NIRS Executive Director Tim Judson delivered the following comments at a press conference held at the Pennsylvania State Capitol on Monday, March 25, 2019:
The disaster at Three Mile Island was not an “accident” – it was, and remains, one of the largest industrial disasters in the world. But it is still too often dismissed as a “near-accident,” a “partial meltdown” that didn’t really hurt anyone. That impression is and has been a deliberate mistelling of history, starting from the moments after the disaster began. Large releases of radioactive contamination began within an hour of the first pieces of equipment failing. As Arnie Gundersen will detail in a few minutes, the disaster resulted in massive radiation releases: up to 100 million curies of intensely radioactive gases — about ten times the amount touted by the Nuclear Regulatory Commission (NRC).
The disaster was also preventable.
Long before March 28, 1979, there were abundant signs of flaws in the design of the reactor and certain components, as well as problems that made the reactor prone to operational mistakes and misjudgments. These were overlooked by the manufacturer of the reactors at TMI, Babcock & Wilcox, and by the Nuclear Regulatory Commission.
Known problems with B&W reactors included: pressurizer relief valves would get stuck, confusing alarms and indicators that led operators to erroneously shut down emergency cooling systems, and electrical circuit failures that caused other systems to go haywire.
Similar incidents occurred at three other B&W reactors in the previous two years before the disaster at TMI-2:
- A valve failure at Arkansas Nuclear unit 1 in September 1978 caused alarms and emergency systems to misfire at Arkansas unit 2. Review of the incident revealed that, if there had been a real emergency at unit 1, electrical circuits may have fused and disabled its backup cooling systems.
- In March 1978, at the Rancho Seco reactor in Sacramento, California, an operator dropped a lightbulb inside of a control panel, which caused a series of electrical malfunctions. This led to a pressure relief valve opening incorrectly and getting stuck, and then a surge in reactor pressure — nearly identical to what happened at TMI-2 a year later. B&W sent a memo about the incident to every owner of their reactors, except for Metropolitan Edison [the owner of TMI].
- In September 1977, at the Davis-Besse reactor near Toledo, Ohio, the primary cooling pumps failed, the pressurizer relief valve opened repeatedly and got stuck, and several other failures and mistakes similar to what happened at TMI-2 occurred. A NRC inspector raised concerns about the incident but was dismissed and told to stay in his lane.
The disaster at TMI combined aspects of all of these incidents, but quickly got out of control.
Here is the sequence of events that occurred starting early in the morning of March 28, 1979:
At 4am on Wednesday, March 28, 1979. TMI unit 2 was operating at 97% of full power when both of the primary cooling pumps shut down. This started a rapidly unfolding series of events.
Four seconds later, the turbine shut down and valves opened to vent steam into the atmosphere. This drop in pressure caused water in the steam generators to boil off, reducing cooling to the reactor core. Pressure in the primary coolant loop surged with the rising temperature, causing the pressurizer relief valve to pop open.
The pressurizer relief valve should have closed automatically after 13 seconds, but it got stuck open. Operators did not realize that it was stuck open until 2 hours and 20 minutes later, losing over two hundred thousand gallons of water in the reactor vessel.
Six seconds later, control rods inserted to stop the fission chain reaction in the core.
30 seconds later, the emergency backup pumps started up to refill the steam generators with water, but none of the water reached the steam generators for over seven minutes, because operators didn’t realize that all of the valves had been shut for maintenance.
90 seconds later, a little before 4:03am, water levels in the reactor became critically low and a loss of coolant accident was underway, but the operators didn’t realize it because they thought the pressurizer relief valve is closed. The emergency core cooling system kicks in to pump water into the reactor vessel and keep the core from overheating, but operators assumed it was an error and shut them down.
- Other alarms are going off in the control room that should have notified operators that the valve was stuck open and the coolant levels were dangerously low. But some of the alarm lights were situated where operators couldn’t see them, and one of the alarms sounded the same as several other alarms that were already going off.
At 4:14am, several thousand gallons of steam released from the reactor vessel had flooded the floor of the reactor building. A sump pump kicked in and ejected the water to the auxiliary building, outside the containment structure. Operators stopped the sump pump 25 minutes later, after 8,000 gallons of water had been transferred. Intensely radioactive gases in the water began releasing to the atmosphere.
Before 5am, surging neutron measurements indicated that the core was uncovered, but operators don’t realize it because they thought the reactor core hadn’t lost water. Fifteen minutes later, pumps to start vibrating violently due to steam pockets in the primary coolant loop. Operators shut down one of the pumps, and then the other 28 minutes later.
Over the next hour, severe core damage occurred. As the fuel rods were exposed and overheated, the fuel cladding burst, releasing intensely radioactive gaseous fission byproducts, fuel pellets spilled out, and large amounts of hydrogen gas were generated.
At 6:18am, a senior operator realized the pressurizer relief valve was stuck open and ordered a backup valve to be closed.
- A sample of water released from the reactor was tested and indicated radiation levels 350 times greater than normal.
- Almost all of the data from this 90-minute period was lost because an operator dumped the alarm data logs to clear an hour-long backlog in the control room printer.
Operators declared a site emergency at 6:50am – initiating contact with PEMA and environmental monitoring in the area. A vent gas monitor read radiation levels 100 times greater than the normal limit.
At 7:45am, a portion of the reactor core collapsed. The NRC was not notified of the incident until 7:50am, nearly four hours after the disaster started.
The White House wasn’t notified until 9am. At 9:02am, the Associated Press reported that a general emergency had been declared, but incorrectly stated that no radiation had been released.
Radiation readings outside the plant were increasing, but at 9:30, Met Ed held a press conference, repeating the claim that no radiation was released and failing to mention the emergency declaration.
- Over the next two hours, operators remained unaware that hydrogen gas voids were building up in the cooling system, preventing the core from being cooled and releasing more radioactive gases into the auxiliary building.
At 1:51pm, hydrogen that had built up in the containment building exploded — bursting the pressure barrier and allowing large amounts of radiation to escape. But because operators did not realize that the core was uncovered and severely damaged, they didn’t recognize the boom as a hydrogen explosion.
Six hours later, at 7:50pm, the high pressure injection system was activated and forced coolant water back into the reactor vessel and primary coolant loop. This continued all night long, but because of the hydrogen and has pockets, the temperature in the reactor core failed to go down.
- Operators decided to “purge” the gas pockets, venting more radiation to the atmosphere.
- Radiation levels continued to increase over the next day, but Met Ed continued delivering inconsistent information to the Governor and NRC, and reported many developments after-the-fact. As a result, the authorities did not have a clear picture of the situation, did not realize that radiation releases were ongoing, and deferred decisions to order an evacuation.
- NRC officials began to realize that hydrogen gas was building up in the containment building and become concerned that it could explode, leading to a large release of radiation.
At 11am on Friday, March 30, another release of radiation was announced. Forty minutes later, the Chairman of NRC recommended to Gov. Thornburgh that he order an evacuation of pregnant women and small children.
At 12:30pm, the governor issued the evacuation order.
Two days later, on April 1, Met Ed was able to start removing the gas pockets from the cooling system and get the reactor core damage and releases under control. On April 4, Thornburgh announced the threat of catastrophe was over. The emergency advisories and evacuation order remained in place until April 9.