 |
Re: NTSB Press Release re: BNSF rear-end collision in Iowa
Author: OldPoleBurner
Date: 05-04-2011 - 23:02
I cannot recall when this actually was, but somewhere in the late 80s or 90s, I read an NTSB report about a rear-ender on the Conrail mainline somewhere in Eastern Pennsylvania (I think east of Pittsburg). It doesn't appear to be on-line at the NTSB website; at least I didn't find it. But maybe someone has a better memory than I, and can fill in the blanks.
The report went on at length about various recorded events leading up to the accident, and then concluded that they had proved the locomotive crew were all "involuntarily asleep" (hypnagogia); brought on by severe fatigue and sleep deprivation. This was one of the first times I recall that the Feds had actually found clear scientific evidence of this huge problem.
During the whole time, the alerter was functioning and was being acknowledged - faithfully; but talking detectors, which had recorders in them on Conrail, were not being acknowledged as required for quite some time up to the accident (they were at first). This prompted the NTSB to hire a sleep disorder laboratory to investigate. They proved that it was possible to be in this altered sleep state, and still perform many mindless actions by 'rote'. It was specifically found that acknowledging a locomotive alerter was one of the possible actions while asleep. And to some extent, even whistling for grade crossings was possible as well!
In light of all this, I have never understood the point of the alerter. It can be very ineffective. So why the FRA has a proposed rule-making requiring an ineffective device, makes reason stare. It would probably be much more effective at preventing accidents, to ensure adequate time off and rest between reasonable length work shifts. Anybody ever hear of the 8 hour workday (at least plan for approximately eight hours most of the time - go overtime only when crap goes wrong unexpectedly)
-------------------------------------------------------------------------
"Anybody know how PTC will allow trains to close up at restricted speed?"
The most important feature of PTC, beyond that of most conventional already existing American ACS/ATC systems, is that the PTC will employ "supervised braking", which compels adherence to a specific worst case speed / braking curve, in which it is always still possible to stop at or before the next target. Whereas, most ATC setups will allow 'forestalling' of a brake application, with no supervision of braking thereafter. It is possible then, with this ATC setup, to forestall in your sleep and then fail to later set the brakes.
This braking curve enforcement means that when moving at restricted speed - say 15mph, you can get a lot closer to the target without a penalty stop, than you could at 79mph. Once you are at minimum braking distance at restricted speed to the target, you must stop or you will be forcibly stopped before passing it.
If that target is a red signal (the usual case), the system must be cutout to close in beyond the red signal. This is one of the areas of PTC that is still quite controversial. Just how to do that and still ensure safety is in still heated debate in many quarters. The basic problem, is that for all practical purposes, existing PTC Technologies cannot reliably determine the location of anything not PTC equipped, such as freight and passenger cars. Given the mad dog rush into to this PTS buffoonery, it is likely that any "closing in" movement beyond a red signal, will require some sort of cutout, or at least holding a momentary override lever against a spring while in motion.
I do not know how this has been worked out (if it has), because I dropped out of working on this stuff when I found out that train location was to be expressed as a statistically calculated 'Norm', accompanied by a 'PUF' factor. That's "Position Uncertainty Factor" PUF - folks!
That was quite enough to set off my BS detectors folks - flashing bright red! And also that of just about every other signal engineer I have talked to in past three years; except one - and he worked for guess who - the PTC vendor!
OPB
The report went on at length about various recorded events leading up to the accident, and then concluded that they had proved the locomotive crew were all "involuntarily asleep" (hypnagogia); brought on by severe fatigue and sleep deprivation. This was one of the first times I recall that the Feds had actually found clear scientific evidence of this huge problem.
During the whole time, the alerter was functioning and was being acknowledged - faithfully; but talking detectors, which had recorders in them on Conrail, were not being acknowledged as required for quite some time up to the accident (they were at first). This prompted the NTSB to hire a sleep disorder laboratory to investigate. They proved that it was possible to be in this altered sleep state, and still perform many mindless actions by 'rote'. It was specifically found that acknowledging a locomotive alerter was one of the possible actions while asleep. And to some extent, even whistling for grade crossings was possible as well!
In light of all this, I have never understood the point of the alerter. It can be very ineffective. So why the FRA has a proposed rule-making requiring an ineffective device, makes reason stare. It would probably be much more effective at preventing accidents, to ensure adequate time off and rest between reasonable length work shifts. Anybody ever hear of the 8 hour workday (at least plan for approximately eight hours most of the time - go overtime only when crap goes wrong unexpectedly)
-------------------------------------------------------------------------
"Anybody know how PTC will allow trains to close up at restricted speed?"
The most important feature of PTC, beyond that of most conventional already existing American ACS/ATC systems, is that the PTC will employ "supervised braking", which compels adherence to a specific worst case speed / braking curve, in which it is always still possible to stop at or before the next target. Whereas, most ATC setups will allow 'forestalling' of a brake application, with no supervision of braking thereafter. It is possible then, with this ATC setup, to forestall in your sleep and then fail to later set the brakes.
This braking curve enforcement means that when moving at restricted speed - say 15mph, you can get a lot closer to the target without a penalty stop, than you could at 79mph. Once you are at minimum braking distance at restricted speed to the target, you must stop or you will be forcibly stopped before passing it.
If that target is a red signal (the usual case), the system must be cutout to close in beyond the red signal. This is one of the areas of PTC that is still quite controversial. Just how to do that and still ensure safety is in still heated debate in many quarters. The basic problem, is that for all practical purposes, existing PTC Technologies cannot reliably determine the location of anything not PTC equipped, such as freight and passenger cars. Given the mad dog rush into to this PTS buffoonery, it is likely that any "closing in" movement beyond a red signal, will require some sort of cutout, or at least holding a momentary override lever against a spring while in motion.
I do not know how this has been worked out (if it has), because I dropped out of working on this stuff when I found out that train location was to be expressed as a statistically calculated 'Norm', accompanied by a 'PUF' factor. That's "Position Uncertainty Factor" PUF - folks!
That was quite enough to set off my BS detectors folks - flashing bright red! And also that of just about every other signal engineer I have talked to in past three years; except one - and he worked for guess who - the PTC vendor!
OPB
