Twenty-five years ago, Hollywood released The China Syndrome, featuring Jane Fonda and Michael Douglas as a TV news crew who witness what appears to be a serious accident at a nuclear power plant. In a spectacular coincidence, on March 28, 1979, less than two weeks after the movie came out, the worst accident in the history of commercial nuclear power in the United States occurred at Three Mile Island. For five days, amid growing alarm, the citizens of central Pennsylvania -- and the entire world -- followed the efforts of authorities to prevent the crippled plant from spewing dangerous quantities of radiation into the environment. This book is the first comprehensive account of the causes, context, and consequences of the Three Mile Island crisis. J. Samuel Walker captures the high human drama surrounding the accident, sets it in the context of the intensely polarized debate over nuclear power in the 1970s, and analyzes the social, technical, and political issues the accident raised. His book will clear misconceptions held to this day about Three Mile Island.
The heart of Walker's narrative is a moment-by-moment account of the accident itself, in which he brings to life the figures who dealt with the emergency: representatives of the Nuclear Regulatory Commission, the state of Pennsylvania, and the White House, and a cast of scientists and reporters. Walker also looks at the accident's aftermath in the area surrounding Three Mile Island, including studies of the long-term health effects of the incident. Three Mile Island is a fascinating window onto the politics of nuclear power and an authoritative account of a critical event in recent American history.
J. Samuel Walker is the historian of the United States Nuclear Regulatory Commission. His previous books include Permissible Dose: A History of Radiation Protection in the Twentieth Century (California, 2000) and Containing the Atom: Nuclear Regulation in a Changing Environment, 1963-1971 (California, 1992).
The first five hours of the graveyard shift that began at 11:00 p.m. on March 27, 1979, at the TMI-2 reactor were uneventful. The plant ran at 97 percent of full power while a staff of six employees monitored its operation and performed routine duties. The generator produced nearly nine hundred megawatts of electricity as clouds of steam billowed out of the plant's two cooling towers. TMI-1 was not operating because it had been shut down for routine refueling.
Like all power reactors built by Babcock and Wilcox and about two-thirds of the nuclear plants in operation in early 1979, both units at Three Mile Island were pressurized water reactors (PWRs). Three nuclear plant manufacturers used the principles of PWR design: Westinghouse, Combustion Engineering, and Babcock and Wilcox. The fourth vendor, General Electric, employed a different design called a boiling water reactor. In PWRs, the water pumped through the pressure vessel (at a rate of some ninety thousand gallons per minute) is kept under high pressure. As the water passes through the core, it is heated to about six hundred degrees Fahrenheit under normal operating conditions, but the high pressure of about twenty-two hundred pounds per square inch (150 times greater than atmospheric pressure) prevents it from boiling. In the TMI-2 plant, the core contained about a hundred tons of uranium encased in 36,816 thin, twelve-foot-long fuel rods. The pressure vessel that housed the core was thirty-six feet high and had steel walls nine inches thick.
Water circulates through the core in a PWR in what is known as the primary loop. After the heated water exits the core, it proceeds to one or more large containers called steam generators; the two steam generators at the TMI-2 plant were each seventy-three feet high. In the steam generators, the heat from the water passing through the core is transferred to the secondary loop, a separate system for circulating water. The water in the secondary loop is allowed to boil, creating the highly forceful steam that runs the turbine. The water from the primary loop becomes mildly radioactive from its contact with the core, but it is isolated from the water in the secondary loop. After transferring its heat in the steam generators, the water in the primary loop returns to the core. The steam in the secondary loop that drives the turbine is condensed back into liquid form and recirculated.
THE CAUSES OF THE ACCIDENT
The chain of events that set off the severe accident at TMI-2 and melted a substantial portion of its core began innocently enough at 4:00 a.m. on March 28. The initial problem occurred when pumps in the condensate polishing system tripped. After steam that drives the turbine is condensed back to a liquid state, it passes through the condensate polishers, which remove impurities in the water. This process is a part of the secondary loop. Operators at TMI-2 had been working for several hours to clear a blockage in one of the eight polishers when the system's pumps unexpectedly shut down for reasons that have never been determined. A polisher bypass valve that would have allowed the water to continue flowing failed to open. One second after the pumps quit, the main feed-water pumps that sent water to the steam generators automatically tripped in response to the cutoff of water from the condensate polishers. Immediately, according to design, the turbine tripped, shutting down the plant. As soon as the turbine tripped, auxiliary feed-water pumps came on. But the flow of water from the auxiliary pumps to the steam generators was blocked by two valves that had inadvertently been left in a closed position. At this point the secondary system was unable to provide water to the steam generators.
The closing of the secondary system caused heat and pressure to rise rapidly in the primary system, largely because the steam generators could no longer remove heat from the water that had come from the core. As a result, eight seconds after the polisher pumps tripped, the reactor scrammed automatically. The control rods entered the core and terminated the production of heat from nuclear fission. But the problem of dealing with decay heat remained, and it was greatly complicated when a critical valve, called a pilot-operated relief valve (PORV), stuck open. This permitted large volumes of cooling water from the primary system to escape. The earlier events in the accident were serious but not unprecedented, irreparable, or particularly alarming. The failure of the relief valve was the principal mechanical cause of what soon became a grave crisis at Three Mile Island.
The PORV sat on top of a large container called the pressurizer, which at TMI-2 was forty-two feet high. The pressurizer performs a vital function in PWRs: using electric heaters and water sprays, it regulates the pressure in the primary system. Maintaining proper pressure is essential not only for operating efficiency but also for safety. A sudden increase can damage pipes and other equipment in the primary system, including the pressure vessel that holds the core (the pressurizer should not be confused with the reactor pressure vessel). If the pressure in a reactor rises so rapidly that the normal operation of the pressurizer cannot handle it, the PORV opens automatically to reduce system pressure. At TMI-2, the PORV opened three seconds after the condensate pumps tripped, exactly as designed. Unfortunately, ten seconds later, after the temperature and pressure in the primary system had diminished, it failed to close as designed. The open relief valve allowed growing quantities of reactor coolant to escape. This was not the first time that the PORV had stuck open at TMI-2, and it was a chronic problem at Babcock and Wilcox plants. The same sequence of events had occurred at Davis-Besse in 1977. In that case, an operator recognized that the valve was open and immediately blocked it.
The operators at TMI-2, however, did not realize what had happened and did not promptly shut off the PORV. Within a few seconds after the accident began, the plant's alarm systems, including a loud horn and more than a hundred flashing lights on the control panels, announced the loss of feed-water in the secondary loop, the turbine trip, the reactor trip, and other abnormal events. But they offered little guidance about the causes of those occurrences and did not differentiate between trivial and vital problems. One of the operators, Craig Faust, later commented, "I would have liked to have thrown away the alarm panel. It wasn't giving us any useful information." To make matters worse, there was no clear signal to show the position of the PORV. A signal light that had been installed during start-up testing a year earlier showed only that electrical current was sent to the valve to open it; by inference the valve was closed when the current (and the signal) were off. The operators checked the signal on the morning of the accident, saw that it was not lighted, and assumed, therefore, that the valve had closed properly. The operators might have determined that the valve was open by looking at a pressure indicator for the reactor-coolant drain tank, which was where the water that poured out of the open PORV wound up. But that signal was situated behind the seven-foot-high instrument panels that were the dominant feature of the control room. The operators had to walk around the tall panels to look at the drain-tank indicator, and did not do so as they attempted to cope with the flurry of confusing signals they were already receiving.
The operators saw no definite signs that the plant was suffering a loss-of-coolant accident and was in danger of core "uncovery," in which the core is not fully covered with water. Their training programs had not prepared them for the conditions they confronted on the morning of March 28. The operators and supervisors on duty were well-qualified professionals, but they were baffled by the information they received. The two operators in the control room, Faust and Edward Frederick, were veterans of the navy's nuclear submarine program, had joined Met Ed in 1973, and had completed operator qualifying programs. The shift foreman, Fred Scheimann, who was in the turbine building trying to unclog a condensate polisher when the accident began, had also served in the nuclear navy. He had acquired fifteen years of nuclear experience and had worked at TMI for six years. Like his colleagues, the shift supervisor, William Zewe, had received his initial nuclear training in the navy. He had a total of thirteen years of nuclear experience and had been employed at TMI for seven years.
The navy provided the foremost talent pool for operators in the commercial nuclear industry, and it gave them solid training in the principles and procedures involved in running reactors. In addition, TMI-2 operators received training from Met Ed and from Babcock and Wilcox, which provided extensive experience on a reactor simulator. They were required to pass examinations administered by the NRC to qualify for operator licenses and to renew the licenses every two years. As a group, operators at TMI scored above the national average on NRC qualifying exams. Nevertheless, the experience and training of the operators on duty at TMI-2 when the accident occurred, and of the reinforcements that they soon called in, did not prepare them to cope with the deteriorating conditions in the plant. Their training courses and testing procedures placed much more emphasis on carrying out routine operating tasks, responding to minor malfunctions, and memorizing course materials than on developing the analytical skills needed to deal effectively with unanticipated problems. Operator training was not a high priority for the NRC or the nuclear industry, and the deficiencies in existing programs exacted a heavy price during the TMI-2 accident.
The fundamental source of confusion for the operators on the morning of March 28 was that the water level in the pressurizer was high but the pressure in the primary system was low. This condition occurred because water was leaving the core and escaping out of the primary system through the open PORV. The water level in the pressurizer rose as coolant flowed through it. There was no instrument in the control room that acted like a gasoline gauge in an automobile to show the amount of coolant in the core. Operators judged the level of water in the core by the level in the pressurizer, and since that was high, they assumed that the core was covered with coolant. They were confused by the seemingly contradictory signals that the water level indicator for the pressurizer kept climbing while the pressure in the core was low.
The operators' primary concern was not the possibility that the plant was experiencing a loss-of-coolant accident but the possibility that the pressurizer was "going solid." Under normal conditions, the pressurizer contains both water and a steam cushion that are used to maintain proper pressure in the primary loop. If the pressurizer goes solid, it fills with water, which eliminates the steam and severely impairs the means of controlling pressure in the system. The operators at TMI-2 had been trained by both Babcock and Wilcox and Met Ed to avoid letting the pressurizer go solid, and they were keenly aware that filling it with water was undesirable and perhaps disastrous. Zewe, the shift supervisor, later explained that "if you go solid, you worry about an overpressure condition; you also worry about an underpressure condition, too, and the uncontrolled aspect of it."
While the TMI-2 staff struggled to sort out conflicting signs and decide on appropriate actions, the plant's emergency core cooling system began to operate as designed. About two minutes into the accident, the high-pressure injection pumps, a part of the ECCS, automatically activated in response to the loss of coolant from the core. The two pumps fed water into the primary system at a rate of about a thousand gallons per minute, which was sufficient to make up for the coolant escaping through the open PORV. The high-pressure injection system, triggered by the low pressure and rising temperatures in the core, performed flawlessly. Despite the fact that the ECCS came on, the operators remained focused on their concern about the pressurizer going solid. In that context, the addition of a large volume of water to the primary loop was not a welcome development because it seemed to increase the chances that the pressurizer would fill with water. Therefore, about four and a half minutes into the accident, Scheimann, the shift foreman, ordered that one of the high-pressure injection pumps be shut down and the other sharply throttled back. He did so, he later recalled, because "pressurizer [water] level at that point was indicating that it was coming up at a rapid rate, and was rapidly approaching your solid indication." As a result, the plant lost much of a vital component of its defense against a loss-of-coolant accident.
The effects of the ill-advised decision to scale back on the flow of water from the ECCS were compounded when the operators also shut off the four reactor coolant pumps. The pumps were huge pieces of equipment, described in one report as each the "size of a small truck." They were a part of the reactor's primary system; their function during normal operation of the plant was to force coolant through the core. A little more than an hour into the accident, the pumps began to shake so furiously that the operators could feel the vibrations in the control room. This was a result of the rising heat in the core and the growing presence of steam in the coolant. The operators still did not recognize that they were dealing with a loss-of-coolant accident, and in accordance with their training, at 5:14 a.m. they shut down two of the pumps to prevent damage to them. At 5:41 a.m. they turned off the other two.
As long as the reactor coolant pumps were operating, they circulated enough water and steam through the core to keep it covered. After the pumps were closed down, however, the steam in the pressure vessel (which provided some core cooling) separated from the water and rose to the top of the vessel, the level of cooling water fell even further, and the fuel assemblies in the core soon became uncovered. At that point the plant was suffering the kind of loss-of-coolant accident that reactor experts had long feared and tried to prevent. As a consequence of mechanical failures and operator errors, what began as a series of minor malfunctions escalated into a major crisis.
In the first one hundred minutes or so of the accident, any one of a number of actions would have maintained adequate core cooling.
Who would have thought that this subject could read like a novel. The well-told story completely transcends its specific subject matter. It is a terrific tale of the dynamics of people, organizations and institutions and how they behave under stress. It should be required reading in the Organizational Management classes at business schools. The author provides an un-biased, beautifully written and comprehensive view of this frightening occurance.
Was this review helpful? YesNoThank you for your feedback.Report this reviewThank you, this review has been flagged.