(from Operational Accidents and Radiation Exposure Experience Within the United States Atomic Energy Commission, 1943-1970, U.S. Government Printing Office: Washington, D.C., 1971.)
[ < Table of Criticality Accidents] [ ^ Trinity Atomic Home Page] [ > Radition Exposures]
In the AEC's operational activities (not licensed) for the past 28 years there have been a total of 26 occasions (see Chart XV) when the power level of fissile systems became uncontrollable because of unplanned or unexpected changes in the system reactivity. On three occasions, the power excursions were planned; however, the fission energy released during the excursion was significantly larger than was expected. There have been a total of six deaths attributable to criticality accidents. The property damage resulting from these excursions has been approximately $4,455,000; however, 98% of the property loss was due to the SL-1 reactor excursion.
Further study of this accident record reveals that nine of the unplanned excursions occurred behind heavy shielding and three of them occurred in facilities remotely located with respect to personnel. Hence, the probability of injuries to people was reduced almost to the vanishing point. It is also noted that fourteen of the accidents occurred during experiments, six occurred in production or processing facilities, and five in reactor activities. In these laboratory, production, and reactor facilities there were, respectively, two, one, and three fatalities.
A review of these incidents has been made by W. R. Stratton, University of California, Los Alamos Scientific Laboratory, Los Alamos, N. Mex. All we have done below is to prepare a brief description of each incident.
Unexpected criticality was achieved in a volume of an aqueous solution of a salt of U235 during a series of routine critical experiments in progress in a well-shielded assembly area of a critical experiments facility. The criticality radiation alarm system functioned as designed, the evacuation of personnel from the building was prompt and orderly, and the excursion was terminated expeditiously by a negative coefficient of reactivity and was prevented from recurring by the action of the safety devices. The fission yield was 1.1 X 1016. Gamma-ray sensitive personnel dosimeters read immediately following the excursion showed no direct exposure greater than 5 mr to any person present. There was no property damage or loss of fissile materials. An estimated 100 cm, of solution (15 g of U) were spilled when a rubber-stoppered connection immediately above the sphere was dislocated.
The purpose of the particular experiment in progress was to establish the critical concentration of a sphere of the solution of uranyl nitrate surrounded by a thick water reflector. In the course of approaching criticality by incremental additions of solution, a small volume of air was observed entrapped in a flexible transparent tube. Supercriticality occurred during an attempt, by remote manipulation of liquid levels, to remove the air.
A nuclear excursion of 4 x 1016 fissions took place in the critical mockup of a high power density reactor. There was neither damage to the equipment nor significant exposure of persons; nevertheless, the incident indicated poor practice and an undesirable interpretation of operating procedures which has been corrected. The reactor mockup is fueled with elements composed of fully enriched uranium in a graphite matrix, and a smaller number of graphite moderating elements. This permits a relatively small core volume (250 liters). The core, housed in a graphite cylinder, drops out of its Be reflector for loading. Control and safety drums are within the annular reflector.
Before the incident, fuel along the core axis was replaced by additional moderating elements to investigate flux-trap effects. Instead of the usual step-wise interchange of elements, the entire moderating island was installed. Then, instead of stepwise multiplication measurements while inserting the core into the reflector, which is proper for initial approaches to criticality, there were no measurements during interrupted insertion. It had been inferred from the behavior of different moderating elements in an earlier mockup that the overall reactivity change would be minor. This was a serious mistake, for the actual change proved to be about $10. Before complete closure was achieved, a very short period and scram (dropping the core and actuating the safety drums) occurred.
A nuclear excursion and subsequent fire took place during a subcritical experiment in a shielded vault designed for critical assembly experiments. The excursion was estimated at 4 X 1017 fissions and was followed by oxidation of the enriched uranium metal in the assembly.
The cause of the excursion is believed to have been directly attributable to mechanical failure.
The total property loss was $94,881.
An unplanned nuclear excursion occurred in a plutonium processing facility because of the inadvertent accumulation of approximately 1500 grams of plutonium in 45-50 liters of dilute nitric acid solution in a 69-liter glass transfer tank. The sequence of events which led to the accumulation of the plutonium the tank cannot be stated positively. However, it is believed that, when a tank valve was opened, the solution from another process vessel overflowed to a sump and was drawn into the transfer tank through a temporary line between this tank and the sump.
When the excursion occurred, radiation and evacuation alarms sounded. All but three employees left the building immediately, according to well-prepared and -rehearsed evacuation plans. Fortunately, they were not in close proximity to the involved system nor in a high radiation field.
The course of the nuclear reaction involved initial criticality (1016 fissions); a subsidence; one or more later peaks; and after approximately one-half hour, a declining rate of fission, which terminated in a subcritical condition 37 hours later. The total number of fissions was approximately 8 X 1017.
Of the 22 persons in the building at the time, only four employees, those who were in the room with the system, were hospitalized for observation. Three of them were the system operators, who were in close proximity to the excursion, and who received estimated radiation doses of 110, 43 and 19 rem. None of them showed symptoms definitely referable to their radiation exposures. The fourth was sent to the hospital only because he was in the room at the time of the incident.
Some fission product activity, airborne via the vent system and the exhaust stack, was detected in the atmosphere for a brief period after the accident. The physical damage amounted to less than $1,000. (See TID-5360, Suppl. 4, page 17.)
A criticality excursion occurred as enriched uranium metal, neutron-reflected and -moderated by hydrogen, was being assembled. The excursion was caused by a too rapid approach of the two pieces of metal used in the experiment.
There was no personnel exposure or property damage. The energy release was estimated to be between 1015 and 1016 fissions. Fission product contamination, both airborne and contained in the metal, decayed sufficiently overnight to allow unhindered continuation of the experiment.
The incident occurred in a critical experiment laboratory specifically designed to accommodate such occurrences, since events of this nature cannot be considered entirely unexpected in an experimental facility of this sort. (See TID-5360. Suppl. 4, p. 14.)
A nuclear excursion of approximately 6 X 1017 fissions occurred in a first-cycle product evaporator at a chemical processing plant. The criticality accident resulted when a solution of enriched uranyl nitrate accidentally surged from a geometrically safe section of the evaporator into the upper critically unsafe, vapor disengagement section. The accident occurred behind thick concrete walls in a processing cell which is part of the first cycle for processing highly radioactive spent-fuel elements.
Personnel response to the radiation alarms and the evacuation signal was prompt and orderly.
Analyses of badges from 65 individuals indicated a maximum exposure of 55 millirem gamma and 0 beta. The maximum thermal neutron exposure detected in the badges analyzed was less than 10 millirem. Analyses of nuclear accident dosimeters indicated that there was negligible fast neutron flux associated with personnel exposures.
The radioactivity released to the atmosphere as a result of the accident was about twice normal background when it left the area. Loss of $6,000 resulted from cleanup of the, incident. (See TID-5360, Suppl. 4, p. 9; 1961 Nuclear Safety, Vol. 3, #2, p. 71.)
A nuclear excursion occurred within the reactor vessel, resulting in extensive damage of the reactor core and room, and in high radiation levels (approximately 500-1,000 rem/hr) within the reactor room.
At the time of the accident, a three-man crew was on the top of the reactor assembling the control rod drive mechanisms and housing. The nuclear excursion, which resulted in an explosion, was caused by manual withdrawal, by one or more, of the maintenance crew, of the central control rod blade from the core considerably beyond the limit specified in the maintenance procedures.
Two members of the crew were killed instantly by the force of the explosion, and the third man died within two hours following the incident as a result of an injury to the head. Of the several hundred people engaged in recovery operations, 22 persons received radiation exposures in the range of three to 27 rem gamma radiation total-body exposure. The maximum whole-body beta radiation was 120 rem.
Some gaseous fission products, including radioactive iodine, escaped to the atmosphere outside the building and were carried downwind in a narrow plume. Particulate fission material was largely confined to the reactor building, with slight radioactivity in the immediate vicinity of the building.
The total property loss was $4,350,000. (See TlD-5360, Suppl. 4, p. 8; 1962 Nuclear Safety. Vol. 3, #3, p. 64.)
A nuclear incident occurred in a process equipment waste collection tank when an accidental transfer was made of about 200 liters of uranyl nitrate solution, containing about 34 kilograms of enriched uranium (91 percent U235), from safe process storage tanks to a geometrically unsafe tank through a line formerly used for waste transfers.
Limited visual inspections and test that no significant, property damage or loss resulted beyond the approximately $60,000 cost to recover contaminated uranium solution resulting from the incident.
Of the 21 personnel directly involved in this incident, seven received
external exposures to radiation. The exposures were 8, 6, 3.95, 1.50, 1.38,
1.17, and 1.17 rem. Two individuals also received external exposures to the
skin of 50 rem and 32 rem. No medical treatment was required for the 21
personnel involved. (See TID-5360, Suppl. 3, p. 14; USAEC Serious
Accidents Issue #163, 4-18-60.)
The chemical operator introduced what was believed to be a dilute plutonium solution from one tank into another known to contain more plutonium in emulsion. Solids containing plutonium were probably washed from the bottom of the first tank with nitric acid and the resultant mixture of nitric acid and plutonium-bearing solids was added to the tank containing the emulsion. A criticality excursion occurred immediately after starting the motor to a propeller type stirrer at the bottom of the second tank.
The operator fell from the low stepladder on which he was standing and stumbled out of the door into the snow. A second chemical operator in an adjoining room had seen a flash, which probably resulted from a short circuit when the motor to the stirrer started, and went to the man's assistance. The accident victim mumbled he felt as though he was burning up. Because of this, it was assumed that there had been a chemical accident with a probable acid or plutonium exposure. There was no realization that a criticality accident had occurred for a number of minutes. The quantity of plutonium which actually was present in the tank was about ten times more than was supposed to be there at any time during the procedure.
The employee died 35 hours later from the effects of a radiation exposure with the whole-body dose calculated to be 12,000 rem +.
Two other employees received radiation exposures of 134 and 53 rem, respectively. Property damage was negligible. (See TID-5360, Suppl 2, p. 30; USAEC Serious Accidents Issue #143, 1-22-59.)
A nuclear accident occurred in a 55-gallon stainless steel drum in a processing area in which enriched uranium is recovered from various materials by chemical methods in a complex of equipment. This recovery process was being remodeled at the time of the accident.
The incident occurred while they were draining material thought to be water from safe 5-inch storage pipes into an unsafe drum.
Eight employees were in the vicinity of the drum carrying out routine plant operations and maintenance. A chemical operator was participating in the leak testing which inadvertently set off the reaction. He was within three to six feet of the drum, while the other seven employees were from 15 to 50 feet away.
Using special post hoc methods for determining the neutron and gamma exposures of the employees involved, it was estimated that the eight men received: 461 rem, 428 rem, 413 rem, 341 rem, 298 rem, 86 rem, 86 rem, and 29 rem.
Area contamination was slight, with decontamination costs amounting to less than $1,000.
During this incident 1.3 X 1018 fissions occurred. (See TID-5360, Suppl. 2, p. 25; USAEC Serious Accidents Issue #136, 8-25-59; USAEC Health and Safety Information Issue # 82, 9-5-58; 1959 Nuclear Safety, Vol. 1, #2, p. 59.)
The "Godiva" assembly was to be used to irradiate uranium-loaded graphite samples. The samples were to be heated in a shielded furnace, exposed to a "prompt" burst of neutrons and then transferred to a counter for evaluation. The experiments are conducted at an isolated site in a building separated from the control room and all personnel by about a quarter of a mile.
On the occasion of the accident, preliminary bursts were being produced. In the process of lowering the top safety block, an unexpected burst occurred that was estimated to have produced 1.2 X 1017 fissions. The energy was great enough to tear the uranium parts from the assembly, knocking one to the floor, and to distort the steel rods in the frame. The uranium was deformed and there was much more surface oxidation than usual.
There were no personal injuries or overexposures. No gamma radiation above background was detected outside the reactor building. Radiation levels in the building were high initially . . . seven roentgens per hour gamma just inside the door (12' from Godiva) and 5,000 to 20,000 counts per minute (per 55 cm2 probe) alpha on horizontal surfaces about the room; therefore cleanup procedures were delayed 2-1/2 days until they could be completed without unnecessary exposure to cleanup personnel.
The total property loss was estimated at $2,400.
(See TID-5360, Suppl. 2, p. 18; USAEC Health and Safety Information Issue #75, 1-8-58.)
Too rapid assembly caused the system to become promptly critical. The burst yield was 3.2 X 1016 fissions.
There were no radiation exposures nor any property damage as a result of the incident.
A homogenous UO2F2, water-moderated critical assembly was made prompt critical by an overaddition of fuel to the assembly. Before reaching the critical point, the hand-operated valve was turned off. However, fuel continued to be added to the reactor because of air pressure in the line. Although the automatic safety system operated, assuring termination of the burst, considerable fuel was displaced from the reactor. The number of fissions in the burst was estimated to be about 1.6 X 1017.
No serious exposures resulted, since all personnel were shielded by a minimum of five feet of concrete. There was no significant property damage and all uranium was recovered. (See TID-5360, Suppl. 1, p. 5.)
The Experimental Breeder Reactor (EBR-I) was undergoing a series of experiments.
Without modification, certain safety instrumentation. would not permit the conduct of the experiment; therefore, reliance was placed on manual control to shut down the reactor.
During an experiment, the scientist in charge told the operator to press the "emergency reactor off" button. This would have instantaneously removed sufficient reactivity. Owing to a misunderstanding, the operator began by withdrawing the control rods at normal speed. This allowed the reactor to reach a higher power than anticipated and resulted in consequent melting of the fuel elements.
Shortly after the accident, there was a rise in the radiation level in the building. The building was evacuated. There were no personnel injuries. There was minor contamination of the sodium potassium coolant. (See TID-5360, p. 30.)
Destruction of the Borax I Reactor released 135 MW-sec of fission energy.
More than 200 safety experiments were made on the Borax I Reactor simulating control rod accidents. For the last test, conditions were set up so that the reactor would be run to destruction.
The tests were carried out by withdrawing four of the five control rods far enough to make the reactor critical at a very low power level. The fifth rod was then fired from the core by means of a spring. In this test, the rod was ejected in approximately 0.2 seconds. After the control rod was ejected, an explosion took place in the reactor which carried away the control mechanism and blew out the core. At half a mile, the radiation level rose to 25 mr/hr. Personnel were evacuated for about 30 minutes.
No one was injured and the destruction of the reactor was part of the cost of the experiment. (See TID-5360, p. 29.)
The experiment in progress at the time of the incident was one in a series designed to study criticality conditions of uranium-water solutions in annular cylindrical containers.
The cause of the accident was a. displacement of the central tube, effectively a poison rod, to a region of less importance. This displacement resulted from a dislocation of the positioning spider by a pin, used to connect sections of the liquid level indicator rack, protruding beyond the side of the rack and engaging a leg of the spider as the indicator was raised. Removing the compressional force from the top of the central tube allowed it to fall against the inside of the 10-inch cylinder. Although the displacement was small, it was sufficient to cause a large increase in the effective neutron multiplication.
The safety system apparently operated normally and the reaction was stopped automatically. All personnel in the building during the incident, were protected by a minimum of five feet of concrete shielding; therefore, no serious exposures were incurred. (See TID-5360, p. 18.)
The incident occurred in the course of an extensive study of the properties of supercritical radiation bursts produced by an assembly of fissionable metal. This study was covered by a specific procedure. A reference check of critical conditions preceded each supercritical burst.
To attain rapidly sufficient power for a delayed critical check, it was customary to set control rods at the position of minimum reactivity and insert a reactivity booster in the form of a fissionable metal slug. This time, when the booster was inserted, radiation indicators and the assembly temperature recorded went offscale (to return in a few minutes), and scrams were actuated. The resulting shock separated parts of the assembly and damaged steel supporting members.
There was no injury. The property loss was an expenditure of $600 for repair of the assembly. (See TID-5360, p. 9.)
Manual withdrawal of a control rod from a. critical assembly caused an accidental supercriticality.
The operation being conducted was the comparison of a series of newly-manufactured control rods. The assembly had been operated with the standard control rod. It was then shut down by inserting all control rods and draining the water moderator, a, standard safe method of shutting down the assembly when core changes are to be made. The standard rod was removed and the first of the series of control rods to be tested was inserted.
The assembly was filled with water with the test control rod fully in and the standard type control rods fully inserted. Withdrawal of one of the standard control rods 32 centimeters caused the assembly to become critical and the power was leveled off while the desired measurements were made. The control rod was then reinserted into the original "in" position.
With the water still in the assembly, the four members of the crew then went into the assembly room for the purpose of replacing the control rod which they had just tested. The group leader went up on the platform, reached out with his right hand and started to pull out the tested rod. As soon as he had withdrawn it about one foot, the center of the assembly emitted a bluish glow and a large bubble formed. Simultaneously, there was a muffled explosive noise. The group leader let go of the control rod which he was removing and it fell back into position. The crew left the assembly room immediately and went to the control room.
Four employees received radiation exposures ranging from 12 to 190 rem. (See TID-5360, p. 23.)
Two stacks of fissionable disks were being built up stepwise to give a slightly subcritical assembly with the two stacks brought together by remote control. The individual stacks were built up by hand in fixed assemblies and the two stacks brought together only by remote mechanisms.
After two members of the operating crew calculated erroneously from previous steps that one more disk could be added safely, the disk was added and, with attempted caution, the system was assembled remotely. Radiation indicators went offscale, actuating scrams, neutron counters jammed, and a puff of smoke was observed on the television viewer. Within three to five minutes indicators and counters returned to operating ranges.
There was no injury, no loss of material, no damage to facilities, and negligible loss of operating time. (See TID-5360, p. 7.)
Upon completion of volume measurements, it was thought that some additional information as to the required dilution could be determined by finding where criticality might occur on the rods. The control rod was pulled first with very minor reactivity effect. Following this, the safety rod was withdrawn intermittently at high speed (2.3"/sec). A waiting period for the delayed neutron effect of about 15 seconds was made just prior to the incident. This was too short a time to determine whether or not the assembly was critical. The operators next heard the safety controls actuate, instrument indicators moved offscale, scalers jammed, and the most startling manifestation was that of the breakdown of "counters" playing back through the public address system. The portable "Juno" in the control room was offscale. Presumably, a further rod withdrawal had been made.
There were no injuries. The building was successfully decontaminated, except for the test room and assembly. Before decontamination of this area was completed, a fire occurred and, subsequently, the building was abandoned because of the respread of contamination. (See TID-5360, p. 14.)
Interactions between two masses of fissionable material in water were measured at progressively decreasing horizontal separations. Remotely controlled operations established the desired horizontal separation of the two components and flooded the system.
After the final measurement, the system was "scrammed" (a rapid disassembly mechanism was actuated). Safety monitor indicators went offscale, neutron counters jammed, and the television viewer indicated steaming. Within a few minutes, indicators and counters returned to operating ranges and indicated a rapid decay of radiation.
There was no injury, no loss of material, and no damage to facilities. (See TlD-5360, p. 13.)
The reactor was being remodeled for higher power operation. As part of the required alterations, two new control rods had been placed in the system in addition to the three existing control rods.
The employee who had built the rod control mechanism wanted to test the comparative fall times of these new rods. He opened the enclosure on top of the reactor and manually lifted the rods, neglecting the possibility that this would affect the reactivity of the reactor because of its higher power arrangement. Heretofore, the three existing rods were sufficient for safety.
Normally, rods are raised remotely from the control room when the control panel is activated by a key switch. Since the rods were pulled out manually with the panel being off, no equipment was turned on except a direct reading temperature meter. Therefore, there were no neutron sensitive devices to record or warn of a rise in the neutron level. It was not observed until after the incident that the reactor temperature had risen about 25 centigrade.
The removal of the two rods probably gave a delta-K of about 0.86 percent, producing an initial period of about 0.16 second. Since the measured temperature coefficient is approximately 0.034 percent k/C, the observed temperature rise indicates the rods were out sufficiently long so that the reactor was stopped by the negative temperature coefficient.
There were no injuries. The employee doing the work received 2.5 rem of gamma radiation according to his film badge. There was no damage done to the reactor and no loss of active material. (See TID-5360, p. 21.)
A senior scientist [Louis Slotin] was demonstrating the technique of critical assembly and associated studies and measurements to another scientist. The particular technique employed in the demonstration was to bring a hollow hemisphere of beryllium around a mass of fissionable material which was resting in a similar lower hollow hemisphere.
The system was checked with two one-inch spacers between the upper hemisphere and the lower shell which contained the fissionable material; the system was subcritical at this time.
Then the spacers were removed so that one edge of the upper hemisphere rested on the lower shell while the other edge of the upper hemisphere was supported by a screwdriver. This latter edge was permitted to approach the lower shell slowly. While one hand held the screwdriver, the other hand was holding the upper shell with the thumb placed in an opening at the polar point.
At that time, the screwdriver apparently slipped and the upper shell fell into position around the fissionable material. Of the eight people in the room, two were directly engaged in the work leading to this incident.
The "blue glow" was observed, a heat wave felt, and immediately the top shell was slipped off and everyone left the room. The scientist who was demonstrating the experiment received sufficient dosage to result in injuries from which he died nine days later. The scientist assisting received sufficient radiation dosage to cause serious injuries and some permanent partial disability.
The other six employees in the room suffered no permanent injury. (See TID-5360, p. 4.)
During the process of making critical mass studies and measurements, an employee [Harry Daghlian] working in the laboratory at night alone (except for a guard seated 12 feet away) was stacking blocks of tamper material around a mass of fissionable material.
As the assembly neared a, critical configuration, the employee was lifting one last piece of tamper material which was quite heavy. As this piece neared the setup, the instrument indicated that fission multiplication would be produced, and as the employee moved his hand to set the block at a distance from the pile, he dropped the block, which landed directly on top of the setup.
A "blue glow" was observed and the employee proceeded to disassemble the critical material and its tamper. In doing so, he added heavily to the radiation dosage to his hands and arms.
The employee received sufficient radiation dosage to result in injuries from which he died 28 days later.
The guard suffered no permanent injury. (See TID-5360, p. 2.)
An experiment was designed to measure the critical mass of enriched uranium when surrounded by hydrogenous material. The enriched uranium was in the form of cast blocks of the metal, 1/2" X 1/2" X 1/2" and 1/2" X 1/2" X 1". The blocks were stacked in a pseudospherical arrangement in 12 courses in a 6" X 6" X 6" polyethylene box. The voids in the courses were filled with polyethylene blocks of appropriate dimensions. The polyethylene box was supported by a 2-foot-high stool within a 3-foot cubical steel tank. The tank had a 2-inch opening in the bottom through which it could be filled and drained by means of supply and drain hoses attached to a 3/4-inch tee. The opening in the tank was fitted with a shutoff valve, as was the drain hose. A polonium-beryllium source of about 200 mc strength was placed on top of the assembly. A fission chamber and a boron proportional counter were used to follow the experiment.
The immediate supervisor was absent from the scene when the experiment was begun. According to one of the operators, the water level was raised above the polonium-beryllium source with the supply valve almost fully open. At this point, a slight increase counting rate was observed, which corresponded with what had been observed previously when the source alone was immersed in water. A few seconds later, the counting rate began to increase at an alarming rate.
At this point, the supervisor returned, walked to within three feet of the tank and noted a blue glow surrounding the box. Simultaneously, the two operators were hastily closing the supply valve and opening the drain valve. The building was evacuated.
The three individuals involved received excessive radiation exposures, estimated in two cases as about 66.5 rem, and in the third as 7.4 rem. The doses delivered to the head and neck of these individuals may have been considerably greater. They were hospitalized for observation, but no untoward symptoms appeared. No significant changes in blood counts were observed, and sperm counts on one occasion, sometime after the incident, were normal. It is not believed that the individuals concerned received any significant radiation damage. There was no damage to equipment, no loss of active material, and no local contamination problem. (See TID-5360, p. 10.)
This was the first reactor designed to generate prompt power excursions. Prompt critical was obtained by dropping a slug of UH3 in styrex through a vertical hole in a small assembly of the same material, which was diluted with polyethylene and reflected by graphite and polyethylene. Near the end of the planned sequence of burst of increasing power, a 6 X 1015 fission burst blistered and swelled the small cubes comprising the assembly matrix. No material was lost, there was no contamination, and there were no