Categories
Toxic Custard newsletter transport

Detailed Metro train stats revealed

This morning The Age published more detailed train service data than we usually get to see. Some information is routinely published, but we rarely get an insight into the breakdown between AM, PM and off-peak punctuality, for instance.

In some ways the data was no great surprise — in the first week of March, hundreds of services were altered, including 71 Loop bypasses (City and Altona), and 399 shortened services, with 95% of them at peak hour in the peak direction. This matches the anecdotal evidence often heard from daily users.

Also not surprising is that peak services are less punctual than off-peak. As Jarrett Walker long ago wrote in his Human Transit blog, peak is when the system is at its most stressed — from numbers of trains and passengers on the network, causing congestion and longer dwell times at stations, with any delays snowballing much faster.

The Age: Metro disruptions

Some lines are clearly much worse than others — these figures have more detail than we usually see, but it’s reflected in the aggregate figures published in the Track Record monthly reports. The worst lines tend to be those with single track sections (which quickly causes delays to escalate) and those in growth corridors (more trains on the line, and more people getting on and off them)… with some unfortunate lines such as Cranbourne having both those attributes.

Lines directly or indirectly linked to those less punctual lines, such as sharing Loop tunnels, tend to get affected too.

What perhaps is surprising is that until March, the reliability and punctuality data wasn’t automatically captured. It was gathered by Metro themselves, and a sample was cross-checked by PTV. (In contrast, the tram network has had automated monitoring for decades — the data from it is used to feed into Tram Tracker. Buses are mostly monitored manually, with only a tiny sample ever being reported on — a small enough number to make it meaningless, though steps are underway to automate it.)

Crowded train home

The sheer number of Loop diversions — about 10 per day, most likely concentrated at peak periods — is also surprising. This can cause a lot of disruption for people, and has flow-on effects to other services as people change trains. That’s if they’re told on time — I’ve been aboard a service that was diverted to bypass the Loop after leaving Richmond, giving no chance for people to change. Many were not happy.

The reasons for specific alterations weren’t included in the data, but we know this is gathered, as Metro get exclusions from performance penalties for problems they have no control over — which is fair enough.

Given we all pay for public transport services (as both passengers and taxpayers), is it not reasonable that this type of detailed information is published regularly? That would provide better visibility of delays and alterations, where and why they occur, and would cast light on specific parts of the network, what the problems are, and how they can be fixed — so voters can hold the operators, authorities and politicians to account.

Fixing the problems

It’s also important that the state government make sure Metro is only altering services for good reasons — such as a counter-peak service altered so a peak service can run on time, rather than just to help the punctuality statistics.

Metro may need to be pulled into line in the short term. How? Well former Labor transport minister Martin Pakula, while in opposition, seemed to think it was perfectly possible:

FORMER Labor transport minister Martin Pakula today called on the state government to force Metro to stop its practice of skipping stations to improve punctuality.


Mr Pakula says the situation could be easily resolved by Transport Minister Terry Mulder.

“There is the franchise agreement (between Metro and the state government) and there is common sense,” Mr Pakula says.
This can be resolved by the Transport Minister getting onto Metro and telling them it is not on.”

Herald Sun, 19/4/2012

Metro should be willing to listen, given you’d imagine they’re seeking an extension to their current contract, which expires soon.

Longer term? Line-by-line targeted investments can make the system more reliable, starting with those single track sections. And the new contracts (due during this term of government) need to be made more watertight against strategies like station skipping, to ensure the service is run in the interests of passengers.

Does the frequent part of the network need timetables?

A change in emphasis should also be considered. As the system transitions to a more “metro”-like network, with segregated lines running frequently, it’s arguable that specific train times matter less than keeping the service running frequently. For instance, if a 5 minute service is in place, it doesn’t matter if the 8:00 train arrives right on 8:00 — instead the contract might be structured so penalties apply for gaps between trains of more than 5.5 or 6 minutes.

The current regime has undesirable impacts right now. For instance, South Yarra sees dozens of trains every peak hour to the City, but some have to wait there for the timetable to catch up to them. This doesn’t make sense. If the 8:51 arrives early, and there’s a slot for it ahead to get into the City, and there’s another train right behind it, let it leave early.

Equally, if trains are running every 10 minutes down the line, and one gets cancelled, a big 20 minute gap eventuates. To even out the loads better, if it doesn’t cause any other problems, it might be better to hold the train before it and run it 5 minutes later, creating two 15 minute gaps instead.

If trains are frequent enough, people don’t bother with timetables. Eventually, if the network and the contracts are structured the right way, the operator could work to provide a frequent reliable service, where you know you’ll get to where you’re going quickly, rather than trying to meet specific train times which don’t matter anyway.

An issue to think about for the future.

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.

25 replies on “Detailed Metro train stats revealed”

“such as a counter-peak service altered so a peak service can run on time, rather than just to help the punctuality statistics”

If that counter-peak service is going to form a peak service, it’s almost a given that altering it to arrive on time is going to help the peak service run on time. If it’s forming an off-peak service that, for example, runs on a single-track section, and which would delay yet another train if it’s late, isn’t that also a good reason? I would suggest that altering a service to run on time would be for a “good reason” at least 95% of the time. Is it worth the hassle of checking if there are good reasons for the other five per cent??

“Well former Labor transport minister Martin Pakula, while in opposition, seemed to think it was perfectly possible:”

So you’re suggesting that a politician knows enough about detailed railway operations to know how possible it was? I can recall a (transport?) minister during the redevelopment of Spencer Street station when some platforms were closed saying that a train driver wrongly routed to a closed platform should have gone to the right platform instead, apparently blissfully ignorant that it’s not the driver that sets the route.

“For instance, if a 5 minute service is in place, it doesnโ€™t matter if the 8:00 train arrives right on 8:00…”

Sure, okay. You are due to arrive for an appointment at 8:15, the trains are running every five minutes, but they are taking 15 minutes longer in their journey time, so you arrive late for your appointment. No big deal, I guess.
Your argument assumes that passengers are joining a late-running train and consequently arriving late; you overlook them joining one that is on time and then becomes late.

“If the 8:51 [at South Yarra] arrives early, and thereโ€™s a slot for it ahead to get into the City, and thereโ€™s another train right behind it, let it leave early.”

Okay, and the 8:51, unlike all trains for the next 20 minutes, is the one that you are catching to go through to Laverton. I don’t think the fact that there are other trains to other destinations soon after would be much comfort.

“To even out the loads better, if it doesnโ€™t cause any other problems, it might be better to hold the train before it and run it 5 minutes later, creating two 15 minute gaps instead.”

Yes, perhaps, but given that running it late might cause it to miss bus connections, be late for a single-line cross, conflict with another train at a junction, or not have enough time at the terminus before the return journey, there are a lot of things that need to be considered before you can be sure it won’t cause other problems.

“If trains are frequent enough, people donโ€™t bother with timetables.”

Okay, so I have a train leaving my local station for the city every five minutes, and I need to be in the city for an appointment at 11 a.m. Without any further information (i.e. without a timetable), what time do I need to be at my local station to catch a train? You’re correct up to a point, but you’re overstating it.

I comment from a position as a casual off peak user. You didn’t focus on trains missing stations, or turnbacks before the ultimate destinations. Not that much time can be regained by skipping stations and I really don’t think this should happen. Terminating trains before the ultimate destinations can make the service better for the rest of the day for far more people than those inconvenienced by one train not going all the way. It rather depends on if you the person who did not get to their destination or the many who’s train was on time. There could be a train sitting at the end of the line with a driver ready to leave at the delayed train departure time, but are we prepared to pay for this?

Delayed trains are the problem and why they are delayed should be the focus. Peak hour service on the Upfield line is a twenty minute service. That a train is cancelled or skips stations resulting in a forty minute service is so wrong. Even I won’t wear the argument of getting trains back on time in that situation.

The most annoying “loop bypasses” are those where a train which is supposed to go through the Loop runs direct from Flinders St to Richmond, inconveniencing passengers waiting at Loop stations. Yes, generally there’s an announcement (and sometimes it’s even clear enough to understand), but frequently this is at the last minute and by then it’s too late to catch another train to get to Richmond on time. Come to think of it, why don’t the displays inform us of cancellations?
However, without wishing to be thought of as an apologist for Metro, just sometimes it might make sense for a train to skip stations, if it’s so late that the following train is only a couple of minutes behind.

On the importance of timetables, it should be noted that train travel times would still be provided, so you’d be able to make a judgement on how early you would need to catch a train. Frequencies instead of times have also begun to be listed for some tram stops, but I am not convinced that timetable information should not be provided for 10 minute services. For more frequent services, sure, but 10 minutes is a bit long for there to be running according to headways only. In the case of trams, however, the provision of real-time information means that this isn’t a big deal, but for train and bus services, it would be beneficial to still use timetables. Trams also suffer from routine congestion and delays, so in that case, a headway based system where trams are dispatched according to headways may be beneficial; but I still reckon that 10 minutes is too long.

If we had a metro style 10 minute service as you suggest the system would be vastly improved. It is a poor service, tonight I was going to catch the train 2 stations to Camberwell to do shopping but it was a 15 minute wait for the train when I arrived so I decided to walk. During peak frequency at Glenferrie is not an issue but in the evening it is poor.

@John:

“If itโ€™s forming an off-peak service that, for example, runs on a single-track section, and which would delay yet another train if itโ€™s late, isnโ€™t that also a good reason?”

Perhaps, though the nature of peak is that there should be plenty of trains arriving that can form off-peak services, as many are just going back to stabling. We know they transpose trains at Flinders Street for this type of reason.

I take your point about some ministers not knowing everything!

“given that running it late might cause it to miss bus connections, be late for a single-line cross, conflict with another train at a junction”

Well, I did qualify with “if it doesn’t cause any other problems”!

I guess ultimately I’m saying look at metro operations overseas: for cities like London, NYC, Vancouver even, there is no public timetable. You know roughly the travel time, and that there’s a train every few minutes. Connections aren’t an issue if the connecting bus is also frequent, though you might arrange to hold some services (eg last of the day, or late-night when they’re infrequent) for a specific train.

But hey, this is precisely the sort of discussion around these issues that I’m talking about.

@Campbell, I agree. 10 minutes isn’t frequent enough to remove timetables. I’d say 5 is okay, though slightly higher (less frequent) might also work for people.

I can only recall one shortened service on the Sandy line in the past 10 years. Not a bad record given I use the service in peak times.
I recall many loop bypass trains when Sandringham trains went around the loop on weekdays.
It would be good if trains ran every ten minutes on my line at weekends, rather than every 20 mins.

there are a small number of morning loop trains on the Frankston line and they get altered.

I’d like to know if it’s worth getting on a loop train at all or just grab any train and change at Richmond regardless.

I dislike changing at Richmond for a loop train but it’s annoying to stop all stations from Seaford then have it bypass the loop when I could have got an express train and change at Richmond anyway.

The lack of shortened train services on the Sandringham line is not all that surprising with no single line sections, growth areas or junctions to create delays, a short length and only two places to cut services short at (Elsternwick and Brighton Beach). And people from Middle Brighton in would be unlikely to find out about inbound cut shorts from Brighton Beach and may not find out about outbound ones either.

I gave up worrying about time-tables long ago; they are pretty much useless so an approximate time is good enough. What I would really like is information at the station from a person, or a billboard, or a pa system (anything so long as it is clear; I don’t have a smart phone so I can’t access whatever information is posted on metlink’s site or any other). For example, this morning I got on the 8.56 (the scheduled time, I wasn’t paying attention to the time it actually arrived) from Hughesdale to South Yarra, only for the train to go about 100 m and stop. It took about 50 minutes to get to South Yarra. (and I still had a 20 minute walk to get to work). Fine; the driver announced there were severe delays (no reason), and I read several chapters of my book. I later read the delay was due to a passenger accident at Malvern at 8 am. These things happen unfortunately and sadly. But surely the information could be given to other commuters before they board a train so they can make an appropriate decision – I would have worked at home instead. It’s not much use telling me there’s a severe delay once I am on the train and stuck between stations.

@ John

“Okay, so I have a train leaving my local station for the city every five minutes, and I need to be in the city for an appointment at 11 a.m. Without any further information (i.e. without a timetable), what time do I need to be at my local station to catch a train? Youโ€™re correct up to a point, but youโ€™re overstating it.”

I think it is quite easy to work out the time you need to get to a train you just add five minutes to the normal travel time to the city. You can also be just as likely delayed on a scheduled train and an unscheduled train can make the best amount of time as they are not restricted by timetables. This is common on many subways around the world and can happen here as well provided we catch up on maintenance funding, modern signalling and infrastructure capacity.

Daniel: “Well, I did qualify with โ€œif it doesnโ€™t cause any other problemsโ€!”

And I did agree that if the qualification is met, you’re right! But I questioned whether one could be sure of it being met often enough to be useful.

“I guess ultimately Iโ€™m saying look at metro operations overseas: for cities like London, NYC, Vancouver even, there is no public timetable.”

When I visited Paris and rode the Metro I consulted the timetables they had posted.

“But hey, this is precisely the sort of discussion around these issues that Iโ€™m talking about.”

And I’m happy to oblige! :-)

Llib: “I think it is quite easy to work out the time you need to get to a train you just add five minutes to the normal travel time to the city.”

In my example scenario, all you knew was that train ran every five minutes and you had to be at your destination by a particular time. I said that without a timetable, you had no other information. So how do you know what the “normal travel time” is?

But I’ve seen timetables that show full details up until a given time, then say “then every x minutes until”, then give more full schedules. Assuming x is, say, five or ten minutes, you have enough information, so you don’t need a full timetable. But you still need some sort of timetable.

In some cases, you could also do what the VR used to do on the St. Kilda and Port Melbourne lines: List departure times from each end and how many minutes journey time to each station. This works for short lines with consistent times, but becomes less practicable for longer lines and where different journeys are quicker or slower than others.

> There could be a train sitting at the end of the line with a driver ready to leave at the delayed train departure time, but are we prepared to pay for this?

The M&MTB used to regularly use “block cars” at busy times, a tram sitting at the terminus that left at the scheduled departure time. When it crossed with the late tram, the drive and conductors swapped trams and continued on their way, the late tram becoming the next “block car”.

Andrew and mgm,

It’s impracticable to do with the trains what you (mgm) describe the M&MTB doing (and I’m not suggesting that you were suggesting that for the trains), but the railways have long had several trains “on standby” to cover cancellations or late running. The problem is that it’s not practicable to have them at the ends of all the lines (too many, plus lack of suitable place to sit them, in some cases). Instead, they are either close in to the city or at selected non-city locations.

But they don’t solve all problems, they have limited flexibility, and any shortage of drivers means that they are about the first shifts to be cancelled (which at least means that you’ve saved cancelling a scheduled service).

@roger & @tom. You can’t easily terminate short on the Sandringham line. Elsternwick is a locked ground lever – requiring special staff to attend, obtain the keys, and hand signal trains. The delays in terminating a train would far outweigh any savings. Brighton Beach has a signal box, but it is normally switched out and would require the station staff to attend.

Other lines have permanently staffed signal boxes – Caulfield, Westall, and Dandenong, for example, making short arrivals and departures trivial to implement.

As someone who catches trains in the Werribee line every day, the figure of 69% for AM peak is all too realistic. Combined with the fact that it is the largest growth area in Melbourne and the trains are almost all completely crush loaded and you’ve got a service that is clearly inadequate.

Can I seek compensation? 69% is well below their target

@ John
Having a timetable gives a poor sense of security on assuming a reliable journey on Melbourne’s trains as all the signalling failures, congestion and other issues has heavily affected timetables and made them highly unreliable as is the case with Werribee in the AM peak. What’s the point of the timetable if more than 30% of services cannot get to within 5 minutes of the timetable. In my eyes this makes the timetable more like a rough guide and not anything you can rely on.

The major problem with Melbourne is poor and outdated infrastructure. To successfully implement a service with no timetables in busy services you need to upgrade the infrastructure (extra capacity, signalling, electrical, tracks, bridges, level crossing removals etc) and invest in enough rolling stock to get the services up to 5 minute headways in core areas. At least there is an effort to work on level crossing removals and some backlogged maintenance from the previous government but the key now is to start work on MM1 and then MM2, upgrade signals to in cab signalling as well as beefing up rolling stock orders.

Llib: I notice that you avoided answering my question about how you know how long the journey is, by changing the subject from the usefulness of the timetables to the reliability of them.

@John
I haven’t avoided the question and the goal here is to improve the performance of the system to compete with the car. Speaking of the car which do not run on timetables they seem to have plenty of idea of journey times without having to look at a timetable, instead they use guestimation of distance and time of day to work out travel times and of course include possible variables such as congestion etc.

Getting back to PT services with no timetables you could easily work out travel times from the relative distance of the destination combined with the average speed of the service. The PTV could also gather those statistics and publish them to help travellers work out an approximate time they need to be at the station.

In regards to those trams that don’t have a timetable, I have never heard of any complaints about the fact that they don’t have a strict timetable.

I believe they should be introduced on trains and frequent bus services on a trial basis and then statistics should be gathered on whether patronage improves and if there are significant problems with operations. Until it is trialled there is no way that anyone can say it won’t work without evidence of it failing in action.

Llib, you HAD avoided the question, as is obvious from you trying to answer it now rather than point out where you had answered it before.

“The PTV could also gather those statistics and publish them to help travellers work out an approximate time they need to be at the station.”
Or, to make things even easier, they could provide a timetable!

“In regards to those trams that donโ€™t have a timetable, I have never heard of any complaints about the fact that they donโ€™t have a strict timetable.”
Trams have timetables. See http://tt.ptv.vic.gov.au/tt/TTB/20150326-135125/vic/03006F_ttb_TP.pdf for instance They are not posted on every stop, but they exist, so people who want to can consult them.

“Speaking of the car … they seem to have plenty of idea of journey times without having to look at a timetable … include possible variables such as congestion etc”
Some idea, not “plenty” of idea. With a car, you (a) allow plenty extra, (b) know from experience, (c) get it wrong and arrive late, or, these days, (d) use a journey planner to calculate it. Telling people they have to allow plenty of extra time because you couldn’t be bothered publishing a timetable is bad form and bad PR. You can’t know from experience unless you /have/ experience. I’ve numerous times been late when driving because I didn’t allow enough for congestion (hence I did not have “plenty of idea”), and of course (d) is the equivalent of having a timetable.

“I believe they should be introduced on trains and frequent bus services on a trial basis…”
Sorry, /what/ should be introduced? Timetables? They already have them. More-frequent services? That’s an entirely separate question.

The key goal of non timetabled services is to have turn up and go services that make it much more effective at attracting riders than infrequent, slow and unreliable timetabled services. This will also allow more efficient transfers as the wait between transferring to different services is minimised.
It is also more efficient for the operator because all he has to do is provide enough rolling stock and drivers to keep the services at a high frequency, the operator of the service can also catch up time more easily if there is a cancellation or other problem because the service is not held back by a timetable. The speed of the service is also much better on average because if the service is running ahead of time it does not have to slow down to keep with the timetable.

This is how metros and busy trams and buses work around the world as consulting timetables is not really realistic for most passengers in busy metro areas.

In my opinion superior frequency, travel times, convenient transfers and better utilisation of train paths and PT infrastructure is much more important and will attract more patronage than sticking to timetables that usually aren’t adhered to anyway.

In the western suburbs we can only dream of a turn up and go service. Ours is a turn up and wait possibly up to 44 minutes if you just miss a train and the next is cancelled. A friend moved to the eastern suburbs and ran for a train but noticed no one else did. They simply caught the next one in a few minutes. Is the west of the city less deserving of a decent public transport system? It seems that way!

Okay, Llib, perhaps we are making progress. You are drawing a distinction between “turn up and go services” and “infrequent, slow and unreliable timetabled services”. But what about frequent, fast, and reliable timetabled services? Since when does a timetabled service have to be infrequent? I never suggested slow timetable services. Rather, my point was that increasing frequencies to the point where people can simply turn up and go doesn’t completely eliminate the need for timetables. You are arguing against something I never suggested.

“It is also more efficient for the operator because all he has to do is provide enough rolling stock and drivers to keep the services at a high frequency, the operator of the service can also catch up time more easily if there is a cancellation or other problem because the service is not held back by a timetable.”

Whether timetables are published to the public or not, the operator will still have timetables that it runs the trains to, and rosters for the drivers driving them. And timetables rarely hold services back. The normal problem is keeping up to the timetable, not keeping down to it.

misguidedjenni,

The “east” tends to have more frequent services because the geographic centre of Melbourne has long been to the “east” (i.e. east of the CBD and the centre of the rail network), and therefore the eastern lines are longer, and therefore busier, and therefore require more frequent services.

But this is changing. The “west” has had more line extensions than the east, and greater increases in frequencies than the “east”. The last two extensions of the suburban rail system in the east were to Pakenham in 1975 and Cranbourne in 1995. Since then, the “west” (and I include north here) have had extensions to Werribee (1983), Laverton (from Altona) (1985), Sydenham/Watergardens (2002), Craigieburn (2007), South Morang (2011), and Sunbury (2012). Following on from these (and other factors) have been increases in patronage and frequencies. They don’t /always/ match the best of the eastern lines’ frequencies, but they are getting there.

Hi, I am from Hong Kong. I am just wondering how Melbourne Metro doing since 2009? Is the performance getting better or worse compare to Connex? What is the customer satisfaction level? And do the public wants to renew the contract with Metro?

Thanks

Comments are closed.