The entry says the reactor had been speaking for months in a language everyone at the plant understood. This piece goes through what it said, as the Atomic Energy Commission’s board of investigation and the contractor’s own interim report took it from the operating logs, the rod history logs, the night order book and the testimony of the people who kept them. Nothing in it caused the accident on its own. All of it shortened the distance a hand had to travel to cause it.

The margin as designed

A reactor built to run three years without refuelling has to start with far more reactivity than it can use, and the design of SL-1 held the surplus down in two ways: five cadmium control rods, and strips of boron fixed to the fuel assemblies, a burnable poison that was meant to burn away over the years at about the pace the uranium did. Argonne had wanted the boron inside the fuel plates. To get the plates made in time it was left out of them, and went on the outside instead, as thin strips of aluminium-boron alloy tack-welded along the side plates of the assemblies, full-length strips on forty of them and half-length strips on the bottom halves of the sixteen in the centre of the core. The boron was fully enriched in boron-10, the isotope that absorbs neutrons.

The hazards report Argonne wrote in 1958, ANL-5744, estimated how the reactivity of the core would move over its life as the boron burned out faster than the fuel: it would rise, pass through a maximum, and come back down. The maximum, it reckoned, would be about 0.8 percent above the fresh core, which it expressed, as the board’s Annex P puts it, as “one or two inches in the position of the central control rod in the region of this rod’s largest differential worth.” The design shutdown margin for the cold reactor, the amount by which the five rods fully inserted could hold the core below critical, was 4 to 5 percent in reactivity. The board’s reading of the early measurements was that the actual margin when the reactor was new was probably nearer 3.5 percent, “considered adequate.”

The critical experiments of August 1958 had also measured something that was true of the core for the rest of its life. With sixteen fuel assemblies in a square and no boron strips, and the four outer rods fully in, the reactor went critical with the central rod alone withdrawn 14 to 14.5 inches. Eight more assemblies made the core so much livelier that the central rod had to go back in to 9.25 inches to hold it critical. The full core of forty, with its boron, needed more of the rod out, not less of it, but the fact was recorded in the first week: “The reactor could have been made critical by withdrawal of the central control rod only, anytime after startup of the reactor.” Combustion Engineering noted afterwards that “the type of accident which actually occurred was not discussed in the Hazards Report and the consequences of a partial core meltdown and expulsion of fuel from the reactor vessel to the personnel normally in the facility was not evaluated.”

The boron

Combustion Engineering took over the plant on 5 February 1959 with, it said, three documents: the hazards report, a set of plant drawings that was not up to date, and “a brief Standard Operating Procedure.” It began a test run on 6 March and cold critical experiments on 30 March, and ran the reactor at 3 megawatts of heat, with shutdowns for maintenance in August 1959, January 1960, November 1960 and 23 December 1960.

The first inspection of fuel came in September 1959, when assemblies were pulled and looked at by the contractor’s people and Argonne’s. The bowing of the boron strips, in the three-inch lengths between tack welds, was first seen then. More inspections followed in October 1959, August 1960 and November 1960. The one in August 1960 is the one that mattered. “It was observed that large amounts of the boron strips were missing from some fuel elements,” the board found, and the assemblies at the centre of the core “were extremely difficult to remove, by hand. Removal caused plates to fall off and flaking of material. A considerable number of flakes were collected from the bottom of the vessel.” The conclusion drawn at the plant was that pulling the fuel out to inspect it was doing damage, and that “further removal of fuel elements might cause further loss of boron.” The inspections stopped and the reactor ran on. Nelson’s memorandum of May 1961 put this on his list of the things that might have prevented the accident: “The formal recommendation of a report on the loss of boron from the reactor core, after intensive review of the problem, was to terminate a previously established inspection routine of the fuel elements and to continue to operate the reactor.”

The boron that was gone showed up as reactivity. By 500 megawatt-days of operation, around mid-1960, the reactor “was gaining reactivity faster than predicted,” and the rods had to sit deeper in the core, at power and cold, than the hazards report’s curve said they should. The contractor’s estimate was that the boron had been worth about 1.1 percent of reactivity at the start of core life; a rough observation of the gain above prediction led to a rough estimate of how much of it was missing, on the assumption that the loss was uniform. The board, with the caution of people who knew how much depended on assumptions about a core nobody could now see, gave its own figures: at the shutdown of 23 December 1960, the shutdown margin for the cold reactor “was probably 2 to 3 percent, assuming rod worth was essentially unchanged from earlier measurements and calculations.” The Los Alamos reviewers of 2000, with General Electric’s later measurements in hand, put it at 3 percent, about four dollars of reactivity, against 3.5 to 4 percent when the reactor was new.

On 11 November 1960 the contractor pushed strips of cadmium into two of the empty T-rod shrouds, the channels Argonne had left for rods that were never fitted. The effect, measured at power, was about 1 percent in reactivity, two-thirds of what the lost boron had cost. The table of critical rod positions tells the story in inches. At 2.56 megawatts with equilibrium xenon the rod bank had to be at 17.8 inches in September 1960; after the cadmium went in, 19.3 inches in November. At the same time the contractor was proposing to raise the plant’s power. The Idaho office turned down 8.5 megawatts as an unacceptable hazard in radiation levels, and accepted operation up to 4.7 megawatts as “not a significant change,” which under the Commission’s manual meant that no review by the Division of Licensing and Regulation was needed.

The rods

The control rod drives had given trouble from the start. The first record is Argonne’s, in its report of December 1959 on the initial testing: to meet the scram-time requirement, which was ten inches of fall in a second and thirty in two, the drive of rod 7 had been replaced and one of the two constant-velocity springs removed from each of the five drives. The design had put the rack and pinion in saturated steam above the vessel, with water fed to a five-ring seal to keep the steam in; particles built up in the seals, the pinion bearings and the rack housings, and a rod that fell slowly could usually be brought back to specification by changing the seal water flow or taking a spring out. “The rate of flow of seal water affected the performance of the rod drives,” the board found, “as did the presence of foreign matter.” The chief operators had a mechanics’ speciality and a way with the problem, and the contractor was running a study of how the scram performance varied with seal water flow. It was “not considered a serious problem.”

The contractor sorted the sticking into three types: a rod that fell too slowly but did not need its motor; a rod that stopped and needed the motor to push it home; and a rod that could not be driven at all, by clutch slip or a sheared key. It counted sixty-three sticking incidents in drop tests or scrams in the two years of its operation, forty of them in the first twenty-two months. Set against the number of scrams and drop tests in those months, “approximately 2.5 percent” had gone wrong. “In the one month period prior to shutdown (November 18 to December 23, 1960) the incidents of sticking during scrams and rod tests increased markedly to approximately 13.0 percent.” The board’s list, Annex J, gives them rod by rod: a rod that stuck on the drive out, a rod that “dropped part way, stuck, then dropped to zero,” a rod that “would not drop at all on selected rod drop,” a rod that “had to be forced by hand.”

The contractor’s best reading was that the sticking on withdrawal came from the clutches, whose torque was low, failing to overcome the friction of the system and its misalignments and corrosion. It tested the hand assists that the operators had used four times, by its count, to free a rod: three men applying torque to the pinion coupling with one hand produced 125 to 147 inch-pounds, and with two hands 250 to 295, about what a new clutch could deliver; “a hand assist would not do more than apply a torque value that could be delivered by a new clutch.” On 19 December 1960 a pipe wrench was used to withdraw rods 1 and 5, which were sticking on their way out at 28 inches. The cadremen said that the wrench, hung on the coupling, turned it by its own weight enough for the clutch to pick up the load once the sticking point was passed.

Something else may have been happening by then, and the board heard two views of it. The boron strips were bowing outward from the fuel assemblies, and the central assemblies, each with two strips, had become hard to pull. One chief operator with a mechanics’ speciality testified that he believed the clearances inside the shrouds had closed and were pinching the blades, and he had a piece of evidence: a dummy aluminium rod made for the empty number 4 shroud could not be fully inserted until several inches had been cut off its bottom. Others felt there was no sign of the shrouds closing, only of crud on the surfaces, and that exercising the rods kept them free. The board did not decide. It noted that after November 1960 the cadmium strips had pushed the rods into a higher part of their travel than they had ever worked in, “a new region of the mechanical structure,” and that if the rods had begun to stick in the shrouds just before the Christmas shutdown, “the fact that sticking because of seal difficulties was an old and familiar problem might have been responsible for failure to recognize this latter development or to bring it to the attention of higher supervision.”

Rod 9, the central rod, had the best record of the five. The contractor counted 130 scrams in the six months before the last shutdown with one momentary hesitation; it had a new face-type seal fitted in November 1960; and in the lifting of rods for the washer and nut, “there is no indication either in the logs, or as stated by military and CEND operating personnel, of any rod ever sticking. One of the military crew members stated that he performed this operation approximately 300 times, and another more than 250 times with no sticking.”

The night orders

The night order book carried the plant superintendent’s instructions to the shifts. On 20 December 1960 he wrote: “Each shift will perform a complete rod travel exercise at approx. 4 hours after the start of shift. This rod exercising will be required of each shift until further notice.” On 22 December, with a special run in progress to get equilibrium data at 2.56 megawatts: “Do not perform control rod exercises during 2.56 MW power run.” The operators complied with both.

On the morning of 23 December the reactor was shut down for Christmas, and the log reads, in part: “0825 Dropping rods to secure reactor. Rod drop times #1 no drop. #3 dropped 1/2” and stuck. #5 clean drop in 0.82 sec. #7 no drop. #9 clean drop in 0.81 sec. 0827 Driving rods 1, 3, and 7 to zero. 0830 Controlling bypass steam flow to cool down to 2°F/min. 0835 Rod #3 dropped from 9” to 0 in .5 sec. Rod #1 dropped from 16” to 9” in 1.3 sec.” The assistant operations supervisor remembered commenting that this was worse than usual, and thought it was because the rods had not been exercised during the run.

The contractor’s project manager and the assistant director of its nuclear division testified to the board “that they were not aware of any significant difficulty with the operation of the control rods and also were not aware of the entries in the log books over the past several months describing these difficulties.” The board’s finding on the lines above them is in one bracketed paragraph: “The testimony indicates that the loss of boron was well known within the Division of Reactor Development at AEC Headquarters, although it was not categorized as a serious condition in the reports transmitted to Headquarters. The difficulties with operation of the control rods appears not to have been known at Headquarters, and very little knowledge of the extent of the difficulty was known by the AEC staff at ID.”

The procedures

The contractor had been asked, when it took the plant over in 1959, to rewrite the Army’s manuals, which the Army Reactors Branch had said were not satisfactory for the contractor’s use. The rewriting was still going on. “Reactor operating procedures, completely satisfactory to the AEC, have never been completed by Combustion Engineering, Inc., although they have been in the process of preparation and revision since mid-1959.” A reactor safety committee existed at the site, made up of the contractor’s supervisors and its health physicist, and it reviewed new test procedures; it “did not routinely review reactor operating experience or procedures unless specific problems were brought to it.” A full technical review of the reactor had been made once, in February 1959, by a committee from the contractor’s offices in Connecticut, and no other since. The Commission’s Division of Licensing and Regulation had reviewed the hazards report before start-up, as had the Advisory Committee on Reactor Safeguards, and neither was asked again. Visits by the Idaho office’s staff “did not include inspection of the nuclear safety of reactor operations.”

The last piece of paper is the letter of 29 November 1960, from the contractor’s part-time project manager to the Commission’s contracting officer’s representative, “requesting written confirmation of the oral agreement that CEI shift supervisors were not required for routine supervision of plant operation during the night shifts.” The contractor would supervise non-routine work on any shift, and the operators were encouraged to telephone the off-duty supervisors if anything unusual came up. The letter had not been answered on 3 January. The board found that the Commission’s own people believed night supervisors “would defeat a part of the purpose of operating the reactor under the existing arrangement, i.e., to obtain plant operating experience with only military personnel.” Reassembling the control rod drives was routine work.

Sources

• US Atomic Energy Commission, SL-1 Accident: Atomic Energy Commission Investigation Board Report, June 1961, sections 3 and 4 and Annexes H, J, K, N and P, with Curtis A. Nelson’s memorandum of 10 May 1961: the boron history, the shutdown margins, the cadmium, the rod drive experience, the night orders, the scram log, the testimony, the procedures and the reviews. inldigitallibrary.inl.gov
• Combustion Engineering, Inc., SL-1 Reactor Accident on January 3, 1961: Interim Report, IDO-19300, 15 May 1961, sections II B and II C: the transfer from Argonne, the three documents, the metallurgical and reactivity history, the three types of sticking, the sixty-three incidents and the 2.5 and 13 percent, the hand-assist tests of March 1961, the pipe wrench, rod 9’s record, and the 300 and 250 lifts. inldigitallibrary.inl.gov
• Thomas P. McLaughlin and others, A Review of Criticality Accidents, 2000 Revision, LA-13638, 2000, page 97, for the shutdown margins as later estimated. osti.gov