digital screens and automated announcements wrongly timed

[34D]

Hello,

Hope someone can offer some insight to this issue. I live on the Leeds-Garforth-Selby/York line. Within the past year or so, all stations (not 100% sure about Ulleskelf or South Milford) have been fitted with digital screens and audio announcement facilities.

At East Garforth, the announcements come too late: ie "please stand clear of the edge of platform 2" comes as (literally) the fast train is in the platform.

At cross gates, they come too early. Example "please stand clear" then 5 minutes later "platform 2 for the 1145 service to York" then after another minute the fast train flies through.

Also, I was at Knottingley and Pontefract the other day, on a train to Leeds that was approx 5 mins late and had completely disappeared from the screens (which were saying first train Wakefield).

Can anyone suggest why? More to the point, can anything be done, and if so is it a TOC issue or a Network Rail one?

 

[YorkshireRider]

I was at Cross Gates the other week and the 1041 to Blackpool was showing due at 1046, it then rolled in bang on time at 1041

 

[Peter Mugridge]

On the Underground the Bakerloo Line invariably announces "Stand clear of the doors" after they have shut...

 

[greg tingey]

“Please stand behind the yellow line when the train comes into the platform, & let the passengers off first … use all avaialable doors down the whole train”
MEANWHILE …
The train was already in the platform, with all its’ doors open & had been so for at least a minute.

Victoria line UndergounD ....
LUL are really, really bad about this sort of thing

 

[asylumxl]

I'm not sure about how these screens are set up, but in the case of the ones you find down here I believe the announcements are set via the passing of a signal +/- a delay.

In this case the signal may be very close to the platforms and as such the train is shooting through by the time the announcement plays.

As for incorrect time keeping, I believe this is done via timing points along the route and is purely an estimate. As such if there are very few timing points the estimate can vary widely.

I might be totally wrong though.

 

[thelongestroad]

FGW have a similar problem on parts of the Reading to Gatwick line. The signalling block sections are so long that the CIS isn't very accurate at times.

 

[VTPreston_Tez]

“Please stand behind the yellow line when the train comes into the platform, & let the passengers off first … use all avaialable doors down the whole train”
MEANWHILE …
The train was already in the platform, with all its’ doors open & had been so for at least a minute.

Victoria line UndergounD ....
LUL are really, really bad about this sort of thing

It's certainly okay on the Northern line, years of semi active use of it and never had an announcement fault yet

 

[Class377/5]

Hello,

Hope someone can offer some insight to this issue. I live on the Leeds-Garforth-Selby/York line. Within the past year or so, all stations (not 100% sure about Ulleskelf or South Milford) have been fitted with digital screens and audio announcement facilities.

At East Garforth, the announcements come too late: ie "please stand clear of the edge of platform 2" comes as (literally) the fast train is in the platform.

At cross gates, they come too early. Example "please stand clear" then 5 minutes later "platform 2 for the 1145 service to York" then after another minute the fast train flies through.

Also, I was at Knottingley and Pontefract the other day, on a train to Leeds that was approx 5 mins late and had completely disappeared from the screens (which were saying first train Wakefield).

Can anyone suggest why? More to the point, can anything be done, and if so is it a TOC issue or a Network Rail one?

Note that not all systems work the same. There's a major London TOC who uses 3G (using a dongle in the CIS system) to transmit limited info to the CIS. Northern updates some of its CIS by a train pulling into a platform and using wifi to transmit data. Going off the description you OP gives I sounds like the main system used, which is based of NR berth data.

In this situation, technically it's a Network Rail issue as the announcements are based on berths prior to the station/platform in question. There is often a problem with the berth being triggered too late (or indeed too early) and the announcement being some what pointless by the time it's being made.

But as the berth are not for announcement systems (in the eyes of NR) it falls to the TOC to pay for it as its a purely TOC problem. However getting it moved and the signalling that goes with it isn't cheap and unlikely to happen.

In some places it may be impossible to provide timely info (ie where junctions are) as the first location the system can report a train may indeed be the platform. In this situation then your looking at no easy soluation.

 

[Addington]

More to the point, can anything be done, and if so is it a TOC issue or a Network Rail one?

Try talking to Network Rail, but be prepared to go around in circles. When I was in Wales, the auto help points died. When I spoke to Arriva, they said they manage the stations, but the infrastructure was a Network Rail problem. Network Rail will then tell you the TOC manages the station, so talk to them.

 

[Westboy]

This happens freuently at Bridgwater. "The Train now appproaching..." is probably still in Highbridge when she first announces it, and then she announces it about 6 times again before it gets there.

Even more annoying is that there is gap in the speech, so it "First Great Western.........Serivce to.". Grates.

 

[moggie]

Note that not all systems work the same. There's a major London TOC who uses 3G (using a dongle in the CIS system) to transmit limited info to the CIS. Northern updates some of its CIS by a train pulling into a platform and using wifi to transmit data. Going off the description you OP gives I sounds like the main system used, which is based of NR berth data.

In this situation, technically it's a Network Rail issue as the announcements are based on berths prior to the station/platform in question. There is often a problem with the berth being triggered too late (or indeed too early) and the announcement being some what pointless by the time it's being made.

But as the berth are not for announcement systems (in the eyes of NR) it falls to the TOC to pay for it as its a purely TOC problem. However getting it moved and the signalling that goes with it isn't cheap and unlikely to happen.

In some places it may be impossible to provide timely info (ie where junctions are) as the first location the system can report a train may indeed be the platform. In this situation then your looking at no easy soluation.

It's clearly a TOC problem as in most instances it they that insist on provision using their own procurement systems. They could of course ask NR to provide it but TOC's would still have to specify their operational requirements.

Irrespective of which organisation provides CIS equipment these automated systems all rely on real time train running information derived in some way from the NR TD system. This in turn uses raw data of track section occupancy from the signalling system. The signalling system is ONLY designed for the safety critical operational requirements of the railway system - i.e it is not optimised for (to use the term) non-vital information systems. The location of station platforms is unrecognised by the signalling system albeit that signals are often positioned optimally at the end of platforms, so the postion of a train at those signals may coincide with a train at the platform.

With the advent of technology (and an increase in public expectation) that permits economical application of CIS systems at most platforms it is increasingly the case that use of raw data derived from the signalling system is often going to be unsuitable for use and display of accurate train positioning info to the public for the reason stated. While it is possible to write various sofware enhancements within the CIS system to more accurately configure the system it will always be the case that this will not be entirely foolproof.

The answer must surely be reliable interface to the train via GPS style systems until such time as the signalling system uses simliar information (ERTMS / ETCS) - don't hold your breath!

 

[M60lad]

I know at Bradford Interchange is one of those stations were the screens can't be trusted sometimes as I've been there and seen them saying that there's a Manchester Victoria service running about 5mins late only for the train to turn up on time which can be confusing at times.

I've also seen other screens saying that a train is delayed but grossly exaggerating the delay, saying its delayed for about 15mins when in reality its only a 5min delay, makes me wonder sometimes how the computer operating the screens works out the delays

 

[CNash]

Sometimes, a train can be sat at a terminus for a good number of minutes before a platform number is advertised for it; other times, the platform number appears before the train has arrived. Is this related?

 

[VTPreston_Tez]

Sometimes, a train can be sat at a terminus for a good number of minutes before a platform number is advertised for it; other times, the platform number appears before the train has arrived. Is this related?

I believe that this is because the train needs to be prepared. At some stations (Bradford Interchange as an example) it's announced in advance as there's only a reversal, not a termination,

 

[Class377/5]

It's clearly a TOC problem as in most instances it they that insist on provision using their own procurement systems. They could of course ask NR to provide it but TOC's would still have to specify their operational requirements.

Don't think you understand how most commonly implemented system works. It's track/signal related meaning it has to work with NR kit so blaming it all on a TOC kits is missing the point.

Irrespective of which organisation provides CIS equipment these automated systems all rely on real time train running information derived in some way from the NR TD system. This in turn uses raw data of track section occupancy from the signalling system. The signalling system is ONLY designed for the safety critical operational requirements of the railway system - i.e it is not optimised for (to use the term) non-vital information systems. The location of station platforms is unrecognised by the signalling system albeit that signals are often positioned optimally at the end of platforms, so the postion of a train at those signals may coincide with a train at the platform.

No they don't. As I explained in my post, there are multiple systems and not all use real time data. And some are completely unconnected to the berth system. I know this as Ive used systems they aren't use this during 2012. My post was from experience.

With the advent of technology (and an increase in public expectation) that permits economical application of CIS systems at most platforms it is increasingly the case that use of raw data derived from the signalling system is often going to be unsuitable for use and display of accurate train positioning info to the public for the reason stated. While it is possible to write various sofware enhancements within the CIS system to more accurately configure the system it will always be the case that this will not be entirely foolproof.

The answer must surely be reliable interface to the train via GPS style systems until such time as the signalling system uses simliar information (ERTMS / ETCS) - don't hold your breath!

What's your answer for underground locations? Such as Thameslink?

The problem with real time data is its real time time. Meaning it can never be 100% correct as trains do not always go from point A to point B in the same time. Simple things like passenger loadings can cause minor delays, which some systems don't cope well in showing. Then again some passengers cannot cope with a few minutes delay!

There are ways of getting some data from signal box across (we're currently working on using more data) but there is no one correct soluation for every point. Indeed there is no soluation currently for some locations.

Sometimes, a train can be sat at a terminus for a good number of minutes before a platform number is advertised for it; other times, the platform number appears before the train has arrived. Is this related?

A train might not be showing the correct headcode so won't show properly on the boards. Or it old be because it's been hidden for operation reasons such as cleaning the train.

 

[wintonian]

Nothing unusual.

Although SWT's CIS has improved over the years you still get the occasional "please stand clear" etc. after the train has passed or "the next train to call at platform X..." immediately just before a bit of freight whizzes past at 90 mph, or the previous terminating service is still in the platform. The most common and unhelpful one IMO is the on board announcements that go "this train is for ..." etc. after the doors have been locked.

 

[Lrd]

The one I hear at Southampton from time to time is "Please stand away from the train at platform 1 as it is ready to leave" as it powering it's way out of the station.

 

[Solent&Wessex]

At Keighley, trains on platform 1 are always announced as 3 minutes late. They are nearly always on time. On the West Yorkshire side of the Calder Valley line the new screens are manually controlled from Northern control in York, so it relies on the folk in control keeping it up to date.

 

[wintonian]

At Keighley, trains on platform 1 are always announced as 3 minutes late. They are nearly always on time. On the West Yorkshire side of the Calder Valley line the new screens are manually controlled from Northern control in York, so it relies on the folk in control keeping it up to date.

That reminds me you never quite know what is going to turn up on plt 7 (nee 4) at Reading when everything due there is either on time or a couple of minutes down. It's not been unusual for the CIS to show the Plymouth train as next only for the Hereford and then Oxford trains to turn up first.

 

[Marton]

The displays and announcements in Middlesbrough seem to be wrong much of the time.

Announcing trains after they have left is the usual problem.

A particular problem for the 1739 to Newcastle an the 1740 Whitby which share platform 1, but go in rather opposite directions.

 

[moggie]

Don't think you understand how most commonly implemented system works. It's track/signal related meaning it has to work with NR kit so blaming it all on a TOC kits is missing the point.

No they don't. As I explained in my post, there are multiple systems and not all use real time data. And some are completely unconnected to the berth system. I know this as Ive used systems they aren't use this during 2012. My post was from experience.

I probably understand as much about CIS systems as you do about main line signalling systems.............

My response was to the OP who asked why there were glaring inaccuracies with the information presented to the passenger and which organisation was at fault.

My response attempted to explain in general terms (and I accept not all systems use the same derived info) the information is dependent on various trigger points ultimately derived from the signalling system. I simply sought to explain that the signalling information is not optimised for use by CIS as this is not its designed purpose.

If the TOC's choose to use sub-optimal input data from a system designed for other purposes (safety critical control of the railway) that's fine but imho it's hardly NR's fault if they do so and the data quality is not optimal for CIS use. Likewise as most 'modern' NR signalling systems use standard signalling comms protocols one would expect TOC CIS systems to be specified to use the same if they choose to derive their input data from the NR systems. Of course that means that in some instances they will also have to cope with various legacy systems still in use with NR - so be it.

It informative to know that some TOC's are developing their own stand alone systems that do not rely on signalling data - that in my view is the approprate way forward until such time as the core safety control infrastructure is aligned with all peripheral information systems which I assume will be an ETCS / ERTMS advanced capability.

 

[3141]

There's an interesting one at Salisbury in the later afternoon when a 3-coach unit enters service from the depot and stops at the front end of the platform for Waterloo. (It's a while since I've been there and I cannot remember the platform number.) A few minutes later a train from Exeter arrives and couples to the three coaches in front.

The announcement for the train that's entering service says "Stand well back on platform X. The approaching train is not scheduled to stop at this station." But of course it is scheduled to stop and it does. They haven't got a pre-recorded announcement to cover a unit entering service, which will wait in the platform before a second unit couples to it.

 

[jon0844]

When FCC switched its systems to the 'new' Atos one, there were loads of issues that didn't exist before - and I am sure a lot of the raw data being fed in hadn't changed. It had incorrect timetable data, with trains for King's Cross calling at all stations to Moorgate - and then King's Cross. You had destinations with spelling errors and much more - a total and utter mess. The standard announcements also repeated far too often, while apologies for delays to non-existent services stacked up and were read in one long loop that could go on for 5 or 10 minutes!

But, while a lot of that was fixed quite quickly - you had a spell where information from the track circuits was being delayed and affecting other systems too. Many trains showed as late because they'd already passed a point but it hadn't been registered, which also meant that trains turned up 'early' and probably caught people out expecting it in 2-3 minutes and choosing to grab a quick coffee (me? No, of course not - well, only the one time!). It also looked bad from a customer perspective that got used to seeing 'delayed' on most trains on the boards, their smartphone app etc.

I'd say it's taken FCC the best part of two years to iron out most of the bugs on what I know believe is end-of-line software that won't probably get any new updates beyond bug-fixes. So, if your TOC has switched over to the Atos system then maybe that will give you a ballpark figure on how long it could be until it all works as intended!

What I find annoying is that the Atos system is set so it only announces the train when it is about 20-25 seconds away. The good thing is that hearing that announcement tells you the train really is approaching (before it crosses the nearby signal, it won't trigger) but the downside is that a) you don't get any advance announcements at all - so anyone that is blind won't know anything until the train is there b) it doesn't repeat the announcement when the train has stopped so staff get bombarded with 'Is this the King's Cross train?' when they run in, or simply haven't heard the original one (perhaps because a non-stopping train came flying by).

I asked FCC why they (effectively) axed the others and they simply said they only have to do one announcement. So I guess in the mind of the current management, doing the bare minimum to tick a box is better than doing a bit more to benefit passengers. FCC also told me Hatfield is too small to be zoned, even though the old system was...

 

[VTPreston_Tez]

Nothing unusual.

Although SWT's CIS has improved over the years you still get the occasional "please stand clear" etc. after the train has passed or "the next train to call at platform X..." immediately just before a bit of freight whizzes past at 90 mph, or the previous terminating service is still in the platform. The most common and unhelpful one IMO is the on board announcements that go "this train is for ..." etc. after the doors have been locked.

At least it has working announcements for fast trains unlike at Rose Grove!
(Disclaimer: This was the last time I was there, mid-late 2012)

 

[MidnightFlyer]

At least it has working announcements for fast trains unlike at Rose Grove!
(Disclaimer: This was the last time I was there, mid-late 2012)

Or any of the other local stations along the East Lancs line, which is most probably because they aren't programmed to announce them A good example of their usefulness however was at one station along there, where the platforms announced (e.g. Platform 2 for Colne) didn't match up to the newly-installed numbers on the platforms...

 

[wintonian]

Saw a twitter comment yesterday from someone concerned about the "the next train on platform * does not stop here" announcement at Portsmouth Harbour again, mid you I get almost as concerned when I hear it coming into plt 3 on the Soton stopper - at least there I don't require a life-jacket.

 

[causton]

The London Overground CIS screens on the platforms at Clapham High Street and Wandsworth Road aren't set up for passing trains - they just keep displaying the next service... I don't remember if the old yellow CIS screens did beforehand (but I think they did, as the parliamentary 1610 from CLP was announced as a non-stop I'm sure!)

 

[Class377/5]

I probably understand as much about CIS systems as you do about main line signalling systems.............

My response was to the OP who asked why there were glaring inaccuracies with the information presented to the passenger and which organisation was at fault.

My response attempted to explain in general terms (and I accept not all systems use the same derived info) the information is dependent on various trigger points ultimately derived from the signalling system. I simply sought to explain that the signalling information is not optimised for use by CIS as this is not its designed purpose.

If the TOC's choose to use sub-optimal input data from a system designed for other purposes (safety critical control of the railway) that's fine but imho it's hardly NR's fault if they do so and the data quality is not optimal for CIS use. Likewise as most 'modern' NR signalling systems use standard signalling comms protocols one would expect TOC CIS systems to be specified to use the same if they choose to derive their input data from the NR systems. Of course that means that in some instances they will also have to cope with various legacy systems still in use with NR - so be it.

It informative to know that some TOC's are developing their own stand alone systems that do not rely on signalling data - that in my view is the approprate way forward until such time as the core safety control infrastructure is aligned with all peripheral information systems which I assume will be an ETCS / ERTMS advanced capability.

My job requires me to understand bit of both actually. The rest I get from the signallers I work alongside daily.

The problem with TOCs creating thier own system is you'd have multiple set ups and TOC equipment on the tracks so this set up is easier for all. After all TOC has no right to have access to track and indeed a lot of them have no track based staff other than drivers, shunters and fitters.

And there is of course the fee NR charge for this data to the TOCs which is why NR properly prefers this set up.

ETCS/ERTMS as I understand it won't be giving public data on locations etc. as Thameslink will be the first major implemation, we've no plan to move from the current system. GSM-R may provide location data but again that's using NR intrastructure to allow TOCs to get the information.

I'd say it's taken FCC the best part of two years to iron out most of the bugs on what I know believe is end-of-line software that won't probably get any new updates beyond bug-fixes. So, if your TOC has switched over to the Atos system then maybe that will give you a ballpark figure on how long it could be until it all works as intended!

What I find annoying is that the Atos system is set so it only announces the train when it is about 20-25 seconds away. The good thing is that hearing that announcement tells you the train really is approaching (before it crosses the nearby signal, it won't trigger) but the downside is that a) you don't get any advance announcements at all - so anyone that is blind won't know anything until the train is there b) it doesn't repeat the announcement when the train has stopped so staff get bombarded with 'Is this the King's Cross train?' when they run in, or simply haven't heard the original one (perhaps because a non-stopping train came flying by).

I asked FCC why they (effectively) axed the others and they simply said they only have to do one announcement. So I guess in the mind of the current management, doing the bare minimum to tick a box is better than doing a bit more to benefit passengers. FCC also told me Hatfield is too small to be zoned, even though the old system was...

That's not true, FCC is not using an end of line software, in fact it's just received a major, 12 months in development upgrade with a possible future one coming. As for other TOCs, they have a choice of which part of the ATOS system to use, for example Southeastern are moving over to the same system and having to learn the differences between thier old and good system. Some parts are better, some aren't.

The ATOS system runs of track data so it's not set to a certain time as much as its based on incoming data. It's not ideal as discussed above but other alternatives will be much more expensive. There is a way to get to advanced data onto the CIS system but when you have junctions near by it requires signaller to set up routes in advance and dependant on that route being kept. There is simply no way to be sure when junctions are either end of a station as its the signaller who will decide which platform and no system can pre-judge that correctly.