Metro trains crash: time for automated equipment

The November 3, 2004 accident at Woodley Park-Zoo/Adams Morgan stationTwo Washington Metro trains collided today during afternoon rush hour. At least two deaths and multiple injuries have been reported.

The cause of the accident was reportedly due to a broken track that caused a derailment. Officials are not sure yet if the train collided head-on or if the derailed train was hit from behind.

Often train collision, such as the Los Angeles Metrolink train crash in September 2008, are due to human error. Even though in this case it appears that a broken track was to blame, the escalation of this event into a catastrophe may have been preventable.

This is not the first time Washington Metro trains have been involved in accidents. With the exception of one incident in 1996, which was attributed to the inability of train operators to override automated breaking systems, tragic accidents in 1982, 2004, 2005 and 2006 were in part attributed to human error.

Hopefully this event can be used as an awaking for transit systems to move to more advanced train control system, perhaps fully automated systems. Derailments may not be entirely preventable, but many collisions are. Perhaps we can use this tragic accident as a catalyst for upgrades to our transit systems in the United States. As many systems, such as BART in the San Francisco Bay Area, are due to fleet replacements this may be the opportune time to not only increase safety, but lower operating costs.

Link to BBC Video of the crash.

*NOTE: the image pictured above is from a 2004 incident.


8 Responses to “Metro trains crash: time for automated equipment”


  1. 1 Jersey Mike June 25, 2009 at 5:48 am

    For someone with a transit blog you sure don’t seem to know a lot about transit systems work. The DC Metro uses an automated system. There was most likely a track circuit detection failure in the signaling system. There was a similar incident in 2005 where the on board driver hit the emergency brake and got his train stopped when the automated system was about the drive the train into the one in front.

    ZPTO makes trains less safe, not more. It might work in an airport, but not a real transit system.

    • 2 Switching Modes June 28, 2009 at 8:56 pm

      Jersey Mike,
      Perhaps I should have added the name ‘advanced’ before automated equipment. The point is that our train control technology is out of date. I disagree with your statement that automated equipment makes trains less safe; to the best of my knowledge there is not advanced automated train control system in the world that has had a fatality.

      NOTE: I wrote the article to be simple, so I didn’t go into different technologies. When I speak of advanced automated systems, I mean automated system which incorporate positive train control. In short, these systems put the control of the train in the hand of an on board computer, rather than a central control unit. The two primary advantages are that they are safer and less costly to operate. Additionally, such systems can allow trains to run at lower headway times.

      Sorry you might have misinterpreted the article.

      • 3 Jersey Mike June 29, 2009 at 9:56 am

        Advanced automation technologies usually mean various types of CBTC (wireless) train control and these are far less reliable and completely untested when it comes to their relative safety. The #1 driver behind technologies like CBTC is cost reduction. They cost less to maintain and less to train operators. In return they are buggy, allow for slower speeds and are often unreliable. Philadelphia replaced a traditional block signal system on its sub-surface trolley line in the interest of “safety” and it double trio times from 20 to 40 minutes. How many people will die in their cars because this “upgrade” has made the system half as fast?

  2. 4 anonymouse June 26, 2009 at 4:33 pm

    In this case, the train was almost certainly running in full automatic mode, and the cause of the crash was a track circuit that failed to detect the train ahead. I don’t think that automation actually had anything to do with this: if the train operator were running the train in manual mode, she would still have seen a cab signal indication allowing full speed, and the train would have been going just as fast. With conventional trackside signals, the train would have gotten a green signal. I don’t think automation would help or hurt in this case in particular, but perhaps we need to be much more cautious of newer signalling technology whose designers may have forgotten the lessons learned in prior decades, lessons paid for with people’s lives.

    • 5 Switching Modes June 28, 2009 at 9:03 pm

      anonymouse,
      I agree with your statement that newer systems should take into account the short comings of systems that preceded them, but I disagree that newer automated systems would neither “help or hurt in this case.” Please read my response to Jersey Mike’s article…

      • 6 anonymouse June 29, 2009 at 10:15 pm

        Let me put it as simply and clearly as I can: the root cause is not the architecture of the signal system or how “advanced” it is or is not. Systems of the type used in Washington can be made to be safe, but safety is not something you buy in a box. Safety takes constant vigilance, competence, and a strong safety culture. This incident once again shows that WMATA does not have some or all of the above. The root cause of the accident is the action of a signal engineer who failed to understand the implications of his design, or of a signal maintainer who failed to understand the consequences of his actions. And ultimately, it is the failure of the system as a whole to provide a check on their actions, to ensure that when people inevitably make mistakes, they are caught and fixed before they become dangerous.

        As for “advanced” systems, I think they can provide benefits, but also have their cost. And the main cost is that implied by the very term “advanced”: complexity. Complexity makes them harder to design, and harder to verify that they’re acting as they’re supposed to. And if they’re not tested fully, then they may fail, and when they do, it will be in some particularly unusual case where conditions are already far from normal, and that makes for a very dangerous situation.

  3. 7 Adirondacker June 29, 2009 at 1:26 am

    If you are defining positive train control as “stop the train if it moves through a stop signal” – which is the signal the train in DC should have had – then that’s been around for more than 100 years. You don’t need computers to enforce it. Don’t even need electricity to enforce it.

    http://www.nycsubway.org/tech/signals/stops.html

    • 8 Jersey Mike June 29, 2009 at 5:20 am

      It wasn’t a problem with the method of signaling, but instead a Wrong Side Failure that gave a false proceed.


Leave a Reply