During operational tests, the check valves on the pumps, meant to prevent backflow if something stopped working, would slam shut instead of closing gently but firmly, as they were supposed to. This flaw was re-engineered, and further extensive testing would ensure no problems from the pumps.
Still, working with a large volume of pure sodium would prove challenging. Tests were being performed in an abandoned gravel pit about 20 miles north of the building site on August 24, 1959, when a load of sodium exploded in air. Houses in the nearest neighborhoods, Trenton and Riverview, were damaged by the blast and six people were hospitalized. By December 12, 1962, there was enough of the reactor assembled to test the main cooling loop. An operator was at the control console watching the instruments as the sodium circulated at full speed through the system. The temperature started reading high on a thermocouple gauge. There was no fuel in the reactor, so where was the heat coming from all of a sudden? Were sodium and water mixing due to a flaw in the steam generator? Thinking this through, the operator reached over and hit the red water-dump button. Water gushed out of the secondary cooling loop into a holding tank, taking it quickly out of the steam generator and away from possible contact with the sodium loop. Unfortunately, this caused a sudden vacuum in the system, which was designed to hold high-pressure steam and not an airless void. A safety disk blew open, and sodium started oozing out a relief vent, hit the air in the reactor building, and made a ghastly mess.
No one was hurt. When such a complicated system is built using so many new ideas and mechanisms, there will be unexpected turns, and this was one of them. The reactor was in a double-hulled stainless steel container, and it and the entire sodium loop were encased in a domed metal building, designed to remain sealed if a 500-pound box of TNT were exploded on the main floor. It was honestly felt that Detroit was not in danger, no matter what happened.
Cost of the Fermi 1 project reached $100 million, and it was too far along to turn back. The fuel was loaded on July 13, 1963, and the fuel car was not acting well.[149] The first startup was a few weeks later, on August 23 at 12:35 P.M. A system shakedown at low power would continue until June of 1964. A few problems surfaced. The number 4 control rod delatched from the drive mechanism, leaving it stuck in the core. The large, rotating plug in the top of the reactor vessel, used to move the refueling arm between the fuel rotor and the core, jammed and wouldn’t move. A sodium pump had to be repaired, and the cap on the reactor vessel had to be rebuilt so that it would fit correctly. Some electrical connections and cable runs were defective, causing instrument problems. These glitches were all knocked down.
By January 1966, the Fermi 1 plant was wrung out and ready to go to full 200 megawatts of heat in several cautious steps. August 6, 1966, was a day of triumph. The thermal power was brought up to 100 megawatts, enough to make 33 megawatts of electricity, or about half what the backup diesel generators could produce. The project cost had also reached a high point, at a cool $120 million, and critics pointed out that the reactor had so far been able to generate measurable electricity for a total of only 52 hours.
At this half-power level, unusually high temperatures were indicated in fuel assemblies M-091 and M-140, the steam generator started leaking steam, and control rod no. 3 seemed to stick in the guides. The next day, August 7, the positions of the hot fuel assemblies were swapped with trouble-free fuel assemblies, to see if the problem moved. There seemed no correlation between the specific fuel assemblies and overheating. The problem seemed to be the position in the core, and not the fuel.[150] Could it be that the thermocouples in those locations were just reading wrong?
The operating crew was ready to try another cautious power-up on October 4, 1966. The Fermi 1 reached criticality at 11:08 P.M., and it idled at low power while every little thing was checked. At 8:00 a.m. the next day, the operators were ready to bring the reactor to half-power, but a steam-generator valve seemed stuck. It took until 2:00 P.M. to resolve that problem, and then the feed-water pump in the secondary loop was not working. They powered down and worked on it. By 3:05 P.M. they had resolved the problem and the power was increased to 34 megawatts and rising.
Something was not right. The neutron activity in the reactor core was erratic and bouncy. There was no reason for the neutron level to be anything but smooth and steady. The power ascension was halted while Mike Wilbur, the assistant nuclear engineer in the control room, contemplated the meaning of these instrument readings. Based on previous problems, Wilbur had a hunch. He stepped behind the main control panel to take a look at the thermocouple readouts on the fuel assembly outlet nozzles. These instruments were not considered essential for running the power plant, so they were not mounted on the main panels. They were included in the hundreds of instrument readouts, lights, and switches in the control room for diagnostics, and here was a problem that required a deeper look.
Fuel assemblies M-140 and M-098 were both running hot. At this power level, the temperature of coolant flowing out the top of fuel assembly M-140, which had given trouble in the past, should have been 580°F. It was reading 700°F. As Wilbur was taking this in, at 3:08 P.M. the building radiation alarms started sounding. It was a rude, air-horn sound: two mind-numbing blasts every three seconds. There were several possible explanations for the radiation alarms, but the assistant nuclear engineer knew deep in his heart that one was likely. Fuel had melted, spreading fission products into the coolant. The only thing that was not clear at all was: Why?
The crew executed emergency procedures as specified in the operations manual. All doors were closed, and all fresh-air intakes were closed in the building. Detecting radiation in the building was an emergency condition. It was supremely important to not let it leak out into the world outside Lagoona Beach, which would make it a big, public emergency. They executed a manual scram at 3:20 P.M., shutting Fermi 1 down with the floor-trembling shudder of all controls dropped in at once. One rod would not go in all the way. Not good. Was a fuel assembly warped? They tried another scram. This time, it went in. With the neutron-poisoning control rods all in, there was no fear of the core being jostled or melted into a critical condition. There could be no supercritical runaway accident, and if the core were completely collapsed and flowing onto the spreader cone, the uranium would be mixed with melted control material, which would definitely discourage fission. News of the accident, specifying that the engineers did not know what had happened, spread across the land.
The reactor had never run at full power, and only for a short time trying to get to half power, so there was no worry that it could melt down any further in the shutdown condition. There were not enough delayed fissions and fission product decays to cause havoc with high temperatures in the fuel. Over the next few weeks, the operators and engineers tried to find the extent of the damage without being able to see inside the reactor core. One at a time, they pulled control rods and noted the increase in neutron activity in the core. A few control rod locations did not return the activity they expected. Near these, the fuel may have sagged out of shape.
149
It is difficult to find a cross-section view of the Fermi 1 reactor that does not have a big X drawn through the refueling car. It was not a popular accessory.
150
Geek-joke from 1975: How does a Digital Equipment Corporation computer-repair technician fix a flat tire? He swaps the wheels to see if the problem moves. You had to be there….