Since my in-depth look at the Sydney Metro Northwest covering the launch-day observations and experiments received a lot of traction online, I feel like it’s probably a good idea to write a follow-up article now that the first week has passed, covering some of the things I may have missed in the first article and the developments or issues that have come to light since launch day.
I will reiterate, however, that I do not work for and have no affiliation with the Sydney Metro Northwest teams. I have not worked in the rail industry, instead, am just a curious engineer that is a public transport enthusiast. I’ll attempt to detail and explain things from my perspective – one that includes visiting the public transport networks of Singapore, Hong Kong, Shanghai, Seoul, Busan, Daejeon, Daegu, Gwangju, Tokyo, Kyoto and Osaka. Hopefully I’ll come around to detailing my adventures on those other networks when time permits.
First Week Sentiments and Issues
On the launch date, it seems that sentiment surrounding the Sydney Metro Northwest was somewhat unfairly negative. It’s as if people wanted the system to fail, criticising every little fault or flaw by claiming it should have been fixed by opening time. I’ve seen complaints that the doors are dangerous, the trains themselves are old, noisy and dangerous and more. This was made worse when there were issues with doors causing trains to be held up – this has been reported as “hopeless“, “disaster“, “chaos” and “a debacle”. In reality, the coverage couldn’t be further from the truth as the system rapidly recovered (as designed) after the disruption was taken care of and it was the first day of operations. Every engineer that has deployed a system would know that testing only goes so far – as soon as things go into production, you will encounter issues which may not be foreseen purely because the conditions have changed. I suspect many were eager for the metro to fail because Gladys Berejiklian was behind the project. To judge the performance from a “free ride day” is really not representative of regular service loads.
By the time the first weekday come about, sentiment became more positive as commuters began to understand just how valuable the metro would be to reducing their commute times to and from the city. It was reported by the Sydney Morning Herald that the metro had passed its first real test, with one commuter (Paul Nijar) claiming that “it’s up there with Japan now.” and it seems that some are even looking forward to Stage 2. While the issues with overshooting did occur occasionally, it is not unexpected. Even the Twitter messages looked more positive overall.
However, unfortunately, the metro made headlines several times in the week with numerous door malfunctions, failure to open doors at Chatswood, fire-alarm and power related issues and a train stalled due to communications issue at Cherrybrook. It would seem that the press were again exaggerating what were essentially teething issues.
The issue at Chatswood was the incoming train didn’t open its doors for commuters to disembark, instead going through the turnaround process taking about five minutes to arrive at the adjacent platform. While this is an inconvenience, as passengers have to push through boarding commuters and interchange to the other platform via the underground area, this is hardly the rogue train situation the headline seems to imply. Instead, the choice of the word rogue almost unfairly conjures up an image of a runaway train, of which this was not. Even on opening day, there was a train that pulled into Tallawong with the doors shut for minutes, only opening once the other platform was cleared. It seems that the sequencing may be buggy with services that are terminating, not allowing enough time for the train to be cleared … at no point are the passengers in any danger, merely just a little inconvenience. But of course, people might be anxious being unfamiliar with driverless train technology and perhaps eager to make a tight interconnection. Staff communication could probably have been better.
Fire alarm issues at Macquarie Park station may well have been initiated by an actual event (e.g. someone smoking near a detector) or a faulty detector – whatever the cause may be, evacuating the station and having trains pass the station without stopping is the “safe” option. This in itself is hardly a noteworthy event – this happens with some regularity on some overseas systems (e.g. London Tube). As for the blackout, who know as to the cause – but since trains and station infrastructure relies on electricity, it would make sense that a loss of power would mean a loss of service. This would be the same for regular trains – again, making this hardly noteworthy.
But the biggest take-away from that article was this part:
The rail workers’ union said the system was proving to be the debacle their members had predicted.
“(It) has failed to prove it is capable of doing the two basic tasks required of a driverless train — driving and stopping,” Rail, Tram and Bus Union NSW secretary Alex Claassens said in a statement.
“We’ve seen commuters forced onto replacement buses, people left stranded in between stations with no information, delays and serious automatic door problems.
“We’re lucky we haven’t seen any major safety incidents as yet.”
I think that this shows exactly the angle the writer was going for – of course, train drivers and the rail union in general would be against driverless technology which may be (correctly) perceived to be a threat to their jobs and hence would do anything to discredit it. But the truth is that driverless technology has been proving itself to be capable of being safely and efficiently implemented overseas – so even though it may stop and drive correctly the first time 99.9% of the time, they seem to be focusing on one-off problems.
You don’t see my calling for drivers to be fired whenever they do an over-run of the platform, skipping the stop, overspeeding or passing a signal at danger, all of which happen (although, perhaps less frequently than in the past). The only reason it doesn’t happen more often is because each driver has a lot of money and time invested in training – but if you fix a driverless system once, every carriage will benefit from it. The requirements on screen-door-less platforms is much less stringent as well – so it’s hardly apples-to-apples.
Commuters are claimed to be “forced” onto replacement buses – but keep in mind that the failures are not all the fault of the metro. The same thing happens when signal faults, train breakdowns, tree-fouling, fatalities and other issues happen on the regular trains – some of which are much less likely on a driverless metro. I don’t think it’s “lucky” that we haven’t seen any major safety incidents – the system is designed to prevent them as much as possible.
Likewise, the disruption at Cherrybrook shows exactly how the system is designed to fail-safe. This is how any safety-of-life critical system should be designed – should there be any failure or doubt, the system should take the safest course of action. In this case, the system uses CBTC to enforce train separation, thus continuous communication is necessary to ensure the train maintains its safety distance (amongst other functions such as aligning with the doors, controlling its speed and running announcements). In the case of a loss of this link, it is designed so that the train and the network comes to a halt to prevent any collisions. The system failed safe as designed – causing some inconvenience, but being safe.
I think it would be good to draw an analog with the elevator. Initially, early elevators were “driven” by an elevator attendant using a controller handle. The cab would be “stopped” at each landing by the attendant manually, often overshooting or undershooting the intended position depending on the load in the car, sometimes requiring “jogging” the car up or down to get it closer to level and reminding passengers to “step up” or “step down”. These workers were put out of a job with the invention of electronic elevator controllers – simple ones featuring relay logic and one or two speed control, resulting in some “jarring” starts and stops or imprecise levelling. It took some time for people to be comfortable with the idea of “electronic” control – but now, we don’t even think about it now that we get into elevators with variable speed drives with vectored control that smoothly accelerate and decelerate with nearly perfect levelling every time.
A driverless train is a closed system quite similar to an elevator, just distributed across a wider area with a larger load of passengers and more “cabs” to manage. Unlike driverless cars which have to deal with more variables, the concept of driverless trains is more easily realised in a safe manner and its continued deployment almost seems inevitable.
On the Train Again
As I don’t work in the city, I don’t have much of a reason to catch the metro for the most part. But I was given a reason to board the metro again because of visiting Vivid Sydney Chatswood precinct – so it was back on the customised Alstom Metropolis (Made in India) carriages.
This time around, instead of taking about 42 minutes, it seems that the trains were running somewhat quicker with travel times about 37 minutes. However, it seems this speed-up is not without its downsides. While GPS tracking is difficult to maintain within the carriage, I was able to get a better GPS trace which shows the speed of the train running about 65km/h from Tallawong to Rouse Hill and then 100km/h from Rouse Hill onward, in agreement with the claimed 110km/h operational speed.
One thing I did notice was that the shorter trip time seems to have come at the expense of stop dwell time. This time, I was seeing a few people getting caught in the doors, which may not have sensors (they are a reliability issue) and instead seem to close a fixed time after announcing “Please stand back from the closing doors”. This is why it’s important for metro users to be aware of the PSD operation and not to dawdle along and try to exit or enter at the last moment. Instead, they should be more like the commuters in Asia who are too eager to join or leave the train and have positioned themselves accordingly to make the most of the limited dwell time.
In order to investigate this, I decided to switch on audio recording on my Sharp Aquos S3 phone to record my journey in both directions. To determine dwell time, I listened for the “whoomph” sound that comes about as the doors start to open or when the two doors meet and begin to lock. My measurement of dwell time is “whoomph” to “whoomph”, with actual boarding/disembarking time perhaps two to three seconds shorter due to time taken for door movement.
The measured dwell times seems to differ significantly on some stations, perhaps by programming, whereas others were perhaps due to a lack of clearance ahead of the train due to a hold-up. On the whole, the average dwell time is about 23 seconds, but many of the stations seem to hold to roughly 20 seconds. This is not a lot of time, especially for crowded trains or if you have difficulties with mobility.
It seems that people being caught in the doors is not helped by the lack of dwell time, thus those used to “slowly” sauntering out of the carriage will find themselves in-between the doors sooner or later.
But this time, things got slightly more entertaining when drunkards boarded the train. It seems that the walk-through design of the train encourages them to run up and down the train as if it were a track and field event, disrupting everyone with their uncontrollably bad singing amongst other things. Unfortunately, while the MTS staffer could apologise for the disruptions, it seems he was powerless to deal with the open bottle of alcohol and unruly behaviour.
We thought the train would be happy once the drunkards had left the carriage, but on their way out, they knocked the platform PSD, causing one of the doors to stop functioning altogether. As a result, the train was held as only one side of the door would close. In-between door cycles, the MTS attendant pushed and pulled the door to try and get it back into alignment – eventually it managed to shut, but I wonder if any permanent damage was sustained. I surely hope not – the hardware is supposed to be able to handle a little “rough and tumble”, but perhaps when disturbed, the doors don’t attempt to reclose with the right force …
But perhaps they are a bit fragile – and this would need some fixing – as two doors at Tallawong station were out of service when I was there.
At least, the two trains I rode only overshot once each – both at Tallawong station. One of the stops was a bit marginal, but the doors still opened. Corrections still seem a bit too timid, requiring an extra retry, but is not as big of an issue as people make it to be. Corrections barely take all of half-a-minute, and it’s not the “being stranded” that some people like to claim.
As I was already recording audio, I had another chance to grab the in-carriage announcements. Unfortunately, I wasn’t aware how “compressed” and DSP processed the audio is from my Sharp Aquos S3, thus the audio quality is a bit degraded compared to the norm, but it’s still better than nothing.
While the passenger information displays and announcements worked much better than on launch day, it is not without the occasional issues. On my ride towards Chatswood, a malfunction occurred during the stretch from Epping to Macquarie University, resulting in the distance-to-go indication freezing and continual maximum-volume announcements of door opening and closing while in motion. While not a safety issue, it is very much a surprising inconvenience. It also seems to show that announcement volumes are controlled somehow (e.g. track-side tags? internal measurements of noise?).
Station Facilities Redux
As launch day entailed a lot of gates and crowd control, I wasn’t able to examine some of the finer points of station facilities. I may have missed noting it before – but on some stations, the T-coil signal seems a bit too loud.
At each station, there is a station guide that breaks apart each level of the station and where the facilities are. On stations with interconnections with buses, a graphical route map illustration is provided as well. This is quite convenient and is not something that’s often seen on regular train stations even though this is the standard overseas.
At Chatswood, the interchange signs within the building seem to have the metro “M” symbol coloured blue, similar to the bus icon. Perhaps this was the original colour, prior to the teal green, or perhaps someone made a mistake?
Passenger Information Displays
Passenger information displays seemed to be working better, with more services populated on the screen at Tallawong and times seeming sensible, however, are probably still based on train distance rather than real-time running forecasts.
At Chatswood, outside the station, one of the screens is now dedicated to the metro line.
Inside the gate, two screens are dedicated to the metro line, but both show the same thing. It’s better than having one screen show a list of terminating services, but in a stroke of genius, it seems someone had the foresight to reserve a screen for when Stage 2 goes live.
Unlike some of the other systems overseas which I have ridden, the platform emergency switch is not a push button but a key-switch. As a result, only authorised parties can declare a platform emergency and bring trains to a halt – whereas overseas, there are generally stop push buttons that anyone can actuate. Perhaps they don’t trust us not to use it improperly.
As someone who is interested in radio, I thought I’d take a look at the Metro from the perspective of radio. As the regular trains have moved from Metronet to GSM-R, I was wondering what the metro was using. As a result, I did some site and client-based searching on the ACMA Register of Radiocommunications Licenses database which resulted in 33 licenses for METRO TRAINS SYDNEY PTY LTD and 12 assignments for NSW Government Telco Authority at the Sydney Metro Train Facility Administration Building at 47 Tallawong Road. This suggests the administration building has facilities for the NSW GRN APCO P25 trunking system.
A look at the licenses seems to suggest that most of them are relatively low power 5W stations with an emission designator of 25K0D7W. What is 25kHz wide and digital? It’s likely to be TETRA. Looking at the list of TETRA users, it’s seen that Hong Kong MTR is using Motorola equipment, so it seems this is another “legacy” of MTR that we’re bringing over. The use of TETRA instead of GSM-R would be a bit disappointing as it means that there’s no direct interoperability with Sydney Trains radios, which are also incompatible with the APCO P25 radios carried by most of the public services workers.
A list of all the frequencies sorted by frequency shows that the lower frequencies are used for reception at the repeater sites (i.e. transmitted by mobile stations), with the higher frequencies used for transmission from the repeater site (i.e. received by mobile stations).
There are a total of 22 frequency pairs across the 33 licenses, thus some frequency re-use is in place.
Sorted by site, it seems that most sites have two or three frequency pairs in use. As TETRA is encrypted, there’s not much to listen to.
Another major point with the metro is the supporting infrastructure, which includes bus connections. Oddly enough, a few people were quite excited about the buses – one thing I neglected to mention was the cessation of the Station Link buses.
North West Night Bus
Signage informing passengers about the North West Night bus running from Sunday to Wednesday nights was seen in the stairwell in Chatswood. Notice the wording “Limited hours only.”
This is because, unlike the nightride services which cover the hours when trains are not operating, this bus really only covers for the missing metro services with two routes each running every ten minutes.
The on-demand service seemed to have captured some interest, although not everyone could see how it would work. Cooee Busways is the operator covering my area, but there is also Metroconnect. This week saw Cooee Busways advertising at stations, handing out leaflets and reserving all of Stand 1 at Schofields station, resulting in 751 and 734 services in both directions stopping at Stand 2, causing some congestion. The “unofficial” Stand 3 is home to the Elara private shuttle bus, which when combined with the mess at Schofields of excessive “kiss-and-ride” pick-ups, results in substantial delays during peak times.
Despite this, it does seem some people are using the Cooee service including even school children, although it is probably still far from being packed. I haven’t found myself using the Cooee at all, primarily due to the additional cost incurred using it and the rescheduled route buses actually better connecting with the train to the point I can’t guarantee a time benefit booking a Cooee compared to just catching the cheaper route bus. Sorry, Cooee – you’re going to be for emergencies, perhaps when it’s not too late on a weekday.
But people have (as expected) begun to complain about limited parking at metro stations filling up early. One of the answers? Leave your car at home and use public transport – route buses or on-demand. But the benefit of the metro has already been felt as some drivers who are still driving to the city are benefiting from reduced traffic jams.
Public Bus Stand
The public bus stand display now has finally booted but does not show any services. Hmm.
On the edge of the display, there is a blinking LED, but there are “clear” portions in the class in an ascending pattern like signal bars. Taking a photo with flash reveals an IR receiver and multiple LEDs – I suppose if someone knows the protocol, they could control the screen.
But by far, the most interesting thing to most was the solar-powered real-time bus display using dual eInk displays. This is a pretty cool device, as it means it updates in real-time, shows you the present time and never needs to be altered even on route changes. In the shots above, you can see a full-page refresh, and various selective-refreshes of the screen.
Who makes this? Well, an Australian company called Mercury Innovation. You’ll find their logo hidden on the rear.
It’s still early days for the Sydney Metro North-West and teething issues are to be expected. While initially, sentiment seemed to be a bit negative, it improved once the benefits of the metro became apparent even though the media still seems intent on blowing up some minor incidents out of proportion, perhaps inspired by various political agendas.
The service seems to run well enough for now. While the passenger information display has improved, the voice announcements still have timing issues and occasional glitches and overruns still happen from time to time. This is not a major concern though, not from the perspective of safety.
What may be a concern is the short dwell time at stations and the propensity for people to be caught in the doors if unprepared. Perhaps the doors can flash their lanterns half-a-second prior to moving to provide some level of warning and the announcements on platform can be better synchronised. Another issue is the seemingly fragile nature of the PSDs which can be put out of alignment by a shoulder-barging drunkard escaping the train.
The metro has already helped many people save time on their commute, whether they ride the metro or continue to drive to the city due to reduced congestion. Unfortunately, people on the North Shore T1 line may well be the ones losing out as the trains are becoming overcrowded in the morning with those transferring at Chatswood to ride into the city.
While many people seem to be looking forward to Stage 2, I’m not entirely sure it’s the best way of going about it. Having many more stations along the line (i.e. all of the Bankstown line) will result in a game of statistics – more stations means more probability of disruptions which could perhaps delay the whole line. Instead, perhaps the system should be separated into independent lines which may require an interchange (or two) to avoid the “flow-on” effects of a delay as regular trains currently experience.