Train services vs patronage

A follow-up to last week’s post mapping out the number of train services per station.

Commenter Andrew suggested I compare it against patronage – happily I’d been working on this anyway!

Here it is – and as a bonus I’ve got hold of 2017-18 station patronage data.

The darker the blue, the higher the boardings per service. Click on a station to see the numbers.

What’s interesting is that the top ten stations for boardings per service are not what you might expect.

  • Melbourne Central 51.060
  • Sunshine 46.142
  • Southern Cross 45.337
  • Box Hill 44.133
  • Flinders Street 42.08
  • Williams Landing 41.522
  • Watergardens 39.459
  • Footscray 39.360
  • Essendon 38.933

Remember, these are figures across the entire day. Peak would be a lot busier.

I was surprised that two of the top five are well outside the CBD. And if you’re boarding at Williams Landing or Watergardens and you think it’s busy, it’s not your imagination.

Is it an anti-west bias? I wouldn’t jump to that conclusion. Historically, development in Melbourne has spread further to the east and south than the west and north. So the train lines are longer, with more stations, and thus overall higher patronage. So more train services are needed.

Of course the north and west are now growing fast. Is provision of PT keeping up with demand? Nope… but that’s the case across most of the city.

Charting the numbers

Here is the data on a chart – services per weekday on the X axis, and boardings on the Y axis.

Note the patronage (Y axis) is on a logarithmic scale. It was even messier without this, because just a handful of stations have daily boardings up between 10,000 and 100,000 per day.

The middle range, between 500 and 10,000 per day is a bit of a dogs breakfast because of a large number of suburban stations in that range – and in many cases, they also have identical service levels, thanks to being on the same line. More on these below.

Broadly, you can see that the higher the patronage, the more services a station probably gets.

But does patronage follow service, or service follow patronage? A little of both – authorities tend to try and act to demand/crowding, but the convenience of frequent services attracts more passengers (including more development around stations – hello South Yarra!)

The underground stations at Parliament and Melbourne Central (and to a lesser extent Flagstaff) get a lot of people going through them, despite only having four platforms and about half the number of train services as Flinders Street.

The highest ranked Zone 2 station is Box Hill, followed by Dandenong. Some zone 2 stations have seen a big increase in passenger numbers since the fare cap was introduced in 2015.

Zooming in on the suburban stations of under 7000 boardings per day, things become a little less muddled:

A few stations like Newport, Camberwell, Clifton Hill and Burnley are getting more trains than you might expect for the patronage, thanks to their junction status. They are important for interchange, and different lines converge there, resulting in more services. (In contrast, Caulfield gets slightly more services, but has the patronage to match thanks to the adjacent university campus.)

Some inner-city stations also get more services than the patronage might suggest – but this certainly is not an argument to cut services – one could expect recent infill development has seen the numbers of passengers continue to increase in recent years. Consistent stopping patterns are also important – running some trains express burns up track capacity and creates confusion.

There are also some very busy middle and outer suburban stations that appear not to be getting enough services, such as Sunshine, Essendon, Glen Waverley, Watergardens, Werribee and Hoppers Crossing.

This Channel 9 story in February (which is where the patronage data came from) highlighted the mismatch between demand and supply:

Regulars here would know my message from all this: as Melbourne grows, so does overall travel demand. The number of services provided around the rail network needs to continue to grow, including outside peak hours – as well as boosting tram and bus services.

  • See more on the Channel 9 patronage story from February
  • Patronage in 2017-18 at some stations was affected by major works, for instance Rosanna (and surrounding stations), the Dandenong line and Frankston all had closures.
  • Projects such as the metro tunnel will enable increased train services on many lines, but the timing and number of these is unclear.
  • I’ve been reminded that Philip Mallis posted a different set of data, showing annual patronage at stations, including both V/Line and Metro.

Visualisation: how many services from your station?

I don’t know why I didn’t think of this before: slicing and dicing GTFS data is not as intimidating as I thought it might be, and it’s pretty easy to import into Google Maps so you can visualise it.

Once you find what you want and figure out its quirks, you can get some pretty useful stuff.

Let’s start with this: a map of the total number of weekday Metro train departures per station. Click here (or use the View Larger Map icon) to see it bigger.

It’s easy to import into Google Maps, though there is limited control over the groupings and colours.

So, the stations are grouped into:

  • red (under 140 departures per weekday)
  • yellow (140-202)
  • green (203-262)
  • and cyan/light blue (above 262).

Departures includes both directions (including arrivals for terminus stations), per day, for Monday to Thursday. Fridays run a slightly different timetable due to Night Network overnight services.

A few things I noticed

Unsurprisingly the central stations have the most train services, with lines except Stony Point (and Alamein outside peak) converging there.

Junction stations such as Caulfield and Clifton Hill (and inner stations on that line) that serve multiple lines also do well here, as do major stations served in peak by expresses and stopping trains, such as Box Hill, Glenferrie and Newport.

Dandenong probably just sneaks into this category thanks to late-night shuttles from Cranbourne adding to the departure/arrival count.

Green accounts for most stations on the lines with frequent all-day services out to Frankston, Dandenong and Newport, as well as some stations that have very frequent peak service such as Box Hill to Ringwood.

Yellow is mostly lines that run only every 15-20 minutes off-peak, but have a more frequent peak service.

Why do Middle and West Footscray and Tottenham not quite make yellow? Because they are bypassed by some trains from Sunbury in peak, which probably helps balance train loads, but adds to waiting time.

It’s a similar story for some stations that are yellow, not quite green: Highett and Southland, and many of the zone 1 stations on the Ringwood line.

Red stations have relatively infrequent service: all-day 20ish minutes (such as the Altona Loop, Williamstown, Alamein and Upfield lines) or good peak but infrequent off-peak (such as the outer sections of the Sunbury, Hurstbridge lines, and the Ringwood and Dandenong branches) or just very few services such as the Stony Point line.

The interesting thing to look for is red stations close to the City. Some of these I expected, such as the Upfield line, with only a basic service all day, in part due to the single track.

I didn’t expect inner sections of the Hurstbridge line to show up in red. Stations such as Westgarth fall just below the 140 threshold, thanks to only 20 minute frequency off-peak, and some peak express trains bypassing them.

Craigieburn line at Parliament Station, 2:57pm on a Monday

The need for upgrades

The nature of a visualisation like this is that not all stations can be cyan and green.

But the challenge for government is to boost services to those red and yellow stations, particularly those close to the CBD, to keep up with demand – not just train crowding, but overall suburban travel demand, including that sparked by infill development.

While adding peak service is tricky on some lines due to capacity, a boost to all-day off-peak/weekend/evening frequency is easy – and would bring huge benefits to passengers by cutting waiting times and crowding, and improving connections.

Old photos from May 2009

It’s the end of May, so it’s time for a batch of photos from ten years ago.

I quite liked this ad for iiNet in Flinders Street Station:

iiNet advertising, Flinders Street Station
iiNet advertising, May 2009

I was snapping photos for Tony’s web site “Our Fading Past” (currently offline, but the Google Map is up), highlighting old signs around Melbourne. This classic is in Ripponlea on the side of a heritage-listed pharmacy.

Old Kodak advertising on side of chemist in Ripponlea, May 2009

On the 70s-era Z-class trams, you can still find these things. Originally these trams had front-door boarding, so you could have your ticket checked by a conductor. The rear doors were used to alight, and these little screens would light up to indicate you could press the button to open the doors. I think these days most of them are wired up to the Next Stop buttons.

Z-class tram, Open Doors indicator, May 2009

Also arriving on trams: new Myki readers. Myki became active on Melbourne trains at the end of 2009, and on trams in mid-2010, with the touch-off requirement removed to make it work with the slow readers.

Then new Myki reader on a tram, May 2009

Excitement! New screens arrive at Richmond! Evidently some teething problems initially.

New screens at Richmond station, May 2009

At this time, there were still plenty of CRTs around the network. These were at Parliament station.

Parliament station, May 2009

A visit to the Dandenongs…

In the Dandenongs, May 2009

…wouldn’t be complete without a photo of one of greater Melbourne’s worst bus stops for accessibility.

Bus stop in the Dandenongs, May 2009

A foggy morning at Glenhuntly station

Glenhuntly station, May 2009

…and here’s a view at the other end of the day from Richmond. Note the lack of platform cover, and the Hitachi train.

Richmond station, May 2009

How does the new local bus route fit into the network?

Good news: New bus route 627 starts in June, running from Moorabbin to Chadstone via East Bentleigh and Murrumbeena.

Excuse the micro-transport-blogging about this specific route in my local area, but (as usual) there are considerations that are relevant across the network.

This new route fills some gaps in the local network, including East Boundary Road (which is meant to be part of the Principal Public Transport Network, but currently has sections with no public transport), and sections of Tucker Road and Jasper Road. New bus stops are under construction.

Route 627 bus stop under construction

The route will make trips to the huge Chadstone Shopping Centre easier for people in Moorabbin and parts of Bentleigh and East Bentleigh who currently don’t have direct services – remembering that changing along the way is not a good experience given low bus frequencies, especially on weekends, which are the busiest shopping days.

The 627 will add another connection to the Frankston line from the Dandenong line, which can be useful when major delays (planned or otherwise) hit one line or the other, and it’ll serve the new McKinnon Secondary campus, though it won’t connect to the existing campus.

It adds to frequency on about 3.5km of route 822. (It also shares parts of routes 701 and 703, though not really in a way that adds very useful frequency.)

It runs more frequently on weekends than many other suburban bus routes: every 40 minutes instead of the typical hourly frequency. It’ll mostly run half-hourly on weekdays.

The bad news: The route structure is a zig-zag. Not particularly intuitive. This is partly due to challenges with the T-junction at the southern end of Tucker Road.

This route has been laid over the existing local bus network, without any other changes. Ideally this would have been an opportunity to straighten out route 822 to run along East Boundary Road, with this new route filling the gap, and/or straighten out route 626 along McKinnon Road (so that route could connect the two McKinnon SC campuses).

Combined timetable

I wanted to talk in detail about the timetable.

The new 627 adds useful frequency to the section of route 822 between Chadstone and Leila Road (and as far as south as Centre Road if people can walk for a few hundred metres).

But the timetabling doesn’t make the most of this.

If I were planning it, I’d be particularly careful to aim for even frequencies southbound, departing from the most likely trip sources on the shared section: Murrumbeena station (particularly on weekday afternoons) and Chadstone (particularly on weekend afternoons).

Ideally northbound frequencies would be as even as possible too of course, but this is a secondary priority, since people can more easily time their departure from home. Timing your connection from a train arriving at Murrumbeena, or the arrival of another bus or the end of your movie at Chadstone, is a bit harder.

Unfortunately the planners haven’t quite hit the target here. Here’s how the combined service looks at Murrumbeena southbound on weekdays after 3pm:

  • 3:13 (822)
  • 3:23 (627)
  • 3:43 (822)
  • 3:54 (627)
  • 4:14 (822)
  • 4:24 (627)
  • 4:43 (822)
  • 4:55 (627)
  • 5:18 (822)
  • 5:25 (627)
  • 5:48 (822)
  • 5:56 (627)
  • 6:19 (822)
  • 6:26 (627)
  • 6:45 (822)
  • 6:55 (627)
  • 7:14 (822)
  • 7:28 (627)
  • 7:48 (822)
  • 7:58 (627)
  • 8:28 (627)
  • 8:44 (822)
  • 9:04 (627)
  • 9:28 (627)
  • 9:37 (822)
  • 10:04 (627)
  • 10:47 (822) (last bus)

Trains are every 10 minutes or better until 10pm, so the aim isn’t for buses to meet specific trains, but instead to provide a good frequency so that nobody has to wait too long for a connection.

Instead the combined bus departures are irregular. At the commuter peak, gaps vary as widely as 7 to 23 minutes. Not so good.

Given a total of 4 buses per hour, a consistent 15 minute combined service would have been better. (Back in the 1980s, the predecessor to the 822 was the 655, which ran every 15 minutes in peak as far as Stockdale Avenue, East Bentleigh – a bit further than the combined 627/822 route will run.)

These shortcomings aside, there’s a clear opportunity to grow commuter patronage to and from the station, which is great.

The last two bus departures for the night are scheduled 5 minutes after (then not so frequent) train arrivals – which is good – even better if bus drivers can wait a little while if the train is late.

Bus 822 at the old Murrumbeena station

There’s a quirk with travel time.

On the new route 627, Chadstone to Murrumbeena is timed at 7 minutes.

On the older route 822, it’s 9 minutes… despite the two routes being identical on that section. And there’s a similar discrepancy in the other direction.

It appears route 822 hasn’t been re-timed since the level crossing was removed and the route was straightened out to avoid the side street detour so it could stop outside the old station. That’s an issue which affects many routes serving stations rebuilt through the Level Crossing Removal Project.


How does the Saturday afternoon shopping rush look from Chadstone? Much better: each bus leaves every 40 minutes, and they’re pretty evenly spaced until 5:31 when the 822 drops back to hourly – then it’s a bit messier, for instance both routes are timed to depart at 6:11pm.

And weirdly, the 627 gets more frequent after 7:30pm.

  • 2:11 (627)
  • 2:29 (822)
  • 2:52 (627)
  • 3:09 (822)
  • 3:31 (627)
  • 3:51 (822)
  • 4:11 (627)
  • 4:31 (822)
  • 4:51 (627)
  • 5:11 (822)
  • 5:31 (627)
  • 6:11 (822)
  • 6:11 (627)
  • 6:51 (627)
  • 7:11 (822)
  • 7:31 (627)
  • 8:04 (822)
  • 8:06 (627)
  • 8:36 (627)
  • 9:00 (822)
  • 9:06 (627)
  • 9:36 (627)
  • 10:00 (822)

What about Sunday afternoon from Chadstone? Not so good – route 627 is every 40 minutes, but route 822 is only hourly, so it’s messy.

  • 2:08 (822)
  • 2:11 (627)
  • 2:51 (627)
  • 3:08 (822)
  • 3:31 (627)
  • 4:08 (822)
  • 4:11 (627)
  • 4:51 (627)
  • 5:08 (822)
  • 5:31 (627)
  • 6:08 (822)
  • 6:11 (627)
  • 6:51 (627)
  • 7:08 (822)
  • 7:31 (627)
  • 8:00 (822)
  • 8:06 (627)
  • 9:00 (822)
  • 9:01 (627)
  • 10:00 (822)

The obvious solution would to upgrade the 822 to match the 40 minute frequency on Sundays, and to also tweak the weekday timetables to match better.

Bus 822 navigating a side street in Bentleigh East

Shared stops?

No doubt along the common part of the route, the two routes will share stops, including at Murrumbeena station.

At the Chadstone bus interchange? Unclear. Hopefully either a shared bus bay or adjacent bays.

Route number

Another point: Why was the new route given the number 627?

627 used to be the number used for what is now the 625 and 626 when they were one, long, confusing route.

Would it be better as, say, 820, to help sell the shared section with 822? (The route numbers 821, 823, 824 and 825 are already taken.)

Or are there future planned network changes that are coming into play here?

Queues for buses at Chadstone, Boxing Day 2018

The big picture

I’ve focussed on a local route, but the same principles apply across the public transport network.

The addition of extra services (and extra service kilometres) is very welcome. But as I noted in a 2014 blog post, these upgrades should not be planned in isolation.

The real aim is to help each new upgrade improve the overall network.

While the bus system needs more resources, there are lots of gains to be made from reviewing and revising the network: moving towards direct routes, where possible along main roads, and using the efficiencies gained to increase frequency.

The patronage growth in Smartbus routes shows this works, but it’s also been shown in other areas such as Brimbank.

The windy indirect routes are not useful to most people – and as this Twitter thread shows, it’s something of a myth that buses that stop everywhere suit people with limited mobility.

Monday’s State Budget provided not a lot of new funding for buses, so it makes sense to ensure the network is as efficient as possible at meeting suburban travel demand.

Hopefully this new route will help, and other upgrades will follow.

PT service kilometres vs population

In the world of public transport, services are what counts.

Infrastructure is important, but ultimately, infrastructure is only built to enable services.

If the train only runs every 40 minutes, it’s a long wait whether the station is 100 years old, or shiny brand new.

Today is State Budget day, and it’s a good chance to look at the overall level of service around the network. The Budget Papers (the Service Delivery book) includes figures for total service kilometres for each mode.

To put these figures in context to see how the network is expanding against population growth, we can calculate service kilometres per person.

So how’s that tracking? These graphs are based on the 2018 Budget Papers:

(Edit: 7/6/2019: I messed up and used the wrong numbers on this graph. Corrected and updated below.)

Service kms per capita

On the face of it, that looks okay. It’s going up. Bus kilometres in particular show some growth.

However one should always be cautious about the most recent years. 2017 was the expected outcome, and 2018 was the target. We’ll see what those say in the 2019 budget papers.

(My assumptions: I’ve used metro train, metro bus and tram, and a portion of V/Line service kilometres, because V/Line serves parts of Melbourne’s west. And I’ve used Census figures for greater Melbourne’s population.)

Focusing on just Metro Trains, which is the backbone of the public transport network:

Again, an uptick. But the underlying figure for 2018 was a target of 23.8 million kilometres, a rise from 23.1 – or a bit over 3%.

Has that actually happened? Probably – the 7km Mernda rail extension opened, and there were some worthwhile service increases for the Dandenong line and a couple of others.

But is the overall network keeping up with population growth, both on Melbourne’s outskirts and the inner and middle suburbs? Is every line seeing more services to combat crowding across the day?

I’ll aim to update these figures later today – to see if reality has matched the forecasts, and what the plan is from here.

Update 8pm: Today’s budget includes some investment in additional train services (Metro and V/Line), expected from 2020 onwards – though the 2019-2020 expected service kilometres is the same as this year. It’s just as well something’s happening, as Metro patronage is continuing to grow – it’s looking like almost 5 million trips above the target for 2018-19.

There’s no increase in the target for tram service kilometres, and a minor increase in buses.

As a result of all this, for 2019-2020, the target service kilometres per person for Metro trains, and overall across the network, will drop slightly.