Washington Metro Deaths: The Accident that Wasn't Supposed to Happen

One of the things I like the most about visiting Washington, DC is its efficient, clean, and easy-to-use subway system. Whenever I need to visit Washington, I usually fly to the Baltimore-Washington Airport and then take public transportation from there. A car is a liability in Washington, as far as I'm concerned, because you can get to most places of importance on the subway. Washington's Metro is one of the newer of the nation's major public subways, having opened in 1976, and from the start it embodied computer-controlled systems of signaling and braking. So along with many other admirers of that system, I was shocked to read last June 23 of an accident the previous day that ultimately claimed the lives of a train operator and eight passengers. What went wrong?

The National Transportation Safety Board (NTSB), in its slow, methodical way, is still investigating, but a compilation of news reports on Wikipedia can allow us to get a fairly good idea of what may have happened. The accident happened during the afternoon rush hour, just after 5 PM. As is often the case, Train No. 214 on the Red Line, one of the major north-south arteries, was stopped between the Takoma and Fort Totten stations, waiting for another train to clear the Fort Totten stop.

Trains have been waiting on tracks ever since there were trains and tracks, and in the pre-automation days a system of block signals was devised to warn oncoming traffic that the block of track ahead was occupied. When the engineer saw a yellow signal, he was to slow down, since it meant that there was another train in the block beyond the next one. Blocks were spaced far enough apart to allow plenty of room to stop the fastest train moving at a legal speed, and for the most part the system worked as long as engineers were paying attention and the signals were working.

While we will never know exactly what the oncoming train operator saw (she was killed in the accident), we do know from records that her train, No. 112, left the Takoma station at 4:57 PM. From what I understand, the Metro trains can operate in either an "enhanced" manual mode or a completely automatic mode. Operators can delay the train's departure to accommodate laggards who won't pull their arms out of the doors, for instance, but the system is also automatically supervised by braking sensors that presumably stop the train long before it could collide with another one. However, if the operators do their job, these emergency systems rarely come into play, and I can see how both operators and maintenance people might get lazy about making sure the automatic safety systems are in working order. According to reports, Train 112 was in automatic mode when the collision occurred.

At any rate, some time before 5:03 PM the operator of Train No. 112 applied the manual brake. This wasn't enough to keep her train from plowing into the rear of 214, telescoping onto the rear cars. "Telescoping" in a train collision means that one car rides up on top of another, usually doing great damage to both cars since the stationary car smashes much of the moving car's insides and vice-versa. Any people who happen to be in the way do not usually make it out unscathed. Although members of the U. S. Army who were present made heroic rescue efforts, they were not sufficient to prevent the nine fatalities and numerous injuries that resulted.

In tests on June 25, NTSB officials found that the track circuit located at the site of the stopped train failed to detect a test train placed on it. This is highly significant, since a similar failure would account for the accident. If the automatic systems didn't receive a signal that the No. 214 was stopped on the tracks, they would not have engaged and the only thing that would have averted the accident was a manual intervention by the operator. Why wouldn't the track circuit work?

The apparently simple task of figuring out whether a multi-ton railcar is present on a set of tracks is not as easy as it seems. Signaling currents have to be detected in an electrically noisy environment with large AC power currents running nearby to run the cars. The system has to work despite corrosion and oil on tracks, varying pressure on wheels, and a number of other factors. Down here in Texas, a new surface commuter rail system in Austin has been delayed repeatedly for many months partly because the train-detector track circuits have not yet worked properly. Perhaps a whole new concept of train detection is called for if the present systems are so flaky. But I'm not a railway engineer, of either the driving or designing kind, so I will refrain from second-guessing those whose job it is to fix such problems.

While we will have to wait to see what the NTSB ultimately concludes about this accident, it looks like it may well have been a fatal example of the old "garbage in, garbage out" saying. If the automated distributed braking systems don't have valid data to work with, they're not going to stop the train. And a harried, overworked (or perhaps inattentive) operator can't always be counted on to take action when she sees that a collision is imminent. Some of the Metro tunnels snake around and bend in ways that make it impossible to see far enough ahead at all points to avoid a collision, even if you paid all the attention in the world.

In the meantime, the next time I go to Washington, I'll still take the Metro. But I may make it a point to get into one of the middle cars.

Source: The Wikipedia article on this accident appears at http://en.wikipedia.org/wiki/2009_Washington_Metro_train_collision and was my primary source for this blog.