Atomic Accidents, page 20
It was time to change out the flux wires, and the previous shifts had done all the heavy lifting. The reactor was basically put back together, with the vessel filled to 9 feet with water and the head screwed down. All the night shift had to do was reconnect the controls and put the big concrete blocks that shielded the top back into place. It was a three-man job. Byrnes and Legg would reconnect the rack to the main control while McKinley would stand off and act like a health physicist, pointing a “cutie pie” ionization chamber at Byrnes and Legg while they worked. The day shift would actually have an “HP” on the staff, monitoring all the activities in the reactor building to constantly check for abnormal radiation, but the night shift was a minimum crew.
The rack that engaged the pinion for up-down motion was screwed to the top of the control, which was at its lowest position in the core, keeping the reactor at cold shutdown status. A C-clamp had been tightened on the rack to hold it in a slightly raised condition, just above the top of the shield plug. This position allowed a three-foot metal rod to be temporarily screwed into the top of the rack so that a man could handle it with the motor disconnected.
The instructions were clearly mimeographed. Byrnes would take hold of the handling rod with all ten digits and lift the 100-pound center control by one inch. With the load off the C-clamp, Legg, crouching over the shield plug, would unscrew it and lay it aside, and then Byrnes would gently lower the control until it rested on the bottom of the core structure. McKinley was standing off the reactor top, in front of one of the man-sized concrete shield blocks, idly watching the show and pointing his radiation detector.
All that we know for sure is that at 9:01 P.M., Byrnes, against the written directions and everything that the instructors had drilled into his head, with one massive heave jerked the master control clean out of the core as fast as he could. If it were lifted four inches, the reactor would go critical, blasting the three workers with unshielded fission radiation. Byrnes managed a full 23-inch pull, and the reactor went prompt critical with a 2-millisecond period, producing a steam explosion in the reactor such as has never been seen before or since. The water covering the reactor core instantly became superheated steam.108 The four-foot slug of still water over the core, not becoming steam, was pushed with incredible speed to the top of the reactor vessel, through the 4 feet 6 inches of clear space, until it hit the screwed-down top like a very big hammer. The force of the hammer-hit picked up the 13-ton steel vessel and shot it nine feet out of the floor, shearing away its feed-water and steam pipes. The nine shield plugs on top of the vessel shot off like cannon shells, burying control-rod fragments in the ceiling.
Byrnes and Legg were killed instantly, not by the intense radiation surge, but by the explosive shock of two billion billion fissions, 15 megawatt seconds of energy, and an air pressure wave of 500 pounds per square inch. McKinley died two hours later of a massive head wound, inflicted by the concrete shield block as he was thrown backwards. All three men had fission products, built up by the reactor running at full power for two years and turned to nascent vapor in the sudden heat of prompt fission, buried deep in their bodies. There was no way to simply wash away the contamination. It seemed embedded in every tissue.
Legg was pinned to the ceiling with a piece of the master control. The reactor internals were an unrecognizable tangle of twisted parts, and water, gravel, and steel punchings were scattered all over the reactor building floor. The crude steel building, which was meant only to house the equipment and keep the rain off, managed to prevent a scattering of radioactive debris, and outside it was hard to tell that anything had happened. The plant was a total loss, and everything would have to be carefully disposed of, leaving not a trace of radioactive contamination or subjecting any worker to an abnormal dose. The three bodies were so deeply and severely contaminated, they would have to be treated as high-level radioactive waste. Autopsies were performed quickly, behind lead shields with instruments on 10-foot poles.
A commendable job was done analyzing the accident and cleansing the site of any trace of it. It required a great deal of skill and planning to decontaminate the site, as the inside of the reactor building was too radioactive for normal work. A full-scale mock-up of the building was constructed for decontamination practice and to figure out the actions of Byrnes and Legg before and during the accident. Television cameras were inserted into the reactor core using remote manipulators, and a Minox subminiature camera was used on the end of a pole to take photographs through small openings. A seldom-mentioned technique called the “gamma camera” was used to spot where highly radioactive fragments of the reactor had landed in the ceiling and on the floor of the building.
The gamma camera is a variation of the old “pinhole” camera. It is possible to make a picture on a piece of photographic film by mounting it on one inside face of a light-proof, square box. On the opposite face of the box is a tiny pinhole. Light enters the box through the hole, very dimly, and it forms an image on the film without the use of a lens. Light can only travel in a straight line, so individual rays of light from the scene outside the box are organized into an image simply because they all have to come through at the same point, the pinhole. If the box is made of lead, through which radiation cannot pass, then the pinhole on the front face of the box will image gamma rays onto the film the same way it uses visible light. Gamma rays are photons, just like visible light but at a much higher oscillatory frequency and energy. To use the camera, the investigators first made a light image of the ceiling in the reactor building using the open pinhole, then covered the pinhole with a light-tight but gamma-transparent shutter. The gamma ray image was allowed to expose the film for 24 hours through the same pinhole, superimposing a gamma-ray image atop the light image. When the film was developed, it identified radioactive objects in the picture as shining brightly, like points of light.
The reactor building and its contents were buried in a trench 1,600 feet away from the original SL-1 building site. The cleanup took 13 months and $2.5 million. Much was learned about managing power reactor disasters, and the question of what had happened was answered in great detail, down to the millisecond.
What was never answered by detailed forensic analysis was: Why did Byrnes pull that control out? He knew good and well what it would do if it came up to four inches. Why jerk it out all the way? There are, to this day, a few conflicting opinions, ranging from it was a murder-suicide to he was exercising the rod to clear the bent boron. In my opinion, Byrnes was showing off for McKinley, the new guy from the almighty Air Force. The Air Force was running the dangerous, super high-tech HTRE experiments down the road at Test Area North, and the Army was stuck with this cheap, low-power rig that was just sitting here making a slight turbine-hum. Byrnes wanted to give McKinley a thrilling blip on the cutie-pie radiation detector he was holding by bouncing the main rod. He knew that if he could bring it up to supercritical for just a split second, the power would drop again quickly as the control went back down. No harm done, but he bet himself that he could make Air Force lose control of his bladder. The thing was heavier than it looked. He wiped his sweaty palms on his pants, braced, and put both arms into it. Up she came. They never knew what hit them. Their nervous systems were destroyed before the senses had time to register the violent event.
It was a tragic loss of life, and the general attitude toward small, simple, cheap nuclear power was affected. The Army went on to deploy the PM-2A portable medium-power reactor in Greenland, the PM-3A at McMurdo Station in Antarctica, the MH-1A mobile high-power reactor at the Panama Canal, the SM-1A stationary medium-power reactor at Ft. Greely, Alaska, and the PM-1 at Sundance, Wyoming. They even developed the ML-1 at the NRTS, the first nuclear power plant that would fit on the back of a truck. It all ground to a stop in 1977 when the need for remote power and the money to support it both drifted away. It was not at all a bad idea, but commercial nuclear power production moved off in the opposite direction, toward bigger, more complicated, more expensive installations, and the SL-1 incident was a reminder of the dangers of making it too simple. The accident had proven that there was no such thing as foolproof.
The snowcap over Greenland into which Project Iceworm was dug turned out to be one big, slow-moving glacier. The rooms and tunnels so carefully carved out of the ice deformed and shrank with time. Every month, 120 tons of ice had to be shaved off the inside walls, and by the summer of 1962 the ceiling in the reactor room had come down five feet. In July 1963 it was clear that this bold step in missile positioning was not going to work, and the PM-2A reactor was shut down and shipped back home. By 1966, Camp Century was unlivable, and it was given back to nature. When last visited in 1969, it was completely wrecked and buried under a great deal of new snow.
The deep snow cores taken at Camp Century are still in use today. These long cylinders of snow cut from the glacier are a record of the climate and atmospheric conditions for the last 100,000 years, and they have been used to map the carbon dioxide history on Earth since the emergence of mankind.
The last steam explosion in Idaho was the final in a series of experiments with the SPERT-I, or the Special Power Excursion Reactor Test One, on November 5, 1962. One would think that enough was learned in the BORAX-I and the SL-1 blow-ups to pretty much give us what there was to know about water boiling too fast in a reactor tank, but this odd experiment was funded by the AEC. It, like BORAX, gave a bit more of a show than was anticipated, and it went down in history as another accidental power excursion at the NRTS. The incident was all recorded as a movie in slow-motion at 650 frames per second.109
The reason for the SPERT experiments had to do with the expanding need for nuclear specialists in the 1950s. The AEC was aware of a projected shortage of nuclear engineering and health physics graduates in American universities, and technical campuses needed small research reactors in place to encourage these studies and excite some interest in nuclear topics. The Aerojet General Corporation had already introduced an inexpensive “swimming pool” reactor for use in schools. It was simply a concrete-lined pool of water, sunk in the floor, with uranium fuel assemblies and control rods clustered in the middle. “Is it safe?” asked the AEC. “What would happen if a control rod came loose and dropped out the center of the core?”
It would explode, of course, with the water shooting out the open top of the reactor and hitting the roof really hard, but that was not a sufficient answer. Details of everything that could possibly happen to an open-topped, water-moderated, low-power reactor were demanded. The SPERT contract was given to the Phillips Petroleum Company, and a site was chosen about 15 miles west of the eastern boundary of the NRTS. The first of four SPERT reactors was started up in June 1955, and it was run through a series of torturous accident scenarios.
The setup was very much similar to BORAX-I, but it was enhanced with a metal shed covering the reactor. Controls were inserted through the bottom of the reactor core leaving the top completely open, with one master control in the center. Two periscopes looked down into the core from above, and a tilted mirror showed the open reactor top from the side. These optical features were used to make motion pictures of every experiment simulating wrongful operation by undergraduates horsing around with the controls.110 Nuclear instruments and recording procedures had improved since BORAX, and better data collection was anticipated. Experiments with fast power transients blew the water out the top of the reactor, just as seen in the BORAX excursions. The fuel was the same as was used in most of the test reactors at NRTS, thin plates of aluminum-uranium alloy clad with pure aluminum. It was not meant as a high-temperature fuel, and some plate warpage and slight melting were observed in the core.
November was the end of the open experiment season in Idaho, as the temperature began to drop to Greenland levels, and the team was out of tortures for the SPERT-I. SPERT-II was designed and ready to build. As one last experiment for 1962, the team wanted to simulate AEC’s worst fear, that the main control rod would fall out. The control-rod drive was modified to break free and fall with gravity, and the metal roof was removed so that the driven water would have nothing to blow away. A 3.2-millisecond period was predicted, with some fuel melting this time. The test was not cut out to be such an event as the BORAX-I excursion. SL-1 had taken the thrill out of seeing water reactors explode.
It was a sunny day in the desert, and the wind was calm. Three … two … one … RELEASE. The main control rod went into freefall. The little reactor suddenly lit up with a blue flash in the mirror as 30.7 megajoules of energy came alive and started heating the fuel. All 270 fuel plates melted, and the fission process shut down completely, as predicted. Everything was going according to plan for about 15 milliseconds, and then all hell broke loose as the unexpected transpired.
The melting fuel sagged and changed shape. Heat transfer between the fuel and the water was suddenly improved, and the resulting steam explosion was more energetic than expected. It completely destroyed the reactor. Contents of the reactor vessel were pounded out of shape and thrown skyward. There was no roof to carry away, but a flying periscope hit a steel roof-beam and bent it outward. Bits and pieces of reactor were scattered all over the place. There were no injuries except to a few egos, and the large bank of instruments could find no harmful release of radioactive gasses into the atmosphere. I will not say that the fine engineers at the NRTS were slow to learn, but this sort of behavior in a suddenly uncontrolled water reactor was hardly a new finding.
This was not the last reactor blown up at NRTS, but all the others were predictable, controlled, and not considered accidents. NASA planned to send up a nuclear reactor into space in the System for Nuclear Auxiliary Power program, and the SNAP-10A nuclear power generating station was scheduled to be launched into orbit in 1965.111 There were, of course, concerns that the booster could fail before achieving orbit and a power reactor could come down in the ocean. What would be the hazards if this worst launch failure happened?
At NRTS the recently vacated ANP test facilities at Test Area North were converted to test the SNAP-10A in a simulated high-speed crash into the Pacific Ocean. It was correctly predicted that a severe nuclear power transient would result from slamming into the salt water. The reactor was naked of any shielding, and it was moderated by zirconium hydride mixed with the uranium fuel. Hitting the surface would suddenly introduce extra moderating material favorable to fission (water) and the reactor would go prompt supercritical.
A huge water tank was mounted on one of the ANP double-wide rail cars that carried a nuclear jet engine with a SNAP reactor mounted in the middle. The reactor was protected from the water with a Plexiglas shield until an explosive charge threw it out of the way and let the water crash in. The resulting fireball on April 1, 1964, threw reactor fragments far and wide.112 To keep thermocouples and radiation instruments from being melted and lost in the explosion, the measurements were done remotely, using infrared pyrometers looking into a mirror behind a lead wall. Just to be sure, the test was run three times in experiments SNAPTRAN-1, 2, and 3, and three perfectly good SNAP reactors were destroyed.113 They took movies.
There was also the LOFT (Loss Of Fluid Test) reactor and Semiscale, which was not really a reactor. Both experiment series were used to find what would happen if a water-cooled reactor breaks a major steam pipe. The LOFT reactor was a half-scale, fully operational pressurized water reactor, while Semiscale was a slightly safer experiment using an electrical heat source instead of fission to make the steam in a simulated power plant. These programs and the accidental experiments dating back to EBR-I were all valuable in finding how to build a safe power reactor and how not to build one. This knowledge was put to use in designing the Generation II nuclear reactors that now produce 20 percent of the electrical power in the United States. There would be commercial reactor accidents in America, but never a steam explosion, and the three men who stood on top of the SL-1 were the last people ever to die in a power reactor accident in this country.
Today, there are no ongoing reactor safety experiments, anywhere in the world.
Torturing nuclear reactors to make them give up the secrets to safe power production was not the only activity at the NRTS. Among the first building sites at the new reservation in 1948 was the Idaho Chemical Processing Plant, known to all as the “Chem Plant.” If the techniques for building civilian power plants were to be sorted out in the Idaho desert, then it made perfect sense to also work on fuel reprocessing and waste handling. A model plant was built to recycle the fuel used in reactor experiments and to develop practical methods for extracting the various components made in the fission process.
Commercial reactor fuel is uranium oxide, with two out of every three atoms in the solid material being oxygen. The uranium content in fuel is usually enriched to 3.5% fissile uranium-235. The rest is uranium-238. After about 4.4% of the uranium-235 has fissioned, the fuel can no longer support the self-sustained chain reaction, and it must be replaced. Approximately 20.5% of the waste product embedded in the spent fuel is plutonium resulting from neutron activation of the U-238. The rest of the waste, 79.5%, is fission products, 82.9% of which are stable. That leaves 1.1% of the spent U-235 as radioactive waste which must be either disposed of or put to use as industrial and medical isotopes. The idea of fuel reprocessing is to remove the remaining U-235 and Pu-239 from the spent fuel and return it to the energy production process. As you can see from the breakdown of the used fuel components, the waste to be buried is a tiny part of the original fuel. In 1948 when uranium was thought to be scarce, it made no sense to bury an entire used fuel-load without breaking it down and sorting the portions.
The Foster Wheeler Company of New York, experts at making oil refineries, designed the plant, the Bechtel Corporation out of San Francisco built it, and American Cyanamid ran it. The Army had wanted to operate it under military control, but the AEC wisely argued that if it was to model a commercial process, then civilians should learn how to do it. Construction took 31 months on 83 acres of flat-as-pool-table desert north of Big Southern Butte. The first shipment of fuel arrived to be processed in November 1951.
