The graph below is a hypothesis rather than anything that's been measured.
It seeks to show the relationship between service levels and complaints received regarding frequency at off-peak times on the Melbourne metropolitan rail network.
X axis is service frequency, Y axis is volume of complaints related to frequency.
Click graph above to enlarge
The red line is the cost of running the service by frequency. This is assumed to be fairly proportional to the service level added until you hit constraints such as rolling stock, signalling and line capacity in which case the marginal cost of adding extra frequency can become extremely expensive and not quick to do. So expensive that you instead think about alternatives like adding carriages, price signals like incentives to travel off-peak, diverting passengers onto parallel quieter lines etc.
Conversely if service is already frequent except for some narrow time bands (eg Sunday mornings or evenings) then the marginal cost of adding frequency in those time bands is small. When expressed in terms of maximum waits the cost of halving maximum waits on some lines can be tiny as only a few percent more trains need adding to the timetable.
The blue line is more speculative. For a suburban rail system where trip lengths of 30 to 60 minutes are common, it makes the assumption that basically no one would complain about a 10 minute frequency (unless it's peak times and even this is insufficient to avoid crowding).
Whereas 30, 40 and 60 minute frequencies are universally despised. Especially in cases where you don't have full control over your arrival time due to having just come off another train, tram or bus or having an activity such as work that finishes at a particular time.
Note that these curves would be system and trip specific. Frequency is less important on regional train networks with high average speeds and long trip distances. On the other hand for metro systems it is more important with even 10 minute headways contributing to excessive variability for large volumes of short trips. What you see above is a 'best guess' average for the Melbourne suburban system. Much of it is based on social media and other posts about train frequencies. That is you rarely hear objections to high frequencies but you often do for low frequencies such as the 30 or 40 minute gaps that exist on most of the suburban network at certain times.
If this holds true then some great opportunities present themselves. Going from every 40 to every 20 minutes is adding 1.5 trains per hour each way. Going from every 30 to every 20 or every 20 to every 15 minutes is adding 1 train per hour each way. In all cases maximum waits (and thus travel time variability) is greatly reduced, as are likely complaints.
Note that these curves would be system and trip specific. Frequency is less important on regional train networks with high average speeds and long trip distances. On the other hand for metro systems it is more important with even 10 minute headways contributing to excessive variability for large volumes of short trips. What you see above is a 'best guess' average for the Melbourne suburban system. Much of it is based on social media and other posts about train frequencies. That is you rarely hear objections to high frequencies but you often do for low frequencies such as the 30 or 40 minute gaps that exist on most of the suburban network at certain times.
If this holds true then some great opportunities present themselves. Going from every 40 to every 20 minutes is adding 1.5 trains per hour each way. Going from every 30 to every 20 or every 20 to every 15 minutes is adding 1 train per hour each way. In all cases maximum waits (and thus travel time variability) is greatly reduced, as are likely complaints.
In contrast, going from (say) 9 to 12 trains per hour is more expensive (3 trains per hour more) and is unlikely to make as big an inroads into complaints (unless the higher frequency is required to relieve crowding).
The moral is that governments wishing to generate goodwill in an election year would do well to start by cutting the maximum waits across the whole network first. After they are low then you can consider other frequency upgrades. This has lessons for the current government that has tended to avoid even small frequency boosts with the new Metro Tunnel timetable getting a less warm reception than it might have due to previous over-selling coupled with under-delivery of service on less favoured lines where 30 and 40 minute gaps remain common.
The recommendation going forward is to get as many train lines as possible down to a 20 minute maximum wait - as opposed to the widespread 30 or 40 minutes. This was done for Werribee, Williamstown and Frankston in 2021. There wasn't anything more on Metro lines for 5 years but the 2025 state budget funded Craigieburn and Upfield with implementation later this year.
This then leaves Mernda, Hurstbridge and Sandringham (maximum 40 minute gaps) and Belgrave, Lilydale, Glen Waverley, Alamein (maximum 30 minute gaps) to resolve. For context Sydney has a maximum 15 minute wait at most stations while Perth's maximum is 15 minutes at every station during the day all week, dropping to 30 min after about 8 or 9pm.
The recommendation going forward is to get as many train lines as possible down to a 20 minute maximum wait - as opposed to the widespread 30 or 40 minutes. This was done for Werribee, Williamstown and Frankston in 2021. There wasn't anything more on Metro lines for 5 years but the 2025 state budget funded Craigieburn and Upfield with implementation later this year.
This then leaves Mernda, Hurstbridge and Sandringham (maximum 40 minute gaps) and Belgrave, Lilydale, Glen Waverley, Alamein (maximum 30 minute gaps) to resolve. For context Sydney has a maximum 15 minute wait at most stations while Perth's maximum is 15 minutes at every station during the day all week, dropping to 30 min after about 8 or 9pm.
1 comment:
Another thought on the same tangent is what kind of targeted service increase would get noticed and hopefully praise by passengers and voters.
For example, families and workers heading home at 10:30-11pm after evening sporting events or concerts would soon notice if they never had to wait more than 20 mins (with average wait times dropping from 15 mins to 10 mins), regardless of publicity.
Whereas arriving the station after the soccer at 8:15pm on Australia Day to find you have a 24 min wait for a train to Glen Iris, Oakleigh or Preston leads to complaints and ultimate people either choosing to drive or even watch events on TV.
Especially cases where you have to squeeze in the train when it finally arrives and then stand for 30 mins to the middle suburbs as the train gradually clears out by say Moorabbin or Mitcham, or you go on a endurance race on evening bus replacement with kids either falling asleep or becoming cranky.
Similar sentiments would apply for Sunday morning trains and evening trams.
In contrast to the above, increasing daytime trams from 12 mins to 10 mins would be a blip on the radar for many passengers (especially on combined corridors), yet also involves one additional service per hour and likely require dozens more drivers.
Frequency increases can also reduce the impact of cancellations, as the next service is never too far away. While on a 20 min frequency, it can spoil your whole morning (see Upfield Line and Altona Loop complaints).
Overcrowding on suburban buses around school times ans complaints by the community is felt most where the next bus is 30, 40 mins or even 60 mins later (eg Wellington Secondary College). Whereas if kids have several buses to pick from, they can simply wait another 10 or 15 mins and it becomes a minor inconvenience.
And issues with poor online information resulting in complaints and ultimately a general distrust in public transport (or at least buses) can also be avoided with higher frequency.
As reported by Bus & Coach Society of Victoria yesterday, website timetables for buses were incorrectly showing Sunday timetables & frequency (GTFS data in 3rd party apps and the PTV journey planner was displayed correctly).
Had a similar data problem occurred for trams, many wouldn't have even noticed because trams largely act as turn up & go (Sunday mornings and most evenings as an exception).
If you want people to choose public transport, it has to be at a standard they never feel a need to complain - it's also the complaints politicians and operators don't get but the hundreds of cars/Ubers clogging Punt Rd after the footy or dozens of parents in cars waiting at school pick-up than kids getting the bus home that show transport policy is often not meeting community expectations.
Post a Comment