Page 1 of 1
| Driver "re-education", Japanese style 20/04/2015 at 09:32 #71258 | |
maxand
1637 posts |
Australian forum members might be interested in a documentary of the 2005 Amagasaki rail crash disaster, part of the "Seconds from disaster" TV series currently showing on ABC television. If you missed the broadcast you can watch it again here until 8:45pm on 3 May 2015. Quote: Drivers face financial penalties for lateness as well as being forced into harsh and humiliating retraining program, known as "Nikkin Kyoiku" ("dayshift education," including weeding and grass-cutting duties during the day)[4] retraining programs.[5] The final report officially concluded that the retraining system was the most probable cause of incident.[6] This program consisted of violent verbal abuse, forcing the employees to repent by writing extensive reports. Also, during these times, drivers were forced to perform minor tasks, particularly involving cleaning, instead of their normal jobs. Many saw the process of "Nikkin Kyoiku" as a punishment and psychological torture, and not as driver retrainingAccording to the program, JR West (West Japan Railway Company) still practice this type of re-education and claim that they are as efficient as their competitors, although as a result of the inquiry they intend to improve their methods somewhat. Their timetables are very strict, down to the second. Quote: The Japan Railway company is very strict when it comes to punctuality,[3] and commuters often depend on near-perfect timing on the part of trains to commute to and from work on time. This is because at stations (including the derailed train's next scheduled stop at Amagasaki Station) trains meet on both sides of the same platform to allow people to transfer between rapid and local trains running on the same line.It appears passengers are allowed only a few seconds to move from one train to another. This is the other face of the much-vaunted Japanese "efficiency". I don't think it would work too well in either the UK or, God forbid, Melbourne. Somehow I can't see Australian drivers being made to scrub bird s**t off active rail lines. Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 10:01 #71259 | |
jc92
3631 posts |
Also available on Youtube.
"We don't stop camborne wednesdays" Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 11:59 #71261 | |
|
Jersey_Mike
250 posts |
That episode really shows how "making up time" has become a lost art in the west. Go back to the 1800's and all the best locomotive engineers had an uncanny ability to make up for delays even on the order of 30 or 60 minutes over a long enough distance. A lot of this had to do with just how variable a steam locomotive's performance was compared to the push-button diesels and electrics used today, but also because railroads generally trusted their employees not to derail the train, even if it meant bending the rules. Some railroads only had lists of speed "restrictions" and any section of track not restricted allowed for the driver to attain whatever speed they could. The New York Central's Niagara routinely exceeded 110 or 120mph on the Water Level Route and the Pennsylvania's GG1's often ran with their speedometers pegged at 100 on a line officially rated for 80mph. Out west early passenger diesels were often geared for 110 or 120 and many city-to-city speed records were set that still stand to this day. Unfortunately, as in Japan, these practices would occasionally lead to accidents and over time oversight became more strict and the trains were slowed down. Some of this was probably justified, especially as level crossings became more hazardous. Of course past a certain point we have to ask ourselves if the economic consequences of such risk aversion is really worth the marginal lives saved. Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 13:10 #71262 | |
|
maxand
1637 posts |
Over longer distances it might have been easier to make up time, as Jersey_Mike pointed out, but this became more difficult as more and more stations appeared enroute and the distances between them diminished. Dead man's controls to limit runaways were introduced into electric streetcars in the 1920s, but soon found their way into locomotive drivers' cabins as single-person operation became widespread with diesel and electric power replacing steam. Before that, a fireman usually knew how to slow or stop a train if the driver became incapacitated. The type of scenario causing the Amagasaki rail crash should become much less common with the advent of driverless trains. I forgot to mention in my original post that one of the few good things that came out of this disaster was that Japanese train operators were asked to install automatic train stopping systems ahead of curves to prevent overspeeding. See the Wikipedia link for greater detail. Last edited: 20/04/2015 at 13:12 by maxand Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 16:55 #71265 | |
GeoffM
6292 posts |
Without some active reinforcement (ATP/PTC etc), driving at virtually the limits is always going to end unhappily. The south London problem of drivers constantly running on yellows is often cited as a big risk and it is perhaps surprising there haven't been more problems. I recall going to Cleveland airport in the US by train. One could sit at the front and watch the driver/engineer/operator. This particular operator's driving method was full throttle until the overspeed kicked in, ease off, full throttle, overspeed, ease off, full throttle... :doh SimSig Boss Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 17:23 #71267 | |
|
Jersey_Mike
250 posts |
" said:Without some active reinforcement (ATP/PTC etc), driving at virtually the limits is always going to end unhappily. The south London problem of drivers constantly running on yellows is often cited as a big risk and it is perhaps surprising there haven't been more problems.The problem with ATP type systems is that they often stack two or more margins of safety on top of each other. The track speed is about 20% lower than the overturn speed and then the ATP enforcement has another 10% or 20% margin of error on top of that. One ends up leaving a lot of performance on the table. Interestingly enough I think that working "on sight" is poised to make a comeback, but instead of the driver's vision it would be a computer vision system that doesn't have to use beacons or transmissions to position itself. If we have cars that can drive autonomously in real life traffic the rail use case is already solved because you don't even have a "steering" variable to contend with. " said: That's often more efficient than having the operator try to use more gentle acceleration. Well designed propulsion packages are set up for the operator to just peg the throttle with the electronics providing all the finesse. I can't speak to constantly hitting up against the overspeed, but a rapid transit line in Philadelphia placed signs on the wayside informing drivers when to brake, coast and apply power. BTW the Cleveland line to the airport is classified as a heavy rail rapid transit system and is not part of the national railroad network. I am unsure what safety features are fitted. Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 17:59 #71268 | |
|
Muzer
718 posts |
You know, I suspect a hybrid system might not be too bad an idea. I don't know how realistic it would be, but something like a system to be controlled by the (in-cab) signalling up to the end of the final block before an occupied section, then automatically proceed slowly "on sight" up to the train in front. Obviously would require ATO. And obviously this is for plain line, and most likely suitable for things like the London Underground more than mixed traffic railway. Log in to reply |
| Driver "re-education", Japanese style 20/04/2015 at 20:27 #71277 | |
|
GeoffM
6292 posts |
" said:The problem with ATP type systems is that they often stack two or more margins of safety on top of each other. The track speed is about 20% lower than the overturn speed and then the ATP enforcement has another 10% or 20% margin of error on top of that. One ends up leaving a lot of performance on the table.What's your source for this? Certainly nothing real life that I've ever worked on. Can't remember which automated railway it is but the managers prefer the trains to be driven themselves, rather than by humans, because the computer works to the curve; the humans drive before the curve, more cautiously. The computer is better at timekeeping. " said: Interestingly enough I think that working "on sight" is poised to make a comeback, but instead of the driver's vision it would be a computer vision system that doesn't have to use beacons or transmissions to position itself. If we have cars that can drive autonomously in real life traffic the rail use case is already solved because you don't even have a "steering" variable to contend with.  There are a couple of fundamental problems with your thoughts. Firstly, trains need much more distance to stop than cars: one cannot "see" the train in front visually. Secondly, trains DO need to, um, "steer" as they don't all follow each other on exactly the same route: how else would a train know to stop in front of a set of points that have yet to be set in the correct direction?" said: " said:This is not about using full power to accelerate but constantly driving in such a way that you get a sawtooth speed graph, with passengers repeatedly being thrown forward by the emergency brakes, then back as it accelerates again, then forward as it brakes again, then back...That's often more efficient than having the operator try to use more gentle acceleration. Well designed propulsion packages are set up for the operator to just peg the throttle with the electronics providing all the finesse. I can't speak to constantly hitting up against the overspeed, but a rapid transit line in Philadelphia placed signs on the wayside informing drivers when to brake, coast and apply power. SimSig Boss Log in to reply |
| Driver "re-education", Japanese style 21/04/2015 at 08:13 #71291 | |
|
kbarber
1712 posts |
" said:
If we ignore the idea that the computer doesn't need beacons/transmissions, I think Mike just invented moving block. You know, that system the DisneyLand Railway and the UndergrounD have been busy implementing over the last decade or so. Log in to reply |
| Driver "re-education", Japanese style 21/04/2015 at 12:31 #71308 | |
|
Jersey_Mike
250 posts |
" said:The computer works to the speeds that were designed for human operation. They are certainly more efficient at achieving those speeds, but if you compare an ATO system to one where a human operator has no speed enforcement the human operator will do better than the computer because the human has a greater degree of leeway to exceed the speed limits. If you look at the actual overturn speeds they are significantly greater than the posted line speeds because it is not safe for human operators to operate the trains on the edge of disaster. However an automated system has better reaction time and therefore would need less of a safety margin. Second, many train speed restrictions are in place for passenger comfort or to prevent wear and tear. Both are important, but again neither is safety critical. Finally human operators are a better judge of conditions, where automated system often have to make worse case assumptions about train performance. Looming over all of this is liability. If an employee messes up its simple human error, but if an automated train is going 2mph over the track speed and derails for a reason completely unrelated to speed the signal vendor etc can still be liable. Modern train protection systems apply a safety margin to comply with speed restrictions that already factor in a safety margin. Especially when a human is still operating the train, the "red line" speeds should probably be 5-10% higher than the official rulebook speeds so that the driver can hit the maximum performance without risk of speed penalties. " said: You use the same traditional signaling system. The computer vision system would be able to recognize wayside (or cab) signals and signs and comply with them (or provide an enforcement mechanism). No need for all sorts of expensive signaling kit. You can even add in whatever code SimSig uses to detect a misroute to do the same and send a text message the signaler. Quote: If we ignore the idea that the computer doesn't need beacons/transmissions, I think Mike just invented moving block. You know, that system the DisneyLand Railway and the UndergrounD have been busy implementing over the last decade or so.I've mentioned before that the Boston Green Line trolley system operates at a frequency too high for CBTC systems because all current CBTC systems have significant error bars around where the train is. Because of system latency and the chance of wireless messages being dropped a trains location is always +-some significant number of feet. Only something like adaptive cruise control will allow a following train to pull up to the buffers of the leading train. Technology that allows for autonomous road vehicle operation can be used to automate (or supervise) rail vehicles without any signaling modifications, full stop. A box in the cab can detect and respond to wayside signals, follow speed restrictions, stop at stations and apply the brakes if something is on the track. Last edited: 21/04/2015 at 12:35 by Jersey_Mike Log in to reply |
| Driver "re-education", Japanese style 21/04/2015 at 15:25 #71320 | |
|
GeoffM
6292 posts |
Sigh. There are so many problems with your post, Mike, that I don't even know where to start. What gets me is you seem to be ignoring real world evidence that directly contradicts many of your statements. :doh
SimSig Boss Log in to reply |
| Driver "re-education", Japanese style 21/04/2015 at 15:56 #71325 | |
|
Jersey_Mike
250 posts |
" said:Sigh. There are so many problems with your post, Mike, that I don't even know where to start. What gets me is you seem to be ignoring real world evidence that directly contradicts many of your statements. :dohWhy don't you try to start. The whole theme of this post is the propensity of human operators to exceed operational speed limits in order to make up time. The practice of making up time was possible because of a gap between what was laid out in policy and what was possible in practice. An ATP system that enforces those existing policy limits removes any ability for that gap to be closed. The Federal Railroad Administration actually said that the ATP mandate in the United States will reduce capacity over non-enforced human operation. Regarding computer vision I am inspired by the current problems with the ACSES train control system currently being installed by Amtrak. Because of limitations of the system the positive stop enforcement stops trains up to 1000 feet away from the signal. At stations with a stop signal at the end of the platform the rulebook states that trains will not be able to platform properly unless a proceed indication is displayed. "Traditional" methods of allowing trains to locate their position are both costly and time consuming as trains slow and creep like an elevator aligning with a floor. However if autonomous road vehicles can operate in traffic based on what it "sees", then a cost effective SPAD prevention becomes possible through on board identification of the stop point and continuous calculation of the braking curve. It could also enforce safe permissive operation. Last edited: 21/04/2015 at 15:58 by Jersey_Mike Log in to reply |
| Driver "re-education", Japanese style 21/04/2015 at 16:16 #71326 | |
|
GeoffM
6292 posts |
" said:" said:I don't have all day.Sigh. There are so many problems with your post, Mike, that I don't even know where to start. What gets me is you seem to be ignoring real world evidence that directly contradicts many of your statements. :dohWhy don't you try to start. " said: The whole theme of this post is the propensity of human operators to exceed operational speed limits in order to make up time. The practice of making up time was possible because of a gap between what was laid out in policy and what was possible in practice. An ATP system that enforces those existing policy limits removes any ability for that gap to be closed. The Federal Railroad Administration actually said that the ATP mandate in the United States will reduce capacity over non-enforced human operation.What the FRA thinks might happen is different from what actually happens in the real world. Remember that the FRA is not exactly renowned for its forward thinking, or its independence. Back to the real world: why don't you watch some YouTube videos of ATO in action: trains entering a station platform at 25mph+ and stopping at the far end on a sixpence (many have to: platform edge doors have to align with train doors), with just the last few feet at a slow alignment speed. " said: Regarding computer vision I am inspired by the current problems with the ACSES train control system currently being installed by Amtrak. Because of limitations of the system the positive stop enforcement stops trains up to 1000 feet away from the signal. At stations with a stop signal at the end of the platform the rulebook states that trains will not be able to platform properly unless a proceed indication is displayed."Inspired" is not a word I would use. You do know that ACSES is not new ("currently being installed")? It's being upgraded to support PTC but it's not new. One crappy implementation doesn't define the rest of the world, you know. " said: "Traditional" methods of allowing trains to locate their position are both costly and time consuming as trains slow and creep like an elevator aligning with a floor. However if autonomous road vehicles can operate in traffic based on what it "sees", then a cost effective SPAD prevention becomes possible through on board identification of the stop point and continuous calculation of the braking curve. It could also enforce safe permissive operation.Again, I recommend you actually visit the Real World (TM) where your problems mysteriously don't seem to be an issue. Finally, despite being repeatedly asked, you've not provided any sources. Not Yahoo Answers perchance? SimSig Boss Last edited: 21/04/2015 at 16:17 by GeoffM Log in to reply |