Page 1 of 1
| Realistic failure rates 11/12/2013 at 08:29 #52651 | |
Hawk777
386 posts |
Hi! I was wondering which failure rate setting in SimSig corresponds most closely to real life. I don’t live in the UK, so I can’t really rely on personal experience to say how often various pieces of Network Rail infrastructure fail. Based on experience from where I live, it seems like even the “Low” failure rate in SimSig is unrealistically high, with a fair number of failures in a day—compare this to my own experience (only as a customer, but one who gets alerts when trains are delayed significantly) on a local light rail system on which report-worthy delays tend to be measured in the two or three per month range, if that. Of course, this is a much newer system than Network Rail, and also probably much smaller than most SimSig simulations (even if not smaller in geographical area, at least smaller in terms of number of points, etc.), but still, the failure rates seem so very different. It’s hard for me to believe Network Rail realistically gets a dozen or more pieces of important hardware failing on a single day within a single IECC’s area of coverage—or maybe my local controllers are just very good at hiding failures from customers (a good thing)! I realize this might be something NR employees can’t talk about, but maybe people who just live in the UK could comment on this from their experience using the network? Log in to reply |
| Realistic failure rates 11/12/2013 at 10:41 #52660 | |
|
kbarber
1712 posts |
I can certainly comment as an ex-signalman. I'd agree the Simsig failure rates are considerably higher than I'd expect in real life. I was a signalman at Marylebone for about 12 months in 1980/81 (in those days a mechanically framed box, manually worked points with just one motorised exception and a mix of colour light and semaphore signals). We worked Track Circuit Block (i.e. automatic signals worked by track occupation rather than the manual Absolute Block system that is now increasingly rare); this had been commissioned as far back as 1923, complete with the first colour light signals for daylight use on a main line, all of which were still in use. I think I'd have regarded it as a pretty bad run if I'd had 2 or 3 failures in a month. In fact we'd have called it a bad run if there had been that many spread around all the 3 shifts. Admittedly motor points introduce a potent source of failure. So did the early time-division multiplexing systems by which signalbox control areas were extended beyond the limits where direct-wiring was financially viable. When I was a Station Manager my area of responsibility included Hackney Downs, which was an NX panel and included such a TDM to control Clapton Junction. The Clapton 'Westronic' (nicknamed 'Catastronic' by the installers who built it in 1960, I later found out) was truly notorious for its poor reliability (and the junction points were not considered the best either) but I don't suppose I had to go out there more than 4 or 5 times in the whole of my 2-year sojourn in that job. Log in to reply |
| Realistic failure rates 11/12/2013 at 11:38 #52663 | |
|
MrBitsy
121 posts |
Busy boxes can experience several problems in a day on a regular basis, so Simsig on a low setting can be realistic.
TVSC Link 4 signaller - Temple Meads, Bath & Stoke Gifford Last edited: 11/12/2013 at 11:40 by MrBitsy Log in to reply |
| Realistic failure rates 11/12/2013 at 13:25 #52665 | |
|
maxand
1637 posts |
I'm still not clear as to whether signal failures are classed as track circuit failures, in real life or in SimSig, and whether signallers are notified as to the difference, also in real life or in SimSig. I realize SimSig is a simplification in many ways, but it might make the simulation more interesting if the lists of causes of delays and equipment failures could be expanded a bit. A couple of my recent posts (1,2) about disruptions to service in Melbourne Metro reveal an amazing variety of causes, such as: Quote: A train driver made a mistake and picked up passengers on the Sandringham line when he should have been running the train empty.. Passengers left the train at North Brighton. Also: Quote: Did You KnowAll this and more at http://www.melbournetrains.net/. Developers please note. This has become my second most favourite railways site after SimSig. With events such as these, who needs TC failures anyway? 
Last edited: 11/12/2013 at 13:38 by maxand Log in to reply |
| Realistic failure rates 11/12/2013 at 13:37 #52666 | |
|
clive
2738 posts |
" said:Hi! I was wondering which failure rate setting in SimSig corresponds most closely to real life.The settings aren't the same between simulations; they're part of the simulation data. Picking Fenchurch Street, because I have the data to hand, the "Low" rates are: - points fail once in every thousand movements; - signals fail once in every 2000 aspect changes; - track circuits fail once in every 1000 clears plus once every 116 days. (These are average rates; the actual failures are random.) The "Medium" rates are 400/1000/400/46 and the "High" rates 200/500/200/23. Log in to reply |
| Realistic failure rates 11/12/2013 at 14:15 #52668 | |
|
mfcooper
707 posts |
Sometimes we can go days without a failure across the 5 panels at Victoria (Central), and some days we have lots in a couple of hours. It is completey random, and the settings on SimSig allow the user to tailor their own experience. " said: I'm still not clear as to whether signal failures are classed as track circuit failures, in real life or in SimSig, and whether signallers are notified as to the difference, also in real life or in SimSig.(I'm sure this has been spoken about before.) The following are all "In real life, in the UK". "Signalling Failure/Problem" is the phrase used to tell the general public that there is a problem with the signalling equipment. This equipment, of course, includes signals, but also includes other stuff like points & track circuits. However, It can also refer to buttons on a control panel, the computer having a hissy fit, or even the signaller on duty making an incorrect regulating decision, causing delay to one or many trains. The wording to the public is very general as very few people know what track circuits (and points!) are. Signallers have no absolute way of telling if a track circuit showing occupied is a train or a failure. Through regular monitoring, one can tell when a train has left a track circuit occupied after the train has passed (and the track circuits ahead of the failure start clearing again), and occasionally a track circuit will occupy randomly where there is no train nearby, and this would cause a No Description Alarm (NDA), which is usually '***' or '*X**' in the train describer. Sometimes this "random" occupation is a metallic item completing the track circuit, such as a shopping trolley dumped on the line, but it can also be a "signalling failure", where a cable has been damaged. Because there is no absolute way to tell if the occupied track is a train of not, the first train that travels through an assumed track circuit failure must be told to examine the line in case an obstacle is encountered. If it is an object, like a shopping trolley, then the track circuit is operating correctly and there is no failure. If no obstruction is found, then there is a failure somewhere, be it wiring or some technological piece of kit, which technicians will have to go and investigate and hopefully fix sooner rather than later. NB: It is increasingly common to now have Axel Counter track circuits, where a treadle counts train axels into and out of a section of line, which occupies and clears the track circuits on the signaller's display. So if a counting unit goes wrong, the TC will remain occupied. And this type of TC cannot detect the presence of a conductive object on the line. Log in to reply The following user said thank you: maxand |
| Realistic failure rates 11/12/2013 at 17:48 #52673 | |
|
TomOF
452 posts |
" said:Thats an interesting link, could get a bit addictive. Drifting off topic slightly At the Newcastle meet last year many photos where taken of the panel with a view of eventually doing a Tyne And Wear Metro sim. A Melbourne metro sim would be nice to do one day if enough info could be gathered.... Log in to reply |
| Realistic failure rates 12/12/2013 at 07:40 #52690 | |
|
Danny252
1461 posts |
" said:Sometimes this "random" occupation is a metallic item completing the track circuit...My favourite failure cause is one where a track circuit immediately in front of the box would immediately occupy upon a specific signal being cleared, and then show clear as soon as the signal was put back. After a bit of investigation (mostly involving several people having a go at pulling the lever and watching the TC light every time), it turned out that when the wire for this signal passed under the track in question on its way out of the box, and when the lever was pulled, it became taut and came into contact with the undersides of the rails, shorting out the track circuit; having worked this out! The solution in the end was to put an insulator on the underside of one of the rails, which can still be spotted by observant passers-by... Last edited: 12/12/2013 at 07:40 by Danny252 Log in to reply |
| Realistic failure rates 12/12/2013 at 08:05 #52691 | |
|
Hawk777
386 posts |
Great, thanks for all the info—sounds like low is probably the most reasonable choice.
Log in to reply |
| Realistic failure rates 12/12/2013 at 12:22 #52693 | |
|
maxand
1637 posts |
Thanks MFCooper for shedding light on this. Quote: "Signalling Failure/Problem" is the phrase used to tell the general public that there is a problem with the signalling equipment.But from the signaller's point of view, can he/she distinguish between a failed signal, e.g., stuck at red - maybe by vandals who interfere with them hoping that the train will stop long enough for them to cover it with graffiti (see an earlier post of mine) and a failed track circuit for such reasons as you and Danny252 mention? Or is the only indication a "failed" TC (red segment)? Another point of interest is a bug at well-documented locations in some sims where a train proceeds after a route is set for it, but loses its description, becoming a No Description train ("****"  to which the signaller needs to assign a description before it leaves his area. As this routinely occurs at the same location(s), one assumes it is a bug and that it will eventually be fixed. Your comments imply that this can happen randomly due to a number of causes. Is this something that could/should be included in SimSig for added realism?
Last edited: 12/12/2013 at 12:28 by maxand Log in to reply |
| Realistic failure rates 12/12/2013 at 12:36 #52694 | |
pedroathome
887 posts |
" said:I believe that in some places (please correct me if I am wrong here, just taking from what I have previously read somewhere on here) but places such as Brighton Station, a TD does not in real life step from the buffer stop end of a platform onto the Hove, Preston Park or Lewes lines as the train passes. In my opinion though, playing with delays turned on is much more realistic James Log in to reply |
| Realistic failure rates 12/12/2013 at 14:42 #52695 | |
|
Danny252
1461 posts |
" said:Thanks MFCooper for shedding light on this.A failed signal doesn't show as a failed track circuit is the simple answer. Surely you can see this yourself from Simsig! Quote: Another point of interest is a bug at well-documented locations in some sims where a train proceeds after a route is set for it, but loses its description, becoming a No Description train ("****"I'm sure we've covered this with you on umpteen occasions, but non-stepping TDs are not a bug, nor do I recall any specific case where they have been identified as such. All that he stated, regarding "****" TDs, is that they are what are displayed if a track circuit becomes occupied when there is not a valid previous TD to step from - either because that route does not have TD stepping facilities provided (such things were expensive when they had to be implemented as relay logic), that the train is moving without a route being set (and hence the previous TD cannot be determined by the interlocking system), or because the TC has failed. Log in to reply |
| Realistic failure rates 12/12/2013 at 16:28 #52696 | |
GeoffM
6282 posts |
" said:Another point of interest is a bug at well-documented locations in some sims where a train proceeds after a route is set for it, but loses its description, becoming a No Description train ("****"Obviously it wouldn't be a widescale bug! I believe most of the sim manuals tell you how to use the TD berths at a terminus - here is Brighton, for example: Brighton manual. Yes I did just edit it to clarify a little more though the previous wording was still correct. In simple terms, always interpose an outbound working into the signal end berth, and all inbound trains step to the buffer end berth. SimSig Boss Log in to reply The following user said thank you: maxand |
| Realistic failure rates 12/12/2013 at 18:00 #52697 | |
Steamer
3922 posts 
|
" said:But from the signaller's point of view, can he/she distinguish between a failed signal, e.g., stuck at red - maybe by vandals who interfere with them hoping that the train will stop long enough for them to cover it with graffiti (see an earlier post of mine) and a failed track circuit for such reasons as you and Danny252 mention? Or is the only indication a "failed" TC (red segment)?A failed signal (i.e. a signal displaying no light) is shown, as in SimSig, by a hollow grey circle in place of a filled colour circle, and triggers an alarm. A track circuit will show as occupied when the current to the electromagnet in the relay is interrupted. This normally occurs when a train's wheels short-circuit the relay, however any conducting material placed across the rails will do the job. Wikipedia diagrams here and here. Note that in real life the output from the relay won't lead directly to a signal, instead it goes in to the Interlocking system. The vandals point is interesting. If the vandals operate a Signal Post Replacement Switch (UK, I don't know what the Australian system has), I'd suspect the signal would be displayed as a red signal in the usual way. I don't know if the operation of the switch would set an alarm off. The behaviour of the signalling system would depend on where the vandals interfered with the electronics. "Don't stress/ relax/ let life roll off your backs./ Except for death and paying taxes/ everything in life.../ is only for now." (Avenue Q) Last edited: 12/12/2013 at 18:03 by Steamer Log in to reply The following user said thank you: maxand |
| Realistic failure rates 12/12/2013 at 18:08 #52698 | |
|
Forest Pines
525 posts |
" said:A failed signal might appear to a signaller as an unlit indicator (like in SimSig) or as an indicator showing a different indication to what the signaller expects. This is a completely different to a failed track circuit, where the failed circuit appears to the signaller to be occupied. In the case of semaphore signals, the indicator ("repeater" will show "on", "off", or "wrong", which means either that the signal is not properly in one position or the other, or that the equipment which detects the signal's position has failed. With non-mechanical colour light signals, things are slightly simpler; circuitry is provided to check that the relevant lamps are drawing current as expected, and if they're not the signal indication will show it to be unlit (and the previous signal will revert to red).In the real world a colour light signal could "partially fail" but remain lit - for example if a signal has a route indicator and the route indicator does not light properly the main aspect will be held at red. I'm not sure how this is reported to the signaller, though. Log in to reply The following user said thank you: maxand |
| Realistic failure rates 12/12/2013 at 18:14 #52699 | |
|
Forest Pines
525 posts |
" said:I could be wrong but I seem to recall that the Great Train Robbers stopped their target train by simply covering up a signal's green aspect and showing a red light to the driver by putting a battery across the lmap terminals. Presumably in that case the signalman remained unaware that anything nefarious was happening. Log in to reply |
| Realistic failure rates 13/12/2013 at 11:17 #52712 | |
|
kbarber
1712 posts |
I think, yet again, what the bobby sees will depend quite a lot on the age of the installation (or of the most recent modification to it). In old fashioned mechanical working, there might not be any TCs to fail (I can think of one such box, still active in the early 1980s, and I'm sure there were others). But as others have said, the indication the signalman sees will be identical whether there's a train on the TC, or there's something shunting it (a shopping trolley would be more visible than a length of thin scrap metal strip - a regular problem when such stuff was loaded at Upton Park and a bit dropped off as the train came out of the sidings!) or there's a broken rail or a broken bond at a rail joint or something wrong with the gubbins of the TC itself. Some earlier mechanical installations had fewer but longer TCs than we've come to expect in modern practice. So it would be possible to find that a TC extended from somewhere in rear right through a trailing crossover and up to the starting signal, with room between the crossover and the starter to stand a train before it set back through the Xover. In that case neither the crossover points nor the shunt signals reading back through them would be locked by a TC showing occupied. An example can be seen here http://www.signalbox.org/diagrams.php?id=152 - neither crossover 23 nor shunt signal 24 is locked by TC148. Correct operation of a signal was indicated with a Mk I eyeball, except for those which were out of sight of the box for which a repeater would be provided. That meant that a pretty wide range of angles could be accepted as 'OFF' for a signal visible from the box but there was an objective test - have the repeater contacts made up? - for those at a distance. In a lot of panel boxes (and indeed where auto sections were supervised from a mechanical box), automatic signals would not be repeated in the box (think 'panel signals' in the F3 menu). In that case, if a TC failed it would be obvious that the signal in rear of it would be at danger. But if any other part of the apparatus failed (and it wasn't only lamp filaments that failed, I have to say) there would be no indication at all in the box until a driver rang in from a red signal. Controlled colour light signals were usually indicated in the box but mostly only with 2 indications - danger or off. That was latterly the case even where the signal aspect was visible from the box - Marylebone 38 signal, for example; I don't know whether it might have been different in much earlier installations but I personally don't recall any. An exception to the 2-indication principle was the Southern Railway and the Southern Region, who often indicated all 3 or 4 aspects of a signal (as can be seen here http://www.wbsframe.mste.co.uk/public/Waterloo.html - scroll down for a close-up of part of the frame with 3-aspect indications behind the levers or here http://www.wbsframe.mste.co.uk/public/West_London_Junction.html where the 3rd picture down shows a good view of illuminated 4-aspect indicators.) In an older colour light installation, there are filament failure alarms for both auto and controlled signals. They vary according to the age, the manufacturer (and sometimes, I suspect, how the designer was feeling that day). It was most common to group several autos into one alarm indication. The signalman then had to watch the panel/diagram to see what signal was changing aspect as the alarm triggered (it was usually possible to work out what aspect it was, too, by observation over a period). However the 2nd yellow of a 4-aspect signal wasn't usually detected, so it was possible for a driver to see G followed by Y without anyone having realised anything was wrong. At Marylebone, the auto signals had no filament failure alarms at all - the only way you knew a lamp had blown was when a driver rang in from a black signal! (They didn't have a secondary filament either, nor any way of keeping the signal in rear at danger. Oh, how innocent they were when those signals were installed back in 1923...) In the UK, signal post replacement switches are worked by a key so a vandal would need access to such (and the knowledge to use it). I think it's unlikely a graffiti artist would have either. The great train robbers certainly had the knowledge and, as I read it, were indeed able to prevent the signalman receiving any failure indication (they covered up the properly-lit green lamp with a coat as well as putting their battery across the red, I believe). Descriptions are a whole other can of worms. I think it's important to recognise that train describers are not - as far as I know, at any rate - safety-critical apparatus. Therefore you aren't allowed to rely on a train describer for information that might affect the safety of the line. They do fail, in all sorts of ways (again dependent, I suspect, on the age of the installation). There aren't the detailed alarm indications - although there's usually some kind of 'not described' alarm somewhere; again that will vary, sometimes only operating for a description not sent to the next box but in other cases for any berth (or sometimes, perhaps, for any berth where no route was set into it, so it also implicitly acts as a SPAD alarm - I have an idea London Bridge behaves like this). The not-safety-critical point was certainly relevant with the describers originally fitted between Liverpool St and Shenfield in 1949 - they specifically indicated which trains were electric, but were not permitted to be used for signalling non-electric trains into an isolation (the signalman was required to stop each train and ask the driver what the traction was). So as usual, you pays your money and you takes your choice. I imagine Simsig implements current practice rather than trying to simulate the actual practice of the time each installation was built. Log in to reply The following users said thank you: maxand, Forest Pines |
| Realistic failure rates 13/12/2013 at 12:36 #52713 | |
|
maxand
1637 posts |
Thanks for all your contributions, particularly Keith's. (Steamer) Quote: A failed signal (i.e. a signal displaying no light) is shown, as in SimSig, by a hollow grey circle in place of a filled colour circle, and triggers an alarm.I must say I haven't seen one before, hence my confusion. Will look out for this from now on. (GeoffM) Quote: In simple terms, always interpose an outbound working into the signal end berth, and all inbound trains step to the buffer end berth.Nice tip, thanks Geoff. Log in to reply |
| Realistic failure rates 13/12/2013 at 12:54 #52714 | |
|
Forest Pines
525 posts |
Regarding hard-to-fix TC failures, I recall reading on uk.railway back in the mists of time about a location somewhere in England (Cheshire maybe) where rock salt in open wagons was a regular freight traffic flow. If it was a rainy day, each train would leave every track circuit failed behind it, as the rainwater running off the wagons made the ground conductive enough for the track relay to drop. Not very much you could do to clear a failure like that I imagine!
Log in to reply |
| Realistic failure rates 13/12/2013 at 22:07 #52731 | |
|
splash
102 posts |
That is true. Trains used to leave Hartford up the WCML and the salt got into the track circuits and cause failures all along the WCML. It took them along time to fix. If you followed one off the salt trains you would expect to get delayed because of the failures
Log in to reply |
| Realistic failure rates 27/12/2013 at 19:07 #53291 | |
|
clive
2738 posts |
" said:It is usual that a description doesn't step beyond a controlled signal if no route is set from it. At least one Railway Inspectorate accident report has used this fact to determine which train in a collision SPADed. Log in to reply |
| Realistic failure rates 27/12/2013 at 19:59 #53295 | |
KymriskaDraken
963 posts |
" said:I remember one of the Senior S&T guys, Peter Day, telling me that when I was working Bristol Panel and was involved in a freight train SPADing at Sodding Chipbury. He said in order for the TD to step the signal has to be off and the track circuit immediately in advance of the signal has to be occupied. The slight lag between the track occupying and the signal going back to danger is long enough for the TD to step forward. Last edited: 27/12/2013 at 19:59 by KymriskaDraken Log in to reply |
| Realistic failure rates 27/12/2013 at 20:09 #53296 | |
|
KymriskaDraken
963 posts |
" said:" said:I had a similar thing at Little Mill Jn. The techs cured it by using a high-technology piece of orange drainpipe to run the signal wire under the track.Sometimes this "random" occupation is a metallic item completing the track circuit...My favourite failure cause is one where a track circuit immediately in front of the box would immediately occupy upon a specific signal being cleared, and then show clear as soon as the signal was put back. After a bit of investigation (mostly involving several people having a go at pulling the lever and watching the TC light every time), it turned out that when the wire for this signal passed under the track in question on its way out of the box, and when the lever was pulled, it became taut and came into contact with the undersides of the rails, shorting out the track circuit; having worked this out! The solution in the end was to put an insulator on the underside of one of the rails, which can still be spotted by observant passers-by... Log in to reply |
| Realistic failure rates 27/12/2013 at 22:35 #53304 | |
|
clive
2738 posts |
" said:That's usually the case for automatic signals. For controlled signals it's usually triggered by overlap track circuit with route set (with the destination berth determined by the route). Log in to reply |
