This is Barry Quigley’s nightmare: It’s 95 degrees in Chicago and everyone with air-conditioning has it cranked to high. The Byron nuclear power station south of Rockford is running at full steam, pumping out 2,500 megawatts of electricity to the midwest grid. Quigley, a senior reactor operator, is at the helm in the plant’s control room, supervising other operators who are watching banks of lights, display screens, and alarms. Quigley’s shift is nearly over, and so far it’s been an easy day. Suddenly a cacophony of bells and sirens screams out, jolting everyone in the room into action. One of the plant’s two reactors has been shut down by control rods automatically inserted into the reactor to stop the atomic chain reaction that produces heat. But the reactor is still hot–more than 600 degrees and rising. Inside the control room, signals tell the operators the outside source of electricity to the reactor and its cooling systems has failed. If Quigley and his crew don’t get cooling water to the reactor, the decaying radiation will keep heating up, the reactor core will melt, and Byron will become the next Three Mile Island.

Immediately, operators start the emergency diesel generator, which sends power to a circuit breaker the size of a refrigerator. The breaker should send electricity to the backup feed water pump, which should then transport cooling water to the reactor. But today the circuit breaker catches fire. The pump cannot start and no water is getting to the reactor. Normally, operators would try the second diesel generator, but today the reserve generator is out of service for maintenance. Quigley and his crew still have options. Next they try to cool the reactor’s core indirectly by starting a diesel-driven backup feed water pump. Except today it doesn’t work. “We’re going down a bad road there,” Quigley says. “At that point, it would be difficult to maintain core cooling.”

The scenario haunts Quigley because he believes deregulation is allowing nuclear plants to cut too many corners in their bid to be the cheapest producers of electricity. And he’s not alone. Other nuclear plant employees, regulators, industry consultants, and citizen groups are concerned, too. Utilities like Exelon, the new parent company of Commonwealth Edison, have landed on top of a market shakeout in which a handful of companies created by mergers bought up nuclear plants around the country. Com Ed plants are making more money and producing more electricity than ever before. But they’re doing it in part by cutting jobs, asking remaining employees to work more overtime, and performing less maintenance and fewer inspections. At the same time, Congress has forced the U.S. Nuclear Regulatory Commission to ease regulations, cut its inspection staff, and stop fining nuclear plants that violate federal rules.

Nuclear boosters claim plants today are safer than they’ve ever been, and they say more layoffs and fewer regulations are needed to make plants even more competitive. At the urging of the White House and Republicans in Congress, as well as the industry, the NRC is quickening its review of applications to extend the lives of aging reactors and to build new ones. The White House energy plan, released earlier this year, endorses nuclear power as a clean, “green” alternative to coal-fired power plants. “If you want to do something about carbon dioxide emissions, then you ought to build nuclear power plants,” Vice President Cheney said. “They don’t emit any carbon dioxide. They don’t emit greenhouse gases.”

But nuclear power presents other environmental problems, its opponents say, including radioactive waste and daily low-level emissions of radiation. And they worry that pressure to produce electricity at the lowest cost could lead to a nuclear accident exposing thousands of people to high doses of radiation. They also worry, especially after this month’s deadly attacks on the World Trade Center and the Pentagon, that American nuclear power plants could be targeted by terrorists. “The [Federal Aviation Administration’s] version of security was asking passengers two inane questions,” says Dave Kraft, director of the Evanston-based Nuclear Energy Information Service. “That’s the same attitude we see [with the Nuclear Regulatory Commission]–going through the motions of regulating without providing substance.”

In Quigley’s nightmare, there aren’t enough workers on hand to respond when the reactor shuts down, and the ones available aren’t fast on their feet because they’re so tired. The diesel feed water pump, which would have helped cool the reactor, doesn’t start because an exhausted worker misaligned a valve in the pump weeks ago. And, because plants are doing less maintenance and the NRC is inspecting less often, the misaligned pump valve, along with frayed wiring in the breaker, was never detected.

“They say that scenario would happen once in 100 million years,” Quigley says. “But you’re cutting staff, you’re working overtime, the NRC is inspecting less, you’re doing more maintenance on-line and taking stuff out of service. You’re cutting corners and putting more work on fewer people, and sooner or later they’re going to start screwing up.”

Back in 1997, executives at Chicago’s Commonwealth Edison were debating whether to shutter their six nuclear plants in Illinois. The Zion plant, 50 miles north of Chicago, had been closed for good after a series of safety sins that culminated with the discovery of a nitrogen bubble inside the nuclear reactor. The La Salle plant, 84 miles west of Chicago, was temporarily shut down as its staff tried to atone for numerous mishaps, including the partial clogging of the tunnel that delivers water to the reactor. The Dresden plant, 54 miles southwest of Chicago, had been named for the sixth year in a row to the NRC’s watch list, a distinction reserved for the worst nuclear plants in the nation. Com Ed had paid nearly $1 million in fines to the NRC in 1997 alone for violating a litany of federal nuclear safety laws at all of its Illinois plants.

That same year, the Illinois legislature voted to deregulate electricity sales. This meant that in two years Com Ed would start to lose its monopoly over metropolitan Chicago and the northern part of the state. Under the old system Com Ed was guaranteed customers, and the huge debt from building nuclear plants was factored into the state-approved rates that consumers paid. But deregulation was a new game entirely. Supply and demand would determine rates, and if Com Ed couldn’t provide cheap electricity to Illinois, another utility–a coal-fired plant in Ohio or a natural-gas-powered turbine in Wisconsin–would be happy to sell its electricity to Com Ed’s old customers.

And Com Ed could not compete. It owed billions on nuclear-plant mortgages, its older plants needed new equipment, and the NRC was bearing down on the utility for its dismal safety record. Com Ed lobbied for legislation that would allow it to recoup its nuclear investment; the plants, after all, had been built to satisfy the public need for electricity at a time when Com Ed was the area’s sole provider. To now be stuck with these “stranded” costs, the company reasoned, would prevent it from competing against utilities using low-price generators like gas and coal. The Illinois legislature listened, and it wrote provisions into the deregulation bill that saved Com Ed. First, it said customers who switched to a cheaper utility would still have to pay for a portion of Com Ed’s nuclear costs through fees. Second, it said Com Ed could sell off its physical assets to raise money to cover its costs.

The legislation did force Com Ed to lower its rates, but the utility’s critics say customers were left to foot the bill for years of cost overruns and mismanagement at Com Ed’s nuclear plants. “The legislation allowed Commonwealth Edison and Illinois Power to reap multibillion windfall gains by selling their coal plants and transmission lines that the public had paid for over many years,” says Howard Learner, an attorney with Chicago’s Environmental Law & Policy Center, a consumer advocacy group that has challenged Com Ed’s rates in court. “Then the utilities [have] kept the billions of dollars in profit and not passed those on to consumers.”

News that Com Ed would be able to recover stranded costs and sell assets buoyed its flagging stock in late 1997. Even so, so many plants had been closed or were limping along with safety problems that Wall Street still doubted whether Com Ed and other nuclear utilities could compete.

That’s when Oliver Kingsley Jr. and John Rowe entered the picture. Kingsley, a burly ex-navy man with a reputation for whipping troubled nuclear plants into shape, took over Com Ed’s nuclear operations in late 1997. A few months later Rowe came on board as the utility’s chief executive officer. Rowe didn’t want to sell the company’s nuclear plants; he wanted to turn them into moneymaking machines. He announced a risky plan to sell Com Ed’s gas-fired plants and to concentrate on boosting the productivity of the floundering nuclear fleet.

At the same time, big changes were under way on Capitol Hill. Utility executives and lobbyists began an aggressive campaign to soften federal nuclear regulations. The Nuclear Energy Institute (NEI), the industry’s lobbying arm, argued that nuclear plants were safer than ever. The industry did not deserve–nor could it afford–the stringent and expensive regulations imposed after Three Mile Island. The partial core meltdown near Harrisburg, Pennsylvania, in 1979 had forever changed nuclear regulation in the United States. The Nuclear Regulatory Commission grew up and got mean. Its staff ballooned, its budget multiplied, and Congress expanded its powers. Inspectors no longer trusted utilities to do the right thing. They demanded maintenance records that proved pumps were being tested; they looked at valves and pipes and emergency plans; they started employee training and testing programs; and they levied huge fines when utilities failed to fix problems. The high cost of regulation had been factored into the state-approved rates that customers paid, thus guaranteeing stable dividends to shareholders. But with the end of the monopoly system, the industry knew its days were numbered if the rules did not change. In June 1998, utility analyst Steven Fetter of the New York bond rating firm Fitch IBCA, Inc. told the Senate’s nuclear oversight committee that NRC enforcement stymied investment in nuclear power. “Needless to say, a situation like this creates a great deal of unease among debt and equity investors and inhibits new investment in nuclear as competition comes to the electric sector,” Fetter said.

Utility executives complained that regulations were outdated and NRC inspectors were out of control. They told stories of excessive regulation, such as being cited for putting two asterisks instead of one next to a note in a control room logbook. Their claims had some merit. The NRC was by its own admission inconsistent, unpredictable, and overly subjective. Some inspectors abused their authority and cited plants for violations that had nothing to do with safety.

But there was another side to the story. Some plants got away with egregious safety violations for years because NRC inspectors failed to look hard enough or turned a blind eye to problems. When the General Accounting Office reviewed NRC enforcement at three nuclear plants in 1997, investigators found that the utilities “failed to fix substantial and recurring safety problems in a timely manner.”

“NRC allowed these licensees repeated opportunities to correct their safety problems,” Gary Jones, associate director of the GAO’s energy office, told the Senate’s nuclear safety subcommittee in 1998. “However, implementation of corrective action plans were never fully completed, and although management promised to fix problems, [they] did not always follow through.”

In 1996, the NRC, acting on a tip from a whistle-blower, found that a Maine nuclear plant had deliberately falsified safety data so that it could produce more electricity even though it knew its cooling system could not support the increase. At the same plant, NRC inspectors found that 16 feet of cable in a high-pressure water pump’s control circuit had been inadvertently cut and removed three to four years earlier. If the pump had been needed in an emergency, it wouldn’t have been usable. In 1998 the NRC discovered that a Michigan plant operated for 13 years with a severed piece of piping that in an emergency delivers borate solution to the reactor core. In an accident, there would have been no backup way to shut down the reactor. In Cordova, Illinois, Com Ed discovered during a 1997 drill that a fire would have cut off power to all its emergency pumps, leaving the Quad Cities plant with no way to cool the reactor. Twenty-two years after the NRC required all nuclear plants to revise their fire plans, Quad Cities was still vulnerable. In each of these cases, and in scores of others, equipment failure could have led to accidents as bad as or worse than Three Mile Island. Why didn’t those accidents happen? “These plants did not experience an initiating event which required the backup equipment to function,” says Dave Lochbaum, a nuclear engineer with the Union of Concerned Scientists, which lobbies for better regulations. “Safety was achieved by not challenging the equipment rather than by having the equipment successfully fulfill the required safety systems. In other words, luck.”

Not surprisingly, nuclear lobbyists never mentioned these examples to Congress when they made their case for “improving the regulatory environment.” In 1997, there were ten reactors nationwide closed by the NRC because of safety problems, including La Salle and Zion. Under new leadership, the agency levied $7.4 million in fines in 1997, double the 1996 amount. And inspectors were issuing more citations for small violations they believed added up to bigger trends, such as bad management or inattention to detail.

The industry fought back hard, arguing that regulators were ignoring clear signs that plants were safer. As evidence of their equipment’s greater quality and reliability, they pointed to plant capacity factors that have jumped from around 64 percent to almost 90 percent industrywide in the last decade, according to the NEI. This means plants are producing more electricity relative to their physical limits than ever before. Also, the number of automatic shutdowns has declined significantly since 1980, indicating more reliable equipment. Twenty years ago, plants automatically shut themselves down an average of 7.3 times annually. In the mid-90s, the average dipped below one. (Critics contend that operators manipulate the numbers by manually tripping their reactors before they automatically shut down.) From 1989 to 2000, the number of unusual events the industry reported to the NRC declined from 197 to 18. “The safer a plant is, the less time it’s involuntarily down,” says Mitch Singer, an NEI spokesman. “Consequently, safer plants run more efficiently and more profitably.”

The NEI told Congress that regulations should be based on actual safety risks associated with rule infractions, rather than the infractions themselves. If a valve was found accidentally closed, the NRC should calculate the risk of that leading to a serious accident, given the multitude of backup systems available. Nuclear energy lobbyists argued that plants with good safety records should be allowed to police themselves, with limited NRC inspection. And they said that plants should not be fined for rule infractions.

A few powerful congressmen got the message, and over the following years they received hefty campaign contributions from the industry in return. Republican Pete Domenici, then chairman of the Senate subcommittee that decided the NRC’s budget, threatened to slash the agency’s 1999 funding by $20 million if the NRC didn’t change its ways fast. In June 1998, Republican Jim Inhofe, then chairman of the Senate nuclear safety subcommittee, convened the first NRC oversight hearing in four years. Inhofe made clear the reason for the gathering: “In order to encourage growth in the industry, we must reform the NRC,” he said in his opening remarks.

Mike Stein was a lawyer with the NRC’s Office of Enforcement when Congress started pressuring the agency in 1998. He was part of the team that issued tickets and fines to utilities that broke rules. “[Congress] put a real scare into the NRC,” says Stein, who now works for the Federal Energy Regulatory Commission. “That’s when the industry got its foot in the door.”

After the congressional hearings, the Nuclear Energy Institute’s people became regulars at the NRC offices in Washington, Stein says. “They started meeting with our staff to develop the new regulations,” he said. “By 2000 it was de-enforcement. We were bending over backward to ensure ‘green’ findings.” In 1998 the Office of Enforcement proposed $5.2 million in fines, $2.2 million less than the previous year. By 1999–when Richard Meserve took over as chairman–the NRC issued only $1 million in fines, and by 2000 that number dropped to $447,000. Congress spared the NRC the deep budget cuts it had threatened, paring the agency’s budget by only $7 million. But NRC funding has been declining since 1993, and the agency now has $70 million less to spend and 500 fewer employees than it did eight years ago. Vicki Bier, a human performance analyst at the University of Wisconsin who recently completed a study on deregulation for the NRC, says, “The NRC is now at pre-Three Mile Island levels at a time when the world is about to turn upside down [because of deregulation] in the industry.”

With the promise of a kinder and gentler NRC, big nuclear utilities like Com Ed started looking at their nukes with renewed affection. Rowe, the new CEO, reasoned that selling the company’s battered reactors would raise only pennies on the dollar. But if Com Ed could boost the output and capacity of its reactors–by improving equipment and running it more efficiently–the utility could produce lots of cheap power.

Oliver Kingsley Jr. was the linchpin in Rowe’s risky plan to make Com Ed nukes pay off. A former nuclear submarine commander, Kingsley seems to inspire fear, panic, or admiration, depending on whom you talk to. He’s an imposing man with a big southern voice that booms when he’s giving an order or demanding an explanation. He is best known in the utility industry for the turnaround of the Tennessee Valley Authority’s nuclear program. But his critics say Kingsley’s real success is at buttering up the NRC and intimidating employees.

Ann Harris worked for Kingsley at the TVA’s Watts Barr nuclear plant. When she repeatedly raised safety concerns about the plant’s electrical systems, Harris says TVA managers, including Kingsley, harassed her. The Department of Labor agreed. From 1990 to 2000, Harris prevailed in six whistle-blower lawsuits against the TVA. “TVA, under Oliver Kingsley’s regime, had the largest turnover of employees, senior management, and middle management of any time in TVA’s nuclear program,” Harris says. “He cuts away people he cannot control. He wants ‘yes’ people around him. He rants and raves. I’ve seen his face turn purple in meetings when someone disagreed with him.”

Despite scores of whistle-blower complaints from Watts Barr employees, the NRC allowed the plant to open in 1996, a staggering 23 years after construction began. Harris says Kingsley worked “sweetheart deals” with the NRC. “He knows how to negotiate,” she says. “He says, ‘We’ll put this in writing.’ And the NRC says, ‘As long as it’s in writing and you say you’ll fix it, we’ll trust you to fix it.'”

Repeated requests to interview Kingsley for this article were denied by Exelon public relations officials.

As he did at the TVA, Kingsley has turned around plant productivity at Com Ed. By the end of 1999, all the Com Ed nukes (except Zion) were back on-line, and the fleet was generating more electricity than ever before. Exelon expects to make equipment improvements that by 2003 will generate an extra 1,000 megawatts–the equivalent of bringing a new plant on-line. That’s enough new energy to light 750,000 homes.

Within two years of Kingsley’s arrival, the NRC had lifted an “adverse trend letter” against Quad Cities and removed La Salle and Dresden from its watch list. Soon after, the NRC eliminated the list altogether–a longtime industry goal. Having plants on the watch list caused “a crippling drag on earnings for the companies affected,” Ralph Beadle of the NEI told Wall Street analysts last year.

By the summer of 1999, Unicom, then Com Ed’s parent company, was talking about buying more nukes. After raking in $4.8 billion from the sale of its coal-fired plants, Com Ed had money to invest. And Corbin McNeill Jr., CEO of Philadelphia Electric Company (PECO) and another former submarine commander, was looking for a partner. PECO had already teamed up with British Energy of Scotland to form AmerGen, a holding company that would soon buy two nukes in New Jersey and Pennsylvania, including the remaining undamaged reactor at Three Mile Island. The company also bought the Clinton nuclear power station, 40 miles west of Champaign, from Illinois Power for pennies on the dollar. While companies with only one or two reactors, such as Illinois Power, wanted to rid themselves of nuclear plants, PECO, British Energy, and a handful of other big nuclear utilities saw opportunity. Their strategy was to buy up reactors on the cheap and run them more efficiently through economies of scale. AmerGen bought the Clinton plant, which was valued at $4.25 billion, for $200 million in December 1999, the same month it paid $100 million for Three Mile Island. By the fall of 2000, PECO and Unicom’s marriage was approved, and Exelon, a $16 billion company with the largest nuclear fleet in the nation, was born.

Wall Street immediately took a shine to the merger. A year later, 12 of 15 analysts surveyed by Thompson Financial/First Call of New York said they were bullish about Exelon stock, advising clients to buy. And Exelon recently reported second-quarter earnings gains of 18 percent over the same quarter last year.

But producing more electricity and making more money doesn’t mean plants are safer, says Dave Kraft of the Nuclear Energy Information Service. “Some of these improvements have come about by reducing regulations and instituting things like on-line maintenance, which means they can do more maintenance while the reactor is operating,” Kraft says. “They are testing to the max all these limits, and now they’re bringing in economic constraints that weren’t there before. We think it’s a bad recipe, and somewhere along the line something big and awful will happen.”

It’s a scary feeling when the lights go out at a nuclear power plant–those ten seconds of held breath between darkness and light, when workers pray one of the two emergency diesel generators will kick in. The darkness means the outside source of power has failed. Maybe lightning hit the transformer. Maybe too much electricity was flowing through the grid–a new problem with deregulation–and the whole system tripped.

When it happened at the Byron plant in northern Illinois two years ago, the generator did kick in. The lights came on and workers exhaled. Then the work began. First they had to figure out why the external power failed. At the same time they had to monitor all the gauges, screens, and alarms to make sure the reactor was being cooled by the backup water system. And they had to get the primary power source restored. But by the time everyone had been assigned a job, there was no one left to monitor the generator. So for two hours the plant ran on backup power without the staff knowing if the generator was operating properly. “Those engines are just like a car. They can overheat or have low oil pressure,” says Quigley, who has worked at Byron for 16 years. “When we run [generators] for maintenance or testing, we have someone in the room. But when we run them because we need them, we’ve got no one to monitor them.”

After the incident, Quigley wrote a report in which he told management that a shortage of manpower had left the staff unprepared to deal properly with an emergency. Nothing happened. He says the company has consistently ignored reports that he and other managers have written regarding staffing concerns. An Exelon quality-assurance employee who asked not to be identified said he sees the same pattern at other Exelon plants in Illinois. The quality-assurance department is like the internal affairs division of a police force. Its employees are supposed to make sure the company follows federal regulations. “Guys that operate reactors continually write that people are making mistakes because they are too busy, because there are not enough people,” the employee says. “And management continually does nothing. I specifically looked at how those problems were dispositioned, and at every site, in every case, they were closed with no action taken.”

In 1997 there were 886 workers at the Byron plant. Today there are 799. At Byron, Braidwood, Dresden, La Salle, and Quad Cities, Com Ed employed 4,541 people four years ago, according to the company. Now, Exelon says its roster at those plants lists 4,004 employees–a reduction of 537. (Zion and Clinton were not included because Zion is now closed and Clinton was not part of Com Ed in 1997. But according to inside sources and corporate data, the number of workers at the Clinton plant declined from 936 to 672 between May 2000 and July 2001–a 28 percent reduction in staff.) At the Dresden plant in Morris, there are 168 fewer workers today than there were four years ago. Quad Cities has 131 fewer people; 103 jobs have been cut at La Salle; at Braidwood, 48 workers have lost their jobs.

Staffing cuts don’t jeopardize safety, insists Exelon spokesman Dave Knox. Exelon, with 17 reactors nationwide–11 in Illinois–is consolidating many jobs that once existed at each separate plant. For example, while the number of engineers per site may be smaller, the company is able to move engineers from site to site as needed, Knox says. He argues that moving employees around improves efficiency because lessons learned at one plant can be applied at others.

“Having fewer people is not necessarily a problem if you have the right people with the right responsibilities,” says Vicki Bier, who studied deregulation for the NRC. “I think downsizing is necessary. But it is easy to make mistakes in downsizing and not realize you have cut too far in a certain area until something bad happens.”

Bier worries that plants are losing too many senior employees to early-retirement and attrition packages. Many of the employees who are leaving have been working at nuclear power plants most of their careers, she says, and are better able to make quick decisions. “Maintenance technicians have a lot of information in their heads about the past history of a piece of equipment–like ‘this problem only happens in hot weather’ or ‘the last time this happened, we solved it this way,'” she says.

Knox says Exelon was careful to retain workers with institutional knowledge or special skills. “We don’t want to lose our best people,” he says. “It was our decision who left…. It’s not just ‘Let’s reduce people.’ Where we can reduce to improve efficiencies, we do.”

Part of the problem in evaluating the choices made so far by Exelon and other nuclear utilities is lack of information, Bier says. The NRC doesn’t track staffing levels at plants, so regulators and the public don’t know the depth of staffing cuts. The NRC requires minimum staffing levels only for control rooms. There are no regulations governing the number of people who work in other areas of the plants, such as engineering, maintenance, radiation waste, training, quality control, or administration.

Michael Fox, a nuclear energy consultant in the state of Washington, used industry data to analyze staffing patterns at plants from 1993 to 1998. Fox found that during that period total staffing levels declined 14 percent industrywide, and he feels certain that plants have trimmed their rosters more since then. If they haven’t, he says, they will. “To survive, utilities need to reduce significantly their operating and maintenance costs. The only way this can be done is by reducing staffing levels.” Fox’s study found that some jobs, such as those in control rooms and in administration, have been unaffected, while training-department and quality-control staffs declined by almost 30 percent from 1993 to 1998. This troubles Bier, who says more training, not less, is needed if fewer employees are taking on more duties, often in new areas. “For every one position that’s cut in the power plant, there are two cuts in training,” says an Exelon training employee who asked not to be identified. “Training is getting the hardest hit.”

The NRC doesn’t track staffing at plants because there’s no indication that staffing affects safety, says David Desaulniers, a human factors analyst at the agency’s headquarters in Maryland. “You could have a very large staff that’s potentially not well managed, that’s inefficient in its operations and not well operated from a safety perspective,” he says. “Whereas a well-managed, tightly run operation could be very efficient and safe. The NRC tries to keep its focus on safety and performance, rather than something that is so much more removed.”

But Bier says that if the NRC reacts only to performance, it might not recognize staffing levels that are eroding safety margins until it’s too late. In Bier’s study of deregulation in other industries–including airlines and railroads, and also nuclear plants in Great Britain–the problems she saw “were not problems with equipment and reliability, but problems with human performance, safety culture. And NRC is de-emphasizing these issues.”

Staffing cuts also have led to an increase in overtime at Exelon plants, especially among nonunion employees such as control-room supervisors, engineers, instructors, and line supervisors. Com Ed used to pay overtime wages to these management employees, but since Kingsley came on board that has stopped.

“They have been abusing people really bad,” says one Exelon manager who asked not to be identified. “This week alone, I’ve put in 17 hours of overtime. Today [a Wednesday] I worked 14 hours. I went as long as I could until I couldn’t see anymore, then I went home.” The employee says safety is being sacrificed because “everybody is running on overload. Unit supervisors and shift managers, they’re working all the time. They’re so busy they can’t really focus on the plant. They’re too busy doing other jobs.”

Morale is sagging among employees who are working overtime without extra pay, Quigley says, and this by itself can jeopardize safety. “When you’re working 60 hours a week and getting paid for 40, you’re not inclined to identify problems and solve them,” he says. “It’s more work. If someone is working 20 hours a week for free, why should he identify a problem that will cost him another 10 hours a week?”

The NRC has a policy, but no regulation, discouraging “routine” or “excessive” overtime at plants. According to the policy, workers with safety-related jobs should work eight hours a day, 40 hours a week while the plant is operating unless there’s a temporary need for longer shifts. Then the workers aren’t supposed to work more than 16 hours in a 24-hour period, more than 24 hours in a 48-hour period, or more than 72 hours in a seven-day period. Deviations are allowed only under “very special circumstances” such as an emergency. But about a third of the plants that responded to an industry survey said they routinely authorized more than 1,000 deviations–and up to 7,500 deviations–annually. “There are no firm standards for what constitutes excessive overtime and no requirements for reporting that type of data to the NRC in the first place,” Bier says. “So the NRC doesn’t know how much overtime is being worked, how much staffing levels have changed, and has no basis for determining where overtime is increasing, whether it’s increasing to a problematic point, or of determining if their regulations about overtime are being satisfied.”

Larry Lawson, a control-room operator at Byron, routinely works 56 hours a week, including at least two 16-hour shifts. In a recent week his schedule looked like this: in on Monday at 3 PM, out at 11; back on Tuesday at 3 PM, out at 7 AM Wednesday; back on Wednesday at 3 PM, out at 11; back on Thursday at 3 PM, out at 7 AM Friday; back on Friday at 3 PM, out at 11. “That’s kind of brutal, but it’s all within the guidelines,” Lawson says. “You have to be alert [on the job]. It’s hard to do when you work 16 hours and come home and get 6 hours to sleep at the most–assuming you can sleep in the day.” The most important part of Lawson’s job is monitoring safety indicators in the control room. If something goes wrong, he’ll have to make the right decisions fast. But at the end of a 16-hour shift, at the end of a 56-hour week, it’s hard to think at all. “Have you ever been drunk? It’s a lot like that,” Lawson says. “Your judgment is impaired. You’re slower. It takes longer to absorb a complex problem. Even reading something, sometimes you have to read it over again.”

As shifts started getting longer, Quigley began to worry that people were making more mistakes because they were tired. But every time he raised the issue, he says, the company ignored him. “Whenever someone made a mistake, they never looked deep enough to see whether it was because they were tired.” The “root cause” was always something like “inattention to detail.” In 1999, Quigley went to congressmen Ed Markey, John Dingell, and Ron Klink with his concerns. He asked why the federal government limits hours for truck drivers and airline pilots, but not nuclear plant employees. The congressmen wrote the NRC: “We are concerned that industry cost cutting in response to the emerging competitive electricity market could be leading to staff reductions that will exacerbate overtime and degrade the ability to safely handle plant emergencies.” In her response to their letter, the NRC’s then chairman, Shirley Ann Jackson, acknowledged, “The NRC has no regulation limiting the use of overtime for personnel working at nuclear power plants.” But she wrote, “The NRC has attributed few events at nuclear power plants to personnel fatigue….Of those events that have been attributed to personnel fatigue, the NRC has found that they were typically of limited safety significance.” The letter revealed that the NRC planned to end “routine review of licensee use of overtime,” which it has since done.

“There are no events [attributed to fatigue] because we don’t look hard enough,” Quigley wrote the NRC in response. “To rely on shallow or missing data to conclude that there is no problem with fatigue borders on irresponsibility.”

In September 1999, Quigley filed a petition asking the NRC to create an enforceable regulation limiting nuclear plant employees to 60 hours a week and 108 hours in two weeks. He asked that 16-hour shifts not be allowed to run into the hours between 11 PM and 7 AM and requested that no more than two 16-hour shifts be allowed in a seven-day period. If employees were asked to work two 16-hour shifts, he asked that the second shift be preceded by 24 hours off. “Consider a 16-hour shift from 3 p.m. to 7 a.m.,” Quigley argued in his petition. “Assume the worker arises at 8 a.m., after a restful sleep, on the day he is to work….Near the end of the shift, the worker will have been awake for almost 24 hours. Australian researchers show that at 24 hours awake, the performance degradation is equivalent to a blood alcohol content of .10.” The level at which a driver is considered legally drunk in Illinois is .08.

The industry said in response that Quigley hadn’t proved fatigue was responsible for any worker accidents, and it claimed that both the current NRC policy and its “fitness for duty” program addressed fatigue. The fitness-for-duty program, established to eliminate drug and alcohol use at nuclear plants, relies on supervisors to identify signs of fatigue in workers. But Quigley says it’s naive to depend on supervisors to do this. First, he says, they’re no longer trained for it. “For example, a recent annual exam contained only one question on how to identify aberrant behavior,” Quigley says. “It asked, ‘What type of behavior is it when you see someone sitting in the middle of the floor crying?’ This in no way, shape, or form qualifies a person to identify a fatigued or impaired worker.” Second, he says the monitoring of field employees has become much less rigorous because supervisors are “spread thin and spend a great deal of their time fighting fires.” Finally, he says some workers don’t have supervisors, especially on weekends.

In its response to Quigley’s petition, the industry reminded the NRC that it would have to prove that new overtime regulations would increase public safety to such a degree that the cost to the industry would be justified. That standard, called the backfit rule, applies to any new regulation the NRC proposes.

Last June, after reviewing the petition for nearly two years, the NRC staff issued a report to the five-member commission urging it be adopted: “The staff reviewed the current and proposed federal limits on work hours for personnel in six other industries in the U.S. and Canada, as well as for nuclear plant workers in eight other countries….The staff found that the NRC’s guidelines were collectively the least restrictive.” As of mid-September, the NRC had yet to make a ruling on overtime.

It was a Saturday afternoon last May in Braidwood, Illinois, a town of 5,200 people 50 miles southwest of Chicago. Workers at Exelon’s Braidwood nuclear power station were trying to get an auxiliary transformer back on-line. The transformer sends power to backup emergency systems, but it had been taken out of service for maintenance. The workers had never done this operation before, so a supervisor had accompanied them to the switch room where they were reviewing procedures. Then the supervisor’s beeper went off, and he left to take a phone call. The workers who were expected to pull the fuse that would reactivate the transformer weren’t wearing protective clothing, so another worker who was suited up volunteered to pull it for them. This worker pulled the wrong fuse, causing the reactor’s cooling pumps to lose power. The reactor automatically shut down, something known in the industry as a scram.

The incident highlights a trend that some critics say is compromising safety. Doing maintenance, especially to backup safety systems, while the reactor is running was unheard of in the nuclear industry a few years ago. But under new pressure to make money by keeping plants running as many days as possible, the amount of time that Exelon plants are shut down for maintenance and refueling has been reduced from about 100 days a year to an average of 22 days. That means each plant can produce about $39 million more of electricity every 18 months. Much of the maintenance that used to be done while the plant was cool is now done when it’s running at 100 percent power.

At Braidwood, the emergency systems worked last May when the reactor shut down. But doing maintenance on-line increases the probability that a reserve safety system will be unavailable during an emergency or that a worker will make a mistake that causes the reactor to shut down. When something goes wrong, the plant is designed to shut down automatically, says Dave Lochbaum of the Union of Concerned Scientists. But “during a scram, there are so many things going on at the same time that if you encounter something unexpected, you have a lot of balls up in the air to juggle and you have to make sure you don’t drop any of them. It’s more complicated and therefore more likely that something will go wrong. Both at Chernobyl and Three Mile Island, the first step was a reactor trip.”

Backup safety systems like transformers, emergency diesel generators, and reserve cooling pumps are regularly taken out of service for maintenance while the plant is hot. Because there are at least two generators per reactor, the industry and the NRC say the risk of losing all power is remote. And even if both generators were lost, other backup systems would be available to keep the reactor cool enough to avoid a meltdown.

But Lochbaum notes that backup generators have failed before. At the Vogtle plant in Georgia, a truck backed into the switchyard in 1991 and knocked out power to the reactor’s cooling system. One generator was out for maintenance and the other one didn’t work. Luckily, the plant was in a refueling outage, so the core temperature rose from 90 degrees to only 136 degrees in the 36 minutes that it took to get backup electricity restored. If the plant had been on-line, the fuel would have started at 600 degrees and the temperature would have risen much faster.

“There are virtually no restrictions on what can be done on-line now,” Lochbaum says. “Exelon is one of the leaders in on-line maintenance.”

At several Exelon plants in Illinois, workers are now being ordered to do maintenance inside the building that houses the reactor while it is splitting atoms. “That’s when it’s most radioactive,” said one Exelon maintenance worker who asked not to be identified. “That’s the scary part. They’re trying to do so much on-line, regardless of the safety aspect or the health of workers.”

The nuclear industry argues that on-line maintenance actually improves plant safety. “If [equipment] is performing in a degraded condition, you don’t wait for an outage to fix it. You fix it right away,” says Tony Pietrangelo, NEI’s director of risk- and performance-based regulation. In addition, on-line maintenance allows employees and contractors more time to do repairs because they aren’t as rushed, Pietrangelo says.

But Quigley says workers are more rushed–and more stressed-out–than ever before. “Outages used to be 90 days long. You worked your ass off for 90 days but then it was back to normal,” Quigley says. “Now you have an outage pace continuously. It’s busy all the time.”

In the past, workers conducted routine tests and repairs when the plant was on-line, but they saved major tests and maintenance projects for outages, which occurred every 18 to 24 months. Today workers have to juggle both kinds of projects while the plant is hot. And there are fewer workers to do the jobs. “They’ve increased the overall workload on a daily basis to minimize time they’re in outages,” says an Exelon quality-assurance employee who asked not to be identified. “People are making mistakes because they’re handling too many distractions at one time. I’ve seen guys in the control room who have oversight of the reactor, and I have seen them handle three to five tasks in seven minutes and be interrupted by three phone calls and be interrupted by individuals at least twice. It diminishes the amount of focus that they have on each activity because of those interruptions and phone calls.”

Water is the key to controlling nuclear reactors. If the water level inside a reactor gets too low, the radioactive core could overheat. That’s why nuclear plants are built with series of automatic relays that monitor water. If its level dips too low, the relays trip the reactor, stopping the fission process. At the Braidwood and Byron control rooms, operators used to test relays four times a year. Now they are tested once every 18 months.

In Illinois and throughout the nation, the NRC is allowing plants to inspect and maintain much of their equipment less frequently–this at a time when aging plants are producing record amounts of electricity for longer periods of time. Pumps that used to be tested every month are now tested every three months. Valves that used to be checked quarterly to ensure that they opened and closed now are tested yearly. Measurements of pipe strength taken every year are now taken every five. A few years ago, utilities were required to test the amount of radiation leaking from the reactor’s concrete housing three times every ten years. Now it is tested once every ten years, saving the industry as much as $1 billion over 15 years, according to the Nuclear Energy Institute. Taken together, the reductions in testing frequency and maintenance requirements will save the nation’s nuclear plants hundreds of millions of dollars annually, according to Public Citizen, an advocacy group that monitors nuclear plant safety.

“As humans get older, we go to doctors more often to make sure we maintain our health at optimum levels. As nuclear plants get older, fewer checkups are done,” says Lochbaum.

The new testing variables are part of an overall trend toward so-called risk-informed decision making by the NRC. In the past, the NRC required plants to test and maintain equipment according to the vendor’s recommendations. Now utilities are being allowed to revise their testing and maintenance schedules according to how well different components have performed over the last two decades. The risk of failure is calculated based on historic failure rates.

“Risk-informed is more realistic because it takes into account the likelihood of something happening,” says Mark Ring, an NRC branch chief in Lisle, Illinois. “In the beginning of nuclear power, there were no analytical tools for determining risks. You compensated by being overly conservative. With deregulation, if you have five overdesigned things, you’re paying for the fact that you have four you don’t need.”

But Lochbaum argues that the approach is fundamentally flawed. Plants are most likely to malfunction when new equipment is being broken in and when old equipment starts to wear out. But utilities are using performance data from a period when many plants were in the middle of their lives and equipment was most likely to work. “This neglects the fact that they’re heading toward the part of the curve where they’re entering the wear-out phase,” Lochbaum says. “If you reduce how often you test and inspect, you may miss the elbow where it starts rising again and equipment may not function as you want it to. Since components do wear out, you may not be testing it often enough to catch that failure early on.”

Lochbaum also worries that if an employee flips the wrong switch or incorrectly calibrates an instrument, the equipment won’t work if it’s needed. With less monitoring, utilities are less likely to detect problems caused by employees, he says. Tony Pietrangelo of the NEI insists that human error and equipment age are both considered when determining how often equipment is tested and maintained.

Risk plays a big role in the NRC’s new way of enforcing regulations. After an 18-month test at 13 plants, the NRC introduced its new “reactor oversight process” to all plants in April of last year. Under this plan, inspectors spend less time looking for rule infractions, allowing utilities to police themselves much more. When problems are found–either by federal inspectors or the company–the NRC no longer issues a violation or a fine. Instead, regulators determine the probability that the infraction could lead to a serious safety problem.

Nuclear plants were designed with so many layers of backup systems that in theory, an accident would be averted even if the main cooling pipe to the reactor broke, off-site power was lost, and an earthquake hit simultaneously. A handful of such postulated catastrophes became the backbone of the NRC’s regulatory program, even though the likelihood of any of them happening is “infinitesimal,” according to the NEI. For instance, if inspectors found a switch was faulty, and that switch would have been necessary during a major accident, the NRC would have cited the utility. “Under the new system, that would not be considered an event that was risk significant,” says the NEI’s Mitch Singer.

“We’re looking at the probability and the consequences of the event, and the combination of those is the risk,” Pietrangelo says. “We’re not just guessing at how often a pump will fail. We have hard data on how often that pump will fail and we can plug it in to plan how significant that is.” Some systems are more important than others. A ruptured steam tube would rank higher on the risk list than a broken switch with almost no chance of contributing to an accident.

Risk-informed proponents argue that focusing on the most important safety problems will lead to quicker action by the NRC because time won’t be wasted on minor issues, and that it will create a more realistic and objective standard for holding plants accountable. Lochbaum, who often criticizes the NRC for inaction, says he thinks some elements of the risk-based plan could work if the NRC follows its own rules.

Critics outside and inside the NRC, however, say that under the new system very few violations have been deemed “safety significant.” The new oversight process rates the severity of infractions according to a color. Minor infractions with “little safety significance” are green. “Serious” violations are classified as white, yellow, or red, depending upon the likelihood of the problem actually causing a nuclear accident. Green violations don’t lead to any increased oversight by the NRC. The company must simply explain how it is going to fix the problem. The NRC has inspectors assigned to each plant; but when it issues a white, yellow, or red finding, it sends in more inspectors to investigate. According to the NRC’s Web site, in the past 12 months 306 violations at nuclear plants were green, 23 were yellow, and one was red.

“The general concern is ‘Gosh, everything is green–there are no problems anymore.’ The concern is that the threshold is set to where nothing is a serious finding anymore,” says Gary Wright, manager of nuclear facility safety for the Illinois Department of Nuclear Safety. The state agency employs resident inspectors who work alongside NRC inspectors at all six Illinois power plants. Illinois is the only state with such a program.

Jim Luehman, deputy director of the NRC’s Office of Enforcement, acknowledges that inspectors aren’t completely sold on the new program. “The concern that many inspectors have is you could have an infinite number of greens,” Luehman says. “But green is not good. If you have multiple greens in an area, isn’t that telling you something? Are we going to wait until there’s actually a problem or are we going to be more proactive and act on the lower-level problems before they become big problems?”

Luehman says most managers and staffers are starting to come around. When the General Accounting Office asked NRC managers and staffers about the new plan in late 1999, 75 percent said they believed utility and industry groups had too much influence in developing the process. Sixty percent said they believed the new process would reduce safety margins. But a 2001 survey conducted by the NRC indicated that 88 percent of the managers and staffers believed the new process “provides appropriate assurance that plants are being operated safely.” Nevertheless, the survey tapped a vein of skepticism. Nearly half the respondents said they don’t believe the new process identifies declining safety performance before there’s a significant reduction in safety.

“The proof will be in the pudding,” says NRC branch chief Mark Ring. “The recipe is OK, but you don’t know until you’ve used it for a while if it really works. With licensees left to their own devices, will they continue to maintain high levels of maintenance or will they run more pieces of equipment to failure?”

For its part, Exelon insists that it cannot make money unless its plants run safely. “We know in our hearts and we have proven on paper that the best-run, safest plants are the most efficient,” says spokesman Dave Knox. “Cutting corners is not an option.”

But many critics of the new world of deregulation worry about eroded safety margins. “It is essential to evaluate the cumulative effect of cost-cutting measures on nuclear safety levels, instead of considering their consequences individually,” Lochbaum writes. “A bee sting is painful, but it is generally not life threatening. A thousand bee stings are a completely different story.”

Research for this story was conducted with financial support from the Fund for Investigative Journalism and the Academy for Alternative Journalism.

Art accompanying story in printed newspaper (not available in this archive): illustrations/Mark DeBernardi.