Follow us on Twitter

Speed versus Frequency

The new train timetables that started up on September 19th have been running for almost exactly a month now. A major change in those timetables as an increase in frequency for Southern Line and (particularly) Eastern Line commuters, but a corresponding loss of almost all the express trains on the rail network. Gone went the Western Line express (which was strange considering peak time frequencies didn’t increase there and gone went most of the Southern Line peak trains.

Effectively, ARTA made a decision to prioritise frequency over speed. The problem with express trains is that they tend to catch all-stopping trains in front of them, unless you leave a large gap in the timetable or (in an ideal world) if you have dedicated tracks for express services. From next year, when we get up to having 10 minute frequencies at peak times on our three main lines, it seems somewhat unlikely that we will have any express services – because slotting them in would just be too difficult – unless we switched to something like Wellington’s timetable system, where the lines have a number of different stopping patterns, which does reduce the frequency of trains at particular stations.

There is a natural tension between high speed and high frequency, because of the issue of trains catching each other. But I do wonder whether the right approach is to focus completely on frequency at the cost of speed. Do we really need all our trains stopping at all our stations all the time? On the Western Line in particular, the inner section is painfully slow due to poor track geometry – while it also seems that the majority of patronage is from stations west of New Lynn. Would it make more sense to have every second train skip some of the lower patronage inner stations on the Western Line – like Avondale, Baldwin Ave, Morningside and Mt Eden? How much time would you gain from changing with the stopping pattern, and would it be worth the hassle (and the reduction of service from a train every 10 minutes to a train every 20 minutes at those stations)?

Personally, I think it’s important that we keep at the top of our mind the necessity that our Rapid Transit Network is, well, rapid. The average speed of trains on the Western Line is a painfully slow 30 kilometres an hour – pretty disgraceful in many respects as around half a billion dollars has been spent on upgrading this line over the past five or so years. If we had bus lanes along the Waterview section of Great North Road I am pretty sure it would be faster to catch the bus between New Lynn and the CBD at peak times than the train – something that should simply not be possible.

Of course the ultimate solution is to triple track our main lines (or quad track them). Then we have the ability to run both express and local services on the same line without having to worry about getting frequencies too high – because the faster trains could just pass the slower ones on different tracks. New York City’s subway system benefits from the express/local division enormously. But in Auckland that’s only likely on a short section of the southern line (Westfield to Wiri) any time in the near future. So really, I think we do need to have a good long think about what’s more important: speed or frequency? We also need to have a good think about what could be done to speed up our current trains (such as being more efficient when it comes to dwell times).

I don’t really know the answer here – what is more important to train catchers. Speed or frequency?

34 comments to Speed versus Frequency

  • rtc

    It’s impossible to please everyone, people travelling from Henderson in will say speed is most important, those travelling from Mt Eden will say frequency.

    One thing that comes to mind is how much time do the trains bypassing Baldwin Ave save?

  • karl

    I’d say for a regular user, speed is slightly more important than frequency, while its reversed for casual users. Since regular users will LIKELY stay on board, and casual users are POSSIBLE converts to regular use, in terms of the train system success overall, I think its a good choice.

    Further, more trains = more capacity for passengers, able to take growth until we get electric trains (and since longer trains have their limits due to station lengths, “more trains more capacity” holds true even then).

  • Matt L

    As a user from out west I think speed is a big issue and there are lots of little things that could be done to really help the situation regardless of the frequency, the main problem I have though is that trains still struggle with the timetable, especially the new 6 car trains.

    Faster dwell times are essential, cutting them down by 10 seconds a stop would save 2.5 minutes (not including Newmarket). Passengers will quickly learn that they need to be ready to jump on or off the train when it pulls into the station compared to the current practice of some people of remaining in their seats until the doors are open. Pilots or something to prevent the 3 minute end change at Newmarket could also save about 2.5 mins so I think there is the potential to save up to about 5 minutes just from this alone.

    I personally would close Morningside and Baldwin Ave and combine them in a station at St Lukes Rd. That would spread out those inner isthmus stations a bit and make things slightly faster while also providing a good opportunity to integrate with a crosstown bus along St Lukes Rd. Morningside could be retained as an event only station when there are big games at Eden Park.

    My understanding is the loco hauled SA trains don’t handle the curves that well compared to EMU’s like we will be getting. Once our new trains arrive that, along with the faster acceleration and better braking should give us a bit more of an increase as well.

    Of course the biggest way to speed up the western line would be the CBD tunnel. It knocks about 3 km off the journey to town which is quite considerable when you think that it is only about 27 km currently from Britomart to Swanson so there is the potential for some quite big time savings. I think that post tunnel we could see a Britomart to New Lynn time of about 20 minutes with it being only an additional 10 minutes to Henderson. That would make it fairly rapid and the sort of time you would struggle to do in a car at any time of the day

  • sj

    Surely what really needs to happen, in the medium term (ie, prior to triple- or quadruple-tracking) is to speed up the trains on the existing network? If our very slow trains became not so slow, then speed would increase, to a certain point, without sacrificing frequency. Why is it that our rail system (not just in Auckland, but nationwide) is so disgracefully slow anyway?

  • rtc

    “Why is it that our rail system (not just in Auckland, but nationwide) is so disgracefully slow anyway?”

    Lack of investment in proper maintenance or modern rolling stock comes to mind.

    “My understanding is the loco hauled SA trains don’t handle the curves that well compared to EMUs like we will be getting.”

    Don’t forget none of the SAs will be going, they’ll just be replacing the diesel loco with an electric one.

    • Matt L

      I remember seeing an electrification document that stated the EMU’s will run on the western and eastern lines while electric loco hauled SA sets will run on the southern

  • cateye

    There is two tracks, why not have express trains pass all-stopping ones on the other track? Considering the location of all trains is known, it would be very easy to identify when a train could make this passing manoeuvre. It would be especially easy at stations where the other train is stopped anyway. Does anyone know how many sections actually have that bit of rail between the tracks allowing this? There are some, and more could be constructed for a relatively cheap cost.

  • Geoff

    @cateye, you can only overtake on double track if there is a big gap between trains going the opposite way, which in Auckland there isn’t.

    • If you have a 4 platform station and a soundly designed timetable, Express trains passing All-stops can work. Or having the all-stops terminate at this station, so continuing passengers must transfer to the Express at this point.

  • I use the bus routinely when I’m living in Dunedin, but as a fairly frequent taster of public transport around the world, I think greater frequency brings some real advantages, mostly at a psychological level.
    1. Waiting time seems like wasted time, unless you are stuck in a bus in traffic (which is horribly frustrating), being on the move makes it seem like you are marking progress.
    2. Given the choice, in most climates (very hot, cold, or wet), I’d prefer to spend more time in the cooler/warmer/dryer environment than at the staion or stop.
    3. Faster is obviously ideal if there is bus/train at the time that you wish to travel. This I guess relates to more frequent being better for casual users.
    I don’t know if it would work on any of the Auckland lines, but it appears that the Cais Sodre Cascais line in Lisbon has a 4 platform midpoint station, and the timetable is set up such that all-stops trains wait at this station and are overtaken at this point by the express. And it really seems to work. The problem with implementing this in Auckland would presumably be the Britomart bottleneck (unless your all-stops train were to stop at Newmarket rather than switch ends, although it would really need to link tightly with another train there; and I have no idea if there is room).

    Also, on making the trains faster, having spent a lot of time messing with TomTom travelling around of late, it has really reinforced several things about moving faster. Time spent stopped is the biggest killer. No amount of driving faster etc. can really undo the time lost to a stop. Second, although shorter routes are not necessarily faster (esp when comparing 130kmh autoroute with 90kmh windy road), reducing the distance travelled (ie cutting out a diversion to look at something) really helps as well.
    Thus, I’d go with higher frequency, the CBD tunnel, and maybe the smart use of Express trains. It might just be using express trains at peak times, and having them ‘chased’ by an all-stops train.

  • They are two sides of the same coin. Both are functions of total travel time. Travellers tend to measure door to door time.

    People tend to value waiting time more highly than travel time, so generally will prefer to spend 5 minutes on a train than 5 minutes waiting for a train.

    Any decent economic modelling of this trade off should be able to give the answer. At peak times if it is assumed that the majority of users become timetable familiar, then the waiting time becomes less important (so there can be gaps).

    However, it is pretty critical that public transport be total travel time competitive with motoring. That means speed does count and should result in end to end travel times at least around the same as car (at worst 5 minutes longer).

  • Cam

    “People tend to value waiting time more highly than travel time, so generally will prefer to spend 5 minutes on a train than 5 minutes waiting for a train” I agree with this i think frequency is the most important in attracting new users in particular.

  • Matt L

    As a result of this post I decided to time the stops this morning to see just how long they were. For those that don’t know the process is: Train stops, driver enables the doors to be opened, Train Manager (TM) then opens the doors, passengers board/alight, TM closes all doors but his, closes his door and signals the driver to leave.

    With this in mind I noticed that the TM had the doors open at most stations within a few seconds of stopping however on a few occasions this was as long as 10 seconds. After that the doors were generally open for between 15 and 30 seconds and once the doors started closing it was around 12 seconds before the train left.

    The quickest stops were 30 seconds and the longest at just over 50 with the average somewhere in the range of 40-45 seconds. Definitely seems like there is plenty of room for improvement here. Is there an international best practice for how long a train should stop for.

    • Thanks Matt, very interesting.

      I don’t know how easy it would be for you to do, but getting a cumulative total of how long the train spends stationary would be interesting. I wonder what percentage of its trip-time involves it not moving.

      • Matt L

        I only did it from Henderson onwards and I can roughly remember what the times were. Including a two minute stop in Newmarket we were stationary at stations for about 10 minutes in total across 12 stations (skipped Baldwin Ave) and ended up 3 minutes late into Britomart. Total travel time was 50 minutes.

    • I did a little pseudo-experiment like this in Melbourne not too long ago, I found that off-peak or at light stops it was as little as 20sec, averaged around 30 or so when it was busy, however in the CBD with loaded trains it was 40-50sec at times. Apparently these are considered excessive here, as chronic passenger congestion has chipped away at dwell times for the last five years.
      A curious exception to this is Flinders St station, which despite being just one of the loop stations is treated as a terminus where timekeeping stops are made, drivers change over etc. Any time of day or night the dwell time here can be five, six, seven minutes. Very frustrating if you are headed to the next loop station along.

  • cateye

    @geoff define a large enough gap? I would disagree with you there, passing a stopped train (say 1000m movement conservatively) at a speed of say 30km/h (9m/s) would take two minutes. Allowing for a safety factor on top of this there is more than enough time. What Auckland line are you thinking of?

    However I agree with James, what to do about the Britomart bottleneck? With the CBD rail tunnel the travel times may be fast enough on the Western line, but implementation on the Southern line with a peak service from Puke/Papakura could be worth it regardless.

  • ingolfson

    “so generally will prefer to spend 5 minutes on a train than 5 minutes waiting for a train.”

    Much worse – the time spent WAITING at the stop is *psychologically* counted at 2-3 times the time on the vehicle, so a 5 minute wait feels like 15 compared to a 5 minute journey. So this too, would make higher frequencies seem the better choice over speed, if you can’t have both being excellent at the same time.

    Overtaking stations seems a good idea for the mid-term.

  • Jon R

    I would suggest passing loops are installed at a number of stations to allow “express” type services to pass all stops trains. The problem is all the new stations have been built without this in mind, so again, we would need to rebuild some stations.

    Passing loops could be built at:
    Western Line
    Mt Eden
    Morningside (has loop)
    Mt Albert
    Fruitvale Rd

    Southern Line
    Remuera
    Greenlane
    Penrose (has a lot of space on down line and has passing loop on up line)
    Papatoetoe
    Homai
    Te Mahia

    Eastern Line
    Orakei
    Meadowbank

    Realistically a third track is needed between Puhinui and Otahuhu (entrance to Westfield yards) so freight and express trains can be removed from all stops services.

  • greenwelly

    If you strip out the farnarkling that goes on with the newmarket-britomart section, the route is 27.5 from New market to waitakere and takes 47min @an average of 35km/h

    For reference in Wellington the Hutt line is 32.4 KM long and the all stopping service (17 stops) takes 45min @ 43Km/h the express does it in 40min @48.6 Km/h

    The Paraparamumu line is 48 KM and and the all stopping service (14 stops) takes 58 mins @49 km/h the express has 6 stops and takes 52 minutes @55km/h

    It would be interesting to see some dwell time numbers, but I think the EMUs should help a little,

    But the Hutt speeds of 43 km/h are probably as good as they will get which would lower the time from Waitakere- Britomart to around 38 Min.

    • Matt L

      I’m not sure what the lines are like in Wellington but up here the western line was never built for any decent kind of speed. Between Newmarket and Mt Albert there are many curves the trains have to wind their way through with stations spaced as close as 800m so trains can’t really get any decent speed up. I do think that EMU’s will help a bit here so we can only hope it really makes a decent difference.

  • Luke

    On the Madrid metro the dwell time was no more than 30secs, even on the busiest stations in the CBD.
    I think Auckland has much room to improve in this regard.
    Searching on Google gives plenty of examples where studies have been done into this so wouldnt be too hard to do by just looking at overseas experience.
    It would have to be multi-party approach with Veolia, Kiwirail, AT and NZTA (safety) all involved and may need changes to the safety case for rail operations.

  • Also, I know that this is obviously an extreme example*
    but the TGV Paris Bordeaux takes 3h09
    same route, 1 stop 3h18/3h20
    same route, 3 stops 3h39.
    The time spent stopped at the station is not terribly long, but the slowly down adds as well. Obviously Auckland trains don’t run quite as fast, but I think it does illustrate the impact of stopping, not just in the stationary time, but the time taken to decelerate and accelerate.
    Perhaps, if instead of having passing loops, if there are very closely spaced stops, especially early in the journey, having alternating trains hitting alternating stops, even it if is just to lose 2-3 stops might be an effective outcome (although the service frequency would ideally need to be close to 10 minutes or under).

    *Actually, it could have been more extreme, most of this route is not Line Grand Vitesse, and so the trains are only travelling at 220kph or so. However, it was this route that drew my intention to the amazing impact of only a single stop.

  • 5689hfjijkghjkdf

    Heres an interesting read on rail network capacity, in particular the part about how having all stops and express trains on the same track affect capacity:
    http://www.core2008.org/assets/papers/tuesday/1045/Ray-Bartlett/Ray%20Bartlett.pdf

    • That’s a great paper. Really, really interesting. I don’t think that Auckland is yet at the point that express trains would massively interfere with all-stops (the talk has the only been about missing very few stations), which would mean that the express train would not be as steep as illustrated in the diagram in the paper, but the point does still hold.
      Also relevant to the idea of express versus all-stops trains is the turnback notion, where suburban versus regional trains clash — which is equivalent to the idea of having express (regional) passing all-stops (suburban) trains.

  • Luke

    I think the airport line could skip Remuera and Greenlane.
    Also Papakura services could run non-stop between Manurewa and Papakura.
    The stations in between would be served by the Papakura – Manukau service. This service would also stop at Tironui, Glenora.
    As an aside should we be worried about the capacity of the Manukau station? By 2025 we should have trains to CBD, airport and Papakura. All on a terminus station with 2 platforms!

    • The problem with having Airport trains skipping stations between Penrose and Newmarket is that they’ll be likely to catch the trains in front of them. Assuming we run Airport Trains and Southern Line trains at 10 minute frequencies (or even higher post CBD tunnel) that’s a train every 5 minutes or less along that section of line – which means you probably need them to all have the same stopping pattern.

      Greenlane station is one of the saddest stations on the whole network. So much potential where it is – but I’ve spent a lot of time around Greenlane and I’m yet to have actually spotted the station.

      • Haha. I’m not even from Auckland and I know where the Greenlane station is ;)
        It’s not in Greenlane, it’s on the southern motorway. When I saw it, I was very confused. It was only after looking on google maps, I found out that it was a railway station.

        • Yeah I was just having a look at it now. I wonder if you tried to get a narrow little link road down there in front of it you could develop apartments next door – though their view would be straight out onto the motorway.

          Gah – why do we have so many useless stations in Auckland.

          • Matt L

            Why would you need a link road, there is access of Greenlane East. With the right noise proofing it shouldn’t be to much of a problem to build apartments there and that section of the motorway could be a prime candidate for covering and building over top of which would also better connect the two sides of the motorway

          • Link road so that the station feels like it fits into the urban landscape rather than is hidden at the back end of a whole pile of properties. That way houses across the road from the station could face the station and so forth.

          • Matt L

            The station isn’t really that different from what Ellerslie was and I think the station just needs an upgrade with some things to really indicate that there is a station there i.e. one of the big station plinths. Making the footpaths wider and the pedestrian motorway bridge feel safer would probably help quite a bit as well.

  • Luke

    I’ve been to events at Ellerslie Rasecourse and Greenlane was well patronised then.
    Not the nicest place on the way home though when waiting for the first train of the day on a Saturday…
    Need better signs at the station and surrounding streets as a minimum. Surely the station doens’t need a full upgrade just to get a little signage?
    With regards to skipping stops I thought they might be able to get away with the train skipping 2 out of three stations and not be a problem. Maybe a 6min – 4min pattern may be required.
    With regards to the line south of Manukau a third track would be required for this pattern to work.

Leave a Reply