Three Mile Island: The Inside Story

Digging Deeper into the Wreckage

Figure 10.1. This drawing of the interior of the TMI-2 reactor vessel represents the state of knowledge attained by 1989. Learn more

By April 1984, five years after the accident, the fact of the total destruction of the upper 30-40 percent of the core of TMI-2 was accurately known and fully accepted as a result of the sonar and video inspections. Yet almost nothing was known about the condition of the lower 60 percent. This appears graphically in Figure 9.1, published in April 1984, which gives a drastic depiction of the upper part of the core, while representing the lower part of the core as entirely undamaged. In fact, the actual condition of that lower 60 percent, while entirely different, was in important respects far worse than the mere void that obtained above it.

It would require another three years for a fairly complete picture to emerge of the condition of the lower core—and with it a good understanding of the course of the accident. The conclusion that a substantial part of the material then resting in the lower half of the core was once molten was resisted until the evidence was beyond argument. This was also the case with the further and still more alarming conclusion that in the course of the accident some twenty tons of that molten material had spilled into the bottom of the reactor vessel.

Rather, the long preferred supposition was that the production of the void in the upper part of the core involved no melting of fuel rods—or, anyway, no melting of the uranium in them—or, at least, very little melting of uranium. The views varied, but all excluded substantial melting. Instead, the preferred explanation of the void was that the portion of the fuel rods standing above the water level in the reactor vessel, being extremely hot, shattered when flooded with cold water at 7:20 a.m. And this view, which was partially, but only partially true, made it easy to continue to picture the fuel assemblies as essentially intact underneath the rubble of their shattered upper portions.

In September and October 1983, shortly after the sonar survey, and again in March 1984, long-handled tools—also designed at INEEL—were inserted through the same control rod channel and retrieved samples of the material lying on the floor of the cavity, and lying as far beneath the cavity as these probes could be thrust into the bed of rubble (ref. 16, p. 852).

Figure 10.2. Retrieving samples of the debris on and below the floor of the cavity, autumn 1983. Learn more

Not immediately, but eventually the study of those samples led INEEL metallurgists early in 1985 to the conclusion that a substantial amount of fuel had melted. Yet, to Dennis E. Owen, who led this effort, it seemed in retrospect that the acceptance of substantial melting came not from the chemical and metallographic analysis of the material retrieved, but from the in-vessel video images showing rounded edges on the wreckage and other obvious visual signs of melting having occurred (telephone conversation, April 29, 2004). In this persuasiveness of the pictures there is a peculiar irony, for those obvious signs were there to be seen from the earliest video inspections. Yet without a willingness to believe the conclusion, even the pictures were not persuasive.

After the overhead crane was declared operable, the reactor vessel’s head was unbolted and lifted away in July 1984. Then a much wider range of investigations became possible, as we have seen in Section 9. In December, when the upper grid structure was jacked up, a video camera was snaked down between the core former (the side wall of the core) and the inner wall of the reactor vessel. The camera could see large quantities of debris lying at the bottom of the vessel. By the spring of 1985, INEEL researchers were describing that debris as “slag-like.” And in July 1985, another of their cleverly designed long-handled tools grabbed a bit of material lying in the bottom of the vessel (ref. 16, p. 853). But the conclusion that the debris had gotten there in molten form—by pouring—was not yet drawn.

In July 1986, core-boring equipment adapted from the petroleum industry brought up 2½-inch (6.4 cm) diameter cores from the reactor’s core (ref. 16, p. 854; ref. 12, ch. 5, p. 18). The custom made bits—tungsten carbide faced with artificial diamond—were good for only one core, so hard and abrasive was the solidified slurry of grains of uranium dioxide ceramic imbedded in an alloy of steel, zirconium, uranium, and miscellaneous other metals. (M. R. Martin, conversation April 20, 2004.) Three months later, after unsuccessful efforts to dig out this consolidated material, the boring machinery was set up again and provided with solid-faced bits for a “Swiss cheesing operation” that reduced the mass to removable rubble (ref. 12, appx A, p. 19).

Figure 10.3. Core boring produced unequivocal evidence of a mass of once-molten material in the center of the core. Learn more

By the end of 1986 it was indubitable that much of the core had melted, but it would be two years and more before it was clear what a serious danger that had occasioned. After “Swiss cheesing,” the mass of solidified material in the lower part of the core was laboriously cut and chipped apart, and then fished and sucked up out of the water-filled reactor vessel. Then the structures supporting the core had to be cut out before access was gained to what lay in the bottom of the vessel. Only then, early in 1989, was it discovered just what that “slag-like” material lying in the bottom of the reactor vessel was—and thus, also, that there had been one really horrific moment in the course of the accident. That was the moment portrayed in Figure 3.7 when twenty tons of molten fuel broke out of the center of the core and cascaded down the side of the reactor vessel into its bottom.

Figure 10.4. Overall view of deployment of plasma arc (MDM = “metal disintegration machining”) apparatus for cutting prismatic samples from inside of bottom of TMI-2 reactor vessel. Learn more

Thus the last question to be investigated, after all that rubble had been excavated, was the condition in which the bottom of the reactor vessel had been left by that thermal assault—why it had withstood so unexpectedly well the softening that megawatts of heat at so high a temperature were expected to effect. Special plasma arc cutting equipment—designed, as always, to be operated from above the reactor vessel and under water—carved prismatic samples out of the inside of the 5-inch (13 cm) thick vessel wall. With the utility pushing, as always, to move as quickly as possible to complete the cleanup and put an end to that expensive operation, 15 samples were extracted in 30 days early in 1990, working 24 hours a day, 7 days a week (ref. 7). These specimens were then distributed to laboratories around the world for analysis, in a research program that was concluded only in 1994 (ref. 25). In all, TMI showed that—contrary to common belief—a disaster inside a nuclear reactor does not necessarily lead to a disaster outside the reactor.