Real-time delay information

[math1985]

Passengers have different ways to get informed about possible delays: overhead next-train displays, tv-like displays, the National Rail site/app, the Network West Midlands site/app (in the West Midlands), and sites like http://www.realtimetrains.co.uk/. I noticed that these systems are not always in agreement with each other.

For example, today a southbound Cross City train was delayed. As a result, the overhead displays at Selly Oak announced a delay for the northbound train as well (predicting the delay as a consequence of the delayed incoming train). However, RTT indicated an on-time train. In the end, however, the northbound train was on time, because the train was cancelled between Barnt Green and Redditch. Was RTT smarter than the overhead displays because they knew about the part-cancellation, or is RTT less intelligent because they don't use information about delayed incoming trains (and the correct prediction was in this case just luck)?

In general, I wonder how the different systems calculate the delays. I often see TRUST mentioned in this context. But if all systems get their information from TRUST, I wonder why the systems display different delays? Does anyone have any insight in how these systems work? And which of the systems is typically most accurate?

 

[infobleep]

I don't have any insight into how it all works but I would love the systems to be more in agreement. On National Rail Enquiries I would also like reasons for delays to be against trains which are seriously delayed rather than ignoring those at times but still putting reasons against those delayed by only a few minutes if that.

I would like them to be clearer on what a minor delay is. Recently South West Trains services were being delayed by 30 minutes but the delays on the overview page were only considered minor. May be that is correct. I would have thought 30 minutes to be major.

Another morning both South West Trains and National Rail Enquiries were reporting delays of up to 20 minutes through Wimbledon or cancellations but that wasn't much consultation for people at stations between West Byfleet and Esher, where there was no train service for an hour, all services having been cancelled. In that case far better to say how delayed passengers might be.

I was once told on Twitter that they couldn't put the maximum time trains would be delayed as for most people that wouldn't be their experience. Well if that is the case don't put up to. Put on average.

Anyway hopefully someone or some people with influence are interested in improving this and in time it will get better and more constant between systems.

I fully admit it's not always easy during periods of disruption but I still feel improvements could be made.

A guard once said to me, don't believe what you read online. Incidentally the National Rail Enquiries App appears to include any train service entered and then cancelled. Then if staff reenter it, it appears again. This doesn't affect their Web Site, just the app.

I don't know if they have given up on developing the app or something better is in the pipe line. I do like however the fact you can go back in time. You can't do that on the National Rail Enquiries Web Site. However it doesn't go back in time passed midnight. It would have been better if they had used the railway day, rather than the calendar day.

Sent from my Nexus 5 using Tapatalk

 

[Bald Rick]

Passengers have different ways to get informed about possible delays: overhead next-train displays, tv-like displays, the National Rail site/app, the Network West Midlands site/app (in the West Midlands), and sites like http://www.realtimetrains.co.uk/. I noticed that these systems are not always in agreement with each other.

For example, today a southbound Cross City train was delayed. As a result, the overhead displays at Selly Oak announced a delay for the northbound train as well (predicting the delay as a consequence of the delayed incoming train). However, RTT indicated an on-time train. In the end, however, the northbound train was on time, because the train was cancelled between Barnt Green and Redditch. Was RTT smarter than the overhead displays because they knew about the part-cancellation, or is RTT less intelligent because they don't use information about delayed incoming trains (and the correct prediction was in this case just luck)?

In general, I wonder how the different systems calculate the delays. I often see TRUST mentioned in this context. But if all systems get their information from TRUST, I wonder why the systems display different delays? Does anyone have any insight in how these systems work? And which of the systems is typically most accurate?

TRUST does not and can not predict delays. All it does is record delay. Systems downstream then do the predicting. In general the downstream systems will just add the existing delay to the rest of a train's schedule, subtracting any timetable allowances, to get to the predicted delay at your station. It appears some systems also subtract additional dwell time, ie where scheduled dwell time is over the standard allowed for that station. Generally, anything else displayed has been subject to manual intervention.

Trains returning from terminal stations is a difficult one, and unless inbound and outbound share the same head code, will always be shown as on time until they have left their origin point - unless subject to manual intervention in the system.

Similarly no system is yet clever enough to predict delay, eg if every train is delayed 15 mins through a section, it will still be shown on time until it is delayed, again unless there has been manual intervention.

What is coming over the hill is the Traffic Management systems. The predictive train graph and conflict resolution components will show the actual future times of a service, and with the diagrams* plugged in will also link inbound / outbound services at termini. Future delays at a given section, eg for a speed restriction, can also be forecast. This all goes to the downstream systems, making CIS much more accurate.

However even TMS can't predict incidents that have yet to happen, and ...

* rolling stock diagrams are easy, but driver / conductor diagrams gets a bit more difficult in terms of predicting turnaround delays. It is possible, but the nature of crewing arrangements, and that the system is predicting people rather than a machine, makes it more of a challenge.

Hope this helps.

 

[PHILIPE]

One of the most annoying things is trains disappearing from the screen when they haven't turned up as expected.

 

[infobleep]

Thanks for the reply Bald Rick. Fascinating.

Sent from my Nexus 5 using Tapatalk

 

[Bald Rick]

One of the most annoying things is trains disappearing from the screen when they haven't turned up as expected.

I agree. I'm not sure what causes that, I have always assumed that the 'system' thinks the train must have called but has somehow been missed so auto deletes.

It is a similar annoyance to that when you are cruising down the motorway and the matrix signs flash "60", then "40" for a couple of sections then give the all clear with no congestion, accident or other reason for the restriction. You start to lose faith in the system.

 

[math1985]

Thanks.

I take it any manual intervention happens after the stage when RTT gets the data? Is the manual intervention done by the TOC, or by Network Rail?

 

[Qwerty133]

Not even all departure boards at a station will display the same time, once I saw 3 different times. for the same service (2 different boards and the NRE app or RTT)

 

[Parallel]

I agree. I'm not sure what causes that, I have always assumed that the 'system' thinks the train must have called but has somehow been missed so auto deletes.

Yes, this is annoying - Happened last Thursday (12th June) during the signalling chaos at Westbury. There were passengers waiting for the Cheltenham Spa via. Melksham service at 19.32 and it had an earlier delayed service to BTM that was expected to arrive at the same platform at 19.32 but ended up departing at xx:40 something, and everyone seemed confused as to where their train had gone! Then it came back on the screen, and then disappeared again. The actual train then went north of the station from the sidings, and then driving back to platform 3 when the announcement said "the next train is not scheduled to stop" and it displayed "Cardiff" in the window! Because it was due to start at Westbury, I guess it didn't have any track sensors to go over to begin with so kept getting deleted off the board! Eventually it left nearly 25 mins late.

This happens a lot at the Cambrian coast stations. Due to the few track sensors at the request stops (the "no reports" on RTT), e.g. If you arrive at Dyffryn to get to Harlech in plenty of time, the train will just be wiped off when it reaches it's scheduled or expected time (probably worked out by how late it is by Barmouth) and you can only see the next service to Birmingham International, even if the Harlech service is still over 5 minutes away from actually arriving!

 

[infobleep]

This morning I am certain the 7.04 from Redhill to Victoria left a minute late or if not just 1 second before, based on the station clocks. However the National Rail Enquiries App show it as having left on time.

The app however also shows the train as arriving a minute late into East Croydon. How do they know that given it left on time?

Lately I've noticed a reasonable number of trains showing as being a minute late to stations before they have even started their journey. I've noticed this on both Southern, First Capital Connect and South West Trains services. Is it due to the working timetable being different?

Sent from my Nexus 5 using Tapatalk

 

[The Planner]

TRUST rounds down so anything up to 59 seconds late counts as the same minute or on time.

 

[math1985]

This morning I am certain the 7.04 from Redhill to Victoria left a minute late or if not just 1 second before, based on the station clocks. However the National Rail Enquiries App show it as having left on time.

The app however also shows the train as arriving a minute late into East Croydon. How do they know that given it left on time?

RTT seems to give the answer: the time is also measured in the stations where it doesn't stop. For example, although it left Redhill on time, it passed Merstham 2 minutes and Coulsdon South one minute late, so presumably that was used to calculate the delay in East Croydon.

 

[PaxVobiscum]

The 'UK Train Times' app sometimes gives an apparently random predicted delay at one or several stations for trains which have still to depart. Not sure the following is a good example, but I have certainly seen some which were highly improbable and did not actually happen when the train ran. I have often wondered why it does this.

 

[greatkingrat]

In that case it is a GBTT / WTT difference. That train is advertised to depart Glasgow Central at 1950, but the WTT shows a 1952 departure, hence the app is showing it as 2 minutes late.

http://www.realtimetrains.co.uk/train/G54131/2014/06/16/advanced

 

[math1985]

The 'UK Train Times' app sometimes gives an apparently random predicted delay at one or several stations for trains which have still to depart. Not sure the following is a good example, but I have certainly seen some which were highly improbable and did not actually happen when the train ran. I have often wondered why it does this.

Very good example.

That app uses the same data as National Rail (NR): http://ojp.nationalrail.co.uk/service/ldbdetails/9PVacmR0MWnlUqbkRf5mBQ==

If you look on RTT, you see that for some reason, the WTT is 1,5-2,5 minutes later than the GBTT (probably the WTT has been pushed 2 minutes later).

Now let's look at how NR calculates the delays. Delays of 1 minute are indicated as 'on time'. NR apparently uses the differences in arrival time between WTT and GBTT to calculate the delay. The stations that are indicated as '2 mins late' have a arrival time delay of 2 minutes. The other stations have an arrival time delay of less than that.

Let me know if you have any more examples like that.

 

[The Planner]

Delays of 59 seconds are on time, 60 seconds is 1 minute late.

 

[math1985]

Delays of 59 seconds are on time, 60 seconds is 1 minute late.

Who are you replying to?

 

[Crossover]

It is a similar annoyance to that when you are cruising down the motorway and the matrix signs flash "60", then "40" for a couple of sections then give the all clear with no congestion, accident or other reason for the restriction. You start to lose faith in the system.

Agreed - that really annoys me (thinking I'm going to end up stuck in traffic!)

 

[Nicholas43]

Trains returning from terminal stations is a difficult one, and unless inbound and outbound share the same head code, will always be shown as on time until they have left their origin point - unless subject to manual intervention in the system.

In my experience at Paddington, this human intervention fails to get a grip - but very likely I underrate how difficult it is. Typical scenario is that arriving train whose stock is forming departure at (say) 2000 is known to be, say, 30 down, and therefore there is no chance of departure at 2000. But "on time" will be mindlessly displayed until a human admits otherwise. Then we get "delayed". Then "expected (say) 2020". Then, in a recent case, I presume because control wanted a quick turn-round, "boarding". Intending passengers rush to platform, to find no train at all, let alone a train ready for boarding. This seems to be a big lack in the vocabulary at Paddington. In this situation, why not reveal just the platform number, and a realistic departure time?

 

[infobleep]

Do train companies get fined if they don't give passengers enough information to be able to board their train in time? I suspect not. Thus it doesn't matter so much if they leave without them, providing it doesn't reach the newspapers.

Sent from my Nexus 5 using Tapatalk

 

[DJL]

Do train companies get fined if they don't give passengers enough information to be able to board their train in time? I suspect not. Thus it doesn't matter so much if they leave without them, providing it doesn't reach the newspapers.

Sent from my Nexus 5 using Tapatalk

I was once the only passenger on a long train (at least 8 cars) in the rush hour.
And I mean literally the only passenger.

The train went non-stop from London Bridge to Orpington in about 16 minutes then terminated.

Obviously this was during some fairly major disruption. I only spotted the train by accident on the departure board for that platform (2 if memory serves). It wasn't on any other board or announcement.

Whilst I was extremely grateful for my personal limo-on-rails home it did strike me as a colossal waste.
I have no idea how long it was for the next train but it was undoubtedly crush-loaded.

 

[infobleep]

A few months ago I goon a delayed Poole service from Clapham Junction. There was fatality and they had told passengers to get the slow stopping service to Basingstoke, which may have run via Chertsey as it was from one of those platforms. It was crush loaded so I decided to head back to to Poole train.

It soon left with people on board but it wasn't that busy. Certainly less so than normal for 23.15 on a Saturday night.

Sent from my Nexus 5 using Tapatalk

 

[infobleep]

Interestingly this evening the National Rail Enquiries live departure board for Woking talks of delays between Portsmouth / Havant and Farham of up to 60 minutes. Where as at Guildford and also Clapham Junction, they only talk of delays to Motspur Park. Needless to say Woking doesn't have services to Motspur Park but Guildford does have services to Portsmouth and Havant and Fareham although only 1 in either direction to Fareham.

Makes you wonder how online information is added to live departure boards. Surely they don't have to manually select each station. Given that the other day there was delay information about services around Birmingham during the day and only 1 service in either direction heads there from Guildford and not at the time of the disruption.

Sent from my Nexus 5 using Tapatalk

 

[Bald Rick]

In my experience at Paddington, this human intervention fails to get a grip - but very likely I underrate how difficult it is. Typical scenario is that arriving train whose stock is forming departure at (say) 2000 is known to be, say, 30 down, and therefore there is no chance of departure at 2000. But "on time" will be mindlessly displayed until a human admits otherwise. Then we get "delayed". Then "expected (say) 2020". Then, in a recent case, I presume because control wanted a quick turn-round, "boarding". Intending passengers rush to platform, to find no train at all, let alone a train ready for boarding. This seems to be a big lack in the vocabulary at Paddington. In this situation, why not reveal just the platform number, and a realistic departure time?

Understandable frustration. Control may have a habit of doing a set swap, so putting up a realistic departure time of what that train should be working, say 30 min delay, then find out that control have stepped up resources, means plenty of people might miss the train having gone down the pub for half an hour. Such control decisions can be made at very short notice, like 5 mins or less.

Also, the reality is that the info screens will be worked by one person who may also be doing announcements and the station radio.

 

[infobleep]

More delays at Fareham and Havant and Portsmouth but today they have put announcements on the Guildford live departure board. Delays are only listed as up to 20 minutes but even that appears to be wrong because if a train or two are cancelled, your delay won't be 20 minutes I wouldn't have thought.

Anyway Farncombe has no live departure board announcement so they have managed to find at least one station not to inform online. Lol.

I'm sure getting the right delay information out will improve in time.

UPDATE: There is no information in the Clapham Junction live departure board but yesterday that had information when Guildford didn't! Nothing like being constant!

Sent from my Nexus 5 using Tapatalk

 

[swt_passenger]

Interestingly this evening the National Rail Enquiries live departure board for Woking talks of delays between Portsmouth / Havant and Farham of up to 60 minutes.

Last night, at around this time there was a person hit by a train in Cosham Station. That would have mainly affected trains between Portsmouth and Fareham or Havant and Fareham, but would not necessarily have had any or much effect on the main line services up and down the Portsmouth direct and via the other side of the Farlington triangle.

I think if NR had issues between Portsmouth and Havant they'd have been shown differently.

So in this particular case you shouldn't necessarily have assumed they were also implying delays between Portsmouth and Havant, but the info provided for Woking enquiries would have been useful for people making connections at Havant?

 

[infobleep]

However there was a train last night running very late via Guildford so clearly Guildford should have been included.

This morning Clapham Junction wasn't included when trains most definitely stop at that station.

Sent from my Nexus 5 using Tapatalk

 

[infobleep]

Got off the late running 18.54 Woking to Basingstoke which unusually arrived into platform 4. After I alighted the guard blew his whistle whilst I heard an announcement on platform 5 telling customers to go to platform 4 for the Basingstoke train.

As I cross the footbridge and saw people running I heard the doors of the Basingstoke bound train reopen. Perhaps the guard saw the people running. He could quite easily have left without them I imagine. I'm very grateful on behalf of the other passengers that he didn't.

Sent from my Nexus 5 using Tapatalk