How does Realtime Trains predict platforms?

Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R

RailUK Forums

Clip

On Moderation
Joined
28 Jun 2010
Messages
10,616
As far as im aware all platforms are part of the time table or else you couldnt really make them work could you?
 

StephenHunter

Member
Joined
22 Jul 2017
Messages
534
Location
London
In many cases, the platforms are pretty standardised because of the layout of the line. For example, TfL Rail services at Romford use the northernmost 'electric' lines so will use Platforms 4 and 5 unless engineering work intervenes.
 

Bald Rick

Veteran Member
Joined
28 Sep 2010
Messages
12,921
It’s what is in the timetable data, which includes platforms. For more simple stations, it’s done by line code - ie if it is on the ‘up’ line st a 2 Platform station it must be routed to the platform on the up line. Ie, it’s what is in the timetable.
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
It was more intelligent than just using the WTT. It, currently, just shows the WTT until it knows demonstrably otherwise. Up until about 6 months ago, it would infer a future platform by inferring the platforms further down the line based on a 95% likelihood. This was removed in favour of a system currently under testing, where it is in place in (I think) 12 stations at the network. There will be an announcement about that at some point soon.
 

M60lad

Member
Joined
31 May 2011
Messages
500
Its certainly a good way of finding out at London Euston what platform a train is going from even before it shows on the station departure boards, it usually comes up on Realtime Trains about 5mins or before the departure boards are updated.
 

PHILIPE

Veteran Member
Joined
14 Nov 2011
Messages
11,692
Location
Caerphilly
Its certainly a good way of finding out at London Euston what platform a train is going from even before it shows on the station departure boards, it usually comes up on Realtime Trains about 5mins or before the departure boards are updated.

That would no doubt be the the booked platform that you see on RTT beforehand
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
That would no doubt be the the booked platform that you see on RTT beforehand
While the platform light grey it is either the booked platform or some other logic to forecast the platform. Once it goes black (previously when it went bold), that is when it is interposed into that platform.
 

Ianno87

Established Member
Joined
3 May 2015
Messages
7,746
That would no doubt be the the booked platform that you see on RTT beforehand
But the actual platform will appear when the inward train has arrived and the signaller interposes the outgoing headcode into the signal berth, usually immediately as the train arrives.
 

43096

Established Member
Joined
23 Nov 2015
Messages
7,397
While the platform light grey it is either the booked platform or some other logic to forecast the platform. Once it goes black (previously when it went bold), that is when it is interposed into that platform.
Doesn't it go red if it is a platform alteration i.e. not the booked platform?
 

Tom Quinne

Established Member
Joined
8 Jul 2017
Messages
1,811
The WTT specifics the booked platform which is what the station PIS uses, however once the route is called by the signaller that’s updates all the integrated systems to show the reality.
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
Doesn't it go red if it is a platform alteration i.e. not the booked platform?
Yes, indeed.
The WTT specifics the booked platform which is what the station PIS uses, however once the route is called by the signaller that’s updates all the integrated systems to show the reality.
The WTT does not include detail of platforms at every station, which the CIS requires, so most systems have some mechanism of adding the platforms that are missing and that includes overwriting the WTT sometimes. I am not convinced that there are any CIS that update on platform change until it's either manually done or the train steps into either (a) a berth of that platform normally or (b) very rarely, a point of no return berth, but I am happy to be corrected.
 

Ianno87

Established Member
Joined
3 May 2015
Messages
7,746
The WTT specifics the booked platform which is what the station PIS uses, however once the route is called by the signaller that’s updates all the integrated systems to show the reality.
Yes, indeed.
The WTT does not include detail of platforms at every station, which the CIS requires, so most systems have some mechanism of adding the platforms that are missing and that includes overwriting the WTT sometimes. I am not convinced that there are any CIS that update on platform change until it's either manually done or the train steps into either (a) a berth of that platform normally or (b) very rarely, a point of no return berth, but I am happy to be corrected.
It usually updates when the train steps into the platform berth.
 

IanXC

Emeritus Moderator
Joined
18 Dec 2009
Messages
5,768
Yes, indeed.
The WTT does not include detail of platforms at every station, which the CIS requires, so most systems have some mechanism of adding the platforms that are missing and that includes overwriting the WTT sometimes. I am not convinced that there are any CIS that update on platform change until it's either manually done or the train steps into either (a) a berth of that platform normally or (b) very rarely, a point of no return berth, but I am happy to be corrected.
There is a system which uses ARS route setting (clearly only where available) in a similar fashion to your "point of no return berth" concept.
 

Horizon22

Member
Joined
8 Sep 2019
Messages
641
Location
London
Booked platform using WTT data and also syncing with Darwin. I believe Red is an alteration, italics is predicted or not yet published (such as empty stock hidden from CIS screens)

It was more intelligent than just using the WTT. It, currently, just shows the WTT until it knows demonstrably otherwise. Up until about 6 months ago, it would infer a future platform by inferring the platforms further down the line based on a 95% likelihood. This was removed in favour of a system currently under testing, where it is in place in (I think) 12 stations at the network. There will be an announcement about that at some point soon.
Would be very wary about this at terminals. Grief as passengers start storming down platforms and then a 'platform change' (NB: it didn't actually change in small letters on most of apps it says "predicted") causes crowding issues and complaints with staff. If you're in the known there's an 80-90% chance you'll be right, but there's always last minute changes and signallers deciding differently for operational (or occasionally not!) reasons.
 

Tim_UK

Member
Joined
9 Jan 2019
Messages
73
Booked platform using WTT data and also syncing with Darwin. I believe Red is an alteration, italics is predicted or not yet published (such as empty stock hidden from CIS screens)
Can you just clarify if you are suggesting that Realtimetrains uses Darwin?

I don’t think it does.
 

AM9

Established Member
Joined
13 May 2014
Messages
8,454
Location
St Albans
There is a system which uses ARS route setting (clearly only where available) in a similar fashion to your "point of no return berth" concept.
Maybe not quite the same but at certain stations where quite wild deviations in platforms can occur (e.g. Euston) I also run Open Train Times Maps where (in the case of Euston) the set route in green also gives the route into the berth.
Having just opened OTT to check, it seems that routes set aren't being displayed at the moment.
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
Booked platform using WTT data and also syncing with Darwin. I believe Red is an alteration, italics is predicted or not yet published (such as empty stock hidden from CIS screens)
RTT does not use Darwin. I have made this clear in the past and I will make it clear again. It never has, does not at the moment and in the current environment is unlikely to ever happen. My reasons for this are public: https://blog.realtimetrains.com/2020/02/thoughts-on-information-during-disruption/

Would be very wary about this at terminals. Grief as passengers start storming down platforms and then a 'platform change' (NB: it didn't actually change in small letters on most of apps it says "predicted") causes crowding issues and complaints with staff. If you're in the known there's an 80-90% chance you'll be right, but there's always last minute changes and signallers deciding differently for operational (or occasionally not!) reasons.
Any app following Darwin data requirements will be suppressing the platform off using the platSup field (if via push port) until such a time at which that goes false. They should never be showing a forecasted platform while that field is held to true. Wouldn't it be great if that kind of data could be used openly? The reason why not - see above.

For this reason, RTT operates based on berth occupation within the signalling system for departures from terminal stations and this has always been the case. For the rest of the network, I had, in the grand total of 7 years of using the previous system, one complaint about it getting a platform wrong when it turned up on the other face of the same island. Nevertheless, it can be improved on and this new system that I have hinted about is in action at several stations on a line of route and is proving to be reliable.

Maybe not quite the same but at certain stations where quite wild deviations in platforms can occur (e.g. Euston) I also run Open Train Times Maps where (in the case of Euston) the set route in green also gives the route into the berth.
Having just opened OTT to check, it seems that routes set aren't being displayed at the moment.
Euston does not have signalling state data to display this. Other areas do.
 

AM9

Established Member
Joined
13 May 2014
Messages
8,454
Location
St Albans
RTT does not use Darwin. I have made this clear in the past and I will make it clear again. It never has, does not at the moment and in the current environment is unlikely to ever happen. My reasons for this are public: https://blog.realtimetrains.com/2020/02/thoughts-on-information-during-disruption/


Any app following Darwin data requirements will be suppressing the platform off using the platSup field (if via push port) until such a time at which that goes false. They should never be showing a forecasted platform while that field is held to true. Wouldn't it be great if that kind of data could be used openly? The reason why not - see above.

For this reason, RTT operates based on berth occupation within the signalling system for departures from terminal stations and this has always been the case. For the rest of the network, I had, in the grand total of 7 years of using the previous system, one complaint about it getting a platform wrong when it turned up on the other face of the same island. Nevertheless, it can be improved on and this new system that I have hinted about is in action at several stations on a line of route and is proving to be reliable.


Euston does not have signalling state data to display this. Other areas do.
OK thanks.
 

Horizon22

Member
Joined
8 Sep 2019
Messages
641
Location
London
RTT does not use Darwin. I have made this clear in the past and I will make it clear again. It never has, does not at the moment and in the current environment is unlikely to ever happen. My reasons for this are public: https://blog.realtimetrains.com/2020/02/thoughts-on-information-during-disruption/


Any app following Darwin data requirements will be suppressing the platform off using the platSup field (if via push port) until such a time at which that goes false. They should never be showing a forecasted platform while that field is held to true. Wouldn't it be great if that kind of data could be used openly? The reason why not - see above.

For this reason, RTT operates based on berth occupation within the signalling system for departures from terminal stations and this has always been the case. For the rest of the network, I had, in the grand total of 7 years of using the previous system, one complaint about it getting a platform wrong when it turned up on the other face of the same island. Nevertheless, it can be improved on and this new system that I have hinted about is in action at several stations on a line of route and is proving to be reliable.


Euston does not have signalling state data to display this. Other areas do.
Apologies I did know that, was in auto-mode as I was ironically working on Darwin at the time! As for the other comment, Trainline and some other apps not specific to rail have done this before. I appreciate the desire to give accurate platform info in advance, so if it uses berth occupation that's fine, but otherwise it will never be 100% correct.

I use RTT regularly for my journeys because of the nice linkage to see if a train is "delayed" as many apps use, whilst RTT gives you information on whether the train has actually moved and/or departed its origin.
 

cactustwirly

Established Member
Joined
10 Apr 2013
Messages
5,695
Location
UK
The example I've witnessed stood on the platform so to speak is Bradford Interchange.
I think Paddington uses it as well.
The platform of suburban services has been announced before the train has entered the platform, when I've caught such services in the past.
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
Apologies I did know that, was in auto-mode as I was ironically working on Darwin at the time! As for the other comment, Trainline and some other apps not specific to rail have done this before. I appreciate the desire to give accurate platform info in advance, so if it uses berth occupation that's fine, but otherwise it will never be 100% correct.
I am led to believe that Darwin itself has done it in the past too, with fairly poor results. RTTs model for this is essentially at origin terminals using berth occupancy and then further down the line it uses other logic which is 99.9% of the time right if the platform is 'confirmed'.

I think Paddington uses it as well.
The platform of suburban services has been announced before the train has entered the platform, when I've caught such services in the past.
This will likely just be manual intervention as is common for most London terminal stations.
 

Horizon22

Member
Joined
8 Sep 2019
Messages
641
Location
London
I am led to believe that Darwin itself has done it in the past too, with fairly poor results. RTTs model for this is essentially at origin terminals using berth occupancy and then further down the line it uses other logic which is 99.9% of the time right if the platform is 'confirmed'.

This will likely just be manual intervention as is common for most London terminal stations.
The CIS at stations can be placed into auto or manual modes. Auto relies on a headcode/ID being occupied in the platform by the signaller. This can be before the train is in the berth and if the headcode is wrong (manual error by signaller), it also won't advertise. Manual requires an extra step for staff in a control room to confirm platforms, but can also be in advance if required/wanted. Normally only applicable at London stations as you state.
 

Indigo2

Established Member
Joined
16 Apr 2010
Messages
3,643
Location
Reading
An example from 5-6 years ago, at Twyford during the Henley regatta. Most of the year Henley trains virtually always use platform 5 if arriving from Henley and terminating to go back again, and only use platform 4 if arriving from Reading or Paddington before continuing to Henley. But during the regatta week the trains were strengthened such that they were too long to fit in platform 5 and so used platform 4 all day instead. RTT correctly showed platform 4 while the train was approaching Twyford, but then switched incorrectly to platform 5 when the train was arriving, showing it as a platform change. I always assumed this was due to some sort of learning algorithm combined with ambiguities in the signalling berth data.
 

Tom

Member
Joined
19 Jan 2008
Messages
360
Location
35,000ft
If it was going to platform 5 on arrival, then it's highly likely to be some weirdness with the berth stepping or mapping. RTT only really got good 'independent' full coverage of the network about 3 years ago I'd say so it may have been an error from the original NR reference data for that area.

I hold the reporting logs from near enough the start of RTT so can look into it if you know the exact dates. It's good to know these things fairly quickly as there are tens of thousands of these things and sometimes something just doesn't happen enough for us to be on top of it normally.
 

Top