What’s faster, a less frequent direct service or a more frequent service that requires a transfer?
In my post last week about needing a metro timetable to match our upgraded rail network a few things came up that caught my attention
- That testing didn’t go well
- There were a number of suggestions about west-Newmarket services – with some similar comments I’ve seen from time to time in other places.
So let’s look at these
First up, AT have admitted that testing didn’t go well, hopefully this means they do make changes to remove some of the complexity and give us a more metro-style timetable.
“The testing was a valuable learning curve”, Ms van der Putten says.
“It highlighted several challenges, particularly around network congestion, which our teams are now working through. We’ll tweak a few things before we run the simulation again in the April school holidays, to help us finalise a robust timetable and provide reliable services from day one.”
The main focus of this post though is about west-Newmarket services. Talk about retaining some services is understandable given it’s been an important destination for Western Line rail users up until now. It’s also personal for me, not only do I live out west near the rail line, so have an interest in the connection but my wife works in Newmarket so any change will impact her.
However, nostalgia and individual benefit alone isn’t a good enough reason to keep direct services to Newmarket in the face of such a transformative change as the City Rail Link is.
As a reminder, the CRL does three fundamental things for the rail network.
- It adds two new stations in the heart of the city and at K Rd, significantly widening direct access to so many more destinations from its current sole downtown edge point, making it easier to get directly to tens of thousands of jobs and to the universities and other education facilities in the city.
- It increases the capacity of the rail network – currently 18 trains an hour arrive at Waitemata but with the CRL, on day one, up to 32 trains an hour (16 per direction) will be access the city. Eventually, with additional investment around the network, this could go as high as 48 trains an hour. More trains means more capacity but often overlooked is that they also mean faster journeys as how long you need wait for a train at a station is reduced. Even just moving from 6-trains per hour to 8 is like being a station closer to your destination.
- It significantly shortens travel times from the Western Line to the city. A trip from Maungawhau to Waitemata should take around 9 minutes, about half what it does now via Newmarket. For trips to destinations near the new stations the total travel time saved is significantly higher.
When it comes to the Western line, there are three general options for how services could be run:
- All services to the city and then out to either the Southern or Eastern lines. Passengers needing to get to Grafton or Newmarket would need to transfer.
- A loop around around the city and Newmarket and then back out west again – this could be either all in the same direction or some having alternating directions e.g. half go via the city first and the other half via Newmarket first.
- Have some services go to the city and some run a west-south pattern.
To help explain why I think Option 1 is the best, let’s dive into some numbers.
Limited train slots
In a perfect world we’d have a far more developed rail network, with additional tracks on our lines to enable more services to run, no level crossings flat junctions to create constraints. We don’t live in that world and mostly we’re limited to one track per direction. The Western Line has the additional disbenefit of having a lot of level crossings.
AT are concerned that more trains on the network means barriers will be down for longer and that will result in more people taking risks around crossings. There is currently no funding to replace those level crossings and current projections are it will take decades to remove them all.
Currently at peak there are six trains per-hour in each direction, or 12 in total. Once the CRL opens, AT have said that due to the safety concerns they can’t increase the total number of trains crossing those level crossings so have opted to redistribute how they’re used, with eight trains per hour in the peak-direction and four counter-peak direction. So eight trains per hour in the peak direction is our limit, the question then is, where do we send those eight trains.
As an aside, I think there’s more they could do with things like enforcement to reduce the safety risk at crossings and if the CRL is as successful as expected I won’t be surprised if they resort to other measures in order to enable more services.
Changing Passenger Demand
To help answer the question of where to send the trains, looking at where people currently catch them to is a useful starting point. The station boarding and alighting data I have is about six years old now but while the exact numbers might not be the same, the general trends for where people are travelling to/from is likely still relevant.
Looking at all people boarding at a Western Line station (west of Grafton) and heading in a city-bound direction we get the following breakdown of destinations.
The numbers might be slightly different at peak times but as you can see, about a third of all boardings are heading to Waitemata with over a quarter heading to another station along the Western Line. Trips to Grafton and Newmarket account for just over 20% of all boardings. Notably, that’s off-set by just over 8% of people who would directly benefit from the currently proposed service pattern.
Interestingly, Grafton and Waitemata also have the largest (and near identical) discrepancies in the number of people boarding and alighting due to something we call ‘downhilling’ – where people (most likely students) are getting off at Grafton and catching a bus down to Uni, then at the end of the day continuing downhill to catch a train home from Waitemata. I suspect there are a lot of trips currently using Grafton that will likely be replaced by travelling to Karanga-a-Hape and Te Wai Horotui and a short walk or sometimes a bus.
With the changes noted earlier in the post, the CRL will significantly change Western Line demand. For example currently public transport use from the west to around Midtown and the Karangahape Rd area is lower than it is around Waitemata. The CRL will change that.
It would not be unreasonable to expect that over the short to medium term that overall usage would double. I’m not sure what the modelling is predicting but let’s assume that within that doubling of usage, those non-city trips increase by about a third – a not insignificant number. The percentage of trips to Grafton and Newmarket would drop and even if you included the inner-Southern line stations like Ellerslie, overall that’s likely less than 20% all trips. That’s because it would mean trips to the city would increase from about a third now to well over half of all trips. Notably, currently about 54% of boardings on the Southern and Eastern lines are destined for Waitemata – though you’d expect their share of trips to the city to also increase.
The point of all of this is that while it may seem like a lot of people are going to Grafton and Newmarket today, once the CRL opens a lot more will be wanting to travel to the city and the question becomes, is it worth diverting some trains for what might be 10-20% of trips?
At 10-20% of trips that’s maybe two trains per hour.
Would it actually be faster?
A key feature of Auckland’s public transport network development over the last decade or so has been the focus on making much of our PT network frequent enough that it becomes turn-up-and-go, so there is no need for a timetable. That’s important for making it easier to use.
In terms of time spent on the move, a direct service from Western Line stations to Grafton and Newmarket will always be faster than one that requires a transfer at Karanga-a-Hape.
Based on existing timetables and planned frequencies, the alternative – a train from Maungawhau to Karanga-a-Hape, then transferring to another train to Grafton – would take between 3 and 10 minutes longer than a one-seat ride to Grafton. On average this is likely to be about an extra seven minutes of travel time compared to a direct service. But with perhaps just two trains an hour – one every 30 minutes – what option is faster for overall journey time.
One of the cross passages between platforms at Karanga-a-Hape and access to the escalators up to Beresford Square
To answer that question I built a little model to compare the two options: direct journey vs transfer at Karanga-a-Hape. The model assumes someone just turns up at a station hoping to go to Grafton or Newmarket, which option gets them there faster?
Even in the worst case scenario – you catch a train to into Karanga-a-Hape to transfer, and discover you’ve just missed a train heading to Grafton, leaving you a 7 minute wait for the next one – 60% of the time it will still be a faster option than the less frequent direct service.
If, on the other hand, you’re lucky with timing – you step out of the train, stroll across the platform, and your next train turns up within a minute – then the transfer journey will have been the faster option 83% of the time.
Remember too, Karanga-a-Hape offers a short, all-weather, level, cross platform transfer.
Focusing on travel times like this is also why options like the loop pattern won’t work in this situation like ours – the loop is big so people travelling to the far-side of the loop have to go the long way around, which again would be slower than just transferring. And attempts to answer that by alternating loop direction to give every destination a one-seat, not only makes the network much more confusing to operate and to use, but halves frequency for everyone.
Frequency is freedom, provide enough of it, on a regular legible pattern and people will transfer, because overall their journeys will be quicker, easier, and more predictable.
No one likes a longer journey but that seems like a decent trade-off given the wider benefits of CRL. Furthermore, as frequencies go up, that transfer option becomes even more likely to be the faster option
Greater Auckland’s work is made possible by generous donations from our readers and fans. We’re now a registered charity, so your donations are tax-deductible. If you’d like to support our work you can join our circle of supporters here.
Processing...
Yep, until the level crossing situation is resolved, Option 1 appears to be the best interim solution. Not only does it allow a higher frequency but, I’d suggest it potentially offers more reliability.
Didn’t the new network do away with all these weird low frequency overlapping services already? Like the eastern line used to alternate half the trains to manukau and half to papakura, and buses used to do all sorts of crazy stuff.
Why are they going back to the old inferior model that was so successfully done away with?
Like I can’t for the life of me understand why, when they’ve just spent five billion dollars building a tunnel to access the middle CBD and stop the trains needing to go around via Newmarket… the first thing they do is send trains away from the CBD via Newmarket?!
Another consideration – regardless of which is actually quicker we need a system that is easily understood – only a minority are likely to be timetable nerds.
As to level crossing removals on the Western line I am assured that public consultation about options for each of the existing crossings is coming though nothing definite just now. Some of the business case documentation is becoming available though with numerous redactions.
My concern is the way they’re approaching the level crossing removal is it will be too broad – do you want a lump of coal or a pony type stuff – without proper costs or trade-offs. Everyone will want the crossing they used separated which is simply unaffordable, especially in any reasonable timeframe.
AT’s narrative on reducing the frequency of trains to reduce vehicle risks at level crossings isn’t at all transparent. It is not, and never ever was, about “safety”. It has only ever been about limiting disruption to cars. “Safety” is always a an easy way of justifying measures that could genuinely benefit PT users but instead used to satisfy drivers.
“Even just moving from 6-trains per hour to 8 is like being a station closer to your destination.”
Matt this is a great way to communicate the power of increased frequency in PT.
The proposed post CRL network design appears to me to have wandered away from the target of higher frequency, on a regular pattern, clearly communicated.
Instead they seem to trying to run an over-complicated, harder to use, harder to operate fiddly pattern to attempt to satisfy lots of infrequent one-seat rides.
Probably the model is in charge.
Modelling and technical work should always be checked back to the higher order strategy, so what is that, or what should it be? Most simply put: Metroisation.
Running our little urban railway as much like a Metro as possible. So on a regular, un-complicated, high frequency pattern, clearly and simply communicated.
Remember everyone agrees the CRL is “transformational”. Yes that is an overused term, but one that i believe is appropriate in this case, as you’d hope for the size of the investment!
Let’s be clear what transformational properly means – it means this changes everything. Modelling is heavily based on extrapolation, on the continuation of what there is now.
Even when new assumptions are added, increased population etc, these almost always remain bounded by what there is now. Modelling is very poor at picking discontinuities, by definition.
We don’t know how people will respond to the changed opportunities offered by the CRL. A key strategy in this situation is to remain flexible, to not try to second guess and predict use.
Of course railways are pretty inflexible things, especially heavily interlined mixed railways with a lack of track, like ours. All more reason to start with the simplest most regular, and highest frequency pattern possible, with a plan to review after a decent period of operation.
This is not a call to react immediately to early ridership patterns- it will take a year or more for people to get into new rhythms, but it is a call to keep it super simple and clear and direct. Easy to communicate to new users, this is a once in a lifetime opportunity to gain new riders.
Or to phrase it in governance language: what is the bigger risk: some current users will lose their direct ride (that is certain from the west) be grumpy, and get amplified by media searching for clickbait, or that through complexity and/or outages appeal to new riders falls flat?
Given that the number of non-users of the Auckland train system is a way bigger than the number of current users heading to Newmarket/Grafton, and that THE WHOLE PURPOSE of this massive investment is to significantly grow ridership, clearly the strategic intent of the opening operating pattern should be simplicity, frequency, easy of use. Make it simple, repetitive and as reliable as possible.
Frequency gives shorter journey times, that will get new users coming back, so that is the killer app here.
Of course no operator wants to disappoint current users, but there are trade-offs and
1) a frequent transfer is not a terrible alternative (likely to be better, as Matt shows).
2) eye on the prize! New users, in volume, spreading praise by word of mouth…
It is a shame though, the most expensive rail system on a $ per km basis in the world has not dealt with the level crossings on the western line. They have been a known issue with multiple studies done over the last 25 years going all the way back to ARTA. The fact is fixing them is never going to be any cheaper than now, when the CRL works were being completed. The idea of building network reliability and demand then shutting down again to fix a known issue makes no sense. The fact the majority of Aucklands intensification has happened and is happening on the Western line further raises questions about lack of investment in level crossing removal and limiting train frequency as a result. Putting it in the too hard basket is shifting the costs to another generation and is mostly a cowardly approach. The level crossings on the Western line need to be sorted. Until then shift intensification somewhere else.
Swanson to Manukau via Newmarket? And run the S-C loop to Ōtāhuhu via Ōrākei instead. The O-W would still be the direct but less frequent connection, and still maintain the ability to terminate peak services from Manukau at Newmarket. Westies could also transfer to another E-W train to Newmarket like option 1 as well (during peak)
Your piece reminds me of an observation from a Hong Kong-based transport enthusiast Jacky Lim (I don’t like his Hong Kong political stance after 2022, but let’s put it aside and that issue doesn’t concern non-Hong Kongers anyway). He made a remark on social media a few years ago that in his opinion, Singapore’s bus (and public transport) service is objectively worse than that of Hong Kong, because there are fewer point to point services, you have to travel to nodes then transfer to the final destination. In Lim’s view this is less than optimal from end user experience.
I think you all had written an article many years ago that the best practice for a sustainable public transport network involves users/passengers often needing to transfer between services just to reach the final destination. This improves efficiency. Your article here seems to be running on the same theme. But to someone like Jacky Lim he thinks the end users of public transport dislike such best practice design around transfers between modes and/or services.
How would you respond to Lim’s misgiving about the whole concept of having to transfer just to get to one point to another? Surely we can strike a balance in between?
Thanks.
(PS: Jacky Lim does YouTube programmes and also writes on Facebook. I don’t think his podcasts or social media contents are in English) https://www.youtube.com/@JackyLimChannel
Frequency is key. For me, I like to feel moving. So if I get off one train K-Road, have a quick walk inside a nice station to the next platform, and then catch the next train with barely any waiting time, that is great. Ideally, the frequency of both trains is so high that I can catch it immediately and a slower person (handling a stroller or walking aid) would be able to catch the next one with limited waiting time.
If there is only one bus/train every half hour from the station where I need to switch, I have very long wait times, and even nice stations will get boring quickly. In those situations, I would prefer a one-seat ride (even if it is just every 30 or 45 minutes), so I can do the waiting at the beginning/end.
That probably explains some differences: given Hong Kong’s population density, even least frequent public transport runs every 20 minutes at the minimum, so waiting between services is not an issue in Hong Kong’s contexts. But it doesn’t work in most other countries and certainly not New Zealand.
The majority of users post-CRL won’t be transferring. This discussion is about a relatively small portion of of users that will lose a direct frequent service as a result of an overall network improvement.
One assumption in the article is a bit questionable. It’s not such a short walk from the Karanga-a-hape station to the university. There’s no direct bus. You’d have to change at the Karangahape-Symonds corner or hike from Albert St up through Albert Park.
Bereseford Square -> Alfred Street is 8-9 minutes. A direct bus at peak time would take longer.
Why would you use Karanga-a-Hape to get to Uni and not Te Waihirotiu with a short walk along Victoria or Wellesley Streets
It is a short walk to the unis from Te Waihortiu, and there are a gazillion buses running up and down Wellesley st if that is too much.
K Rd is a nice level walk or a short frequent bus ride to the hospital and Medical School.
Surely you would get off at Te Wai Horotiu to go to the Universities.
K Rd is flat/downhill.
What happened to the simple clock-face timetables and predictable patterns? AT seems to be able to do it for buses very well so the principle is clearly well understood. Yet for the trains we’re going to end up with weirdly spaced timetables (judging by the number of photos of the departure boards at the stations taken during the trial and posted on social media) and weird running patterns. Is there something else at play here? Like KiwiRail forcing some freight train slots on AT (particularly for the western line?)
Within AT the rail team is a bit of a silo so tend to ignore what happens in other parts of the org (or PT team).
We’ve raised issues like these directly with the team but been told they know best.
Great summary, I have been trying to get my head around all this! One question: According to KT, the issue with changing trains at K Rd to go to Newmarket is that there will be no train going in that direction at peak times – is that not right?
Yes there are trains going from K Rd to Newmarket
He is absolutely wrong, there will definitely be trains going from K Road to Newmarket at rush hour.
This is partly AT’s fault for creating a service pattern so bloody confusing.
Surely AT could do the timetabling so Western Line trains arrive at K’Rd 2 or 3 minutes before the Southern line trains arrive to make for an easy transfer
Ideally yes but the challenge is in Auckland we already have quite a few locations where we need to balance timetables for, notably all of the flat junctions, like Quay Park, Newmarket, Penrose, Westfield, Wiri etc. timetabling to hit K Rd for quicker transfers might cause other issues around the network.
So grade separating the Westfield and Wiri junctions is something that would help improve timetabling and reliability
Wait, so Western line in leak is 8 one way and 4 the other….
Does that mean 4 are empty running back or are they deliberately not being run and used somewhere else?
That’s mental
This is why at peak times there’s an eastern line service that runs though the CRL to Newmarket, if the level crossings weren’t there they’d be being sent out west
If you run single unit trains then they could be doubled up at Newmarket for the return less frequent journey.
I seem to remember there was triple tracks at Avondale back in the day when it was single track I am just wondering if crossing loop or a third line could be used to increase frequency to allow for more west south trains to run without interfering with trains travelling through the loop