Recently the New South Wales Government released a long term rail strategy for Sydney. It makes for interesting reading. Like Auckland, Sydney is experiencing relatively fast population growth and expects to grow from its current population of 4.6 million to around 6 million by 2031. The map below shows the location of employment across Sydney in 2031 – highlighting that the city centre is expected to continue to be by far the largest area of employment:

Australian cities have made conscious efforts to try and decentralise their employment over the past few decades, but while places such as Parramatta, Macquarie Park and North Sydney certainly do have a lot of jobs – the CBD still stands out. For comparative purposes I think Auckland’s CBD has around 80,000 jobs, showing that as a proportion of population Sydney is a bit more concentrated than us.

Sydney’s rail system is a rather strange hybrid of inner-suburban, outer-suburban and inter-city trains all competing to use the same tracks and stations. A lot of different operating patterns are run, especially at the peak times, which makes the system incredibly complex and therefore fragile to failures. The diagram below shows an example of this on Sydney’s western line – detailing all the different conflict points:

Because of continued population and employment growth, Sydney’s rail system is struggling to meet growing demand. On certain parts of the network there are significant capacity constraints. Importantly, overloading of certain services leads to slow boarding times which actually ends up meaning fewer trains per hour can be operated than planned – further reducing capacity and further creating overcrowding. This is succinctly described below:

A useful diagram shows where crowding on the network is likely to occur and how severe it will be – without significant upgrades. Something like this would be interesting for Auckland’s public transport network (bus and rail) in 2030 or 2040 if the CRL isn’t built:

The main proposed solution to Sydney’s impeding rail crush is quite interesting and proposes to effectively split the network up into three parts – a metro-like “Rapid Transit” for single-level trains, a suburban network and an inter-city network. This is shown in the map below:

There has been a long debate over whether the double-decker trains in Sydney are well suited to the role they play and it seems that the latest rail strategy quite cleverly says that the system needs to stop trying to be so much of a “jack of all trades” and actually start specialising a bit: much like how Paris has its Metro, RER and Transilien networks.

The rail strategy realises that implementing this long term vision is going to take time and will be expensive (I suspect the new cross-harbour tunnel would be the most costly part of the whole scheme). Therefore it outlines a staging of the plan, with many of the earlier phases around service improvements or relatively minor infrastructure upgrades (along with a big project here and there). This is sensible to ensure that maximum value is squeezed out of the existing rail network:

The really amazing thing about this strategy is that it’s promoted by a centre-right state government of New South Wales. Interestingly both Victoria and New South Wales have seen centre-right governments elected in the past few years with core parts of their election platforms being to improve public transport in Sydney and Melbourne. I wonder why Australian politicians of all political hues seem to understand the importance of good public transport, whereas in New Zealand our centre-right politicians (particularly in central government) seem stuck in the stone age on the issue. I suppose in Australian cities enough people catch the train that they can’t be ignored – something that will happen in Auckland over time.

An Auckland version of this rail strategy would be damn useful as well.

Share this


  1. The current NSW government will be cursed for generations IMO.

    (a) Creating the NWRL with a tunnel bore too small for Sydney’s double deck trains is inappropriate
    (b) The needed Western Express aka CBD Enhancement is cancelled.

    Probably enough for now.

    1. I’ve had this debate with a number of people too and arrived at similar conclusions to you. Why is it that a heavily suburban area (Rouse Hill) is being given “single deck metro-style” trains, but Parramatta, Hurstville, Hills District and the dense Inner West are relegated to suburban double deck trains? Seems like a bit of topsy-turvy planning to me if ever I did see it.

      1. That does look like politics, the right loving the no labour possibilities of metro trains so determined to not bore the extra 400mm diameter that would allow new lines to be able to take the high capacity, but manned, double-deckers.

        1. There are two very key reasons why the planners favour driverless single deck metro for *new* lines in Sydney:

          1) Driverless means much lower operational costs, allowing for high frequency and span of service. Most people don’t get just how massive staffing costs are, and how much extra service you could afford to run without them. It’s not a hatred of labour, its a hatred of not being able to run service because labour costs so much.

          2) Low profile single deck metro has a much smaller cross section than double deck heavy rail, and less constrained track geometry to boot. 400mm less diameter is not insignificant, although the difference between a Sydney double deck (needing 5m vertical clearance) and a single level metro would be a metre or more. A 5m tunnel bore is 20m2 cross sectional area, a 4m bore is 12.5m2 area. One metre less means a tunnel close to half the size. Consider a 10km long line, that’s 75,000 cubic metres less earth to move and replace with tunnel structure with a 1m smaller bore. When you’re building long new tunneled lines it can save billions on capital expenditure.

          Long story short, much much cheaper to build, and much much cheaper to run at good frequency. It’s not a perverse techno fascination as much as taking advantage of advanced technology to make their plans feasible. The fact that the existing lines remain double deck bears no relevance to the design of new lines. There is obviously little point in changing existing double deck lines to single, although making them driverless would be a huge gain.

  2. It is not just the Australian centre right governments who recognise the need to build public transport networks, for example the UK is proceeding with planning for High Speed 2 and the Crossrail project also continues under the Conservative led coalition. Think France continued to expand the TGV when the UMP was in power. We seem to have more of an American political approach. Does it sometimes make you wonder if vested interests are at play?

    1. Yes – vested interests all right. The previous transport minister happened to be National’s chief fundraiser. The Road Transport Forum (previously headed by a former National minister and now headed by a former ACT MP) happens to be one of their chief funders.

  3. Yes that’s a good question. I am guessing the answer is that it’s too difficult to change some areas over from double-deckers to single-deckers plus it’s cheaper to build new lines in a way that only needs to handle single-decked trains.

    Could be wrong though.

    1. Changing from double-decker to single is not really problematic, however, the loading gauge of double deckers is significantly different to single (if I understand correctly) thus changing from single to double isn’t easy to do as evidenced on the UK network which is hampered by loading gauge restrictions. Here in Sweden, double decker trains are run on the same lines as single deck trains no problem, but that is because we have very generous loading gauges.

      If they do build the new tunnels in Sydney to a restricted loading gauge that doesn’t really improve interoperability of the network and will segregate it and restrict rolling stock to certain lines. This is already bad enough given the Epping-Chatswood link cannot be operated by anything other than M-sets and newer. In summary, there should be no restriction (other than rolling stock availability) as to why current CityRail lines cannot be run with single deck trains, but of course given they’ve already got new double deck rolling stock on order, they’re not going to back track and convert older lines to single deck trains.

  4. Why does it sound like some people think double-deckers are the wrong thing to go for? I must be missing something? Surely the extra capacity is good, even with a greater space need in tunnels?

    1. It’s the loading/unloading speed of double deckers – same with double decker buses. Sure, they work fantastically well for longer distance trips, but for short trips they’re a nightmare. Sydney, being a hybrid system without another form of rail backbone as is the case in Paris, is hampered by attempting to run a metro-like system in the city centre with the same rolling stock also serving Penrith. It can’t really be done efficiently or effectively. There is a reason that in the “model” systems, be they in Japan or Europe, double deck stock is used only for longer distance travel with single deck stock fulfilling a more frequent stop model (often with more doors per carriage too).

      1. I can see that delay thing, but wonder whether that it isn’t worth the 50-80% extra capacity… plus, I am sure you could add more doors to a train design, and still end up with a 30-40% passenger boost compared to a single level car, even once you deduct the space lost by more doors?

        1. You lose a considerable amount of space for each of the doors due to the need for stairs too. Those stairs also slow things down as it funnels passengers through a tight area. The saving in dwell times are something like in the order of 30 seconds or more per station so would easily add up to enough to allow a number of extra trains to run which actually increases the overall capacity of the system.

          1. There’s obviously a diminishing returns issue here – but I still think there might be a way (at least theoretically, if starting from a clean slate) to produce optimised double-deckers which give more capacity at similar dwell times. They might only have 125% or 130% of the capacity of a single-decker, but that is still a 25% boost to your capacity!

            Since once you get up to your maximum frequency on a line, the issue can’t be solved that way anymore…

          2. All the suburban trains in places like Zurich are double decker and I’ve never experienced issues with dwell time, so it is possible to use these trains. My experience of the Sydney double deckers was poor, they seemed quite crushed and uncomfortable in comparison, which suggests the Swiss tunnels are bored bigger.

          3. My understanding is that there are bifurcated platforms at the main stations, is that correct? The purpose of those is to allow one train to be at the platform while another one approaches and is done because otherwise the dwell times would require frequencies to be cut because they aren’t safely sustainable otherwise.

        2. In Melbourne they looked at introducing double deckers about ten years ago. If I recall correctly the capacity loss due to increased dwell times was about the same as the capacity gain from the extra deck, so they didn’t bother.

        3. In reality you can’t usually have more than two doors per car in a bi level design. That’s because the doors are located at the ends over the bogies with stairs up and down to the two levels. Like Matt says, if you want to add a third set of doors in the middle with another pair of stairwells leading up and down either side you’d be cutting massively into your usable floor area.

          The thing with single level automated metro is you can run them at headways of 75, even 60 seconds. So they might have 40% less capacity per train, but you can run (and afford to run) two or three times as many trains. At the end of the day the metro would have greater capacity.

          1. You could keep the middle doors for the lower level only – then you need no stairwell, and only lose some seats (which you regain as standing space, so less / no capacity impact). As an engineer, I just don’t accept the “can’t be done” argument 😉 There’s always possibilities – they are just sometimes too expensive or have other drawbacks (in this case ones that would not affect capacity OR dwell time). That said, I accept that it may be an academic issue as there may be no appetite / no window of opportunity to do this.

          2. The lower level is below the level of the platforms so would needs stairs up to the middle but that would cause issues with the upstairs part due to ceiling heights etc.

            I saw a document which said that an 8 car double decker train has about 800 seats with standing space for another 400 so 1200 people all up. It said studies showed that any more people standing and dwell times blew out really badly and started impacting on how many trains could be run. An 8 car single deck train in metro configuration would have seats for up to about 600 people, but that people could get on and off faster due to extra doors. That means that the overall capacity is increased.

          3. Max, you’d still need a platform level door with stairs down to the lower level. The point is with double deckers there are three floor levels, one floor level is below platform height, the other above, and the doors are halfway between at platform height.

            Matt L, one thing with the comparison is that doubles get lots of seats but the standing area near doors is often quite limited. In fact you try not to have anyone standing because it conflicts with your critically limited circulation space around the doors and stairs. An eight car double with 800 seats and 400 standees would be incredibly crowded around those passageways. On the flip side, an eight car metro set with 600 seats could easily handle 600 more standees with people getting on and off easily.

  5. A couple of points:
    – Paris RER has double deck and 3 doors per carriage, but significantly longer carriages and presumably less curved platforms.
    – Only reason for putting the single deck in the NW is that it came later. Seems an obsession within the public service to have single deck. Cityrail are far too expensive for what they do, so not entirely baseless.
    – Nick R’s technical comments ignore the politics. Sydney wanted a line to the NW. Cityrail couldn’t do it without taking paths from other lines. The Liberals promised it anyway and this is what Sydney is left with.
    – Crewing is only a small part of Cityrail’s expense. Guards only cost about $100m, so imagine about $150m for drivers. Put that against the $2-3bn for the whole shebang.

    1. Crewing is a relatively small part of total expense, but it is the majority of the marginal cost of running extra services outside the peak of the peak. It’s a seemingly small distinction that most people don’t seem to grasp, but it is incredibly influential on the service you can provide.

      Consider the fact that in Sydney they have the rail tracks, the trains, the signalling etc to run a service on every line every few minutes, and they do do that in the peak. So why not run trains every few minutes 24 hours a day then? Because the marginal cost of staffing them is too high, they can only afford to run them at times when they are full or close to it. So staffing currently costs $250m a year. If they wanted to double the number of runs across the week to give good all day frequencies, they would need another $250m. That’s a quarter billion at the margin just to use the lines and trains they already own properly. So what happens, they run poor all day/all week frequencies.

      Take Vancouver on the other hand. They have less population and less density than Sydney, but run each of their lines every three minutes 18 hours a day. The reason they can afford to do this is purely because they run automated metro.

  6. I would contend that the poor off peak services have no basis in logic. $1/4bn – Hardly anything in the scheme of Sydney’s transport budget. Might cost something to run improved off peak services, but you are also increasing your fare revenue as well as increasing your benefit from the system. Of course, periodical tickets do reduce the farebox recovery from improved off peak services.

      1. They don’t see political advantage in doing so.

        Before 2005 there was 4/hour to Penrith in Sydney. Then there was a drivers shortage and they trimmed off peak services. They hired more drivers but didn’t reinstate services. Either they were wrong before the drivers shortage or after.

  7. I would say because there were no political points to be gained by doing so. Although that ignores the political points that were lost from the increase in traffic congestion. I am presuming that they just didn’t think of that.

    BTW, Wellington studied it here: page 72

    BCR of increasing off peak to 20 minute frequency = 1.1. What has happened with this? Nothing, hasn’t it?

    Wellington’s marginal labour costs are far more than Cityrail AIUI.

  8. One more thing, notice how as soon as Melbourne was privatised they started increasing off peak services. It has been similar with Sydney’s private bus services which have lobbied the govt for service increases far more than the STA have, at least from what I can gather.

  9. In inner Sydney, with existing signalling, practical line capacity is about 20 per hour for double deck trains and 24 per hour for single trains (assuming shorter dwell time with three door cars).

    The theoretical standing capacity of double deck cars cannot be achieved in practice, because the vestibules become over crowded long before the upper and lower decks are full.

    The upshort is that single deck service has about the same capacity as double deck – but with a smaller proportion of seats, of course..

    Double deck cars are better for maximising peak period seating. Single deck cars are better at all other times – they are cheaper to buy and operate, and they offer a more pleasant travelling environment – higher ceilings, no stairs, bertter sight lines. Stairs discourage the elderly, who are an increasing proportion of the population. Good sight lines along the train are important for people’s sense of security at low use times.

    Which you prefer depends on how you value the costs and benefits of each. I think that frequent, comfortable offpeak service should be valued highly. Accordingly, I think that Sydney going double deck was a mistake.

    If Sydney was single deck, existing peak direction sitting capacity could mostly be maintained by workarounds such as running a few more peak period outer suburban trains to Sydney terminal.

  10. This article starts with a great map of Sydney showing employment distribution. Is there one of these knocking around for Auckland? would be fascinating to see, as I’ve always understood Auckland’s employment to be very dispersed (outside of the CBD that is), with very few concentrations around our region.
    Anybody got one, or can do one?

    1. There is this from the economic development plan but the groupings are quite large. It isn’t too hard to get the Stats NZ data for census area units and put together something.

Leave a Reply