Taiwan’s Mini Metros
And what we can learn from them.
The previous one got too long, so this one is a continuation. In this part, I focus more on the medium-capacity rail systems that can be found in Taiwan’s urban cities and the lessons we can learn.
If Matra doesn’t pull
There are things that have to be seen in person to be believed. The Wenhu (Brown) line in Taipei is one of them, a souped-up people mover system much more akin to our LRT lines. Regretfully, I have no videos, only this.
With part of it — the Muzha Line — being the first line of the Taipei Metro to open, the story here is, in any case, very akin to what happened with the Bukit Panjang LRT. Yes, down to headline-grabbing fires during testing, and worse still Matra, the developer of the VAL system, walking off the job due to the many construction delays.
Consequently, as then-Taipei Mayor Chen Shui-bian famously said,
馬特拉不拉 我們自己拉
(if Matra doesn’t pull, we’ll pull it ourselves)
And the Taiwanese did just that, instead of arguing whether the Muzha Line was an afterthought conceived through political pressure. They made it work. They pulled through. Then they extended it into the Neihu District of Taipei — placing even more stress on the system. At least they saw fit, at that time, to resignal the line with Bombardier’s Cityflo 650 CBTC system as part of this extension — coincidentally the same to be used on the Bukit Panjang LRT.
A 4-car Wenhu line train carries around 450 passengers, which is much less than our 3-car lines or even the upcoming JRL. In order to meet the travel demands of Neihu District, the kind of service they need to run stands at 80 seconds between trains. What does that look like? It looks like this. I can confirm that I personally was quite dumbfounded to stand there and witness it for myself.
They have not had to compromise on reliability to deliver this high level of service either, hitting an MKBF of nearly 5.4 million car-km. Normalized to what we see in Singapore, that’s around 1.3 million train-km, close to what Singapore MRT lines are delivering. And perhaps with much better service outcomes than our medium-capacity lines. In 2021, 62 trains — 124 2-car pairs — were needed to operate the Wenhu line in the morning peak, a utilization rate of 82%. I reckon it’s higher now that the 80 second peak has been increased. They only have 152 2-car pairs forming only 75 trains, so maintenance must keep every train in the best possible condition.
And I’ve also been told that work has been put in to increase speed limits and speed up round trips too — though I never got to validate this claim. But if true, this is also a further example of how Taipei squeezes every bit of performance it can from its systems before throwing hands up and asking for capital investment.
Of course, why do this when platforms are built for 6 cars? Surely longer trains, despite a reduction in service level, would be able to carry more people sustainably? Alas, that is a question I’m not confident answering definitively, but I’d suspect the greater political environment plays a part; considering how the DORTS as a government agency and the Taipei Metro as a state-owned company, relies on government funding, which varies with the ebb and flow of Taiwanese politicians.
But overall, I can’t help but think that Taiwanese railway culture also plays a big part. The sense of “get it to work” against the many engineering and even political obstacles may have been there long before the MRT, and it draws a big difference compared to how in Singapore we tame the beast.
In the South
If anyone wants to see the war on cars happen they can simply go to Kaohsiung, where intelligent traffic cameras have issued over 900 summons in just under a month, in the hopes of reducing the accident rate on the Kaohsiung LRT (read: tram). Has it been successful? We’ll see, but two accidents in a week is not a good look; even if only 10 throughout the entire year.
And I saw one happen. Here’s a picture to prove it.
In any case, is the Kaohsiung LRT delivering? I doubt; at only 10 minutes in peak hours and 15 minutes off peak, that’s far worse than many trunk bus services in Singapore. You can imagine the frustration of waiting over 10 minutes for a tram, going two stops, and getting into an accident. A bit of an extreme case, I’ll admit.
Yet, technically, Singaporean tram advocates might want to sit up and take note of several features, including the use of supercapacitors to power trams instead of overhead lines, the latter of which the Singaporean LTA seems to be ideologically opposed to. Natural green track is also widely used along the line — which could provide some sort of replacement for the grass verges so widely seen in Singapore. And of course, the intensive use of traffic cameras to catch traffic violators around the tram line.
However, at such low levels of service, I can’t help but wonder whether the Kaohsiung LRT is actually supposed to be good public transportation, or just meant for the tyre-spike types to fight their war on cars. There are other ways to fight a war on cars, I’d think. Or could this be an implementation issue? I’m not so sure, but considering that like the Taipei MRT, the Kaohsiung LRT is reusing parts of old TRA alignments, and has dedicated track. It doesn’t have to run in traffic like San Francisco streetcars, just across traffic.
And it’s not the only tram system either. New Taipei City — basically the suburbs of the Taipei metropolitan area — has plenty of tram networks too. But unlike Kaohsiung, New Taipei invested in grade separating these trams, and operate a model closer to San Francisco where multiple routes are interlined into these grade separated areas, which may be also topographically similar to the way we’ve planned our LRT networks. By avoiding traffic lights at road level, New Taipei’s trams can also be much faster than Kaohsiung’s. The use of trams also allows New Taipei to build smaller stations, with passengers expected to cross the tram tracks at grade instead of using bridges.
At least Kaohsiung also has a full-size MRT network — a network capable of using 6-car trains but which uses only 3-car trains for now — and that MRT network is capable of making most of the journey, with the tram a last-mile option similar to buses.
Small goes big
The last city I visited was Taichung, which has a certain notoriety in some of my circles. Why?
While most light metro projects literally mean just that, with shorter, smaller cars, the Taichung MRT uses trainsets that have each car just slightly smaller than our TEL ones, at 22m by 3m. And like the TEL, the Taichung MRT train cars have five doors per side. That also means the Taichung MRT has similar routing restrictions and technical requirements as our heavy rail lines; but run only 2-car trains.
The similarities don’t end there — with trains by Kawasaki and control systems by Alstom, I might even say that the Taichung MRT may well be what the TEL might look like if it were an elevated line. The oversized stations, such as at Taichung City Hall and Wenxin Chongde, also show us what we can expect on the JRL. At least JRL trains are longer and justify the larger stations.
In Taipei they’re a bit more sensible, and smaller vehicles are used on the Circular (yellow) line, at 17m by 2.65m wide. That way, maximum curve radii can be reduced as much as possible; and it’s very much needed, for the Circular line runs aboveground over roads much narrower than Singaporean roads, in denser parts of New Taipei City. The lighter vehicles may also permit a lighter structure, that doesn’t carry so much weight, to be built, which saves costs.
But on many of these light metros — both the Circular line and the Taichung MRT — and maybe even some of the grade separated tram lines, something that really bothered me was how the trains basically slow to a near halt to pass through very sharp curves. This is most noticeable at Banqiao station, whereby to bring the Circular line as close to the railway station as possible, a diversion involving two 50m radius curves has to be built. And this makes trains really, really slow as they enter the station. At least I’m thankful that the smaller Singapore LRT vehicles allow us to take such curves at a faster speed than 10–15kph.
And in fact, the Taipei area is building at least two more such systems. The Sanying line will use trainsets similar to the Circular line, but only in 2-car configuration. That’s probably going to be only as much capacity as a Singaporean LRT! Similarly, Taoyuan’s Green line will have trains that are as long as the Sanying line, but are 3m wide.
Common threads
This is not to say these light metro systems are entirely accessible.
A common design feature in both Taipei and Taichung is to have platforms and tracks over the road, then the station ticket hall and access in a building off to one side. Thus, one might find themselves having to cross the road, enter the station, then cross the road/train tracks again. There are several stations where a more “regular” design is used with the ticket hall directly above/below the platforms, but these are a minority. To be fair, urban land use and land acquisition may have forced them into this.
Consequently, to get to the opposite platform, it can be quite a trek, involving getting to one platform then crossing a bridge over the tracks to the other. A particular station of note is Jingan station, where getting from the lower of the two underground Orange line platforms to the Circular line — built above an elevated viaduct similar to our TWE — can involve up to seven escalators. Whatever it is, I’d hope that the EWL-CRL interchange at Gul Circle will never be this complex.
Similarly, the Circular line also introduced the idea of unpaid interchanges to Taiwan, at Banqiao and Xinpu/Xinpu Minsheng with the Blue line. The existence of Circular line Banqiao station and the hoops jumped through to bring the line there may point to Banqiao being more of a transfer point to TRA and HSR services than to the Blue line.
Otherwise, how can this be explained, where probably at least 25 or so people are queueing for a single lift to take them up from the underground passageways of the mainline station to the elevated Circular line?
At least the Xinpu/Xinpu Minsheng transfer is straightforward — exit one station, walk down the road, enter the other station. But like at Tampines in Singapore, I do wonder if just a bit more investment could have built an in-system transfer linkway there.
These are extreme examples, but they show how hard it is to build a good interchange between an elevated line and an underground line. At least we don’t have these problems with the JRL — we’ve even managed to build cross platform interchanges on the JRL — but they’re still worth noting as we expand our underground network and need to connect it with existing elevated lines, most notably at Clementi and Gul Circle.
Another common thread is the significant amounts of noise barriers being used on elevated sections in Taipei — not only on the automated metros, but even on the LRT elevated sections, and even on the rubber-tyred Wenhu line. Could it be that the Taiwanese have a lower tolerance of railway noise? It’s not impossible. While not a “mini” metro per se, the Xinbeitou branch line of the Taipei Metro has ridiculously low speed limits due to local resident complaints, resulting in what should have been a relatively quick ride become a slow trundle above Taipei streets. Specially built half-length trains have to be used for this route as well.
Plans to operate full-length through trains from central Taipei to Xinbeitou were also scrapped because of this, resulting in a situation where one must transfer from short-turn trains at Beitou platform 3, to shuttle trains at platform 4. Say what you will about the Changi Airport branch, at least there’s a cross platform interchange (for now).
A model?
I would even go as far as to say that the light metros of Taiwan may well be an example of what living along the JRL might be like when testing starts in 2026. Considering that this is the first time the LTA is building an automated, elevated steel on steel system, and in such close proximity to residential estates, there are considerable challenges and similarities we can see once we look at the Taiwanese mini-metros and LRT systems.
Similarly, they may also give us solutions towards new town transportation options, when simply spamming buses — and the accompanying congestion and manpower issues that arise — are no longer a sustainable option; which we may already be seeing in Tampines and Yishun. Traffic, like it or not, will always be part of transport, and planning around it — or to avoid it altogether — must take a key part in planners’ minds.
Of course, there are the lessons in operations, open data accessibility, and getting the thing to work, that are the most important. Taiwan has weighted the tradeoffs differently from us, and while we’re more conservative with many of these, they take more risks. No doubt some could work for us, like elevated trams or squeezing our peoplemovers. Some might not work well, like full-size trains towering way above the surrounding buildings.
But what we cannot deny is that the full spectrum of what is possible exists in Taiwan, which can be used as a reference. We might need to build a tram someday, be it with ground power supplies or supercapacitors. Or the LTA might see fit to reintroduce peoplemovers. Or even build a monorail (OK, Taiwan doesn’t have this).
Like what you read? Join the Telegram Channel for updates!
