Slashdot Mirror

← Back to Stories (view on slashdot.org)

TGV Accident Caused By Excessive Speed (railwaygazette.com)

Posted by Soulskill on from the vigilance-is-important dept.

Cochonou writes: Analysis of the black boxes of the TGV (Train à Grande Vitesse) which derailed on Saturday revealed that the accident resulted from excessive speed and late braking. The test train entered a 945m-radius curve at a speed of 265 km/h, far over the maximum speed of 176 km/h. The French national railway company ruled out any other cause, such as mechanical failure or track mishap.

During test runs, a number of security features are disabled, in particular parts of the TVM system, which would have prevented any overspeed during normal service. This leaves the train speed under the sole responsibility of the driver.

The accident, which killed 11 people, occurred on the last run of the scheduled trials on the new high-speed line between Paris and Strasbourg. As more details on the accident surface, it becomes evident that this last run was performed in a festive spirit, with relatives (including children) of the employees on board, and seven people present in the train cab instead of train. This casts a shadow on the security procedures of the French national railway company: it appears that the high-speed train technology is considered so safe that the risks inherent to trials runs were somehow neglected. The two drivers and the traction inspector have been suspended pending possible criminal charges. Other changes in the management structure will probably follow.

13 of 96 comments (clear)

  1. Black box by Dan+East · · Score: 2

    Black box audio from the train revealed seven spectators in the cab chanting "Plus vite! Plus vite! Plus vite!"

    --
    Better known as 318230.
  2. The life of a test pilot ... oh wait. by goodmanj · · Score: 2

    I was gonna say, "well, seeing what happens when you go too fast is part of a test pilot / driver's job", until the article mentioned bringing kids along. Ugh, that's reprehensible.

    1. Re:The life of a test pilot ... oh wait. by Brett+Buck · · Score: 4, Insightful

      You would have been wrong, anyway. There's no mystery whatsoever what happens for a particular turn radius, center of gravity, turn bank, and speed. Someone with the necessary information can calculate the derail speed within a few mph. There's nothing to be learned by trying to test the limits.

  3. My question is... by taiwanjohn · · Score: 2

    Why isn't this automated? I know... they say it's a test run, so certain safety features are disabled, but ffs, can't you at least find an operator who knows wtf he's doing? This is just sad.

    --
    XML is like violence. If it doesn't solve your problem, you're not using enough of it. --AC
    1. Re:My question is... by Penguinisto · · Score: 2, Insightful

      I used to think the same way... then I had to go from Portland to Seattle on business... a lot.

      Turns out that the train takes the same amount of travel time (esp. when you factor in traffic), and when you add up gas and the cost of parking in Downtown Seattle (where even hotels will charge something like $40/day), it is actually somewhat *cheaper* than driving. Seattle is small enough size-wise to make most of it walkable without too much trouble.

      I'd much rather sit in a fairly cozy seat on the train, plug in the laptop, maybe grab something to eat, and have a drink or two (even coach does this). Much superior to shouting at traffic IMHO.

      --
      Quo usque tandem abutere, Nimbus, patientia nostra?
    2. Re:My question is... by nukenerd · · Score: 3, Informative

      Turns out that the train takes the same amount of travel time

      Just goes to show how slow trains are in the USA if you need to highlight that they are as "fast" as a car.

    3. Re:My question is... by Applehu+Akbar · · Score: 2

      And that's at slow US train speeds. TGVs in the most frequently used corridors would give us much better door-to-door times for regional travel, and would cut the puddlejumper clutter at our major airports. Flying would be a much better experience if it were reserved for long distances.

    4. Re:My question is... by Anonymous Coward · · Score: 2, Interesting

      It's not automated first because they're operating outside commercial speed envelope (they're authorized to overspeed as munch as 10% over commercial speed limit to check engineered margins actually exist) and second because those new tracks are often used with new hardware, and new hardware can have bugs that require emergency human action to avoid crashes. Therefore the test protocol gives maximum freedom to the engine operator.

      (likewise new planes are tested by test pilots outside normal envelope, except trains need a track to run, track is too expensive to be dedicated to tests in any meaningful length, so tests are conducted on new track before it's commercially used. If you look at train speed record on wikipedia you'll see they were all achieved on new track just before it was opened to commercial service, for similar reasons)

      The problem is that this was the final test run (meaning all the previous tests were successful, or more tests would have been scheduled), on hardware marginally improved from previous models with excellent safety record, during a week-end, and the test team treated it like a private risk-less celebration and forgot they were operating with safeties off.

  4. Re:Excessive Speed? by Anonymous Coward · · Score: 2, Informative

    Unless my math is wrong, 10.5 gun related deaths a year per 100,000 population comes about to about 93 gun related deaths a day in the US. ((322,200,000/100,000)*10.5)/365.2425 = 92.6. So you are not wrong there.

    And yes, when you're all hyped up on shit like guns, of course you make bad judgments and accidents happen.

  5. Re:It didn't have to happen by msauve · · Score: 2

    Drivin' that train, high on Champaigne...

    --
    "National Security is the chief cause of national insecurity." - Celine's First Law
  6. Re:Automate trains by GuB-42 · · Score: 2

    There are two computer-driven subway lines in Paris, with plans to develop this technology to other lines.
    I think the problem is mostly one of certification. Full authority autopilot code is held to much stricter standards than driving aids, even though they are essentially the same thing on trains.

    Also, the human driver is usually the only technical person on the train and may need to deal with various problems like various types of obstacle or malfunctions. An autopilot may guarantee the train safety by breaking before the obstacle but may be clueless about what to do afterward : people (healthy or not), animals, rocks other trains or track damage are not dealt with the same way. Should the police be called ? Is there a medical emergency ? For how long will the traffic be interrupted ? Is there a danger for other trains on the same line ? Automated subway lines mitigate the problem with an array of sensors and passive measures (walls) that would be impractical for long distances.

  7. Re:Automate trains by onkelonkel · · Score: 3, Interesting

    What resistor is that? Please elaborate.
     
    I know a bit about signals and I know of no such resistor. The equipment and control logic for US signal systems are fail-safe designs, based on the standard AREMA guidelines and any failures will cause the signals to go all red.

    --
    None of them can see the clouds; The polished wings don't care.
  8. Re:Automate trains by squiggleslash · · Score: 4, Informative

    We're very close to that. Problems are:

    1. The technology is mostly there but rarely all there. The US, for example, is rolling out PTC, which is 90% of the self-driving-train solution (though it's intended to be merely a safety upgrade), but PTC will not be universal. While Europe is way ahead of the Americas on this, largely because they're not stupid, boneheaded, and corrupt when it comes to transportation policy (and thus they take trains seriously rather than deliberately running them down, making them all but unobtainable, and then claiming nobody wants them when nobody rides once-a-day museum relics whose stations are 50 miles away from anywhere you want to go and whose speed rarely breaks 50mph) PTC is still not universal.

    2. You do, still, need equivalents of the technologies going into, for example, Google's self driving car. Did a tree fall on the track? Has heat bent the rails out of shape? Is there an idiot driving parallel to the train who's likely to jump the tracks at the next crossing (well, in fairness, human engineers can't generally deal with that either, and usually have to suffer the trauma associated with slamming on the brakes, getting out, and finding bits of someone's head on the track.) What about a washout?

    3. Yeah... unions. I hate blaming unions for anything, largely because 90% of the time when someone claims unions are the thing that killed a particular industry or stops needed reform from happening, they're making it up or at the very least massively exaggerating. In this case, however, the unions have this issue on their radar and have been fighting smaller crews, and expressed concerns that PTC = 1 engineer or eventually no engineers.

    There are automated systems out there, but they generally run in completely enclosed subway tunnels and have a high degree of human monitoring. Until PTC can be augmented with techologies that can visually and non-visually verify the tracks ahead are safe, we can't really automate any major conventional intercity railways.

    But I bet it wouldn't take a year for, say, a team made up of Google's self driving car engineers to create those technologies.

    --
    You are not alone. This is not normal. None of this is normal.