Smithsonian - National Museum of American History, Behring Center
Three Mile Island
Unit 2 nuclear power plant

Three Mile Island: The Inside Story

The first looks inside the reactor

First entry into the reactor containment building, summer 1980.

Click to enlarge imageFigure 5.1. First entry into the reactor containment building, summer 1980.

Learn more

In the months following the accident, “the actual condition of the core was the subject of intense, often heated, speculation.” And yet, as this General Public Utilities engineer in charge of much of the cleanup acknowledged as the job was nearing completion, “the most severe prediction was short of the mark” (ref.23). The prevailing view among the reactor engineers was that “most of the core damage was to the cladding”—i.e., to the zirconium alloy tubes holding the uranium oxide fuel pellets—and “damage to the fuel pellets . . . was minimal at Three Mile Island” (ref.13, p. 33). Although the independent inquiry instituted by the Nuclear Regulatory Commission surmised that some fuel had melted and run down the channels between the fuel rods, it concluded, “Despite this amount of damage, a core meltdown, as normally considered, did not occur” (ref. 29, v.2, pp. 535-36).

To some few experts, one piece of evidence strongly suggested that a considerable part of the fuel in fact lay in the bottom of the reactor vessel, where none should be. A few hours into the accident the neutron detectors toward the bottom of the inner wall of the cylindrical concrete shield surrounding the reactor vessel (solid black in Figure 2.6) began reporting anomalously large numbers of neutrons. As only the uranium fuel itself, not any of its highly radioactive fission products, could be a significant source of neutrons, the onset and continuance of such high readings from detectors in the vicinity of the bottom of the reactor vessel pointed to a “relocation” of a considerable part of the fuel. But as no one could imagine that a considerable part of the fuel had melted, there was no plausible explanation for how such a “relocation” could have taken place.

The “relocation” hypothesis was so much at variance with the tendency of nuclear power plant engineers and operators to take at every stage an overly optimistic view of the seriousness of the damage to the core, that relocation was not accepted until finally it could be seen by video cameras some six years after it had occurred. And even when seen and accepted as fact, the conclusion that it had poured down molten was not drawn.

Nothing but speculation was possible without inspection of the interior of the reactor pressure vessel. The first step in this direction, entrance into the containment building, was not possible until—more than a year after the accident and amid great but unwarranted public outcries—the radioactive krypton gas, which had been released from the ruptured fuel rods and had accumulated in the sealed containment building, was exhausted into the atmosphere.

The first results on the condition of the reactor’s core came from analyzing debris accumulated on filters in the reactor’s cooling-water system. The fact that “only” 6 percent of this material was uranium, while most was zirconium (the main component of the tubes surrounding the uranium fuel) along with the relatively low melting point elements from which the control rods were alloyed, was construed optimistically as indicating some melting of control rods but not of fuel rods (ref. 19).

The first active intervention to test the conditions within the reactor’s core—the operation of the so-called axial power shaping rod (APSR) assemblies—took place in June 1982, two years after first entrance into the containment building. The APSRs are, in effect, a special sort of control rod. Although they play no role in starting up and shutting down the reactor as control rods do, the APSR contain, like control rods, not fuel, but neutron-absorbing elements. Those elements are so chosen and so distributed along the 12-foot (3.5 m) length of an APSR that power generation along the axis of the reactor’s core—i.e., from top to bottom—can be made more uniform by appropriate degrees of insertion of the APSR assemblies. (Because the concentration of neutrons is greater at the center of the core, fuel in the center “burns” more rapidly and generates more heat. Hence the need for a corrective.)

An axial power shaping rod (APSR) assembly comprises 16 rods, held by a "spider" attached to a leadscrew.

Click to enlarge imageFigure 5.2. An axial power shaping rod (APSR) assembly comprises 16 rods, held by a “spider” attached to a leadscrew.

Learn more


At the time of the accident, TMI-2 was operating with all eight of its APSR assemblies so positioned that 75 percent of the total length of the rods was inserted into the core, while 25 percent remained in the space in the reactor vessel above the core, the so-called upper plenum. No steps were taken to raise or lower the APSR at any time during the accident or in the months afterward.

In initial planning, looking inside the reactor vessel seemed difficult to accomplish. But it was thought that the operation of the APSR, if carefully prepared and extensively instrumented with accelerometers and vibration analyzers, could provide reliable information about the condition of the core. Such instrumentation was tested in an engineering laboratory and then in TMI-1. In June 1982, the TMI-2 APSRs were activated and driven down into the core as far as they would go. The result of doing so is represented by the thin black rods hanging down into the cavity in the Lucite topographic model pictured on the home page and in Section 8.

Axial power shaping rod dynamic test.  There were eight APSR assemblies in TMI-2, roughly equally spaced around a circle about two-thirds of the distance out from the axis of the core.

Click to enlarge imageFigure 5.3. Axial power shaping rod dynamic test. There were eight APSR assemblies in TMI-2, roughly equally spaced around a circle about two-thirds of the distance out from the axis of the core.

Learn more


Although four of the eight assemblies were practically immovable, four did move freely inward to the full extension of their drive mechanisms. Again this was, at first, interpreted optimistically “as proof that the core was not badly damaged” (ref. 12, ch.8, p.12; similarly ref. 20 as optimistic). When, however, the written reports on this test were prepared five to ten months afterward, the interpretation was reversed: the ability to move four of the APSR drives freely “probably results from the nearly total breakup of the fuel assemblies and the APSR assemblies at these locations, such that only the APSR leadscrews, the spider and perhaps the uppermost portion of the APSR clusters were actually moving” (ref. 10, p. 38).

What brought about this 180-degree reversal of the conclusions initially drawn from the APSR test? It was a look inside the reactor vessel. This took place in July, 1982, just a month after the APSR tests—much sooner than had initially been thought feasible.

The concept of “lowering a video camera through a Control Rod Drive Nozzle and manipulating it around the plenum and down to the top of the reactor core” had emerged early on as a way to get beyond mere “analysis and speculation as to the condition of the reactor core.” But these quotations from a 1981 planning document (ref. 11, abstract) do not envision getting a camera down into the core.

Contemplated at that early date was the removal of an entire 2¾ inch (6.8 cm) diameter control rod drive mechanism, an operation that would require the removal of the heavy “missile shields” above the reactor, and that, in turn, would require operation of the heavy-duty polar crane at the top of the reactor building. But after a year of being exposed to the high levels of radiation that existed in the building before the purging of the accumulated krypton gas, and to the 100 percent humidity at higher than tropical ambient temperatures—not to mention the near explosion of burning hydrogen that had occurred several hours into the accident—the overhead crane needed to be tested carefully before doing any heavy lifting. Thus video inspection seemed some while away.

The Technical Assessment Advisory Group, experts who had been assembled by General Public Utilities to advise it on problems of cleaning up after the accident, pointed out that lifting of the overhead missile shields would not be necessary if, rather than removing an entire control rod drive mechanism, only the control rod leadscrew were removed—removed by cutting it into short lengths as it was screwed out of the reactor vessel. This would create a 1½ inch (3.8 cm) diameter access hole in the top of the reactor vessel.

The lowering of a video camera down a control rod guide tube near the center of the reactor.

Click to enlarge imageFigure 5.4. The lowering of a video camera down a control rod guide tube near the center of the reactor.

Learn more


There were limitations to the “Quick Look” concept. While removing an entire control rod drive mechanism would have allowed the camera an open view of the plenum, removing only the leadscrew meant that the (much smaller) camera would be inside the leadscrew’s guide tube down to almost the top of the core. And this was all the greater a limitation as hardly anyone expected that the camera would be able to get any farther down than the top of the core. Still the undertaking seemed worthwhile even if all that one could see was the tops of fuel assemblies in the immediate vicinity of the control rod guide tube. A test of the concept in TMI-1 showed that this much was feasible (ref. 9).

The first “Quick Look” video examination, July 21, 1982, is recalled as a dramatic event by those present, and by many who only heard others tell of it. “There were many exclamations of surprise, disbelief, and confirmation,” as the video camera was lowered below the level of the top of the core without anything at all coming into view (ref. 12, ch. 5, p.13). The camera was able to see only 3 inches (8 cm) ahead, due to its small size, incorporating lens and light in one unit, and the turbidity of the water in the reactor vessel. Not until five more feet (1.5 m) of cable had been paid out did the operator call out, “Got something!”—rubble lying at the bottom of the cavity in the core (ref. 18, p.40). A steel probe inserted down through the same access port showed that the layer of loose rubble was about a foot (0.3 m) thick (ref. 12, ch. 5, p.13).

With two further entries in the following weeks, a total of six hours of videotape were obtained from two control rod positions, one at the center and one somewhat off center. These videos then were reviewed by a panel of experts. And here once again the tendency was to draw optimistic conclusions where the video did not clearly show the contrary. What the camera could not see was supposed to be still OK. The conclusion drawn was that the damage was confined to “the top center” of the core (ref. 1, abstract). Likewise, the fact that the “Quick Look” operatives had failed to get a camera into the core through a third channel created by removing a control rod lead screw that was still farther from the center of the core “led to the belief that the only damage was in the center of the core and that all peripherals were intact” (ref. 12, ch. 5, p. 13).

"Quick Look" conclusions.

Click to enlarge imageFigure 5.5. “Quick Look” conclusions.

Learn more


This hopeful conclusion was not, however, the only one drawn from “Quick Look.” The authors of ref. 10 inferred, as quoted above, the nearly total breakup of the fuel assemblies in the vicinity of the APSR even though the camera could not see as far out from the center of the core as the location of the APSR.