DC Metro Collision

[aegrotatio]

 Regarding edblysard's long post, the comment that I have to make about Metro's operations is that the trains can and do "creep up" to the next train.  The headways are very short.  At stations you can see the following train about a platform's length behind you.  The trains stop in tunnels and creep forward 10 yards or so at a time several times.  It happens every day.

Regarding blocks, it sure appears to me that Metro allows more than one train in a block in the course of normal revenue service, and those trains are both computer-controlled and manually-controlled.

Heck, before the 1996 accident, the operators weren't even allowed to drive the trains manually because of excessive flat spots.  The system was running for 20 years by that time.

 

There are sensors between the rails every few yards.  Every description I have ever read about the DC Metro is that the central control center (in DC) knows where every train is and that's due to these embedded sensors.

[aegrotatio]

 Hi, zugmann, DC Metro's rush-hour timetable states that trains arrive at stations less than 2 minutes apart.  I ride the DC Metro and at Metro Center it's well under 2 minutes between trains.

 

[oltmannd]

aegrotatio

Regarding blocks, it sure appears to me that Metro allows more than one train in a block in the course of normal revenue service, and those trains are both computer-controlled and manually-controlled.

I'd bet that blocks are pretty short, some <1000 feet, which gets pretty close to a platform length.

The control center may know where every train is but that has nothing to do with the ATP system - which is the safety system.  The control center is not part of the safety system. 

[ValleyX]

zugmann

You can't compare a freight operation to a mass transit operation running every 7 minutes. There would be so much talk on the radio that it would be useless. There should be no need to talk every time a train stops. That is why we have signals. Yes signals can fail, but how many precautions can you justifiably take for an extreme event (all the while being able to move people efficiently)? What's next? Hire rear flagmen for the metro? The radio could fail as well. Better send someone out with fusees and torpedos. Even in freight ops we don't announce to the world every time we stop for a signal. That is why we have a signal system. If there's a train stopped ahead, we get a restricting and proceed prepared to stop in half the distance. No radio needed.

Except, depending on who you work for or where you're operating, you do exactly that in freight operations. CSX requires that every fifteen minutes, you announce your stopped position. That's the only one I know of that requires a constant rebroadcast. One thing I've learned from reading this and other forums is to never write a generic answer based on your own experiences to the exclusion of everything else. That might be the way YOU do it but not the way EVERYONE does it. OTOH, its sometimes hard not to snap at some railfan answer that is completely off base. Patience! LOL!

[zugmann]

I wasn't trying to snap at anyone.  But there's a difference b/t a section of busy class-1 mainline with a dozen or so trains on it, and a metro line at rush hour with trains spaced every 2 minutes.  (2 minutes...wow).   We do sometimes tell trains pulling in behind us where exactly we are and what our lenght is, but that info isn't 100% neccisary as that train behind us must be following restricted speed rules. 

 

When you're as vusy as metro, I doubt it would be very safe to have every train announce every time tehy pause for a few seconds until they can occupy the next block. Unless something goes wrong, they shouldn't be sitting there very long. 

 

But that's my opinion.  Take it for the electrons it's written with. 

[nagle]

 I'm not seeing much solid information here. First, read the July 1 NTSB report.  The NTSB has already determined that it's a track circuit problem. But they're still not sure what failed.

For background, read one of the previous NTSB reports on Metro collisions. The signaling system is classic General Railroad Signal relay-based fixed blocks with audio-frequency track circuits. That's century-old technology, and Metro is still using the original 1970s trackside systems. There are no computers in the basic train protection system.

The NTSB currently suspects problems with an impedance bond between two blocks. (An impedance bond passes traction power but not signal info.) This puzzles me. An impedance bond that fails should not result in a false clear indication. A short between adjacent blocks should result in a train in either block appearing to be present in both. An open should result in a false indication of train presence. Adjacent blocks should use different audio frequencies, so leakage isn't normally a problem. How did this fail?