Sabotage at Three Mile Island?

Sabotage at Three Mile Island?

Investigators suspected sabotage at Three Mile Island

There is evidence to suggest that sabotage played a role in the "accident" at Three Mile Island. (This publication details only the evidence that has been documented by official government or NRC investigations.) Several days before the emergency, an unannounced NRC inspection of the plant's physical protection discovered access control infractions. Previous announced inspections found TMI to be in compliance with regulations. At the time of the accident, Three Mile Island was not required to enforce the then new "two-man rule." The two-man rule was designed to prevent a worker from being alone in vital areas. Additionally, TMI had not met the deadline for other newly required security upgrades.

In the first moments of the accident, emergency feedwater was prevented from entering the system because the "emergency feedwater valves" were closed. Indicator lights on a control room panel should have alerted the operators that these valves were closed. The two lights were hidden from view by a maintenance tag that was covering them. The valves are supposed to stay open so that emergency pumps can deliver water to the steam generators if the normal circulation is interrupted. The steam generators remove enormous amounts of heat from the reactor. Without feedwater, the steam generators boiled dry within two minutes. The temperature and pressure soared inside the reactor vessel.

The licensee's internal investigation did not consider intentional closure. The NRC Office of Inspection and Enforcement reasoned that it would take a monumental effort to interview each of the more than 750 people who had access to the emergency feedwater valves. The NRC claimed its investigators from the Office of Inspection and Enforcement were sensitive to any evidence of sabotage. But there is some disturbing and eye-opening evidence that wasn't criminally investigated. In fact, the NRC never even discovered the initiating event.


The accident started at exactly 4:00:37am on March 28, 1979. This was precisely to the minute of the one year anniversary of start-up or what is known as criticality. This aroused suspicions of worker celebrations involving drinking. The workers testified that they had their normal coffee and doughnuts only.

The trouble started somewhere in the condensate polisher system. Some unknown event caused the polisher outlet valves to close. There are several ways that a saboteur could have made this happen without being detected by plant telemetry or subsequent investigations.

The NRC Office of Investigation and Enforcement hypothesized that the initial failure was a result of a stuck-open check valve allowing water to pass into an instrument control air line and thereby cause the condensate polisher outlet valves to close. The investigators tried to duplicate this condition to test their theory. Despite pouring 15 gallons of water into this line, they could not cause the valves to shut. But, this remained the best guess as to what the first failure might have been. Because the NRC believed that the accident could have been averted at several points if human errors weren't committed, they were satisfied with not knowing the initiating event. Still, the investigators did conclude, "The problems encountered with the condensate system and condenser vacuum significantly detracted the operator's attention from the accident."

Then in the first seconds of the accident, a condensate polisher pump failure was followed by the immediate shutdown of its paired pump. The NRC investigators reported that a "wiring error" caused this second pump to quit when the first one had. A criminal investigator never assumes that an error is "only an error."

A broken air line in the condensate polisher system was ignored by NRC investigators who believed that air was prevented from leaking out by the actuation of another automatic valve. But, at least one worker testified that he had heard the broken line blowing air during the emergency. The licensee claimed that the air line was broken by a water hammer which caused equipment to shift two or three feet. (A water hammer is a sudden pressure change or a slug of water like the one that can rattle your household pipes when turning off a water faucet.) The NRC investigators reported that based on their visual inspection, the air line movement was not as great as the licensee claimed. The cause was never determined or considered necessary.

An hour into the accident, workers needed to re-establish water circulation by opening a bypass valve. The handwheel was missing from this important valve. A search for the handwheel delayed bypassing the condensate polisher system where the failed pumps were located.

The radiological releases began when a safety valve on top of the reactor failed to close. This valve opened to relieve the rapidly increasing pressure. Control room operators did not know that the Pilot Operated Relief Valve (PORV) was still open because the telemetry system was improperly engineered. The operators were fooled by a panel light which only indicated that an electrical signal had been sent to close the valve and not its actual status. Thousands of gallons of water in the form of steam spilled out of the reactor in what is known as a loss of coolant accident. For a short while the contamination was contained inside the reactor building. Although these valves had failed before at other plants, the PORV at Three Mile Island has yet to be inspected. A TMI engineer who believes that the valve simply failed said that sabotage could not be dismissed.

(Eighteen months before the TMI accident, the reactor at the Davis-Besse plant in Ohio started going out of control in what was actually a precursor to the Three Mile Island emergency. The PORV stuck open and operators struggled to understand the situation. Another design problem caused confusion about the water level inside the reactor. This problem reoccurred at TMI since both reactors were designed by Babcock & Wilcox. Davis-Besse was operating at only 9 percent compared to 97 percent at TMI when the troubles began. The Davis-Besse operators were able to return the plant to a safe condition. Afterwards, an investigation of the reactor revealed that an electrical relay had been removed from the PORV. Someone suggested sabotage. The reactor manufacturer finally decided that the relay was probably "borrowed" for usage in another part of the plant since it was compatible with several systems.)

The highly radioactive water steaming out of the TMI reactor would normally be pumped into an immense holding tank inside the reactor building. For some unknown reason the valve for this sump pump had been switched so that the contaminated water was transferred into the auxiliary building. From here the radioactivity was released to the environs through open vents.


In June 1979, an NRC special review group conceded that the NRC investigators of the TMI accident had "no training in investigative techniques or knowledge of the laws of evidence or criminal procedures." The NRC investigators did not have the authority to administer oaths and felt that the quality of the information they had obtained would have been enhanced if oaths were given. The NRC actually did have the authority to administer oaths and didn't appear to know this until after the interviews were conducted.

The report also said:

".... a trained investigator should have been dispatched with the initial response team to organize and retain portions of the supportive evidence (notes, logs, etc.) which were lost during the initial days of the accident."

Additionally, the review group found that the NRC investigation was hindered by the delay of receiving transcripts of worker interviews

(Also noteworthy is that the control room alarm printer fell behind by almost two hours. The printer was designed to store alarms in its memory until they can be printed. So many alarms were going off in the early stages of the emergency that the control room operators had to dump the stored alarms to get to the current ones. The information was forever lost.)

A technical investigator for the President's Commission on the accident questioned the adequacy and efforts of the Office of Inspection and Enforcement. Nuclear Regulatory Commission investigators had not even arrived at the plant until two weeks had passed. He also questioned the licensee's internal investigation.

The President's Commission obtained an internal TMI memo which had been written ten months before the accident. It said, "It's time to really do something on this problem before a very serious accident occurs. If the polishers take themselves off line at any high power level the resulting damage could be very significant."

The Chief Counsel for the President's Commission requested the licensee to examine its personnel files for "any person who might have long-standing grievances against the company." This was requested specifically as an attempt to discover workers who might have had incentive to close the emergency feedwater valves. Interrogation of the five workers who were identified by the company was considered.

On August 7, 1979 the President's Commission requested the FBI to determine the feasibility of an investigation into the circumstances surrounding the closed valves. The President's Commission had the authority to ask for assistance from any Executive agency and by vote had decided that the FBI was needed. But, the FBI went right back to the NRC which informed them that human errors and equipment failures were to blame for the accident; therefore, an investigation was not necessary.

An encrypted telegram sent by the FBI to the White House Situation Room around April 6, 1979 informed the President that sabotage was not responsible for the accident according to the NRC's Harold Denton. There was no reasonable way for Denton to have drawn this conclusion. The telegram which is now in the National Archives is labeled "encrypted for transmission purposes only." Portions of it are blacked-out even though it has been unclassified.

On August 15, 1979 the President's Commission asked NASA to perform an inspection of the condensate polisher system. Three Mile Island did not even have the "as built" technical drawings needed for a proper inspection. How could the NRC inspectors have done a thorough job without these? The fact was that they didn't. Investigators from NASA's Office of Flight Assurance found wires that were disconnected at five of the eight polisher panels. Operating and engineering personnel didn't know when or why they were disconnected. They also noted that an instrument air valve on the back of the polishing system control panel permits the air to be shut off and thus cause the outlet valves to close. Paul Leventhal, co-director of the US Senate investigation of the Three Mile Island accident (now director of The Nuclear Control Institute), wanted to perform a special sabotage investigation. "The initiating event was always so mysterious in that so little was known about it," Leventhal divulged in an interview. "I wanted to hire someone like a former FBI agent to do an investigation but the Minority co-director objected."

Just four days into the accident, the FBI had already announced that sabotage was ruled out and the investigation was closed. Maybe they were trying to quiet the fears of the public which had just seen the new film "The China Syndrome." (Some people actually wrote to the NRC accusing Hollywood of a sick publicity stunt.) In actuality, the FBI was planning to meet with confidential sources who believed that sabotage was to blame. An openly public source was Pennsylvania State Representative Joseph Zeller.

Both the Senate and President's Commission investigations were called off the hunt and instructed that a criminal investigation was not their responsibility. It is not entirely unusual for a valve or switch to be in the wrong position, but this many "errors" should have been investigated for criminal activity.

Soon after the emergency, the Los Alamos Scientific Laboratory concluded:

"There was very little protection against insider sabotage. ...There was very little or no control of the whereabouts of people inside the vital area; so it cannot be said that sabotage to the auxiliary feedwater system was impossible."


"...some vital area doors that should have been locked or guarded were found to be open and unguarded. Actually, there was very poor protection against the sabotage actions of the insider."


"The conclusion can be drawn that the protection against the activities of an insider is still inadequate at TMI..."
And an embarrassing incident did happen several months after the TMI accident when a newspaper reporter was hired as a security guard. He told of entering the control room unchallenged (only armed guards were permitted access). There was no lock on the door and a piece of clothesline hung where the doorknob should have been. A college textbook used this incident as an example of poor security. The book cited the reporter's headline -- "Three Mile Island: It's a Paradise Island for the Saboteur." General Public Utilities sought an injunction to block publication of the article on the grounds that it could compromise national security.

[return to nuclear terrorism homepage]