|
|
Post by metrider on Jun 28, 2017 19:19:23 GMT
Continuing from another thread about the new DMIs but concentrating on the recently installed DMIs at Harrow on the Hill P5/6 From the other thread… metrider: “The new DMIs at Harrow on the hill P5/6 also display confusing info sometimes. I was hypothesising that the DMIs were collaging information from a couple of different sources and sometimes displaying the same train twice (with slightly different times). I was thinking that in HoH's case the sources would be the Uxbridge branch, the other lines, and the signaller, or possibly other prediction software and the signaller. Where does the information that feeds these new DMIs come from? . . .”-and- DistrictSOM: “As its a manually controlled frame and the signaller may not be able to follow the timetable to the extent of the correct platform for what ever reason that may be the system for the TD is not able until the route is selected know what platform the train is coming in”.. This evening, I just missed a train and had a few minutes in which to play ‘guess which platform the next s/b train will depart from”. Today the sequence of the displayed info was confusing to a degree. There were no time warps involved (see the More Park thread), but I needed headache tablets afterwards. The DMIs displayed as follows: P5: [1. Aldgate 4 mins / Calling at ... / Aldgate service on Platform 6] and confusingly P6: [1. Aldgate 4 mins / Calling at ... / Aldgate service on Platform 5] Within a minute, this progressed to: P5: [1. Aldgate 2 mins / Calling at ... / Next Southbound Service in 4 mins] P6: [Aldgate service on Platform 5 / Next Aldgate Service in 4 mins / Next Baker Street Service in 6 mins] and then (about 2 mins later): P5: [1. Aldgate / Calling at ... / Aldgate service on Platform 6] and confusingly P6: [1. Aldgate / Calling at ... / Aldgate service on Platform 5] At which point a train arrived on P6 (to a green starter, so in this case no doubt that it wouldn’t go first), sat there for about a minute before departing. It was away before the advertised train arrived at P5. My trade is computer systems, with programming knowledge and am usually able to fairly accurately analyse anything along those lines that I encounter. In this case I find it very difficult to get inside the systems head. I know it’s just poor programming (or probably more accurately poor algorithms behind the programming) causing the odd display order, but I can almost imagine somebody with a warped sense of humour sitting behind the CCTV monitors watching people migrate between platform sides each time they change a setting  , Maybe they have competitions on how many times they can get away with doing that between successive departures (with bonus points for puzzled expressions), I don’t know!  Does anybody have better explanations, or ways to predict what will actually happen? I still think that a more sensible solution would have been a single board covering both platform edges with platform numbers added as the signaller sets the routes into the platforms.... metrider |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 28, 2017 19:22:29 GMT
I agree on your last statement
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2017 15:03:20 GMT
These data feed problems look set to continue until the phased the roll out period on 4LM as the signalling and train data feed systems on the Met are brought up to the same standard system. Currently stations between West Harrow and Uxbridge can only be describe trains by line not destination. Train description south of Moor Park is more comprehensive so this data jumble with Harrow's manual train describer and route clearance system feeding into modern DMI's is bound to cause issues. Migration Area 9 will automate tracks between Preston Road, Moor Park and West Harrow closing Harrow signal cabin, Migration Area 13 will close Rickmansworth cabin automating Moor Park to Watford, Amersham and Chesham (Met only) and Migration Area 14 will close Rayners Lane cabin automating the Uxbridge branch (Met only) as the last part of 4LM. The installation and upgrades of DMI's and their feeds will be progressively rolled out before many areas go live but pairing Harrow's DMI's with what is effectively 1940's analogue technology is a problem that won't and in some cases can't be remedied immediately.
|
|
cso
Posts: 1,043 
|
Post by cso on Jun 30, 2017 15:51:38 GMT
Migration Area 13 will close Rickmansworth cabin automating Moor Park to Watford, Amersham and Chesham (Met only) What happens for Chiltern trains then if the automation is for Met only? |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2017 16:15:16 GMT
Migration Area 13 will close Rickmansworth cabin automating Moor Park to Watford, Amersham and Chesham (Met only) What happens for Chiltern trains then if the automation is for Met only? Standard Network Rail four aspect signals will be in place on the fast lines between Harrow and Amersham for the Chiltern turbos. These same signals will be capable of displaying a blue aspect when Met trains use the fast lines in ATO. |
|
|
|
Post by metrider on Jun 30, 2017 19:07:52 GMT
Thanks Martin12, I thought that would be the case. Merging those disparate data sources is bound to be a nightmare. I can sort of think of some ways around that, but not worth doing if the new system is on the way. Any rough time scales on that that are not too top-secret to be hinted at?
If the current arrangement takes exactly the same information that the old system had then no wonder it does not know the order in which trains will run south of HoH.
That does beg another question. How will the new arrangement deal with situations like Harrow on the Hill, where running order of departing trains might not match the arrival order and/or where departure time may be affected by booked stands?
|
|
|
|
Post by crusty54 on Jun 30, 2017 19:30:56 GMT
The last time I was in the Harrow signal cabin everything was being worked by hand.
Real blast from the past.
|
|
North End
Beneath Newington Causeway
Posts: 1,769
|
Post by North End on Jun 30, 2017 20:19:12 GMT
Thanks Martin12, I thought that would be the case. Merging those disparate data sources is bound to be a nightmare. I can sort of think of some ways around that, but not worth doing if the new system is on the way. Any rough time scales on that that are not too top-secret to be hinted at? If the current arrangement takes exactly the same information that the old system had then no wonder it does not know the order in which trains will run south of HoH. That does beg another question. How will the new arrangement deal with situations like Harrow on the Hill, where running order of departing trains might not match the arrival order and/or where departure time may be affected by booked stands? The Seltrac system will have the timetable loaded into it, and knowing the logic the VCCs work to means it can make a fairly accurate prediction of what will happen. It can't second guess if the signaller is going to intervene however. Kennington southbound is a good example if this. In the evening it may well predict a train off the Bank branch will be the first to Morden and this is indeed what would happen if left alone. However it's often preferable to put a hold on platform 4 and let a late runner off the Charing Cross branch go first, this keeps the trains in order, and is more sensible as if there's going to be a gap you are better to have the emptier Charing Cross train run first as it will absorb more passengers at Stockwell compared to the crush-loaded Bank train. However the by-product of all this is the platform indicators will say the Bank train is first so everyone will pile onto that, and then get to watch the Charing Cross train pass by probably with a few spare seats to add insult to injury. Just one of those facts of life unfortunately. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2017 20:22:50 GMT
That does beg another question. How will the new arrangement deal with situations like Harrow on the Hill, where running order of departing trains might not match the arrival order and/or where departure time may be affected by booked stands? Platform alterations are of course inevitable in locations with flexible track layouts such as Harrow and Earls Court. The only difference will be that the system will pro-actively detect a deviation from the timetable and will be able to switch/alter information regarding platforms on DMI's much quicker than the current system as all the information is centrally controlled thus being easier to manage and coordinate. A more rudimentary version of this set up exists on the Westbound Platforms 1&2 at Acton with Piccadilly line services towards Uxbridge some times being routed onto Platform 1 which is usually the domain of District services to Ealing Broadway. The video below shows this move in action. As soon as Signal WL 25 gives the divergent route for P1, the board on P2 will flash up *Service Alteration* and should (if it behaves) say underneath "Rayners Lane service - Platform 1". Unfortunately the camera only focuses on P1 but it demonstrates how the system at Harrow should would work hopefully with a bit more advance warning as route clearance is no longer dependent on the location of physical signals and therefore should give a bit more of an advanced warning. Time lines for 4LM are mixed and varied depending on, this is just based on from what little I've heard over the past months (feel free to correct) SMA's 0.5 to 5 (Circle) by 2021 SMA's 6 to 9 (District east and Met trunk to Moor Park) by 2022 SMA's 10 to 14 (District west and Met branches) by 2023 |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2017 20:28:32 GMT
The last time I was in the Harrow signal cabin everything was being worked by hand. Real blast from the past. Quite so, thankfully the preservation of the lever frame (in 2009) was given more thought than the light-boxes.  |
|
Chris M
Global Moderator
Forum Quizmaster
Always happy to receive quiz ideas and pictures by email or PM
Posts: 19,758
|
Post by Chris M on Jun 30, 2017 21:26:50 GMT
What happens for Chiltern trains then if the automation is for Met only? Standard Network Rail four aspect signals will be in place on the fast lines between Harrow and Amersham for the Chiltern turbos. These same signals will be capable of displaying a blue aspect when Met trains use the fast lines in ATO. Presumably if the signalling gets confused and shows any aspect other than blue to a Met train in ATO, the driver would (?disable ATO and) react as they would to such a signal when driving manually. What happens though if a blue aspect is shown to a Chiltern train? Even if the system should not be capable of going wrong in this way, it's such an obvious failure mode to a human that is simple to deal with from an operating perspective* (all you need to do is write one or two procedures) that I'd be surprised if it hadn't been considered. *those dealing with it from a signal maintenance perspective however may not see it as such of course! |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2017 22:07:40 GMT
Standard Network Rail four aspect signals will be in place on the fast lines between Harrow and Amersham for the Chiltern turbos. These same signals will be capable of displaying a blue aspect when Met trains use the fast lines in ATO. Presumably if the signalling gets confused and shows any aspect other than blue to a Met train in ATO, the driver would (?disable ATO and) react as they would to such a signal when driving manually. What happens though if a blue aspect is shown to a Chiltern train? Even if the system should not be capable of going wrong in this way, it's such an obvious failure mode to a human that is simple to deal with from an operating perspective* (all you need to do is write one or two procedures) that I'd be surprised if it hadn't been considered. *those dealing with it from a signal maintenance perspective however may not see it as such of course! Interesting points raised, according to Thales a green aspect for an ATO Met train accounts to a systems failure/malfunction or loss of communications and would by default bring all trains in the surrounding area to a halt for absolute assurance of no conflicting movements. There are special contingency plans for working in manual on the fast lines in the event of a failure so your theory would seem plausible in an absolute worst case scenario. If a Chiltern encounters a blue aspect then it must attempt to stop short otherwise it will get a SPAD as the trips on ATO Met line trains will be cut out so blue aspect signals will have their train stops raised. In a worst case scenario the Chiltern could be given permission to proceed via an Incompatible Train Movement (ITM) similar to how Bakerloo 72's make their way to Acton via the Jubilee except during traffic hours during a hypothetical Met ATO failure although I believe it would be treated akin to permissive working. |
|
|
|
Post by MoreToJack on Jun 30, 2017 23:57:23 GMT
Presumably if the signalling gets confused and shows any aspect other than blue to a Met train in ATO, the driver would (?disable ATO and) react as they would to such a signal when driving manually. What happens though if a blue aspect is shown to a Chiltern train? Even if the system should not be capable of going wrong in this way, it's such an obvious failure mode to a human that is simple to deal with from an operating perspective* (all you need to do is write one or two procedures) that I'd be surprised if it hadn't been considered. *those dealing with it from a signal maintenance perspective however may not see it as such of course! Interesting points raised, according to Thales a green aspect for an ATO Met train accounts to a systems failure/malfunction or loss of communications and would by default bring all trains in the surrounding area to a halt for absolute assurance of no conflicting movements. There are special contingency plans for working in manual on the fast lines in the event of a failure so your theory would seem plausible in an absolute worst case scenario. If a Chiltern encounters a blue aspect then it must attempt to stop short otherwise it will get a SPAD as the trips on ATO Met line trains will be cut out so blue aspect signals will have their train stops raised. In a wort case scenario the Chiltern could be given permission to proceed via an Incompatible Train Movement (ITM) similar to how Bakerloo 72's make their way to Acton via the Jubilee except during traffic hours during a hypothetical Met ATO failure I believe it would be treated akin to permissive working. Not doubting you, but that goes against everything that I've been told and understand about CBTC... and also the basics of signalling. There is already precedence for blue (and white) aspects on the Victoria (and Central) line(s), and all indications thus far at that they will work in the same way. That is to say, a green aspect is the 'least restrictive' and is regarded as a proceed aspect to all trains, whether ATC fitted or otherwise. The associated train stop will be lowered and any train operating in tripcock mode can pass unimpeded. A red aspect is, as you'd expect, the most restrictive, and all trains - regardless of mode of operation - must stop (unless prior authority is given), and in the case of tripcock-fitted trains they will be tripped on the raised train stop. ATC fitted trains would be emergency braked. The blue aspect is obeyed by all trains, and means different things dependant on the mode of operation: tripcock-fitted trains and all trains not running in either automatic or protected manual modes must treat this aspect as though it were a red, and the associated train stop will be raised. If a SPAD occurs, the train will be stopped. However, trains operating in a mode where ATC codes are available (i.e. automatically or in protected manual, not restricted manual) can pass this aspect and must drive to the limit of movement set out on their in-cab displays. Essentially, on the most simplistic level, it boils down to: *Green: the signalling section and overlap ahead are clear of trains *Blue: the signalling section and overlap ahead are occupied, but there is a sufficient moving block for a train to proceed *Red: the signalling section and overlap ahead are occupied, and there is not a sufficient braking distance between the train in front Yellow aspects (I understand just single, all references thus far have been to LU 3-aspect signalling, not NR 4-aspect as you've mentioned) can also be used to increase throughput for non-ATC fitted trains. Having to emergency brake your Met line train, which has been diverted up the main instead of the local at short notice ahead of the next Chiltern, because the signal has just bobbed from green to blue is inherently not a good idea. |
|
|
|
Post by MoreToJack on Jun 30, 2017 23:59:46 GMT
Anyway folks: we have plenty of threads already dedicated to the 4LM resignalling, including many of these signalling-related discussions. I'll dig out a link in the morning, but if we can keep this thread for the Harrow DMIs that'd be great.
(And hopefully myself or another of the friendly passing mods will be able to merge this into the previous thread on Harrow DMIs, again where many of these answers were provided. Perhaps a dedicated DMI 'super thread' is made as the answers to the questions about the Picc, Uxbridge branch and more have also been answered previously...)
|
|
|
|
Post by metrider on Jul 1, 2017 7:41:59 GMT
Thanks all. I was aware of the other discussions on the deficiencies in the detail of the information available on the Met branches, and unless I''ve missed something the previous Harrow DMI threads were more to do with 'raising the flag' that they were being installed and their positioning rather than hard information as to any improvement in the short term quality of the predictions (prior to the signalling upgrade). I was interested if in the interim any additional information was being fed into the system to aid those (pre-upgrade) predictions, or any automatic intelligence was implemented to work around the shortcomings in the other systems. It seems not. however confirmation that the old shortcomings still apply without any enhancements is useful information in itself. I'll continue treating the predictions of new DMIs with skepticism, rolling the information displayed in with my own observations, knowledge of service patterns at the times that I travel, and what the signaller is likely to do when things are a little awry. There are some times when we can become tuned to blindly rely on low quality information. A non-railway parallel that comes to mind is a common sight in the clubhouse of the airfield I fly from. On a potentially nice day, the other pilots can be seen in the clubhouse head down glued to their laptops analysing various weather forecast details (of varying reliability) to decide when to fly. Someone had the bright idea of a lower tech solution - they attached an empty, clear, picture frame to the window facing west and labelled it "Weather Actual"...... I suppose I can apply the same realism to a different kind of MET forecast. [edited to add terrible pun to end of post. sorry 'bout that ] |
|
|
|
Post by tjw on Jul 1, 2017 14:40:39 GMT
I wonder what will come first, an accurate DMI that will quickly update as the situation changes... Or everybody using their mobile phone (3rd party apps) with train running information provided by a online signalling feed.
|
|
|
|
Post by metrider on Jul 1, 2017 15:31:14 GMT
The third party apps are hindered by the same inadequacies in the data as are the new DMIs, so I can't see them giving a better picture either, not until the data feed improves (when the signalling is upgraded), and then hopefully the new DMIs will start making mode sense.
There are third party apps around at the moment and they certainly have the same issues. In fact the one I use does not even attempt to predict HoH.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 1, 2017 15:40:07 GMT
I wonder what will come first, an accurate DMI that will quickly update as the situation changes... Or everybody using their mobile phone (3rd party apps) with train running information provided by a online signalling feed. Im putting my bet on the DMI's because even TFL's website for departures at Harrow (which uses the same API for 3rd party apps) is displaying different information to what is shown on the ground. Take Northbound Platforms 3 and 4, on the ground the DMI's usually display (in off peak hours) all Uxbridge services on 4 with Watford, Amersham and Chesham services on 3. However on the website it duplicates Northbound and Southbound departures on each platform. Third party apps echo this leading me to think that if data feeds improve it will be for LU trackernet systems before 3rd party API feeds. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jul 1, 2017 16:36:10 GMT
I wonder what will come first, an accurate DMI that will quickly update as the situation changes... Or everybody using their mobile phone (3rd party apps) with train running information provided by a online signalling feed. The DMI's believe it or not will be more accurate |
|
|
|
Post by metrider on Jul 1, 2017 17:09:08 GMT
Does trackernet just display the position of trains and the state of signals/points or does it stray into predictions of arrival times and (yet to be set) future routing?
I've been assuming that the granular position information from trackernet was being further analysed to predict arrival times at future points, and that was used to feed the APIs. Is the the case?
I had noted the 'sorting' of trains on the new P3/4 DMIs.So it appears that there IS room for a 'fiddle factor' to be computed and inserted between the raw data and the DMIs. For the northbound services that would be be very simple. The fiddle factor would simply be that Uxbridge branch services and Harrow terminators are displayed on P4 and everything else northbound on P3 UNLESS/UNTIL a route that contradicts that premise is set. Add to that that the raw data for northbound trains is more complete/accurate, and that would explain why those DMIs seem to make sense (well they do when I've seen them anyhow).
Any such fiddle factor for the southbound DMIs sounds like it would be impossible to do without more data being available to the fiddle factor code. My technical instinct tells me that it might not be too much extra that is needed, but still would be hundreds of times more complex than the P3/4 enhancement, and the information to base the prediction would need to come from several places very distant.
|
|
, Maybe they have competitions on how many times they can get away with doing that between successive departures (with bonus points for puzzled expressions), I don’t know! 
