A Not-So-Capital Plan, Part 1:
Perhaps the Most Expensive Subway Train in the World
Part One of a look at the MTA’s 2025-29 Capital Plan.
Cover images [1].
November 21, 2024
Introduction
It will come as no surprise to any observer that the MTA has a problem with cost control. The price of the agency’s capital construction projects keeps rising, most visibly with new projects like the extension of the Second Avenue subway to Harlem, now projected to be the most expensive subway line in the world. To this point, however, the cost of the MTA’s rolling stock—its subway trains, passenger coaches, and locomotives—has been reasonable. Sadly, this is no longer the case. In the MTA’s recently released 2024–2029 MTA Capital Plan the agency’s cost for new subway trains has ballooned far in excess of inflation. In fact, the trains that the MTA is planning to acquire for New York City Transit (NYCT) are now projected to cost more than double the subway trains currently being delivered today. That price makes them arguably the most expensive piece of metro rolling stock anywhere in the world.
Nor is high cost the only issue with these new trains. As designed right now, they will lack many of the features found on rolling stock around the world that improve both passenger experience and operational efficiency. In this case, these missing features and high costs arise from the same origin: the MTA’s excessively conservative approach to specifying and ordering trains. In some ways, this approach is understandable: the last time the organization experimented with major new features on its rolling stock, during the 1970s, the resulting fleet was disastrous. In fact, the trains were plagued by so many faults that they would bring down the companies that constructed them—in the process, all but eradicating the domestic American rolling stock industry. Ever since this experience, the MTA has followed a far more back-to-basics approach, one which has produced, to the agency’s credit, a series of highly reliable trains. Time and technology move on, however, and solutions that worked in the past are proving no longer effective in the modern day. Today’s riders would be far better served by purchasing essentially off-the-shelf modern trains rather than building costly, overly-customized units.
As recent political events have shown, New Yorkers are growing increasingly frustrated with their public sector. It is high time for our public agencies to demonstrate to a skeptical public that they can govern effectively. The subway is the lifeblood of New York City, and it should be an example to both the country and the world at large of how to follow best practices and operate transit well. The region needs to make sure we are buying the best, most cost-effective trains we can get our hands on, trains we will happily ride for decades to come.
Cost Explosion: The World’s Most Expensive Trains
According to research conducted by the Transit Costs Project at NYU Marron, rolling stock generally does not cost much more in the United States than it does in the rest of the world. This is in stark contrast to actual subway construction costs, where American costs are many times higher than those across the globe. By way of comparison, Chinese subway trains of today typically cost around $100,000/m [2] (in 2023 PPP dollars), while modern European subway, light rail, and commuter trains cost only somewhat more—for example, the new trains for Rome Metro’s Lines A and B cost $120,000/m. Even compared to its domestic peers New York has generally done well acquiring rolling stock at reasonable costs. The newest cars on the system, for example, the R211, cost $165,000/m for the base order, with the first and second options costing $164,000/m and $169,000/m, respectively, all in 2023 dollars. Conversely, contemporary West Coast rolling stock orders now average almost $200,000/m. In the Bay Area for example, BART’s new trains cost $164,000/m, while Muni’s recent light rail orders were $195,000/m (base) and $277,000/m (option).
In the MTA’s new capital plan, however, the cost of planned subway rolling stock has soared to an average of $272,000/m for the R268 [3] (B division) and $221,000/m for the R262 (A division) [4], in Year of Expenditure (YoE) costs. This cost is in a near-tie with a few other orders, such as the Toronto Line 2 replacement, depending on exact inflation adjustments. Most likely, among subway rolling stock procured without spare parts and maintenance, the R268 is the single most expensive train in the world.
Left: An image of the current R211 train. Right: A mockup of the proposed R262 train [5].
It is imperative that the MTA study why its costs have reached such an unprecedented level, and find ways to bring them more in line with costs around the world. This is especially important given recent political events. New York is now facing a period where accessing federal dollars to finance its transit needs may no longer be possible, making getting the most bang for the buck more important than ever. Perhaps just as importantly, today’s political winds have no patience for the dysfunctional status quo. The MTA and New York as a whole have an opportunity to demonstrate how effective management can create a sustainable, world-class transit system, even under difficult financial constraints.
The cost for these new cars is so high that it would be prudent for the MTA to get its rolling stock costs down to at least their historical levels before ordering any more cars than are absolutely necessary. Unfortunately, however, some of the new train orders are time-sensitive. For instance, the MTA must order replacements for the 38-year-old R68s and the 36-year-old R68As in time for when CBTC, the subway’s next-generation signaling system, Communication-Based Train Control, starts testing on Broadway (the N, Q, R, and W lines). At that point, the entire B division[6] (except for the small Franklin Ave Shuttle) will have significant sections wired for modern signals, requiring CBTC-capable trains. While newer trains (all from 2000’s R142 on) are designed to be able to be easily upgraded to CBTC, earlier trains (everything from 1988’s R68A and older, inclusive) are simply not CBTC-capable.
Conversely, the A division’s conversion to CBTC has not yet been planned (in large part thanks to its newer but still electro-mechanical signals). As such, there is less of a rush to replace the R62 fleet, which although built in 1983, still has a decent mean distance between failures (MDBF). Their siblings, however, the slightly newer R62As currently used on the 1 and 6 trains, not only have a poor MBDF, but also finicky air conditioning units that often break in the summer and are difficult to replace since they’re not modular. This has a major impact on customer experience, so replacing these older A division cars should be prioritized as soon as costs are under control.
Conservatism Born of Fire: New York City Transit Rolling Stock History
The recent cost explosion in NYCT rolling stock has to be understood in terms of subway history going back to the 1970s. During that decade, as the city lost more than a million residents, the subway system infamously began to fall apart. Train failures were constant, and graffiti was an uncontrolled scourge. The 1970s were also, however, a time when the subway attempted to innovate technologically—sadly, with disastrous results. The failures of this era have colored how the MTA has approached its rolling stock ever since, pushing it to adopt a very conservative approach. This served the agency well from the mid-1980s through the early 2010s, but since then, technology has moved on, while the agency’s approach has not.
In the 1970s, NYCT bought the R44 and R46 trains, using a number of then-new features:
Formation as eight 75-foot cars rather than 10 60-foot cars to lower maintenance costs by reducing the number of cars and trucks per train.
Then-new technology such as air conditioning, electro-pneumatic brakes, air springs for passenger comfort (on the R46s), and even the now iconic “bing-bong” door-closing sound.
Automatic Train Operation (ATO) capability for the soon-to-be-cancelled 1970s incarnation of the Second Avenue Subway.
A higher top speed, breaking the world speed limit for subway trains in tests (88 mph).
Lightweight truck designs (on the R46s) to reduce track wear and increase performance.
Other tests on one-off trainsets included hydraulic brakes and carpeted floors.
A new stainless-steel body design that would become standard for future subway rolling stock.
At the time, the United States still had a domestic rolling stock industry—albeit one that had shrunken considerably from its early-20th century heyday. Under Buy American[7] regulations, the MTA ordered its new trains from two venerable manufacturers: the St. Louis Car Company for the R44 and Pullman Standard for the R46.
Left: A refurbished R44 at Broad Channel. Right: The interior of an R46 train at 96th St / 2nd Ave. [8].
Unfortunately, both types of trains were plagued by major defects and operational difficulties. The 75-foot trains wouldn’t fit along major parts of the network, namely the BMT Eastern Division (L, J, M, Z) due to its tighter curves [9]. Although the cars were longer, they still only had 4 sets of doors per side, meaning that loading and unloading passengers took far longer during busy periods. Longer cars also meant more dramatic motion where cars met, and it was deemed too dangerous to allow people to walk between cars, a then-common subway practice. The doors on the 75-foot cars remain locked for this reason, which also presents problems during emergencies.
Subway yard workers had difficulty adjusting to the new electro-pneumatic braking systems. Rust began appearing on the R44s where a painted body stripe was built from carbon steel instead of stainless steel. The high top speed and automatic train operation were of no use once most major expansions were canceled. And worst of all, the R46 trucks began cracking in operation, forcing NYCT to bring back the older R16 trains while the R46s had an emergency refurbishment. To this day, the R46 truck debacle is often highlighted as a key example of the poor state of the subway system during the 1970s. NYCT sued both vendors for delivering defective products and won, leading both to exit the rolling stock business.
Just as the 1970s are infamous in the history of both the city and the subway for collapse, the 1980s are famous for the beginning of regeneration. For the subway, this began with Richard Ravitch’s appointment to head the MTA in 1979, and the subsequent establishment of five-year capital plans and the creation of the State of Good Repair program.
The MTA’s rolling stock purchases of the 1980s were an understandable reaction against the failures of the 1970s, pursuing a conservative technological course. By then, air conditioning was mature enough that it was included, but there were no new technical innovations in the first 1980s train, the R62. The R62 was, uniquely, fully made and assembled in Japan, as President Ronald Reagan’s budget cuts had defunded aid to transit agency capital plans, leaving the MTA free from Buy America restrictions (which only apply to federally-funded purchases).
Left: An R62 train at New Lots Ave. Right: An R68 train at 18th Ave. [10].
In subsequent years, federal transit funding returned, and with it Buy America regulations, even though by that point, the United States no longer had any domestic rolling stock manufacturing capacity to speak of. Instead, international vendors like Kawasaki and Alstom would establish plants in the United States to produce American orders, a situation that exists to this day. But the technological conservatism remains, leading to the R62A, R68, and R68A trainsets. All of these models have roll signs, lacking even dot-matrix electronic destination boards, due to the backlash against the attempted innovation during the R44 and R46 orders. The trains proved to be more reliable; to this day, the R62 and R68 have high MDBFs despite their age, and the R62A and R68A, while having much lower MDBFs, still do considerably better than the R46s did until their midlife refurbishment.
Rolling stock purchases since the 1990s have maintained the same basic conservatism. Before buying the more technologically advanced R142, R142A, and R143 trains, the MTA first bought demonstration trains in 1989, called the R110s, and tested them before incorporating their new technology, such as the digital display boards, into the production trains. The 2000s and 2010s orders have largely hewn to the same standards.
Left: The interior of an R142 train, which has served as a basis for all future MTA rolling stock. Right: An R160 at Avenue P [11].
This technological conservatism has, until recently, produced a series of reliable, cost-effective trains, entrenching a strategy of maintaining the status quo in MTA operations. Over time, however, this staid approach has begun to lead to a deterioration in quality and a rise in prices. As technology advances and passenger preferences change, so must procurement strategies.
By way of analogy, consider the smartphone market. In 2010, the BlackBerry was a solid if not cutting-edge technology for a smartphone. The firm’s technological conservatism, however, led it to attempt to maintain its status quo as its market position was quickly overtaken by the iPhone and Android phones. Eventually, BlackBerry was forced out of the market, and today, a consumer who insists on getting a BlackBerry instead of an iPhone or Android has no choice but to look to the secondhand market. Were a large firm to insist on procuring something akin to a BlackBerry phone, they would have to pay a large premium to get a manufacturer to build something that is no longer in regular production.
Today’s train vendors do not make trains similar to the R160, R179, and R211 any more than phone manufacturers make BlackBerries. The MTA needs to adapt to this new reality, not just to minimize cost, but to streamline operations and maximize passenger comfort.
Modern Train Design: Lessons from Around the Globe
It is time for the MTA to abandon its excessive conservatism and buy trains consistent with the normal technology of the 2010s and 2020s. Given its limited experience with these new technologies, the organization should not try to innovate beyond this. Right now, the most important thing for the organization is to get a handle on 21st-century rolling stock designs and buy trains compatible with them. Only then will the organization have the in-house expertise to innovate further.
Examples of rolling stock from around the world. Left: An Alstom Metropolis in Amsterdam. Right: A Siemens Inspiro in Warsaw. [12].
To this effect, the MTA should use a modular product like the Alstom Metropolis, Siemens Inspiro, or Stadler METRO. These trainsets simplify maintenance through standardization and include many features that remain rare or nonexistent in New York or that the MTA does not use to its full potential:
Open gangways, enabling passengers to move between cars for better circulation, increasing capacity, and improving safety by making subway surfing more difficult. New York recently took delivery of its first batch of open gangway cars as part of the R211T pilot, but this should be a regular feature and not just a pilot. All future trainsets should include open gangways, including the immediately upcoming R211 second option order.
R211T open gangway. From Khyber Sen, ETA.
Fire and smoke detection and suppression systems and improved ventilation. The MTA, in Bulletin 12-24, forbids the R211T from running on long express runs, specifically 59 St to 125 St and Hoyt-Schermerhorn St to Euclid Ave. And according to another rule, the R211T can only be used on lines where the spacing between every stop is less than 4 minutes, which is only the C and R lines. However, due to delays, trains often wait more than 4 minutes in between stations on local lines, too. The MTA has not given the public an explanation for why this is the case, but a Reddit account claiming MTA affiliation alleged that the R211T has an inadequate smoke clearance system, and others that it does not even have smoke detectors. Yet other MTA affiliates said that smoke could travel farther in the train, but that the air-circulation is overbuilt, and that this was an incidental finding that poses little actual risk, and that the MTA was being overly conservative here. If they need to be upgraded to be in line with international open gangway standards, the MTA must do so immediately for the upcoming R211 second option order, which they should specify as open gangway R211Ts. Finally, more powerful ventilation could also double in improving air quality (subway tunnels and stations are full of unhealthy steel dust and high PM 2.5) and reducing the spread of airborne disease.
CCTV of subway car interiors and exteriors, with train crew and dispatcher access. NYCT has begun adding security cameras to cars, and is including them in newly procured vehicles like the R211. Providing crews access to this feed will help improve security by increasing visibility in emergency situations. This technology can also be used to increase operating efficiency when turning trains at the end of lines where the train turns in non-revenue layup tracks beyond the station. Right now, NYCT requires conductors to manually check that trains are empty before moving them to tail tracks or yards. Verifying that a train is clear using video feeds would substantially speed up this process when the train is already empty, saving work for the train crew.
Cabs that allow the selection of a different driving position along the train without venting the air brake system. Current rolling stock requires venting the brake line when train operators switch to a different cab, such as at non-loop terminals. This slows down train turnarounds, reduces a line’s capacity, and produces a loud noise on the platform.
Future cabs that can support one-person train operation (OPTO) and even fully automated driverless (GoA3) or unattended operations (GoA4) in the future if technology and labor relations permit it. Even though the labor situation today makes implementing these procedures difficult, this may well change over the 40-year lifetime of a train car. If utilized, these features would enable significant off-peak service increases by reducing operating costs, which is important because all ridership growth now occurs off-peak [13].
A guaranteed emergency brake rate (GEBR) of at least 3 mph/s, in line with the performance specs of modern Movia, Metropolis, and Inspiro trains. Current NYCT trains are only theoretically capable of such braking rates; in real-world measurements, they performed notably worse. The IRT Capacity Report explains that the Queens Boulevard Line CBTC system currently assumes a GEBR of only 1.4 mph/s, and the best that current rolling stock can achieve is 1.8 mph/s, even though these trains are theoretically supposed to be 2.3–2.4 mph/s.
The staggered doors on a B car of an R142 2 train. From Khyber Sen, ETA.
Doors at the same locations as the existing rolling stock that will remain in service on each division, to avoid adding conflicts for platform screen door (PSD) installation. Historically, the MTA procured some B division trains as 75-foot cars, which have different door locations compared to the 60-foot cars. And the new A division trains have their doors staggered from each other, while the old A division trains and B division trains have doors facing each other. Ordinarily, the decision is a matter of internal train circulation, which staggered doors supposedly do better, but it is more important to have consistent door placement for all trains using the same line. Therefore, the R262 should stagger the doors like the R142 whereas the R268 should be like the R160 and not stagger them.
A reduced dynamic envelope to make PSD installation significantly more feasible, as recommended by the MTA’s PSD study. The dynamic envelope of a train car refers to the envelope within which it may sway while moving, even during rare equipment failures like the loss of suspension on one side of the train. To avoid strikes during this swaying, PSDs must be placed outside of this envelope [14]. Current B division trains have a dynamic envelope of about 11” from the train exterior, which forces PSDs to have at least an 11” gap between the train and PSD. Such a large gap is problematic for multiple reasons. For one, it necessitates the PSD being placed much further back, which may be impossible on NYC’s narrow platforms (without expensive structural reconstruction) that often contain columns and stairways very close to the platform edge. The MTA’s PSD study claims that PSDs are only feasible at 27% (128/472) of stations, largely due to violating ADA clearance (154 stations). Significantly reducing the dynamic envelope would free up valuable inches for ADA clearance, making the widespread installation of PSDs imminently more feasible. Secondly, the gap between the PSD and train would be large enough for a person to fit in, risking entrapment. This is dangerous and necessitates costly and inaccurate sensor-based gap detection. Instead, the dynamic envelope could be reduced, allowing the train to be nearly flush with the PSD. This is how most trains with PSDs operate around the world, like the 124-year-old Paris Metro Line 1, and even domestic ones like the Honolulu Skyline and JFK AirTrain right here in NYC, where sideswipe PSD strikes are extremely rare, just as platform strikes are extremely rare on the NYC subway. Clearly this is technologically feasible, so it is a matter of aligning the MTA’s conservative specs, which trains are certified to, to the global norm. For example, the MTA’s dynamic envelope assumes simultaneous loss of one suspension and operation at high speed, even though these cannot occur simultaneously [15]. This is exactly what Paris did as it installed PSDs on its old lines, tightening and refining the dynamic envelope standards to the actual physical capabilities of its trains and their signals. Furthermore, reducing the dynamic envelope has numerous other benefits, from track worker safety to smoother trains to reduced tunnel clearance (for example, allowing B division-sized trains to fit in the East NY Tunnel).
Left: A diagram showing the Limiting Line of Line Equipment (LLLE) and dynamic envelope (DE) of a B division train, and the 11” gap it creates. From the MTA PSD Study, but labeled by BQRail.
Right: A picture of the small gap between train and PSD on the JFK AirTrain. From BQRail.
Modern passenger information screens (PIS) that are capable of displaying any website on a modern standards-compliant web browser and are capable of intermittently connecting to the internet, at the very least anywhere there is cell service [16]. This would avoid issues with bespoke vendor software that is difficult to update by using global, proven standards and open-source software and websites that will update in real time. The MTA has taken great strides forward in its PIS designs, but because the software remains bespoke and inflexible, last week, the MTA had to resort to using paper signs over its brand-new R211 screens that couldn’t handle a reroute to 2nd Av.
Left: R211T PIS. From Khyber Sen, ETA. Right: Paper sign because PIS was too inflexible to update in time. From Jason Rabinowitz
Adopting the above rolling stock innovations from the international industry would serve double duty. It would create better passenger service than is available on the current rolling stock. But it would also reduce costs, by reducing the need to redesign trains from the international norms of the 2020s to those of the 1990s.
Conclusion
The subway cars that the MTA buys today could carry passengers for the next 50 years, much as the cars purchased in the 1980s have done reliably since then. While conservatism once served the MTA well, today’s situation is far removed from that of the 1970s and 1980s, and the MTA needs to adapt to changes in both the agency’s own situation and the subway rolling stock industry.
Today, it is more cost-effective to procure off-the-shelf subway cars with modern features. The global rolling stock industry is fully capable of supplying the MTA with equipment tailored to its requirements without needing to conform to holdover ideas from the 1970s-80s crises. With federal funding for rolling stock purchases far from certain over the course of the 2024-2029 capital plan, it is even more imperative to buy the best equipment at reasonable prices. And no matter the situation, New York should not own the dubious crown of having purchased arguably the most expensive subway train in the world.
At this quarter-mark into the 21st century, it is clear that transit governance in New York is in urgent need of improvement, and New Yorkers have become rightfully skeptical of governance without seeing improvements on the ground. By adopting best practices in rolling stock procurement, the MTA can start demonstrating that good governance is in fact possible and become an example for cities and regions across the country.
Footnotes
Background image from page 2 of the MTA’s 2024-29 Capital Plan.
R262 image by the MTA,Marron uses cost per meter of trainset length as its chosen measure, as the trains compared vary greatly in both number of cars and length of each car; in contrast, width is unimportant.
$1,775 million / (355 cars * 60.21 ft) = $272,450/m
$3,950 million / (1140 cars * 51.33 ft) = $221,451/m
R211: EmperorOfNYC, “A relatively brand new R211A subway car on the A line, awaiting departure from Inwood-207th Street” (July 7, 2023 https://commons.wikimedia.org/wiki/File:R211A_A_Train_@_Inwood-207th_Street_July_7th_2023.jpg).
R262 image by the MTA,The subway system is split into two divisions which use trains of different sizes: the A division, made up of the numbered lines (and the 42nd Street Shuttle), and the B division, made up of the lettered lines.
Not to be confused with the 1982 Buy America Act; Buy American was passed in 1933.
Adam E. Moreira, “"A" train entering the Broad Channel station, Broad Channel, NY,” (May 4, 2007, https://en.wikipedia.org/wiki/File:NYCSubway5274.jpg).
EmperorOfNYC, “R-46 Q Train terminated @ 96th Street,” (October 18, 2021, https://commons.wikimedia.org/wiki/File:R46_Q_Terminated_@_96th_Street_October_2021.jpg).Subway rolling stock is generally divided into two categories, with shorter, narrower trains on the A division (the former IRT, today’s numbered lines, with 8.5’ wide, 51’ long cars), and longer, wider trains on the B division (the former BMT and IND, today’s lettered lines, with 10’ wide, 60’ long cars). 75 foot trains could only operate on the former IND and the southern division of the BMT, meaning they could not be used on the L, J, M, or Z trains.
HazmatSchizo, "A Kawasaki R62 with General Electric propulsion departs Pennsylvania Avenue, bound for New Lots Avenue," (August 20, 2024 https://en.wikipedia.org/wiki/File:R62GE3TrainToNewLots.jpg).
MTAEnthusiast10, "A Gravesend bound D train made up of R68 subway cars departing 18 Avenue on the West End Line," (October 16, 2021 https://en.wikipedia.org/wiki/File:Diligent_on_West_End.jpg).
EmperorOfNYC, "An R142 awaiting to start service at 149th Street-Grand Concourse on a rerouted 6 line," (July 20, 2020, https://en.wikipedia.org/wiki/File:R142_6_Train_@_149th_Street-Grand_Concourse.jpg).
Mtattrain, "Jamaica/179th St-bound MTA NYC Subway F train of R160 cars arriving at Avenue P," (July 30, 2012, https://en.wikipedia.org/wiki/File:MTA_NYC_Subway_F_train_arriving_at_Avenue_P.JPG).
Willem_90, " Het teststel type M5 van de metro van Amsterdam tijdens de perspresentatie in september 2012," (September 13, 2012, https://en.wikipedia.org/wiki/File:Metro_Amsterdam_M5_Kraaiennest_4.JPG).
Janusz Jakubowski, "Metro, second line (M2), station Dworzec Wilenski," (March 8, 2015, https://en.wikipedia.org/wiki/File:Siemens_Inspiro,_Metro_Warszawskie_(16753246185).jpg).
The Hub Bound Reports of 2019 and 1989’s respective Table 14 data show that, where weekday subway ridership into the Manhattan core rose over this period from 1,838,503 to 2,227,922, ridership in the 7-10 am peak period actually fell from 1,009,010 to 923,095.
The MTA uses a slightly different term for this, the Limiting Line of Line Equipment (LLLE), the line beyond which no line equipment (such as PSDs, a tunnel wall, or wall equipment) can be placed. For the MTA, the LLLE is slightly larger than the dynamic envelope. The above recommendations about the dynamic envelope all equally apply to the LLLE as well.
Automatic Train Protection (ATP), a safety critical SIL4 (safety integrity level 4, or a one in a billionth chance of failure per hour) system, would not allow high speeds near stations.
As cell service is rolled out in more tunnels, the intermittent connection would shift to continuous connection.
.jpg){kind=link}
