How much would trains every ten minutes cost?

One of the gaping holes in Melbourne’s public transport system is the lack of an all-day every day frequent service on the backbone: the Metro suburban train network.

Melbourne is one of the few cities in the world, outside North America, which doesn’t have frequent all day trains.

Other Australian cities are moving towards this. Perth has now trains every 15 minutes to most stations until around 8pm. Sydney does even better – they’re every 15 minutes until around midnight, and the new Sydney Metro line runs every 4 minutes in peak, and every 10 minutes the rest of the time.

The PBO costing

A few months ago the State Parliamentary Budget Office published their cost estimate for the Greens policy of trains and trams every 10 minutes, every day, all day until 9pm: $214 million per year – with $173 million of that being for trains.

Some people thought this was cheap. I actually wonder if it might be even cheaper. One source suggested to me that the train upgrades would be closer to $100 million per year.

How did they calculate it? The PBO’s source document doesn’t have much detail. I did ask, and they said they only included scalable costs, which in theory takes into account the cost savings inherent in using existing assets more efficiently.

The Greens policy scope is larger than some other proposals, as it includes 10 minute frequencies all the way to Pakenham and Werribee.

It’s unclear if the PBO took into account potential efficiencies from driver shifts – a driver who might currently run 1-2 peak services in a shift may also be able to drive some additional off-peak services for no extra cost.

Departure signage at Flinders Street station

The importance of the network effect

The cost would be partly offset by increased fare revenue.

The PBO’s estimate of increased revenue was based on work by Infrastructure Australia, but it’s not clear if this takes into account the network effect. That is, if you run a single route more frequently, that’s good, you get some more passengers. But run a lot of routes more frequently and you make exponentially more journeys more time-competitive, and get a lot more passengers.

It’s like when they made it so text messages weren’t confined to one phone network, but you could SMS anybody. Usage grew exponentially.

(Yes, I am old enough to remember these things.)

Apart from connections between different bus, tram and train routes being easier with high frequencies, it also makes life easier during works and disruptions, when sections of rail lines are replaced by buses, as connections (especially bus to train) involve less waiting around.

Objections to frequent trains

Interestingly, when faced with the idea that most of Melbourne’s trains could run every 10 minutes all day, a few people object to the concept.

Here are some of the points that I’ve seen raised.

“Huge infrastructure upgrades are needed first!”

No they’re not. Although there are sections of single track that are a barrier, most lines can run trains every ten minutes with no issues, because they already run more frequently than this in peak hour.

Comeng trains at Craigieburn TMF

“There aren’t enough trains!”

Yes there are. While there might be some adjustments needed to maintenance, again, there would still be plenty of trains unused outside peak hours, so this shouldn’t be a big problem.

“It’s too expensive”

Big transport networks are expensive. According to the Budget Papers, the fees paid to MTM for running and maintaining the entire Metro network amount to about $1.1 billion per year.

So we might be talking about a funding increase of around 10% per year to make the network vastly more useful for people. And that doesn’t count increased fare revenue.

(Total rail network costs, including Metro, V/Line, V/Line coaches and the strange Capital Assets Charge, which is an internal government accounting trick, are about $3.9 billion. Against that higher cost, this is a tiny increase of about 3%.)

“Nobody travels in the middle of the day”

Melbourne is now a big city. Plenty of people travel outside peak hours.

Vicroads data shows road demand as strong right throughout the day, and weekend demand is nearly as strong as weekdays.

All-day frequent service can also help spread the peak load, by making it more attractive to travel outside peak.

Off-peak trains aren’t crowded

In some cases they are. For instance, western suburbs lines get very crowded during weekday off-peak and weekends – see below.

This plan would solve that, but the main benefit is sparking more demand by cutting waiting times (including connections from other services) to make public transport a more attractive option. This is because transport is supply-led.

Express trains would be better

Some argue that instead of trains every 10 minutes, there should be alternating stoppers and expresses every 20 minutes.

This is messy, eats track capacity, is harder for new users to understand, and means only a fraction of stations get the cut in waiting times. And each station skipped only saves about a minute of journey time.

Studies indicate that perceptions of waiting time can be up to 2.5 times that of travel time. Cutting waiting time is the priority to get more passengers on board.

It’s opex

Yes, it’s operating expenditure, recurring funds, which can be seen as bad for government budgets, unlike once-off capital expenditure. But this is the reality with a public service. You don’t build a hospital and then not staff it properly.

Better services would maximise use of the (substantial) rail infrastructure and fleet, and the cost would be partially recouped by increased fare revenue.

Spreading peak demand, just as was the case with the Earlybird fare, can also seen as a way of saving on upgrades to peak capacity (on public transport and on the roads), which are very expensive.

More broadly, it would assist economic growth by providing more opportunity for people to reach employment and education.

“Every 10 minutes? It should be every 5!”

Perhaps eventually, but let’s walk before we run.

With feeder bus services generally poor, who would use the trains if they ran every five minutes? The risk is they’d be under utilised.

Better to build up the patronage, then see which lines need a further boost.

“What about buses?”

Buses are important too, and many bus routes suffer the same problems as off-peak trains – more so, given some routes run only hourly. So yes, buses need upgrades.

But if you had to do just one mode, I’d start with trains:

  • the fleet is ready to go
  • they serve both short and long distance trips
  • carrying capacity is much higher, including passengers per additional staff member deployed
  • upgrading a network of just 15 lines provides far greater frequent service coverage across Melbourne
  • trains are largely immune from road congestion, so the investment in new services is maximised

“It would lock up the road network”

Unlikely. Outside peak times, local arterial roads (the ones that typically have level crossings) are not under the same stress as at peak hour, and the proposal is for fewer trains running than at peak hour.

The introduction of 10 minute all-day trains on the Frankston, Dandenong and Ringwood lines has not caused chaos on the roads.

And the extra trains would attract trips out of cars onto public transport, giving more people a way to avoid road congestion.

That said, the continued removal of level crossings means there is the opportunity to boost train service levels with less effect on the road network.

Citybound train at Murrumbeena skyrail station

Frequent trains mean huge benefits

Melbourne continues to grow, and all day traffic levels continue to grow.

Running the trains more frequently all day brings huge benefits for many, including better connections across the network, without breaking the bank.

You know the joy of walking onto a station platform and finding there’s only a few minutes until your train? This experience makes using the system far more attractive. That’s the power of high frequency service.

The State government is going great guns on infrastructure, but it’s time they moved on upgrades to services as well, especially a no-brainer like this.

By Daniel Bowen

Transport blogger / campaigner and spokesperson for the Public Transport Users Association / professional geek.
Bunurong land, Melbourne, Australia.
Opinions on this blog are all mine.

30 replies on “How much would trains every ten minutes cost?”

Hi Daniel

Unlikely that there would be any real efficiency gains leading to cost savings from drivers doing more trips in a shift, let alone the 20% suggested.

There are real safety and break time issues which restrict that. And.. truly, drivers /staff.are not the main cost in running Metro. But they are the main variable.operation cost, because most other operating costs are essentially fixed.


I agree we should quickly move to as a less a time as possible. 5 minutes would be a dream! I think though that the government thinking behind the waiting times on some lines has to do with the level crossing removal program. For example the lines that have a 10min wait time have had a lot of level crossings removed compared to the Werribee line. I suspect the government is trying to please the road network as well as the train network at the moment.. Besides when the metro tunnel is completed isn’t the train network going to run every 2-3 minutes?

How about counterpeak services as well – take out the Footscray to Sunshine expresses, and there is an 18-minute gap outbound at times on the Sunbury line (at West Footscray the outbound gaps from 7.12am are 11, 12, 9, 18, 7, 11, 18, 12, 6, 12, 6, 12, 6, 12, 9 before the off-peak 20 minutes starts. Plenty of times I’ve watched my Geelong train (every 20 minutes) pass by while I’m in that 18-minute gap for a 6-minute trip.

But directly on the topic – a 10% increase seems pretty cheap for the utility, partly offset by increased fare revenue.

As a regular user of the Sydney network, with the minimum service level on major lines of 4 trains per hour at all times that services operate (0400-0000 Sun to Thu and 0400-0100 Fri to Sat), which applies even to stations 50km from the CBD (eg Penrith, Macarthur), when I go to Melbourne it feels like the system is trapped in the 1970s by comparison. And the irony is that Melbourne is the city with more nightlife!

Don’t dismiss the infrastructure improvements needed so quickly. The Cranbourne line actually cannot run trains every 10 minutes because the section between Lynbrook and Dandenong is 7 minutes of single track, and from Lynbrook to Cranbourne there’s about another 6 minutes of single track.

The best frequency the Cranbourne line ever runs at any time of day is 14 minutes as a result, at a few points during peak. However during most of the peak period it can only manage 16 – 19 minute frequencies, which is not much better than the 20 minute off-peak frequency.

” Sydney does even better – they’re every 15 minutes until around midnight,”

That’s not really correct. For example, there is a train about every 5 minutes on the line to Hurstville, but due to various stopping patterns many of the stations, for example Banksia and quite a few others, only get a train each 30 minutes. That’s a ridiculous long time to wait for a trip which is only 15 minutes long.

A few years ago, many of the Sydney lines only had 30 minute frequency off peak, worse than almost everywhere in Melbourne.

That has been improved in most places. Checking some of the timetables, it seems that the 30 minute frequency only persists on the suburban network for:

the Richmond/Windsor line,
the Carlingford line, which is about to be closed,
the lines between Birrong and Cabramatta and Birrong and Lidcombe.
Asquith, Mt Colah and Mt Kuringgai, beyond Hornsby.
some minor stations on the Illawarra line.
bizarrely, travel between Auburn and Parramatta.

No doubt, there are some others which I missed.

I think that people who travel off peak during the day are fixed train travellers and an improvement in service intervals to ten minutes would not make much difference to patronage. While it is annoying if you have just missed a train and you have to wait 15 or 20 minutes for the next, phone apps have made a huge difference to how people use trains. You know when to walk quickly to the station or dawdle.

I was not surprised to hear from my brother that there is a train every 20 minutes to Pakenham. The service seems much more frequent when stuck at the railway booms that are down much before they are needed, the problem being terminating trains and through Vline trains.

However on the Sandringham line, it should be a ten minute day time service. It is busy enough, and passengers should not be dumped at Flinders Street Platform 22 and a Half and should go to proper Flinders Street Station platforms.

Never mind how daytime crowded western suburb trains can be. I’ve experienced it and it is awful. I hate to think what it is like in peak times.

Having just spent time in Rome, Milan and Munich using their trains – it was amazing to wander onto a platform and have a train arrive in just a few minutes every time. All if these cities have smaller populations than Melbourne.. Then I come home to Melbourne and every single day in my first week the not too frequent trains are delayed. Disappointing.

1) Does the entire network really need to be? Not much point doing it on lines where there just isn’t demand. For example you could stagger Lilydale and Belgrave lines with 20 minutes frequencies ensuring that you do indeed have a 10 minute frequency for the length of City to Ringwood.

2) Much of the network continues to be in disrepair with signals, switches, poor quality track and foundation, and the need for works like a flyover at Clifton Hill being essential infrastructure that has long been neglected and would impede such frequent operation.

There’s certainly a case for better daytime frequencies, but whether it needs to be every line and every station, not so sure. But the real need lies in evening frequencies – 30 minute intervals are just a joke.

Hi Andrew/Kent

The frequency does influence how people use trains in off peak. There is plenty of research on that. But let me give three practical examples.

In Canberra the switched the evening service on buses – which used to be as infrequent as 60-90 minutes, to a cloverleaf where buses would leave Civic every 20 minutes to say, Gungahlin, then drop off/pick up people on various routes in Gungahlin and then return ( you could text/call with your stop, and the driver would know to go there for pick up). Full buses, less DUI, less city assaults. People took the bus, even though travel time was greater

I drive into the city on the days I have to work until 8pm- and it costs me much more- because of the wait times for the train in the evening makes the trip too long- because of the 30 + minute wait for the train. All other times

Look at the Uber stats of people spending $ in the evening ex cbd.

In areas and periods with frequent train or trams services, a journey just becomes a series of boardings and interchanges. Frequent PT services (every 10 minutes or less) minimizes interchange time so the total journey time becomes comparable with driving and parking. More frequent services greatly increases the network effect and the opportunites to just “turn up and go”.

They all need to be frequent. Latent demand is quite likely to be high, so it is not right to claim that ‘there is no demand’ and that therefore certain lines can make do with lower frequencies. In Lilydale, for example, there are a lot of people who will not travel on trains because they know they will have a high risk of a long wait at either end of a journey, particularly at night when trying to get home. As a result, they drive.

10 minute trains are one thing if you have convenient walking or interchange access to a station, and will certainly be a boon for nights, where you’re too tired to wait for trains. But daytime off-peak? I mean once the station car park is full, it’s full: there might be ten minute services but I might still need to drive.

This is especially true of end of line stations like Mernda, Lilydale, and Hurstbridge, in my experience, where bus connections aren’t great and a lot of people are driving non trivial distances just to get to the station.

Ten minute trains on the Sandy line would be a dream! I would use the train at the weekends and evenings instead of using the car.
When will it be a reality? A long way off I suspect.

In terms of OPEX, I would imagine it will mostly be in the operational and staffing costs of running the trains themselves. Platforms and stations are already running regardless of whether trains run 10, 20 or 30 minute frequencies.

One of the things I’ve noticed on weekends is that the platforms are even busier than usual and this is a potential safety issue. At peak times, crowded platforms might last some 10 to 20 minutes (unless there are disruptions) before clearing when the trains arrive. On weekends, they will stay crowded for much longer, and it’s not helped by the fact that the trains may sit on platforms for much longer.

This estimate needs to be refined as a research project rather than a quick calculation, as has been done here. Infrastructure Australia would have supplied a “typical value” that for a doubling of frequency the patronage increases by 50%. There would be data from Melbourne when frequency was increased on the Frankston and Dandenong sectors that could help refine this value, but was not used in the cost estimate. Once a new frequency is introduced patronage tends to increase over time until it reaches a new equilibrium, which was not acknowledged in this estimate.

When the 10 minute service was introduced on the Frankston and Dandenong lines were introduced, some of the additional costs were offset by not needing to run special services for events such as AFL games, which are relatively expensive in staff time per service. These savings were not included as part of the cost estimate.

The estimate for the train network was for a 10 minute service for the entire line, whereas patronage is most likely to be responsive to the additional frequency in the inner sections , such as to Watergardens on the Sunbury line (saving 23% of service hours, so 23% lower costs relative to a 10 minute service to Sunbury), to Broadmeadows on the Crigieburn line (saving 21%), to Greensborough on the Hurstbridge line (saving 32%), to Ringwood on the Belgrave and Lilydale lines (saving 33% and 25% respectively).

What a dream, not having to wait 27 minutes at Croydon every day because the buses are crap and the trains somehow manage to be even worse (it’s a joke when it’s faster to catch one of two buses from Croydon to Ringwood than it is to wait for the train – the 670 or the clockwise 380; of course though, they suffer from Melbourne bus syndrome and all but disappear after 6PM so it’s only feasible between peaks).

How hard can it possibly be to run more Mooroolbark and Upper Ferntree Gully trains all day between the Lilydale and Belgrave services? Turn the 680 and 693 into train link services and they can be the surrogates for the inadequate frequencies caused by the single track sections. I’m sure Ventura would be happy to run them, especially with the $$$$$$$ involved (which is paid regardless of whether there are 50 passengers or 0 passengers).

As for buses, introducing more SmartBus services would generate a lot more patronage, even if the only thing done was merging two bus routes together and doubling the frequency (or in the case of hourly buses, tripling or quadrupling the frequency); it’s basically what SmartBuses are anyway once you remove the LED signs (not that Transdev even cares to run actual SmartBuses on the 900-series routes anyway). Currently we have just nine: 703, 900, 901, 902, 903, 905, 906, 907 and 908 (why don’t they call the 703 the 904 already, the proposed 904 is dead and buried). Nine semi-decent bus routes in the entirety of Melbourne (not counting the inner city buses like the 216/219/220/246 etc.)

Not giving the 703 a 900-series route could be due to the fact that it’s very often late. In afternoon peak they can run as much as 20 to 30 minutes behind schedule.

On a side note, Daniel you’re right that off-peak trains can be crowded too. Dandenong line trains at 12pm are often just as busy as they are around 9 to 10am. At least half to 2/3 of the seats are usually taken.

Even with current bottlenecks, conflict of movements, and other obstacles, would that still allow daytime trains (weekdays and weekends) running every 15 mins (an extra train set per hour) for all double-track routes? (Essentially meaning most stations closer to the city than Eltham, Dandenong, Ringwood, Sunshine etc)

The Glen Waverley line should have 20 min train in the evenings, and nothing is stopping the government from doing so.

Just returned from eight days in Sydney using trains, ferries and buses. Went from North Ryde to Kiama by train, return $2.50 (senior – 296km return) In fact $2.50 per day for all forms of public transport. Only had to wait for 5 minutes for a suburban train.

Has NSW been able justify increasing train frequencies because their economists use different values to those in Victoria? Transport for NSW recently released their updated “Economic Parameter Values”, which shows that a doubling of off-peak train frequency increases patronage by 50% in the short term, but importantly it adds that in the longer term it would increase by 100% (Table 51 on page in linked document below). So after a few years as users readjust their travel patterns and the higher frequency attracts new development, off-peak trains are just as busy as were previously at the lower frequency. This longer term change was not included in the modelling conducted by the Parliamentary Budget Office, and leads to an overestimate of the costs because the longer term increase in fare revenue has not been taken into account. I expect that users of the Dandenong and Frankston lines have experienced this, that the trains now with a 10 minute frequency look just as busy as under the previous 20 minute frequency.

The NSW report also includes train operating costs, which can be publshed in NSW as it is under publich ownership and operation, whereas in Victoria would be commercial-in-confidence because of thie private operation.

Comments are closed.