Page 1 of 1
| is ARS getting smarter? 05/07/2013 at 18:38 #46549 | |
jc92
3685 posts |
while working marylebone, and the station in particular, I have 5R11 which departs at 06:37 from P5, and 2H02 arriving at 06:40 (booked) into P4. this presents a clash in the station throat. I was expecting ARS to hold 2H02 at ME10 and let 5R11 out first, as the tt dictates. as 2H02 is running with plenty of recovery, and i had a 2 min window to sneak it in before 5R11 departed, I went to set the route, only to find ARS doing it for me! is this prototypical behaviour? ie can ARS predict and exploit gaps, or is it in fact being stupid and its just fortunate coincidence? Joe "We don't stop camborne wednesdays" Log in to reply |
| is ARS getting smarter? 05/07/2013 at 18:47 #46552 | |
GeoffM
6376 posts 
|
" said:while working marylebone, and the station in particular, I have 5R11 which departs at 06:37 from P5, and 2H02 arriving at 06:40 (booked) into P4. this presents a clash in the station throat.That's not a clash, that's just tight margins. Need to check the ROTP but I think the allowable margin is less. " said: I was expecting ARS to hold 2H02 at ME10 and let 5R11 out first, as the tt dictates. as 2H02 is running with plenty of recovery, and i had a 2 min window to sneak it in before 5R11 departed, I went to set the route, only to find ARS doing it for me!If it thinks it can get an early train across without delaying its rival then it will do so. But yes the ARS has steadily improved over the years, particularly recently. It's still not perfect and neither is the real thing. SimSig Boss Log in to reply The following user said thank you: jc92 |
| is ARS getting smarter? 05/07/2013 at 18:49 #46554 | |
|
jc92
3685 posts |
" said:" said:sorry - a clash in the sense of you cant physically set both routes at the same time, just to be clear that the routes conflict.while working marylebone, and the station in particular, I have 5R11 which departs at 06:37 from P5, and 2H02 arriving at 06:40 (booked) into P4. this presents a clash in the station throat.That's not a clash, that's just tight margins. Need to check the ROTP but I think the allowable margin is less. thanks for the prompt response I'll have to keep an eye out in future! "We don't stop camborne wednesdays" Log in to reply |
| is ARS getting smarter? 05/07/2013 at 18:51 #46555 | |
guyh
54 posts |
As it happens the Marylebone signallers always leave the ARS switched out for the up line, but leave it on for the down line. Therefore all arrivals have to be signalled in manually from ME12 to ME10 and ME10 to the platform, while all departures will usually route themselves out (TRTS will usually be done 2 mins prior but the ARS is restricted to only route 60 seconds prior to departure time). There are a number of reasons for not having the ARS on for arrivals:- 1) "Chiltern do too many swaps". 2) The ARS can't route into occupied platforms, and there is a lot of multiple occupancy at Marylebone. 3) The signallers work off Chiltern-produced simplifiers which are issued weekly, often with different versions for each day; there is little faith that Network Rail's planning team will have amended all the platforms in ITPS to match Chiltern's latest simplifier. 4) Having worked this way for years (I've been working for Chiltern since '95 and in this field since '98 - they haven't used it since then apart from one (successful) Sunday trial we did in 2011 or 2012(?)), the signallers don't have confidence that the ARS wouldn't mess up decisions in the station throat. Log in to reply The following users said thank you: GeoffM, jc92, Splodge |
| is ARS getting smarter? 05/07/2013 at 19:16 #46559 | |
|
guyh
54 posts |
Oh, and yes, the ARS in SimSig is not the same as the real one, but it's very very close now. The real one works in the following way (approximately): - I need to route a train in order to keep it on two greens; - I will check whether any other train in the area should be routed across this junction first; - For every other train in the area that hasn't yet gone across this junction, I'll work out (a) what the delay to both trains will be if this train goes first and (b) what the delay to both trains will be if the other train goes first. I'll simply add the delays together so I get a total for (a) and a total for (b). - If this is a converging move where the trains share a section of track going forward for several signal sections, I'll simply look at their booked times at the start of the shared section and the end of the shared section and work out the delay assuming that they can follow each other at minimum headway at each end (I'll be a bit dumb and not consider that their calling patterns might prevent this); - Having wasted milliseconds calculating this, I will always choose to route the trains in the booked order, UNLESS the delay caused by routing wrong order is considerably less than the delay routing in the booked order. In fact I have some programmed figures that no-one dare change because they don't really understand the likely effect! For a crossing move (eg Down Met conflicting with an Up Wycombe line at Neasden): if the delay for wrong order is more than 2 mins less than the delay for right order then I'll route them wrong order; if they are converging moves, then I'll need a bigger result - I'll need at least 5 mins better result for wrong order to be applied. There's some great bits about this:- 1) As it only looks at pairs of trains, you can have a sequence that should be train 1 up Wycombe, train 2 down Met, train 3 Up Wycombe. However train 1 is late as following train 3. ARS will look at train 2 approaching Neasden and say, ooh, I need to hold this for late running train 1. It will look at train 3 approaching Neasden and say, ooh, I need to hold this for train 2 that is due first. And it will continue to never route anything because it wants train 1 to be first but train 1 can't get there first! 2) For up trains approaching Neasden, it will consider their shared path to be from the junction to just past ME10, and might calculate that an on time / early up Met can go before a late running Up Wycombe (assuming a 5 minute saving), however it fails to notice that the Up Met won't be able to arrive in its platform im Marylebone early as another train will be there until just before departure time. So the trains will come up wrong order when right order would have allowed the late runner to arrive. 3) Plenty more that I can't think of now, but the main point is that the signallers are experts at this and are excellent at placing reminders, switching out ARS areas, setting trains to non-ARS etc to effect the best regulation. The freights which arrive from Greenford early are particularly good at letting themselves out onto the patch destructively early or late if ARS thinks it'll be ok (bearing in mind that the trains that might conflict between Aylesbury and Princes Risborough don't "exist" at that time). Log in to reply The following users said thank you: UKTrainMan, Jan, Temple Meads, John |
| is ARS getting smarter? 05/07/2013 at 20:29 #46572 | |
|
jc92
3685 posts |
thanks for the detailed response Guy, maybe we should have a "hints and tips from the real thing" section on the wiki? edit: Grrrr back to old tricks - 2G08 let routed out of south ruislip JUST before 5P08 enters at wembley and wants to pass it! learning the hard way! "We don't stop camborne wednesdays" Last edited: 05/07/2013 at 20:35 by jc92 Log in to reply |
| is ARS getting smarter? 06/07/2013 at 22:03 #46656 | |
Soton_Speed
285 posts |
" said:In fact I have some programmed figures that no-one dare change because they don't really understand the likely effect!How long has it been like this?!   Interesting comments from GeoffM here (theiet.org) - it's amazing what google dregs up! In Zone 6, no one can hear you scream... Log in to reply |