Why is the Night Bus stop for Bentleigh station not at Bentleigh station?

Sometimes the little things matter. Sometimes a bit of thought can result in a better outcome.

Logic says that when they put in a new bus route to serve the station, it should stop at the station.

When the signs for the new 979 Night Bus went up at Bentleigh, I noted that they were not at the route 703 stops next to the station.

Eastbound it shares the 701 stop, which is about 100 metres from the station (positioned so the 701 can do a loop around so it can terminate there). Westbound it shares the 703 stop 120 metres away from the station.

At the time this seemed fair enough — it was the end of 2015, and the station precinct was about to get knocked down and rebuilt for level crossing removal. The closest bus stops would be out of action for a time.

But it’s been many months since the new station opened. The 703 now stops just metres from the station entrance.

Having the Night Bus use the stops closest to the station would be beneficial:

  • Easier train/bus interchange
  • Better security at the bus stop from being visible from the station concourse, where the PSOs patrol and station staff are on duty
  • Passengers could use the benches and the waiting room while waiting for the bus — useful on cold winter nights
  • A more logical, intuitive location for a Night Bus stop apparently designed to serve the station
  • As with the current location, no clash with daytime bus services, as the 703 doesn’t run after midnight

This is just one example of course. Are there others where the service nominally connects at the station, but the stop is some distance away?

And this issue may be obvious, but it’s a pointer to other problems.

Other problems with Night Bus

Some routes have stops chosen for their proximity to stations or hospitals, but stop names have been based on the nearest obscure cross street, which is meaningless to most people:

In some spots, they haven’t thought too much about where the Night Bus stop signs should be installed:

Timetabling is also a problem. In the case of Night Bus 979, there are hourly trains and hourly buses, but they don’t connect at Bentleigh – hop off a train from the city at 33 past the hour; you’ve just missed the eastbound bus at 25 past the hour. Instead they have timed connections at Elsternwick on the Sandringham line. It makes sense in some ways, but how does one explain that to passengers?

These issues might underscore problems with the design of the Night Bus network. While the 24-hour tram and train services have been relatively successful (despite poor frequencies, particularly on the trains) the buses have been less so. I suspect a big part of that is the routes being totally different from the daytime routes.

Even in the Doncaster area, where the backbone of the network is four Smartbus routes, they chose not to run those routes through the night, but instead run a completely different route that does a long loop around Bulleen and Templestowe. (I know there are proposals to revert to using the Smartbus route structure. Even at lower frequencies, this makes a lot of sense.)

Possibly there are some areas where it’s logical to provide night service that is not based on the daytime routes, but these should be the exception rather than the rule.

The stop placement, stop names, passenger familiarity with the routes — all these are solved by running 24 hour routes instead of special night-only routes. This means there’s a great opportunity to improve Night Bus services, and I hope PTV’s review of the Night Network will give this serious consideration.

V/Line peak patronage 2012-2017

Five years ago I published some figures on V/Line’s peak patronage, based on their capacity reports, where they show the number of seats occupied in each service in peak hour.

Here’s an update comparing with their figures for February 2017.

Line 2012 2012 2017 2017 Change 2012 2012 2017 2017 Change
  Trains AM ๐Ÿ˜ Trains AM ๐Ÿ˜ AM ๐Ÿ˜ โฌ†๏ธ Trains PM ๐Ÿ˜ Trains PM ๐Ÿ˜ PM ๐Ÿ˜ โฌ†๏ธ
Geelong 11 2917 16 5920 102.95% 8 3292 13 4800 45.81%
Ballarat 10 2771 13 4131 49.08% 9 3059 11 3978 30.04%
Bendigo 10 2612 8 2229 -14.66% 7 2572 6 1609 -37.44%
Seymour 4 792 4 883 11.49% 4 831 3 820 -1.32%
Gippsland 4 746 4 741 -0.67% 3 859 3 619 -27.94%
Total 39 9838 45 13904 41.33% 31 10613 36 11826 11.43%

Apologies for the emojis; this blog format is not well-suited to wide tables. The face means it’s measuring the number of passengers. (Yeah they’re not smiling.)

AM peak is all trains arriving in Melbourne before 9am; PM peak is all trains leaving between 4pm and 6pm. And note that the maximum load quoted by V/Line is 100% — if a train has all seats filled and passengers standing, it is under-represented in these figures.

Why is AM peak higher than PM peak? Because it covers a wider span of time. A number of people would travel into Melbourne in AM peak, but travel back either before or after the PM peak, and those numbers aren’t published by V/Line.

Overall, clearly there is a lot of growth (41% in AM peak, 11% in PM peak), but this is focused on some lines far more than others, partly reflecting the changes on the V/Line network since then.

The biggest change in this time is that the Regional Rail Link project was completed, giving trains from Ballarat, Bendigo and Geelong their own tracks through suburban Melbourne. It’s not perfect, and V/Line’s operations on the line leave a lot to be desired, but it has opened up the way to far more trains and passengers.

Waiting to board a Geelong train at Southern Cross, PM peak

The Geelong line has seen the biggest growth, with a 53% increase in the number of train services, but a massive jump of passengers of 103% in AM peak, and 46% in PM peak. A big factor is that the line now serves the new stations at Wyndham Vale and Tarneit (the latter being the busiest V/Line station apart from Southern Cross) as well as Deer Park.

The irony is that having fixed the problem of large numbers of suburban commuters on the Bendigo line (see below), Regional Rail Link re-created the same issue on the Geelong line. There are proposals to extend Metro services to Wyndham Vale and Tarneit, but don’t hold your breath on that.

The Ballarat line is also showing strong growth, with 49% more passengers in AM peak, 30% in PM peak. Caroline Springs opening in January would have contributed, but I suspect there’s far more growth at the existing stations of Melton and Bacchus Marsh, as well as Deer Park, which are all growth areas.

Note that V/Line train lengths can vary. In the case of Ballarat, 19 peak hour trains in 2012 vs 24 in 2017 is a 26% increase in services, but in that time the seat counts have jumped from 3195 to 4811 in AM peak (up 50%), and from 3711 to 4375 in PM peak (up 18%), or a total increase of 33%. PM peak in particular is well below the growth seen on the line, so clearly more carriages are needed.

(Sorry, I don’t seem to have the detailed figures from 2012 for the Geelong line handy.)

Over on the Bendigo line, patronage has actually dropped, as has the number of trains. Why? Because the Sunbury electrification opened in November 2012, taking a lot of passengers from one the Bendigo line’s busiest stations to Metro services — the definitions for peak are different, but it’s about 1500 people from Sunbury, and about 180 from Diggers Rest.

While passengers at Sunbury still have the choice of boarding V/Line trains, they only run about once an hour, even in peak, compared to a Metro train about every 12 minutes, and the travel time is much the same, with no fiddly changing trains if you want one of the Loop stations or Flinders Street.

With the limited figures available in public, it’s difficult to determine how much Sunbury patronage has grown since electrification, but I’d expect the rest of the Bendigo hasn’t lost any other passengers overall, and has probably grown.

The Seymour line seems to have seen moderate change, with 11% growth in the AM peak, and a very slight decrease in PM peak. I’m surprised there hasn’t been a bit more growth thanks to the price cut in 2015 — $4.10 one way is dirt cheap to get all the way out to Wallan or Heathcote Junction, but there might be other factors — some stations might have no or very poor connecting buses, a hostile environment for walking and cycling, and a full car park. As always, the fare is only part of the equation — service quality matters more.

Gippsland line patronage is stagnant in the mornings, and well down in the evenings. This is probably due to a lot of Pakenham passengers having shifted to Metro — of the three PM peak services, two are marked as Pick Up only at Pakenham (this changed since 2012), and one runs express — all designed to ensure seats are saved for those travelling longer distances.

Boarding a Geelong train, evening peak

Investment in public transport usually goes where the crowding is. It might be some time before we see major infrastructure changes on the Seymour or Gippsland lines, though there are proposals to duplicate and upgrade the latter to counter the job losses in the Latrobe Valley by providing better access to employment in Melbourne.

The Ballarat line is being duplicated to handle more trains, and cut reliability problems, which is good. Hopefully the Bendigo line is coping for now. It’s unclear what will happen with the Geelong line in the short term, but clearly V/Line and the government need to deal with the growth that’s occurring across the network.

Caroline Springs train/bus connections: could be a lot better

Caroline Springs Station opened in January, and is already being used by a fair number of people.

But the station has a weakness: it’s in the middle of nowhere, a not-very-pleasant one kilometre walk from the nearest houses.

Which means you pretty much have to use transport to get there: bike, bus, or car.

Of course you shouldn’t have to own a car to be able to use public transport, and relying on getting passengers to the station by car is not only extraordinarily expensive, it doesn’t scale up as the station gets bigger.

Bus to train; train to bus

So given the trains are fairly infrequent most of the time, and the connecting 460 bus is fairly frequent (3 per hour on weekdays; better than most of Melbourne’s appalling middle/outer suburban buses), and the connections are critical to the usability of the station, how good did they make the connections?

It’s easy enough to compare the train and bus connection times. I looked at “up” (bus to Caroline Springs, then the train to the City) and “down” (train from the City, then the bus from Caroline Springs) connection times. Of course some people might travel between the suburb and destinations further out, but I’d assume most would be travelling towards the City.

Caroline Springs Station

All the connection times are in a spreadsheet here. (Did I mess anything up? Leave a comment.)

Summary results:

Day Direction Trains with bus connection Total trains Average wait time % 10 mins or less
Weekday To City 23 24 0:15 26%
Weekday From City 24 25 0:11 50%
Saturday To City 13 13 0:13 38%
Saturday From City 13 15 0:15 38%
Sunday To City 11 12 0:22 18%
Sunday From City 10 12 0:13 50%

Overall the connections are pretty good at peak times. Most connection times are under 10 minutes, meaning a minimum wait for changing between the trains and bus.

But outside peak times it’s a real mixed bag. At lunchtime weekdays you might face a 28 minute wait for the bus. And if you arrive on the 3pm train from the City during school terms, it’s a 12 minute connection, but outside school terms (eg this week) there’s no bus until after 4pm!

On weekends they’ve clearly prioritised trips to the City in the morning, and trips back in the afternoon. If you live elsewhere but want to visit friends or relatives in Caroline Springs for the day, you may end up facing a 25+ minute wait in both directions. It’s probably quicker to take the bus instead to/from Watergardens, where the trains are far more frequent and the connection times will be shorter.

On Sundays, most connections to the City are terrible. In 8 out of 11 cases, you’ll be waiting more than 20 minutes for the train.

Caroline Springs Station

Where the bus connections really fall down is in the late evening.

  • Monday to Thursday, the last two trains (arriving Caroline Springs at the not exceptionally late times of 9:58pm and 10:47pm) have no bus connection.
  • On Fridays, those two do, but the later arrival (11:58pm) doesn’t.
  • On Saturdays, the last train (arriving 12:17am) has no bus connection.
  • On Sundays, the last two trains (arriving 9:17pm and 10:37pm) have no bus connection.

Oh, and if you’re up early on a weekday and hoping to catch the first train to Melbourne at 5:17am, you’ll have to find another way to get to the station; the first bus arrives an hour later.

Bus stop at Caroline Springs Station

Should connections be guaranteed?

Timed connections have improved since PTV started overhauling them when South Morang station opened in 2012. Before then, there appeared to be little or no thought given to suburban bus/train connections.

Of course, even good connections are only good if services are on time. If they’re not, it’s not clear what the protocol is. Trains can’t really wait, but a departing bus should be able to wait if the train is delayed — and let’s face it, with ongoing punctuality problems on V/Line, this is not unheard of.

Maintaining that connection depends on the bus drivers having information about when the train is expected, and when I looked I saw no evidence that this was the case.

Do V/Line’s operational people notify the bus company if that happens?

There’s nothing in the literature I saw that indicates any kind of guarantee of connection. This is especially important late at night. You don’t want to be missing the bus home because the train was 6 minutes late and the bus driver took off before it arrived. And if you can’t rely on the connections, will you use them?

PTV-liveried train, tram and bus

Connections are important

You wouldn’t build a road without connections to other roads, and nor should you run public transport that way.

Connections are essential for ensuring individual routes form a network, making each route useful for far more people.

Connections of course are a lot easier when one or both services are frequent. One of the benefits of the 10 minute suburban trains where they exist is that even if the bus is late, you’ll never wait very long to continue your journey. And if you’re coming off a train onto one of the tram or Smartbus services (at least when they’re running frequently) then likewise, a high level of coordination isn’t necessary — it just works regardless.

But this isn’t possible everywhere, so authorities need to get better at how they plan and manage good connections around the public transport network.

The ins and outs of fare gates

Fare evasion is a drain on the public transport network. But sometimes the “tough on fare evasion” rhetoric isn’t matched by practical, non-confrontational approaches to reduce it.

Across the public transport network, there is a mix of strategies. On Melbourne’s train network as on many big busy systems, there are too many passengers to check all tickets manually, so other means such automated fare gates come into play.

Before Myki, and before Metcard, there were “manual” fare gates. At central city stations, paper tickets would be visually checked by staff as you passed through a gate/past a booth. Checking an individual ticket was quicker, but this required huge numbers of staff to cover peak times. On weekends in the 80s and early 90s there were usually minimal staff on duty, often resulting in queues to enter and exit places like Flinders Street Station even at relatively quiet times.

And suburban stations had physical gates, which not only stopped people running in and trying to catch a departing train, but also enabled station staff to do ticket checks. By the late-80s, staff cuts meant this became more and more rare, and they were all removed (or opened permanently) during the 90s. (In fact, just before Metcard arrived in the 90s, station staff were often so scarce it was impossible to buy a train ticket at some times of day.)

Caulfield station gate

How many trips go through gates today?

Automated fare gates were initially installed with Metcard at central city stations only, but they’ve gradually spread.

Nowadays the policy is that all Premium Metro stations (staffed first to last train) have gates installed when they are built or rebuilt. By my reckoning, that means that the number of stations with gates has grown to 25:

  • Flinders Street (apart from an ungated entry/exit at Elizabeth Street sometimes opened at peak times due to crowding)
  • Southern Cross
  • Flagstaff
  • Melbourne Central
  • Parliament
  • Richmond
  • South Yarra
  • North Melbourne
  • Bayswater
  • Bentleigh
  • Box Hill
  • Camberwell (platforms 1+2 only)
  • Caulfield (platforms 2+3 only)
  • Dandenong
  • Footscray
  • Frankston
  • Glen Waverley
  • Glenferrie
  • Mitcham
  • Ringwood
  • South Morang
  • Springvale
  • St Albans
  • Sunshine
  • Williams Landing

(Have I missed any? Leave a comment.)

Myki gates being repaired

So, based on the 2015 patronage (station entry) figures on the PTV site, we get the following:

  • Total boardings per weekday: 750,050 / per week: 4,362,850
  • Boardings at gated stations (including partially gated) 400,880 (53.4%) weekday / 2,344,910 (53.7%) per week
  • Boardings at ungated stations: 349,170 (46.6%) per weekday / 2,017,940 (46.3%) per week

So 53% of station entries are through a gate. But how many passenger trips is that? That is, what proportion of trips pass through a gate at the start or the end?

A calculation based on some assumptions:

  • Assume that 95% of entries into central (City Loop plus North Melbourne and Richmond) stations are going to the suburbs (rather than to another Central station)
  • Also assume that most passengers take two trips per day, from home to a destination, and back again. Therefore their destinations when entering at a central station are distributed to the gated or non-gated suburban stations to the same ratio as the entries into them.
  • Treat the partially gated stations as fully gated.

My brain was hurting after figuring out a way to do it in Excel, but eventually I got a figure of 82.4% (on weekdays) of passenger trips having a gate at the start or end of the trip. (For the whole week it was 82.3%)

Also notable in the figures:

  • That 95% assumption doesn’t actually make a huge difference to the final result if it’s tweaked to say 90% or 98% or even 100%.
  • There are far more suburban entries than central entries, which means there must be a (perhaps) surprising amount of suburb to suburb travel — about 23% of all station entries. I suspect this reflects the growing importance of activity centres like Glenferrie, Caulfield, Box Hill, Footscray, Frankston and Dandenong, as well as places like Huntingdale (for Monash University Clayton).

So anyway, about 82% of Metro trips are going through a gate. In theory.

Southern Cross V/Line gates

Keeping them closed

But fare gates only work at discouraging fare evaders if they are kept closed.

Sometimes, a gate is simply left open. There’s no consistency, and it’s a long-running problem that includes central city stations.

Fare evaders can of course jump gates, but why would we make it so easy for them that they don’t even need to do that? (Not that Melbourne’s fare gate design makes them particularly difficult to jump.)

“But Daniel!” you say, if gates are kept closed you need a staff member standing next to them all the time! Not so. And this is where we get into station design.

Parliament station (north end) Myki gates and bypass gate


World’s best practice is to design station concourses so that the booking office is adjacent to the gate, and the staff member inside has control to release/open the gate to let through people who can’t get through themselves — those with wheelchairs, awkwardly large luggage, prams and so on.

With many stations being rebuilt during level crossing removals, there’s an opportunity to get this right. This is precisely how the new Bentleigh station is configured.

Which means that sometimes, you’ll find all the gates closed, as they should be, and the staff member on duty watching from behind the window, ready to help.

Caulfield station, inbound passengers during evening peak


At interchange stations, there are problems with passengers having to leave the paid area (and often go out onto the street) to change platforms. Prime examples include Caulfield, Camberwell and Footscray. Ideally these need over or underpasses for interchange, to make it a quick process that doesn’t require touching on and off along the way.

Another, far more expensive option is to reconfigure the lines and the platforms so that the most common changes of train are cross-platform.

Caulfield is notorious for delays in evening peak coming off platform 4. And it’ll get far worse when the metro rail tunnel opens next decade, requiring more changes between the Frankston and Dandenong lines.

Gate numbers and speed

The number of gates, and speed are issues to ensure delays don’t snowball. The newer Vix gates are about twice as fast as the older Myki gates. And on a bad day, the older gates are hopelessly inconsistent.

Flagstaff, shown below in 2011, used to clog up at busy times. The installation of faster gates, as well as an increase in the total number of gates (achieved by moving the booking office) has mostly fixed those problems.

Queue at gates, Flagstaff station, 8:52am (back when there were hybrid Metcard/Myki gates)

Configuring for peak

Another issue is something central city station staff are used to, but suburban station staff are still figuring out (and/or may need training on): balancing the peak vs off-peak directions.

Sometimes you’ll find queues to exit the station because there’s insufficient gates switched to the peak direction. This should be easily fixed.

Station gates

Ticket checks on every trip

Getting people to pay their fare is a mix of making it easy to do so, and hard to avoid.

Ultimately, fare gates can be used to help reduce fare evasion, being a cost-effective way of giving people the expectation that they will get checked on every trip.

But if we’re going to have fare gates around the network, and if they’re going to work well, they need to be designed and used properly.

* * *

See also:

  • Marcus Wong: How fast are Melbourneโ€™s ticket gates? (the pre-Vix variety)
  • Want to make the system free? First find $700m per year to pay for it. Then show how it will get people out of cars when the overriding factors are about service quality, not fare cost.
  • There is ongoing debate about the merits of gates vs Proof Of Payment systems (eg with random checks for compliance). Some city metros (particularly in Germany) have gone to POP.
    In Melbourne, this is how tram fare enforcement works. My personal view is it has been shown to be problematic in terms of low checking rates and confrontations between inspectors (Authorised Officers) and people who haven’t paid, either inadvertently or deliberately.

PS. An industry insider has noted another side effect of fare gates is that because staff need to be nearby (visible) this has a significant effect on safety, and perceptions of safety — far more so than stations without staff, and even staffed stations that don’t have gates. Interesting.

The State Coalition pledge for 10 minute trains

On Sunday the Coalition pledged to upgrade more lines to trains every ten minutes off-peak.

Here’s the Channel 9 story – have a watch:

There’s not much detail yet — nothing on David Hodgett’s web site that I can see (though it’s a bit broken; you can’t see news items beyond the first page). Nothing on Matthew Guy’s web site, nor the Liberal Victoria web site.

But from Hodgett’s quotes in the story, what they appear to mean is 10 minute services on every line where the infrastructure supports itwhich is most of them.

This would have huge benefits across the rail network, and for the PT network as a whole, by making anywhere-to-anywhere trips much easier by cutting waiting/interchange times, and help grow capacity and patronage without putting extra stress on peak hour.

Given increasing travel (and traffic congestion) at what used to be seen as “off-peak” times in the middle of the day and on weekends, as well as our growing population, this is just what’s needed.

And it would bring us up to the type of train service offered in most big cities around the world.

It’s not actually that radical. PTV and Metro already have an unfunded plan for just such a rollout. It’s really a matter of timing.

Flagstaff station

For such an obviously good idea, the cynics really came out on this one — perhaps far moreso than if Labor had pledged it.

The Libs never do anything for public transport!

Most of the 10 minute services we have now were actually introduced under the Coalition government in 2012. It was also under the Libs in 1999 that the appalling Sunday 40 minute services mostly got doubled to 20 minutes.

Both sides do good stuff and bad stuff. It’s certainly not as simple as saying Labor’s good at PT while the Coalition is bad. (Indeed, one sweeping and partly flawed generalisation is that Labor is better at infrastructure, and the Coalition is better at services.)

There’s no capacity for this!

This does not put stress on signalling, track capacity or the train fleet, because it’s off-peak, not peak. It does not need major infrastructure such as the metro tunnel.

It needs extra drivers, and extra maintenance capacity, as well as other running costs such as power.

It’ll cause chaos at local level crossings! It’ll devastate communities!

No it won’t. We’re talking about fewer trains than in peak, and at times of less traffic on the roads as well.

It doesn’t cause huge problems on the lines that already have it. And in fact improving train services like this is likely to get more people out of their cars, as long as they’re actually promoted to non-train users (which is rare).

There’s only one remotely sensible argument I’ve heard against doing this now (rather than later), and it’s this:

Is it the right time to be bringing lots more patronage onto the off-peak/weekend service when there are so many infrastructure projects in progress? Level crossing removals on most lines, track and signal upgrade works, works to enable new trains in some cases. It adds to stress on the replacement buses and doesn’t provide the best environment to win over customers permanently and produce ongoing travel pattern changes.

I think that’s a fair point, but there are likely to be works such as level crossing removals going on for decades, as hopefully this will be an ongoing program.

Perhaps one way of playing it is to implement frequency increases as lines get their worst level crossings removed. For instance:

  • The Sunbury line for instance now has no level crossings between the city and Watergardens (apart from a minor crossing into a rail yard at Tottenham). And the line has no more crossings scheduled to be removed except Melton Highway, which will only effect the lesser-used outer end of the line.
  • The Sandringham line has no crossing removals scheduled.

In fact, frequent trains can make life a lot easier when bus replacements are in operation. Buses are sometimes delayed by heavy traffic, and train connections not held. If the train is missed, it’s a lot better if the next one is only ten minutes away.

Electric train passing

Other things:

In the news clip, Labor claims to have introduced a large number of services since 2014. Not so. Barely any in fact.

I’d love to see this policy implemented. But before they definitely win my vote, I’m concerned as to what else the Coalition will pledge in their policy platform. They still seem obsessed with the disastrous East West Link.

Nonetheless, this is a great policy pledge from the Coalition. If Labor are sensible, they’ll match it.