New 'On-time' targets proposed

[david.hill64]

'The Times' carries an article this morning suggesting that DaFT is thinking about a new performance regime that would penalise TOCs if a train is more than 1 minute late arriving at its destination. Now 'The Times' has had it in mind for a long time that this sort of measure is required so there may be an element of wishful thinking about it. However, it seems to me to be an entirely impractical target that is likely to result in longer journey times (with more margins built in) and fewer services.

 

I was on a Derby to London service recently that was delayed by a requirement to get two disabled passengers on-board in Derby and detrained at Leicester. If there had been only one, then I think the on-board and station staff might have managed it within the normal dwell time, but two on a severely crowded train gave them no chance. As a result of our subsequent late running there could well have been consequent delays to other services. To punish the TOC for having the audacity to carry more than one person requiring assistance seems to me to be wholly unjustified. Perhaps there will have to be a greater army of 'delay attribution executives' to determine what happened and justify why no penalty is payable.

 

I can also imagine other techniques to claim on-time performance. While working for the Taiwan High Speed Rail project, some colleagues and I had a trip to Japan to study various aspects of the Shinkansen system. The Shinkansen has an enviable reputation for on-time performance, but on one trip we weer slowed by heavy snow and arrived at our destination 40 minutes later than the time that was the advertised arrival time when the train departed. However, the train was still 'on-time' as during its journey, the railway had invoked an emergency timetable. It seems that this is the standard way of dealing with out of course events: if things are not working properly you change the timetable. I wonder whether we will have to resort to such subterfuge.

 

It might just be better to spend the effort getting to the root cause of delays and dealing with it.

 

 

[Zomboid]

Whilst it would be admirable to achieve on time to the minute at the end of a journey, I'd say it's more important to get intermediate stops at the right time, rather than building in a big margin at the end of the journey.

I think its pretty good that you can have 10 minute accuracy on a journey like Inverness to London, though no doubt there's a long wait in the timetable at Edinburgh on that one. Weymouth to Waterloo within 5 is arguably more impressive...

[david.hill64]

Whilst it would be admirable to achieve on time to the minute at the end of a journey, I'd say it's more important to get intermediate stops at the right time, rather than building in a big margin at the end of the journey.

I think its pretty good that you can have 10 minute accuracy on a journey like Inverness to London, though no doubt there's a long wait in the timetable at Edinburgh on that one. Weymouth to Waterloo within 5 is arguably more impressive...

So to get right time at intermediate stops you build in 5 minute dwell times at every station, and suddenly your Sheffield - London service takes 15-20 minutes longer than before: hardly good for business. I agree that right time is important but I think targets such as this will promote undesirable consequences.

[cromptonnut]

Currently "on time" is "less than 5 minutes late".

 

This website breaks it down by company, and you can see how impractical a suggestion it is for everything to run exactly on time unless, as stated above, 'padding minutes' get added to extend journey times.

 

http://trains.im/ppm/

[The Stationmaster]

Having over the years had more than a little to do with passenger (and freight  ) service planning, timetabling, train performance, and punctuality I do seriously wonder about the practicality of such targets?  On the WR London Division we counted '1 late' as right time for the peak period stats but back then it was a less crowded railway with fewer trains and fewer passengers getting on/off them and they are just a couple of the elements which affect performance.

 

Back in the days of Post Office mails using passenger trains it was noticeable in many places that if PO traffic was light, or Postmen were on strike, then performance improved significantly.  The reason was fairly simple - British dwell times tended to be quite tight - especially when compared with most mainland European railways where the opposite was the case and slack dwell times plus advertising public departure times in advance of working book times created further opportunities for lost time to vanish.  In other words slack times have their benefits on performance figures.

 

The other British method, of providing Recovery Time in longer distance and loco hauled trains also could offer benefits as there was always a lump of several minutes on the final stretch to the destination - not a big 'ump' (only 2 -4 minutes normally) but still a handy amount.

 

Things started to get stupid when the Passenger's Charter came in (and not just stupid in the punctuation) as the immediate reaction was to slacken times by shoving in what was known as a 'Performance Allowance' (referred as 'diamond time' in some places as it was shown in a 'diamond' symbol in the WTT.  In reality this reduced timetabling to a farce in some respects as the time was usually dumped in the final stage of a journey thereby reducing line capacity and turning margins at the terminus and any key junctions into a joke and a planning nightmare.

 

The simple answer is just to apply a sensible Recovery Time (there for engineering speed restrictions and properly related to their time cost) plus Pathing Time if required approaching junctions etc and otherwise just use the basic running times - that is the most efficient way to write a timetable - tried and tested over many years and known to work effectively.  The running time is in any case an average of averages allowing for minor variations in train performance, the weather (except extremes thereof) and so on.  The only awkward decision then is in respect of dwell times but ideally they should reflect what happens at each station at the various times of day although that does become a rather excessive sophistication of timetabling process.

 

But that leads to one very big headache for 'somebody' - if relatively pure timetabling rules are applied in that manner and the network is used at or above maximum theoretical capacity (as it is in a lot of Britain) then the slightest thing, such as one or two disabled passengers, or people in the wrong part of the platform if a train is in reverse formation, or a train slightly under par performance wise, will hit timetable performance and, on a busy railway, lead to reactionary delays which almost inevitably will snowball into ever worse figures unless a train is taken out to break the chain.  The simple fact is that a heavily used railway working near to, or in excess of, its theoretical capacity will not be a reliable railway without a lot of timetable padding (which has its own ill effects anyway).

 

To impose a '1 Late' target on such a railway is in reality a nonsense in much of today's NR situation.  But then of course Govt would never accept the UIC fiche on capacity calculation which our continental colleagues have used for years to beat their Govts over the head to get extra track capacity.  If you want that sort of performance you have to provide the infrastructure to do it and that, regrettably, has long not been the British way.

[The Stationmaster]

Currently "on time" is "less than 5 minutes late".

 

This website breaks it down by company, and you can see how impractical a suggestion it is for everything to run exactly on time unless, as stated above, 'padding minutes' get added to extend journey times.

 

http://trains.im/ppm/

What could be very revealing would be to see those stats on the old BR measure where the categories were 'Right Time', 1-5 late, 6-10 late, 10-30, and over 30 late.

[Chris116]

The figures already lie! One of the Cardiff Valley lines the train is timed to take 10 minutes between the last two stations at the head of the valley but only 2 minutes in the reverse direction. What is even worse is that the departure time is actually before the arrival time at the terminus and this is at a single platform station on a single track line!

[Edwin_m]

I suggest there are at least two issues here.

 

For the railway to operate properly, trains need to be within a minute or so of their planned times on busy routes, otherwise they miss their slots and delay other trains.  In the past and to some extent today, all that has mattered to some managers within both Railtrack/NR and the TOCs has been the "charter figure" of 5 or 10 minutes late, but I believe this should relate only to passenger experience as I discuss below.  As someone pointed out in Modern Railways recently, a southbound East Coast presenting 9min late at Hitchin or the Welwyn bottleneck will probably be within Charter time at Kings Cross but will have delayed several other trains possibly beyond their own 5min Charter figures.  So there is some case to incentivise trains being within 1min of time at selected operationally critical locations. 

 

The basis of the Charter figure was someone deciding that passengers would count arrival within 10min for a longer journey or 5min for a shorter one as being acceptably punctual.  This possibly springs from the tendency to treat railways like airlines, where that sort of punctuality is considered pretty good.  But 10min is longer than the connectional allowance at many stations so a 10min delay can often be turned into a much longer one on an onward journey, which can also trigger Delay Repay compensation.  I would suggest the Charter threshold needs to be reduced, perhaps to 5min for all services, but need not be as low as 1min. 

[seahorse]

Can we have the same for buses?

 

And should there be a standard time for all road journeys with fines or compensation for delays?

 

Why is there such an expectation for rail?

 

Everybody knows E=MC squared, even if they don't know what it means. Everybody needs to know that queue time = average service time divided by 1-density. which approximates as 70% load gives queues, and a 10% increase in demand gives a 20% increase in queue - we think that if something is "only" 98% full it is inefficient and we should be able to squeeze more out it. 

 

If we insist on 99% punctuality, then the means must be assigned - among other things  timetables must be altered to be 99% achievable - longer station times, more recovery times, more new lines, take freight off passenger lines, reduce number of station stops, leave passengers behind once station time is reached, more "Norton Bridge"  type flying junctions, and a LOT more money

[Keith J]

Currently "on time" is "less than 5 minutes late".

 

This website breaks it down by company, and you can see how impractical a suggestion it is for everything to run exactly on time unless, as stated above, 'padding minutes' get added to extend journey times.

 

http://trains.im/ppm/

 

Slightly off topic but I think these stats are completely meaningless. In my view it’s not the number of trains that are late that matters it’s the number of passengers affected and whether the train is a commuter train or not. For example if a Monday morning peak time train that is packed to the rafters is delayed the number of passengers impacted is far great than say a train travelling around lunchtime. Also, a peak time train will be full of passengers commuting therefore the number of business hours lost is also far greater than when the passengers are mainly travelling for other reasons like visiting friends or relatives etc. which is often the case with off peak trains. Additionally the amount of compensation that would need to be paid out through delay repay would be significantly greater for peak time trains.

I was looking at the stats for Abellio Greater Anglia where 97% of trains are on time. If the 3% of trains that are delayed are all peak time trains then the number of passengers affected would be far greater than 3%.

[The Stationmaster]

Can we have the same for buses?

 

And should there be a standard time for all road journeys with fines or compensation for delays?

 

Why is there such an expectation for rail?

 

Everybody knows E=MC squared, even if they don't know what it means. Everybody needs to know that queue time = average service time divided by 1-density. which approximates as 70% load gives queues, and a 10% increase in demand gives a 20% increase in queue - we think that if something is "only" 98% full it is inefficient and we should be able to squeeze more out it. 

 

If we insist on 99% punctuality, then the means must be assigned - among other things  timetables must be altered to be 99% achievable - longer station times, more recovery times, more new lines, take freight off passenger lines, reduce number of station stops, leave passengers behind once station time is reached, more "Norton Bridge"  type flying junctions, and a LOT more money

 

Alas it doesn't actually work like.  I would suggest that in practice, and certainly when written, all timetables on the national network are achievable - in fact they will breach what are nowadays called 'Timetable Planning Rules' if they are not achievable and should not therefore be published.  But there is a difference between achievability and reliability.  

 

Achievability is planned on the basis of everything working as it should - and logically there is no other way you can plan anything as complex as most railway timetables unless you are going to waste capacity or resources or both.  But reliability is a very different kettle of fish as it relies on everything working exactly as it should - and that includes the passengers if it is a passenger or mixed traffic timetable and if the line has any level crossings it also includes the motorists using those level crossings.  The only way you can really improve reliability is to build in resilience by using either surplus infrastructure, or trains (as do SNCF for example), or ensuring through its design and maintenance that your infrastructure and train resources are absolutely guaranteed 100% reliable - all of these things mean spending more money.

 

Taking away freight trains makes no difference at all as long as they perform reliably - their presence only becomes a problem if people want to run more passenger trains and aren't prepared to pay for the additional infrastructure they need (think Crossrail on the GWML where as it happens most freights are faster overall than stopping passenger trains).  Reducing the number of station stops reduces journey times, moves demand to other trains (assuming the passengers are not deterred) and reduces the opportunity for station overtime on some trains - but it will be just another part of the plan and still subject to whatever other influences can undermine reliability.  More flying junctions basically increase capacity and they do reduce the opportunity for reactionary delays but apart from the latter they inject no greater reliability into the basic plan so their use needs very careful consideration against their cost.

 

But I'll tell you one thing which has a better impact on reliability than most other things - no passengers, works like a dream.  And within reason dwell times make little difference in terms of station overtime - it's the way passengers react at stations (and getting ready to alight at them) which makes the difference.  Dwell times are, I suspect rather like the NHS in that you can add more time (money in the case of the NHS) and you'll still have problems with station overtime(overspend) - so yes dwell times need to be passenger numbers related but that is not the whole of it when it comes to reliability.

[stewartingram]

Slightly different topic, but just a few comparisons with the Underground. I often think that in many situations the Underground can be considered as the "real" railway, with National Rail just playing, whereas it is usually thought of the other way round (and no offence intended to either party).

Some lines are basically simple such as the Victoria; others are complex routes with junctions, such as the Northern. Just take the Victoria as an example for now. 16 stations end to end, and actually all underground. Most trains run end to end, a few turn back short such as at Seven Sisters, which is the junction for the depot at Northumberland Park, and many of these form staff trains to the depot, the others either stabling or using the depot as a turnback. Should be a doddle to operate then?

Train frequency at the moment is 33tph, impressive. Roughly speaking that is a train in every platform north to south at the same time, and the same south to north. Dwell times are around 30 seconds but vary for each platform due to passenger loading. It can be seen that a delay at any platform will impact on all trains behind it AND all trains in the opposite direction. So then the Line Controller intervenes from within the Control Centre at Northumberland Park. It is commonly not realised that the Underground does in fact run to a timetable, and each train carries a reporting number on the front. If trains are amended, they often change reporting numbers and take over a different timetable service, and if the Controller intervenes, may be turned back early. Yes it does actually make the stats better, but the prime aim is to keep the railway running for the passengers. A delayed train not only impacts on the following trains, but also on the stations where platforms, and further up, escalators and ticket halls, can be dangerously overcrowded. Incidentally, in a very recent RAIL, I read that the working timetables are written to within 10 seconds, with a train scheduled every 100 seconds!

Now the impact on all of this to the public is that with a delayed train (lets say a faulty door, so tip the passengers and run it ECS round the line to take it off the railway at Seven Sisters and on to the depot), is that the passengers will be delayed until the next train is allowed to come in. I think the times that LU quotes is MINOR delays up to 5 minutes, then SEVERE delays after that. But think about it? train 123 fails, and tips then moves off. Train 124 behind it is thus delayed, and 125 behind that. Once 123 has moved on, 124 arrives, followed by 125, 126, etc 100 seconds behind one another. So for Joe Public, service is restored? However, for the operators delays still occur until a specific train (call it 234) actually runs on time to the Working timetable! (There are operational implications such as crew rostering, stabling trains at the right place, maintenance at end of duty etc). LU actually keeps informing its passengers customers there are delays even though they are back to running at 100 sec intervals!.

 

Edit to clarify Delays:

 

Minor - 5-10 minutes from scheduled time

Severe - 10 minutes or more.

I believe these are now correct, though if someone feels the need to correct.....

It is also relevant that the Vic line is fully automated and with in cab signalling; the driver (Train Operator) actually only operating the doors and starting the train in normal service, leading to very efficient headways. However on other lines manual driving, and trackside signal are still used.

 

Stewart

[JDW]

I agree with most of the above posters, in that while its important operationally for trains to run on time, especially on a complex network, penalties for relatively inconsequential delays of a minute or so seems rather silly, even more so on quieter lines.  Anything that involves the general public and the weather, and any number of other external factors, can never be perfect all the time.  I work for a bus company, and we often had problems in certain places heading into the city at rush hour, so we revised the timetable to allow for the average longest time taken to do that stretch of road.  On a quiet day, we can cover the 5 stops in around 4-5 mins.  The schedule allowed 5, but if traffic was queuing, it could be ten or more, so ten minutes were added, to all daytime journeys.  That meant that most buses were on time.  But we had to sit at every stop for a minute or two, with people moaning that the service was slow.  The company pointed out that people had been complaining it was always late, or rather not running according to the schedule, and so now, the service is more reliable because it generally arrives at the advertised time.  Thankfully, the ten minutes was soon reduced to 7, which is just about right.  

[John M Upton]

I can tell you that in the murky argumentative bun fight world of delay attribution, any delay caused by the boarding/disembarking of a disabled passenger (and they don't necessarily have to be in a wheelchair, a little old lady needing help on/off also counts) is not chargeable to the relevant TOC.

[Reorte]

I agree with most of the above posters, in that while its important operationally for trains to run on time, especially on a complex network, penalties for relatively inconsequential delays of a minute or so seems rather silly, even more so on quieter lines.

That's why I'm wondering who this is supposed to benefit; as you say operationally even small delays can cause big problems but if it's aimed at that then any penalties should surely be directed at the trouble caused (a minute late at New Street taking a bigger one than a minute late at Kyle of Lochalsh). From a passenger point of view a delay of a minute is neither here nor there (not that it'll stop some people from grumbling against it). Even five minutes shouldn't really make a difference practically, if it does you've probably organised things far too tight anyway (I assume I'm just taking a risk if I allow less than ten for a connection), although I'll probably frown a bit when waiting for my train. The passengers who do grumble about it though almost certainly won't be planning any car journey they make to anywhere near that precision.

[roythebus]

When I was a driver at Waterloo in the early 1980s we regularly achieved something like 98% RIGHT TIME, yes, actual right time, not up to 5 minutes late, on a regular basis. and that was with slam-door stock, Westinghouse or EP brakes, none of this modern technology! And we the drivers were quite proud of this record.

[IWCR]

Just  working  on  times  is  not  helpful,

My  local  railway  meets  a  ferry  and  most  of  the  business  comes  from  that  ferry.

I  have several  times  been  on  the  ferry  which  is  late  and  witnessed  the  train  leaving  empty,  (if  they  are  late  they  are  fined),  on  one  occasion  I  and about  three  others  made  it  onto the  train,  the  remaining  50+  passengers  arriving  on  the  platform  having  the  doors  closed  on  them,  Surely  the  service  requirement  should  be  to  make  the  connection.

Yes  an  unusual  situation  and  the  train  still  needs to  be  back  on  time  to  make  the  connection  for  a  ferry  sailing  but  some  leeway  should  be   possible.

 

Pete

[Classsix T]

Just working on times is not helpful,

My local railway meets a ferry and most of the business comes from that ferry.

I have several times been on the ferry which is late and witnessed the train leaving empty, (if they are late they are fined), on one occasion I and about three others made it onto the train, the remaining 50+ passengers arriving on the platform having the doors closed on them, Surely the service requirement should be to make the connection.

Yes an unusual situation and the train still needs to be back on time to make the connection for a ferry sailing but some leeway should be possible.

 

Pete

That sounds like a case of being hoist by your own petard Pete.

 

If I was in the position of having to integrate those two forms of transport, I'd have them departing at opposite ends of the clock face (if hourly, natch). Then there's only half an hour wait for either the boat or train plus enough leeway to negate a few minutes delay to either.

 

C6T.

[JDW]

That's why I'm wondering who this is supposed to benefit; as you say operationally even small delays can cause big problems but if it's aimed at that then any penalties should surely be directed at the trouble caused (a minute late at New Street taking a bigger one than a minute late at Kyle of Lochalsh). From a passenger point of view a delay of a minute is neither here nor there (not that it'll stop some people from grumbling against it). Even five minutes shouldn't really make a difference practically, if it does you've probably organised things far too tight anyway (I assume I'm just taking a risk if I allow less than ten for a connection), although I'll probably frown a bit when waiting for my train. The passengers who do grumble about it though almost certainly won't be planning any car journey they make to anywhere near that precision.

 

Yes, I've suggested something similar in the past, to those who tut and moan at every minute or two of a delay - try driving the same route every day by car, at different times of the day, different times of year, and timing your journey.  Try to do it in exactly the same time every time.  And not just end to end, but passing the same waypoints at the same times too.  Regardless of weather.  Or the football match that's on.  Or helping some lost tourists, waiting for an elderly woman or two, and having to all funnel in to one lane because a car has broken down in lane 1.  Good luck.  Of course, we could always have a different timetable for the days when its snows, rains, every second Thursday when slow old Mrs Jones goes shopping, the days when a car breaks down on a busy roundabout, and if we published them all in advance that would solve the problem...

[Edwin_m]

If you're driving you don't usually have to arrive at a particular place in time to catch a connecting car, or miss it and have to wait a long time for the next one. 

[Reorte]

If you're driving you don't usually have to arrive at a particular place in time to catch a connecting car, or miss it and have to wait a long time for the next one. 

Sometimes you do - driving to the station is one obvious relevent example. I just don't think that a minute late and you'll miss your connection is reasonable (and I'd hate to be in such a rush that I need that even when everything is on time, far better to have a slightly longer but less stressful journey). Miss a connection because you're five minutes late and it's reasonable to grumble about the railway, miss one because it's one minute late and you should grumble about yourself.

[The Stationmaster]

I know we live in an age of digital just about everything but I wonder how many rail passengers have watches etc that are so absolutely accurate that they can tell if a train is one minute late or even 5 minutes late?

[Reorte]

I know we live in an age of digital just about everything but I wonder how many rail passengers have watches etc that are so absolutely accurate that they can tell if a train is one minute late or even 5 minutes late?

Plenty (probably most) stations conveniently tell you. But it's a good point. I'd expect most peoples' watches are set to within five minutes' accuracy but probably not 1 (I'm guessing smartphones will be accurate by using a time server).

[Wheatley]

Depends where you set the target of course. If you set the Right Time target at the same level as the PPM (<5mins) target then you'll fail. If you set it lower you won't. Northern has been working to Right Time for a while now (since Alex Hynes took charge in fact) and is currently operating at about 75% RT and 90%+ PPM. From April this year both measures are published.

[Rivercider]

If you're driving you don't usually have to arrive at a particular place in time to catch a connecting car, or miss it and have to wait a long time for the next one. 

If you are driving to take the kids to school, or pick them up afterwards there is a deadline.

My mother lives a few doors from a primary school and I can assure you the standards of driving/parking,

and the choice language that occurs from time to time from  late running drivers should not be heard by children!

 

cheers