A game-changer or the price of doing business?
At first there was the shock — the unbelievable devastation wrought first by Japan’s largest recorded earthquake (recently upgraded to a magnitude of 9.0) and then by the tsunami. Then, as the focus shifted to the daunting cleanup and recovery, another crisis hit — this one at the Daiichi plant in Fukushima prefecture about 50 miles from the epicenter. Reports of partial core meltdowns, radioactive releases, explosions, exposure of spent fuel, and, most serious of all, damage to a containment vessel contribute to a growing sense that events are spinning out of control.
In the United States and Europe, Japan’s nuclear crisis has heightened the debate over the place nuclear energy should play in our energy mix, with some using recent events to argue against it (or at least to call for a reassessment) and others maintaining that events in Japan are an anomaly and nuclear power is “safe.” (More on this ongoing discussion here, , here, here, here and here.)
What Is Safe?
It’s almost as if people expect nuclear power plants can or should be operated accident-free. Fact: there is no way to design and operate a nuclear power plant that guarantees zero accidents. Minimizing risk and the probability of an accident, yes, but implicit in this approach is an acceptance: accidents will happen.
Here’s a brief summary of how most of the world manages nuclear safety. It’s taken from an unpublished report I wrote with Mitch Golden, a former professor of physics.
“The basis for assuring the safety of nuclear power plants … has been a design and operational philosophy known as ‘defense-in-depth.’ This calls for multiple layers of redundancy in safety systems and practices so that a failure in any one or even several components of the system does not lead to an accident. However, such a philosophy by itself does not stipulate the specific level of redundancy needed to assure an acceptable margin of safety.”
To quantify the level of safety attained at a given facility, plant operators carry out a probabilistic risk or safety assessment. This involves estimating probabilities for various equipment failures, human errors, and natural disasters (including attempts to assess earthquakes), and simulating different possible sequences of events that could result in an accident such as core damage, a large release of radioactivity onto an unprepared population, and/or fatalities. By essentially summing up the probabilities of accidents from each possible cause, one arrives at an overall probability or frequency of failure and/or death.
Most of the nuclear reactors in the world (about 420 out of 442) are so-called Generation II reactors. These include the reactors at Three Mile Island in Pennsylvania and the Fukushima plant. According to a 2004 International Atomic Energy Agency report, these plants have an average probability of a core accident of one in ten thousand per reactor-year*.
That means a single reactor would, on average, experience an accident every ten thousand years — not much worry there. But we don’t have just one reactor, we have 420. The time for an accident at one of those 420 can be estimated by multiplying the number of Generation II reactors (420) by the frequency of core damage (10-4) and dividing it into one:
Years for an accident at one plant ~ 1/(420 x 10-4) = ~ 25 years.
So let’s see, Three Mile Island occurred 30 years ago.** And now Japan … right about on time.
That’s for an event with core damage. More serious accidents with large releases of radioactivity and/or significant fatalities are less probable — perhaps by a factor of 10 or more — but still within the spectrum of possible events.
Bottom line: the nuclear accident in Japan should not come as a total surprise. Nuclear accidents are the price of doing nuclear business.
Does that mean we close all the nukes and never build another one? That depends on how you assess risk and danger. For example, the chances of dying in a nuclear accident are far smaller than those from perishing in a car accident, but who’d suggest we do away with cars. Pollution from coal-fired power plants kills far more people than nuclear accidents. And experts say that the next-generation power plants currently on the drawing board should see at minimum a factor-of-10 improvement in safety.
Which is not to say that opposition to nuclear power on safety grounds is not irrational. “The threats from a nuclear accident,” Golden and I wrote, “are fundamentally different from many other dangers. An accident could devastate entire communities, not just random individuals — a singularly painful prospect for society as a whole. Moreover, a nuclear accident threatens the health and welfare of future generations as well as the present.”
The unfolding nuclear crisis is first and foremost about the people of Japan, and our thoughts and hopes are with them. It is also a reminder we’re working with some very dangerous stuff and that will always involve risk.
* The total number of reactor-years is the total number of reactors in operation times the number of years each reactor has been in operation. So if there were two reactors, one in operation for 10 years and another in operation for 30 years, the number of reactor-years would be 40.
** The 1986 Chernobyl disaster in Russia involved a different reactor design and a different probability of an accident, so should not be included here.