The Fermi Chronicles - Part 17: Nuclear Events - Windscale, UK, 1957

As with the Canadian NRX reactor (and subsequent nuclear event in 1952), the Windscale reactor in Cumberland, England (now Sellefield) was an experimental reactor commissioned in the mid-1940s not to investigate the possibility of civilian uses for nuclear power, but rather to manufacture plutonium for nuclear weapons (the cold war was being waged after all). This experimental reactor was gas-cooled (GCR), which is definitely my least favorite design. It was also graphite moderated. Unfortunately, graphite can react to neutron flux at low levels by storing potential energy that at some point could be spontaneously released as heat. This is now known as the Wigner Effect. To release the potential energy stored in the dislocation of its crystalline molecular structure, an annealing process must be initiated periodically. Basically, the material has to be heated to a certain temperature, allowing the crystalline structure to expand and realign, and then cooled very slowly. This process was initiated at Windscale but the reactor, and monitoring instrumentation, simply were not designed for such a process.

To initiate the annealing process at Windscale, the fan speed was reduced through the graphite. Then, the reactor power was increased, but temperature sensors were misplaced in the core. Since the reactor was not designed for the process, damage to the fuel resulted (the fuel was metallic uranium, which has a lower melting temperature than uranium oxide), but was not picked up by any instrumentation. When control rods were inserted back to reduce the power, the energy release decreased faster than anticipated. The thought was that the annealing process was not completed, so the control rods were withdrawn once again. The annealing process began on October 7, 1957. By October 10, operators knew something was awry. Core temperature continued to increase rather than decrease. Radiation sensors at the top of the discharge stack were pegged out. Air samples a half-mile away were 10 times normal! Two operators donned protective clothing and removed an inspection plug where a thermocouple was reading an unusually high temperature. Through the plug hole they could clearly see to their horror that several channels of fuel were glowing cherry red, and that the core had been on fire for at least 48 hours! Turns out, the uranium from the damaged fuel during the annealing process reacted with the graphite and caught fire. Increased air flow simply fueled the fire more. Finally, the core was flooded with water, even though there was a danger of hydrogen generation that could have led to an explosion.

For good reason, the GCR design was never built again. A lot of radioactive material, about 10,000 Curies worth (mostly radioactive iodine), was released into the atmosphere. However, no one was injured or killed in the incident. As a precaution though, radiation levels in the surrounding area as well as food and milk were monitored closely. Also, milk was banned in the area for almost a month as that is where the radioactive iodine would be most likely to accumulate.

As with the NRX reactor in Canada, much was learned from this event. For one, no more GCRs! (that's one most websites don't list in the "things learned" section). Liquid is far more controllable than gas. In addition, the second annealing process was done without a procedure once it was noted that the core didn't behave as expected. Today, when something occurs unexpectedly, everything stops and evaluations take place. Also, much was learned about proper instrumentation. Also, graphite was seen not to be a good candidate as a moderator, although the Russians still used it (Chernobyl was graphite-moderated and the graphite fire there contributed the majority of contamination released into the atmosphere). Some lessons are hard to learn I guess. In addition, the need for an emergency plan for the plant and surrounding area was clearly seen. This is standard at every plant now. The sharing of information between plants was not done back in 1957 as there was still a cloak of secrecy surrounding each nuclear plant. Lessons learned at Windscale could have prevented the 3-mile island incident 20 years later. Again, some lessons were tough to learn...



UPDATE (11/1/2010): YouTube videos of the 'disaster:'

Previously:
The Fermi Chronicles - Part 16: Nuclear Events - Chalk River, CAN, 1952
The Fermi Chronicles - Part 15: The Nuclear Business Model
The Fermi Chronicles - Part 14: Neutron Moderation
The Fermi Chronicles - Part 13: Nuclear Reactor Types
The Fermi Chronicles - Part 12: Generating Electricity
The Fermi Chronicles - Part 11: Worldwide Uranium Availability
The Fermi Chronicles - Part 10: Utilizing Nuclear Reactions To "Breed" More Fuel
The Fermi Chronicles - Part 9: Nuclear Fission
The Fermi Chronicles - Part 8: Neutron Interaction
The Fermi Chronicles - Part 7: Radioactive Decay and Half-Life
The Fermi Chronicles - Part 6: Atomic Structures
The Fermi Chronicles - Part 5: Nuclear Waste Storage
The Fermi Chronicles - Part 4: Radiation Types and Radiation "Dose"
The Fermi Chronicles - Part 3: Radiation Types
The Fermi Chronicles - Part 2: A week of training
The Fermi Chronicles - Part 1: The alpha post