When Science Goes Wrong (25 page)
The remaining portions of the men’s bodies – just a limbless, headless torso in Legg’s case – were now prepared for burial. The bodies were coated in drying powder, wrapped in cotton and then in plastic sheeting. A final wrapping was done with lead sheeting – one-eighth of an inch thick for McKinley and Byrnes, three-quarters of an inch for Legg. More than a ton of lead was used for this purpose. Thus ensheathed, the bodies were placed in hermetically-sealed steel caskets that had been lined with more lead, and the caskets were placed within purpose-made lead vaults. Radiation levels outside the vaults were now reduced to levels well below 1 roentgen per hour, and the bodies were deemed safe for transport.
Eighteen days after the accident, military planes carried the vaults to airports near the intended burial grounds. McKinley’s family chose Arlington National Cemetery on account of his service in Korea, as well as a sense that he had died in service to his country. Byrnes and Legg were buried in cemeteries in New York and Michigan respectively.
The funerals were bizarre affairs. In each case, on the day before burial an extra-deep grave was excavated and a foot-thick layer of concrete was poured and allowed to harden, forming a pad. On the day of the funeral, the rites were limited to eight minutes, with the grieving family members standing at least 20ft away from the vaults. Legg’s family insisted on the casket being taken out of its vault for the rites. This caused radiation levels to double, and the service was limited to five minutes. The vaults were then lowered onto the pads and, after the families had left, more concrete was poured so that the vaults were completely encased in foot-thick concrete – two feet thick, in Legg’s case. Finally, the concrete was covered with several feet of earth.
In their effort to establish the cause of the accident, investigators took two main directions: interviews with SL-1 staff and administrators, and examination of the accident site. The staff interviews produced limited information. The only eyewitnesses to the accident were dead, after all. The workers who had shut down the reactor before the Christmas holiday described how three of the control rods refused to fall freely into the reactor core and had to be driven down. This sticking behaviour was much worse, they reported, than had been experienced earlier.
Engineers described another problem that had developed with the reactor over its two-year operating lifetime. As part of the reactor design, strips of boron had been attached to the fuel plates. Boron, like cadmium, is an efficient absorber of neutrons, and the strips had the function of helping prevent a runaway reaction and extending the life of the core. Nevertheless, the strips had begun to corrode and flake away from the fuel plates, leaving the reactor more excitable than it had been when it came online. Thus, the control rods didn’t have to be raised as far as they did previously in order to start the nuclear chain reaction.
Because of these and other problems that had developed with the SL-1 reactor, officials had planned to replace the entire reactor sometime during 1961. When the accident occurred, therefore, the reactor was being operated on borrowed time. The crews had to keep recalibrating the control rods, for example, to ensure that the control motors moved them up and down within the range that controlled the reactor core. Nevertheless, no one believed that the reactor was threatening to get out of control. There still seemed to be a large margin of safety built into the various operations required for starting, stopping, and running the reactor.
Officials described the background and training of McKinley, Byrnes, and Legg. Because McKinley was a new trainee, there was some suspicion that he might have done something to trigger the accident. Attaching the control rods to the drives – the task that the men were engaged in at the time of the accident – required the rods to be raised by hand by about four inches. Raising the central rod by more than about 16 inches would trigger a runaway chain reaction (a nuclear ‘excursion’, as it is called). Both Byrnes and Legg were well versed in the procedure and knew not to lift the rods beyond the permitted four inches. Could McKinley, out of ignorance, have pulled the central rod much farther out than that?
During the first few days of the inquiry, this scenario seemed especially plausible on account of an unfortunate circumstance: in the haste and confusion of the recovery, the bodies were misidentified. McKinley’s body was originally thought to be the one that was impaled in the reactor room ceiling. This would have meant that he was standing directly on top of the reactor at the time of the accident, where he could have been handling the central control rod. In the course of the autopsy, however, it became clear that McKinley was the man who was pulled out first – the man who was still alive at the time the rescuers arrived. This man had been at some distance from the reactor at the time of the accident, thus he could not have been handling any control rod. Once the bodies were correctly identified, the idea that the accident was the result of a trainee’s mistake fell apart.
Much more information was obtained from the reactor site. During the 11 months following the accident, workers gradually tore down the reactor building while carefully documenting every item that was found. Many of the items, including the control rods, were extremely radioactive. These were taken for study to a ‘hot lab’ on the Testing Station, where they could be handled, cut up and inspected with remotely-controlled instruments.
The investigators first wanted to establish whether or not a nuclear excursion had in fact occurred. The fact that radiation levels in the reactor building were so high did not compel that conclusion: the radiation came primarily from nuclear fuel that had been ejected from the reactor, but the fuel might have been ejected as the consequence of a chemical explosion or some other event within the reactor vessel. To resolve this issue, radiation technologists took one of the dead men’s wedding rings and dissolved it in acid. They found that some of the gold atoms in the ring had been converted from normal gold,
197
AU, to radioactive gold,
198
AU – a transition that occurs by capture of an extra neutron. Thus, the presence of
198
AU was proof that there had been an intense flash of neutrons during the accident, and these neutrons must have been generated by an uncontrolled chain reaction within the reactor’s fuel elements.
The central control rod (minus its cadmium blade) was one of the early items to be recovered, because it was found lying directly on top of the reactor. Evidently it had fallen there after being violently ejected during the explosion and striking the reactor room ceiling. The blade had broken off and remained inside the reactor.
From the control rod, the investigators learned the exact moment in the proceedings when the accident occurred. To re-attach a control rod to its drive, the crew had to take the following steps, which formed an unvarying routine. First, they had to insert a handle into the top of the rod, effectively extending the rod by a couple of feet. Second, a crew member grasped the handle and raised the control rod by about four inches, while another worker attached a C-clamp to the rod. The rod was then lowered back down until the C-clamp rested against the housing through which the rod passed as it entered the reactor vessel. This held the rod in a fixed position for the following steps. The handle was removed and a nut and washer were attached to the top of the rod – these were required for attachment to the drive. Then the handle was reattached and a worker raised the rod just slightly – maybe a quarter of an inch -so that another worker could remove the C-clamp. The handle was removed, and the drive was connected to the top of the rod.
When it was recovered, the control rod had nothing attached to its top end. In fact, the very topmost part of the rod had been broken off and was missing. Then workers found this missing part in the attic above the reactor. Attached to this part was the washer and nut used for joining the rod to its drive, and attached to the washer and nut was the lifting handle.
The fact that the handle had been attached to the central control rod meant that the crew was working on this rod when the accident occurred. Furthermore, since the washer and nut were already in place between the handle and the rod, the workers must have already completed the first, four-inch, lift and attached the C-clamp and the washer and nut. They were now due to perform the second, tiny, lift that was required to release the C-clamp.
This wasn’t all that was learned from the central control rod. When it was found lying on top of the reactor, the rod still lay within its ‘shield plug’ – a kind of steel collar, normally fixed to the top of the reactor vessel, through which the rod could be moved up and down. Below the shield plug was fixed a ‘guide tube’: this formed a sheath around the rod, which moved up and down inside it. Evidently, the rod, the shield plug and the guide tube had been blown out of the reactor as a unit. When found, the rod was only slightly withdrawn from the plug, just as would be expected for the phase of the operation that was being performed at the time. This seemed to argue against the idea that someone had withdrawn the rod the 16 inches required to initiate a chain reaction in the core.
But quite a different story emerged when the rod and the plug were disassembled and examined microscopically. The control rod was peppered with small impact marks and scratches that had been caused by the explosion. So was the guide tube. It quickly became apparent that the individual marks on the guide tube could be lined up with corresponding marks on the control rod – but only if the top of the rod was withdrawn from the shield plug by 20 inches. Thus the rod must have been raised by that amount at the time of the explosion.
Apparently, someone or something had raised the rod much farther than was prescribed or permitted, and in fact four inches beyond the distance needed to set off an uncontrolled chain reaction. Evidently, when the rod/plug assembly was propelled violently against the ceiling, the tip of the rod (along with the handle) broke off, and the remainder of the rod was driven back down through the shield plug and guide tube, leaving it in a seemingly normal position.
The investigators confirmed the excessive withdrawal a few months later when workers were able to inspect the reactor core. The cadmium blade of the central control rod was trapped by the partially melted and deformed fuel plates, but it was 20 inches above its lowermost position. Thus the entire control rod assembly had been withdrawn by that distance at the time of the accident.
Why had the control rod been raised so far beyond the safe level – especially at a phase of the operation when it needed to be raised by only a fraction of an inch? The most obvious explanation was that the rod had become stuck, and a crew member – or two crew members working together – had tugged mightily to free it. Then, when the rod suddenly broke loose, they unintentionally pulled it far beyond the intended distance.
This scenario sounds entirely plausible: everyone has had similar experiences in the course of wrestling with balky mechanical devices, if not with such disastrous consequences. But could it really have happened at SL-1? To find out, the investigators built a replica of the control rod that could be locked and released at will. They had a large number of workers attempt to free the ‘stuck’ rod by pulling on it as hard as they could. When the rod was suddenly and unexpectedly released during the pull, none of the workers raised the rod more than about 10 inches. This was certainly an uncomfortably large distance, but it wasn’t enough to start an uncontrolled chain reaction in the core. The main reason that the rod didn’t come up very far when it was released was its weight of around 100lb. As soon as the workers’ initial effort slackened, it stopped dead or fell back.
The investigators considered another scenario, which was that one of the crew members had played a prank on the man who was tugging on the rod, perhaps by grabbing or pinching him in the rear. Might a reflex response have led the man to straighten up, causing him to pull the rod upward by 20 inches? To test this idea, the investigators actually perpetrated this prank on a number of men who were in the process of pulling on the mock-up rod. None of the men who were goosed in this way pulled the rod up by any extra distance, however.
Eventually, the identity of the crew member who raised the control rod became clear: It was Jack Byrnes. McKinley, as already mentioned, was not within reach of the rods when the accident occurred. Legg must have been crouching astride control rod number seven*, one of the outer control rods, because this was the rod that ripped through his body, sending him rocketing upward and pinning him to the ceiling. Because of his crouched posture, he was far too low down to have been pulling on the central rod, but he was at an ideal height to detach the C-clamp from it. Furthermore, the pattern of radiation absorbed by the bodies of Legg and Byrnes confirmed that Legg was crouching down, whereas Byrnes was standing above the centre rod in the normal position for pulling on it.
All in all, it seemed clear that the accident happened because Jack Byrnes, an experienced operator, raised the central control rod too far, but the exact reason for his action remained obscure, at least in the official report. The investigators wrote that ‘the reason or motive for the abnormal withdrawal is considered highly speculative, and it does not appear at all likely that there will ever be any reason to change this judgment’.