Booking engines telling passengers to change from HSTs to Sprinters?

[Envoy]

No matter how many times I bring this to the attention of the TOC's, nobody actually does anything about the way the computer system advises customers to change at the earliest opportunity.

Let us take as an example a journey on 1C04 = the 7.30am from PAD to Paignton. Put in for PAD > TOTnes & you are advised to change at Exeter - presumably for a sprinter (which may be crowded). Now, we rail geeks all know that the last opportunity to leave the HST would be Newton Abbot, wait about 10 minutes & then board the sprinter for the short hop to Totnes.

How about travel to west Wales? Same thing happens >
PAD > Fishguard dep 8.45am. What does it tell people to do? Leave the HST at Newport (a.10.05) to dep. 10.24 on a Sprinter (or even a Pacer) -which you would stay on all the way to Fishguard. Now, this train uses the Swansea District Line to go non stop from Bridgend to Llanelli. That means that the last opportunity to leave the HST is Bridgend. So, customers could have stayed on the HST as far as BGN (a.11.09) & then caught the same Sprinter to dep.11.19.

Here is another one:> PAD dep 8.15am for Carmarthen. National Rail website tells customers to change at Newport (a.10.05 / dep.10.24 on a Manchester > Carmarthen service - probably a Coradia). Now, in all probability, this 2 or 3 coach 175 has come down the Marches line pretty full - and will certainly remain so until Cardiff or even Swansea. So, why the heck are customers being told to leave a Swansea bound HST at Newport?
They could have stayed on the HST to Cardiff. Then they could have caught the 10.42 from Cardiff to Carmarthen - this being the SAME Coradia that they were told to cram onto at Newport!

So, customers are being told to cram onto shorter more crowded trains - which are probably inferior. Hmm...

 

[Clip]

OMG!!


However as Im really not familiar with all those stations how many of the ones you have mentioned are better to stay on till have a same platform change?

 

[Envoy]

OMG!!


However as Im really not familiar with all those stations how many of the ones you have mentioned are better to stay on till have a same platform change?

Newton Abbot would be a same (or level) platform change. Bridgend would be a same platform change. Swansea, being a terminus, is a level platform change. Newport is likely to be a same platform change but also has the possibility of having to cross a bridge to switch trains. So, Newport would surely be the least favoured place to change trains.

 

[Shunter_69]

its most likely to be due to the longer connection time. 4 mins in Swansea isn't very much when you have 19 if you change earlier.

 

[Clip]

Newton Abbot would be a same (or level) platform change. Bridgend would be a same platform change. Swansea, being a terminus, is a level platform change. Newport is likely to be a same platform change but also has the possibility of having to cross a bridge to switch trains. So, Newport would surely be the least favoured place to change trains.

Ok thanks. Must be with interchange times then as shunter_69 mentioned.

 

[Envoy]

its most likely to be due to the longer connection time. 4 mins in Swansea isn't very much when you have 19 if you change earlier.

I have made an update - the 8.15 HST from London Pad only goes as far as Cardiff. So, it would make sense to keep passengers heading further west on the HST as far as Cardiff rather than telling them to cram onto the ATW service (which has come from Manchester via the Marches) at Newport.

The time for changing trains at Cardiff in this case, is 20 minutes and would be on the level between Platforms 4 & 3.

 

[DarloRich]

some of it is to do with trying to get passengers to change at less crowded stations or to offer easier connections between trains

 

[iantherev]

The interchange from the 07.30 from Paddington to Paignton to the 10.18 to Penzance is cross platform but while it could equally be done at Newton Abbot, the Penzance train will be in and waiting at Exeter so no waiting on the platform.

 

[SPADTrap]

Who's to say where the train you're changing to will be busier, at the NRE advised station on the last opportunity? You might get a seat if changing when advised? If you change at the final opportunity you might not?

 

[heart-of-wessex]

Depends if getting on later would mean loosing a seat though, that's what I think sometimes when the journey planner suggests getting off at Swindon then changing for the Westbury service via Melksham. Both the HST and unit calls at Chippenham, which is a same platform interchange, but Swindon involves walking from platform 4 to the bay platform 2.

However as it's a 153, depending on the time of day, I'd change at Swindon just to get the seat I want

 

[Envoy]

some of it is to do with trying to get passengers to change at less crowded stations or to offer easier connections between trains

Well, here is another example:>
PAD > TENby departing PAD at 14.45 on a GWR HST. National Rail website tells customers that they will arrive Newport at 16.30 where they should change to a Manchester to Tenby (ATW) train - which will no doubt be 2 or 3 coaches & pretty full. This train will depart Newport at 16.40 - so they have only 10 minutes to make the connection.

Meanwhile, they could have stayed on the HST all the way to Swansea to arrive at 17.43. The aforementioned ATW service from Manchester to Tenby departs Swansea at 18.14. So, the customers could have stayed on the HST and thus freed up space on the shorter ATW train between Newport & Swansea. What's more, they would have had a much longer time for the change at Swansea than the 10 minutes at Newport - which would have been more stressful. Had the HST been late, they could have got off at Newport & missed the connection altogether whereas, had they stayed on the GWR train to Swansea, they would have had more chance of catching the Tenby train.

I think this all boils down to the computer system just selecting the first point at which a change can be made - even though it is not in the best interests of the passengers or the TOC's.

 

[SPADTrap]

Maybe the system could show all possible change points making it clear which is the last opportunity?

 

[greatkingrat]

The problem is the journey planner has no way of knowing what trains are busier, so it would be very difficult to always pick the "right" interchange point without a lot of manual intervention.

It also varies according to the booking engine, for example WebTis advises a 18 minute change at Cardiff instead of the 10 minutes at Newport.

 

[PHILIPE]

The reason is purely and simply the how the Computer programe is set up. Individual TOCs would not be able to do anything about it as it is a National Rail
programe.

 

[D6975]

Newton Abbot would be a same (or level) platform change.

Not necessarily. The last time I did Bristol-Paignton, the connection at Newton for Paignton departed from the up platform. I had to go over the footbridge.

 

[notlob.divad]

The reason is purely and simply the how the Computer programe is set up. Individual TOCs would not be able to do anything about it as it is a National Rail
programe.

Agreed,

If you read up on the way computers are programmed to calculate shortest path algorithms I think you will find the answers as to why this is done. It also produces anomalies like giving you different routes on the way out and the way back. For example, anyone coming down the WCML from the north bound for Oxford will be told to change at Wolverhampton, yet on the return leg they told Birmingham New Street. It isn't a function of the type of train, it is a function of the way the algorithm works. Plus if you aim to change at the first opportunity and miss it, you always get another chance.

 

[D1009]

Does anyone seriously believe the booking systems are clever enough to know which trains are likely to be busy?

 

[neonison]

The reason is purely and simply the how the Computer programe is set up. Individual TOCs would not be able to do anything about it as it is a National Rail
programe.

I am certain that this will be the case. Having to include all possible changing points would make it difficult to understand for many casual users.

 

[tsr]

I was always taught to advise passengers to change trains at the station with the best staffing/facilities if more than one option is available, and if there is little difference, to go as far as possible before changing. Assuming valid connection times and so on, I think this is a common-sense approach.

Here is a fairly basic worked example: if you are travelling from Uckfield to London Victoria, your train from Uckfield will terminate at Oxted, East Croydon or London Bridge. However, at virtually all times of day bar the early morning and late evening, you have a different range of stations at which to alight, as for Victoria you need to change at Hurst Green, Oxted or East Croydon. The problem is that Hurst Green is unstaffed in the evenings and offers very rudimentary waiting facilities. If your train is only going to Oxted then it is best to change at the destination, as that is staffed first-last train and has more space to wait. If your train goes to East Croydon, carry on until then, as there are more trains to Victoria from there, plus it has many more facilities in which to wait, get refreshments if you are delayed, etc. Despite a potentially short connection at Hurst Green or Oxted, should anything go wrong with the Victoria service following the London Bridge one, you will have a long wait somewhere, either on a platform or on a delayed stopping train. Whilst East Croydon is windswept, busy and so on, it has more realistic onward connections. Unfortunately, and this is where the emphasis for this thread lies, most journey planners do not offer a change at East Croydon, but instead rely on a theoretical "best-case" scenario of a same-platform interchange at Hurst Green within a few minutes, which relies on everything running well on a line where trains can't overtake each other and the infrastructure for service recovery is fairly limited. So in this case as well as many others, journey planning software does not appear to look at the best and safest place to wait, simply because ideal timings appear favourable.

 

[Greenback]

Despite my dislike of Newport sometimes it's better to change there. Looking at the Paddington to Tenby example, I feel that there's more chance of getting a seat on the ATW at Newport than at Cardiff, which can be a total scrum at 1640.

Then we see the personal preference come into things as well. While taking the HST to Swansea might be a decent enough option for most people, a lot of those would probably prefer being on a nice warm train instead of hanging around Swansea station for half an hour (although the waiting room is nice and warm even in the coldest weather!)

The morning Fishguard train from Cardiff is usually pretty quiet. I'd rather change at Cardiff to get that one. Others might feel differently. In the end, this is only a computer program, it can't think like a human being can!

 

[Llanigraham]

Does anyone seriously believe the booking systems are clever enough to know which trains are likely to be busy?

Of course they aren't, and most sensible people know that, but any excuse for a good (pointless) rant!

And in ATW land, quite often the train staff will suggest the best places to change, dependent on the ability of the passenger, the time of day, and even the weather!

 

[Greenback]

And in ATW land, quite often the train staff will suggest the best places to change, dependent on the ability of the passenger, the time of day, and even the weather!

Indeed they do, and as you rightly say, there are far more factors involved than the type of rolling stock.

 

[ComUtoR]

Is the booking engine mandating the journey. If I believed I knew better, am I allowed to simply ignore it ?

0730 PAD
1010 EXD
8 mins wait
1018 EXD
1052 TOT
(3hr 22m)

0730 PAD
1034 NTA
6 mins wait
1040 NTA
1052 TOT
(3hr 22min)

Is there a minimum connection time at Newton Abbot ?
8 minutes connection is better than 6 minutes
Getting on at an earlier point means you have less passengers boarding. Ergo, more of a likely-hood to get a seat.

I think the computer wins that one to be honest.

The sensible option

0706 PAD
1006 TOT
(3hrs)

No changes.

 

[ScotTrains]

A similar thing happens when traveling from Glasgow Central to Crewe on the 1840. It tells you to change from a pendolino to a vomiter at Carlisle even though both trains stop all the way down as far as Warrington (with no platform change). When traveling first class it is preferable to stay on the (quiet) pendo with full hot dinner service than change to a (busy) voyager with snacks. To be fair the staff have always said it is fine for me to change beyond Carlisle if I want.

 

[glbotu]

I think the issue is a non-issue. If you assume the computer has no pre-loaded "loading" data, then it has to come up with the most sensible solution a greater proportion of the time. As such, it makes sense to get you to change early.

Railway services tend to go from major conurbation to major conurbation. I would suggest that in most*** cases that is true.

In that case, rail services tend to get fuller as they approach their terminus. This is a function of the fact that we still, for the most part, operate services into large terminus stations in cities, rather than "Crossrailing" the services across them, to smaller destinations. Even then, you're more likely to get a seat on a Crossrail train at Maidenhead than you would at Paddington (although I realise that Crossrail is an awful example).

That is in no way true for all railway services, as the many examples shown here highlight and I'm sure I can think of a bunch of others (Stansted to Birmingham - massive loading Stansted - Ely, much lower as you head out Birmingham-wards).

While a solution that looks at either loading data, or has more of an understanding of the journey patterns of each service would obviously be better, it's almost certainly not good value for money to develop. It would at least double the complexity of the algorithm (or even quadruple it), for the benefit of only people making long distance journeys with significant overlap in routes, with a non-typical route loading distribution.

***To be precise, > 50% of cases, because that's the limiting case.

 

[D6975]

I think the issue is a non-issue. If you assume the computer has no pre-loaded "loading" data, then it has to come up with the most sensible solution a greater proportion of the time. As such, it makes sense to get you to change early.

Railway services tend to go from major conurbation to major conurbation. I would suggest that in most*** cases that is true.

In that case, rail services tend to get fuller as they approach their terminus. This is a function of the fact that we still, for the most part, operate services into large terminus stations in cities, rather than "Crossrailing" the services across them, to smaller destinations. Even then, you're more likely to get a seat on a Crossrail train at Maidenhead than you would at Paddington (although I realise that Crossrail is an awful example).

That is in no way true for all railway services, as the many examples shown here highlight and I'm sure I can think of a bunch of others (Stansted to Birmingham - massive loading Stansted - Ely, much lower as you head out Birmingham-wards).

While a solution that looks at either loading data, or has more of an understanding of the journey patterns of each service would obviously be better, it's almost certainly not good value for money to develop. It would at least double the complexity of the algorithm (or even quadruple it), for the benefit of only people making long distance journeys with significant overlap in routes, with a non-typical route loading distribution.

***To be precise, > 50% of cases, because that's the limiting case.

In rural areas like Devon trains can often get quieter as they approach their destination. Exeter to Paignton in the summer is one example, heaving from Exeter, but not too bad from Newton, having dumped many pax at Dawlish stations and Teignmouth.

 

[DelW]

Then we see the personal preference come into things as well. While taking the HST to Swansea might be a decent enough option for most people, a lot of those would probably prefer being on a nice warm train instead of hanging around Swansea station for half an hour (although the waiting room is nice and warm even in the coldest weather!)

And there's an excellent and very good value café just over the road.

Maybe booking engines should also be programmed with which interchange points have a pub or café handy :roll:

 

[Greenback]

Local knowledge can be very important, though I wouldn't rely on the Espresso bar being open at 1743 when the HST in our hypothetical Tenby journey pulls in!

If a passenger doesn't have any local knowledge, and can't find anything out, then the booking engines can only reflect the way they've been programmed. And there's no way to prgram the best changing points because what is the best will be different for every passenger, because there are so many factors involved.

 

[Paul Kelly]

ATOC data gives each station a status as a small, medium or large interchange. If there is more than one possible change station in a journey, then they are supposed to prioritise changing at the largest interchange. In the Paddington to Totnes example, both Exeter St Davids and Newton Abbot are rated as Medium interchanges and the change times at both (8 and 6 minutes respectively) meet the minimum connection times (6 and 5 minutes respectively), so booking engines are allowed to change at either of them.

I think most booking engines will choose Exeter because there's a longer time there (8 minutes compared to the 6 minutes at Newton Abbot). They generally don't have knowledge of how busy trains are - although it would certainly be possible to maximise time on an "Express" train over an "Ordinary" train. This would solve the Paddington to Totnes issue at least.

 

[TheNewNo2]

I used to have a similar situation with my Nottingham to Bristol journey. There are two services involved - a 170 from Nottingham to Birmingham or Cardiff, and a Voyager/HST from the North or Manchester to Bristol. The standard change points are therefore Derby, Birmingham and Cheltenham.

NRE would recommend changing at Birmingham, but my preference was always to change at Derby as I was more likely to get a seat. That said, Birmingham is the change you would make between the NOT-BHM and MAN-BRI, so it does make a sort of sense to apply that to NOT-CDF and GLC-PNZ.