Keeping the buses on time

by Kerry Wood
The Regional Council recently claimed that its buses run 99.9% on time, but only 99.7% run at all. There was nothing to define ‘on-time’, but it seems to be starting, or maybe completing, a run within ten minutes of scheduled time (or both?). The figures suggest that ‘on-time’ includes ‘early’.

By comparison, Auckland Transport claims that 95% of buses are within five minutes of time, but admits to much lower performance on some routes. This is nominally worse than Wellington but definitely more realistic.

Information from transport economist Neil Douglas is that Sydney, Melbourne and London all consider buses to be ‘on-time’ if they are within a band from one or two minutes early to five minutes late. Zürich takes a much stronger line―‘on time’ is 30 seconds early to 90 seconds late―and achieves it.

GW’s Public Transport Plan explains the importance of timekeeping, if not the specification:

Surveys and research show that the most important consideration for public transport users―and potential users―is reliability; in other words, that a trip leaves on time and arrives at (or very close to) the scheduled time. This is particularly important when trips require connections with other services.

However, timekeeping remains inadequately measured, inadequately displayed and complacently recorded.

A special difficulty in Wellington is the combination of an overloaded route in the central city, which makes most routes unreliable, and infrequent feeder services. Under these conditions feeder-service drivers may need real-time information on the core-bus they are to meet, at least as much as for their own bus. On the plus side, introducing feeder services will itself reduce inner-city bus delays.

New schedules, to be introduced next year, will have more passengers making connections as part of their daily journeys, but is GW ready to introduce reliable connections? Probably not. Nothing has been done, and identifying and sorting out delay-points takes time: about 20 years to perfect the Zürich system.

A worrying oddity is that buses on the golden mile often arrive later than shown on ‘real-time’ displays. A recent example was arriving at the Arty Bees stop to see my bus due in 12 minutes, but it took 18 minutes. This happens consistently. ‘Real-time’ seems to creep by around 50%: buses generally arrive one minute later than originally shown, for every two minutes of waiting time. This suggests errors in the method of calculating ‘real-time’.

Early- or late-running is easily calculated, by comparing actual time with a theoretically perfect timetable, corrected for expected delays on any given day of the week or time of day. But is this what GW does? and if so, why does real-time creep? Of course, the risk of delays is greater than the opportunities to make up time, but realistic predictions should show a consistent spread of delays.

A second problem is buses that appear on real-time displays, then vanish when the scheduled time arrives. Presumably the bus has been withdrawn ― for whatever reason ― but the real-time system seems happy to track ghost buses. Would this problem improve if operators were not paid for running buses that vanished without explanation?

An effective improvement programme might include these points:

• Set a new timekeeping target, such as recording a bus as ‘early’ if it leaves a stop more than one minute before theoretical departure time (simple enough with GPS already available), or ‘late’ if it arrives more than five minutes after theoretical arrival time.
• Set a timetable for achieving say 80% of the new target times within five years, with targets and programmes set and reported for each year.
• Ensure that real-time displays receive realistic data.
• Measure timekeeping at more points than the end of each route: say every hub and every third other stop.
• Use the GPS/real time system to locate and prioritize regular delay points, and draw up a improvement programme.
• Introduce easy-to-remember ‘clock-face’ timetables and ensure that interchange timetables match. A core route running every ten minutes can meet a feeder route running every 20 or 30 minutes, but not 40 minutes (cannot run every hour) and not 15 minutes (missed changes).
• Introduce summer and winter timetables, and use the regular change-over to update timetables as delay-points are improved. Route improvements are little use until they appear in timetables.
• When the ‘on-time’ target is reached, GW may wish to halve both ‘early’ and ‘late’ figures and start again.

This seems a costly exercise, but failing to invest can also be costly. Subsidies per passenger boarding in Wellington are four times higher than on Zürich’s superb system, and the reason is very high patronage in Zürich. Public transport can compete with cars.

 

13 comments:

  1. Anabel, 25. May 2017, 6:49

    The “real time: system was a big waste of the ratepayers money.
    Now when buses are late we are just told by the electronic screen that they are all on time, how crazy is that?

     
  2. Neil Douglas, 25. May 2017, 8:23

    Excellent ideas Kerry, I hope the GWRC takes onboard your sensible suggestions to define, collect and report reliability statistics that the public can relate to and which the bus operators and the Council can monitor and improve their performance.

     
  3. Michael Gibson, 25. May 2017, 8:37

    Kerry Wood rightly questions the statistics about buses being on time. When I last asked, ‘on time’ meant that a bus had left the starting-gate within ten minutes of schedule. Regrettably this unacceptably slack standard was not mentioned in the report in question and Greater Wellington’s elected members have not insisted that such reports should be helpful and accurate rather than simply self-serving propaganda. I am very pleased at Kerry Wood’s continuing efforts on behalf of all of us.

     
  4. Neil Douglas, 25. May 2017, 9:28

    Your recording of comments is unreliable today! I note that the heading said 2 comments then when I went into the comments section there weren’t any and the number went to zero. A bit like some of Wellington’s disappearing buses? [All fixed now, we hope.]

     
  5. David Mackenzie, 25. May 2017, 9:35

    “Due” appears to mean “about to arrive within one or two minutes,” not as one would expect, “arriving at the scheduled time.”

     
  6. Luiz Alves, 25. May 2017, 10:36

    It is very simple why the RealTime system does not work well in comparison to Zurich.

    1 – No real priority to buses on traffic lights and too many of them on the Golden Mile, making them slow down;

    2 – Time to board: In Zurich (and many European cities) you can board at any doors. Many buses have three doors, while in NZ you must board by the front door.

    3 – They don’t accept money onboard, just tickets and passes (now all electronic).

    4 – More and reliable information on bus stops showing on the map where the routes go. Frequently Wellington passengers have to ask the driver if the bus goes/via somewhere, slowing down the depart.

    Ps. Zurich has a medium size trolley bus system, one of the modern ones which use bi-articulated ones and they are the core of the trams-light rail system. Sadly our Regional Council has decided to dismiss them, when they could and should be the core routes.

     
  7. laidbackchap, 25. May 2017, 10:56

    It’s all smoke and mirrors really. I have recently noticed signage on the trains saying if you think they are not running on time then it means your watch is wrong and you need to synchronize it to the clock above customer services.

     
  8. luke, 25. May 2017, 11:31

    i find the locate function on the metlink website to be quite handy for where my bus actually is.

     
  9. John Rankin, 25. May 2017, 13:29

    A problem with the real-time displays at bus stops is that they claim to predict the future. As Kerry notes, they don’t do this very well. Perhaps it would be better if the display said “1.5km” rather than “10mins”. Stating distance a bus is from my stop is telling me a fact and I can draw my own conclusions. Giving me a prediction, without an error bar, is less useful. It appears that “10mins” means “at least 10 mins” rather than “10 plus or minus 2 mins”.

    A benefit of using distance is that it would also distinguish between buses which are being tracked and ones where the system is using the timetable to report arrival times. As many people have said, prediction is hard, especially about the future.

    When I lived in Edmonton, Canada, the buses had up to 14 timing points along the route. The schedules had a bit of built-in “recovery time” so that a bus could arrive a bit late or early at a major stop, but still leave the stop on time. Many trips required a transfer, so schedule adherence was a big deal. It’s not nice missing a connection and waiting 30 minutes when the temperature is -35C. So the buses ran on time, Zürich style, and everyone trusted that they would, barring major adverse weather events.

     
  10. Luiz Alves, 25. May 2017, 16:03

    @John Rankin. Yes, this is a good idea and even better. As I pointed out in my previous comment, trying to sort out the issues that help to delay the buses, showing the distance should be a good thing for our bus system.

     
  11. H., 25. May 2017, 16:28

    Another issue is that of when a bus is running quite late, it can disappear off the realtime information signs altogether, but then the physical bus turns up at about or a few minutes after the expected (late) time. Interestingly, it happens far more with some bus drivers than others. I think that possibly the bus drivers have an override switch that allows them to turn off the GPS for their bus – and that some do this systematically when running late. Presumably (if they have their individual lateness performance recorded & occasionally commented on) this improves their personal record & makes the driver look better.

     
  12. Ian Shearer, 26. May 2017, 18:14

    Using @JR’s “actual distance away” real time displays would reduce another frustration – full buses that pass us by. Knowing that another bus is 500m behind would sooth feelings.

     
  13. Mike Mellor, 26. May 2017, 23:09

    In my experience (and I use the bus most days), the real time information is generally pretty good – perhaps I’m just lucky! I’m not sure what giving the distance away would add: if another bus is 500m behind it’ll already show up as being (say) one or two minutes behind; if you’ve a smartphone you can already see when it is using the locate function on the Metlink website (as luke says); and one thing that has to happen is improvement of speed and reliability at least along the Golden Mile (as Luiz says), which should make time predictions better. And, worldwide, the standard is to give times rather than distances – I’m not aware of anywhere that does the latter.

    But Luiz is absolutely right about information at bus stops (and his other points). It’s very good that just about every stop has relevant timetables, but bad that few (if any) stops have maps showing the individual routes or the network. Such things are standard at main stops in many cities.

    The need will become even greater next year when the new network comes in, with many current through journeys requiring changes – for instance, the new Karori-Seatoun east-west spine route will have to make connections at half-a-dozen different places to replace what are now one-bus trips. This will also need better timekeeping, and if we haven’t got both that and better information I can’t see how the new network will cope.

    GWRC seem to be putting greater and greater reliance on people’s smartphones, but good old-fashioned paper and signage will be necessary for a long time to come.

     

Write a comment: