Wellington Scoop

The future for light rail

by John Rankin
Wellington’s public transport through the CBD needs four lanes — two for buses and two for rapid transit, of which the most cost-effective option is probably light rail.

For light rail to be most effective, it needs a dedicated right of way with priority over other traffic at intersections. Any sections with shared rights of way will potentially lower the quality of service (making it less reliable) and need to be avoided where possible. In Wellington this will be hard in places, but we should still try.

The international data shows that light rail is more cost effective than buses when the ridership exceeds about 3000 people per hour. The Golden Mile currently carries 6000 people per hour during the peaks. Not all bus trips would be suitable for replacement by light rail, but 4000 on light rail and 2000 on buses should be readily achievable, comfortably above the 3000 people per hour threshold.

The WCC predicts the population of Wellington city will grow by at least 50,000 people over the next 25 years. Even if we assume no mode shift from cars to light rail on the corridor from the station to the airport, that’s still 5000 light rail trips per hour (25% growth). This is why there are many cities in France and other countries, the same size as Wellington or smaller, with successful light rail systems. And while Wellington has good public transport ridership in comparison with other Australasian cities, the evidence from comparable Canadian cities is that we could aim to double the ridership per capita, as people choose to mode-shift from cars to light rail.

There is a rule of thumb that when you have more than about 75 buses per hour on a corridor, light rail is a cost-effective alternative. Wellington currently peaks at about 140 buses per hour on the Golden Mile. Auckland is justifying light rail on Dominion Road in part on the basis that there are 130 buses per hour.

We do need to make sure that light rail is affordable. To do this we can ask two questions: How does the cost of light rail from the airport to the railway station compare to the cost of extending the motorway from the Terrace Tunnel to the airport? And which will deliver a congestion-free journey – light rail or a motorway?

The best advice I have ever read for cities contemplating their first light rail line is from Jane Jacobs:

“Fixed transit routes [are] expensive failures when they [are] not preceded by evidence of sufficient demand…. Choose to locate rail routes by observing which bus routes are most heavily used…. [Otherwise], they don’t have enough passengers.”

The high capital cost of light rail, relative to buses, means light rail is best value on a high-demand corridor, operating at a high frequency, all day every day, on a dedicated right of way — if a city compromises any of these (for example, targeting peak hour travel only), the benefit-cost ratio reduces

On high volume corridors, the operating cost of light rail per passenger is significantly lower than that for buses, because one driver is carrying about 6 times as many people (and it’s possible that by the time Wellington builds light rail, autonomous in-street light rail will be commonplace, further reducing operating costs).

Opportunity is the mother of invention and LGWM is a once-in-a-generation opportunity, so let’s take it.

[Parts of this article were first published by Wellington.Scoop as a comment on Sunday.)


  1. Glen Smith, 28. March 2017, 10:17

    John. This is the most intelligent position on transport I have seen- particularly the first two paragraphs. Two across town PT corridors are required. To limit mode transfers one should be bus (logically the Golden Mile) and one rail (and I talk generically- rail has many and varied forms and part of the investigation should be to see what form is best suited, and achievable, in Wellington). The rail corridor should ideally be dedicated and the only realistic option (in my view) is the Quays using 2 of the 6 lanes. This rail corridor should have 2 primary aims 1.To open up the southern CBD to rail users from the north 2. To provide a rapid transit route for PT users who want to get across the city (cars have the motorway- PT has nothing). If technically possible (and Brent Efford provides convincing evidence it is achievable) transfer should be eliminated at the station to provide a seamless rail network. The destination should be the airport ( in my view the only destination, keeping in mind your caution on cost benefits, where rail is likely to be justified in the foreseeable future- largely due to the projected 22,500 daily passengers from the airport by 2030).

    A lot has been said about the up front cost of rail but very little analysis appears to have been done on the ongoing costs of NOT proceeding with rail, which are likely to be huge and soon exceed any initial outlay. This is not only capacity. Anecdotal and research evidence is that the attractiveness of buses is much lower than rail (compare the uptake of our airport bus service with ridership achieved on airport rail services overseas, look at how how few train users switched to buses in the Hutt rail washout- only 27%- and look at overseas research which showed the majority of rail users would go back to cars if rail was unavailable). This lower uptake will translate into increased costs associated with mass car transportation (pollution, accidents, climate change etc). One of the large ones will be increased congestion which the evidence is will soon mount up to millions of dollars per day. But the largest is likely to be excess deferred infrastructure costs- how much extra it will cost our children/ grandchildren to incorporate rail when it becomes inevitable later this century (no planners have yet refuted my assertion it is inevitable) compared to the cost of incorporating it now when, as you say, we have a unique and ideal opportunity to achieve this. It will be interesting to see what analysis our transport planners have put into this (based on previous performance I doubt any has).

  2. John Rankin, 28. March 2017, 12:37

    @Glen, yes keeping the Golden Mile for buses makes a lot of sense for a lot of reasons. I have an open mind on whether rail should go on the northern part of Lambton Quay, to a stop at or near Farmers / David Jones, then run to Jervois Quay (probably via Panama and Hunter Streets).

    Kerry Wood has illustrated one way this could be done. This assumes Lambton Quay becomes a public transport and pedestrian mall (with provision for delivery vehicles).

  3. Ian Shearer, 28. March 2017, 13:16

    Very sensible John and Glen.

    But Glen’s comment: “…. transfer should be eliminated at the station to provide a seamless rail network” would not be a sensible option. It would be a nonsense for all Metro trains arriving at Wellington Station to continue on to the airport. That would require many platforms at the airport and multiple tracks or many passing places – clearly not possible through Wellington.

    Once we accept that MOST passengers will need to transfer onto the ‘airport service’, we then see that the difference between ‘MOST passengers transferring’ and ‘ALL passengers transferring’ is minor.

    Acceptance that all passengers will need to transfer to any airport service, opens the door for acceptance that a separate light rail system could be used for that purpose. The design of transfer facilities including weather shelter etc is clearly very important.

  4. Kerry Wood, 28. March 2017, 14:05

    I should have shown on the layout plan that Bed is buses, Green is light rail.

    Apart from the northern section of Lambton Quay, it makes remarkably little difference to motor traffic capacity: the lost space is mostly car parking. Even the Quays are less affected than you might think, because there is room for the proposed Frank Kitts Park stop in the existing parking lane and footpath (pushed into the park a little), and narrower light rail and traffic lanes will restore most of the lost capacity.

  5. KB, 28. March 2017, 14:19

    Is this article premised on the notion of 6000 trips per hour on the golden mile? As soon as you move past the golden mile, doesn’t that figure drop drastically to the point that rail being cost effective is undercut (I dont know, I’m asking?)

    Also, it doesn’t sound like this light rail plan services many suburbs, going from CBD to the airport via Kilbirnie – does it mean people who dont live near the limited train route (eg most commuters) would need to take a bus further along the train line to complete their trip, thereby turning what is currently one trip on a bus into a two stage train trip and bus trip? Isn’t this idea introducing friction into the public transport network rather than reducing it? Is that likely to increase or decrease usage?

  6. Ben, 28. March 2017, 16:26

    Melbourne’s huge pedestrian Mall works really well with trams going back and forth. It is quite a vibrant place to be.

  7. Keith Flinders, 28. March 2017, 19:01

    KB. If you study the new bus routes, which come into operation next year, many bus users will need to change buses not only once, but for some twice. Few are aware what is in store from them, others are considering going back to using cars. Shuttle buses will operate in Miramar, meaning those without a Snapper card will need to pay for 2 sections to get to Kilbirnie where currently it is a single zone.

    Changing buses at transit locations will not be done under cover, and in some places will require crossing roads.

    Get hold of the “Easier, Smarter, Better” patronising (no pun intended) publications put out by the Regional Council. Having all the buses in the region painted in the same livery is seen as a ‘must have’ as part of the changes.

  8. Kerry Wood, 28. March 2017, 19:36

    KB: good question

    The 6000 trips an hour is the existing buses at the Supreme Court. Capacity falls off fairly rapidly after that, which is normal enough: capacity design has to be for the busiest section. That’s OK: trams don’t have to be full to justify their existence, any more than buses do.

    Light rail might be initially configured for 4000 passengers an hour, plus a generous margin: too many new systems have found themselves leasing trams (or overloaded) because they had not allowed for a large enough rush of new passengers.

    The other 2000 passengers an hour (these are very preliminary figures) would be on the remaining buses, collectively running say one bus every two minutes (about half the present interpeak service).

    In Wellington it is impractical to time light rail for connections to individual trains, so the initial system might be say a tram every five or six minutes. The initial trams might be shorter than planned, then lengthened to suit growing demand (cheaper than additional trams).

    If a tram to the airport every five or six minutes was thought too generous, every other tram could be terminated at Wellington Hospital, or perhaps Kilbirnie.

  9. Glen Smith, 28. March 2017, 19:36

    Ian Shearer. Why would you want all trains to travel to the airport? Most incoming trains would continue to terminate at the Station using multi-carriage units (the existing Matangi units) to service the current demand from people who want to travel to and from the northern end of the CBD. Only one line would be diverted along the Quays using new specifically designed trains of likely around 45-60 m length (due to the physical constraints on station platform length along the Quays – hard to be precise since the planners who should be looking at this haven’t bothered to do their jobs). These trains would be interspersed between the trains terminating at the Station to service people who wanted to travel to the southern CBD or beyond. The frequency of these would be altered according to demand. They would likely alternate between Hutt and Kapiti lines (likely not Johnsonville due to low volumes – modelling required for this). Commuters could either specifically catch the right unit from their point of origin or transfer at the station.

    KB. No transfer at any interchange is required for anybody who wants to go to the CBD from any bus or rail corridor. This is one of the huge advantages of a dual bus/rail system (also used in many overseas cities such as Sydney and London). They complete their trip from start to finish either by bus (along the Golden Mile) or by rail (along the Quays). The only people who need to transfer are across town commuters whose origin or destination are away from the rail corridor but who want to make use of its rapid across-city dedicated corridor. And these would likely only need a single transfer (unless both their origin and destination were away from rail corridors in which case they would require two- this would likely be a small minority- again it would be nice if planners did some modelling).

  10. Chris Calvi-Freeman, 28. March 2017, 21:36

    Glen, I suspect you’ll find that current rail safety rules will not permit light(er) rail units operating on the same network as heavy rail (Matangi) units and freight trains, but I don’t profess to be an expert on this issue. Then there’s the requirement to have doors at railway platform height and others at street kerb level, or else a light(er) rail network with high platforms through the CBD and out to the airport or Kilbirnie – not impossible, but not very attractive in the street environment and not very convenient for bus/light rail transfers. If only a portion of the trains that ran from Kapiti or the Hutt Valley were to through-route through Wellington CBD, many commuters would still have to change trains at Wellington station, as the existing Metlink service runs a combination of express and all-stops trains.

    Nothing should be ruled out at this stage but I suspect you would get more traction (npi) if you didn’t bang on about “planners…who haven’t bothered to do their jobs”. Light rail is in no-one’s current formal workplan so I’m not sure who you think these planners are. If we can get a proper study going, that will change of course.

  11. Ben, 28. March 2017, 21:49

    What about tram-trains which are light-rail public transport systems where trams run from an urban tramway network to main-line railway lines shared with conventional trains. I believe St Kilda’s in Melbourne has one of these lines, where at one stage the tram is on the tram tracks in the road and then switches to traditional railway line with stations along the way.

  12. Andrew, 28. March 2017, 22:01

    Ben, the LUAS in Dublin does that. Out in the ‘burbs, it runs along an old rail corridor, then changes to rails in the road. Dublin is like Melbourne; pretty flat.

  13. Kerry, 28. March 2017, 22:14

    Glen: The only tram-length ‘formula’ that I know of is trial and error: pick a route, pick a length and see how you get on. It is not just a question of block-length, unless you are willing to either buy-out all vehicle access rights, or compromise on platform height so that vehicles can cross.

    I have found that 65 metres is tricky in places, with several chosen locations close to the limit. The limit would be 50m for a stop in Featherston Street. In practice 72m is claimed as the ‘longest tram in the world’ and FIT is now using 63m, so it looks reasonable.

  14. Glen Smith, 28. March 2017, 22:58

    – Brent Efford, who does have expertise in this area, has provided convincing evidence that ‘lighter’ units can run on the existing network lines. The fallback would be a separate ‘lighter’ network with exchange at the Station
    – I specifically designed my proposal with a ‘station’ design so that platforms could be kept at the same height as existing network platforms
    – the logistics of express/ all stops trains along with through trains would have to be worked out. It may be that only a proportion of units to the airport would originate from the Hutt/Kapiti lines with the rest running from the Station to the Airport (or Kilbirnie – there would be a number of options
    – the fact that rail isn’t on anyone’s workplan begs the question as to why this is the case when across town motorways have been examined in minute detail at the cost of tens of millions. The public have repeatedly over many years indicated they want a high-quality rail-based public transport solution that will solve ever-increasing congestion problems and have consistently voted for this in elections. They want planners to do their jobs and examine options for this. You and other councillors are supposed to be the public’s advocates. Doesn’t this bother councillors? Don’t you think it should be on the workplan so that proposals can be presented to central government for consideration of funding? Don’t you think it is time for local body councillors to go in to bat with some fortitude for the public that you represent (as Len Brown did in Auckland)?

  15. Chris Calvi-Freeman, 29. March 2017, 8:14

    Glen, this is precisely why I am advocating a proper study. Stop criticising those who share your motives, for goodness sake. How would you feel if I banged on about the questionable medical advice I’ve received from time to time?

    Ben, this issue would be covered in the study.

  16. Glen Smith, 29. March 2017, 8:31

    Thinking of logistics, the easiest solution might be to have all Hutt and Kapiti peak- hour trains that are going to the Station shadowed (maybe a couple of minutes later) by an across-town unit. The net effect would be as though you had a larger train (say 8 rather than 6 units) with the across town section separating at the Station. Having the train to the Station going first would also limit the problem of commuters who only wanted to go to the northern CBD occupying the more limited capacity of across town units (which I suspect would quickly reach capacity). Additional across town units starting at the Station could be added to service commuters who were heading to the airport but arriving by bus or rail from Newlands, Johnsonville or the Western suburbs

  17. luke, 29. March 2017, 10:07

    seems a lot easier to have light rail south of the station and heavy rail north of the station than making light rail compatible with the heavy rail network. There is nothing wrong with transfers if the frequency is one every 5 or 6 minutes and no fiscal penalty is incurred.

  18. John Rankin, 29. March 2017, 10:37

    @KB and @Glen: I think you are getting to the nub of the challenges we face. Here are some leading questions, with possible answers. (There’s a book to be written, but I promise to be brief.)

    Are the core customers peak hour commuters from suburbia or people along and within the catchment area of the urban line, travelling at any time? If the focus is commuters, then you’d rather give them a one seat journey wherever possible. If it’s all day travellers, you’d rather put a line connecting the major destinations, like the CBD, educational institutions, hospitals, shopping centres, and sports facilities. At the moment, benefit calculations are based on travel time savings, so favour longer distance commuters over local trips. The models give more benefit for a traveller from, say the Kapiti Coast, than for someone taking a local trip from, say Miramar to the Kilbirnie shops. So that means the BCR for an urban service only works if the service targets all day ridership, rather than the peaks. You need several short trips to get the same benefit as one long trip. Things like level boarding at all doors become critical to success, to minimise the dwell times at stops. Some commentators argue that this way of modelling benefits is especially hard on people like sole parents and low wage workers, so has built in a middle class bias, but that’s another story.

    The models also penalise transfers from one service to another, so have a built in bias towards one-seat trips. So how should we approach transfers? As @KeithFlinders points out, the way service connections are currently designed leaves much to be desired. Taking the railway station as an example, a successful connection requires the urban rail service to do 2 things. It needs to be frequent, say every 5 minutes during the peaks, so that incoming train passengers wishing to connect have to wait on average 2.5 minutes (just time to grab a coffee). And it needs to be reliable, so that tram passengers wishing to catch a train have certainty that they will make their connection. This is one reason light rail needs a dedicated right of way with priority at intersections, so it can’t get held up in traffic. A 5 minute light rail service has a capacity of about 5000+ passengers per hour.

    But we also need to ask what is the price we pay for a one-seat journey? The piper always gets paid (AKA the no free lunch rule). In any public transport network, for a given cost you get to choose either low frequency services with a high proportion of one seat trips or a high frequency service with connections. You can’t have both high frequency services and one seat trips, unless you are prepared to throw money at it. This is purely down to the geometry of networks — you can’t fight physics. We can see this in Wellington: lots of one seat options, but low frequency outside peak hours.

    For an urban light rail service to stack up economically (ie get over the BCR threshold), it needs to aggregate demand to connection hubs along the route, throughout the day. I’m aware of at least one study showing that self-driving vehicles will be great at doing this, acting as feeders for urban mass transit corridors and essentially eliminating congestion. Meanwhile, it’s a combination of park and ride, walk, feeder buses, and bikes. If you are Swiss, connections incur zero transfer penalty — the timetables are synchronised so that passengers change from one service to another during the standard dwell time at stops. And for many commuters, they are happy to take advantage of park and ride, transferring from their cars to the train. It’s not that people don’t like transfers, it’s that we don’t like transfers which make us feel like unloved parcels.

    So the choice boils down to this: do we want to focus on providing one-seat journeys for peak hour commuters, or all day every day services that take advantage of connections. For a given cost, we can do one or the other, but not both. In practice, of course, it’s a spectrum and we have to decide where on the spectrum we want to be.

    My personal view, and others’ mileage may vary, is that I’d like Wellington to be a city where living a rich, full, free and empowered life just doesn’t require a car. So I lean towards the all day, every day, connected service option.

  19. TrevorH, 29. March 2017, 12:00

    It seems everyone has their train set out and is wrestling with timetabling the various branch lines. How about his for a thought? Mass transit between the periphery and the CBD will be irrelevant in the future. The CBD itself is an archaic concept. We will increasingly live in villages, close to our workplaces, as many of us do now in the Eastern suburbs. More and more of us will also take advantage of technology that allows us to work from home. So goodbye to the uncomfortable cattle-trucks that convey us to and from Lambton Quay. Our main transport need will be to get through and out of Wellington quickly and efficiently to the wider region and beyond. I for one won’t miss mass transport, especially in the flu season.

  20. Glen Smith, 30. March 2017, 9:21

    Chris: Apologies. I have no problem with you or other councillors, just the process. It is hard to escape the impression that, before consultation with the client (the public), ‘four lanes to the planes’ has become a fait accompli but that the decision has already been made that any high quality rail spine won’t (in your words) be in your lifetime. The result will of course be a disastrous, and expensive, increase in congestion. Councillors can’t, with any integrity, support this sort of unbalanced development. Unfortunately sensible development requires central funding and the Government and the NZ ‘Transport’ Agency are obsessed with roads. Fortunately local bodies are in a strong position and need to take a firm stance. Best wishes with the study.

    Kerry: Looked at your route and the obvious advantage is that the northern section runs more centrally than the Quays (although this route would also be serviced by buses and the scatter diagram of rail final destinations- which I think I have sent you- shows that while the accepted figure is that people are prepared to walk 400m from a high quality PT corridor, in Wellington it looks more like 800m. The whole of the CBD is within 500m of a Quays rail corridor). The disadvantage may be achieving a dedicated corridor (or corridor of high enough quality to run a seamless rail network), slower transit time (a key demographic target is across town commuters) and political difficulty in removing parking/ traffic. Overall I favour a direct Quays route from the east of the station but am open to persuasion based on the study findings.

    John. A very interesting look at the conflict between affordability/ practicality and the need for transfers vs one-seat trips. I wonder whether a rail spine to the airport would be a unique case due to the more constant and high commuter numbers from steady plane arrivals over the day carrying passengers who are generally prepared to pay a small premuim for a regular high quality service. Also I suspect, with dual bus/ rail across town corridors, a good balance can be reached. I specifically included a Kilbirnie Hub Station with feeders from the Eastern Suburbs. This would allow a mix of one-seat trips from the different suburbs by bus along the golden mile (so as not to distract commuters who are put off by a transfer) intermingled with feeder buses to the rail spine that could maintain enough frequency to ensure a regular attractive service. Again the study findings will be interesting

  21. Mike Mellor, 30. March 2017, 11:08

    I’m all in favour of JR’s network option. Done right, that can meet more needs than over-focussing on “one-seat” options – but doing it right is hard (but by no means impossible, as milions of Swiss and Germans demonstrate).

    Looking at the network rather than individual routes also means that we can consider what I regard as the over-focus on the airport. While it’s important, it’s basically just shorthand for the eastern suburbs, and there are other pressing demands to be met. For instance, Cobham Drive inbound is at least as congested in the evening peak as it is in the mornings, largely with what appears to be the growing number of people commuting from the growing Miramar employment hub rather than airport traffic.

    We need to be addressing such needs, noting that current public transport is well-nigh useless for what have traditionally been seen as counter-peak flows but aren’t any longer – the buses are so focused on getting people from the CBD that services from Miramar reduce in frequency just as the peak begins.

  22. John Rankin, 31. March 2017, 10:02

    @Glen and @Mike: the idea of the airport supplying “base load” for rail is an interesting and useful one. How many rides per hour can we reasonably predict? 500 pass/hr? More? Less? And looking at the eastern suburbs, when planning the first line, modelling ought to include a comparison between an extension to Miramar and feeder services into a Kilbirnie hub.

    I tend to think of the airport, hospital, Massey University (Wallace St campus) and Victoria University (downtown campus) as the destinations most likely to generate the off-peak base load needed to support a high frequency, all day rapid rail service.

  23. Mike Mellor, 31. March 2017, 17:51

    John, Glen – off peak there are curently 13 “ordinary” buses an hour to/from the Miramar peninsula, plus 3 Airport Flyers. In my experience they tend to be reasonably well loaded, so I don’t think supplying a base load will be an issue.

    What may be an issue is that when the new bus network comes in, those 13 buses to the CBD will be reduced to 7…

  24. Keith Flinders, 1. April 2017, 19:50

    On the other hand, Mike, all 7 buses will sport the new livery, and as now appears likely they will all be diesels. Why worry about what users require and the environment?

    Wellington, the only 1st world city, post the Paris Climate Change accord, to ditch 100% pollution free electric buses.

  25. Neil Douglas, 10. April 2017, 20:07

    Keith, Spot on comment.
    I despair at GWRC’s decision to pull down the trolley bus wiring without waiting to prove that their proposed and expensive battery buses will actually work. And, I truly despair at the WCC which ‘owns’ the asset on behalf of city ratepayers and has meekly caved in to bullying GWRC pressure in allowing GWRC to pull the wires down.
    We need a air pollution survey on Lambton Quay, Willis, Manners and Courtenay Place by independent experts that is undertaken between 4 and 6pm to understand the level of health risk on these heavily peopled streets from breathing in diesel fumes.
    GWRC and WCC politicians should be made aware of the increased health risk in having more and more diesel buses choking our city streets.

  26. Chris Calvi-Freeman, 10. April 2017, 21:49

    As a member of the LGWM governance group, I presented the case, at its meeting last week, for LGWM to do justice to a complete investigation of light rail as an integral part of the project from this point forward. The LGWM officers have now been directed to report back on how this can be made to happen.

    Meanwhile WCC councillors and staff are working closely with our GWRC counterparts to help ensure the new bus network (July 2018) will be as good as it can be.