird kill impact for all uranium mining operations. This allows a larger number, the full 0.45/GWh.
By itself this is nearly double the avian mortality rate for wind of 0.269/GWh reported in his analysis (more on that below). And we haven’t yet considered the impact from nuclear plants themselves, mainly birds flying into cooling towers. Here he draws from data at four facilities: Florida Power’s Crystal River Generating Plant (0.454/GWh); Limerick in Pennsylvania (0.261/GWh); the Susquehanna plant in Pennsylvania (0.01/GWh); and Davis-Besse in Ohio (0.0285/GWh). There is no stated rationale for the selection of these facilities as representative of nuclear plant bird impacts. An average of these four numbers produces an industry result of 0.188 fatalities per GWh.
As is obvious, the controlling example is Crystal River Generating Plant. More games. His analysis claims to rely on an incident in 1982 when 3000 birds were killed in two successive evenings from collisions with smokestacks and cooling towers. The implication is that nuclear power plants were to blame. Yet the citation he provides says differently. The Crystal River Generating Complex consists of four fossil fuel plants and one nuclear plant. The report in question states: “Two pairs of chimneys associated with separate fossil fuel generating units are 152 and 183 m tall.” In referring to the bird kills, it goes on to report: “On 23 September 1982, l265 individuals from a kill estimated by Florida Power Corporation employees to be at least 3000 birds were collected beneath the two pairs of chimneys.” In other words, the 3000 kills is an estimate and the proximate cause was two pairs of large stacks associated with fossil fuel plants!
In short, the most significant avian impact cited by Sovacool for mining involved a copper mine operation, and for operating nuclear plants, the dominant contributor was a fossil fuel plant.
Aside: (from Rod Adams) Paul apparently overlooked another glaring problem with Sovacool’s power plant bird kill analysis. Crystal River Unit 3, the only nuclear unit on the power station site, is a low profile plant that uses a canal to bring water from the Crystal River to the plant.
The tall cooling towers at Crystal River power station, even though they are often used by the media as an icon for a nuclear plant, are used by units 4 and 5. Both of those units are coal-fired power plants built after direct cooling for new plants was made illegal in the US. The cylindrical building in the foreground is the nuclear unit.
One does not know where to take this except to say calling it junk science is too kind. The only real data he cites linking bird kills to nuclear plant operations is Limerick (0.261 fatalities per GWh), Susquehanna (0.01/GWh), and Davis Besse (0.0285/GWh). And it is unclear if Limerick is taking two years of data and conflating it into one, but let’s not quibble. Absent other data, and with nothing to help us decide if these are representative or not, taking these data alone, we can infer a fatality rate somewhere between the mean (0.07/GWh) and the median (0.0285/GWh) for nuclear power plants. Regardless, either number is considerably less than the avian impacts Sovacool himself reports for wind, 0.269/GWh.
But we can’t stop here. A quick look at the wind numbers shows similar biases in the other direction (no surprise). A critique by seven scientists titled “Bats are not birds and other problems with Sovacool’s (2009) analysis of animal fatalities due to electricity generation,” provides a laundry list of problems with Sovacool’s analysis of avian impacts from wind turbines. The abstract says it well:
“Recently Sovacool (2009) set out to compare North American bird (avian) and, presumably bat (chiropteran) mortality resulting from three methods of electricity generation, an objective we applaud. However, we feel it is important to point out serious errors in biological fact, logic, and data selection in his paper”.
The criticisms extend well beyond issues raised here, but suffice to say Sovacool does not fare well. On the most relevant topic, bird kills from wind, they state: “… Sovacool’s estimate of the average number of birds killed per GWh of wind power is incorrect and omits a large body of easily accessible, published data.” Their corrected number for bird kills at the six sites examined by Sovacool results in 0.653 fatalities per GWh, significantly higher than Sovacool’s 0.269. Using a broader sample of wind sites, they arrive at 1.46/GWh, over five times Sovacool’s estimate. When bats are added, it increases to 2.94/GWh.
The end result is not too surprising – wind power kills lots of birds and bats, by factors of ten more than nuclear power. It’s a major concern and is one reason we stretch the point by claiming wind is “clean” energy. But so what? There are a lot of issues to consider when evaluating wind environmental impacts as well as those from nuclear power. What matters most is that we approach these very important public policy questions with integrity and sufficient openness to make fair and honest decisions. There’s no place for the sort of malicious and dishonest distortions used here to make some short term argument. The problem is these kinds of studies find their way into the literature and can take on a life of their own. It can be a grind, but it’s important to make corrections when we find it being done.
Paul Lorenzini served as the CEO of NuScale from 2008 through December 2012.