auckland: how network redesign can transform a city’s possibilities

When a public transport network has grown cumulatively over decades, but has never been reviewed from the ground up, it can contain an enormous amount of waste.  Careful redesign is the key to unlocking that waste and generating vast new public transport mobility.  Our new plan for Auckland, New Zealand, now open for public comment, is a dramatic example of what can be achieved.  ("Our" because I led the intensive network design work, with a great team of planners from Auckland Transport and my colleagues from MRCagney.)

If you want to get around Auckland at any time of day, on a service that's coming soon, here's where you can go on today's network (or more precisely, a "business as usual" network extended to 2016)

Akl existing
Under the proposed plan, which costs no more to operate than the existing one, here's where you'd be able to go, at any time of day, on service that's coming soon.

Akl proposed

The network still includes coverage to all corners of the city that are covered now, and ensures plenty of capacity for peak commuters into the city.  But meanwhile, it defines an extensive network of high frequency services around which future urban growth can organize to ensure that over time, more and more of the city finds public transport convenient.

What's the catch?  Only the geometrically inevitable one: more people will have to make connections from one service to another, and the fare system will need to encourage rather than penalise that.  

Whenever someone tells you that it's too expensive or hard to encourage people to make connections, ask them how expensive it is to run the only the first network above while spending enough money to run the second.  Networks that are designed to prevent transferring must run massive volumes of half-empty and quarter-empty buses and still have trouble delivering frequencies that make the service worth waiting for.  The waste involved can be colossal, as you can see from the amount of service we were able to redeploy in more useful ways with this redesign.

To see a bit of the structure clearer (and also because it's a cool graphic), here's the central slice of the drawing of the proposed frequent network, by my MRCagney colleague Nicolas Reid.  It's currently all over the media in Auckland, helping people assess the plan.  By streamlining it calls attention to the logic to the network — a logic that's sometimes easy to lose track of when following the details of every right and left.  Look at the whole thing.

Auckland network

I'm very proud of what our team achieved working with the excellent folks at Auckland Transport, and I hope the plan will be further improved as a result of public feedback, as good plans always are.  But as Aucklanders begin discussing the plan, I hope they stay focused on the core question:  Are you willing to get off one vehicle and onto another, with a short wait at a civilised facility, if this is the key to vastly expanding your public transport network without raising its subsidy?  

That is the real question before Auckland now.  The rest is details.

29 Responses to auckland: how network redesign can transform a city’s possibilities

  1. John October 10, 2012 at 5:06 am #

    Amazing expansion of frequency. I would be interested to hear more about where you found redundancies, or maybe where one might typically find redundancies are in an aged transit network.

  2. Bruce Nourish October 10, 2012 at 5:30 am #

    Amazing. Beautiful. Hope it gets implemented, and not Seattled to death like our restructures.

  3. Jonathan October 10, 2012 at 5:34 am #

    I’ll be honest – I have doubts. First, it looks like there’s more miles of route on the proposed 2016 map than there are on the no-change 2016 map. If so, where did this extra service come from? Reducing frequency or service hours on the existing routes (its no good if the bus that comes every 15 minutes is full, because the capacity demands 5-minute frequency), eliminating routes that aren’t frequent enough to appear on the no-change map (have service obligations been met for all areas), or new money (in which case the proposed/no-change is an apple-to-oranges comparison).
    Second, and somewhat related, I’m concerned about the mention of ‘phasing’ of the frequent network service – down on page 29 or so, point 2.6 iirc. Does this mean the frequent network isn’t really 15-minute frequent, but 4-buses-per-hour, or similar? It seems like this might be an attempt to manage insufficient capacity on the frequent network (perhaps caused by overstretching the reach of the frequent network in order to meet access goals to compensate for eliminated, non-frequent services)?
    Are you allowed to comment, Jarett, or are you under restrictions?
    That negativity mentioned, it looks good – particularly the yellow line, and the blue across the top on the 2022 map – a part of the network that seemed obviously ‘missing’ in the 2016 version.

  4. Miggle October 10, 2012 at 6:48 am #

    @JMH: I’d very much doubt that any current frequent routes would be reduced in frequency, given there are only 3 routes in current service with greater than 15 minute frequency off peak (the Northern Busway and Mt Eden Road at 10 minute and Dominion Road at 5 minutes). It’s more likely the savings will come from replacing infrequent routes to many different destinations.
    As in example there are currently over 60 separate bus routes on just the North Shore, and only 2 with 15 minute frequency or better (see page 3 of the link).
    Replacing these with 8 trunk routes (and a reduced number of low frequency connecting routes)) would certainly save money.
    The change is actually far more dramatic than the first two maps imply, as most train routes are not in any way frequent. Only the section between Otahuhu and Papakura has 15 minute service all day. The rest of the train network has half-hourly or hourly service in the off peaks. The newspaper ad at least fudges this and calls them “rapid” routes.

  5. Jonathan October 10, 2012 at 12:48 pm #

    @Miggle – that’s really nice then. It looks like you get a frequent service within two miles of your front door, guaranteed – and in that context, I can easily see how the pressure to have a service within 0.5 miles, or 0.25 miles, to the city centre, could’ve created a lot of services with, say, 1 hour or less frequencies that nonetheless had quite a lot of route-duplication, and how those less-useful services would’ve been built with development. Hopefully new development will appear at the nexus points of this new frequent grid.

  6. Louis M October 10, 2012 at 1:19 pm #

    The new network will be great for Auckland. A job well done by MRCagney / AT. Savings will come from cutting the endless variations of routes that gobble huge amounts of subsidy, because they carry very few passengers.

  7. Jarrett October 10, 2012 at 1:59 pm #

    To be more precise this is a comparison of 2016 “business as usual” vs redesign. Improved rail frequency is already programmed as i understand.

  8. Nicolas Reid October 10, 2012 at 2:16 pm #

    @JHM. Further to what Miggle has said the map above is just the primary frequent layer of the redesigned network. There is a ‘secondary’ layer of regular-but not-always-frequent services below that, plus a third layer of local connectors and peak-only services.
    The purpose of the Business as Usual map is just to illustrate how hardly any of Auckland’s 300+ existing bus routes meet a consistent ‘turn up and go’ service level.
    If you want to see what the whole network looks like now, we’ll you can’t! It is so complex no one has ever captured it on a single map. These area guides cover most services (terribly mapped if I do say so myself), but don’t cover the peak express network properly:
    To answer the general question of “where did all this frequency come from”, the short answer is more than 300 routes were streamlined down to around 120.
    Part of that involved shifting to a connective model to get anywhere to anywhere connectivity, part of it involved cutting out all the little wiggly diversions and variants that only served one or two bus stops (and expecting some people to walk slightly further to get to much improved service), and part of it included minimising the multitude of long one-way peak express routes to downtown in favour of short frequent connections to trunk routes.
    A further factor is that Auckland is blessed by some significant geographic pinch points which are serviced by radial rail and busway corridors. Auckland is in the process of constructing a new fleet of electric trains that will roughly double the passenger capacity of the main west and south lines. The redesigned network takes this into account, and connects a lot more bus service into key rapid transit stations rather than running them across the city in parallel as is done today.

  9. Jeffrey Bridgman October 10, 2012 at 5:36 pm #

    Awesome! I’m jealous 😉
    I hope the full plan gets put in place!

  10. Siemaszko October 11, 2012 at 2:12 pm #

    As someone who’s relied on buses as primary daily transportation in a few different cities, I’ve got this to say:
    Transfers suck. Full buses are just as bad. A whole system of full buses will quickly lead every user who has the means to find alternate transportation.
    During rush hour transfers are the worst. You can’t expect people to get off a full bus and hope/pray that the next bus they need to get onto will 1.) be on time and 2.) have room for new passengers. In my experience as a rider, this type of system has turned 30 minute commutes into 2 hour commutes, even when buses supposedly came every 7 minutes. I’ve seen fist fights break out over people trying to board/unload. I’ve seen drivers attacked. It’s REALLY bad to run at full occupancy, and even worse to suggest transfers at the same time.

  11. Tony October 11, 2012 at 5:12 pm #

    The current Bus Review in Wellington, New Zealand also received the “HumanTransit” makeover to establish core high frequency routes at the expense of higher levels of transfer.
    However, the subsequent public consultation had a large majority rejecting these proposed changed (74% Dislike vs. 12% Like). Peak hour commuters had an even higher dislike level at 83%. See Item 5:
    It should also be noted that the feedback report also stated “A 7% response rate for feedback has enabled a high level of analysis to be undertaken and conclusions to be put forward.” So this is not from a small sample. In fact the regional council has never had a consultation with so much feedback.
    It is to be expected that many commuters will disliked these changes (route changes had the highest ranked dislike). However the rejection of the layered interchange model is clear with disliking “Making Connections” and “Changes in Frequency” being 2nd and 3rd highest dislikes. In Wellington this model would have dramatically increased interchange levels from 3% to 13% with up 30% of commuters transferring (i.e. whole busloads) at some points!
    IMO the model has two very serious flaws:
    1) High frequency between connection points is fine if you live near one, but if you have to transfer onto connecting bus at a 30 minute frequency travelling home, the higher core frequency does not help. In fact having to the interchange will mean a worse service than a direct 30 min service.
    2) Partially because of 1) commuters will tend to drive to interchange points which do have high frequency which will cause major congestion/parking problems around them. This also means connecting peak bus users going to work will have to compete with park & ride commuters to get onto a core route bus.
    Therefore I agree with @Siemaszko’s points.

  12. Jarrett October 11, 2012 at 11:17 pm #

    Siemasko. If you hate full buses, does that mean a system of full buses is a failure? Sounds like you’re falling into the Yogi Berra fallacy: “Nobody [will use transit}, it’s too crowded.”
    Transfers can indeed suck in many circumstances. The question is whether you prefer to have a much-constrained transit system, along the lines of the two maps above. Chapters 12 and 13 of my book take you through the tradeoff much more clearly. However, if a city prefers the first map over the second, that’s a valid choice.

  13. John Smith October 12, 2012 at 5:13 am #

    I often wonder if the ‘benefits’ of ‘networks’ are often overstated? I’ve never been to New Zealand, so can’t comment on this ‘network’; but in my UK experience, most bus users use one bus, on one route, day in, day out, and rarely need ‘connectivity’. Networks always look fine on paper, but I’m having serious doubts about their usefulness in the ‘real’ world. And do significant numbers of people need to travel end-to-end on long, cross-city routes?

  14. Sam October 12, 2012 at 11:36 am #

    I think the network/transfer issue depends a lot on what sort of city you have and who uses the buses. In Milwaukee, where I live, we have a grid/transfer system of mostly high frequency routes in the core that has pretty even ridership throughout the day and on Saturdays. A lot of the highest ridership routes go through areas with really high unemployment (20/30%+) where 30-60% of households don’t have access to a car. In places like here, a transfer system is important because people don’t just take one bus one place. If we didn’t have a grid, we wouldn’t have the ridership because people just wouldn’t go places. On the other had, this system serves the outlying areas really badly, which is why we have popular freeway commuter bus lines which don’t interact at all with the regular system despite being run by the same org. I guess the question might be: “What kind of city is Auckland?”

  15. Siemaszko October 12, 2012 at 12:18 pm #

    “Full buses” might be an oversimplification. I fear riding a bus that is so full that merely trying to enter/exit the bus can lead to physical altercations or hardship. I fear buses that are so full that they can’t pick up all the riders at the stop, especially if there’s no queuing system for waiting riders. Running at “ideal capacity” vs “maximum capacity” is important, especially when considering transfers.
    As to the Yogi-Berra fallacy: It’s not quite that simple. We’re talking about a system running at lower-than maximum capacity in the present. If the new system chases riders away toward other means of transport, then the final result will be an equally inefficient system, only fewer people will be using it. This can easily spiral downward as continued efforts to be efficient cut into total ridership.

  16. MC October 12, 2012 at 3:57 pm #

    I’ll have to look at this in detail, but I sense that unless this is primarily a BRT system, that the transfers could be problematic from a passenger’s point of view – because buses run notoriously late. We know when Portland shifted from buses going directly downtown, to feeder buses to light rail, for example, transit times went up for many and ridership initially went down. If a network requires even an 8 or ten minute wait (which easily becomes 15 or 20 minutes if one bus is off schedule)the ride becomes long enough to make driving the preferred option.
    I think the network concept makes more sense in large-gridded cities like Toronto.
    The routes thru downtown certainly need to run in bus-only lanes in any case or the traffic delays will make any bus system unattractive …
    I have read Brisbane has achieved a highly successful system by running buses in bus-only BRT routes thru the congested center and then fanning out in the suburbs – reducing transfers for those going downtown. This apparently suits Brisbane’s city form – with a concentration of work downtown surrounded by low-density spread-out suburbs – which sounds similar to Auckland. I could imagine bus routes would overlap thus running very frequently along the urban BRT routes and then fan out in different directions (i.e. at lower frequencies) into suburban neighborhoods.
    I wonder if this approach combined with last mile solutions at the suburban end would make more sense. Bicycling could be one last mile solution – Auckland has good weather but on the other hand is very hilly. Perhaps small electric vehicles accommodated at park-n-ride’s?
    From a planning point of view it will be necessary to locate office uses at transit nodes – and not in auto-oriented office parks.

  17. gap October 12, 2012 at 4:20 pm #

    Much of what you have said seems to be highly pessimistic and predicated around worst case scenarios. Are there any specific numbers in this particular plan that you can point to as evidence that crush-loaded buses will be a likely outcome?

  18. EN57 October 12, 2012 at 9:55 pm #

    Identifying and unlocking waste is important. But there’s never really enough explanation with these redesigns as to whether the aim is to increase and maximise transit patronage or to maximise theoretical mobility over an area by transit. I don’t think they’re the same things.
    For US cities with weak landuse planning (dispersed retail, jobs and density) on giant street grid networks, a multidestinational grid might be the best strategy for maximising transit mobility and ridership. I don’t think that’s necessarily the case everywhere. Depending on geography and landuse patterns, radial patterns might work better in some situations.
    Those of us unfamiliar with Auckland or Wellington can’t really tell here if the proposed network redesigns are really on the money – and whether this approach is really reallocating wasted resources in the best ways possible to improve transit mode share in these cities specifically.

  19. voony October 13, 2012 at 2:14 pm #

    Thanks Tony for the link.
    But as I have noticed in on of my post going thru a similar exercise at my humble level ( )
    The few riders affected by a bus change will always be more vocal than the more numerous new customers [it will attracts].
    that is the problem of a public consultation, and the problem a transit agency has to face is gauge this versus the objective potential improvement of the change, (market analysis), especially if the network is called to growth…where today little inefficiency (like servicing a development of the direct way, case of bus 410 in my post), can become very expensive when frequency of the route ramp-up…
    As a side note, the french railway is moving toward clock faced schedule (train departing at regular interval vs esoteric hours): that pretty often translates in 2mn or so change in the trains schedule…What is sometimes fiercely combated by angry commuters, because in some instance that means they have to take a train sooner or later to match their work schedule…
    What they don’t want to see is that the schedule change improve the commute of many others…

  20. Jack Horner October 13, 2012 at 3:21 pm #

    Current users who would be inconvenienced by change will always be more vocal than current non-users who might benefit from change.
    Did the Wellington consultation include a statistically valid survey of current non-users?
    On Siemaszko’s concern about overcrowding: I would hope that the brief for such a redesign included: ‘where a corridor currently operates at higher frequency than the chosen ‘adequately frequent’ default (apparently 4 per hour here) to cope with demand, this should not be reduced.’ Perhaps Jarrett can clarify that.
    Certainly a more transfer based system does depend a lot not only on frequency, but also on punctuality. It’s not such a joy when ‘every 15 minutes’ really means ‘probably within 20 minutes’.
    More effort on bus lanes, traffic light preemption etc should be regarded as an essential part of this sort of redesign.

  21. Andrew October 13, 2012 at 8:33 pm #

    Ottawa needs a redesign of this sort. This will probably happen to some extent when the light rail line opens but needs to start before then. Ottawa has far too many rush hour only express routes from the suburbs to downtown (meaning that the Transitway has a bus every few *seconds* in rush hour), yet the rest of the network is badly underserved especially outside rush hour. Many major routes like 1, 2, 4, 7, 12, 85, 116 etc. have to put up with 30 minute frequencies during evenings and weekends, while all the resources are poured into the rush hour express service.

  22. Jarrett October 14, 2012 at 7:19 pm #

    Jack Horner.
    Re your proposed rule: ‘where a corridor currently operates at higher frequency than the chosen ‘adequately frequent’ default (apparently 4 per hour here) to cope with demand, this should not be reduced.’
    I believe we’ve observed this rule on the midday, which is what these maps show. Existing midday frequencies are so poor that it was easy to improve all of them, with the focus on ensuring that as many corridors as possible got up to 10 minute frequency (city) or 15 (some of the outer suburban lines shown).
    On the peak, we found much surplus service, for example routes converging from two origin areas but then running together for a long distance, each less than half full. In that sense, we did reduce some peak frequencies, but those frequencies were not consistent and intentional so much as accidental results of so many buses converging on the city.

  23. Carl Weckenmann October 14, 2012 at 8:43 pm #

    Any transit planner who has not faced the ire of a customer base who just wants the status quo, often rejecting changes that will benefit them, has probably not done much of significance. The question for the riding public really should be, “How does this affect your travels?” rather than “What do you think?”. Service planning by consensus is a sure way to be ineffective.

  24. Joseph Alacchi October 16, 2012 at 3:18 pm #

    @Carl Weckenmann

  25. BrizCommuter October 16, 2012 at 11:42 pm #

    Interchanging is not a problem if the average wait is less than approx. 5-7.5mins (10-15 min service frequency). As the average wait increases, the attractiveness of interchanging rapidly decreases.
    Facilities at or around interchange points (toilets, shops, food) can also make interchanging noticeably less painful.

  26. Nathanael October 20, 2012 at 5:39 pm #

    “Interchanging is not a problem if the average wait is less than approx. 5-7.5mins (10-15 min service frequency). As the average wait increases, the attractiveness of interchanging rapidly decreases. ”
    This is almost a truism, but it is the basic principle of everyday urban transportation. (People tolerate longer transfer times on intercity trips. But there are still limits.)

  27. Transport in Auckland November 2, 2012 at 3:13 am #

    After viewing at this plan I can say that it will be people friendly.It should be implemented as soon as possible.Transport in Auckland

  28. Torbayite March 24, 2013 at 1:52 am #

    Does the changes apply to all the network or just the frequent ones?

  29. Matt Miller November 24, 2015 at 12:36 pm #

    ‘Whole thing’ link is dead.