Washington subway crash

[glatt]

It makes me wonder about the wisdom of having the trains run by computer and the drivers just there as a back-up. The computers ensure that accidents virtually never happen, but when they do, the driver isn't at 100% attention, so they don't always catch it right away. On the other hand, if the drivers were in complete control all the time so that they were paying attention, human error would eventually creep in, and there would probably be more accidents.

[tw]

Originally Posted by glatt

It makes me wonder about the wisdom of having the trains run by computer and the drivers just there as a back-up.

So that humans can perform their job, roads must be (but not always) designed so that the human can see beyond what is required to stop (which is also why superior cars use orange - not red - rear turn signals). Trains don't do that. Trains are dependent on a signaling system. Therefore the train operator could not stop the train. Train was traveling too fast around a blind corner for the operator to do anything. This is normal on most every railroad.

Blind turns are typically not a problem. Reason why this crash is reported: because train crashes are so close to zero as to be virtually zero. The system works.

And so again, the questions. Why did backup systems fail? For example, one backup system is maintenance people who are empowered to do their job. Why was this train's computer not instructed by track sensors (as latest reports suggest)?

Again, I am going right back to the reason for such failures. How often do you also use the parking brake? Exact same attitude with equal consequences. Deaths and injury created by not applying the parking brake are rarely reported, in part, because it happens.

About six months ago in a nearby town, she got out of the car to get the mail. Car in park started rolling down the driveway. She ran to jump into the car just as the car door hit a tree. She remained there for 45 minutes until someone bothered to look closer. Death directly traceable to someone who did not routinely implement (enable) a backup system. And simply too common to get equal attention.

[tw]

85% of all problems are directly traceable to top management. This crash is a perfect example of management in denial - without any grasp of the problem - without sufficient knowledge to address the problem with all due action - without sufficient attitude to listen to what the employees knew all along. Now the question - what is the education of all top managers in the Washington Metro? They did not even know how bad the system's signalling was?

From the Washington Post of 21 Jul 2009: D.C. Metro Circuit Failures May Be Widespread, Officials Say

Any attempt to selectively quote the more important points would only subvert the entire message in that news report.

[Flint]

Okay, did somebody hack tw's account? I can't read these sentences with a straight face.

[tw]

Originally Posted by Flint

Okay, did somebody hack tw's account? I can't read these sentences with a straight face.

So you loved it when seven Challenger astronauts will murdered for the same reason. So you loved it when so many were murdered on the Washington subway as management knew Metro signalling was failing everywhere in the system? In many cases, the solution was to turn off the safety system. Often what empolyees do when management does not provide the necessary support, attitude, and knowledge. So that the safety system did not cause a crash, the solution was to turn off safety devices - when management refused to address the problem.

[joelnwil]

Well, the problem with Metro circuits is widespread.

http://www.washingtonpost.com/wp-dyn...072301575.html

"According to the National Transportation Safety Board, the track circuit at the crash site has functioned irregularly since December 2007. Its performance deteriorated further five days before the June 22 crash, when it apparently failed to detect the presence of a train outside the Fort Totten Station, permitting another to plow into it. Nine people were killed, and 80 were injured.

The new information from the NTSB suggests that as long as 18 months ago, Metro officials could have found defects in the track circuitry where the accident occurred. Metro officials said Thursday that they did not know the circuit had been intermittently malfunctioning because they did not conduct tests that would reveal that specific problem.

Metro General Manager John B. Catoe Jr. likened the different level of analysis to a blood test. "When you have blood drawn, you can ask for test A through 25," he said. "We were only doing test one." "

In other words, we could have found it if we had looked. I am not impressed.

"Federal investigators found that the circuit began "fluttering," or intermittently malfunctioning, after Metro crews installed a device known as an impedance bond, also called a Wee-Z bond, at the circuit in December 2007, according to a safety board advisory issued Thursday. Metro has been installing new bonds across the 106-mile railroad as part of a project to boost power so the agency can run more eight-car trains, which consume more electricity than shorter trains. Each track circuit has two Wee-Z bonds.

The fluttering indicated a problem with the circuit, according to the data examined by the NTSB. After Metro crews replaced the second Wee-Z bond in the same circuit June 17, the circuit deteriorated to the most dangerous stage: It intermittently failed to detect the presence of a train. Five days later, a train idling in that circuit outside the Fort Totten Station was hit from behind by another train. "

http://www.washingtonpost.com/wp-dyn...072203524.html

"In addition, a source with knowledge about Metro's repairs said Wednesday that 11 track circuits on the Orange Line, from Minnesota Avenue to Deanwood and up to New Carrollton, were recently found to have problems detecting the presence of trains. One was disabled Monday night and put back into service Tuesday, said the source, who asked to remain anonymous for fear of retaliation. "

Which says to me that now that they are looking at stuff they are seeing a lot of problems previously ignored. I think that we were very lucky to go as long as we did without an accident.

I am glad I do not ride the Metro. And I do not think that important information should have to be given out anonymously, "for fear of retaliation".

[Shawnee123]

Sheesh. Have they checked the o-rings?

You think there'd be a whole system in place for testing and calibration and errors and fixes...scary.

[wolf]

I think they don't bother because 99.9998% of rail crashes are attributed to operator error. Not saying that's the actual cause, just saying that's where the blame is laid.

[dar512]

I think you're a bit hasty in flipping the bozo bit, Glatt. I didn't see anything in that article that says that they've been able to reproduce the conditions that caused the accident. And I can tell you from experience that debugging can be hard. Especially so when you've got hardware and software involved. Not to mention that part of the circuit is train tracks that live outdoors.

Before pointing and laughing at these folks I'd want to know if they have access to all the circuitry and microcode involved. If they don't, they're trying to debug this thing blind. I'd also like to know if they've been allowed to try to reproduce this under working conditions. They may not. Metro may not want to risk squishing one or two of their trains to find the answer to this issue.

[glatt]

Originally Posted by dar512

I didn't see anything in that article that says that they've been able to reproduce the conditions that caused the accident.

It wasn't in the article I just linked to, but there was a test shortly after the accident that recreated the conditions and the result was identical to the crash, except they didn't actually smash the trains together.

Article about it here. Three months ago.

Investigators with the National Transportation Safety Board performed the simulation Wednesday night. In the test, investigators positioned a train in the same location as the train that was rear-ended Monday. The system failed to detect that the idled test train was there

I know it's not easy to figure out, and they are only human, but it's been 3 months. They have had the track closed to train traffic for that entire time, available for study. It's just sitting there. They have to single track around the site and it's causing serious delays for the poor saps on the Red line.

The sensor manufacturer is Alstom Signaling, and it's some sort of problem with their sensors. Or a problem with how their sensors are installed into the system.

Shawnee is right with her o-ring joke. We need Richard Feynman to stick one of these sensors into a glass of ice water and figure it out.

[glatt]

Just for the record, they finally opened that stretch of track this week. But they still don't know what's wrong with the sensors, so all the trains are still being operated manually. My guess is that this is now how things are going to run on the system. The new standard.

[tw]

Originally Posted by Shawnee123

You think there'd be a whole system in place for testing and calibration and errors and fixes...scary.

From the Washington Post of 23 Oct 2009:

Metro to test new software for crash-avoidance system
After a fatal Red Line crash in June, federal investigators said Metro's crash-avoidance system was inadequate and called for the agency to implement a real-time backup. …
Metro uses a software program to check for circuit malfunctions. Since the Red Line crash, those software checks have been run twice a day to look for anomalies. If problems are found, crews are sent to inspect the circuit; if necessary, adjustments are made on the spot.

No problem! If a collision avoidance system fails, a program will discover it up to 12 hours later. IOW report why those two trains collided. 85% of all problems are directly traceable to top management. When management is technically dumb, then a protection system (that does not execute during the rush hour and only operates twice a day) is more than sufficient.

After all, trains don’t collide. When the collision avoidance system failed repeatedly previously, trains did not collide. Therefore don’t worry; be happy. What any management says when employees work for them.

Some believe the operator was at fault. Even that is often directly traceable to top management that did not provide the necessary atttitude and knowledge.

[tw]

Another indication of a company that may have management problems. From ABC News on 23 Oct 2009:

Cockpit Voice Recorder Unlikely to Provide Answers in Northwest Overshoot
Investigators trying to determine why two Northwest Airlines pilots were out of contact with air traffic controllers for more than an hour Wednesday night might have a hard time trying putting all the pieces together. ...

Asked how many years of service the pilots have and how many hours they were into their shift, a Delta spokesman said, "We are not sharing as that is all part of the investigation." ...

Monday at 6:05 a.m., Delta flight 60 from Rio de Janeiro to Atlanta landed on a taxiway instead of the parallel runway where it was supposed to touch down.

[tw]

Management failures continue in the Washington Metro system. Over a month ago, a track worker was killed. Trains are supposed to be notified of track work ongoing, also see warning signs that track workers are ahead, and slow to 10 MPH in that construction area. And yet many workers note trains fly through construction areas at 50 MPH - full speed.

Death only in the Washington Metro account for over 40% of all American railroad workers. The 22 Jun crash and death of six passengers did what? Yesterday, add two more deaths. From the Washington Post of 27 Jan 2010:

NTSB investigates Red Line accident that killed 2 Metro track workers
Federal officials investigating the deaths Tuesday of two Metro employees are trying to determine why the driver of a Metro utility vehicle did not know that the men were working behind him on the tracks. ...
Five Metro workers have been killed on the tracks in the past seven months. The safety problems, including a Red Line crash June 22 that killed nine people, have triggered an upheaval in Metro's leadership ...
Tuesday's accident occurred about 2 a.m. near the Rockville Station when a truck, modified to operate on the rails and moving in reverse, backed into two technicians who were working on the tracks as part of a separate crew.

How often do so many failures continue until someone views the only common factor in all events? Death and near death stories have been almost monthly now on the Washington area Metro. From the Washington Post of 8 Jan 2010:

Report finds slapdash safety culture at Metro
ANEW REPORT by the agency that monitors safety on Metro makes chilling reading -- nearly as chilling as the experience of some of the safety inspectors who helped write it. One team of inspectors was nearly run down by an onrushing Metro train on the tracks last month even as it was examining safety conditions for Metro track workers. According to the inspectors, the operator of the train made no attempt to slow down, let alone stop, or even to acknowledge the presence of the inspectors who had to scramble out of the train's path to safety.

That incident, which took place Dec. 10, sheds light on the some factors that have contributed to the deaths of six Metro workers hit by trains since October 2005, including two who were killed in the past five months even as the agency was conducting its review of track workers' safety. ...

They also extend to what appears to be a perverse culture when it comes to safety at Metro, in which train operators and track workers regard each other with open antagonism; Metro safety classes do not bother to teach Metro's own safety rules; and the transmission and accuracy of critical information -- for instance, the presence and location of track workers in the path of trains -- is slipshod and unreliable.

They are not called accidents. They are murder. But when a murder is directly traceable to top management, then we call it an accident.

The review found that train operators were speeding up rather than slowing down, as required, as they approached track workers who had not been fully cleared from the track right of way.

That attitude should result in assault (if not murder) charges. So weeks after the chilling report, what happens? "Two more dead in Maryland"

What did the Navy do when problems existed. Navy had a three day stand down of the entire Navy. Management addressed problems directly traceable to top management.

How did Marchionne fix Fiat? In sixty days he fired all top management. Therefore Fiat became so productive as to even buy Chrysler. Only public silence can explain so many Washington Metro murders - especially when investigations have made the reasons so flagrantly obvious.

[glatt]

The incident this week was caused by human error. There was supposed to be a lookout on the work crew watching for oncoming rail traffic. That publicly unidentified person was not doing his/her job. That's not management, that's worker incompetence. A lazy co-worker killed those two track workers the other night. The track maintenance truck that backed over them was going in reverse and had limited visibility. The track workers were supposed to be watching for it.

But I suppose you are right. Management should identify the slacker lookout and punish him/her and make sure everyone else is doing their job.

Metro is fucked right now. It's seeing reduced ridership because of the economy and because some people are turned off by all the safety problems. So it has reduced revenue. They are holding public hearings shortly to discuss whether they should cut service, or raise prices, or both. That's going to hurt them even more as it pushes more riders into their cars.