Last year, on March 11, a deadly earthquake and tsunami rocked Japan, killing more than 15,000 people. To make matters worse, the natural disaster triggered a major crisis at the Fukushima Daiichi nuclear power station. The subsequent meltdown and radioactive release is the only event in history other than Chernobyl to rate as a “major accident” on an international scale of nuclear severity. Fukushima is not expected to cause as many deaths as Chernobyl, but contamination from the accident is widespread and will be long-lived. One year after the nuclear crisis began, an exclusion zone 20 kilometers in radius remains in place around the reactors.
Could a Fukushima-scale nuclear incident happen in the U.S.? “There’s been a lot of debate on this issue,” physicist Edwin Lyman of the Union of Concerned Scientists said last week at a meeting of the American Physical Society in Boston. “In our view, complacency is as prevalent here as it is in Japan.” (Lyman and a colleague recently released a report [pdf] on the U.S. response to the accident.) One major threat to a nuclear plant is a prolonged power outage, or station blackout, like the one at Fukushima, which deprived the reactors of their cooling systems.
U.S. nuclear plants, Lyman said, are not well prepared to handle severe, "beyond design basis" events, such as major natural disasters, multiple system failures or terrorist attacks. A report last year by the Nuclear Regulatory Commission (NRC), which itself has been criticized for being too lenient with the industry it is supposed to oversee, revealed that many U.S. nuclear plants were vulnerable to extreme emergencies. “Regulators don’t usually impose stringent requirements to deal with these accidents, because they assume that they’re so improbable,” Lyman said. Indeed, the NRC has called the Fukushima crisis “the result of a combination of highly unlikely natural disasters.” That specific combination of mega-earthquake and tsunami, the agency maintains, would be very improbable in the U.S.
As an example of insufficient preparedness, Lyman cited the 2011 NRC inspection of the Edwin I. Hatch Nuclear Plant in Georgia. The NRC noted that the Hatch facility had procedures in place for only a one-hour loss of external power. “It didn’t really look to me that those preparations were something you could count on in the event of a Fukushima-type event,” Lyman said of the Hatch evaluation. After an hour, the Hatch plant’s operators assumed that backup diesel generators would be up and running. But at Fukushima the same cataclysm that knocked out normal electrical power also killed the generators. “Hatch’s procedures do not provide specific guidance for a prolonged loss of normal or alternate AC power, which is outside of the plant design basis,” the NRC reported. In other words, if the plant lost power and was unable to fire up the generators, the operators would quickly be forced into improvisation.
The NRC only requires that plants such as Hatch cope with a loss of power for four hours. (The agency has recently proposed extending that requirement to eight hours.) But at Fukushima, the station blackout lasted not just hours but days. The outage cutoff the reactors’ cooling pumps and caused the catastrophic overheating of nuclear fuel.
Nuclear power plants do keep batteries on-site in the case of a station blackout, but those last only hours. And from there, the situation deteriorates quickly, according to U.S. simulations of how a stricken plant would deteriorate in a prolonged accident. “After batteries fail, you’re going to get core melt after only about eight or 10 additional hours,” Lyman said.
But at Fukushima, a backup cooling system in one of the reactors worked far longer than that. A reactor core isolation cooling (RCIC) system, which runs on steam but requires battery power to control its valves, lasted for three days of the station blackout. “How did RCIC operate for three days? This is still a mystery,” Lyman said. “There’s no explanation for this phenomenon.” And whereas a cooling system outlasting its expected functional lifetime would seem to be a good thing, Lyman used the example to illustrate how little we can predict about what will happen in a nuclear crisis. “I don’t have confidence in the ability of current computer models to simulate a severe accident,” he said.