As I wrote a few weeks ago, one of the biggest things our next council should do is to focus on how we can turn PT up a notch by improving frequencies. In particular, we should look to bring forward Auckland Transport’s aspiration, as expressed in their Regional Public Transport Plan (RPTP), to improve frequent services from being every 15 minutes to every 10 minutes.
As I wrote at the time from my own personal experience.
The difference between a service every 15 minutes and one every 10 might not seem like much but in my experience makes a significant difference to how people perceive the usefulness of PT. I still recall the impact it had on my commute when train frequencies were improved from 15 to 10 minutes at peak times. When the prospect of a just missed train meant a 15 minute wait I would go out of my way to make sure I made it, precisely timing when I left work or as often happened, having to run to make it. That all changed when services were improved to every 10 minutes. Suddenly the wait didn’t seem anywhere near as bad and it means now I often don’t even bother checking the time, only seeing when the next train is when I turn up at the station. I expect improving frequent buses to be every being 10 minutes to have a similar impact for many others across the region and is even easier to market to potential users.
We can’t afford to wait till 2028 to maybe get these improvements, we need them as soon as possible if we want to make PT much more attractive and play a bigger role in Auckland. The biggest challenge to doing this is of course money. So in this post I’ve set out to try and estimate what it may cost to improve frequencies.
How many extra services?
For this estimation I’ve built a model to compare the current timetable with a hypothetical one where frequent services run every 10 minutes. I had no desire to trawl through timetables so using ATs GTFS data, the data that is used by third parties like Google to show PT timetables, I worked out how many buses for each route (and route variation) depart the first stop for each hour of the day. This has been done for both direction on weekdays and weekend days.
While there could be some issues with timing, in general we can assume that four buses an hour means a bus every 15 minutes and six buses an hour is a bus every 10 minutes. For every route variation and branch I’ve also used the data to calculate the distance travelled. As an example of all of this, an NX1 bus from Albany to Britomart is 17.7km while an NX1 from Hibiscus Coast is 30.8km. You can see a sample of this in the table below for weekdays but the data extends for all services and all hours (so is quite large).
The eagle eyed amongst you may also notice that this shows at peak times there are 71 buses an hour, streaming down the Northern Busway to the city, most of which are double deckers. There are other routes that use it too.
There are 28 current frequent routes and I’ve put them into the categories in the graph below. The five that already meet the aspirational target are: NX1, City Link, Inner Link, Dominion Rd (the 25 routes), Botany (the 70 route). A number of the second category are quite close to this too, for example the NX2 and Mt Eden Rd buses largely only needs better frequencies on the weekends for it to qualify while the Outer Link needs one an hour.
That leads us to the second part of the process. I’ve replicated the table above and where things aren’t already, I’ve increased them to six per hour for between 7am and 7pm. In cases where frequent routes branch, such as into two half hourly routes, I’ve increased each branch by enough so that the core frequent route is 6 buses an hour). There are many changes that are tempting to make but the only other one I did was have more NX1 services from Hibiscus Coast and fewer from Albany – currently there are only two buses an hour to/from the Hibiscus Coast Station off-peak and on weekends.
By multiplying the number of services by the distance we can get the total distance travelled on both the current network and the proposed network and compare the two. This showed we would need an extra:
- 14,900km service km’s on weekdays
- 17,600km service km’s on weekend days.
We have about 250 working days a year and 115 weekend or public holiday days. Combing these it adds up to about an extra 5,750,000km per year. That sounds like a lot but to put that in perspective, AT say that following the roll-out of the new network, the total number of service-kms now travelled was about 59.1 million annually. So, getting current frequent services to be every 10 minutes is less than a 10% increase.
What could it cost?
To get an idea of what it would cost to add those extra services I’ve combined some of the PT data the NZTA publish on public transport such as fares, farebox recovery and the total in-service km’s travelled by buses. This works out at about $5.50 per km, which for 5.75 million km would mean a total annual cost of about $32 million.
However, there’s good reason to believe this cost could be a lot lower. For example:
- Much of this will be able to be achieved without needing to buy new buses as just makes better use of existing ones so the marginal cost is much lower.
- I’ve heard in the past that the figure AT use internally to estimate the cost of new/extended services is a lot lower than this.
Taking this into consideration, a lower end of the scale could be as little as $14 million annually.
Obviously these aren’t small figures but they are smaller than I had expected and doesn’t seem like an unreasonable sum to pay, especially for the government who have made a lot of noise about improving PT, including in the Government Policy Statement (GPS).
These figures were just to upgrade the existing ‘frequent’ buses to have even better frequency but that isn’t the only improvement it would be good to see. There are a number of routes it would be good to see turned into frequents, such as the 650 along Balmoral Rd, or services along SH16. The RPTP has a few maps shows all the routes they want to eventually see become frequents. I haven’t worked it out fully but as a guide, using the figures above, to upgrade a 15km route from every half hour to every 10 minutes would cost about $2-4 million annually.
If all of those 2028 routes, with services running at least every 10 minutes, could be delivered in the next few years we’d have a pretty good network.