In Russian roulette, the players--all resigned to the possibility of death--put a single bullet into one of the six chambers of a revolver, spin the cylinder and take turns putting the gun to their heads. It starts as a one-in-six chance of disaster.
Those odds, however, are tame compared to those favored by our government's nuclear power regulators. They've recently been told by top scientists of a potential problem: floating paint chips and other debris can get past a poorly designed strainer and clog a key pump. If uncorrected, the scientists reported, this problem represents a one-in-three chance of disaster at an American nuclear power plant by 2007. The regulators' response? The problem ought to be fixed by, oh, say, 2008. That's the equivalent of chambering two bullets before spinning the cylinder and pulling the trigger.
"I don't see any issue out there that's more safety significant than this one," says David Lochbaum , a nuclear safety expert with the Union of Concerned Scientists . "This problem creates a one-in-three chance of a reactor accident by January 2007. The problem can be fixed in less than one year. The NRC's current plans call for it to be solved [before the first day of 2008]."
Choking on Paint
A nuclear power reactor needs a steady flow of cooling water. Otherwise it overheats and damages itself, and, in the worst-case, melts down . So if a pipe carrying coolant into or out of the reactor vessel should leak, water spills onto the floor, the level of available coolant drops and the reactor gets hotter. At that point, backup systems kick in. One such system is elegantly simple: The water leaking onto the floor drains into a collecting pool--which has a pump at the bottom that sucks it back into the system.
But some loss-of-coolant accidents are violent affairs, which flay insulation and paint off adjacent pipes and equipment. This and other debris gets washed down into the collecting pool--and, it turns out, it can choke the pump.
Is that disastrous? Not necessarily. There's a backup supply of coolant water that can be tapped for the short term. But outside water pumped into a leaking system means more water pouring out onto the floor. So the inside of the containment building floods, and rising water eventually submerges and destroys critical electrical equipment. Eventually, the weight of all that water may even break apart the containment building itself. And in any case, the backup water supply quickly runs out, since it's not being repleinshed.
Pretty scary. But c'mon: Floating paint chips bringing down a nuclear reactor? This sounds like something any neighborhood auto mechanic could fix. Why not just put a better debris-catching grate over the collecting pool? Or install a pump with a larger mouth that doesn't clog so easily?
Government regulators are considering such options. In fact, if you check out the references page  the NRC has put together on problems with "PWR Sump Performance "--the technical label for the collecting pool-pump issue--you'll see they've been grappling with this for nearly twenty years.
The current plan is to come up with a general idea for a fix shortly, have each plant design its own specific version of that general fix by May 2005, have the modifications all made by June 2007 and the paperwork all turned in by December 2007. So far, the agency's latest report adds, they're actually running a wee bit behind schedule.
But it turns out this long-known problem could be more serious than suspected.
Last year, a study the NRC commissioned from the Los Alamos National Laboratory  looked at sixty-nine of America's 103 nuclear power reactors--those designed to keep coolant water under heavy pressure, and so designated Pressurized Water Reactors, or PWRs --and asked how likely it was that their collecting pool pumps ("containment sumps") might clog at critical moments. Los Alamos, which has looked at this issue repeatedly over the years for the NRC, came back with a startling answer: Very likely. So likely, in fact, that a disaster at a nuclear plant is one hundred times more likely than previously believed.
Lochbaum, the nuclear safety expert with the Union of Concerned Scientists, found the numbers shocking. He rearranged the equations to spread the year-to-year risk of a major loss-of-coolant catastrophe across all sixty-nine American pressurized water reactors. His numbers:
Chance of a disaster by next year: one in ten.
Chance by 2005: one in five.
By 2006: one in four.
By 2007: one in three.
And if the NRC, which plans to have the problem fixed by 2008, gets delayed until, say, 2010? The chance of catastrophe rises to one in two.
What if the problem was fixed this this year at all relevant reactors? Then, Lochbaum says, the NRC could truthfully tell the American public that their numbers show a 99.9 percent chance of not having a loss-of-coolant catastrophe in the next year.
NRC: Just "Remove the Conservatism"
The NRC, in a written reply to questions from The Nation, accepted Lochbaum's interpretation of the Los Alamos study. Yes, the NRC wrote, according to its own commissioned study there's a one-in-three chance of disaster by 2007, which is why, the agency explained, it sent the report back for more work.
"The reason that the NRC commissioned Los Alamos Laboratories to analyze the sump blockage issue was to determine the credibility of that safety issue and to determine whether the agency should move forward with requiring a solution," wrote Gregory Cranston, a staff member in--take a deep breath--the Office of Nuclear Reactor Regulation's Division of Systems Safety and Analysis.
"Upon receiving the August 2002 report from Los Alamos Laboratories, the NRC, based on the significance of the conclusions reached, requested that a second report be prepared," one that "removed some conservatism from the original risk analysis." No, it didn't just pull out the bad news: The second report looked at mitigating actions and technologies, "such as standard operator action to recover from the accident," and "appropriate consideration of general design criteria related to leak-before-break technology."
So the NRC got a shocking report from Los Alamos National labs, and sent it back. It told the Los Alamos team to plug in new assumptions and to consider the concept of "leak-before-break technology"--the idea that equipment is carefully monitored in a nuclear plant, so a pipe that starts to leak a bit will sound alarms long before it bursts, leaving time to head off potential problems.
Los Alamos duly served up a new report in February 2003, one that plugged in those new NRC-mandated assumptions.
"Therefore, by removing some of the conservatism and using the same formula that was used by the UCS, the one in three chance...is significantly reduced to less than one chance in 100 (or less than a one percent chance)," Cranston writes.
"Additionally," Cranston concludes, "it should be noted that the safety record of nuclear power reactors has been quite good and, in fact, since the Three Mile Island event nearly a quarter of a century ago, there has not been a serious accident at any nuclear power plant."
Toledo, Three-sixteenths of an Inch From Chernobyl
For years, no one at Davis-Besse noticed as coolant water dripped onto the roof of the reactor vessel, a giant pot of nearly six-inch-thick carbon steel that holds the core. Or maybe they noticed, but decided it wasn't important. Either way, the coolant dripped, and the boric acid in it accumulated into a slushy little pile, and eventually the acid ate entirely through  the carbon steel walls, creating a hole  six inches deep, five inches long and seven inches wide.
Luckily, the inside of the vessel was coated with a three-sixteenths of an inch thick stainless-steel liner, which held off the acid--but the immense pressure inside the vessel forced it back out the hole. The Davis-Besse nuclear power plant, about forty kilometers outside of Toledo, stood three-sixteenths of an inch from a loss-of-coolant catastrophe. No one even knows how long the hole was there.
Nor was the hole formed without warning. In the summer of 2000, the NRC had put out the word: Nuclear power plant operators were to immediately inspect the nozzles of reactor vessels. Cracks in such nozzles had just been discovered at the Oconee nuclear plant  in South Carolina (embarrassingly, just months after the NRC had given the plant a clean bill of health). Similar cracks were found at plants in Arkansas and Virginia. At Davis-Besse in Ohio, the acidic water was dripping from just such a nozzle crack.
After Besse-Davis eventually alerted regulators to the problem in July, 2001, the NRC admitted it failed to properly police the plant and had ignored numerous warnings; a survey of NRC employees found many concerned "that the NRC is becoming influenced by private industry," and that there is "a compromise of the [agency's] 'safety culture.' "Davis-Besse, and the plant's owners, FirstEnergy Corp., have since been in the national doghouse--not least since it's emerged that the August 2003 blackout probably originated  with FirstEnergy's Ohio transmission lines. The good news is: When you've been caught in a ludicrous situation--letting acidic water drip for years onto your reactor and bringing Toledo within three-sixteenths of an inch from a possible disaster--you're free to fess up to other problems without taking much of a hit. You can even sell it as a contrite new dedication to safety.
And if you've shut down the plant anyway for embarrassing repairs, why not make all needed repairs? After all, even if it does end up costing $80 million , it's cheaper that way.
So in December, Davis-Besse became the first of the sixty-nine PWRs to confess that its containment sump pumps would probably not work if needed. "[T]he licensee stated that the recirculation sump had been declared inoperable as a result of the potential for sump clogging due to unqualified coatings [i.e., old paint] and other potential sources of post-accident debris," wrote the NRC.
In May, Davis-Besse also stated that another part of the system--high-pressure injection pumps further down the line--would also choke shut. (The Cleveland Plain Dealer, which has done excellent work on the Davis-Besse story, has a graphic here  that explains it all.)
The Davis-Besse announcements, coming on the heels of the Los Alamos "one-in-three chance of doom" study, galvanized the NRC. So what does a galvanized NRC look like? In June--one month after Davis-Besse's most recent admission--the agency issued Bulletin 2003-01 .
It recounted the Davis-Besse confessions, and also the research from Los Alamos labs. (Strangely for a comprehensive literature review, it cited the second Los Alamos study--the one with all of the "conservatism removed"--while omitting entirely its shocking parent, the August 2002 study, which is: NUREG/CR-6771, "GSI-191: The Impact of Debris Induced Loss of ECCS Recirculation on PWR Core Damage Frequency.")
The bulletin asked PWR operators to confirm, in writing, that they were all following existing rules. (Daring, no?) It also put forth some suggested Band-Aid measures--like arranging extra backup water supplies to draw on if the pumps don't work. At a June 30 meeting between NRC staff, industry representatives and experts, NRC staff also suggested that rather than risk filling the containment building with water to the point of collapse, plants simply leave open the doors, so the water can wash out.
Lochbaum, who was in attendance, afterward protested that to leave the doors open to high-radiation areas is "contrary to many lessons learned from radiation overexposure events in the past."
The Bathtub Curve
The original "one-in-three chance of doom" Los Alamos study actually went reactor by reactor nationwide and calculated the individual risk level for each. A chart in the study lines the reactors up from least at risk to most at risk.
The study did not name the reactors, instead giving them coded numbers. Lochbaum says he was able to crack the code, and determine the two power plants Los Alamos considered most at risk. They were the Vogtle plant  in Georgia, near the South Carolina border, and the Indian Point  plant, which a coalition  led by the environmental group Riverkeeper  has been working to close  on the grounds that, among other things, it's unacceptably close to New York City in a post-9/11 world.
Davis-Besse, meanwhile, is way over at the least-at-risk end of the table. And that evaluation came before it made its recent improvements. This summer it installed a new pump that's twenty-five times larger than the original, installed new grates and made other improvements. And the NRC's stern position is that Davis-Besse cannot restart its reactor until the agency is convinced the new containment sump solves the problem. Lochbaum asks: If the containment sump problem is so serious it keeps shut Davis-Besse, a reactor at relatively low risk, how come reactors with a far higher risk are allowed to keep running? "The NRC's actions and its words don't match," he says.
Consider again the final sentence of the NRC's response to questions for this article: "Additionally, it should be noted that the safety record of nuclear power reactors has been quite good and, in fact, since the Three Mile Island event nearly a quarter of a century ago, there has not been a serious accident at any nuclear power plant."
The accident at Three Mile Island came on the one-year anniversary--"almost to the very minute," Lochbaum says--of the first time the reactor operated. Other nuclear accidents, from Browns Ferry to Fermi Unit 1 to Chernobyl (the last major accident, in 1986)--also occurred in the first year or two of operation. That conforms with the bathtub curve , a classic engineering concept that reflects the likelihood of failure over time. The curve is shaped like a bathtub, because things are more likely to break down when they're new, or when they're old. "All of the nuclear accidents occurred in the first year or two of operation, or on the 'break-in' phase. Plants are now out of that phase. They are in, or heading toward, the 'wear-out' phase where failure rates climb," Lochbaum says. "It's only a matter of time before we start populating the right side of the curve with plant names."