Have trains gotten faster?

Heidelberg to Flinders St train services, weekday 5 AM to 10 AM, 1939 vs 2011 (express services shaded)

Passengers on Melbourne’s beleaguered rail system may suffer overcrowding, unreliable services and even threats to personal security. But at least improvements in technology mean commuter trains are much faster today than they used to be. Aren’t they?

Actually, no. We take technological improvements for granted in almost everything, but the speed of rail in Melbourne seems to be an exception. If my local rail line is representative of the rest of the system, then travel times today aren’t significantly faster than they were three generations ago!

I compared the 1939 train timetable for the trip from Heidelberg Station to Flinders St Station on the Hurstbridge line with today’s timetable. I started with the first train to leave Heidelberg on a weekday morning in both years and finished with the last departure prior to 10 am (see exhibit).

The number of stations is the same (although back then Flinders St was called Princes Bridge) so I was astonished to see the trip is only slightly faster now than it was 72 years ago. The only substantial savings in trip time have been achieved at the cost of by-passing some stations i.e. by increasing the number of express services.

The average duration of the 17 all-stop services offered in 1939 was 25 minutes, just one minute slower than the average for the 12 all-stoppers available today. The slowest trip time in 1939 was 27 minutes and the fastest, which by-passed some stations, was 20 minutes. In 2011, the slowest time is 26 minutes and the fastest – which by-passes more stations than the 1939 expresses – is 19 minutes. So after 72 years of progress, the trip from Heidelberg to Flinders St is one minute faster!

Modern commuters are nevertheless better off than their predecessors in two key ways, both of which are essentially a consequence of suburbanisation. First, whereas in 1939 there were 19 services from Heidelberg to the city up until 10 am, today there are 27. Second, today’s commuters have a greater choice of express services than the mere two that were available to pre-war residents (the shading in the exhibit indicates express services).

And the increased number of expresses doesn’t come at the expense of by-passed stations. For example, Alphington is by-passed by some services, but still gets 23 in-bound services in the morning. Most of the gains from expresses come from by-passing Victoria Park, Collingwood, North Richmond and West Richmond stations – who still do OK because they are also served by the Epping line (not that these stations generate much patronage).

These welcome and important improvements derive from operational decisions rather than from technological improvements. I’m puzzled why, given advances in technology, modern trains on this line aren’t appreciably faster than they were in earlier years. The speed of cars, planes and communications has gone up enormously since 1939 so why haven’t trains, which have the advantage of operating in a dedicated alignment, similarly gotten dramatically quicker?

This got me thinking about whether technological advances have made urban train travel significantly better in any other respects over the period. I’d guess that labour costs are lower today and energy efficiency is higher. There’ve been some design changes like wider doors and more standing room to increase performance. But my real interest is in how technology has changed things for the better from the customer’s perspective.

Air conditioning is definitely a valuable improvement and I’ll guess that train travel is also safer today in terms of the risk of major accidents like derailments and minor ones like falls while boarding (although maybe the latter’s compromised by our more litigious society). Carriages are presumably better lit at night so passengers can spend their time more comfortably. There’s better real-time information about whether a service is running late and I don’t imagine there were in-carriage announcements about the next station back in 1939. But I don’t see evidence of the sort of major leap forward we’ve come to take for granted in other areas. After all, 72 years is a long time. If trains had progressed at the same pace as (say) planes, a commute would cost a fraction of what it was in 1939 and take only half as long.

The reason might be that urban rail was already a relatively mature technology in 1939 compared to cars, planes and communications. The exponential gains in train performance were made in the nineteenth century. The last big advance was electrification but that had already been implemented from Princes Bridge to Heidelberg in 1921. After that there were diminishing returns – the gains from affordable technologies were modest.

The overland telegraph wires and Morse code that were contemporary with early rail were superseded in short order by newer technologies like radio, optic fibre and digitisation. Our suburban rail lines however are still doing duty with much the same level of technology as they had 72 years ago. The reason Hurstbridge trains aren’t any faster today might well be that the infrastructure is still 1930s standard and is constrained to 1930s speeds.

It seems that improving urban rail in Melbourne will come down to old fashioned strategies like more investment and better management.

27 Comments on “Have trains gotten faster?”

  1. Séan says:

    We already have much faster trains available, capable of going at triple the speed of the Comengs that roll along the current tracks. The main reason they’re not used is that the stations are too close together (particularly from Heidelberg to Flinders) for such high speeds to ever be reached. This isn’t a criticism. Indeed I think some stations should be closer together to avoid gaps in coverage.

    Far better improvements in journey time reduction could be achieved by reducing the waiting time by improving the frequency of services. As for technological advances in trains, now that air conditioning is close to standard (when it doesn’t break down) on the Melbourne fleet, maybe free wifi could be a way to entice people to use public transport. Now that smartphones are becoming a common thing, it would be pretty popular.

    • T says:

      I don’t know if I agree with what you’re saying. I’ve been on the subway for example in many cities around the world where stations are much closer together and yet the trains run much faster. In Toronto, I could walk between stations downtown in 15 minutes or less. There is no way you would even consider walking between stations on any line in Melbourne.

      I suspect it’s not the distance between stations, but the current infrastructure which prevents the trains from accelerating and running faster between stations. In fact, we should have more stations with faster trains running more often. Epic fail Melbourne!

      • Séan says:

        The stations may be closer in other cities, but even so the distances between stations in inner and middle ring suburbs of Melbourne (of which the Heidelberg to Flinders route is a typical example) wouldn’t allow anything beyond a minute (more likely 30 seconds tops, in my opinion) to be shaved off a stopping all stations service. Any speed gains made by using quicker trains, improving signaling, et cetera, would easily be lost by any minor delay e.g. Passengers holding open the doors to get in to avoid waiting half a hour for the next service, wheelchair user needing to board and alight ( why couldn’t Melbourne’s platforms be built to allow unassisted boarding by wheelchair users?).

        Apart from station distance gaps, which may be less of an issue from, say, Heidelberg to Hurstbridge, the other issue is passenger comfort and safety. I struggle to remember the last time I was on a train where no one was standing. Add to that potential missiles like bicycles, prams, et cetera, and rapid acceleration and braking improvements are off the table. By and large, we have probably achieved the peak of what the laws of physics will allow in speed improvements on the Melbourne train system and so e lawf of finance suggest that limited resources are better spent elsewhere.

        • T says:

          It is true that one of the many quirks of the Toronto subway system isthe sometimes jarring stops and starts. You can always tell those who grew up in the city, because they’re the only ones who are able to stand on the subway without needing to hold on with both hands to avoid falling into someone.

          I still think there are many times when I’m on the train and feel like my 85 year old grandmother could walk faster… and it seems to happen a lot more than I’ve ever experience in other cities. That may just be my anecdotal observation, but it seems to me there is room for improvement on the speed factor. But I guess more services more often would be better than faster services overall. Ah well, nothing is perfect…

  2. poneke says:

    although back then Flinders St was called Princes Bridge

    They were two different stations.

    Princes Bridge was on the eastern side of Swanston St, under the Gas and Fuel towers which were demolished to make way for Federation Square.

    The Clifton Hill line trains (and a few others at certain times) left from Princes Bridge.

    I don’t think there is anything unusual about train times being the same now as 60-plus years ago.

    This is common on many systems.

    London Underground, Paris Metro, New York Subway — all these and more have much the same times per trip as decades ago as the stations are, like in Melbourne, closely spaced and the infrastructure is much the same now as then, though of course the trains are much newer.

    There have been some improvements in Melbourne… the last being the third track to Box Hill which was about 1970, which enabled expresses from Box Hill to Richmond. That was quite an improvement at the time. Sadly, there has been no improvement like that since, AFAIK.

    • Alan Davies says:

      People who work in transport or have a longstanding interest in the area get used to the way things are. But many lay persons would be surprised, I think, that speeds have barely increased. 72 years is very long in terms of modern “technological time”, especially for people who grew up in a world where there’ve always been ever-faster computers. It would be an interesting exercise to speculate one day on ways that train speeds might be increased.

  3. Tony says:

    You’re right about there being more trains now than in 1939 Alan, but it should also be kept in mind that the 1930s up to the start of WWII were a lean decade for Melbourne public transport – certainly compared to the 1920s and particularly the late 1940s. In part this was due to the Depression, but it also owed something to the politics of the era: a succession of Country Party governments diverted resources from the city to the bush.

    Unfortunately it proves really hard to find actual timetables from the immediate postwar era to count the actual number of trains that were running. The ones you tend to find are from the 1960s by which stage there’d already been some cuts to regular services in response to declining patronage. But even in 1960 there were 9 trains from Heidelberg arriving at Flinders Street between 8am and 9am, compared with just 5 in 1939. And despite us setting a new record for train patronage in 2011, there’s still one fewer train from Heidelberg to Flinders Street in this busiest hour of the peak than there was in 1960.

    The unfortunate conclusion is that today’s train commuter, from Heidelberg at least, is really no better off than their equivalent half a century ago, either in speed or frequency.

  4. Peter Parker says:

    Alan, if the end goal is to improve mobility, comparing in-train travel speeds is interesting but not very useful.

    Instead the ‘gold standard’ for measurement (and fair comparisons between modes) should be ‘random arrival end-to-end journey time’. In-train travel time, which comparing timetabled times measures, may only be half the trip length or less. And, critically, it’s the hardest and most expensive to reduce.

    Unless it’s already reasonably high (true for peak time but not other times at Heidelberg) random arrival end-to-end travel time is more influenced by frequency than in-train travel speeds. Even for stations as far out as 25-30km from the CBD expressing has little impact if the frequency isn’t there first.

    The superiority of increased frequency over express running in cutting end-to-end travel speed is especially marked for trips involving a bus transfer and shorter local trips. Or for people using it as a turn up and go service – ie demanding convenience and flexibility (rather than living lives around timetables).

    If one was to rank what changes would most cost-effectively cut random arrival end to end travel times, our list would be something like this:

    1. Co-ordinating bus with train frequencies, and then in the second iteration bus with train timetables
    2. Increased train frequencies
    3. More direct bus routes
    4. Bus and tram priority
    5. Increased bus frequency (made easier by 3 and 4 above)
    6. Zebra crossings and shortened traffic light cycles to cut waiting at intersections
    7. Replacing roundabouts with traffic lights for the same reasons
    8. Opening up cul de sacs to provide porous pedestrian access
    9. Reopening station pedestrian underpasses and subways (keeping the one under Spencer Street alone would have saved many thousands of person minutes each day)

    Expressing or in-train travel speed is hardly worth worrying about (unless you’re at a critical number of minutes where it could allow a frequency increase with the same amount of rolling stock).

    Individually the above steps might cut by 1 to 15 minutes each (train and bus frequencies have the highest gains), but add several and the trip is considerably shorter. Increasing train frequency from 30 to 10 minutes cuts waits by 10 min on average, so is equivalent to expressing 10 stations (but without all the compromises such as lower frequencies at the skipped stations etc).

    Once frequency is high expressing is a valid way to further reduce travel times. But if it requires extra trains and tracks to achieve there’s a whole heap of more cost-effective ways to achieve simlar mobility improvements, as outlined above.

    • Simon says:

      Wouldn’t increased density, both residential and commercial, also help? It increases the chances of your origin and your destination being a closer walk to the station.

      • brisurban says:

        It would, but how long would that take?

        More people live close to a train station in Melbourne (there are, what around 200 in Melbourne?) than Toronto (only 69 stations, interchanges counted once).
        But Toronto runs frequent buses to train stations, Melbourne by and large, does not (however SmartBus is a start on getting buses going to train stations).

        Increased density improves access, better integration improves mobility. Even with good access people will still need mobility.

    • Alan Davies says:

      Peter, I freely acknowledge that my concern in this post is with what’s “interesting” rather than what’s “useful”! Specifically, I’m intrigued by the absence of technology-related decreases in in-train time since 1939. Acceleration, speed and dwell times are possible theoretical candidates but it seems there’ve been no practical technological breakthroughs in generations that could be applied to this aspect of Melbourne’s commuter trains.

      But turning to the wider issues you raise, I agree we’ve hit the wall on in-train time and need to look at other potential sources for reducing journey times. Looking at your list, two issues occur to me.

      First, better coordination of trains with other motorised public transport modes is important but the great majority of train trips are single mode, so the scope for time savings on this score seem limited to a sub group. Improving walking speeds and kiss and ride speeds is likely to save more time.

      Second, higher train frequencies save time for random arrivals but nothing for travellers who use a timetable (other than making coordination easier – but see above). In fact it would be interesting to see what “Turn Up And Go” frequencies would be required to equate the average and maximum TUAG wait times with those of travellers who manage their trips around a timetable.

      Higher frequencies are desirable for a range of reasons but I’m not sure reducing total journey time for most travellers will be a key outcome.

      • Russ says:

        Alan, like many technologies, the limits of what is possible reaches a human biological limit first. Planes cannot pull more than 3-4G with human pilots. Suburban train acceleration is limited by human comfort; if they accelerated faster than they do then standing customers would fall over. If acceleration is limited then overall speed is a function of stop distance (stops would need to be 1-2km further apart for trains to reach top speed). Dwell times are also human limited, and fixing that is very hard.

        Consider instead, freight transportation. Transport speeds for various modes are not much faster, but the speed that goods ship is an order of magnitude better in the past two decades.

        As an aside, if you were designing urban transport systems from scratch you wouldn’t use trains, you’d probably use a maglev system of computer controlled hop-on-hop-off containers that operated like packets on a computer network. Path dependency matters a lot for technology with massive time-frames and large investments. It is much much easier to rewrite a computer system from scratch.

        • Alan Davies says:

          I think you’ve summed it up, Russ. Very insightful point about path dependency.

          It’s a mature technology where further gains are just too hard. You might’ve seen that video of an HSR train picking up and dropping off ‘carriages’ from stations – no loss of speed and no dwell time. So there are possibilities, but they’re still ridiculously expensive for HSR, much less commuter application.

      • Julian Wearne says:

        “First, better coordination of trains with other motorised public transport modes is important but the great majority of train trips are single mode, so the scope for time savings on this score seem limited to a sub group.”

        This is true in the case of Melbourne, but in cities where coordination is handled well and where networks are based much more on the connection principal, as opposed to the direct service option that is prevelant in Melbourne then the percentage of passengers that arrive by rail is far higher.

        An example is in Paul Mee’s “Transport for Suburbia” on Page 93.
        Apparently in 2,000 76% of passengers on the TTC rail network arrived at the rail station by bus. In Melbourne that figure was 10%.

        So whilst it is true that 90% of passengers that currently use rail and don’t arrive by bus will not see an improvement in travel times, the 10% that do and a lot of passengers that currently catch buses directly to the city would also see an improvement when intsead of creeping through traffic from the suburbs to the city they instead get fed directly into the rail network and go from there.

        If anyone wants to read the section of the book mentioned you can find the first few pages of the Melbourne and Toronto revisted chapter on Google books.

  5. brisurban says:

    Speed on public transport systems are a function of stop spacing and operations. If the stop spacing has not changed since 1939 then I doubt the speed would have either.

    Fewer stops spaced further between = faster services. Perth’s Mandurah line has stops spaced at around 2-3km apart, it is a system designed for motorised (bus or car) access rather than walking.

    Melbourne’s system is the opposite. Stations are spaced much more closely (1km apart or so), it is a system designed for walking access.

    You could try to have more expresses, but this lowers the capacity of the line (as you don’t want the express caught behind the all stopper in front of it). The exception here is if you use something called skip-stop operation where trains skip each other at minor stations.

    Journey time is waiting time + in vehicle time. Waiting time can be quite significant in the off peak at least- If your frequency of trains has increased then your overall journey time has gone down because you wait less for your train.

    So while in vehicle time might not have change, the frequency of service certainly has (Like Frankston line 10 minutes all day), and therefore I would suggest that overall journey times have decreased somewhat even though the trains themselves may not have sped up.

  6. Alan Davies says:

    Close stop spacing has been mentioned by a number of commenters. There’re two points I’d add on that score. First, close spacing of stations partly explains why in-carriage time is slowish but, since the number of stations hasn’t changed, it doesn’t fully explain why times for all-stoppers have stayed static. Second, stop spacing doesn’t explain why the fastest express in 2011 takes 19 minutes and stops at one two intermediate stations; the fastest express in 1939 took 20 minutes even though it stopped at seven intermediate stations.

    • brisurban says:


      I’m a bit confused by your first point. Speed is distance/time. Are there any places in the world where all stops time has significantly decreased by say 5 minutes or more?

      You could get more agile trains that accelerate faster and decelerate faster but I doubt it would do much on a system’s like Melbourne’s. Not sure about the second one, perhaps there is some fat added to take into account some kind of layover or conflict elsewhere in the system.

    • Russ says:

      Alan, a few points, have cars actually increased in speed since 1939 inside the urban area (which is the comparison being made here). On a freeway, perhaps, but therein lies the stop spacing comparison: a freeway way is an express track with stops spaced at 1-2 mile intervals.

      Regarding your question, on a single track with one-way traffic, expresses are limited by the speed of the train in front of them, so the maximum possible express saving is the time between services (which is smaller in 2011 – 6-10 minutes?) minus twice the gap between trains (1 minute?). Remembering that the Epping trains are also using the line over the express area.

      I’m not convinced there is any value in having express services on a line that busy, given the downsides: longer waits from express to standard service than if all were standard, potentially much longer waits for standard service commuters, and uneven gaps between trains from the terminus meaning uneven loading as well. Best service would run FFS-Nth Richmond-Clifton Hill and Jolimont on game days, with an improved Clifton Hill-Richmond, Hoddle St bus service to service the “gaps”.

  7. Chris says:

    Even if the speed is the same, if the number of passengers increases, then the mobility has I proved.
    I read somewhere that cars are actually slowing down in Melbourne on aveage due to congestion.
    If you speed up the trains, it gets uncomfortable for passengers.
    Major improvements to the delays at Flinders street would help.
    Sometime mes the trains still stop at princess bridge, and everyone has to walk along a very narrow platform to exit at the front of the train. These stops are never announced, so you don’t know which carriage to get in – all adding to travel time. Fully agree with peter’s comments.

  8. brisurban says:

    There is a very low-tech way to speed up trains in Melbourne!
    Cheap too!

    Delete every second station from all lines…

    NOTE: This is humour!

  9. Alan

    It’s true that aviation has made bigger leaps in the last century than rail, but at least when it comes to speed, gains have been quite modest. It takes almost as much time to fly to London now as when Qantas introduced the 747 in the 1970s.

    What has changed radically, as you note, is the cost of air travel, something I think we can put down largely to regulatory changes, specifially the gradual opening up of domestic and international markets.

    That’s not an option available to rail. Sure, privatisation has been tried (and failed, in the UK, as I understand it), but airspace is much more freely available than space for rail, and new airline entrants don’t need to lay any tracks (although they do need to buy airport ‘slots’). Seems to me there’s less room for competitition when it comes to rail.

    One thing I don’t think has been mentioned when it comes to rail passenger innovation: free wifi. I imagine it will be ubiquitous in future; if train providers cannot reduce travel time, they can at least enable passengers to use that ‘dead’ time more efectively.

  10. Daniel says:

    According to some research on the net, the maximum comfortable acceleration for standing passengers is between 1.0m/s/s and 1.8m/s/s. We’ll take a figure of 1.6m/s/s.

    I believe there are about 200 stations in Melbourne with 300km of track giving an average station spacing of 1.5km.

    Putting this figure into a calculator yields a maximum acceleration time in our 750m of about 30 seconds. 30 seconds times 1.6m/s/s is 48m/s or 172km/h. Average moving speed is half that or 86km/h. To that we need to add dwell time at stations. We still get a figure above 60km/h, for all stopping services.

    In contrast, I believe the average speed of Melbourne’s train network is about 30km/h, so the laws of physics aren’t the trouble (althrough I’ve neglected to include jerk and curves, etc, I think these would have a minor effect).

    Or to put it another way, a maximum speed system would keep accelerating until the reaching the top speed of the train or half way along the line (plus safety margin). Anyone who has ridden a Melbourne train knows this is not the case. Even express trains trundle along at 60 or 70km/h on some lines.

  11. jack horner says:

    The difference between a typical Sydney/Melbourne train and an efficient optimised-for-metro-style-operation train, assuming max speed of 20m/s (72kph) for both, is probably about 15 seconds per stop.

    The faster train saves 10 seconds on acceleration/braking (using 1m/s/s instead of about 0.7 for the slower train), and 5 seconds on quicker door opening and closing. [1] Thus about three minutes for an all stopper to Box Hill, for example.

    In a then/now comparison, general policy on how slack to make the timetable probably drowns out technological change effects.

    Daniel, I think typical acceleration is about 0.7-0.8m/s/s for Sydney trains, 1.0 for Melbourne trains, and up to 1.3 for tram/bus. It would drop off at higher speed and the rate of drop off would significantly affect the calculation you have made.

    There are many places on the eastern lines, few elsewhere, where curves would cap line speed at 70-80kph on present standards.

    [note 1]. In Sydney average minor stop dwell time (passenger movement time is not at issue) is probably 5-10 seconds longer than it was 30 years ago, on account of adopting plug doors and possibly more rigorous safety routines. Sydney trains often seem to take an eternity between doors completely shut and starting to move.

  12. […] – Have trains gotten faster? […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s