C
cdl42
Guest
Everyting you would ever want to know about the causes of the 1995 Russel Hill subway crash...
From Transit Toronto.
-----------------------------------------
John Barber Quotes the Wrong Mantra
Reader and contributor Mark Brader wrote in with an interesting response to John Barber’s interview with TTC General Manager Rick Ducharme. Specifically, he takes Barber to task for Barber’s own words in the centre of this paragraph:
“The top priority is state-of-good-repair,†he said yesterday, repeating a phrase that has become a mantra at the TTC ever since a worn-out signalling system killed three people on the Spadina line 10 years ago. “Over the next four or five years I need about $700-million a year … and it’s not negotiable.â€
Brader writes:
The 1995 crash did not happen because the “signalling system†was “worn outâ€.
It happened for two primary reasons. [1] An inexperienced and perhaps undertrained driver misremembered a signal, thinking a caution yellow-over-green had been a grade-timing (GT) yellow-over-white. And [2] in order to use a particular component in the signaling system (specifically, in the train stop) the TTC had modified its design wrongly, in a way that nullified its fail-safe behavior in the exact situation when it was most needed. Responsibility for this error is shared, as it actually occurred at a subcontractor.
I said “misrememberedâ€. This is relevant because of a family of secondary causes. [3] The TTC uses lineside signals and not cab signals, and as usual on railways where this is the case, the driver is expected to remember each signal and act accordingly until new information is received from the next one. In fact, it could reasonably be said that this is the most important part of the driver’s job. Further, [4] with the GT signals as they then existed, the signal after a GT signal remained red until the train was about to pass it, so the red signal did not in itself give new information to the driver; only at the last moment, when it cleared to yellow or green, did the driver receive information about the next block. The red signal did even not confirm that it was part of a GT sequence, as the driver was expected to remember that. Finally, [5] a culture of shaving safety margins had developed, which meant that drivers would come so close to the red after a GT signal that they would not see it change, but this meant that they never got confirmation that a GT red actually was a GT red except by passing it without the train being tripped.
In response to the accident, the TTC addressed cause 1 by strengthening training, cause 2 by fixing the design error, cause 4 by introducing a flashing red aspect to distinguish the GT situation from others, and cause 5 by instructing drivers not to do that. The accident also drew attention to risk factors that had not been involved, and these were also addressed; and the lack of a “state of good repair†was one of those.
There is a place where “wear†comes into the story, but it’s normal wear, not something being “worn outâ€. There is a defined profile (the clearance envelope or loading gauge) that a train has to fit within, and lineside equipment (such as the train stop) has to stay outside of. The modified train stop was out-of-gauge as constructed. But the train is not a rigid mass fixed to the track, so the clearance envelope includes some space for dynamic clearance. That is, the train does not, at any one time, fill the entire profile of space that it has a “right†to fill. So it was possible that the train stop would not be struck even though it was out-of-gauge, and as it happened, it wasn’t.
However, normal wear of the wheels and track over time will change the exact position that the train occupies within the loading gauge. In this case it changed it in such a way that the out-of-gauge train stop was not only hit, but happened to be hit in such a way that if the train was moving at full speed, then the raised trip arm was momentarily knocked down and the train it could pass it without being tripped. And that is what happened.
Thanks, Mark!
(Update, 12:34 p.m.): Calvin Henry-Cotnam writes with his comments on Mark Brader’s comments:
I have to comment on something in Mark Brader’s comments. It is the use of the word “misrememberedâ€.
While I don’t totally dismiss this as a possibility, I do not recall the inquiry coming to the conclusion that it was a case of misremembering. This term suggests that the driver may have had an accurate interpretation but somehow forgot it in the next couple of minutes before the crash occurred.
Mark suggests that the driver may have interpreted a yellow-over-green as a yellow-over-white signal, which is a misunderstanding of the situation, something called an inaccurate situational awareness.
I would also suggest that another possibility may have been that a brief reflection on the white indicator resulted in the driver interpretation that the white grade time indicator was in fact lit as he passed it. This would lead to the incorrect situational awareness that the next signal was red for grade time reasons and all the secondary issues that Mark cites fall into place. I have seen such reflections myself, and while the timing for viewing such would be slightly different sitting in the driver’s position, all it takes is a glance at the right moment to register the event (that the white light is lit) in one’s mind. Glancing just a split second before the reflection occurred would support the idea in one’s mind that the light just became lit, which would be consistent with what happens when traffic is bunched up and headways are limited, as was the case at the time of the accident.
I do not believe that it was a case of “misrememberingâ€. I believe that the driver remembered perfectly fine for the two or so minutes the train travelled through the block. Passing a signal that still appeared red without being tripped reinforced this incorrect situational awareness. However, what he remembered was in error from the point he passed the signal, whether he interpreted a yellow-over-green as a yellow-over-white, or a reflection caused him to see a yellow-over-green-over white. Once he had the wrong impression of the state of things, the stage was set for everything else to occur.
(Update: Wednesday, March 15 at 8:28 a.m.): Steve Munro writes to comment on the comment to the comment, and on the comment itself. No comment.
One other matter that was not covered in enough detail was a fundamental design flaw in the signal system. The TTC sets up its block lengths roughly corresponding to the anticipated speed of operation. Long blocks where trains run fast, short blocks where they run slowly. The idea is that you need more clearance in the fast sections in case you have to stop a runaway train.
However, this scheme fails to take into account the situation where you are operating at speed around a blind curve as in the Russell Hill case. In this situation, you don’t have the added safety factor of actually seeing the train ahead of you as would be the case in a straight tunnel section, no matter what colour the signal was. The way to deal with this is with shorter blocks (ie half-blocks so that you have more than the standard two-red-behind-a-train setup) as in fact we find on a few locations in the system. This ensures that a signal is always visible to the operator regardless of speed or tunnel geometry. Ideally of course you have cab signals and even some form of non-mechanical train control mechanism (that’s another big topic I won’t go into here).
Your other writers have picked up the points that needed saying.
Thank you Steve! Thank you Calvin and Mark for producing a very interesting article indeed.
From Transit Toronto.
-----------------------------------------
John Barber Quotes the Wrong Mantra
Reader and contributor Mark Brader wrote in with an interesting response to John Barber’s interview with TTC General Manager Rick Ducharme. Specifically, he takes Barber to task for Barber’s own words in the centre of this paragraph:
“The top priority is state-of-good-repair,†he said yesterday, repeating a phrase that has become a mantra at the TTC ever since a worn-out signalling system killed three people on the Spadina line 10 years ago. “Over the next four or five years I need about $700-million a year … and it’s not negotiable.â€
Brader writes:
The 1995 crash did not happen because the “signalling system†was “worn outâ€.
It happened for two primary reasons. [1] An inexperienced and perhaps undertrained driver misremembered a signal, thinking a caution yellow-over-green had been a grade-timing (GT) yellow-over-white. And [2] in order to use a particular component in the signaling system (specifically, in the train stop) the TTC had modified its design wrongly, in a way that nullified its fail-safe behavior in the exact situation when it was most needed. Responsibility for this error is shared, as it actually occurred at a subcontractor.
I said “misrememberedâ€. This is relevant because of a family of secondary causes. [3] The TTC uses lineside signals and not cab signals, and as usual on railways where this is the case, the driver is expected to remember each signal and act accordingly until new information is received from the next one. In fact, it could reasonably be said that this is the most important part of the driver’s job. Further, [4] with the GT signals as they then existed, the signal after a GT signal remained red until the train was about to pass it, so the red signal did not in itself give new information to the driver; only at the last moment, when it cleared to yellow or green, did the driver receive information about the next block. The red signal did even not confirm that it was part of a GT sequence, as the driver was expected to remember that. Finally, [5] a culture of shaving safety margins had developed, which meant that drivers would come so close to the red after a GT signal that they would not see it change, but this meant that they never got confirmation that a GT red actually was a GT red except by passing it without the train being tripped.
In response to the accident, the TTC addressed cause 1 by strengthening training, cause 2 by fixing the design error, cause 4 by introducing a flashing red aspect to distinguish the GT situation from others, and cause 5 by instructing drivers not to do that. The accident also drew attention to risk factors that had not been involved, and these were also addressed; and the lack of a “state of good repair†was one of those.
There is a place where “wear†comes into the story, but it’s normal wear, not something being “worn outâ€. There is a defined profile (the clearance envelope or loading gauge) that a train has to fit within, and lineside equipment (such as the train stop) has to stay outside of. The modified train stop was out-of-gauge as constructed. But the train is not a rigid mass fixed to the track, so the clearance envelope includes some space for dynamic clearance. That is, the train does not, at any one time, fill the entire profile of space that it has a “right†to fill. So it was possible that the train stop would not be struck even though it was out-of-gauge, and as it happened, it wasn’t.
However, normal wear of the wheels and track over time will change the exact position that the train occupies within the loading gauge. In this case it changed it in such a way that the out-of-gauge train stop was not only hit, but happened to be hit in such a way that if the train was moving at full speed, then the raised trip arm was momentarily knocked down and the train it could pass it without being tripped. And that is what happened.
Thanks, Mark!
(Update, 12:34 p.m.): Calvin Henry-Cotnam writes with his comments on Mark Brader’s comments:
I have to comment on something in Mark Brader’s comments. It is the use of the word “misrememberedâ€.
While I don’t totally dismiss this as a possibility, I do not recall the inquiry coming to the conclusion that it was a case of misremembering. This term suggests that the driver may have had an accurate interpretation but somehow forgot it in the next couple of minutes before the crash occurred.
Mark suggests that the driver may have interpreted a yellow-over-green as a yellow-over-white signal, which is a misunderstanding of the situation, something called an inaccurate situational awareness.
I would also suggest that another possibility may have been that a brief reflection on the white indicator resulted in the driver interpretation that the white grade time indicator was in fact lit as he passed it. This would lead to the incorrect situational awareness that the next signal was red for grade time reasons and all the secondary issues that Mark cites fall into place. I have seen such reflections myself, and while the timing for viewing such would be slightly different sitting in the driver’s position, all it takes is a glance at the right moment to register the event (that the white light is lit) in one’s mind. Glancing just a split second before the reflection occurred would support the idea in one’s mind that the light just became lit, which would be consistent with what happens when traffic is bunched up and headways are limited, as was the case at the time of the accident.
I do not believe that it was a case of “misrememberingâ€. I believe that the driver remembered perfectly fine for the two or so minutes the train travelled through the block. Passing a signal that still appeared red without being tripped reinforced this incorrect situational awareness. However, what he remembered was in error from the point he passed the signal, whether he interpreted a yellow-over-green as a yellow-over-white, or a reflection caused him to see a yellow-over-green-over white. Once he had the wrong impression of the state of things, the stage was set for everything else to occur.
(Update: Wednesday, March 15 at 8:28 a.m.): Steve Munro writes to comment on the comment to the comment, and on the comment itself. No comment.
One other matter that was not covered in enough detail was a fundamental design flaw in the signal system. The TTC sets up its block lengths roughly corresponding to the anticipated speed of operation. Long blocks where trains run fast, short blocks where they run slowly. The idea is that you need more clearance in the fast sections in case you have to stop a runaway train.
However, this scheme fails to take into account the situation where you are operating at speed around a blind curve as in the Russell Hill case. In this situation, you don’t have the added safety factor of actually seeing the train ahead of you as would be the case in a straight tunnel section, no matter what colour the signal was. The way to deal with this is with shorter blocks (ie half-blocks so that you have more than the standard two-red-behind-a-train setup) as in fact we find on a few locations in the system. This ensures that a signal is always visible to the operator regardless of speed or tunnel geometry. Ideally of course you have cab signals and even some form of non-mechanical train control mechanism (that’s another big topic I won’t go into here).
Your other writers have picked up the points that needed saying.
Thank you Steve! Thank you Calvin and Mark for producing a very interesting article indeed.