MARCH 19 THROUGH 20, 2011: “GIVE ME THE WORST CASE” - Fukushima: The Story of a Nuclear Disaster (2015)

Fukushima: The Story of a Nuclear Disaster (2015)



On March 19, eight days had passed since the massive tsunami surged up and swamped the Fukushima Daiichi site. Ironically, what the crippled nuclear plant needed most now was water. Tons of it.

Inside the cores of reactor Units 1, 2, and 3, water levels had dropped—the result of heat boiling away coolant and of intentional releases of steam through safety relief valves to lower pressure so that makeup water could be injected. Even though seawater was being steadily pumped into all three reactor vessels, instrument readings showed that the intensely radioactive fuel in all of them remained exposed; in Unit 3 by as much as six and a half feet, by slightly less in the other units.1 The amount of water in the spent fuel pools, especially in Unit 4, was uncertain (and a subject of continuing debate). One thing was obvious to everyone, however: water provided at a sufficiently high rate offered the best hope that operators might be able to lower the temperature within each core below its boiling point, prevent further radioactive releases, and wrestle the crippled machinery into some sort of eventual shutdown. At the least, it would buy them some time. For the past two days, Self-Defense Forces and riot police had been spraying water on and, they hoped, into the Unit 3 spent fuel pool using helicopters, high-volume water cannons, and fire engines, each capable of delivering six tons of water per hour. These were augmented by a high-pressure pumper provided by the U.S. military. As water hit the hot Unit 3 drywall head, a large steam plume shot skyward through the gaping roof, an event captured by photographers. When the NRC’s team at White Flint saw the images, they were convinced that the pool, along with the Unit 4 fuel pool, was dry. TEPCO vehemently disagreed, however, and had not yet attempted to inject water into Unit 4.

Also murky was the status of the Units 1 and 2 spent fuel pools, which like the others were perched high above the containment vessels. The concrete-and-steel roof atop the Unit 1 reactor building had collapsed into the pool during the explosion there, possibly preventing water from reaching the pool, or at least filling it with debris. But the heat load from the fuel in the Unit 1 pool was smaller than the others, so it appeared to pose a less immediate threat. Unit 2, which had a higher heat load than Unit 3, still had a roof, so it was impossible to see inside. However, just doing the math—pool volume, heat of spent fuel times days without power to provide circulation and cooling—made it clear that the Units 1 and 2 spent fuel pools would soon need cooling restored or they would also be endangered.

Japan’s Self-Defense Forces and riot police employed fire trucks and water cannons in an attempt to cool Unit 3’s spent fuel pool …

Japan’s Self-Defense Forces and riot police employed fire trucks and water cannons in an attempt to cool Unit 3’s spent fuel pool. Even after the top of the reactor building had been blown off, it was unclear how much water had reached the pool. Tokyo Electric Power Company

Adding water to the pools was also critical to reducing radiation levels around the plant site. The water would provide a vital shield against radiation from the spent fuel; however, workers still would face a threat from the radioactive debris thrown to the ground between Units 3 and 4 by the hydrogen explosions. That and the “shine” beaming from the open pools rendered much of the area off limits. A worker on a fire truck driven close to the reactors had picked up ten rem of radiation in just two minutes—double the maximum exposure that workers in U.S. nuclear plants are allowed to receive during an entire year.

A truck known as a kirin or giraffe, normally used to pump concrete at high-rise construction sites, delivers water to the Unit 4 spent fuel pool …

A truck known as a kirin or giraffe, normally used to pump concrete at high-rise construction sites, delivers water to the Unit 4 spent fuel pool. This truck was soon joined by others. Tokyo Electric Power Company

The ability to deliver water to the pools did not improve until March 22, when a giant truck with a fifty-eight-meter articulated boom arrived. It could precisely direct up to 120 cubic meters of seawater per hour (about 32,000 gallons) high into the fuel pools, a far more effective aid than the aerial drops from helicopters or the fire hoses from ground level. The truck could be driven in, parked, and then operated remotely, meaning human radiation exposure could be kept to a minimum. The single truck now in position at Fukushima Daiichi would eventually be joined by trucks from China, Germany, and the United States.

Although getting water to the reactors and pools was a high priority for the Japanese, they were also devoting considerable time and effort to restoring power to Fukushima Daiichi. Workers were slowly bringing in new lines from the outside. They finally restored off-site electricity to Unit 2 on March 20, and gradually brought power to the remaining units over the next week. However, the NRC crew had its doubts about the Japanese push to regain electricity. Connecting cables was one thing, but getting electricity flowing to equipment that had been flooded was quite another.

“We don’t see where the power thing is any solution at all,” Chuck Casto told his colleagues from Tokyo. In all likelihood, water damage had rendered some of the circuitry and instrumentation inside the reactors and control rooms unusable. (As it turned out, they were partly right. Even after the Unit 2 power panel was energized, it took another six days before crews could even turn on the lights in the main control room. But the availability of off-site power eventually enabled workers to switch from fire engines to electrically powered temporary pumps for delivering water to the reactor cores.)

A week after the accident, radiation levels were still so high that sending in workers to make the necessary connections and repairs to the installed equipment would be difficult if not impossible. No one had yet even been able to get close enough to the damaged buildings to map the areas of highest radiation, vital information if human workers were to be dispatched to the site. Readings from the fire truck indicated those radiation levels were deadly, however.

On the phone Casto sounded exhausted—and exasperated. His colleagues at White Flint knew he was running full-out. “Right now he’s basically a 24/7 individual,” Jim Wiggins explained in a briefing. “He’s getting very little sleep, and he’s holding up, but not for long.” Casto’s days—and nights—were filled with meetings, and his to-do list never seemed to get shorter. The U.S. Embassy staff was struggling to devise a contingency plan in case radiation levels rose enough to threaten Tokyo and the embassy itself had to be evacuated. Casto was being pressured for accident scenarios and dose calculations he didn’t have.

Nor did the NRC at White Flint, for Fukushima was exposing huge gaps in the agency’s ability to provide useful advice in real time for protecting people during a nuclear accident. The decades of stylized accident computer simulations were proving of little help in interpreting the events that had already happened in Japan, much less making credible predictions. As a result, everyone was scrambling to come up with high-confidence assessments of how bad the accident was and how much worse it could get.

Marty Virgilio, who was manning the overnight shift at White Flint, assured Casto that the NRC’s Protective Measures Team was working on the dose estimates. “We’re developing calculations to see what we think would be the worst case and the best case with respect to radiation levels in your neighborhood,” Virgilio explained. “That would be fantastic,” replied Casto. The team was aware that the French Embassy, concerned about the levels of radioactive iodine reported in Tokyo, had recommended that French nationals evacuate the Tokyo region. Ambassador Roos might well be wondering if the French knew something that he didn’t.

Relations between the Japanese government and the Americans sent to Tokyo to assist with the accident response were occasionally distant and strained. This clearly frustrated the NRC crew at the embassy as well as back at headquarters. “[T]he political dynamic and the … organizational dynamic, is just, you know, unfathomable,” said Casto over the phone. He chafed at the formalities of some meetings—“it’s probably one of those pretty-face things again,” he said of one coming up. These formalities, to his mind at least, ate up valuable time—and the Japanese didn’t have any to spare.

As for TEPCO, dialogue between the utility and the NRC team remained practically nonexistent. And when the NRC experts finally did get the opportunity to sit down with company representatives, the talks weren’t especially productive. “[T]hey said they didn’t need any help and everything’s in full control, under full control,” John Monninger said.

While the Japanese had never anticipated an accident as complex as the one unfolding at Fukushima, neither had the NRC. This didn’t stop the agency from engaging in backseat driving, though. Monninger took issue with what he saw as the utility’s narrow focus in dealing with an unfolding, multipronged crisis that was far from under control. “They have one priority: Unit 3,” he told White Flint. “And once they get done with that, they’ll determine the next priority.”

For the U.S. team, there was no shortage of priorities—and the list seemed to expand with each phone conversation.

Soon, the most contentious issue to confront the NRC would surface as the crisis worsened. High-level officials in Washington wanted to know one thing: how bad could this get? For the NRC, that was a question fraught with all sorts of implications. The agency had spent years downplaying the risks of nuclear accidents, contending that a real “worst case” could never happen. Now, it was being asked to assume the opposite.

The list of federal agencies working on the response to Fukushima in the United States and Japan included familiar names like the DOE, the Department of State, and the EPA. But it also included a small cadre of government entities few Americans have ever heard of, including the National Atmospheric Release Advisory Center. The center, known as NARAC, uses computer models and geographic data to map the spread of hazardous materials that get released into the atmosphere. Its predictions are intended to help decision makers in an emergency, and its models were capable of predicting the movement of radiation plumes with more accuracy and to considerably further distances than the NRC’s RASCAL code.

At the moment, experts across the U.S. government were grappling with these questions: just what could be the worst-case scenario, and how much radiation might escape from Fukushima Daiichi if that scenario occurred? The answer had ramifications not only for Japan and the American citizens there but potentially for the United States itself. To make those predictions, the experts first had to agree on what’s known as a source term, an assessment of how much radioactive material could actually be released from the reactors and fuel pools into the environment. The types of material and the timing of the releases were also key inputs for predicting where the radioactive plumes would travel and the nature of the damage they could cause. The bottom line issue for the U.S. government was this: what threats would such a release pose to Americans, whether at home or abroad, and what measures had to be taken to protect them? Without a more precise understanding of the source term, the answer to that question would remain frustratingly elusive.


Source term defines the types and amounts of radioactive material released during a nuclear plant accident. The source term depends on many variables, including the initial amount of radioactive material in the nuclear fuel, how much of that radioactivity gets released from damaged fuel, how much of that radioactivity is retained within the plant, and how much is released to the environment, where it can be transported to downwind communities.

Highly radioactive materials like iodine-131 and cesium-137, known as fission products, are by-products of the nuclear chain reaction that drives the nuclear engine. An operating nuclear reactor core contains a mixture of dozens of different radioactive isotopes. The quantities of these isotopes depend on the power level of the reactor and the length of time that the reactor has operated, among other factors.

Radioactive materials are unstable and release radiation seeking to reach a stable form. The time it takes for one-half of a given quantity of radioactive material to decay—called the half-life—varies from a mere fraction of a second to millions of years. In a reactor accident, the longer the onset of reactor core damage is delayed after shutdown, the more time is available for radioactive materials with short half-lives to decay into stable materials, thereby reducing the source term.

Different isotopes have different radioactive properties, which determine the relative hazards they pose to the environment and to humans and other organisms. Certain isotopes, such as plutonium-239, emit alpha particles, which cannot penetrate skin but are particularly hazardous inside the body; thus alpha emitters are very dangerous if inhaled or ingested. Others, like strontium-90, emit beta particles, which are somewhat more penetrating than alpha particles but still do more of their damage if emitted within the body. In contrast, high-energy gamma rays penetrate deeply, and hence gamma emitters like cesium-137 can do their damage from outside the body.

Different radioactive isotopes also have different chemical forms, which determine how they behave within the reactor, in the environment, and in living things. For example, beta-emitting iodine-131 becomes a gas at ambient temperature, can be transported large distances in the environment, and if inhaled or ingested concentrates in the thyroid gland, where it can deliver a high dose. Luckily, though, with a relatively short half-life of eight days, it does not persist in the environment. In contrast, plutonium-239 remains solid up to very high temperatures and is not easy to disperse, but with a 24,000-year half-life and a tendency to deposit in liver and bone, it is very persistent in the environment or in the human body.

These properties determine the relative hazards of the different stages of a radiation release. In the early stages of the accident, people can be immersed in and inhale airborne plumes. In the later stages of the accident, after the plume has passed, people can be exposed to contamination on the ground and other surfaces and in food and water. Contaminated dust particles can also be carried upward by air currents and inhaled.

When an accident causes the fuel rod claddings to overheat and break but is stopped before the fuel pellets are damaged, the release is limited to the radioactive gases trapped in the space between the pellets and cladding, including noble gases, iodine-131, and cesium-137. If overheating is not stopped in time, the fuel pellets themselves can break apart, allowing some of the radioactive materials trapped within the pellets to be released along with the radioactive gases. Ultimately, the pellets themselves can melt, and release an even wider range of isotopes, including plutonium-239 and americium-241. The greater the fraction of the core damaged, the greater the amount of material available for release.

A functioning reactor containment is designed to allow no more than a small fraction of the airborne radioactivity within the reactor—less than 1 percent—from being released. If containment venting is necessary, filters could be used to prevent radioactive materials from escaping into the environment. However, if the containment barrier is breached or bypassed, a far greater amount of radioactivity can escape.

Even if the containment fails, determining the actual amount that may be released over time is difficult because radioactive gases and particles released from damaged fuel may cool down and “plate out”—that is, stick to—various surfaces and be retained within the plant. But even these materials may eventually heat up again and escape.

Once the source term is developed, the weather conditions are defined and populations at risk are identified. Based on this information, the hazard from radioactive materials escaping from a nuclear plant can be estimated. But other variables also play a factor. Wind may blow the radioactive cloud toward or away from densely populated areas. A city forty miles away from a damaged plant may be at greater risk than closer cities if it starts to rain as the plume passes overhead.

Although the source term is used in estimating the risks, the actual hazard depends on questions that cannot be definitively answered during an accident or even long afterward. When was the fuel damaged? How much fuel has been damaged? To what extent has the fuel been damaged? How much radioactive material released from damaged fuel has been retained in the plant? Artful science is applied to estimate the source term based on the most likely answers to these important questions.

President Obama, in an address to the nation on March 17, offered reassurances. “We do not expect harmful levels of radiation to reach the United States, whether it’s the West Coast, Hawaii, Alaska, or U.S. territories in the Pacific,” he said. “That is the judgment of our Nuclear Regulatory Commission and many other experts.” But behind the scenes, those experts were still debating the numbers, even as conditions inside the reactors and fuel pools were unclear.

Radiation monitors at California’s Diablo Canyon and San Onofre nuclear plants had already picked up readings of iodine-131 just slightly above what the NRC described as “the minimal detectable activity level.” It presumably was blowing in from Fukushima, 5,400 miles away. If these levels continued to increase, there was a chance that the president would later have to reverse himself and order countermeasures like banning milk shipments from certain areas. That could result in a major loss of confidence among the public.

Among the scientists and administrators, disparate views on the radiation threat abounded, apparently hashed out in the confidentiality of the White House Situation Room, where officials from a host of agencies gathered. Discussions also were going on at the U.S. Embassy in Tokyo. There had been internal disagreements in both places over the NRC’s recommendation two days earlier for the fifty-mile evacuation for U.S. citizens in Japan. Now, based on aerial measurements between thirteen and twenty miles northwest of the reactors showing exposure rates above one rem over four days, the EPA evacuation standard, the NRC was confident it had made the right call. But the agency continued to take heat for the decision. On March 18, the Nuclear Energy Institute contacted the NRC to complain that the fifty-mile evacuation could undercut the public’s faith in this country’s ten-mile emergency planning zone.

Back at White Flint, Trish Holahan, director of security operations at the NRC’s Office of Nuclear Security and Incident Response, and her colleagues were poring over some alarming dose estimates for U.S. territory that they had received from NARAC. Because the NRC’s own RASCAL model had limited range, the agency had to rely on NARAC to go farther using its more sophisticated plume transport models. Employing source terms supplied by the NRC, NARAC’s models were finding that thyroid doses to one-year-old children in Alaska could be as high as thirty-five rem—seven times the EPA dose threshold that would trigger the need for countermeasures such as potassium iodide administration.

But the NRC thought that some of NARAC’s assumptions seemed a bit far-fetched. The potential radiation exposure of grazing dairy cows in Alaska in mid-March was one example. “The cows are kept indoors,” Holahan told Jaczko over the phone. “Even the water supply is internal because they’re not outside. So we eliminated that dose.”

The assumption of grazing cattle in Alaska in wintertime was just one aspect of a larger issue that the NRC was trying to grapple with: the White House request for a “worst-case” assessment. From the perspective of the president, this approach made sense: if even under the most pessimistic assumptions there was little risk to the American public from Fukushima, then the president could continue to provide reassurances without fear that he would be later accused of underestimating the threat.

Joining the White House in pushing for the worst forecast was Admiral Mike Mullen, chairman of the Joint Chiefs of Staff. “I have been taught by my nuclear power community my whole life to plan around the worst case possibilities,” he wrote to Obama’s top science advisor. “This in great part had a lot to do in keeping our [Navy] plants safe.”

However, the concept of a worst-case scenario was anathema to the NRC’s way of thinking. For decades, NRC regulations and policies had been explicitly designed to avoid accounting for worst-case scenarios, which were believed to be so unlikely as not to merit consideration. Calculating a worst-case source term fell into that same category.

The NRC had already ventured outside its comfort zone when it made the recommendation for a fifty-mile evacuation around Fukushima, based on source-term assumptions that it judged very unlikely. Now it was asked to consider even more extreme cases.

The NRC instead preferred to focus on what it considered more realistic or “best estimate” scenarios. The NRC had just spent several years on a research project called State-of-the-Art Reactor Consequence Analyses meant to calm public fears about nuclear power by calculating “realistic” severe accident source terms. Those numbers were lower than previous estimates, and the NRC was pleased with the initial findings. But now, when a reasoned perspective was needed more than ever—at least in the eyes of the NRC—the White House appeared to be asking the NRC to throw its approach out the window.

The White House wanted the NRC to provide a source term to NARAC that assumed that 100 percent of the fuel in the cores of Units 1, 2, and 3 and in the spent fuel pools of Units 1, 2, 3, and 4 had melted, with the Units 1-3 primary containments failing completely so that the resulting radiation from all seven sources would be released into the environment.

Jaczko was frustrated by this request. He questioned whether the NRC was being asked to hypothesize an accident that was in fact impossible—one that would essentially vaporize all the cores and spent fuel and eject it all into the environment. “[O]bviously, there’s a physical reality at some point that certain things just cannot happen,” Jaczko said. As bad as this accident was, there were no plausible physical mechanisms that could vaporize the entire core of a light-water reactor like those at Fukushima. Even at Chernobyl, a more unstable type of reactor that experienced a runaway chain reaction and massive steam explosions, most of the core material remained within the reactor.

“[T]here’s what’s worst case and then there’s what’s possible,” Jaczko told Holahan and the team at White Flint. “So I think what we should produce [is] a worst-case [scenario] that [is] actually possible.” Instead, he said, the NRC was being asked to envision the nuclear equivalent of a “meteor hitting the earth at the same time as an asteroid strikes.” Holahan agreed to go back and confer again with the reactor safety experts and come up with some new analyses.

Earlier that afternoon, Jaczko had spent forty-five minutes with the Japanese ambassador, making a combined condolence and business call at the embassy on Massachusetts Avenue. The NRC crew hoped the visit might facilitate communications between Washington and Tokyo. “We have a tremendous opportunity here now,” Jaczko had told his team before heading to the meeting. “We have an ambassador who basically wants to be helpful and can pass information to help move a logjam if necessary.”

Chuck Casto, back from a meeting with the chairman of TEPCO and the utility’s chief nuclear officer, called his colleagues at White Flint at about 9:00 p.m. on March 18 to give them an update. “It was a cordial meeting,” he said. In contrast to the few previous encounters, this time TEPCO executives asked for help from the Americans.

Although the NRC was still focused on finding a way to cool the spent fuel pools, TEPCO was more worried about the reactors themselves—in particular, the accumulation of salt inside them. As the fire trucks kept pumping in seawater, large amounts of salt were building up in the bottoms of the reactor vessels, potentially blocking the flow of the water and interfering with efforts to cool down the fuel. In addition, seawater is more corrosive to metal than the freshwater the reactors were built to use. Forced to use seawater at Fukushima, the Japanese wanted to consult about the problems it would create for them down the road. They were now asking for assistance.

Jaczko was listening in. “Do you think that project team should be NRC people or somebody else, like maybe INPO or something like that?” Casto agreed that both the NRC and the Institute of Nuclear Power Operations ought to get involved, along with the DOE, which has expertise in radiation doses and decontamination.

“I think this is a major change in the mission,” Casto told Jaczko. “Sir, I believe that they are looking to us for solutions… . I think they were … desperate for options.”

Jaczko’s reference to a heightened industry role in the U.S. response to the accident was not hypothetical; the details were being finalized even as the men spoke. The expert industry group was set to gather at NRC headquarters. The weary White Flint team hoped the fresh troops might shift some of the load off the NRC’s shoulders; with luck, the involvement of industry might even encourage TEPCO and Japanese officials to be more forthcoming with information and to accept more help.

Participants with special expertise, such as GE, which had designed the Fukushima Daiichi reactors, would be asked to provide guidance on getting water into the fuel pools, for example. Other industry representatives would be invited to help devise short- and long-term solutions in Japan.

It was not going to be simple. For its part, the NRC planned to work through diplomatic channels in Japan to come up with a cooperative agreement that was politically acceptable to Tokyo, and also to try to obtain better information about the conditions at Fukushima Daiichi and the radiation levels.

But better data would only improve things somewhat; equally important was a definitive plan of action from the Japanese. “[T]hey have shifted almost as frequently as the winds have changed there with respect to what their priorities are,” Virgilio observed. “[I]t’s not a good situation.”

“I’m just trying to figure out who the power player is over here,” Chuck Casto said. For the industry consortium to succeed, it had to be consorting with the real decision makers in Japan, and Casto wasn’t sure who those were. “Is TEPCO the right organization, or should we be going to MOD [Ministry of Defense], or who?”

It was shortly after midnight March 19 at White Flint, and Casto had just left a meeting with Ambassador Roos. Roos knew he might soon face a crucial decision: if the situation at Fukushima Daiichi began to deteriorate, should he be ready to order a much wider evacuation of U.S. citizens and close the embassy? Like the White House, he was demanding better data on a worst-case scenario—information that would allow him to decide if it was safe to stay no matter how bad things got. The DOE had given Roos the results of an accident scenario that Casto had not reviewed. The NARAC models, using a source term known as the super core, were indicating that doses at the embassy could exceed one rem over four days, which would require evacuation under the EPA’s guidelines. When Roos asked Casto about the numbers, Casto later told Virgilio, he felt “blindsided.” Even the American experts advising the ambassador weren’t coordinating their efforts.

In response to the previous White House request, the NRC—despite its reservations—was already developing what it considered a worst-case scenario involving Units 1 through 4. (It was also analyzing a more “realistic” scenario to have on reserve.) But now the ambassador apparently wanted projections for an even more extreme scenario in which Fukushima Daiichi Units 5 and 6 also melted down. For Virgilio, this demand went too far, given that Units 5 and 6 were relatively stable. He also presumed that the results would indicate the need to evacuate Tokyo. “[T]hat’s why we’re trying to do a worst-case that really makes sense given the conditions that we have now,” he said. “I mean the team spent half the night last night trying to figure out where do we start from.”

“Well, and I appreciate that work,” Casto said. “I expect he’ll turn to DOE and say, give me the worst case.”

“You know the old adage,” warned Virgilio, “ ‘Be careful what you ask for.’ ”

“I think they’re trying to get something out to [embassy] employees to show that it’s safe … no matter what happens—even if the extreme happens—it’s safe to be in Tokyo,” Casto said.

Finding the right answer—or at least a best guess—for the ambassador became even more urgent. The prevailing winds were about to start blowing toward Tokyo and continue in that direction for twelve hours.

But this was not an easy task. The debate among the NRC, the DOE, and the White House Office of Science and Technology Policy over whether to evaluate worst-case scenarios and in fact, what exactly was the worst-case scenario continued to percolate over the next several days, and was a source of considerable friction among the agencies. Although the NRC’s emergency planning strategy included the ability to estimate source terms based on stylized cases, the urgent need for the NRC to produce something that accurately modeled a real-world event put the agency in a bind. Nobody ever figured they’d have to do it in real time for an accident as complicated as Fukushima.

At one point on March 20, the NRC’s Jim Wiggins said, in apparent frustration, “I still won’t let anyone use the word ‘worst case’ in the room here … because there’s about five worst cases.”

The White House and the DOE were griping about the NRC’s performance as well. For instance, they found mistakes in the source terms that the NRC had been using. When the NRC modeled the Unit 4 spent fuel pool, it assumed that a fairly large amount of iodine-131 had been released. But the fuel most recently discharged into that pool had aged over three months. In that time, most of the iodine-131 should have decayed away. It turned out that the NRC had received erroneous information leading the staff to believe that the most recent batch of fuel had been discharged from the reactor only a month before the accident; thus, the NRC’s model greatly overestimated the iodine release and resulting radiation doses to the thyroid that would be associated with a Unit 4 fire. But before this mistake was discovered, NARAC had used the source term to calculate trans-Pacific doses, finding some alarmingly high results—four rem (forty millisievert) thyroid doses to one-year-old children in California over a two-month period, for example—although these were still below the thresholds for protective action.

Interagency rivalries only compounded the difficulties. Early on, the NRC team at White Flint complained about interference from the DOE’s experts at the national laboratories. The DOE later fired back with its own critique. A postaccident evaluation, prepared by two experts from Sandia National Laboratories, criticized the NRC, saying the agency “did not seem to engage aggressively until four or five days into the event.” In addition, NRC emergency planning personnel were “very reluctant” to engage with their own colleagues on the research staff who had previously done analyses of events very similar to those unfolding at Fukushima, according to the Sandia study. Other Sandia experts also reportedly dismissed RASCAL as a “toy model” that should not have been used to study real-world events.

At 8:00 p.m. on March 19, Brian Sheron led a conference call updating other NRC staff members on the day’s events. The afternoon meeting with the industry consortium had gone well, he reported, running beyond the ninety minutes originally planned. Industry people seemed receptive to working closely with TEPCO, offering suggestions or support. The DOE would use its NARAC resources to project dose rates in Tokyo based on the predicted wind change toward the city, but at this point there appeared no reason to alter the fifty-mile evacuation zone for Americans in Japan.

Less than an hour later, Chuck Casto was calling. “Here’s today’s crisis,” he said to Virgilio and Sheron. After resisting requests for information and offers of help for days, now TEPCO was accusing the United States of dragging its feet.

Ambassador Roos, his staff, and Casto had just returned from a meeting with the utility. The TEPCO officials wanted to hear the NRC’s assessment on radiation levels and the salt accumulation in the reactors—something TEPCO had asked for just a day earlier. “Well, honestly, I didn’t have a wallet in my back pocket on that,” Casto told his colleagues. “I said, ‘Well, you asked me about it yesterday. There’s a lot of information, a lot of analysis, and I believe we’re working on that.’ ” TEPCO was insistent. “They basically said, we need this stuff immediately.”

Now it was time for the NRC to step back a little, Casto told Virgilio. The Defense Department was willing to bring in whatever heavy equipment was needed. “[W]e’re not working the logistics stuff,” said Casto. “It’s out of our lane.” And the new industry consortium could work more closely with TEPCO.

As a trade-off for the NRC’s ongoing technical assistance, the Japanese should be expected to hand over data, Casto said. “[W]e don’t know the condition of the reactors … , what containment pressure is, what reactor pressure is, whether those things are even full. Nevertheless, all that’s moot. The bottom line is get water. They need to get freshwater into that reactor.”2

One way to start getting freshwater for the reactors would be to get desalination equipment up and running, but this would have to wait until power was restored to the site. The team back at White Flint promised to review research papers on the salt issue and get back to Casto and Monninger, who was also on the line from Tokyo.

“I’m glad you brought that up because let me make it clear,” Monninger said. “We really need you guys to be the brain waves and give recommendations. [H]ere we can’t really read stuff and come up with thoughts and recommendations and that kind of stuff. We want to be the, you know, the grease.” The Tokyo group was too preoccupied by the never-ending stream of crises, and perhaps just too beat, to spend precious time scrutinizing scientific research. Earlier, Monninger had told his colleagues he yearned for “100 hours of sleep.”

At 10:00 a.m. Sunday, March 20, industry representatives joined in a conference call with the NRC operations team at White Flint to brainstorm. The industry consortium, working out of the Marriott across the street from NRC headquarters, had agreed to send two of its technical people to Tokyo that evening or the next day to work with Chuck Casto, and then to embed them at TEPCO’s emergency operations center. That, everyone hoped, might improve communications.

Almost from the outset of the accident, the Japanese nuclear industry had reached out to the U.S. nuclear industry for help, leaving the NRC as a bystander. On March 12, field representatives in Japan for GE-Hitachi, a U.S.-Japanese nuclear partnership, had contacted Exelon, the Chicago-based utility, and asked Exelon to run accident simulations for the reactors in its fleet that were of the same design and vintage as those at Fukushima Daiichi. (However, this information did not prove very useful—Exelon’s attempt to model what was going on at Unit 1 on the simulator at its Quad Cities plant predicted a primary containment pressure that was only 3 percent of what was being reported.)

And the Nuclear Energy Institute appeared to have access to valuable information the NRC didn’t have. The NEI had dose rates the commission had been seeking, for example. (The NRC’s information often came from unexpected places. Details about radiation levels in spinach came from the Wall Street Journal. “It’s amazing how people know this stuff and we can’t seem to get it,” marveled a member of the White Flint crew.) Maybe things would improve with industry bridging the gap.

The status of the Unit 4 spent fuel pool still worried the NRC team. Although on March 20 the Japanese had finally begun using fire engines and then pumper trucks to spray tons of water toward the Unit 4 pool, as they had been doing at Unit 3 for several days, workers were shooting the water from such a distance that “you have incredible losses,” Monninger told his colleagues. (The arrival of the kirin trucks was still two days away.) “[T]he media [are reporting] that these fire trucks are going in and out, the helicopters are doing this, the super capacity pumping system. But then, when you actually [get] down into TEPCO and start talking to the engineers, you find out that it really isn’t that effective.” If it had been, radiation levels would have dropped, but, when pushed on the subject, the Japanese reported no change.

The NRC remained convinced the pool was dry—a view at odds with the Japanese belief. “We’ve got to be very careful with that because we got in trouble before by passing up that information,” warned Monninger.

If the pool were in fact dry, the threat it posed was enormous. The intensely hot fuel could now be melting into the concrete pool floor. And sitting below that was the torus, filled with more than a million gallons of water. If the molten fuel reached the torus, it could vaporize that water almost instantly, causing a powerful steam explosion that could propel radioactive core material far and wide. (However, as with the NRC’s earlier belief that the pool was empty, fears about the molten fuel also proved unfounded. A computer simulation the next day indicated the fuel would not be hot enough to melt through the pool floor.)

On another topic, Monninger related that after days of seeking an invitation, he and Jim Trapp had finally made it to the TEPCO emergency operations center. Its scale surprised them. “This place was massive,” he said. “There’s probably 250, 300 people in that room.” In addition to the sheer size of the TEPCO operation, the visit to the utility’s operations center was notable for another thing, Monninger told his colleagues: “There’s huge [numbers of] protesters, cameras, cops surrounding the TEPCO facility.” Official visitors were now whisked in via an underground garage.

Assuming the spent fuel in the Unit 4 pool was still covered with water, it would require seventy-two tons more water every day to cover losses from evaporation. The Japanese were aiming at least that much at it. The Unit 2 pool was targeted for twenty tons a day; Unit 3, ten tons a day; and Unit 1, five tons a day. But no one knew how much of this water was reaching the pools and how much was missing them and flowing elsewhere in the reactor buildings.

“One of the concerns is they turn the site into a swamp,” said Monninger, “but the other is just the contamination and the runoff from all this water that’s not going into the spent fuel pool.”


Water propelled by tsunamis flooded the Fukushima Daiichi site, disabling the power supplies for nearly all of its safety equipment.

Lack of water damaged the Units 1, 2, and 3 reactor cores and threatened to damage fuel in the Units 1, 2, 3, and 4 spent fuel pools—if that had not already happened.

Lacking freshwater sources, workers tried to address those problems by injecting seawater into the reactor cores. But seawater corroded reactor parts more rapidly and left salt behind that might eventually block the cooling water flow.

Water dumped from helicopters and sprayed from fire trucks was intended to refill the pools and protect their spent fuel from damage.

But water that missed the spent fuel pools could end up draining down into the reactor buildings below and flooding their basements. Workers struggling to repower the Fukushima site might find their efforts thwarted if flooding submerged the safety components on the other ends of the re-run power lines.

Ironically, water took away most of the options available to responders and left them none without potentially catastrophic consequences.

The lack of information on the status of the spent fuel pools was just one of many factors causing angst and blocking interagency consensus on a plausible source term. While the NRC had thought that it finally gave the DOE and NARAC the worst-case source term they were seeking, NARAC was continuing to ask the NRC for more information. At 8:30 the next morning, the White House was convening a meeting to bring together the warring parties. The goal of that meeting was to reach a public policy decision based on science, but in the case of source terms, the science was anything but solid. There were just too many variables, to say nothing of biases.

Getting the source term right was far more than an academic exercise. The safety of American civilians and military personnel in Japan and potentially millions more in Pacific island territories, Alaska, Hawaii, and even the continental United States was involved. Was the threat real enough to distribute potassium iodide tablets to reduce the effects of radiation exposure to the thyroid? Underestimating the hazards could leave many people in harm’s way. Overestimating them could result in the unnecessary movement of large numbers of people—which itself could result in casualties.

Now, it came down to whether NARAC should use a “realistic” worst case or a “worst” worst case. But who could say, given all the extreme events that had already occurred, what was truly “realistic” at this point?

To break the impasse, Jaczko’s office wanted the NRC to send to the meeting not only a technical expert but also someone with senior-level credentials and experience to go “nose to nose” with Dr. Steven Aoki, deputy undersecretary at the DOE, home to NARAC, to press the NRC’s position. Charlie Miller, a veteran department chief, was recruited for the task. A successful outcome, according to Virgilio, would be an “agreement high enough up that my folks wouldn’t continue to bang their heads against the telephone back and forth with folks at our level about what assumptions are, and they would actually do some calculations for us.”

This was likely easier said than done, however. As Mike Weber, the NRC’s deputy executive director, told Miller, “probably what you’re going to find out as each party weighs in is everybody has a different definition of worst case as their own … so we’ve got to come to the common agreement to go forward.”

And impatience with the lack of a definitive answer was growing among a host of federal agencies—all with a stake in the decision. As Marty Virgilio put it, “DOD wants to know where to move their ships. EPA and others want to know what to expect on the West Coast. HHS [Health and Human Services] wants to know what kind of levels in order to make recommendations on whether or not they should actually recommend potassium iodide [tablets] at some point. And it goes sort of on and on.”

Much of the give-and-take at that Monday morning meeting at the White House remains secret. But at the heart of the discussions was finding a way to make the best judgment call on radiation risks in the face of such uncertainty. Were conditions dire enough to warrant an evacuation order for U.S. personnel in Tokyo? Some of the data in White House hands indicated it might be necessary. At the other extreme, should the fifty-mile evacuation advisory be lifted, as some were urging? (Ultimately, the fifty-mile evacuation zone remained in place until October 2011, when the State Department reduced it to twelve miles.)

Even as Miller headed to the White House, the White Flint team was receiving important new information that could dramatically alter source term assumptions. The Unit 4 spent fuel pool, although heavily damaged, apparently had water, reducing the risk there somewhat. “Do we have any idea how we got it in there?” someone asked. The answer was no; nor did anyone know how much water the pool contained.

The NRC elected not to try to change the source term numbers at this point. “It took two days to negotiate this source term,” said a member of the NRC crew. “I don’t know if we want to spend another two days trying to negotiate another one.”

Finally, an agreement was reached among the White House science advisors, the DOE, and the NRC on a scenario the NRC pointedly referred to as “the President’s source term.” This case assumed releases from three reactors and four spent fuel pools, but used “best estimate” simulations for the amount of radioactive material released from each source based on computer models of the accident using the NRC’s own computer code, known as MELCOR—introducing what the agency believed was “realism” into the analysis.3

NARAC’s results for the “President’s source term” turned up one disturbing finding: a potential thyroid radiation dose to a one-year-old child on the West Coast of the United States of 4.5 rem, not far below the 5-rem EPA threshold for protective actions such as potassium iodide administration or interdiction of milk supplies. So the fears of some Americans that Fukushima could impact them were perhaps not as far-fetched as the U.S. government had led them to believe. However, because this result was below the threshold, it did the trick: even in the worst case, nothing needed to be done to protect the children in California.

NARAC also decided to use this source term to evaluate the potential doses in Japan as well, and that produced an alarming result. “You’ve got to evacuate [Tokyo] and everything else,” reported the NRC’s Jim Dyer.

This prompted the crew at White Flint to complain once again about the way the other agencies were continuing to engage in the source term exercise. “[W]e ought to just have realistic models, not these ultraconservative worst-case things,” said Bill Borchardt.

But even the “realistic” models weren’t simplifying matters. NARAC was also running the NRC’s “plausible realistic” scenarios, which assumed far less containment damage than the NRC’s MELCOR models and no releases from any of the spent fuel pools. According to these results, Japan’s twenty-kilometer evacuation zone was not too small, but rather larger than it needed to be. Then came another surprise: the NRC soon discovered that it had made yet another big mistake. Its “plausible” source term was too low, even for the “realistic” case. Although the NRC tried to cover its tracks by coming up with a post hoc justification for the error, its bungling of the math did not help the commission’s standing in the interagency debate.

Data would eventually show that the actual source term was greater than the NRC’s “plausible realistic” scenarios but far less than the more extreme cases evaluated by the White House and the other agencies. Radioactive iodine concentrations on the West Coast of the United States never reached the levels predicted by the MELCOR source term. Tokyo at large was never imperiled except for a few hot spots, presumably created by unlikely but unfavorable local weather conditions.

However, the dose rate data did support an evacuation zone of about thirty to forty miles (fifty to sixty-seven kilometers) from Fukushima Daiichi, still a much larger distance than the twelve-mile (twenty-kilometer) zone initially established by the Japanese government or the ten-mile emergency planning zone in existence in the United States for reactor accidents.

For the NRC, Fukushima Daiichi redefined “realistic”—something the agency had stubbornly resisted for decades. Its reluctance to seriously consider the likelihood of a severe accident with a large radiological release, even for planning purposes, reflected the commission’s propensity to view accident risks and consequences through rose-colored glasses.

Up until March 2011, for example, the NRC firmly believed that no realistic accident at a U.S. reactor could be serious enough to require more than a ten-mile emergency evacuation. The NRC had adopted that modest safety standard after an earlier reactor emergency provided the nation’s first reality check: the 1979 accident at Three Mile Island.

Since then, the ten-mile zone had remained inviolate. In the NRC’s mind, an accident like Three Mile Island—in which some fuel melted but the containment held, limiting the release of radioactivity—set the limit for the worst accident that needed to be rigorously prepared for at a U.S. nuclear plant. Now, as a result of Fukushima, the realism of this and other assumptions on safety would be severely tested.