Necessary or unnecessary, redux
With the importance of rail connections, how often should we tolerate service changes?
Another slate of service suspensions was previously announced on the Sengkang-Punggol LRT, with two loops being shut down every single weekend from May to September. At first glance, this may be disruptive, but the thinking goes that loops are loops and it wouldn’t be too disruptive to take a 10 minute detour going the other way.
And after all, single direction operation especially on the Punggol LRT wasn’t that long ago as well due to all the closed stations. That said, could this be a symptom of a greater issue? System reliability comes not only from unplanned breakdowns, but from whether people can count on rail systems to be there to transport them when they need it. Which means one too many planned closures should not be tolerated either.
On the topic of the lived experience, other transport systems around the world make it very clear if there are necessary systems works and service changes, both planned and unplanned. This appears to be our weakness; it may be due to fragmentation, as bus service changes have the same issue, where operators only announce what is relevant to them and the LTA Rail Division still views itself as a projects delivery organization . Then again, they could just learn from their OneMotoring colleagues who can put traffic updates front and center on OneMotoring’s front page.
Still, all these impact the lived experience, and trying to avoid such project-related closures in the first place should be the prime directive if we want to expect people to use rail services and not demand parallel bus routes which can be expensive to operate.
In search of final solutions
The closures of the LRT loops would be okay especially with their stated reason, which is to conduct reinstallation works on the expansion joints along the rollways. That said, though, one would have thought this was a one-off thing. However, from a quick search of the SBST website, 2022 is the seventh year running in which such planned closures have been declared, and it also doesn’t help that the size and scope of such closures basically drag out every weekend all the way to September.
So what is going on here? The expansion joints seen along the SPLRT guideways are similar to those on public roads and flyovers. The need to shut down the LRT loops for maintenance over the weekend has apparently existed since 2015 (which is as far back as I can find on SBST’s website), and it seems that the further back I look, the more I can find.
Interestingly, the Bukit Panjang LRT does not face this issue, instead just having gaps in the concrete rollways directly. Albeit these are actually positioned at a 45 degree tangent to the direction of vehicle movement, presumably to not adversely impact ride quality. Though, again, whether this helps or not is probably out for the jury.
The fly in the ointment, though, is the changing nature of the works. Earlier service changes in 2015, 2016, and 2017 at least allowed for the affected loop to be opened at 5.30pm in time for the evening peak. In early 2018 the situation improved and the affected loops were able to operate for almost the entire day, opening at 7am in the morning — only about an hour plus after the first train on the unaffected loop. In late 2018 and 2019, though, service only began at 5.30pm on the affected loop once again.
From 2020 onward, and in 2021 and 2022, one platform was and will be totally closed throughout the traffic day — not even 5.30pm, let alone 7am. Perhaps the harbinger of such difficulties was the urgent maintenance works declared on the 8th of January 2020, especially after the full Punggol LRT closure on the the 7th of December 2019.
I’m just a bit surprised that they hadn’t thought of final solutions for this issue such as by redesigning the rollways to use a more solid design of finger-joints. If we want to expect people to use the LRT, we need a certain level of reliability; while the looped nature of the LRT mostly mitigates the effect of single tracking, it’s still an inconvenience.
Or worse, what if this is a result of the Khaw-era mindless focus on MKBF and thus a desire to just get ahead of breaking finger-joints — in which case planned closures affect as much as unplanned closures but don’t hit the KPIs. The lived experience still results in the same effects, maybe even worse.
Phasing issues
Over to heavy rail lines, the TEL is planned to open in 5 phases. To open stage two and three, significant service changes to perform “system integration works” were necessary — this resulted in daily 9pm closures during TEL1, and further down the process, the combined TEL1+2 stretch only opened at 8am in the morning on weekends.
To a certain extent this is understandable given the tightly integrated nature of the modern digital railway, especially how we choose to develop our lines as automated and driverless ones. Even the Northern Line in London had to be shut down completely in order to make the necessary hardware and software changes to commission the new extension to Battersea Power Station which opened last year. And also in London, the recently opened Elizabeth Line will not operate on Sundays for now in order to facilitate further test opportunities for through services from the National Rail network, planned to open later in the year.
A second possibility is to allow the teams to catch up on lost productivity during the events of the past two years. The reason cited for the late Sunday opening of DTL in preparation for DTL3 was that trains needed 2.5 hours to complete a full trip between Bukit Panjang and Expo. While TEL1–3 isn’t likely to be that long, the extended hours could easily mean twice as much work can be done compared to a normal night.
In this vein we can probably tolerate, after TEL3, a few more weekends of no or limited service to facilitate stage 4 and 5 related systems integration. But no more than that! After TEL3 opens and people have switched their commutes to rely on it to take them to town, it becomes less excusable to shut down the line willy-nilly. Even the earlier DTL3-related closures took place only on Sundays and the line was reopened at 7.30am so as not to get too much in the way of anyone who might be out early on a Sunday morning.
A further bugbear might be Stage 5 which only has two stations and will open in 2025. It doesn’t help that for now, the official line is that the DTL3 Extension will open in 2024; and with it means Sungei Bedok station will very likely also open in 2024 barring further delays. This means that TEL Stage 5 will only introduce one new, and potentially less-busy station at Bedok South in 2025. Stage 4 construction progress, at time of writing, appears to be quite advanced, which puts paid to any chance of delaying the Stage 4 opening to be in line with stage 5.
So when we’ve established that many stages are bad and we should try to create as big batches as possible, what does this mean? The JRL has two defacto phases — JRL1 and JRL3. JRL stage 2 should be an independent operation by virtue of the stacked platforms at JS3 station. Likewise, the CRL has 4 phases; the Punggol Extension should be separated likewise due to separate platforms at Pasir Ris.
When that happens, perhaps the lessons learnt from TEL with regard to having so many rounds of system integration should also be applied. Maybe, for example, they could try to combine the extension to Changi T5 with CRL3 to Gul Circle, if the timing works out.
Chasing shadows
The MTR crashed a train at Central station in March 2019 during a regular system commissioning test for their Seltrac CBTC upgrade. Subsequent investigations revealed that implementation issues within a tertiary backup system, where steps were not taken to preserve the integrity of data as it flowed from zone computer to zone computer, was what resulted in the crash.
In this case, it was noted by the investigation panel that MTR’s request for a tertiary backup system separate from the two main computers was unique amongst Seltrac installs and that Thales had simply never done something like this before. This added complexity to the MTR re-signalling project. But that may be somewhat understandable for the MTR. With only two control zones controlling the whole Tsuen Wan Line, a zone computer failure could result in significant ramifications to the service, such as losing the critical cross-harbour connection. Perhaps their system architecture folks might have thought it better to have a third offsite backup.
The resulting ramifications from this have basically forced MTR to “reset” the whole project, and now in an about-face they have to provide the old SACEM system on 93 new Q-trains in order to facilitate the withdrawal of their very old M-trains.
Could the same happen in Singapore? I’m not sure, but something doesn’t seem right. Especially considering that the two retrofitted APM 100s, returned to Singapore in April 2021, have not yet re-entered service after 14 months and counting; that’s even before talking about the new APM 300R vehicles. The Cityflo650 CBTC and APM 100 solution is a proven one in several systems including San Francisco and Guangzhou, so what exactly did the LTA ask for to lend so much complexity to what should be a simple copy/paste job, in that the vehicles are still unable to re-enter service after so long? Could it be elements from Bombardier’s proposal to “not require any modification to existing infrastructure and minimal disruption to rail services” Well, with no Service A off-peak, that ship may have sailed.
And why is this important? You can’t fight entropy. Delays to the LRT asset renewal project may have meant more and more resources are needed to keep old vehicles operational. And it also means the system becomes more wobbly in terms of reliability. It got to the point where after two disruptions in a month, the ST had to work with the LTA on putting out a “we’re working on it” news article presumably to reassure people. All the more so with a reduced fleet requiring service changes to be made such as the off-peak Service A suspension. Extended outages result in lower fleet availability, and means longer waiting times for the passenger.
Risk management
Even the Levyist literature disapproves of taking on such added complexity. Whilst Levy focuses more on construction cost of structures, systems do present the same issues. By forcing contractors to do things a certain way, if their systems do not allow for such changes to be easily made, it means additional development risk has to be taken on by the contractor. Even if we are willing to hand over the money, development risk was what backfired for the MTR, and thus it’s still important to make sure that doesn’t happen here.
This can also be seen in practice in Toronto, where concurrent contracts to replace speed control systems, computer-based interlocking systems and to install CBTC were all discontinued. Alstom was then appointed by the TTC to fully take over all signalling system replacement responsibilities, as a single contractor able to introduce proven subsystem interfaces in the Urbalis 400 product.
As we shift from a “construction” mindset where all the time can be taken before opening to ensure the software is perfect, meeting defined levels of reliability, and with all features available, with asset renewal there is now a race against the clock to minimize disruption to those who have grown to rely on the provision of public transport service. Thus, it might be necessary to find a better middle ground between just getting projects delivered on time or mitigating any external events, and throwing in additional things that are nice to have.
Getting a handle on all these project demands could also help more efficient usage of night engineering hours. Maybe we might be able to operate later trains too, instead of the ineffective night bus services that were recently axed.
