By Joe Stephens and Lena H. Sun
Washington Post Staff Writers
Sunday, August 9, 2009
The crash-avoidance system suspected of failing in the recent deadly accident on Metro's Red Line malfunctioned three months earlier, when a rush-hour train on Capitol Hill came "dangerously close" to another train and halted only after the operator hit the emergency brake, newly obtained records show.
At the time of the March 2 incident, the train operator and control-center supervisors did not know that anything serious was wrong, the records indicate. The operator applied the brake because he realized that the train was not slowing fast enough and would overrun the station platform, a fairly common occurrence. About a week later, while reviewing computer logs, officials determined that there was a problem with the Automatic Train Protection system and that the train had stopped just 500 feet behind another.
Despite repeated promises of greater openness about safety, Metro officials did not make public the near miss at the Potomac Avenue Station, and federal investigators said Metro did not tell them about it after the Red Line crash, which killed nine people and injured 80.
The National Transportation Safety Board, which is investigating the June 22 crash, learned of the March incident last week when notified by the little-known Tri-State Oversight Committee, said NTSB spokeswoman Bridget Serchak. Metro officials did not immediately respond to questions about why they did not notify the NTSB.
The Washington Post discovered the incident while reviewing documents obtained through a public records request filed with the oversight committee, which was created 12 years ago to monitor Metro.
In an April 29 letter to Metro's chief safety officer, committee chairman Eric Madison asked Metro to conduct an investigation and submit a report about the Potomac Avenue incident, citing the "potentially catastrophic" nature of it. He said the train "violated a block," meaning it improperly shared a section of track with another train, and "came dangerously close to the leading train." Madison, a planner for the D.C. Transportation Department, wrote that it was only by "coincidence" that a Metro employee later noticed the incident in computer records.
Metro has yet to formally respond to the committee, which is empowered to oversee safety issues and make suggestions but cannot direct Metro to take action.
Metro spokeswoman Lisa Farbstein said Friday that although both incidents might have involved the failure of the protection system, they appear linked to different components and so would be unconnected.
"If a part goes down on the car, it's not necessarily related to the part that's on the track," said Farbstein, who described the March and June incidents as "very, very different."
Farbstein said the March incident, which took place at 4 p.m. on a Monday as a train on the Orange Line headed toward Vienna, was caused by a single failed relay on a subway car that has been fixed. The car was a 1000 series model, the same kind of car on the striking train in the June crash. The June crash is suspected of being caused by a faulty track circuit. Either problem could lead to a temporary failure of the Automatic Train Protection, a fail-safe system that monitors train locations and is supposed to automatically stop a train if it senses it is too close to another.
Farbstein said that because the car component was flawed, the six-car train at Potomac Avenue did not receive a stop command.
"The train operator did use the [emergency brake] when he realized he wasn't slowing down," she said. The train overshot the platform by about 75 feet, the length of a rail car, and halted about 500 feet from the train in front of it, she said. No one was injured, and the train was taken out of service.
Rail experts who reviewed the information questioned whether an operator, apparently unaware that he was gaining on another train, would have hit the emergency brake as quickly had he not been at a station. Rail experts also said the Potomac Avenue incident could be evidence of a significant problem.
"It is shockingly serious," said Ron Tolmei, an electrical engineer and former manager of research and development for San Francisco's subway system. "This sounds strikingly similar to what's a potential cause of the June accident."
Farbstein said that in response to the March event, Metro examined relays on its entire fleet of more than 1,000 rail cars and identified only "one relay that could be tied to the incident."
After the June crash, Metro officials said that the malfunctioning track circuit at the accident site was "a freak occurrence" and that they were unaware of other incidents, including near misses, that stemmed from failures in the safety system.
Metro General Manager John B. Catoe Jr. has promised repeatedly to keep riders better informed after the agency was criticized for failing to promptly tell them about problems with track circuits on other parts of the system. Farbstein said she did not learn of the March incident until Friday, when asked about it by The Post.
NTSB investigators have not pinpointed the cause of the June accident, in which one train rammed another between the Takoma and Fort Totten stations. But the NTSB says it appears that Metro's control system failed to detect a stopped train and that an approaching one did not receive a command to stop. The agency concluded last month that Metro's train protection system was inadequate and urged the addition of a real-time continuous backup.
Federal officials have said the track circuit at the crash site had been "flickering" for as long as 18 months.
The March incident throws a rare light on the Tri-State Oversight Committee, which was created in 1997 under an agreement signed by Maryland, Virginia and the District. The three jurisdictions jointly fund and control the six-member committee, formed as part of a national effort to improve oversight of subway systems, which are not subject to federal regulation. Each jurisdiction has two committee seats.
Minutes from the committee's April meeting said Metro was trying to "recreate" the train protection failure "but has been unable to do so." All related hardware was replaced, but Metro decided to bring in "external resources to assess the hazard," according to the minutes.
Minutes for the May committee meeting said Metro's "assessment of this hazard is ongoing."
The Red Line crash occurred a few weeks after that meeting. The minutes for July 8 indicate that at that point, the panel still had not received a response to its April 29 letter to Metro. Later in the meeting, the panel discussed a June 3 letter in which a Metro worker alleged "that the ATP system was unreliable." Metro declined to comment on the reference.
Staff researcher Meg Smith contributed to this report.