The catastrophic disasters at Chernobyl and Fukushima are among the worst humankind has had to deal with. Both were the result of the inability of scientists and engineers to foresee how seemingly small problems can snowball into disasters of almost unimaginable scale.
Given that most countries with nuclear power intend to keep their reactors running and that many new reactors are planned, an important goal is to better understand the nature of risk in the nuclear industry. What, for example, is the likelihood of another Chernobyl in the next few years?
Today, we get an answer thanks to the work of Spencer Wheatley and Didier Sornette at ETH Zurich in Switzerland and Benjamin Sovacool at Aarhus University in Denmark. These guys have compiled the most comprehensive list of nuclear accidents ever created and used it to calculate the likelihood of other accidents in future.
Their worrying conclusion is that the chances are 50:50 that a major nuclear disaster will occur somewhere in the world before 2050. “There is a 50 per cent chance that a Chernobyl event (or larger) occurs in the next 27 years,” they conclude.
The nuclear industry has long been criticised for its over-confident attitude to risk. But truly independent analyses are few and far between, partly because much of the data on accidents is compiled by the nuclear industry itself, which is reluctant to share it.
The International Atomic Energy Agency rates accidents using a system called the International Nuclear Event Scale, which is related to the amount of radiation released. However, the Agency does not publish a historical database of these accidents, probably because it has a dual role of both regulating the nuclear industry and promoting it.
So it has fallen to others to compile lists of accidents, the most comprehensive of which contains details of 102 events. (By comparison there are 72 events that have a rating on the International Nuclear Event Scale.)
Wheatley and co have significantly increased this number. They refrain from using the data from the International Atomic Energy Agency and compile their own list instead.
The metric they use in assessing each accident is its total cost in U.S. dollars (based on the dollar value in 2013). And they define an accident as “an unintentional incident or event at a nuclear energy facility that led to either one death (or more) or at least $50,000 in property damage.”
Each accident must have occurred during the generation, transmission, or distribution of nuclear energy. That includes accidents at mines, during transportation by truck or pipeline, or at an enrichment facility, a manufacturing plant, and so on.
The team gathered their data from a number sources, such as published reports and peer-reviewed papers but also from press releases, project documents, public utility commission filings, and newspaper articles in English.
They then calculated the cost of each accident based on all the economic losses it caused, such as the destruction of property, the cost of emergency response, environmental remediation, evacuation, fines, insurance claims, and so on. Whenever an accident resulted in the death of an individual, the team added $6 million to the cost, a figure also used by various U.S. agencies in calculating the value of a life.
Wheatley and co acknowledge the imperfections of this technique but say it has the huge advantage of representing all the negative consequences of an accident in a single U.S. dollar figure. And this in turn allows the accidents to be ranked
The resulting list ranks 174 accidents between 1946 and 2014 and includes their date, location, the monetary cost in U.S. dollars, and the rating where available on the International Nuclear Event Scale and on another well-known scale called the Nuclear Accident Magnitude Scale.
The top five accidents ranked by monetary cost are the Fukushima accident in March 2011, the Chernobyl explosion in April 1986, a fire at the Tsuruga nuclear plant in December 1995, a fire at Rocky Flats nuclear weapons plant in September 1957 and an incident in March 1955 at Sellafield, then known as Windscale, two years before the infamous fire at the facility. Indeed, Sellafield appears five times in the list of the top 15 of most expensive nuclear accidents.
The new database contains 75 percent more entries than the most comprehensive list up until now. And this extra data significantly improves the kind of statistical analysis that can be done.
Wheatley and co take full advantage of this. They say for a start that the new database reveals just how poor the International Nuclear Event Scale actually is. For that to be consistent, the Fukushima disaster would need to be rated at 10 or 11, rather than at the current maximum level of 7, they say.
The team go on to calculate that the rate of nuclear accidents costing more than $20 million has decreased steadily from the 1970s. Along the way, the rate dropped significantly after Chernobyl and now sits at 0.002 to 0.003 events per plant per year.
A significant change in the distribution occurred after the Three Mile Island accident in March 1979. The safety improvements introduced after the accident reduced the median size of accidents by a factor of 3.5.
However, the largest accidents appear to follow an entirely different statistical distribution, probably because they occur as a result of set of entirely unforeseen combinations of circumstances.
These kinds of large unexpected events are known as dragon king events and particularly difficult to analyse because they follow this different distribution, have unforeseen causes, and are few in number.
Nevertheless, Wheatley and co say their data suggests that the nuclear industry remains vulnerable dragon king events. “There is a 50% chance that a Fukushima event (or larger) occurs in the next 50 years,” they say.
Fukushima was by far the most expensive accident in history at a cost of $166 billion. That’s 60 per cent of the total cost of all other nuclear accidents added together.
The team calculate that a Chernobyl-scale event, the most severe in terms of radiation release, is as likely as not in the next 27 years. And they say a Three Mile Island event in the next 10 years has a probability of 50 percent.
That’s a brave piece of work. Anybody taking on the nuclear industry is setting themselves up for a sustained critique. But Wheatley and co are surely up to the challenge. Their database is carefully researched and their statistical pedigree hard to match.
Their conclusions will make for uncomfortable reading for the nuclear industry and its supporters. Many countries are currently investing in nuclear energy because it produces carbon-free energy.
But Wheatley and co’s work suggests that a Chernobyl-scale accident is worryingly likely to occur within the working lifetime of the reactors now being built. And when that happens, a once obscure place will enter the lexicon as a synonym for catastrophe, just like Chernobyl, Windscale and Fukushima.
These risks will have to be carefully weighed against the advantages. The question for engineers, policy makers and the general public alike is whether that risk is worth taking, given what’s at stake.
Ref: arxiv.org/abs/1504.02380 : Of Disasters and Dragon Kings: A Statistical Analysis of Nuclear Power Incidents & Accidents
A startup says it’s begun releasing particles into the atmosphere, in an effort to tweak the climate
Make Sunsets is already attempting to earn revenue for geoengineering, a move likely to provoke widespread criticism.
Radar and laser breakthroughs serve humanitarian ends
Innovations in directed-energy systems could save lives and aid disaster recovery.
This is where Tesla’s former CTO thinks battery recycling is headed
JB Straubel speaks about his company, Redwood Materials, and what challenges loom for batteries.
Why EVs won’t replace hybrid cars anytime soon
Plug-in hybrids won’t get the world to zero emissions, but they can help cut climate impacts somewhat. Toyota is betting they’ll stay in the mix for a while.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.