Over the past few weeks the blog has featured a range of posts around the City Rail Link, options for staging stations and whether to include an eastern link from the tunnel to Grafton or not. We’ve also had some discussion from Peter about the merits of converting some of rail system to driverless metro operation, and I’m certainly on the record as advocating a metro style solution for a North Shore rail line.

These are all contentious issues which have drummed up much impassioned discussion, and taking a step back reveals why. At the surface these sorts of discussions are about infrastructure and technology: do we build a tunnel here, a linkage there? Do we want drivers in our trains or computer controlled system?  Would a different type of track geometry save us construction costs? What sort of frequency do we want on evenings and weekends, what could we afford to run?

But if we dig down a little further this isn’t really about infrastructure, it’s about what that infrastructure allows us to do. Reading the blog over the last wee while I can’t help but think we’re all actually arguing about something very mundane yet incredibly important: service patterns. I think it all comes down to one fundamental question. Do we want to run our network under a suburban commuter rail model, or as a rapid transit system that works more like a metro?

I’d like to use this post to outline the differences between the commuter and rapid transit service models as I see them, outline my conclusions, then open the floor for discussion. So without further ado let’s review the alternatives.

 The status quo: Commuter rail

Commuter rail, or suburban rail as it is sometimes known, is the service model we have at the moment. The key feature of this model is a focus on weekday peak time commuter travel to central city locations. This focus comes from framing the purpose of rail transit in two basic terms: getting people to work, and relieving traffic congestion. It’s easy to see why this pattern emerges, the city centre on weekdays is the greatest concentration of employment in any city, plus the greatest cause of traffic congestion. So in economic terms there is a certain logic to this model, putting the rail service in where it can be used by the most workers and have the greatest decongesting benefits.

That’s the goal of the model, so what are the resulting features of commuter rail that influence the way it is used?

  • Lower frequencies, typically not much better than ten or fifteen minute headways even at peak times. Travel on the system is assumed to be “appointment” travel where commuters catch the same timetabled services every day. No need to focus on high frequency because passengers follow very regular travel patterns without spontaneous travel at different times of day.
  • Capacity tends to be increased by lengthening trains rather than running them more often. One reason for this is passenger based. If you are focussed on getting weekday commuters to their desks at 9am then you’ll have a lot of people all wanting to arrive downtown at 8.50 am, so it makes sense to make that train longer rather than add another one that gets people to work too late. The other reason in operational, by linking two trains together under the one driver minimise your operating costs. If you run two short trains one after the other, you’ll need to employ two drivers instead of one and the staffing costs are double.

    The greater Boston commuter rail network. Radial, CBD focus with branches, duplication and mixed service patterns
  • A focus  minimising travel time once you are on board, and keeping those travel times consistent. This often comes from peak services patterns that run express or skip stops to shave minutes off travel times. A single line may have a variety of patterns to serve different markets. For example the same set of tracks  might see some all stops trains, some expresses from the outer suburbs that skip all the inner stations, short runners that make up capacity at the inner stations, expresses that only stop at major stations but skip all the small ones, skip stop trains that alternate stopping, the list is endless.
  • Lots of service from the suburbs to the central city. Service are arranged in radial lines that terminate in the CBD, perhaps after passing through a city tunnel or after looping around a series of city stations to distribute commuters. This is very good if you want a regular and comfortable one-seat ride to your office downtown.
  • Short span of service.  Again the focus is on weekday commuters to the city, so there often very little service during the middle of the day, in the evenings or on weekends. Some systems may not run late at night or on Sundays at all when there are no commuters.
  • Transfers are avoided wherever possible, instead lines branch for coverage. This is where trains branch off the main line, dividing frequency of service to cover more area.
  • Trains are often designed around maximising seating for long distance travellers, and relatively low turnover of passengers. This typically means little standing room and fewer doors per carriage to get more seats in, which may come at the expense of congestion around doors at stops.
  • Travelling for the first time usually requires consulting a timetable and a network map, both of which may be quite complicated and hard to understand at a glance.

 The contender: Rapid transit

Rapid transit can mean several things but I’m using it here as a synonym for a metro style service pattern. I don’t want to use the word metro itself because many people associate that with underground rail systems like those of Paris, London or New York, and it’s not the tunnels or infrastructure that are the important thing. Regardless of whether the tracks are underground, elevated or at ground level, it’s the service pattern the trains run that make rapid transit rapid transit. The focus here is on all day travel across the whole network, any day of the week or any time of day. Under the rapid transit model all parts of the network are usually just as accessible as central city stations, and it’s not assumed that people take the same timetabled service every day. Instead rapid transit is designed so that people can travel sponteneously at any day or time, from any origin to any destination. Rapid transit dispenses with the assumption of servicing weekday city commuters and instead tries to service any kind of travel.

So what does this rapid transit / metro look like?

  • High frequency all the times on every line to give -turn-up-and-go freedom to travel whenever and wherever people like. Usually this means at least one train every fifteen minutes at any time of day or night, and up to five minute headways or better at peak.
  • Long span of service, with normal service from early morning until late at night seven days a week. The sorts of service commuter rail enjoys during the peak is available on rapid transit at any time.
  • Capacity tends to be increased by adding frequency, but can come from longer trains once high frequency is already reached.
  • Connections are utilised for coverage and branching of lines is uncommon. Lines generally have the same frequency along their length and people need to connect from one line to another at transfer stations to get to different destinations on the network. The requirement to change trains to get to certain destinations is offset by the much shorter wait times that single frequent service patterns provide.
  • Rapid transit often has through-routed “pendulum” lines. Most lines pass through the CBD rather than terminating there, while some may not enter the central city at all. A through routed line give roughly double the number of potential direct destinations as a line that stops in the CBD.

    Singapore rapid transit. Through running crosstown lines, interchanges and almost no duplication or branching.
  • Single consistent service patterns. No peak specials, expresses or skip stops. The ability to shave minutes off the trip downtown is forgone in favour of regular and dependable all-stops service that anyone can use to get anywhere. The stopping pattern at 11pm on a Sunday night is the same as it is at 8am on a weekday morning.
  • Trains can be set up with more standing and circulation room on board and more doors per carriage. This comes at the expense of some seating but allows more capacity overall and facilitates a high turnover of passengers at each station.
  • Timetables and route maps tend to be very simple and easy to understand. Timetables often don’t list individual services but just outline the frequency and span (“The blue line runs every five minutes from 7am to 8pm”).

In short, rapid transit relies on connections and regular service to allow people to travel whenever and wherever they want with a minimum wait time. It allows people to rely on the rail system for all sorts of travel, not just their regular city commute to work.

 So which is better?

Reading this post you can probably see a tinge of bias against commuter rail already, and to be perfectly honest I think it is warranted. I’ll leave you to make your own decisions, but I’m happy to come out and say I think a rapid transit “metro” style service model is the one we should plan for in Auckland.

My thinking comes down to a very basic line of reasoning. Rapid transit serves many trip types and destinations well, but commuter rail only serves commuters well. The deal breaker in my opinion is the fact that a well designed rapid transit network can serve commuters just as well as a commuter rail system, but a commuter rail system will fail to effectively serve a wide range of origins and destinations at various times of day or night.

Commuter rail is great if you make just one return trip a day to your job at the same regular time, and you’d like to do that as quickly and comfortably as possible. But it’s not very good if you try and use if for other reasons: to shop, to visit friends, to get to sports practice, to go to the movies, to make a date (and keep it!) regardless of the time or day of the week. I would also argue that commuter rail might not actually be so much use to office workers these days. Modern commuters have varied start times and varied responsibilities, they can often work across different sites, start at different times, work late to finish projects, have meetings all over the city, attend after work functions and come in on the weekend to get reports finished off in time.Commuter rail lacks the frequency, connectivity and span of service to make spontaneous trips where and where you want them. In a nutshell commuter rail doesn’t give you the same car-like freedom of rapid transit.

The reason for this stems from the fact the commuter rail has a very narrow set of goals and objectives. Our rail systems in Australasia tend to follow this narrow focus for two reasons, both of which are somewhat artificial.

The first is that our funding and evaluation mechanisms are framed in terms of road traffic decongestion, in particular weekday peak traffic. Commuter rail is appealing to peak time commuters so it is easy to justify it as a means to bleed traffic off the motorways, in particular longer distance commutes from the outer suburbs. Indeed, many of our public transport models only every consider peak hour travel, and the outputs are often judged in such useless terms as ‘cars taken off the road’ or “minutes saved by motorists”. It could be very hard to fund a rail proposal that moves a lot of people around at off peak times, simply because if the roads aren’t congested in the off peak there is no decongestion benefit to be had. In short, rail can only be considered worthwhile and funded where it relives peak CBD focussed travel, which means we get the commuter model funded.

The other reason is a little more contentious. I believe that commuter rail systems are favoured in our part of the world because our decision makers never take public transport. Because all our politicians and executives drive, they are predisposed to service patterns that most replicate their expectations for travelling by car.  That is to say they focus on vehicle speed, personal comfort and system capacity, the things that are salient when you travel in your own car.  What gets lost is any understanding of span of service or frequency. How do you explain the enormous value of very frequent service to a person who never has to wait even a second to get in the drivers seat of their chosen transportation. How do you discuss span of service with someone who can take their car out of the garage at any time of day or night, seven days a week?

There is a discussion to be had here around economics. On one hand there are undeniable benefits of relieving peak hour congestion and there is the simple fact that our CBD couldn’t grow significantly without more rail access. But there are big inefficiencies with a focus on the CBD peak, the main one being that you require a large amount of infrastructure and lots of trains to operate at full capacity only a few hours a day, five days a week. Outside of the peak all that sunk investment sits empty and unused in the stabling yard. Also the CBD peak is a finite sized market, one that is already relatively well served by public transport and doesn’t have a huge potential for further patronage growth. So what is so wrong with this? Well it leaves us with a system that is expensive to operate, but one that also has a self-limited ability to pick up new paying customers.

The rapid transit model on the other hand utilises much more of that infrastructure and vehicle capacity right across the day and the week, and furthermore it doesn’t limit the market to central city destinations or peak commuter times. While it has a higher operational cost than commuter rail, it also has a much larger pool of potential customers. If done right, rapid transit can be much more cost effective.

 Can we have best of both worlds?

While I think the rapid transit model is the way to go, I can’t ignore the commuter rail basis of our existing network and the demand there is for peak hour travel to the CBD. It seems that a hybrid model might be ideal, a commuter rail system that operates like a metro with a regular all day service pattern and high frequencies in urban areas. Something that works like suburban rail in Swanson or Pukekohe, but runs like rapid transit through New Lynn and the central city. Luckily there are such systems abroad that we can use as an example, including the German S-Bahn and the French RER. These are effectively commuter rail systems that have been linked by underground tunnels in city centres, and upgraded to metro style frequency, span and stopping patterns. Sound familiar?

Share this


  1. Assuming I’ve understood what you’re getting at, the area of the Auckland isthmus is about the maximum physical extent of what you mean by rapid transit. In Auckland’s case that would mean a fifteen minute frequency from outside the isthmus but a much higher frequency within it and with the CRL it would be possible to have central city ‘loop’ trains as well. So calling it the central city loop link is not impractical.

    1. But not desirable either- a loop running pattern that is. Through routing with good frequency will allow easy journeys to anywhere on the the network with no more than one transfer and serve those longer rides directly, so why would you bother with the difficulty of running a loop service?

    2. Don’t worry Ross, you are not the only one thinking of at the minimum seeing if a loop system would work with the CRL.
      Providing the issue with the Onehunga Line between Penrose 3 and Onehunga itself gets sorted (it will have to for the Airport Line in any case) and the politicians see through with The East Link at Mt Eden, running a small loop using the CRL would seem quite straight forward.

        1. I would qualify that statement a little.

          Large one way loops make for very problematic service patterns, especially from the passenger perspective. Those should be avoided at all costs. We don’t want to replicate the Melbourne City Loop in Auckland for example, especially not a version almost twice as long as ours would be.

          Very small one way loops (of only two or three stations) can be ok, but we don’t have that potential in Auckland so it’s a bit moot.

          Two way loops on the other hand can be quite effective from a passenger perspective if they can pick up a variety of shorter tangential and radial trips on the same route. No one is going to ride a loop right around, but many people might ride it a quarter or a half the way round, all around the circle. These still have operational issues with timekeeping, crew breaks etc, but if those can be managed a two way loop can work for passengers. One strategy to fix a two way loop would be to follow Londons example and break the loop at one point, giving it defined start and end points (even of those are at the same station).

          In Auckland I could see the potential for an ithsmus loop line using the inner parts of the western and eastern lines, the Onehunga branch, and the SH20 corridor between Onehunga and Mt Albert. This would have serveral uses, it would boost frequency on the busy inner section of the western line, it would supplement capacity on the inner part of the eastern line (where the AMETI corridor connects), it would provide a semi-radial trip from Mt Roskill and Owairaka to the CBD, it would link together the radial lines and it would provide for some crosstown trips, say Onehunga to Silvia Park.

    3. Actually Ross I was thinking most of the urban area should be covered by rapid transit, say Henderson to Papakura and everything in between. Beyond those there would still be regular service, just not at the full metro frequency. So through the day we might have a train every ten minutes to Papakura and every thirty to Pukekohe, which would simply be every third train starts and end at Pukekohe instead.

      Anyway the key point of this post was about the service model, not the extent of the network. Basically I’m saying Auckland needs a simple service pattern with regular and reliable all day frequent services. We could copy Australia and arrange our network to focus on getting heaps of commuters downtown between 7 and 9am on weekdays, or we could arrange our network to focus on getting people to all sorts of places whenever they want to go.

    4. Singapore (and the transit map above) is approximately the size of CBD to Onehunga, Henderson to Glendowie. The isthmus argument may have worked during the time of motorway construction, but we now have a massive area of industrial and residential development in the east and south that wasn’t there in the 50s. We also have an additional motorway through Mangere/Onehunga, and a lot of associated development south of the Manukau inlet which has broadened the ‘isthmus’ considerably.

  2. Seems like a hybrid rapid transit / commuter service is what is needed for AKL as Nick points out. Can such a service pattern be implemented easily once the CRL is built? Can portions of it be put in place before then with the EMUs coming on stream fully in 2016?

    1. Sure, what I’m insinuating here is that we should design the CRL with an S-bahn rapid transit/metro type of service pattern in mind, not simply a bigger suburban system. By linking up the dead end of our network back onto itself this presents the opportunity for really good through routed metro style lines.

      Before then we’re probably going to have radial commuter style routing, but there is no reason why we can’t run it with a regular frequent all-day timetable to get started with. There might even be some opportunities to arrange our branch lines such that they connect across the other lines rather than compete for space in Britomart. Stay tuned, Patrick has some ideas on that line of thinking which he’ll post shortly.

      1. Because of the network the CRL will give us even through routing can’t avoid operating as semi- loops so in some ways a pro- or anti- loop argument is really only of degrees in practice. For example any train entering the CRL from the western line will have to visit four the stations to Britomart and then can either go out on the eastern line or to Parnell then either south or west. My view, and I thinks Nick’s is simply that the least desirable of those three options is the last one where it ends up heading back the way it came. Primarily because this seems like a less likely journey than either of the other two, and really only seems a good idea if you only imagine people coming into town and out again. Not say going from Kingsland to GI, or New Lynn to Slyvia Park. Anyway the frequency in the inner part of the system will be such that transfers are the key to serving every possible action.

        Will post some running pattern options tomorrow.

  3. One major point you missed in your rapid transit points was that rapid transit by definition must be completely grade separated from all other forms of rail traffic, road crossings and completely closed off from people or objects getting on the tracks by fences and barriers. You can’t run freight or intercity traffic on a rapid transit line or even suburban services (with some exceptions). Auckland could not do this on the current network without huge expensive. A new separate network would need to be built along side it. The mot cost efficient option would be the hybrid model. Melbourne, and Sydney in particular, are considered hybrids systems. In fact some lines in Sydney do run on metro style frequencies. The RER or JR are the best examples of these and are the ones to look up to.

    1. That’s not true Sean, I didn’t miss that at all because it not a requirement. You can run a rapid transit without being completely grade separated and completely closed off from the tracks. Where do you get that from? You can still run a train every five minutes on an all stops pattern with level crossings, it just means the level crossings are down often. Not ideal for local drivers, but no issue for the rail system. Hell many German rapid transit systems have large sections that run on street, through intersections and traffic lights!

      What makes rapid transit rapid transit is the frequency, the span of service, routing and stopping patterns… not the infrastructure.

      The idea that we need a hugely expensive system of separate new tracks to run our rail system under a metro style service model is ridiculous. The simple fact is Auckland and many other cities will/do run rapid transit frequencies and service levels every day at peak times on their commuter networks, what makes it rapid transit is extending this sort of service level across the whole day. If we can do that from 7am to 9am already why would we need fancy new infrastructure to do that from 9am to 3pm?

      1. Ok, so how would you run say 3-8min all day frequencies with freight and intercity trains using the same lines AND keep it reliable? If you look at the largest metros/subways in the world, virtually all of them are on their own right of way seperate from freight & intercity services.

  4. I’m not talking about large metros and subways Sean, that’s why I chose to use the words rapid transit to describe the service model rather than the word metro. I knew people would get too hung up on infrastructure

    Auckland has one intercity service every two days, that’s not an issue. For freights it would be a case of running them in the interpeak or off peak when headways are around ten minutes or so. That’s the current plan anyway, again all this is really about is running a single stopping pattern per line, and ensuring a long span of frequent service. It doesn’t have to be every three minutes all day, just better that fifteen or so to get that turn up and go. So simplify the peak pattern and maintain it across the day.

    Another question, if service is every 3-8 minutes why does it need to be reliable? One of the features of rapid transit is that there are no timetables and you simply turn up and take the next train that comes every few minutes. How could it be reliable if you have no timetable to work to?

  5. BRILLIANT POST. As a newbie to the finer points of commuter rail I learnt a lot and maybe it should be tagged – backgrounder or similar for the next newbie. Are there any more backgrounders like this one I should read. As an aside there’s no need to worry about freight trains disrupting the non existing rapid transit service frequency (and that by the way is the peformance measure)because once they build the holiday highway they’ll have to incentivise all the rail freight onto trucks to be able to do some after the event justifications.

Leave a Reply

Your email address will not be published. Required fields are marked *