Welcome to European Tribune. It's gone a bit quiet around here these days, but it's still going.
Display:
The news coming out in the media is too confused and contradictory for me to write anything diary-length. I commented on the technical aspects of the accident on first news (when it was reported that lightning hit the train) here, and yesterday here. As a further comment, whatever was the real reason, this train control failure and the public management of the accident paint an even worse picture of recklessness than what I inferred from the 'overspeed' practice and the 380 km/h ambitions.

*Lunatic*, n.
One whose delusions are out of fashion.
by DoDo on Sat Aug 6th, 2011 at 04:39:47 AM EST
[ Parent ]
A recent post doesn't give much light on the subject, but I include it for coverage:

http://www.businessinsider.com/details-chinese-rail-crash-2011-8

In general, it seems that both technology and humans failed in a spectacular (and deadly) fashion.

N.F.

by nfotis on Sun Aug 7th, 2011 at 11:48:58 AM EST
[ Parent ]
In the meantime, Chinese government recalled 54 trainsets for checking/retrofit (if I guess correctly):

http://www.rail.co/2011/08/12/china-recalls-bullet-trains-following-high-speed-crash/

This doesn't jibe with the official line that the signaling system was at fault, though...

N.F.

by nfotis on Sun Aug 14th, 2011 at 08:45:20 AM EST
[ Parent ]
From other news, the recall affects the CRH380B type (the improved version of the CRH3C, that is the Siemens Velaro CN):

China recalls bullet trains in new blow to technology - Yahoo! News

Train maker CNR's President Cui Dianguo said the CRH380B-trains had "experienced a number of quality failings," and the recall would lead to a cut in services on the 1,318 kilometer (820 mille) Beijing-Shanghai route, which takes about 5 hours.

In an earlier statement, the state-owned company said it found faults in signaling systems, sensors and communications components on the trains, which it blamed for delays in service.

This must be unrelated (unless they found the train control system error during post-accident safety checks).

*Lunatic*, n.
One whose delusions are out of fashion.

by DoDo on Tue Aug 16th, 2011 at 02:47:49 AM EST
[ Parent ]
Missed this! Instead of the linked blog post (which misinterprets some points), let me go through its source, the original Caing story.

Rapid Rail Horror on a Rainy Zhejiang Night_English_Caixin

D3115 had stopped several times because the engineer had received track warnings from a dispatcher at a control station in far away Shanghai. The dispatcher's messages were based on "red-light zone" messages flashing on his computer monitor.

...Only dispatchers are aware of these zones; locomotive engineers receive no such data.

Of course he receives data; what they might mean is that the dispatcher sees the red-light zone in advance, and sees the whole extent of the red-light zone.

At 8:15 p.m. on July 23, the D3115 had yet to clear a red-light zone. But the dispatcher decided to let it depart Yongjia slowly. A few minutes later, the train stopped again as it was approaching the red-light zone.

Passing a suspected malfunctioning signal at slow speed is standard procedure; you are practically progressing on sight until the next signal which is hoped to function correctly. (On conventional lines without central dispatching, locomotive drivers are authorised to do so on their own.) But here the next signal was red, too, so the signal disturbance appears to have been broader.

Red-light zones can be problematic for dispatchers, who are under pressure to keep trains running on time.

This can only get problematic if dispatchers can overrule the automatic safety system to the extent of ordering them to progress full speed ahead into a red light zone, which would indeed be a glaring safety error. However, this wasn't established in the article; they only spoke of the order for the first train to move ahead slowly.

The dispatcher told the D3115 engineer to proceed despite the red-light zone and ordered "visual driving," meaning the speed could not exceed 20 kilometers per hour. But a few minutes earlier, dispatchers from the same office had given the D301 permission to proceed from Yongjia, at a top speed of 200 kilometers per hour.

Again, the problematic part is the permission given to the second train (D301); and how it could have been given in the first place.

For example, a group of foreign experts told Caixin that locomotive engineers need at least three months of training before they can safely operate a high-speed train. The Chinese ministry, however, sponsors only 10-day courses for new engineers, said a source in the rail ministry.

Huh... though, hard to tell, the two figures might correspond to different types of courses: one as training of high-speed train driving from the ground up, the other as training for experienced drivers switching from one high-speed train type to another.

In addition to fast-paced track laying, the railways ministry pushed Chinese contractors to develop high-tech signaling and safety equipment for the bullet trains.

Now that's probably a key point. We could barely get ETCS L1 going in ten years, yet China created its own version in two years.

A rail traffic management consultant explained that due to the rapid development of high-speed trains in China, "as long as a company manufactures something, it can sell it. It doesn't matter if they produce 100 substandard products. As long as they can produce them, they'll make money. Time is paramount throughout the whole industry."

Indeed this might or might not apply to the problems that hit the CRH380B. If it applies to signalling system parts, it's even worse. But it's not simply quality; it's getting around technology transfer limitations:

A source who participated in past high-speed rail technology negotiations with foreign companies recalled that "the foreign side's technology transfer had a bottom line: They would only transfer design drawings for some parts and teach us how to manufacture them. But some core things, especially software code, they wouldn't transfer."

And then the own replacement was put in in haste (a recuring theme in the article).

An, the Shanghai bureau director, said flawed signaling equipment at the Wenzhou rail station failed after being struck by lightning. After that, he said, signals seen by engineers that should have been red were mistakenly green.

"At the time of design" the signal makers "should have considered that if there were a breakdown at any time, the signals should display red lights," said Peng Qiyuan, a professor of transportation at Southwest Jiaotong University.

Just what I said: if true, this is a fundamental fail-safe deficiency.

Despite the flaws, the railway ministry and its closely linked state contractors have been unwilling to accept companies from outside their circle of business friends.

Eh. This is just my opinion, but methings suppliers from outside the rail sector (which they mean) would have made even more error-prone signalling system equipment, being inexperienced.

Rapid Rail Horror on a Rainy Zhejiang Night_English_Caixin

A backup layer of protection... rear-end collision prevention system... cannot work without properly functioning ground-based signaling equipment, which on July 23 failed.

Given the braking distances involved, some equipment for train location and communication is necessary, and indeed involving ground-based signaling equipment is the only option with current systems. But, if I read this right, the problem was signals giving bad signals, not signals out of order. Hence, it might be that the rear-end collision prevention system could have been designed to ignore the state of signals and derive only train location from the data stream, but wasn't.

Engineer Pan Yihang was driving the D301 through a tunnel and rain as these system failures occurred. As his train emerged from the tunnel, he spotted the other train on the viaduct ahead. He quickly hit the emergency brake, but it was too late.

On impact, the brake handle pierced Pan's chest and killed him.

Poor guy...

Apparently there were technology-related mistakes after the accident, too:

"Why aren't you saving people?" Yang remembers shouting at the rescuers. "There are still people in there!"

He was answered by a police officer: "They used a life detector, and there were no signs of life. There's nothing inside but dead bodies."

Yet the rescue team's life detector, a special piece of equipment for finding disaster survivors, failed to find a 2-year-old girl who was found alive in the wreckage more than 20 hours after the crash.



*Lunatic*, n.
One whose delusions are out of fashion.
by DoDo on Tue Aug 16th, 2011 at 05:13:08 AM EST
[ Parent ]
Got some more data from a Greek railfan forum
(yes, there are Greeks who love trains and everything related to these!)

We got a video link:
http://www.youtube.com/watch?v=yh7UWZZ75Yg&feature=player_detailpage#t=302s

(obviously, I do not know Chinese, if someone of the readers here can verify it, we would be grateful)

Here is the list of the events according to the video and a reader:

  • Signals near Wenzhou South were found to be malfunctioning, all showing red.

  • D3115 arrived into Yongjia (north of Wenzhou South) 4 minutes late

  • A decision was made that the signals are to be over-ridden and manual operation commenced from Wenzhou South control centre

  • Manual operation commenced from Yongjia control centre

  • D301 was informed of manual operation and made an unscheduled stop at Yongjia at 20:12

  • D3115 departed from Yongjia 27 minutes late

  • D301 was given the clear to proceed

  • D3115 reached the start of signal failure area and stopped

  • D301 departed from Yongjia at high speed, with his signals all showing clear

  • D3115 began running towards Wenzhou South at 20 km/h

  • D3115 driver reported a passenger operated emergency stop, unaware his train has just been rear-ended.

Some comments from a Greek poster:

It's not known why they didn't wait for the preceding train to reach the next station before sending off the next one during the manual operation
(N.F. this is supposedly the only safe way to run in 'dark territory')

It was written that the trains got directions from two different control centers, there was pressure to minimize delays due to bad publicity of previous days, and that they were getting a bonus when they reduce delays.
Due to the language barrier, it is impossible to verify anything.

The dead toll could be MUCH worse, since 1600 people were in the two trains.

There was no understandable motive behind the decision to bury the damaged coaches without giving clear and consistent explanations, raising a major outcry:

N.F.
by nfotis on Thu Aug 18th, 2011 at 08:24:22 AM EST
[ Parent ]
D3115 driver reported a passenger operated emergency stop, unaware his train has just been rear-ended.

Ah, that inconsistency is cleared up now, too.

It's not known why they didn't wait for the preceding train to reach the next station before sending off the next one during the manual operation
(N.F. this is supposedly the only safe way to run in 'dark territory')

Ah, it finally starts to make sense. So all this dispatching was radio dispatching after the automatic signalling system was ruled malfunctioning, and the key mistake seems to be the launch of two trains within station distance.

It was written that the trains got directions from two different control centers

I would think normally, each control centre at the two ends of a section with failed train protection would dispatch trains in one direction. If D3115 and D301 were dispatched by two different control centres, that explains the two trains within station distance, and constitutes another very fundamental mistake. But, even so, it remains unexplained how the dispatcher of D301 could have missed the train ahead on his screen.

there was pressure to minimize delays due to bad publicity of previous days, and that they were getting a bonus when they reduce delays.

This is discussed in the Caing article, but, without seeing how exactly it was technically possible in the system/within the rules to dispatch two trains into the same section, and how it was possible to miss the train ahead, I'm not sure if this is relevant.

*Lunatic*, n.
One whose delusions are out of fashion.

by DoDo on Thu Aug 18th, 2011 at 09:51:43 AM EST
[ Parent ]

Display:

Occasional Series