A report going to the Board of Auckland Transport on Monday about the hopeless on-time performance of Eastern Line trains reminded me of a pet annoyance of our rail system – and that is how utterly pathetic its performance statistics are. Even though we measure punctuality in an extremely generous manner, with a train that runs eight or even ten minutes late throughout its service but manages to catch up to being merely 4 minutes and 59 seconds late when reaching its destination being recording as “on time”, the headline statistics are still utterly horrific: almost 1 in 5 trains couldn’t even meet this standard in May:
To make matters worse, these numbers don’t distinguish between peak trains and off-peak trains – which clearly have very different load numbers. The chances of a train at peak times being delayed is probably greater (more passengers, tighter headways etc.) meaning that the proportion of passengers whose trips do not meet the punctuality standards might well be much higher than the table above indicates. I would really like Auckland Transport to start measuring delays in this way – weighting the trains according to their loads so we get a clearer picture about how reliable and punctual the rail network actually is for passengers.
Looking at what’s happened over time tells an equally depressing story. Despite the huge investment in improving the rail network, things aren’t getting any better.
Now clearly some of these delays are due to infrastructure improvements (for example the Eastern Line’s dip in January was because of temporary speed limits on the line after an upgrade), but if infrastructure was the main issue then I don’t think we’d see such a huge spread in punctuality across the different lines. It would seem, as a general rule, that the shorter the line is, the higher the punctuality is – with Onehunga standing out as a beacon for what the rest of the network should be doing.
The fact that punctuality seems to get a lot worse as the lines get longer suggests that the root of the problem is not about what happens to the trains between stops, but rather what happens at each stop. In other words, something called dwell management. Presumably, every timetable has a certain amount of time set aside for the period when the train is stationary and the doors are open – but (and anecdotal evidence from people who catch the train far more frequently than myself backs this up) train managers are really struggling to stick to this set time. Whether that’s because they can’t get through the crowds to the door, whether it’s because of the 17 times they seem to check whether everyone’s on or off the train, whether it’s because passengers are too slow in getting on or off or whether it’s because train managers get stuck counting out coins to give change in our 1930s-style ticketing system I’m not sure. But it seems as though that’s where the problem is.
Compare Auckland’s performance with a couple of nearby cities: Sydney and Melbourne – which have far more complicated and busy rail networks than Auckland, and often systems riddled with lots of equipment that is much older than what we have here (aside from our decrepit trains, the Auckland network has been pretty much completely rebuilt over the past decade). Melbourne, in the last couple of months has has punctuality of around 90-92%, while Sydney – which reports on peak-time punctuality (probably a tougher standard to meet due to higher loads), has always been above 90% and often about 95%: 
Sydney and Melbourne are probably not even particularly good examples of particularly punctual rail systems from around the world. I was reading the other day that Singapore’s Metro system has an on-time performance statistic of 99.7% – and that’s with a standard of a train being measured late if it’s more than a minute late! So only 3 out of 1000 trains on the Singapore Metro system are more than a minute late – now that’s the kind of standard we should be aiming for!
The Auckland Transport Board Paper is depressingly hopeless in outlining plans for improving our utterly terrible punctuality statistics. One of the key proposals, it seems, is to slow the timetable down so everyone can suffer a much longer trip in order for the statistic to look a bit better:
Already most of our train timetables are significantly slower than they were a decade ago, even though literally billions of dollars has been spent on the rail network over that time to improve the system. A trip from the old Auckland station out to Waitakere used to take as little as 48 minutes. It’s over an hour these days from Britomart (admittedly a bit further away from Waitakere but not by that much) to Waitakere station.
These are not new problems, with the rail network being so unreliable. We’ve been hearing about the problems for years, throwing a huge amount of money to fix them and make the trains faster and more reliable. Yet it seems like it’s not working, which is hugely depressing and explains the rather exasperated tone of this blog post – we’ve been promised solutions for years! If I were Auckland Transport I would put in place a requirement for 90% on time performance across all lines within three months and if Veolia can’t reach that then they lose the contract. Melbourne’s on-time rail performance has improved hugely since they gave Veolia the boot – maybe it’s time for us to do the same.
Processing...
I often find myself chuckling at the train manager system for receiving payment for or clipping fare tickets. I expect it would be difficult to set up a barrier system where people swipe or check in their tickets to gain entry for trains because our stations are so open. For example, you couldn’t do it a ellerslie because some people may be accessing the overbridge or underpass rather than seeking to use the train.
One area I wonder if it could speed things up is if the doors opened automatically, so without the need for a TM key open. Once the train has stopped the driver allows door opening. Probably still need a signal for doors closed, although it does happen automatically on metro systems (which I keep being told our system is).
I think it’s terrible the idea of slowing the schedule to improve performance statistics.
I think passengers are too slow getting on and off. When the train is slowing down for my stop, I get up and start walking to the door so I am ready as soon as it stops. Yet how many wait for the train to be stopped, then think about getting up.
trains can only go where the tracks go which means they don’t have to compete with other traffic but if there is a problem somewhere on the tracks trains cant go anywher
There are many reasons behind the slowness, but from an operational point of view, the two main reasons are:
1) Conflict points (Quay Park, Newmarket, Wiri etc).
2) Coupled routes, i.e., a train set may run to Onehunga, then west, then east, etc.
The first requires KR/AT to get serious about ditching the “all junctions must be at grade” mindset. Not much chance of that if the CRL planning is anything to go by. Cost cutting seems to be more important than providing a workable network where trains can actually keep moving. Just look at the refusal by KR to acknowledge that the single track section at Newmarket Junction is an issue. Trains stop either side of it multiple times every day, often for 3 or 4 minutes.
The second can be resolved quite easily, but again, the mindset of AT and Veolia will need to change.
Geoff, can you expand on why coupled routes are a bad thing? For me I think it’s a good idea as it prevents the possibility of a issue on one line from spiralling out of control. For example if there was a TSR on one line, having all sets restricted to running on that one line only would result in each service running further and further behind time — whereas the current arrangement means a set will run across multiple lines and thus distributing the delay across multiple lines rather than affecting one line heavily.
I fully agree however with your no-grade-separation issue. Wiri, Westfield, Newmarket and Quay Park should all be grade separated. However I think this should be done (at least in the case of Newmarket and Quay Park) as part of the CRL as any reconfiguration of these would depend on the approach chosen for CRL services. But as you said there does not seem to be any acknowledgement that junction grade-separation needs to happen to ensure reliability for CRL services.
James, what to say to that one! Amplifying delays on to as many trains as possible is a very bad idea.
Geoff is correct. There is an insane number of conflict points on AKL’s network and the Manukau branch adds yet another. Without that a third platform and some extra track at Westfield would have enabled the Eastern Line to not conflict outbound from Orakei, much like what is soon to happen at Lidcombe, NSW.
The ideal operating pattern post CRL would also include a quad from Newmarket to Penrose for the Onehunga line and a grade separation near Parnell as well as between the W/E link to the CRL, and then you would have a system which could be expected to be reliable.
What do you mean by ‘quad’? Four tracks?
Actually post CRL a great deal of pressure will come off the Parnell Line, well so long as some kind of eastern link is built at the CRL southern portal. This is what I don’t get; any money saved there by under building it will just require spending elsewhere. I think it is time to build something properly first time. To have at least this junction grade separate will take one big stress out of the system. Newmarket will still be a restriction but with the ability to route trains west [via Grafton] to the CRL will pull a heap of pressure off Quay Park and the Parnell Line.
Yes, quad=four tracks. That is the ideal infrastructure. Otherwise trains coming out of Aotea to Newmarket need to coordinate with trains coming out of Parnell. That is a bad practice, even if there were grade separation there.
I agree that some pressure will come off with CRL, but you will still be left with a sub-optimal rail system.
Cityrail have taken a number of decades to learn this lesson, but finally did with the Clearways plan of the 2000s.
Haven’t we just reduced the rail ROW through Ellerslie in order to expand the motorway even further? Making it even harder to add capacity to the Penrose to newmarket route?
I don’t think this is too much of an issue. Sydney has about seven main routing paths for ten lines across three pairs of tracks on the City Circle system.
If we design ours properly we could have just two routings for two lines across two tracks. While separate tracks for every line would be ideal, I don’t see why we could need or justify four tracks south of Newmarket when all of the CRL and Britomart would only have two. If we can manage to get two lines sharing each CRL track we can do the same through Ellerslie.
If you want to talk about Sydney’s city circle lines, they are quadding Kingsgrove-Revesby and building an enormous double flyover junction at Glenfield. This will leave conflicts at only Sydenham in the AM peak, Birrong and at Cabramatta. Misguided merges will apply at Cabramatta and Glenfield, the latter could easily be removed by logically terminating the relevant trains there. The conflict at Sydenham was planned to have been removed by a six-track to Erskenville.
The quad to Revesby will prevent the need for trains which have already diverged after leaving the CBD from having to interact with other trains again on their journey. Should tell you something.
@James – If one line has a major problem that stops trains from running, the other lines lose their source of train sets. This means everyone goes into crisis mode and urgently have to rearrange all the operations pretty much network wide. Not just the trains, but staff and meal break arrangements all have to be taken into account. The logistics of rearranging so many trains/staff with immediate effect are pretty much impossible, and I’m sure I don’t need to outline the sorts of things that end up happening from a customer point of view. Everyone reading this who regularly uses the trains will know that already.
It’s much easier to isolate the operation of each line. That means you only have to rearrange things on the one line affected.
I think slowing the timetable down is a cop out and it won’t actually deliver any results, it will be the same as loosening the belt in the hope that you will lose some weight. I predict that what will happen is that staff will relax thinking that they have more time to do things and our punctuality will be no different to what it is now. I have been on packed trains that have managed to stick to the timetable and I have also heard stories from drivers where they have to deliberately slow their trains down to stop them getting ahead.
When trains pull up the driver has to give a signal to allow the TM to open the door, this step needs to be removed so the doors open on that signal which is how I understand that it used to be. TM’s need to be more ruthless, so many times I see them just letting passengers stroll along the platforms to walk to another carriage slowing everyone down (this is on top of some of the other reasons you list).
There is though a bit of information on passenger delays, the patronage report lists to the total delay minutes which is the delay for each train multiplied by the number of passengers (although I think this is only a rough number). I have been tracking it for a little while and it gives some quite interesting results.
http://greaterakl.wpengine.com/2012/03/27/measuring-pt-performance-better/
This is the result up until Feb:
I think the timetable adjustments will help to a point — I would point to the Western Line timetable that introduced 37 minute frequencies (replacing the impossible TranzMetro-designed timetable) as an example of how a timetable can improve punctuality. However I suspect the biggest cause for the issues on the Eastern Line is due to the high usage of ADK sets on that line. These sets are very unreliable and I think it’s not a coincidence that the line suffering the most issues is the one that has heaps of ADK sets being used. I don’t think we will see any improvements until the EMUs arrive, replacing the ADK sets.
The Tranz Metro timetable wasn’t impossible, but rather it was for ADL’s. In the 90’s and early 00’s the trains were mostly on time. The timetables were slowed to take into account DART works and SA sets.
“I think slowing the timetable down is a cop out”
I agree. It’s basically saying “we can’t fix the problems that cause delays, so we will just build the delays into the timetable and let them happen”.
Unfortunately, it looks like the new EMU schedules will be slower than the old Tranz Metro DMU schedules. And it only cost $1.6 billion to achieve!
Management of a commuter rail network is the work of a professional organisation which regrettably, Veolia is not. MTR Coporation must be encouraged to bid for the AKL rail services tender when it comes up for renewal in 2014. I surmise that Veolia-TransDev are looking for an exit from NZ anyway since their parent company is shedding transport services from the corporate portfolio and NZ will be not be on the list of key markets of growth going forward.
Peter, I found your comment about peak vs off-peak trains to be interesting when I consider my travel experiences. I travel every day on the Western Line from Kingsland to Avondale and return. It’s my experience that most of the peak trains tends to run to schedule pretty well. However it’s the off-peak trains that I have found to be slightly less reliable. I do travel in the contra-peak direction so perhaps reliability in the peak direction (i.e. city-bound in the mornings and vice-versa in the afternoon) is different. Though I do keep in mind that my trains home in the City-Bound direction can only run to time if the set that operated it was able to operate on-time in the West-Bound direction prior to operating my service.
I would prefer the current reporting to remain as it is — perhaps with another metric just reporting on peak trains. However I fear this may encourage Veolia/AT to abandon improving off-peak performance and focus on improving what is potentially already pretty good peak performance. I wonder if a holistic approach would be more efficient to encourage Veolia to work on improving performance across-the-board.
I have noticed that timeliness seems to be a function of who is the train manager on that service — good ones are very good and these ones often has services running to even 1 minute ahead of schedule! Some of the Western Line services I have been on in the past few months has had me wondering if I should complain to Veolia for running too far ahead of schedule — for instance the 1702 from Avondale to Britomart often arrives 1-2minutes ahead of schedule! Other train managers are hopeless and trains under their management often run several minutes behind schedule for no obvious reason.
Yes, the current door system and procedure is definitely part of the problem. Problem is the current door system can’t be easily changed for driver opening, as the driver cannot select which side the doors open on. Might as well wait for the EMUs for that one.
What needs to stop is having Train Mangers (guards) responsible for both door operations and fare collection. Since we can’t relieve them of door operation responsibilities due to the above, it has to be fare collection. Personally I think guards should definitely not collect any fares on peak services at the very least.
HOP will help with this when it eventually comes along, but I think it needs to be coupled with a much bigger difference between cash fares and HOP fares to discourage cash use.
When HOP comes along there will be machines at stations selling individual tickets. What that means is that there will be NO on board ticket sales anymore.
Hooray!
Really, ticket sales at stations? Wow. Bring that on! I know there’s been plenty written about HOP but I wish they’d hurry on with this. Do I understand that Infratil’s involvement in our ticketing system has slowed things up?
@Jefft I think one of the notes from the latest AT monthly business report was that the ticket machines would start to be deployed at Stations from mid-June (ie now)
Wait, they’re NOT going to eliminate cash fares?
Why not?
Well you’ll still be able to pay with cash at the station ticket machine, which I believe is still crucial to have. I can’t say I’ve used every rail system in the world, but from my own experience, even the world’s best systems haven’t completely done away with options for cash.
We will be eliminating on-board cash fares which are what slows the system down. That’s good enough for me!
Maybe a mindset change is needed. Instead of “performance” which measures delays, why not measure “frequencies” and bin the timetable altogether…!?? For major bus routes, trains and ferries (aka. the spine – all rail, b-lines and devonport ferry), there should be no timetable, rather frequencies. For the minor bus routes, where it’s impracticable to provide 15 minute frequencies or less, then sure, timetabling is the way to go.
Of course, the operators will still need to have their own timetables of departures from the depot etc, but sticking to a fixed timetable for frequent services is just unreasonable. Look at the Outer and Inner Link buses… Why do they have to wait to a scheduled departure? why not flexi it up a bit and only consider the amount of time the previous bus has left the bus stop – providing effectively a regular frequency for passengers, and not timetable. Same would apply to trains… am I off my rocker, or is it too complicated in this time and age to implement successfully?
In that senario, you would need to run rail services every 15mins or more, on everyline from the first service till the last, and every day of the week. I think that would be impractical in Auckland for the time being.
If we had a driver-less metro, that wouldn’t be a problem. Vancouver seem to manage this just fine. But because we’re stuck with heavy rail for the time being, then yes, we have a small problem. We could instead introduce “high frequency” periods eg, 6.30am-9.30am, 3.30pm-6.30pm, when there is no timetables, but frequencies.
Well that’s pie in the sky kinda stuff really, and would be many years away if it were to happen. I think we need to be realistic and look at steps we can’t take in the intermediate term.
The plan when we have all the new trains as I understand it is for 6 tph, or 10 min frequencies on all lines [except the single tracked Onehunga], which is certainly much much more like a turn-up-and-go metro system than the current set-up.
Is 6 tph possible with the Manukau Junction as it is? 14 tph through Newmarket and 20 at Britomart
That would only be at peak times tho, I’m pretty sure it won’t be all day everyday
Excuses. There are always more.
You can’t compare the Auckland system with a true Metro, they have many more doors for loading, much higher frequency and fixed dwell times.
In the last few years a significant number of changes have been made that slow the performance here.
Requirement that TMs not give the gong until after their door is closed (net loss 1 minute per 6 stations).
New braking procedures for SA sets (net loss? not sure but definitely has an effect)
Huge increase in patronage does have an effect on loading/dwell times regardless of TM efficiency. Again 10 seconds per station is 1 minute lost per 6
Between the loading and operational issues, the Western line schedule change is basically explained.
Also of course, the old schedule was basically never met due to the single track issue.
When the new EMUs arrive (which hopefully will be driver-only operation), and the full roll out of HOP, that’s when you “should” see significant change for the better. Even if they still have a Guard on the trains, they wont be collect fares and they wouldnt be in the passenger carriages so should make everything significantly faster. Speaking of which, does anyone actually know if they’re going to keep TMs when the new EMUs come into service?
I’m pretty sure the business case for electrification relied on some pretty massive operating cost savings so I’m guessing not.
The one thing that could kill it, is if the ‘culture’ doesnt change when HOP and EMUs are fully implemented. What I mean by that is, the drivers still drive the EMUs like they drive the current fleet and dont push them to their full capabilities and also that the TMs still take ages to dwell at stations and are slow at opening and closing doors. I wonder if when the EMUs come they’ll keep the timetable to the same running times they’re at now, so then their on time running performance will lift and be high, essentially what happens on the outer link now and have intermediate stops for several minutes to catch up to timetable if they get too far ahead. God I hope I’m wrong on that last suggestion.
As Peter says the plan had been not to keep them but I understand the thinking has now changed and there is talk of leaving at least one staff member on the train for ‘Customer Service’
My day job is spent tackling these sorts of issues – and getting to the bottom of what is happening is the heart of it. The reliability (=service delivery) numbers are on a par with what I expect, but the actual timekeeping to five minutes (we use the same benchmark here in Scotland) is depressing.
To help fix it, we have management information systems which tell us:
* When the delay happens (some trains can be particular poor performers)
* Where the delay happens (often needed for identifying network or timetable issues)
* Why the delay happens – is it train reliability issues, high numbers of passengers, or those hardy annuals like signalling?
Other factors would include where trains have been delayed because of a late-running freight service.
I suspect that Veolia do have this information, as it is not difficult to gather. It would help if more of it could be released, though operating companies get rather twitchy at the prospect of this.
Classic example this morning, my western line train turned up at Newmarket roughly on time, as we were waiting for the driver to change ends a southern line service turned up late and was given priority ahead of us so eventually took off down the hill. We then had to wait for a service heading to Swanson to cross the junction to head to platform 1 before we could leave. At the bottom of the hill we again had to wait on the curve behind Vector when an eastern line service turned up and again got priority over us delaying us further. As a result we ended up being about 10 minutes late into Britomart and it took us 18 minutes to do that final leg of the journey.
Looks like a strong argument for grade separation of the junction at the southern portal of the CRL to me…
The Sydney Morning Herald this morning analyses [Sydney] City Rail train punctuality statistics: http://www.smh.com.au/nsw/revealed-busiest-trains-hit-by-delays-20120617-20ick.html. As I read it, the analysis suggests that ATs preference of ‘slowing down’ the timetable is unlikely to succeed in fact the only thing that this measure is guaranteed to do is to further piss off passengers. Pity we don’t get this sort of investigative reporting on issues that matter from our esteemed (ahem) local broadsheet.
The Herald is on the case this morning I see: http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=10813976
Here’s a good example of a holdup – this section of this video details the knock-on effects of one train having its doors blocked from closing by 15 seconds:
BBC “The Tube” series 1 episode 5, time 21:57.
Why can’t NZ rails learn from Countries that have solved their rail problems successfully? There are many of them. Is it money problem, attitude problem or what?