The operator positions the discharge machine over a hole in the floor, grabs a fuel cartridge down in the reactor, and pulls it up into the shield. The machine then rolls on rails to the end of the room, where the well hoist grabs the cartridge away from the discharge machine and lowers it on a cable down five floors, through the concrete-shielded discharge well. It drops into a lead barrel, which then rolls out on rails to the transit hoist for a trip to Windscale for fuel reprocessing. After you have discharged a few hundred thousand fuel cartridges, it seems routine and foolproof; but one day early in 2001, one fuel cartridge out of 10,176 had a problem.
The operator sent the cartridge down the discharge well and into the lead barrel. For some reason it did not release, and the cable came back up, ready to snag another cartridge after it had been snaked out of the core. Instead of an empty grabber, up came a live cartridge, broadcasting gamma radiation all over the refueling floor. Alarms jangled everyone’s nerves as they leaped from the chairs on their discharge machines and hot-footed for the door. Nobody was injured by the radiation, but it was considered a serious accident, and inadequate design, improper operation of the discharge machine, and statistical probability were blamed. All refueling of the early MAGNOX reactors at Chapelcross and Calder Hall was halted while the incident was investigated.
In July 2001, the workers at Chapelcross discovered that steel drums filled with depleted uranium trioxide tend to rust and develop holes when they are left out in the rain for several years. It was decided to substitute stainless steel barrels.
Later in July, corrections to the problem experienced earlier that year, when a fuel cartridge grabber would not let go, seemed to cause the opposite problem. After hearing something heavy crash into the door at the bottom of the discharge well, investigators sent a remote-controlled TV camera into the area and had a look. Twelve fuel cartridges had let go of the grabber and fallen over 80 feet into the transport barrel, splattering the irradiated uranium over the discharge bay. A careful cleanup was accomplished.
In August it was time for the annual maintenance and cleanup at Chapelcross, starting with Pile No. 1. There was a disturbing finding. It was known that graphite could “grow” in just about any direction under neutron bombardment at temperatures lower than 300 °C, but in Pile No. 1 the graphite had shrunken. The pile was now not quite as tall as it had been, and the steel charging pans, designed to guide new cartridges into the core when refueling, were now hanging in space, supported perilously by the nozzles on the burst-can-detection gear. The other piles were found to have similar defects, but not as bad as Pile No. 1, which was built differently.
Only Piles 2 and 3 were in decent enough shape to be restarted. In June 2004, the entire power station was shut down for good. The sister plant, Calder Hall at Sellafield, had been shut down permanently in 2003. The British graphite reactors had made a good run of it, but they were now obsolete and it was time to call it quits. By 2004, there was no longer a need to manufacture plutonium, and better ways to build a power reactor had been developed. On May 20, 2007, at 9:00 a.m., the four 300-foot hyperboloid cooling towers, visible on a clear day from a distance of 50 miles, were brought down in 10 seconds by demolition explosives, and that was that.
The fuel-reprocessing industry in the UK has reported only one criticality accident in which fissile material managed to come together accidentally in a supercritical configuration. Just about every other country that has tried to separate plutonium from uranium in spent reactor fuel has experienced at least one such excursion, and this was Britain’s. This incident at the Windscale Works on August 24, 1970, is described by the criticality review committee at Los Alamos as “one of the most interesting and complex because of the intricate configurations involved.” An entire book could be written to describe this accident, but I will be brief.
It is the goal of every chemical engineer working on a reprocessing plant design to allow no vessel large enough or of an optimizing shape, such as a sphere or tomato can, to exist in the complicated pipe-works of a reprocessing plant. Occasionally, however, when there is no conceivable way that fissile material, such as plutonium, can collect there, a round tank finds its way into the layout. At Windscale, plutonium was recovered from spent MAGNOX fuel, and near the end of the process it was refined in 300-gram batches. The plutonium was dissolved in a mixture of tributyl phosphate and kerosene and fed to a conditioner vessel, where the amount of plutonium in solution was adjusted to between 6 and 7 grams per liter. This was safely less than the concentration required for criticality.
From there, the small batch was lifted through a pipe to a transfer vessel by vacuum, where it would be fed through a U-shaped trap to a refining operation called the pulsed column. The transfer vessel was a short, cylindrical tank having hemispherical top and bottom. It was a near-perfect size and shape to house an impromptu reactor, but surely no such concentration of fissile material could be fed to it from the conditioner in one gulp, and the trap prevented any backflow into it. It was a mistake to think this.
It turned out that after every transfer of a subcritical amount of plutonium to the transfer vessel, a small amount of plutonium was stripped out of the solution by water sitting in the bottom of the tank. Ordinarily, that would be no problem, because the concentration of plutonium in the tank would never be greater than 7 grams per liter of solvent, but the amount of plutonium dissolved in the water grew gradually. This went on for two years, until the bottom of the tank held a whopping 2.15 kilograms of fissile plutonium-239.
On the last batch through the system, 30 grams of plutonium were dumped in on top of the water, mixing the solvent and water together and causing a supercritical nuclear reactor to suddenly exist in the plumbing for 10 seconds.
It was not a fatal accident. Nobody was injured, and the intricate piping in the plant looked exactly the same before and after the supercriticality, but the fact that it happened in such a carefully designed process was unnerving. It goes down in history as proof of how difficult it is to predict what will happen in a maze of pipes, valves, tanks, and traps carrying a fluid for which its volume and shape are important. The concept of impossibility becomes murky.
Windscale became Sellafield, and the THORP, or the Thermal Oxide Reprocessing Plant, became operational in 1997. It took 19 years to build the facility at the place formerly known as Windscale. Its mission is to take used reactor fuel from Britain, Germany, and Japan and separate it into 96 % uranium, 1 % plutonium, and 3 % radioactive waste, using a modified PUREX process.
In July 2004, a pipe broke and started filling the basement with highly radioactive, pre-processed fuel dissolved in nitric acid. The loss of inventory after nine months had climbed to 18,250 gallons. This went unnoticed until staff reported the discrepancy between solution going out of one tank and not arriving in another tank.
As it turned out, the liquid flow was monitored by weighing it periodically in an “accountancy tank.” This tank had to be free to move up and down as it accumulated the heavy uranium-plutonium waste in dissolved form, and the extent of sag as it filled was translated into weight. As the tank was being installed, it was decided to leave off the restraints that would keep it from wagging side to side as it accumulated liquid. This move, thought to protect it in case of earthquake, proved to be too much for the pipe connecting it to another process downstream, and it broke under the floor where it could not be seen.