Perhaps more than any other post on this blog, the one which really got me fascinated by Auckland’s transport future and convinced me I “wanted in” was a post by Nick R about how driverless trains – the kind used by Vancouver’s Skytrain system that I’m so fond of – could have a role in making rail to the North Shore far more affordable and feasible than perhaps we have ever thought before. There are some key elements to what Nick calls “driverless light-metro”, which make it such an incredibly appealing transit technology:

  • The driverless operation means that the connection between frequent and operating cost is broken (you don’t need to add a driver for every train you add). This allow off-peak service frequencies to remain high, shorter but more frequent trains to be run and operating costs of the system to be kept pretty low. Vancouver’s Skytrain system, I have heard, makes an operating profit.
  • The linear induction motors, the lighter vehicle weights and the technical details of these trains allow for sharper bends and steeper gradients than would ever be possible with conventional heavy rail. Nick’s posts on the technology suggest that 1 in 10 gradients are OK (the CRL is really pushing the envelope at around 1 in 28), while 35 metre radius bends are also possible – yet again much sharper than for conventional heavy rail. A more forgiving geometric requirement means much much cheaper construction cost.

Essentially, a driverless Metro is far cheaper to build and far cheaper to operate than conventional heavy rail. It almost sounds too good to be true – so what’s the catch?

Well effectively there are two main catches. Firstly, because the trains are driverless they need to be operating on a system which is completely grade separated and completely protected from pedestrian intrusion on the tracks. Secondly, the highly specialised traction technology and the less forgiving gradients mean that light-metro tracks are pretty much passenger service only (no freight) and also limited to the particular type of train you run on them – so no inter-city passenger trains or future EMUs running on these tracks. Just the driverless light-metro trains.

These restrictions create an interesting conundrum. While there’s a highly compelling case for all new rail infrastructure to be in the form of a driverless metro, for the far cheaper construction and operating costs, because we have an existing heavy rail network, which we run freight trains along and which we are also investing heavily in maintaining/upgrading as conventional heavy rail, we’re left in a tricky situation of wondering whether, and how, this fantastic technology could be used in Auckland.

Nick’s suggestion was that the North Shore Line be constructed as a Light Metro, operating pretty much independently of the existing network, with possible future extensions along SH16 and SH18 to form some sort of northwest rail loop. The case for rail on the North Shore being constructed in the form of a light-metro is, I think, compelling. Firstly the harbour crossing itself is going to be far far cheaper than for conventional heavy rail (Nick has pointed out that it could sit underneath a road tunnel or potentially even under the existing harbour bridge as unlike conventional heavy rail it would be able to handle the gradient). Secondly, one would imagine that it would be much easier to turn the Northern Busway into a light-metro line than into a conventional heavy rail line – once again because of the more forgiving geometry of the light-metro technology. With a study recently estimating that a whole heavy rail line from town to Albany up the busway being approximately $2.5 billion in cost, a light-metro line may well be significantly less than this (very significantly less if you can sling it under the existing bridge).

What has thrown a few “spanners in the works” of this plan over the past couple of months has been the general thinking of us bloggers around future operating patterns for trains once the City Rail Link is completed. In particular, the general agreement that linking up the western line and the eastern line via the CRL and the North Shore with the southern line via another tunnel, would create the most logical and best long-term operating pattern for trains passing through downtown Auckland. That creates an outcome something like this, as nicely illustrated by Patrick’s post on the matter:  This operating pattern has some hugely attractive attributes:

  • By effectively creating four independent lines into the city centre (both directions on both lines) you have a simply huge amount of passenger capacity. If you ran 24 trains per hour each way along both lines, for example, you’d have nearly 100 trains per hour bringing people into central Auckland – around 75,000 people per hour with 750 passengers on each train.
  • You create a really logical route structure for Auckland’s whole network (setting aside the question of how we deal with Grafton station). There’s a basic north-south line (the blue one) and a basic east-west line (the red one). They cross over in the very heart of Auckland’s city centre.
  • We do away with the incredibly slow bend around Vector arena (though I’m sure you’d keep the tracks there, at least you wouldn’t need to use them for regular service).

Of course, by linking up the Southern Line with the North Shore Line, we’ve just created ourselves one heck of a headache when it comes to our idea of that North Shore line being a driverless light-metro. Or have we actually opened up an opportunity here?

What if we tried to make that “blue line” above fully driverless Light Metro? Let’s explore that idea.

If we remember back to the start of this post, the two big restrictions for driverless Light Metro is that it can’t share track with freight trains and it can’t share track with any other kind of passenger train. Effectively, it has to be its own independent network. That does create use a few headaches. But potentially they’re not impossible to solve. Let’s just say we built the line in blue below as a light-metro line: Yes, yes I know there are issues, but first let’s look at the positives. We probably have a cheaper construction cost for the Airport Line due to the easier geometry of Light Metro. We also have much lower operating costs. There’s a direct line between the North Shore and the airport, which would probably generate quite a lot of patronage and would certainly ease traffic on what’s a pretty big “through movement” at the moment (Waterview Connection eases this pressure on arterial roads but not on spaghetti junction except for people up around Albany who may use SH18/SH16).

The main issue, obviously, is that we have existing sections of track along this alignment – from Parnell right through to Onehunga and Otahuhu. However, if you add in the conventional rail network which would provide the main “south/east-west” connections, there actually isn’t much overlap between the lines at all – just between Westfield and Otahuhu by my calculations: The other key consideration is, obviously, rail freight. But from what I know the Newmarket-Westfield section of the southern line isn’t really used much by freight trains (they prefer the easier gradients of the eastern line), so the only section which would require side by side conventional and light-metro tracks would be between Parnell and Newmarket, unless some other solution can be found to send freight out west via the Avondale-Southdown line (including the Onehunga to Southdown link which isn’t shown above).

I actually kind of think all of this could work, with Auckland ending up with two completely independent rail networks. While that has some disadvantages in terms of route flexibility and the need for transfers for trips from south of Otahuhu to Newmarket (for example), I think the cost savings (both capital and operating) which would arise from being able to build both the North Shore Line and the Airport Line (at least the northern link, the eastern one is something that probably required a bit more thought) as Light Metro lines would probably run into the many billions of dollars.

Which means it’s something worth looking into. Driverless light-metro indeed could play a very important role in Auckland’s rail future.

Share this


  1. Peter-

    Love the hanging under the existing bridge idea.

    That knocks what, a couple of Billion off the cost.

    Might even get support from Maggie Barry?

  2. Keep in mind that the North Auckland freight line runs along the blue line from Westgate to the Port, and can’t be re-routed through the CRL tunnel due to the steep grade.

    Of course, there’s a good chance that line will be mothballed at the end of the year…

      1. Dammit, I thought I had edited that to be less redundant, less factually incorrect, and more substantive. I meant Westfield, not Westgate.

        Here goes, again.

        I’m on the fence about mixing light metro in with the existing rail system. On the one hand, other cities do operate a mix of light and heavy rail (eg LA). And there’s no reason why we should limit our investment options based on decisions about track gauges and train routes made over a hundred years ago – technology’s changed, and we should act on that basis.

        On the other hand, mixing the two systems could also have the effect of limiting options for developing new routes in the future. In effect, the two lines of the cross would be a separate train systems that intersected at Aotea. Looking at your last map, it would probably prevent a direct service between the airport and west Auckland. Which would be a shame, as there’s potentially a lot of latent demand for good PT connecting those areas. (This should be implicit in the priority placed on linking SH20 to SH16.)

      2. Freight trains between Westfield and the North Auckland Line could always travel via the eastern line then Parnell, until such a time as the Avondale Southdown line is built.

        It wouldn’t be too hard for people from the west to get to the airport under this system. Just catch a train into town and transfer or catch a train to Onehunga once the Avondale-Southdown Line is completed, and transfer there.

  3. – I like the idea of the light rail to the North Shore, since it would be easier and probably cheaper than heavy rail. I’ve supported this idea before.

    – I’m not sure that the plan to build lines in a big ‘X’ is really feasible – at least for heavy rail. It would mean building two CBD tunnels (and as we can see, even building one is hard enough), and as you say, it would remove Grafton from the network. Given that all we really gain from it is a station at the University, which will be fairly well served by Parnell, Britomart and Aotea already, it seems a bit of a waste.

    – I’m also not sure that duplicating the existing network is a good idea. Duplicating the Parnell – Newmarket section would also mean duplicating the Newmarket tunnel. I’m also not quite sure how you’d fit in light rail tracks through that section. I don’t see the point of extending light rail to Otahuhu, when there will be a perfectly good electrified train line already there to cover the same route.

    It seems to me that we should be thinking of light rail for routes that the existing heavy rail network doesn’t cover – the North Shore, Tamaki Drive, Dominion Rd, perhaps along the NW motorway, and perhaps lines to Botany and Howick. These lines would be able to connect to the existing heavy rail network at various points.

    1. Sigh. It’s not light-rail Sean. Or anything like light-rail actually.

      Light metro is completely different. It’s high volume, high capacity, high speed rapid transit. As different to dinky trams as you can possibly get.

      1. Thanks for the link Louis. The thing is that if the Light Metro was built with long enough platforms for six car trains, I don’t actually see how the capacity is “medium” at all. 6 car trains at 90 second intervals (closer frequencies than possible with driver operation generally) is huge capacity.

  4. Well Newmarket is the big problem isn’t it? So much to thank the last round of privatisation for; including allowing Fay Richwhite to get hold of the nation’s railway land and flog it of to shitty developers and then defraud the public of the windfall profits and doing nothing to even maintain the existing infrastructure. Amazing that we’ve got that idea being forced on us again….. But I digress, great idea, and do able if:
    1. the NAL is trashed by the gov permaently so freight isn’t an issue or
    2. if Avondale Southdown is built so freight can bypass this route or
    3. Or if one track is left as conventional rail serving the western line and the other two replaced with light metro…

    Any other ideas?

    1. Fay Richwhite didn’t get any rail land, as the land always remained with NZRC. It was NZRC who sold off rail land throughout the 90’s and early 2000’s.

  5. I do like the idea of light metro but I have some concerns about it.
    1) In the future should it ever be decided that new route north to Wellsford from the North Shore line is needed to serve intercity trains and freight then it would be impossible to build. The reasons for that to be create a direct connection to Northland. BUT that would mean we have freight trains running through the CBD which is not at all a feasible outcome unless we did some expensive connection from along SH18 and onto the Western line line and then down the ASL, and in any case

    2) Obviously the ASL would be mandatory if a pattern like you have proposed is chosen as otherwise freights would have nowhere to go without using the Parnell line.

    3) The Airport line is likely to built long before the North Shore line is built, therefore you’d have a line running into the CBD with nowhere that it could connect to.

    4) I think personally think rail to SE Auckland i.e Botany should seriously be considered. A busway is hugely inferior in this particular case. I guess you would also have a doubling up problem there , and you’d only be able to do it as a separate metro line in addition to the current NIMT tracks due to the freights. Or you could build a separate, but very expensive route under Remuera Road.

    1. Triple tracking (two for light metro and one for freight) between Parnell and Newmarket is not impossible. There is already another tunnel for a single track.

  6. As kiwirail own the corridor I can’t see them giving it up easily as it gives them flexibility and is also required for freights heading north. At the very least that would likely require the SAL to be built which could easily cost $1b and is going to be difficult for other reasons. I’m also unsure if we ever could hang the crossing under the harbour bridge, even if the grades allow it and I suspect the NZTA would be very sceptical about it.

    Those organisations will of course change over time but I think their fundamental issues with it will remain.

    1. Nope. They would go via Parnell in the pre Avondale-Southdown scenario or via the Avondale-Southdown line once that is built.

  7. How about the metro: Takapuna, Akoranga (interchange), Wynyard, Hobson St (above ground, but with free transfer to Aotea at Victoria St.), Dominion Road, Onehunga (interchange), Airport.

    Now you have something built largely, or possibly entirely above ground, which given the (lack of) weight it would be carring compared to a conventional bridge would make it massively less expensive.

    There is plenty of space on Hobson St and it is still fairly central; There is an old offramp at Nelson St to get part-way over the CMJ; Dominion Road is crying out for a higher capacity solution, and a station at Bellwood Ave would provide another access point to Eden Park; There is a corridor on the South-Western some of the way to Onehunga; And the airport routing could be more flexible.

    (if you like I can put together a post :))

    1. Sorry you’ve made the same mistake as Sean did above. This is not light rail, this is light metro. It is absolutely incompatible with operating at grade so could never run along Dominion Rd.

      Maybe we need a new name? Driverless Rapid Transit?

      1. I had no intention of running it at grade along Dominion Road – it is just a viable corridor for the patronage.

        (and I’ve got myself excited, so I’ll put together something this evening to illustrate the concept for fully)

        1. I’m hoping to achieve it with some long columns to smooth out the gradient (they may cause a few visual envionmental issues I will be fair) especially on approach to Onehunga. And I’m taking the 1 in 10 over short sections (<500m) you've mentioned as reasonable, especally where it's around stations (slow vehicles will be more acceptable on station approaches). But I'll keep an eye on gradients as I go through 🙂

        2. My convern wasn’t the grades, it was the phenomenal expense of tunneling, vs. destroying the Dominion Rd evironment we a huge long concrete viaduct. Maybe we can use elevated along the motorway or through places like Albany, but not down Dominion Rd!

      2. Sounds like a better name. I wonder whether North Shore rail and then in the future Northwestern rail would be the best place for DRT. But it would obviously require a difficult tunnel at Aotea station. In the mean time it would probably be okay just for the North Shore line to terminate a new pair of platforms underneath Aotea station. If we assume same capacity of Britomart at 20tph then that should be fine for one line only.

  8. If the construction and operational costs are better for Driverless metro then it is worth considering for totally new rail lines and new PT corridors. The argument for the North shore line is fairly compelling and I would suspect it would make a fairly good option for the Airport line and for the AMETI corridor as well. I dont see a massive problem having 2 independant rail networks as one would allow frieght to operate on it and the other would be specifically for PT.

    The cost of a second cross town tunnel seems to be the biggest weakness in this plan.

    1. You will need a second tunnel for North Shore rail no matter what technology you choose. It simple can’t hook into the CRL.

      1. yes, but it dosnt have to be a 3km+ cross town TBM bored tunnel. There are other options to link a north shore line into the Eastern line via a much simpler and shorter 1.5km mostly cut and cover tunnel.

  9. It would be unfortunate if we went with two independent rail networks, as you lose economy of scale. The only saving is on driver wages, but that’s not a biggie, and it woulod be offset by the fact that you would need duplicate facilities and all new equipment. Then there’s the inability to run freights and so on. So one pro, and a long list of cons.

    I disagree that there are operational advantages such as more frequent trains or faster acceleration. Such things are unrelated to whether or not the train has a driver.

    1. Not a biggie? Staffing will comprise 60-70% of marginal operating costs on our new trains! Twice as expensive as electricity and marginal maintenance combined. Cut out the staff and it costs only a third per kilometre to operate a train, or in other words we can triple the frequency for the same money.

      That is where the massive increase in all day frequency can come from. You don’t need peak hour crush loadings to justify running trains every five minutes, you can do that all day and most of the night seven days a week. The point is we will always have enough trains and infrastructure to run a very frequent peak service, but going driverless means we could afford to keep running those trains frequently all day instead of parking three-quarters of them in a stabling yard for 18 hours a day.

      Frequency is most certainly related to the driver, because the staff comprise two thirds of the costs you consider when deciding whether to run a train again or not.

      What facilities would you need to duplicate with two types of train technology in the region, that you wouldn’t have to duplicate anyway if we built a couple of new lines? Obviously with any new line you would need new equipment, regardless of the technology. Sure you’ll need new stabling and maintenance facilities if we built a North Shore line or a Botany Line, but again you’d need that anyway even if it was operated by standard Auckland EMUs.

      The inability to run freights is a potential con, but would we want to run freights on any brand new rapid transit rail lines anyway? Would we want to run freights through a harbour tunnel for instance, and through the city tunnel? I know they won’t be allowed in the CRL, so why would the be allowed in the next one? Given that we already have main north and south freight lines why would we need any more, apart from short links such as the ASL that link those two together more effectively?

  10. You’ll obviously still have to have someone onboard, as there are too many essentials that require a person, such as evacuation procedures and security, so the cost saving will only come from the difference in pay rates.

    Duplicated facilities in that with a single network you only need stabling in Albany, not an entirely duplicated workshops.

    And which expensive grade separation will be chosen for this second system? Elevated (which the public will never allow), or underground (which will make the CRL project cost look easy by comparison)?

    I just don’t see it happening myself.

    1. Nope you don’t need anyone on board.
      They have no body on board on the systems in Vancouver, Kuala Lumpur, Copenhagen, London Docklands and New York Airtrain just to give a few examples. Yes, totally unstaffed, except for some systems maintain roving security and/or customer service operators that make rounds of the stations. In fact the Copenhagen metro is entirely unstaffed: no drivers, no clippies, no station attendants and no security. Evacuation and security is handled from a control room, for example in Vancouver a team of up to eight operators monitor the whole system, compare that to sixty odd drivers they’d need to run their interpeak timetable. Huge cost saving.

      This isn’t pie in the sky stuff, these systems have been in daily service since the early 80s.

      As for grade separation, I would say why not use the kind we already have!

      The northern busway is entirely grade separated on the main section from Constellation to Akoranga by way of overbridges and underpasses, so just like that. It might need a bit more fencing or whatever to prevent intrusion, but the trains do have track intrusion detection systems. Platform screen doors are used in the newest systems, presumably that helps a lot. Otherwise in the CBD we’d be using a tunnel anyway, and a little elevated/viaduct might be appropriate for places like Albany or Westgate or where they are alongside a motorway… but generally it would be a ground level corridor grade separated the way our existing motorways and railways (mostly) are.

      1. Grade separation of driverless trains requires the whole route to be included, not just crossing roads. The Northern Busway is not grade separated in that manner.

        The overseas examples you give are not even close to being valid comparisons with going staffless on Auckland’s network. The environment and requirements under the railways act are totally different, and you’ll never see remote evacuation introduced in New Zealand.

        Not even Wellington’s automated cable car is staffless.

        1. The railway act is a piece of legislation and the government could change it if they wished

        2. You’ll never see a legislation change in New Zealand that relaxes safety. The culture is far too ingrained.

        3. Sorry Geoff but I don’t understand why you mean when you are suggesting that the whole busway would need to be grade separated. It is fully grade separated, even if it runs at ground level. Are you suggesting that it would need to be entirely elevated? Well that’s not the case in the overseas examples. Going back to Vancouver it is an excellent case for what has been proposed above: a long section of their Expo line runs at ground level alongside freight tracks in an old freight corridor, while on the Canada Line there is a long section alonside a motorway and across open ground to their airport, likewise at ground level. These are still fully grade separated, much like our busway.

          This video is a good example from Vancouver that shows a whole variety of grade separation methods, it starts of at an elevated station, runs along a ground level fenced corridor (with at-grade stations and road overbridges), up onto viaduct then into a long tunnel:

        4. Those tracks are not on the ground, they are still raised on a concrete structure, albeit a low one. But the bulk of the route is prominently elevated or underground, and the entire route is on a bulky concrete structure. Not cheap, and doesn’t exactly blend into the surrounds. That’s a lot of expenditure going into an effort to cut pay rates.

          We could achieve a lot more bang for buck by building more conventional railways, than sinking billions into massive concrete pours.

        5. They’re simply bolted to a concrete slab at ground level. A somewhat higher upfront cost than ballasted track but it means never replacing or cleaning ballast, tamping etc ever again.

          It the same construction as the busway so no drama. I really doubt we could get more bang for back out of conventional railways, due to the cost of providing gentle enough grades and curves. The difference in cost between bolting rails to concrete slab or laying ballasted track would pale in comparison to reconstructing the busway to suit 1 in 33 grades, for example. Meanwhile light metro tracks could be laid straight into the existing alignment without so much as a single cut.

  11. Instead of sending one of the blue line stubs to Otahuhu, why not send it sideways from across Panmure or Silvia Park to Botany, and eventually on to Manukau? You still get the same network connectivity with the red line, just one station further along.

    1. I guess I’m taking the AMETI busway as a given and thinking that any upgrade of it to rail in the next 30-40 years is incredibly unlikely.

  12. Just doing some rough maths on the ongoing savings only;
    Say 40 drivers @ $60,000 per year = $2.4 million per year. That is less 0.1% of the cost of the line. Not a huge saving in the scheme of things.

    1. Pretty sure to operate regular trains across the whole line for 18 hours a day, 7 days a week, you’d need a lot more than 40 drivers.

      Veolia charge Auckland Transport around $35 million a year, excluding pretty much everything except staffing costs. And that’s on a network probably not hugely bigger than the driverless lines proposed in my post.

      1. You can’t compare operating costs to capital expenditure.

        A line from Albany to the Airport would be 40km long, or 80km return. 40km/h is about the average speed in Vancouver, so at that speed one return trip would take a driver two hours. Running the line at five minute headways would therefore take 24 man-hours per hour. Multiply that by 18 hours a day, seven days a week and we need 3,024 man hours a week. At a price of $35 an hour to cover salaries and staffing costs thats $105,840 a week on drivers. So driverless trains would be a massive saving in operating costs that would allow us to actually run it every five minutes all day.

        1. I’m not trying to say it isn’t a good thing and wouldn’t save any money. I was just trying to bring some perspective of the size of the savings. Maybe my initial figures were too low but yours still only give just over $5m p.a. which is still only 0.2% of capital expenditure above.

          You can compare on-going operating costs to capital expenditure. The capital outlay can be converted to an on-going finance cost. Using a %5 rate the on-going saving correlates to $100m in upfront capital spending.

          Not much in terms of the total spend that we are talking about…

  13. Sure it’s true from that perspective, but that belies the value of minimising opex. The fact is the operating expenditure determines service levels, not the upfront capital cost.

    With a driven line we might run it every five minutes at peak, but cut that right back to half hourly during the inter peak, evenings and weekends to save costs. Much like we do now.

    With the same system driverless we could run it every five minutes all day every day for the same money. That’s what they do in Vancouver, a city of similar density and land use patterns to Auckland. Over there they actually cover their operating costs with fares, so after building the line it doesn’t actually require any ongoing funding at all.

    1. The saving on operating costs is significantly less than the cost of building long distance elevated structures or long distance undergrounding. Also, operating costs are not the total cost of running a train. You also still need somebody onboard and other people in a control room somewhere, and all of this qould require a lessening of the safety culture, which has zero percent chance of ever happening.

      1. Geoff, the value comes from the saving in marginal operating cost, not operating cost overall.And as we’ve noted above you don’t need elevated structures or undergrounding, the way we build or motorways and busways is perfectly adequate in Vancouver and other cities. This sort of system would be far cheaper to build in the first instance, and much more likely to have very frequent regular service all day every day in the second.
        You don’t need anyone on board, and they systems are much much safer than conventional suburban passenger railways. Vancouver has never had an accident in service, nor ever had a train pass through a signal despite running 18 hours a day for the last thirty years. The only incidents they have every had were both while being driven by humans in the stabling yards. How many times has Auckland had those sorts of problems in the last year alone? I’m not sure how shifting to an incredibly safe system with almost no potential for human error would be a lessening of safety culture.

  14. Just setting aside the driver issue for a moment, as with sufficient grade separation any railed system can be more or less automated, threre is the question of whether the savings in capex of a lighter system are not offset by the costs in fitting the new system to existing rail routes?

    Conventional rail to the Shore may be just as cost effective as a cheaper new system that also requires a rebuild on existing lines. Of course comparing these two competing ideas for the same project is exactly what the BCR model is good at doing: Apples with apples.

    Unlike comparing rural motorways and urban transit systems.

  15. Is the Vancouver Skytrain the only example of light metro which uses linear motor propulsion? Certainly London’s Docklands Light Railway doesn’t. Seems to me that without the linear induction principle the claimed advantage of being able to handle steeper gradients than heavy rail disappears (Swiss heavy rail successfully operates conventional locomotive-hauled mainline trains on grades of 5% by adhesion alone!). So are the advantages of introducing a separate and incompatible system in Auckland really there?

    1. The grade advantages are still there without linear induction, due mostly to lighter vehicles and more drive wheels. It’s like with trams, they can run three times as steep as heavy rail.

Leave a Reply

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