The Accident

(7)

March 28, 1979: just before 4 a.m.

In a routine matter, operators sealed off tank 7 of Unit 2 to flush out the BB-size balls of resin that clean the water that is used in the cooling system before it is sent back to the reactor.  As usual, the resin balls stick together in the output.  Initially, the jerry-rigged tactic of fluffing the balls with a shots of high-pressure air does not work.  The resin constipates the pipes, until  finally the honey-consistency sludge beaks free.  (1)  Unnoticed to the workers, a small amount of water leaked in around a valve from the built-up pressure in the pipes during the clog, and traveled through the pneumatic pipes that service the varied systems.  As a safety precaution, these controls trip (or close) automatically if they lose air pressure or if something (like a little water) get in the way. (2)  

Information Overload

At 4:00:37 a.m., hundreds of white alarm lights blink in the control room and the printer starts chattering away with error codes. 

http://en.wikipedia.org/wiki/File:Tmi-2_schematic.svg

Stuck Valve

Unit 2’s electricity-generating turbine has tripped (or shut down), and two secondary feedwater flow valves have also tripped.

In normal operation, the heat produced by the nuclear core’s reaction is carried by the closed-loop primary feedwater system to steam generators that convert that heat to produce steam supplied by the secondary feedwater system.  If the secondary feedwater system no longer can convert into steam and run the turbine, the closed primary system overheats rapidly.  A pressure relief valve is designed to open in such an occasion and release the superheated water in a waste tank.

(8)

Within 15 seconds of the initial incident, control rods slip into the uranium core to stop the nuclear reaction, but this “scramming” does not remove the residual heat from the primary loop.  Auxilliary pumps are designed automatically activate and pump emergency water into the primary loop, but recent routine maintenance had left them closed and unresponsive. (3)   Foolishly, the design of valve indicators does not show whether the valve is open or not, but rather if the computer has given the valve a command. (4)  Unbeknownst to the operators in the control room, four failures are converging: auxiliary valves that will not open, a pressure relief valve that will not close, a faulty pressure-reading gauge in the steam generator, and no useful knowledge about the true conditions in the core.

The two valve switches (left top and bottom), which control the emergency feedwater flow, were found closed. (9)

Reacting to faulty assumptions and ignoring unexpected readings, the operators turn off the emergency flow to the primary system, believing that the system is equalizing.  However, the pressure relief valve is still open and the replacement water is draining away, leaving very little left.

The reactor’s water is boiling off!

Ten minutes after the initial warnings, the faulty reader in the steam generator finally indicates that there is not enough pressure for normal operation and the workers reengage the emergency pumps to the core, but the water continues to drain through the pressure relief valve into the storage waste tank.

Overflow

Fifteen minutes into the mishap, the waste storage tank is full.  Not designed to hold so much water, a disc blows out of the side and a sump pump drains the radioactive water into an auxiliary building.  That, too, overflows and is about to be released into the atmosphere. (5)

Isotope iodine molecules react with silver in the control rods, and hydrogen bubbles begin to form in the core’s cooling channels, choking off flow.  

The temperature steadily rises….

If it hits 5000 degrees, the core will melt through the floor of the plant, the eight-inch thick steel pressure vessel, the twenty-foot concrete foundation, and maybe even the bedrock of the Susquehanna River. (6)

(1) Mike Gray and Ira Rosen, The Warning: Accident at Three Mile Island (New York: W. W. Norton, 1982), 73.

(2) Mark Stephens, Three Mile Island (New York: Random House, 1980), 9.

(3) William Keisling, Three Mile Island: Turning Point (Redmond, Washington: Express Publications, 1980), 20-21.

(4) Ibid, 27.

(5) Ibid, 22.

(6) Gray and Rosen, Warning, 101.

(7) Ibid, 78.

(8) Keisling, Turning Point, 119.

(9) Ibid, 20.