 |
Re: to the Old Pole Burner
Author: OldPoleBurner
Date: 12-18-2009 - 11:24
>> 1. Is the new, mandated positive train control system a g.p.s.-based system using satellites?
>> 2. Is such a system fail-safe?
>> 3. On a double-or-more-track railroad, can the system determine with certainty which track he train in on?
>> I would like you to explain why the new p.t.c. is not adding any safety to operations.
I don't have a lot of time today - the boss is pressing me, but I'll give it a quick go.
The vendors pushing this beyond and before A.R.E.M.A. signal committee consensus, are pushing a GPS based system. Most of the experimental work I was involved with, used fixed location ground based radio beacons. The GPS is a radio ranging method that measures instantaneous distance to several satellites at once. Radio ranging is done by timing the ethereal propagation delay between broadcast at the transmitter, and reception at the object being measured.
By a complex process of mathematical triangulation with several satellites, the earthly coordinates (long. lat. ele.) of the receiver can be approximated. However, in the case of satellite based radio ranging (GPS), the location of each transmitter is variable due to the non-geosynchronous orbit of each satellite.
This then requires the use of a dynamic database called a satellite ephemeris, which predicts where each satellite will be at each instant. Because of significant wobble and imperfect orbit, this ephemeris must be updated frequently. The GPS receiver must constantly update itself and do all these calculations on the fly - in real time.
While this is already complicated enough - we ain't done yet! The approximation of earthly coordinates (long. lat. ele.) of the receiver, must now be converted again; to determine where a particular consist is, linearly along the track (what milepost +/-). Now, a second ephemeris must be referenced, describing the location of the trackway, every single feature on the right of way, and the instantaneous status of every single appliance, such as signals, switches, grade crossings, beacons, tags, etc, along with boundaries of slow orders and work limits.
The approximation obtained from the GPS is now compared to the trackway ephemeris, and the closest location, in linear position on the trackway is selected. That data is then used by the train itself, to compare its position along the trackway with other track features, appliances and signals, to generate a movement authority and a current speed limit - which is then enforced.
Much of this trackway ephemeris is extremely safety critical. It's safety critical data is also its most dynamic - having a safe half-life of approximately one second. After that the data is stale and unsafe - must then not be used in any safety critical way. Therefore it is a must that dynamic field data be refreshed continuously. This is no small task, and requires extremely robust and triple redundant CPU hardware and extremely hardened software.
Considering the complexity of the task we are asking of this safety critical machinery, it definitely pushes beyond the state of the art. Just the task of performing the math, alone; has no known safe default (i.e. defaulting to a more restrictive aspect). Therefore, it cannot be made failsafe by any known methodology.
Moreover, with GPS, we are speaking of approximations. No two raw radio ranged measurements are ever the same - so the question begs - which measurement do we rely on? This is due to a number of atmospheric issues, and as in any nano-second realm - plain old "electronic jitter". Radio ranging methods therefore periodically take many raw measurements to generate each approximation in a very complex (again) mathematical / statistical process, commonly referred to as a Kalman filter - a sort of averaging if you will.
Whereas a stationary receiver, will eventually resolve down to just a few feet (some claim inches - but I'm skeptical) if left there long enough; moving objects being measured by other moving objects gets downright dicey. In actual experience, even the ground based systems I worked on, could not reliably and safely locate a moving object to closer than +/- 100 feet and sometimes as bad as +/-300 feet. No GPS system alone, with its in-motion transmitters, will ever even match that, for safety critical measurements.
Of course, it is now obvious why we speak of location along the trackway, instead of along the track. No radio ranging method alone can currently prove which track you are on. As long as only one track in a r.o.w is thus signaled, that won't matter - such as at Harmon's installation out of Detroit. But with multiple tracks so signaled , the train must know "Vitally" which track it is on - and must be continuously detectable. This is currently a major stumbling block, which may require extensive new hardware all along each parallel track.
Most proposals so far, involves very sketchy and intermittent detection of such, which if lost (it happens often), leaves that train and all tracks in the area in limbo. This has proved to be an quite unacceptable operationally. So manual entry of which track your on, gets resorted to - which of course, re-introduces human error to add to the already error prone machinery.
I did participate years ago in an accident investigation, where a heavy maintenance hi-railer reported itself on the wrong track and was hit cab-car first, jackknifing the train in the process. If any human or machine error CAN cause an accident - it WILL!
GPS based technologies will be very useful in a lot of important ways to railroads, especially as a means of finesse train handling, timekeeping, and such. But as a means to provide safety and collision avoidance, it will need a lot of help from a lot of new wayside appliances.
But since some of those additional wayside devices that would make it all work, can already provide all the "Positive Train separation" desired, all on their own; I fail to see the point of spending tens of billions on new but inadequate stuff, when we could spend a order of magnitude less for the same benefit - and much quicker too.
Hope this answers your questions, Bert . . . I gotta run. . . .
OPB
>> 2. Is such a system fail-safe?
>> 3. On a double-or-more-track railroad, can the system determine with certainty which track he train in on?
>> I would like you to explain why the new p.t.c. is not adding any safety to operations.
I don't have a lot of time today - the boss is pressing me, but I'll give it a quick go.
The vendors pushing this beyond and before A.R.E.M.A. signal committee consensus, are pushing a GPS based system. Most of the experimental work I was involved with, used fixed location ground based radio beacons. The GPS is a radio ranging method that measures instantaneous distance to several satellites at once. Radio ranging is done by timing the ethereal propagation delay between broadcast at the transmitter, and reception at the object being measured.
By a complex process of mathematical triangulation with several satellites, the earthly coordinates (long. lat. ele.) of the receiver can be approximated. However, in the case of satellite based radio ranging (GPS), the location of each transmitter is variable due to the non-geosynchronous orbit of each satellite.
This then requires the use of a dynamic database called a satellite ephemeris, which predicts where each satellite will be at each instant. Because of significant wobble and imperfect orbit, this ephemeris must be updated frequently. The GPS receiver must constantly update itself and do all these calculations on the fly - in real time.
While this is already complicated enough - we ain't done yet! The approximation of earthly coordinates (long. lat. ele.) of the receiver, must now be converted again; to determine where a particular consist is, linearly along the track (what milepost +/-). Now, a second ephemeris must be referenced, describing the location of the trackway, every single feature on the right of way, and the instantaneous status of every single appliance, such as signals, switches, grade crossings, beacons, tags, etc, along with boundaries of slow orders and work limits.
The approximation obtained from the GPS is now compared to the trackway ephemeris, and the closest location, in linear position on the trackway is selected. That data is then used by the train itself, to compare its position along the trackway with other track features, appliances and signals, to generate a movement authority and a current speed limit - which is then enforced.
Much of this trackway ephemeris is extremely safety critical. It's safety critical data is also its most dynamic - having a safe half-life of approximately one second. After that the data is stale and unsafe - must then not be used in any safety critical way. Therefore it is a must that dynamic field data be refreshed continuously. This is no small task, and requires extremely robust and triple redundant CPU hardware and extremely hardened software.
Considering the complexity of the task we are asking of this safety critical machinery, it definitely pushes beyond the state of the art. Just the task of performing the math, alone; has no known safe default (i.e. defaulting to a more restrictive aspect). Therefore, it cannot be made failsafe by any known methodology.
Moreover, with GPS, we are speaking of approximations. No two raw radio ranged measurements are ever the same - so the question begs - which measurement do we rely on? This is due to a number of atmospheric issues, and as in any nano-second realm - plain old "electronic jitter". Radio ranging methods therefore periodically take many raw measurements to generate each approximation in a very complex (again) mathematical / statistical process, commonly referred to as a Kalman filter - a sort of averaging if you will.
Whereas a stationary receiver, will eventually resolve down to just a few feet (some claim inches - but I'm skeptical) if left there long enough; moving objects being measured by other moving objects gets downright dicey. In actual experience, even the ground based systems I worked on, could not reliably and safely locate a moving object to closer than +/- 100 feet and sometimes as bad as +/-300 feet. No GPS system alone, with its in-motion transmitters, will ever even match that, for safety critical measurements.
Of course, it is now obvious why we speak of location along the trackway, instead of along the track. No radio ranging method alone can currently prove which track you are on. As long as only one track in a r.o.w is thus signaled, that won't matter - such as at Harmon's installation out of Detroit. But with multiple tracks so signaled , the train must know "Vitally" which track it is on - and must be continuously detectable. This is currently a major stumbling block, which may require extensive new hardware all along each parallel track.
Most proposals so far, involves very sketchy and intermittent detection of such, which if lost (it happens often), leaves that train and all tracks in the area in limbo. This has proved to be an quite unacceptable operationally. So manual entry of which track your on, gets resorted to - which of course, re-introduces human error to add to the already error prone machinery.
I did participate years ago in an accident investigation, where a heavy maintenance hi-railer reported itself on the wrong track and was hit cab-car first, jackknifing the train in the process. If any human or machine error CAN cause an accident - it WILL!
GPS based technologies will be very useful in a lot of important ways to railroads, especially as a means of finesse train handling, timekeeping, and such. But as a means to provide safety and collision avoidance, it will need a lot of help from a lot of new wayside appliances.
But since some of those additional wayside devices that would make it all work, can already provide all the "Positive Train separation" desired, all on their own; I fail to see the point of spending tens of billions on new but inadequate stuff, when we could spend a order of magnitude less for the same benefit - and much quicker too.
Hope this answers your questions, Bert . . . I gotta run. . . .
OPB
