9.08.2009

Why Metro's reaction to the 2005 incident is troubling

Time and time again on this blog, I have asked why the 2005 near-miss on the Blue and Orange lines did not serve as a wake-up call for Metro. The Washington Post wrote a feature piece this past weekend, highlighting the catastrophe that almost happened under the Potomac River. The Post's focus was on Metro's reaction to the 2005 article, and how it appears Metro did not keep the region's oversight body informed of the steps taken following the incident.

While the Post's article does have some good imagery from the train operators, the article fails to ask the bigger picture questions. In fact, since the June 22 crash on the Red Line, I have seen very little coverage of the larger matter at hand. My petition to remove John Catoe was aimed at stirring up this debate, but for the most part the media attention surrounding the petition also ignored these issues.

Metro has a large PDF file that "responds" to the Post's article. This PDF file also includes the steps taken by Metro after the 2005 near-miss. Looking at these steps, I just shake my head.

From WMATA's release:
The first recommendation was that Metro should evaluate the track circuit design to determine the extent to which current and future designs can be modified to prevent a recurrence of that type of failure.
My question: Why didn't this incident prompt Metro to consider other potential failure points of the track circuit system. This near-miss made it clear that all that stood between a circuit failure and a horrific accident was an alert operator. Did anyone in the safety department consider this?

WMATA:
The second and third recommendations were that Metro should continue to monitor track circuit behavior and formalize the process. To implement these recommendations, Metro created the computerized tool for detecting loss of shunt incidents, or incidents in which a train is not detected. Metro also created the computerized tool for detecting loss of shunt incidents, or incidents in which a train is not detected. The tool was used weekly to monitor track circuit performance for one year. As the tool had not identified any serious problems in that time, it seemed reasonable to use the tool monthly after July 2006.
Did the lawyers really let Metro write this?

So let me understand this, Metro developed a tool in 2005 to detect when trains disappeared or circuits malfunctioned. But they only used this tool weekly and in 2006 decided to use it monthly? (NOTE: Following the 6/22 accident, the tool is used twice a day). What is this? Do they have to pay $1 million per use to run the tool? Is it powered by a hand-crank? Why isn't this tool run hourly? Why not real-time? What reason could Metro possibly have for only running it once a month? They had a tool that would tell them if a circuit had failed and they didn't use it.

If this tool had been running constantly, it may have detected the problem between Fort Totten and Takoma, preventing the deaths of 9 people. Was Metro expecting that as the system aged, circuits would get more reliable? What was the reasoning here?

No one is asking these questions. Metro certainly isn't answering them. There is a consistent, documented history of failure at Metro. Safety is generally only viewed in a reactive sense, as in Metro taking all steps necessary to resolve current issues, to fix existing problems. Safety is not viewed as an on-going proactive priority. Until that changes, the system cannot truly be considered safe. Metro needs change, and it needs it now.

5 comments:

  1. Depressing. I hate to be selfish, but I am glad I don't generally have any reason to ride Metro anymore.

    ReplyDelete
  2. Well, the BIG question is who knew what and WHEN did they know it!? You find out, Rusty! I mean, Dave Stroup.

    ReplyDelete
  3. Dave,

    The "computerized tool" Metro created is probably an analysis tool that takes data from various sources and correlates it to detect "loss of shunt" or "train not detected" problems. Due to the legacy nature of Metro's systems (built in the 70's), I'd bet that that the process has to run as a batch job, rather than in real-time. I'd also guess that it takes somebody with a data coordinator skillset to properly pull the data use the tool. Even some of the data may have to be hand-keyed and verified. This is probably why the tool was used only periodically.

    What it sounds like you're envisioning is a real-time analysis system, detecting events as they occur, with klaxons blaring and red strobe lights pulsing at Metro central control when an event is detected. Sounds nice, but it's probably not possible.

    ReplyDelete
  4. ...pull the data AND use the tool. It's a bit too early for me and I'm not caffeinated yet.

    ReplyDelete
  5. I figured it might require significant data analysis, but if it can run twice a day now, why not before? perhaps real time is too much to ask but monthly seems a bit too infrequent given the potential for disaster.

    ReplyDelete