Tuesday, September 02, 2025

TT 211: Trains and People - Who gets the most service in the suburbs?


ONE MINUTE EXECUTIVE TAKEAWAY Last year I established that there is little relationship between a line's patronage and whether it was likely to have frequent all day service. Lines like Craigieburn had double the waits of the Frankston line yet carried similar numbers. The difference in all week frequent service was even starker with Frankston getting 90 hours per week versus 20 to 30 hours for Werribee and Craigieburn. Both the latter will get added trips when the Metro Tunnel opens but this won't materially change the all week ratio.
Today I check catchment population versus off-peak service levels for middle and outer suburban stations across Melbourne. Again there are big disparities. Popular stations on less served lines get just one-ninth the service relative to population versus stations on the best served line.

Both exercises support the view that Melbourne's Metro network lacks funded and implemented master timetable planning that aligns all day rail frequency with objective criteria such as usage or catchment population. Instead historical inertia and political factors have been most powerful in determining the service that suburban stations on electrified lines get.
If you don't want to read tedious station by station comparisons just scroll to near the end for the pre and post Metro Tunnel animated frequency map. There's an action plan to fix this right at the end.  

  

 Approaches to planning suburban rail frequencies


There's at least three approaches to deciding how frequently suburban trains should run. The transport authority (which in our context includes the state government and DTP) could: 

a. Operate a decent minimum service level (say every 15 min) on all lines across the week. That has the virtue of simplicity - rock up at any station at any time and you'll have a train soon. The idea that supply (in this case frequent service) is a major driver of public transport demand so it should be high, especially given that most railway costs are fixed and the marginal cost of off-peak frequency is low. This is basically what Sydney and Perth do with minor imperfections (Sydney not quite all stations, Perth not after 9pm). Both networks are good patronage performers relative to their size and historically. 

b. Operate a low network-wide minimum service standard (eg every 30 min) and run only the densely populated section of the network network frequently all day with most of the rest frequent in peak times only. This is based on the idea that public transport usage is demand-led so it is futile to give everyone a good service all day. Brisbane does this. It's a failure with low patronage relative to both network size and (even more so) rail employee numbers. 

c. Similar to b. above but have an even lower minimum service standard (eg every 40-60 min) with history or electoral politics determining which stations get all week frequent serviceThis is what Melbourne has allowed to happen. That wording is deliberate because my hypothesis is that metropolitan electrified rail service planning is a spasmodic set-and-forget affair that lacks an annual funded continuous improvement program that targets the highest needs first. 

Weak patronage rebound for Metro network

While usage of Melbourne's large rail network remains substantial, BITRE reports that it has been the slowest to bounce back in patronage since the pandemic. Contributing factors likely include frequent construction shutdowns, our high working from home rate and long waits at the very times travel activity has grown fastest due to a lack of service reform. 

There has also been a structural shift in that we have effectively paused rail electrification to growth suburbs with V/Line diesel trains increasingly taking on the growth area suburban transport task in the west and north. Unlike Metro, V/Line does have a strong and continuous program of proportionately large frequency improvements with these often being (honoured) election promises. The same BITRE  report says that V/Line has the fastest growing patronage of Australia's regional rail operators. 

Subsequent comments here apply most to metropolitan electrified services. 

Method 

Today I'll look at the 5km population catchment of metropolitan stations relative to the service it gets.

I used the Population Around a Point website by Tom Forth to calculate station population catchments. The service is midday weekday trains per hour. That off-peak measure is good for many reasons including off-peak frequency being a better proxy of a line's usefulness for diverse trips all day, the growing relative importance of off-peak versus peak travel and off-peak's higher patronage/service elasticity. 

So that these circles (mostly) capture only a single train line and do not have a tram option, I picked a ring of suburban stations about 20 to 40km out from the city. Most circles contain several stations but these are on the same line so typically get the same waits between trains.  

Below are the two most contrasting examples. Williams Landing on the Werribee line has a high catchment population but gets few trains while Carrum on the Frankston line has one-fifth the population but has twice the train frequency. Together that makes over a 9:1 disparity in service per person in Carrum's favour.    


Williams Landing is a comparatively new station, opening in 2013. The Werribee line on which it is situated has generally had a fixed 20 minute midday service for much longer. This compares to the Frankston line which also ran every 20 minutes until it was upgraded to every 15 minutes in the 1990s and every 10 minutes in the 2010s. On the population side, Williams Landing's catchment has gone from mostly open space to heavily populated in the last thirty years.

Today's 9:1 disparity is directly due to the less populated line getting lots more midday service while the more populated line stayed stagnant regarding midday service.  

Least to most served stations by service / population

I've already discussed the extremes of Williams Landing (lowest service per head) and Carrum (highest service per head). This is my list with twelve more. Each person represents 20 000 people within a 5km radius of the state station. Below is a list of all fourteen stations I reviewed with comments below each one. 


Werribee line. Discussed above. Top ranked due to its high population density with only a basic 20 minute midday frequency on the Werribee line. Laverton to Werribee portion serves a fast growing catchment.


Sunbury line. Similar story to Williams Landing with high population but only basic midday service. Middle Footscray to St Albans section densifying, Watergardens is the nearest station to some growth areas. Metro Tunnel timetable will soon double its midday frequency to 10 minutes to provide a service that better reflects usage.


Pakenham line. Like stations in the west and north this outer south-east station combines a high population catchment with a 20 minute service midday service. Station is beyond the split at Dandenong which enjoys a 10 minute midday service. This service commenced in July 2014 with midday trains previously only every 15 minutes to Dandenong and every 30 minutes on the branches. Should gain from Metro Tunnel with frequency improvements most likely evenings and Sunday mornings (not midday).
 


Geelong line. Busiest V/Line station apart from Southern Cross. Growth area catchment. Midday trains every 20 minutes which is sometimes better than some peak period gaps. This is a new station on a new section of line that did not exist before 2015. Timetable has larger evening gaps than Werribee line and late morning starts on weekends. 


Lilydale line. Not particularly densely populated but this and the Belgrave line has some of Melbourne's longest gaps between midday trains with Adelaide or Brisbane style 30 minute headways. Waits are already longer than at more distant stations such as Melton and Geelong with Sunbury to surpass Croydon's service levels when the Metro Tunnel opens. Midday weekday service levels have been unchanged for many decades though weekend trains got improved to every 20 minutes in 2012. This make the Lilydale line (along with Belgrave) alone in having a better midday weekend service than weekday service for no rational service planning reason except inertia. More details on the More Trains Melbourne's East Facebook page

Belgrave line. Everything said above for Croydon station applies to Boronia on the Belgrave line. Both lines lag the similarly configured Pakenham and Cranbourne branches which enjoy a higher (20 min) interpeak weekday service.  

Mernda line. Has the typical 20 minute frequency common on northern and western suburban lines as has remained for decades. Line is continuously populated in Melbourne with stations to the south having densifying catchments and stations from Epping to Mernda with growth area catchments including Wollert (which will soon be getting upgraded feeder buses). Line has no infrastructure limitations preventing it going to every 10 minutes. Liberals promised to introduce a 10 minute midday service on the (then) South Morang line in 2014 but lost government. There have been negligible timetable upgrades in the decade since with evenings and Sunday mornings remaining at every 30 and 40 minutes respectively.  


Werribee line. Major centre with growth area catchment. See Williams Landing above. 


Craigieburn line. Another continuously populated line this is another northern suburb line that has lacked significant service uplifts for many decades with its 20 minute midday headway seemingly stuck in stone. Possibly now Melbourne's most crowded, the line serves growth areas north of Craigieburn. Metro Tunnel Business Case envisaged midday service would improve to every 10 minutes but government appears to have walked back on this, with 2025 budget funding aiming only to get maximum waits down to 20 minutes from their current evening and Sunday morning 30 to 40 minutes. 


Cranbourne line. Large growth area catchment with many residents hoping for an extension to Clyde. Beyond the branch at Dandenong. Midday 20 minute service commenced in 2014 with this being an upgrade on the previously 30 minute intervals. Everything said for Narre Warren also applies here. 

Hurstbridge line. Inner portion of line to Eltham has a similar basic service level as Mernda, including 20 minute midday headways. Similarly its timetable has been largely stagnant over many years. 


Ballarat line. As part of the V/Line network, Melton has enjoyed a stronger trajectory of service improvements than most Metro lines. That includes midday weekday services improving from approximately one to three trains per hour. Weekend services have also improved but not by as much with the most recent improvement from every 60 to every 40 minutes. The line serves one of Melbourne's fastest growing corridors and significant further investment in improvements is envisaged. 



Pakenham/Cranbourne line. One of a string of very busy station between Caulfield and Dandenong. Currently enjoys a 10 min midday service with likely evening and Sunday morning gains after the Metro Tunnel opens. Had a 15 minute midday service prior to the 2014 upgrade. Arguably now justifies a 5 min midday service which would be assisted by its (now) lack of level crossings. 

Frankston line. Discussed extensively before, Carrum enjoys suburban Melbourne's highest train service level with trains consistently every 10-20 minutes at all times but in early hours of Sunday morning. This was not always the case with a steady stream of upgrades basically doubling service since the 1990s, assisted by the line having several political swing seats. Bayside placement and inland Green Wedge contributes to catchment population of less than 60 000. But even doubled population would retain Carrum's spot as Melbourne's highest served station for its distance from the CBD. 

Discussion of results

There is no connection between a metropolitan train station's catchment population and how frequently its midday trains run. 

The most important factors appear to be (i) population distribution of 30 or 40 years ago and (ii) political priorities including the distribution of historically marginal and safe seats.

Inertia in metropolitan (though not regional) rail frequencies has been king for much of the last decade. The threshold to trigger a metropolitan service upgrade (even the cheapest types) is high. You basically need middle class commuters left behind on platforms to get a substantial change to train timetables, such as occurred about 10 to 15 years ago. 

The 2025 state budget may represent a turning point with exactly the required lower cost upgrades being funded along with Metro Tunnel operations. That will deliver a major all week frequency upgrade for Watergardens, the second ranked station here. And at least Williams Landing, Narre Warren, Werribee, Broadmeadows, Cranbourne and Springvale should get some gains at other times. However several more like it will be needed before we can say it represents a decisive change from the 'infrastructure over service' stance that dominated this government's first decade. 


The bigger picture

Here's a map showing the Metro network with the most and least amount of frequent service across the day. Black lines have no frequent service, brown lines are frequent in peaks only while green lines are frequent in the off-peaks too. It alternates between existing pre-Metro Tunnel and what I understand will be the service levels post-Metro. Click map for better view.  



Frequent is defined as every 15 minutes, a common threshold and base frequency for Australian transit systems. The thickest green lines are frequent seven days. Everything is daytime only as no single route or line in Melbourne has consistently frequent evening service (although some overlaps do).  

You can see massive service disparities across Melbourne. In a nutshell the west+north get a peak only frequent service while many lines in the east+south enjoy a frequent service all day on weekdays and sometimes also weekends (more here). It's a network of haves and have nots, as then Opposition transport spokesman David Hodgett said in 2016

The Metro Tunnel timetable will give the west its first ever individually frequent all week line (to Watergardens). Craigieburn and Upfield (who see their maximum waits reduced from 40 to 20 min) and Werribee (extra peak trips) also get handy gains. But the overall distribution of who gets all day frequent service remains very skewed. Want this fixed? Keep reading!   

Potential 'Big Service' Metro upgrades 

Existing high usage and high catchment populations are both good reasons to boost train frequencies at midday and other times. What are the best lines to start with? 

1. Belgrave/Lilydale lines - every 10 min to Ringwood, every 20 min Belgrave & Lilydale

On the basis of cutting the longest waits first and being the cheapest to do, a Belgrave/Lilydale timetable rethink could be a good starting point. Key priorities for a Burnley group greenfields might include 20 minute maximum waits for Belgrave and Lilydale, a 10 minute maximum wait at Ringwood and simplified peak timetables with fewer stopping patterns.

However this doesn't fix the east-west divide, which needs the next steps listed to address. 


2. Craigieburn line - every 10 min midday

Craigieburn line passengers pay the same fare as Frankston line passengers yet have double the waits, half the service and more crowded trains with higher unique section line ridership. The Metro Tunnel's business case from 2016 was based on the Craigieburn line operating every 10 minutes. This need remains today with its catchment population growing strongly, exacerbated by limited V/Line service at Donnybrook. Scheduling work may be simpler than a Burnley group greenfields timetable so there may be potential for this to be done first. 

3. Werribee line - every 10 minutes  midday

Though it has fewer stations, Werribee is another strong contender for better midday service. The line also serves a growth area. An upgrade would take some stress off the under pressure Geelong line at Wyndham Vale and Tarneit, especially if weekends used the express line like weekdays do. 

4. Mernda, inner Hurstbridge and Sandringham lines - maximum waits cut from 40 to 20 min

A package of low budget evening and Sunday morning upgrades that replicate what the 2025 state budget funded for Craigieburn and Upfield. Sandringham would be especially cheap since it requires a handful of trains added per week. Mernda and Hurstbridge (to Eltham) upgrades are best done simultaneously as part of a Clifton Hill group package, which if split could see Sunday mornings and evenings until at least 10 pm done as the first phase.

Due to relatively simple scheduling and the few extra trips needed the government may wish to make this first priority to provide early and widely spread benefits on lines long neglected. 

5. Mernda and inner Hurstbridge lines - every 10 minutes midday

Doubling of daytime service to reduce waits from 20 to 10 minutes. Serving a growth area the Mernda line is the most important but a decision may be taken to deliver similar to part of the Hurstbridge line as part of a Clifton Hill group upgrade. A 10 minute frequency would make train travel more popular in densifying inner areas where passengers must currently choose between slow frequent trams or faster but infrequent trains. It would also encourage bus reform and speed access to La Trobe University (via the existing 301 shuttle). 

Other off-peak frequency upgrades not discussed here but which could be desirable are those associated with the Upfield, Glen Waverley, Melton and Geelong lines. 


See other Timetable Tuesday items here


11 comments:

Jim Gleeson said...

Exc info & analysis!
Reg reader here.

Anonymous said...

Fascinating and disheartening stuff, Peter. What explains this pigheaded refusal to increase frequencies? Is it the spectre of ratcheting operational costs - ie adding a permanent burden to the budget? If so, how great a burden? What, in your view, would it take to run an essentially timetable-free network (ie 10-min frequencies everywhere on the electric network) between 6AM and 10PM?

Tommo said...

All valid comments on the lack of frequency. As a Mernda line user, the 20 min off peak waits have long been an annoyance and especially so knowing the Clifton Hill group can sustain higher frequencies. Writing to local MP and transport ministers gets the usual response on Infrastructure builds but nothing suggesting a service upgrade as in the Metro tunnel or PTV NDP (2012) suggestions. Perhaps that will be thrown out as an election promise come November 2026 as no funding was allocated in the last state budget for service uplifts other than Werribee, Sandringham. Perhaps the State thought they'd be further progressed on Infrastructure before the metro tunnel opens and not be in a poor financial health, delaying the ideas of service uplifts yet again across the network. For now, it seems to be chasing votes out west first and fixing decades of neglect again. Let us hope that once that is complete, more widespread reform takes place quickly by end of 2026.
If the tunnel opens as planned, I see no reason why the Clifton Hill group can't go to a better off peak frequency before that though even at 15 min although 10 min would be preferred.

Peter Parker said...

Anon - the Parliamentary Budget office came up with $200m when asked by The Greens about this. There may well be offsetting savings that might be able to make this cheaper. I just don't think the government's been put under enough pressure to identify them all, and they've been lucky to have a weak / disunited opposition.

Tommo said...

Peter is that $200m pa ?? That seems small fry even at double that as you'd expect more patronage too as PT becomes actually useful.

Heihachi_73 said...

Regarding more frequent trains on the Ringwood group: It has to be frequent *and* express, not stopping all stations all day like the weekend timetable. The trip is already too long and the line speed is too slow to be taken seriously as an alternative to the car. The amount of timetable padding between Box Hill and Chatham is ridiculous.

Belgrave and Lilydale trains need to run express west of Box Hill at all times, with shorter runs to Blackburn and Alamein stopping all stations full-time. The Box Hill-Camberwell-Glenferrie-Richmond express needs to be permanent from first to last train in order to actually compete with cars on the Eastern Freeway. If Union station needs to be a permanent stop for expresses, ditch Laburnum instead; some trains already do this.

Scrap the other express variations to make the timetable more legible; the Box Hill-Richmond super express saves all of 2 minutes over BCGR (literally 1 minute each at Camberwell and/or Glenferrie). The mini-expresses should also be axed, it is a complete waste of time having a train stop all stations from Belgrave or Lilydale all the way to Camberwell only to skip all of four inner-city stations. It already takes 45 minutes to get from Belgrave to Box Hill, it makes the Wyndham Vale V/Line look like the gold standard for rapid transit.

That just leaves the too oddballs: the 9:16AM Belgrave 'Puffing Billy Express' from Flinders Street, and the 8:09PM super express from Belgrave which also skips Heatherdale, Mitcham and Nunawading (the only train in that direction to skip those three stations, although it makes up for it by inexplicably ending the express at Camberwell, making it the only express service to stop at Auburn, Hawthorn and Burnley).

Heihachi_73 said...

The Hurstbridge line beyond Eltham seriously needs something more than just a train every 40 minutes. Given that there is next to zero chance of the Hurstbridge line being upgraded in this lifetime, upgrade the buses to be a train surrogate instead, providing a joint 20-minute service (e.g. bus every 40, train every 40, but in between each other, not at the same time).

Extend the 580 to Hurstbridge in place of the 343 beyond Diamond Creek, with the 343 shortened to run from Diamond Creek to Greensborough. Add one or two extra evening trips to both routes to bring them up to standard and upgrade the 343 to run seven days. Abolish the 30-minute evening train timetable on the Clifton Hill group so that people beyond Eltham don't get stranded for an hour after the last bus has finished (no way would the 343/580 run until midnight seven nights when we can't even run the orbital SmartBuses).

Naturally, more services between peaks, given that even the Hurstbridge train itself runs every 20 minutes for a handful of times in peak. Wattle Glen, much like Officer, Cave Hill and "Lilydale" (the houses are almost up to Coldstream station), is now beginning its Big Build housing phase, which unfortunately looks like it's also going to be typical outer-suburban black-roofed McMansions shoehorned into lifeless car-centric spaghetti roads without a single shop for miles, in stark contrast to the detached houses, strip shops and even fast food restaurants in Diamond Creek and Eltham, or the apartments in the inner/middle suburbs.

Anonymous said...

Nearly all the population in a 5K radius around Craigieburn Station is to the west of the railway track. Similar with Roxburgh Park though not to the same extent.
I'm doubtful of the accuracy of the Donnybrook catchment from the site referenced. Actually, no, I'm confident of the inaccuracy of that number, and hence doubtful of all other population numbers given.
Epping North/Wollert could really do with a railway line and station. Despite a large gap between the population or residential areas of the Wollert and Donnybrook suburbs, the further north any Wollert extension can go to Donnybrook, the handier it would be those residents who rely on the existing VLine service or Craigieburn metropolitan service.
The analysis is a snapshot of the present time. If it will lead to changes (read, "improvements") in the services or infrastructure, then how long until those changes are implemented and what will be the population catchment at that time?

Craig said...

Regarding outer Hurstbridge Line frequency mentioned in the comments - since the recent duplication at Montmorency and Diamond Creek, the peak service is now every 20 minutes right out to Hurstbridge, inculding in the counter-peak direction (used to be as bad as hourly late weekday afternoons).

The government and Metro Trains have failed to promote this at all beyond the duplication project newsletters.

Arguably a 20 min frequency could be done full time (even nights) with a 10 min daytime service to Greensborough added later on (maybe even allowing the Hurstbridge train to skip smaller stops like Eaglemont, Darebin & Alphington to save a couple of mins from the outer end)

The 343 is largely redudant as a train supplement these days, especially if you decide to run a 20 min service to Hurstbridge interpeak.

There remains some argument for a parallel bus to serve the gaps between the stations - but it probably doesn't need to be every 20 mins, especially with the current density past Diamond Creek.

A bus along there every 40 minutes would be sufficient, potentially providing savings to fund a bus around Panton Hill and St Andrews (which has strong community desire).

Peter Parker said...

@Craig - I've largely ignored Hurstbridge as it's not a contiguous part of the urban growth boundary. Its catchment within 5km is only 13000. That's much less than even Somerville or Hastings. You might do it so you can claim that every station has a maximum 20 min wait but a heap of other things would have higher priority.

Craig Halsall said...

@Peter - I agree but also feel Diamond Creek is short changed currently...back of the envelope calculation suggests a 20 min service only to Diamond Creek is also possible but involves those trains laying there for about 25 mins to slot in between Hurstbridge trains (when could otherwise make it to the end of the line & back)

Either way they aren't using a Big Build asset to it's full advantage outside peaks