The SL-1 accident was initiated by the rapid withdrawal of the central control rod. Starting from a fully shutdown condition, the action produced a condition in the core technically known as a prompt criticality.
When the SL-1 reactor achieved prompt criticality, a number of events happened in rapid succession. The core power level pulsed to nearly 20,000 MW, more than 6000 times as high as its rated power level. This power pulse lasted about 4 msec, ending due to a combination of void formation and fuel element heating. The total nuclear energy release was calculated to be approximately 140 Mw-seconds with an additional 24 Mw-seconds released in a chemical reaction between fuel and water.
The nuclear power transient caused some of the fuel material to reach its vaporization temperature of 2060 C (3740 F). This fuel boiling caused the fuel plates to swell and the cladding to fail, allowing approximately 20 percent of the fuel to be released.
A large steam bubble formed in the core, which lifted the mass of water above it at a rate of approximately 49 m/sec (160 ft/sec). This water hammered into the core head approximately 34 msec later, ejecting the head shielding and causing the pressure vessel to lift out of its support structure. It jumped nearly 3 meters to collide with the overhead crane before settling back into its original position. Essentially all water left the pressure vessel.
The vessel ejection sheared off all piping connections. The steam bubble also caused the pressure vessel to bulge, increasing its circumference in one place from 4.31 m (14.14 ft) to 4.627 m (15.18 ft).
While the pressure vessel was up in the air, the reactor was not shielded. Calculations indicate that the operators were exposed to an integrated neutron flux on the order of 1013 n/cm3. This explains why the victims became radioactive and could not be decontaminated.
Though the initiating event was a prompt criticality – an event which is at least theoretically possible in some other reactor designs – the consequences of the SL-1 accident were made worse because of several unique design features that proved to be mistakes.
The reactor was more reactive than planned due to problems with the aluminum-boron alloy burnable poison strips. Apparently, these strips had begun to rapidly deteriorate, causing some of the poison material to be lost from the core area. The physical deterioration of the poison strips also seems to have contributed to a growing problem with sticking control rods.
Highly enriched uranium helps to improve the compactness and responsiveness of a nuclear reactor. Often, this is a benefit, in that it allows reactors that can respond to rapid changes in power demand, but it can lead to more radical consequences in the event of a rapid insertion of reactivity.
Uranium-aluminum alloy has a much lower melting and vaporization point than the uranium dioxide that is commonly used as the fuel form in modern reactor plants. Uranium-aluminum will vaporize at 2060 C, uranium dioxide will not even melt until its temperature exceeds 2800 C.
Aluminum is not the optimal cladding or structural material in a reactor that must operate at elevated temperatures. It will corrode, deform and melt in accidents more readily than several other available materials.
The low design pressure for the vessel allowed it to catastrophically fail at a pressure that would cause little or no damage to most other reactor pressure vessels. Essentially, the SL-1 accident was a boiler explosion, a type of accident that has caused death and destruction since the beginning of the Industrial Age.
Any time there is water and a heat source in a confined space, there is the possibility of forming a steam bubble rapidly enough to cause a significant pressure surge. A 400 psi design pressure leaves little safety margin for a steam boiler containing a nuclear reactor.
The small number of rods made each rod far more important than it should be. One of the results of the SL-1 accident analysis was to ensure that all reactors adhered to the “one stuck rod” criteria that was already in use in several other reactor design programs.
One partial success story is that of the reactor building. Though not designed as a containment, it still retained essentially all of the non-volatile fission products. The total release was on the order of 100 curies of Iodine-131 and other gases. More than 99.99 percent of all fission products remained in the reactor building.
It is clear from reading the historical records that the SL-1 project was run by weak managers. The plant procedures had well documented weaknesses and inadequacies, but corrective action was not taken.
For a developmental program deemed important to national security, the project appears to have suffered from serious budget constraints and from being understaffed in comparison to the responsibilities assigned.
The plant managers claimed during post-accident hearings that they had no knowledge of sticky control rods, even though the problems were clearly documented in the engineering logs in a series of entries that began several months before the accident.
Despite knowing of problems with the burnable poisons and other growing difficulties with the reactor core performance, the decision was made to continue operating the plant until the new core was ready. The managers and others in positions of responsibility apparently did not want to take any action that would slow down the project or the deployment of the production reactors.
Many good lessons were learned following the SL-1 accident. Organizational lines of responsibility were more clearly defined, reactor design codes became more useful, operational oversight was increased and several sensible new criteria (like the “one stuck rod”) were implemented.
Unfortunately, poor understanding of the specific causes of the accident has slowed the development of small, distributed nuclear power stations. The idea of small, lightweight nuclear power plants that can perform a variety of functions with a small operating crew is a good one. It should not be lost because the first people who tried it failed to understand the need for durability and robust system design.