washingtonpost.com
Metro Control System Fails Test
Technology Should Have Averted Crash

By Lena H. Sun and Maria Glod
Washington Post Staff Writers
Friday, June 26, 2009

A train control system that should have prevented Monday's deadly Metro crash failed in a test conducted by federal investigators, officials said yesterday, suggesting that a crucial breakdown of technology sent one train slamming into another.

Investigators with the National Transportation Safety Board performed the simulation Wednesday night. In the test, investigators positioned a train in the same location as the train that was rear-ended Monday. The system failed to detect that the idled test train was there, the NTSB said. Investigators did not say what caused the malfunction, and they stopped short of saying the system failure caused the crash.

The test results are significant because they confirmed earlier findings of "anomalies" in an electrical track circuit in the crash area.

The findings suggest that the oncoming train in Monday's crash might not have received information that a train was stopped ahead on the rails north of the Fort Totten Station. The stopped train was struck by a train operated by Jeanice McMillan. She and eight others were killed in the crash; 80 people were injured.

If a malfunctioning circuit failed to detect the stopped train, it would have assumed that the stretch of track in front of McMillan's train was clear and set the speed of her train at 59 mph, sending it hurtling into the stopped one.

The steel rails show evidence that McMillan activated the emergency brakes before the crash.

An NTSB spokesman yesterday said that no one was available to discuss the findings.

Jackie L. Jeter, president of Amalgamated Transit Union Local 689, which represents most of Metro's front-line employees, including train operators, said she interpreted the findings as proof that the system failed.

A senior Metro offi cial knowledgeable about train operations said an internal report confirmed that the train control system failed to detect the idling train when the crash occurred about 5 p.m. on a curved section of track between the Fort Totten and Takoma stations.

Metro has temporarily reassigned the top official in charge of the train control systems that are supposed to prevent crashes. Matthew L. Matyuf, superintendent of the Automatic Train Control Division, has been assigned to a "special project," spokeswoman Lisa Farbstein said yesterday. She said she did not know the reason for the move. Matyuf did not respond to two messages left yesterday at his Loudoun County home.

The operator of the idling train, who was hospitalized after the crash, was also interviewed by federal investigators yesterday. Unlike McMillan, he was operating his train manually. He told investigators that he saw a train in front of him at the Fort Totten Station and stopped, officials said. While stopped, he felt a "hard push" from the impact, according to the NTSB.

Farbstein said the safety board's testing will help pinpoint "what went wrong so we can fix it." The agency took several steps yesterday to assure the public that its trains are safe. Metro General Manager John B. Catoe Jr. said personnel began inspecting all 3,000 track circuits Tuesday. Trains will be operated manually instead of by onboard computers until the completion of the inspection, which is likely to take several weeks. Trains on the Red Line will also be restricted to speeds of 35 mph.

The union and some members of Congress are also recommending that Metro sandwich its oldest cars, the ones that made up the striking train, between newer cars. The oldest cars, purchased more than 30 years ago, have been criticized as being prone to fold into themselves, like a telescope, during a crash. The lead car of the striking train in Monday's crash incurred the worst damage and was compressed by two-thirds. The cars, manufactured by Rohr Industries, make up more than 25 percent of Metro's fleet.

Metro's fleet of 1,128 rail cars is composed of six series of cars. Operations personnel say using the same series of cars in a train allows for smoother operations. No decision has been made about train configurations, officials said.

A House committee is scheduled to hold a hearing on the crash July 14.

Catoe said he will take action "at any suggestion of problems" and not wait for a formal report from the NTSB on the cause of the crash, which could take a year or more. Metro board chairman Jim Graham, while pledging full cooperation, questioned the daily release of investigation details by the NTSB.

"I believe it is very important to gather and determine the facts first and at an appropriate time release the facts," he said, instead of "putting out piecemeal information, theories and possibilities" about the crash.

Five to 10 percent of the track circuit inspections have been completed, and no immediate problems have been detected, Deputy General Manager Gerald Francis said yesterday. Federal investigators said maintenance crews were working on the track circuits in the area of the crash earlier this month. Metro's rail chief, Dave Kubicek, would not say what that work was or how frequently track circuits are inspected, saying personnel have "different levels and different strategies" for inspecting the circuits.

A track circuit is an electrical circuit that includes a length of running rail and allows the presence of a train to be detected. The circuit also communicates commands and instructions between the track and the train. If a train tries to approach too close to the rear of another train, information provided by the track circuits is used to slow or stop the second train before there is danger of a crash.

Metro's automated trains operate under several control systems. The train protection system is made up of circuits embedded along the track, which range from 150 feet to a half-mile long. As trains cross the circuits, signals are transmitted down the line to following trains. The signals automatically set speeds, slowing or stopping a train so that it doesn't crash into the one in front.

The railroad is divided into blocks, which are varying lengths of track, and computers are set to keep two blocks of distance between trains. As an added layer of control, another electronic system regulates train speeds and spacing and stops the trains as they enter stations. A third system controls overall train movements to maintain proper routing and keep trains on schedule; it is monitored by workers in Metro's downtown central control room.

If the train protection system is working as designed, when one train begins to enter the buffer zone, the computers deploy the brakes on the second train and force it to stop.

In the past decade, Metro has run into trouble with components in its signal system. In 1999, the agency discovered that a handful of key devices known as relays were prematurely failing. Relays, about the size of a hardcover book, transmit the signals that automatically control speed, braking and switches. They were designed to last 40 years, but some were malfunctioning after 25 years, sending erroneous signals to trains on at least four occasions.

Metro ripped out all 20,000 relays and demanded that the manufacturer, Alstom Signaling, replace them. Alstom installed signals at a cost of $8 million to Metro, a process that took 20 months. In that time, Metro trains ran in manual mode. Neither Metro nor Alstom ever determined the cause of the failures.

Catoe and other top officials have been meeting with families of the victims and injured and writing checks to help cover funeral, medical and other expenses. Officials said the payments are not intended to avoid lawsuits against the agency.

Staff writer Lyndsey Layton contributed to this report.

View all comments that have been posted about this article.

© 2009 The Washington Post Company