Category: Transportation

When There’s Nothing Left To Burn, You Have To Set Money On Fire

Two recent news items have driven home the point that American construction costs are out of control. The first is the agreement between the federal government and the states of New York and New Jersey to fund the Gateway project, at a cost of $20 billion. The second is the release of more detailed environmental impact studies for high-speed rail on the Northeast Corridor; I previously expressed tepidly positive sentiment toward the NEC Future concept, but now there are concrete cost projections: the only full HSR option, Alternative 3, is projected to cost $290 billion. As Stephen Smith noted on Twitter, Alternative 3 is twice as expensive per km as the mostly underground Chuo Shinkansen maglev. As such, I am going to ignore other issues in this post, such as whether to serve Hartford on the mainline or not: they are real issues, but are secondary concerns to the outrageous cost figures.

Although both Gateway and NEC Future have extreme costs – too high for me to be able to support either project – the causes of those high costs are different. Gateway includes not just a new tunnel across the Hudson but also substantial unnecessary scope in Penn Station South; however, I suspect that even if the scope is pared down to the minimum required to provide four tracks from Newark to New York, the budget would still be very high. The bare Gateway tunnel (including Penn South) is to my understanding $14-16 billion; the maximum cost that can be justified by the extra ridership, unless additional operating improvements (which can be done today) are in place, is about $7 billion. As with Second Avenue Subway, there is a real problem of high unit costs. I emphasize that there is too much scope in Gateway, but the scope alone cannot explain why 5 km of tunnel cost many billions, when expensive non-US projects such as Crossrail top at a billion dollars per km and the geologically more complex Marmaray tunnel cost (in PPP terms) about $400 million per km.

The situation with NEC Future is different, in two ways. First, if Gateway cuts a zero from the budget, I will consider it a solid project, perhaps even an inexpensive one given the wide river crossing. (For reference, in 2003 the projected cost was $3 billion). In contrast, if NEC Future cuts a zero from its budget, I will still consider it too expensive – perhaps worth it because of the benefits of HSR, but certainly too high to be built without further inquiry. $29 billion for 720 km is justified for a line with a fair amount of tunneling and entirely greenfield construction, whereas the NEC has long segments that are already nearly ready for HSR and requires very little tunneling.

But second, and more importantly, NEC Future’s unit costs are not high. Read appendix B.06, which discusses cost: on PDF-p. 28 it breaks down cost by item, and other than the tunnels, which at $400-500 million per km are several times as expensive as intercity rail tunnels usually are, the infrastructure items’ per-km costs are reasonable. And the NEC doesn’t require much tunneling in the first place: Connecticut may be hilly, but HSR can climb 3.5% grades and ride on top of the hills, and only in Bridgeport is tunneling really necessary. Make it perhaps 5 km of required tunneling, all around Bridgeport. When I said $10 billion would build full-fat HSR on the NEC, I assumed $200-250 million per km for the Bridgeport tunnel. I also assumed $750 million for new tunnels in Baltimore, whose cost has since risen to $4 billion in part due to extra scope (4 tracks rather than 2). So 2 extra billions come from more expensive tunneling, and 278 extra billions come from bloated scope. Perhaps a subset of the 278 comes from high unit costs for systems and electrification, but these are not the main cost drivers, and are also quite easy to copy from peer developed countries. In the rest of this post, I will document some of the unnecessary scope. I emphasize that while Alternative 3 is the worst, the cost projection for Alternative 1, at $50 billion, is still several times the defensible cost of improvements.

Let us turn to chapter 4, the alternatives analysis, and start on PDF-p. 54. Right away, we see the following wasteful scope in Alternative 2:

  • Full four-tracking on the Providence Line, instead of strategic overtakes as detailed here.
  • A bypass of the Canton Viaduct, which at a radius of 1,746 meters imposes only a mild speed restriction on trains with E5 and Talgo tilt capability, 237 km/h.
  • An entirely new tunnel from Penn Station to Sunnyside, adding a third East River tunnel even though the LIRR is not at capacity now, let alone after East Side Access opens.
  • A tunnel under Philadelphia, so as to serve the city at Market East rather than 30th Street Station.
  • Two new HSR-dedicated tracks in New Jersey parallel to the NEC, rather than scheduling commuter trains on existing local tracks as detailed here.
  • Two new HSR-dedicated tracks alongside much of the New Haven Line, even in areas where the existing alignment is too too curvy.
  • Extensive tunneling between New Haven and Providence (see PDF-pp. 69-70 and 75), even in Alternative 1, even though HSR trains can climb the grades on the terrain without any tunnels outside the Providence built-up area if the tracks go west.

Alternative 2 also assumes service connecting New Haven, Hartford, and Providence, which I do not think is the optimal alignment (it’s slightly more expensive and slower), but is defensible, unlike the long proposed tunnels under Philadelphia, totaling around 30 km. The overall concept is also far more defensible than the tunnel-heavy implementation.

Alternative 3 adds the following unnecessary scope (see PDF-pp. 58 and 76-83):

  • Full six-tracking between New York and Philadelphia and between Baltimore and Washington.
  • Tunnel-heavy alignment options bypassing the New Haven Line, including inland options via Danbury or a tunnel across the Long Island Sound.
  • The new Baltimore tunnels are longer and include a new Baltimore CBD station, where the existing station is at the CBD’s periphery.
  • If I understand correctly, new platforms at New York Penn Station under the existing station.
  • Tunnels under the built-up area of Boston.

According to the cost breakdown, at-grade track costs $20 million per km, embankments cost $25 million per km, elevated track costs about $80 million per km, and tunnels cost $400 million per km. When I draw my preferred alignments, I assume the same cost elements, except tunnels are cheaper, at $200 million per km. (I also add 20% for overheads on top of these base costs, whereas these documents add contingency on top of that.) This should bias the NEC Future toward above-ground options.

Instead, look at the maps in appendix A. Alternative 3 is PDF-pp. 76-81. The options for getting out of the New York urban area include an almost entirely tunneled inland alignment, and a tunnel under the Long Island Sound; making small compromises on trip time by using the New Haven Line, and making up time elsewhere by using better rolling stock, is simply not an option to the planners.

Let’s go back to Gateway now. Although the cost premium there is not as outrageous as for NEC Future, it is a good case study in what the US will fund when it thinks the project is necessary and when there is sufficient lobbying. Paris has the political will to spend about $35 billion on Grand Paris Express, and London is spending $22 billion on Crossrail and is planning to spend much more on Crossrail 2. Between Second Avenue Subway, the 7 Extension, Fulton Street Transit Center, the PATH terminal, East Side Access, and now Gateway, New York is planning to have spent $43 billion on public transit by the middle of next decade. And now people are talking about Second Avenue Subway Phase 2. The political will to build both rapid transit and HSR in the US exists; the government spends tens of billions on it. But due to poor cost effectiveness, what the US gets for its money is almost nothing.

The $20 billion that the federal government and both states are willing to set on fire for Gateway prove that, were there a plan to build HSR so that trains would go between Boston and Washington in three and a half hours on a budget of $10-15 billion, it would be funded. This is not a marginal case, where the best plan still elicits groans from anti-tax conservatives: those conservatives ride trains between New York and Washington and want them to be faster. Instead, it is purely about excessive costs. Gateway’s $20 billion could build the tunnel and also full HSR on the NEC, and the $290 billion that NEC Future wants to burn on HSR could build nearly a complete national HSR network, serving most metro areas above 1 million people. It’s no longer a question of political will; it’s purely a question of cost control. 95% cost savings are possible here, and this is the only thing advocates for better intercity rail in the US should be focusing on.

Public Rapid Transit and Private Taxis

Cap’n Transit is writing about how, given that the political system in the New York area is hostile to public transportation expansion, private taxi companies are filling the gap, and this is fine (update: see the Cap’n’s comment below). This is not the first time I see people in the US claim that private shared-ride services are a substitute for public transportation; on Vox, Timothy Lee wrote that a ride-share service offered by Lyft is “the beginning of the end” for public transit. The tones are different – the Cap’n is hopeful that these services would get people out of cars, Timothy Lee is denigrating public transit and its supporters – but the message is the same: ride share is a substitute. I would like to explain why not only is this not happening, but also any hope of Uber, Lyft, and similar companies making the jump to conventional public transit is unlikely.

First, let us consider costs briefly. The biggest marginal cost of bus service is the driver, leading various futurists to fantasize about driverless taxis vastly reducing costs and competing with large buses. The only problem is, it costs too much to operate a car even without taking the driver’s wage into account. In the US, households spend 19% of their income on transportation; nearly all of this is private cars rather than plane travel or public transit. This works out to around $1.5 trillion a year, or about 30 cents per vehicle-km. Taxis have to pay this, and more, due to the cost of either the driver’s wage or the technology involved in automation. This is within the range of US urban public transportation‘s cost per passenger-km; the New York subway is 21 cents, and to be accessible to the masses it is subsidized. Of course, given automation, the subway would cost substantially less to operate.

This means that the only way taxi services can be affordable is if people share rides; Uber and Lyft are indeed moving in that direction. The problem is that sharing a car with a stranger ends the entire advantage of being in a car rather than on a train or bus. Slugging is not a popular mode of transportation; Wikipedia mentions a few thousand daily users in various US cities, whose subway systems get multiple hundreds of thousands of users. To offer even somewhat reasonable fares on their shared ride services, UberPool offers $5 promotional fares, and a maximum of two unrelated riders per car; sometimes, when the Uber system can’t find a second rider, there is just one rider, paying an express bus fare for private taxi service. It is not possible to make a profit in this manner.

Now, what the private sector can do, beyond taxis, is to scale up and offer vans and buses. It happens every day in the urban parts of New York beyond subway range: these are the dollar vans of various immigrant neighborhoods in Brooklyn and Queens, and the private services running in Hudson County. It’s possible that Uber and Lyft will eventually go that route. So far, tech startups involved in transportation have tried to reinvent the wheel, for example Leap’s failed attempt to provide premium buses within San Francisco, but it’s possible that a well-capitalized private company will instead try to offer more conventional bus service.

The problem is that the private sector has never in recent history scaled beyond that. This was not always the case: the London Underground, the New York els, and the Chicago L were built by private companies, often in competition with each other; in Japan, there are many private railroads, which built commuter lines by themselves in the prewar era. However, in recent years, rapid transit outside Japan has always been built publicly; when private companies exist, they either operate trains by contract, as in Singapore, or were initially public and only privatized after they were already running trains, as in Hong Kong. Japan belongs in the same category as Hong Kong, with one complication: the private railroads still build commuter lines in the suburbs, but, at least in Tokyo, they rely on the publicly-built subway for passenger distribution within the city core, as (due to prewar government regulations) the private lines do not enter Central Tokyo. Let us examine why it’s the case that the private sector no longer builds subway systems.

In the biggest cities of the world in 1900, the urban geography was simple: people worked in city center, or in their own neighborhood. This monocentric arrangement made it easy to build streetcars and rapid transit privately, since all a company needed was to build a line from the center to some suburb or outer-urban neighborhood. Network effects were weak, and transfers were not so important. The Manhattan els radiated north from South Ferry because there wasn’t much demand for east-west transportation; the Brooklyn els, the Chicago Ls, and the London Underground lines similarly radiated from the center in all directions.

Developing-world cities are in a similar situation. As they build their CBDs, they create situations in which people work in their home neighborhoods or in the CBD. For example, Nairobi’s matatu network is CBD-centric, with not much crosstown service, because the jobs that require commuting are concentrated in the CBD. Of course, there are many local jobs within neighborhoods, but usually people work in their own neighborhoods rather than commuting crosstown. However, construction costs in the third world are typically higher than they were in 1900 in what is now the developed world. When New York built the Dual Contracts – already at public expense – the cost was $366 million, which is (contrary the link to the cost figure) $8.6 billion today. This is around $50 million per km, about 42% underground. This cost is not unheard of today, but it is low; in China, $160 million per km is more typical of underground construction. See examples here, here, and here. Moreover, in the poorest countries considering transit expansion today, incomes are a fraction of the level of the US of 1913 ($6,500 in today’s dollars): Kenya’s GDP per capita is $3,000, Ethiopia’s is $1,500. Thus, rapid transit is less affordable. India, at $6,000, is more comparable to the US a hundred years ago, but it has high construction costs, and an urban geography that’s diverging from the monocentric layout I’m describing here.

In the developed world, construction costs, while higher than a hundred years ago, are more affordable, because the GDP per capita is not $6,000 but $30,000-$60,000. However, the cities are no longer monocentric; even relatively monocentric Stockholm has major secondary centers in the universities and in Kista, with high peak demand for subway service. In a polycentric city, a single line is no longer enough; the transit lines must work together as a network. The entire philosophy of Jarrett Walker‘s network restructures (i.e. the frequent grid) is based on this fact, taking bus networks that have not changed much from back when the cities were monocentric and updating them to reflect modern-day everywhere-to-everywhere travel reality.

With network effects so great, private startups can’t really step in and supplant the public sector. The barriers to entry are large, which is why the only companies doing so have a long history of corporate existence, either as private Japanese railroads or as recently-privatized companies, and are not startups. Of course, online social networks have large network effects as well, but they operate in a young industry, whereas transportation is a mature, conservative industry, without much opportunity to offer new service that does not yet exist. Advances come from engineering and network design and are slow and cumulative, unlike the situation in the tech sector.

Of course, the government could structure its regulations in a way that lets the private sector tap into public-sector network effects. For example, it could compel operators to cross-honor one another’s transfer tickets. But this is the exact opposite of how tech startups work, which is without such regulations. You can’t send a Facebook message to a Twitter account. It’s also not how European private ventures that run on public tracks and compete with public operators work: the Italian private high-speed rail service, NTV, does not cross-honor tickets from the public operator, Trenitalia, and vice versa. Once the government mandates free transfers between companies, and joint planning for network optimization, and schedules that are more cooperative than competitive, we’re back in the world of public planning, and the private companies just run service by contract, as they already do in such cities as Singapore and Stockholm.

Improving public transit, then, requires improving the public side of transit. Taxis are a niche; so are buses that can be run privately, to the CBD or to the public subway network. The core of transit ridership, in the cities where public transit usage is high, consists of a mesh of buses and rapid transit that cannot be grown spontaneously by the private sector. If the government can’t provide this, the city will be auto-oriented. Good transit advocates have to then work to make sure the government is more competent and can build this network, rather than hope successful private ventures will save them; there is no alternative.

Local and Global Knowledge

Adonia Lugo has a post criticizing Vision Zero, an American movement that aims at reducing the number of pedestrian and cyclist deaths from car accidents to zero. Adonia makes a lot of criticisms regarding lack of diversity within US bike advocacy, which I’m not going to discuss because I’m only tangentially familiar with it, via the general urbanist connection to Streetsblog. Instead, I’m going to zoom on one criticism, in which Adonia also invokes transit: Vision Zero activists look to a slate of European countries for guidance on making streets safer, including Sweden (which, alongside Norway and Denmark, has nearly the lowest car accident death rates in the world), and Denmark and the Netherlands (which are famous for their urban cycling facilities). Adonia’s response is,

With my inclusion filter on, it sounded like another example of white bike advocates looking to Northern Europe for solutions instead of turning to urban communities in the U.S. to find out how they’ve managed to get by walking, biking, and using transit all these years.

This is where I lost sympathy. What Adonia is asking, essentially, is for more respect for her (and her peer group’s) local knowledge, which is based on American cities in which few people who can afford cars take other modes of transportation. In the entire US, the only city where significant numbers of people who can afford cars take public transportation is New York, and there is not a single city where significant numbers of people who can afford cars ride bikes to work. This means that any discussion of improving transit access must include at least some knowledge of what happens outside the US.

Local Knowledge and Denigration

The problem is that talking about what happens outside the US shifts the locus of expertise from people with local knowledge to people with global knowledge. If an American city talks about adopting ideas from one of its neighborhoods, or even from a nearby city, there’s a lot of local knowledge, in the form of people who live or have lived in that area, or know many people who live there, and can evaluate a policy as to its success or failure. Internationally, there isn’t any, outside specialized forums; even highly educated Americans are usually monolingual, have never lived outside the US, and aren’t really plugged into the political debates in other countries, except maybe Canada.

The result is denigration. I’m not very plugged into cycling advocacy, so I’m going to use public transit for concrete examples. I have accepted that whenever I propose that comes from another country, someone is going to say “that’s there, this is America.” I definitely got this response when I started proposing modernizing regional rail in New York: “you are not a real New Yorker.” New York is the worst in the US in that it resists any ideas from other cities, even domestically.

It’s ultimately a defense mechanism against something that’s literally foreign, which the activist cannot evaluate because they and the people they trust haven’t really seen this in action. Thus, many Americans choose to believe that US public transportation is not a failure, that it’s just in bad circumstances and has little to learn from Europe. I’ve seen New Yorkers make remarks such as “there is no history of underinvestment in Europe” (yes, there is – look at Berlin during the Cold War, or at the removal of streetcars in postwar France and West Germany).

For example, I’ve found that bringing up Stockholm as an example of good transit in the US gets me accused of trolling, repeatedly, more so than bringing up London or Paris. The reason is that, to New York-based readers, London and Paris are almost peer cities, and to other Americans, London and Paris are equivalent to New York; therefore, they match the perception that public transit works in old, huge cities, but not in smaller or newer ones. In some ways, I think Stockholm is a better example of what US cities should aspire to, precisely because it is a small city. It is also not as old as London or Paris; between 1950 and 2010, Stockholm County’s population grew by a factor of 1.9, whereas metro Philadelphia’s grew by a factor of about 1.6, Boston’s by a factor of 1.4, and Chicago’s by a factor of 1.7. Boston in particular had a very good public transit network in 1950, and it systematically dismantled it and bypassed the remains, so that the metro area public transit mode share is only 11%. Expressed differently, metro Boston has 55 annual rail trips per capita, whereas New York has 95 and Stockholm 200. Of course, the cities of the US Sunbelt have had far more postwar growth than Stockholm (though many are comparable to Vancouver) and even lower transit usage than Boston and Philadelphia, and there indeed wholesale imports of European ideas are less practical. But it says a lot that in the oldest US cities, with the street layout most similar to most of Europe’s cities’, transit usage is still very low.

Adonia denigrates ideas she considers racist, but this denigration cuts across political and tribal lines in the US. I have seen considerable denigration from American urbanists that city centers could ever be family-friendly whenever I mention generations of families living in the central parts of Tel Aviv, or Vancouver, or Stockholm. There’s even a Twitter account dedicated to this denigration, The Suburbanist. Of course, what’s missing is the history of white flight and racism – not that Israel, Canada, and Sweden are less racist than the US, but their racism did not involve leaving inner cities to low-income minorities. But mentioning that cities aren’t bad places for families reminds certain people that they’re leaving the cities because they don’t like minorities, so they lash out. Nowadays, the Suburbanist engages in open racism, but this wasn’t the case a few years ago, nor is it the case with a large number of Americans who, in comments on various blogs (never here as far as I remember), yell at me for bringing up foreign cities.

Not Invented Here, Periphery Version

When planners and managers denigrate foreign ideas, this is called Not Invented Here syndrome. It is common in American transportation planning. I believe the reason Vision Zero sticks to “it works in Scandinavia” is to at least try to confront those planners with the fact that, by international standards, they have failed to promote road safety, especially for pedestrians and cyclists. Certainly this is the reason I bring up the failure of every US city except New York to maintain respectable public transit usage.

Now, the two centers of public transportation innovation in the world – Europe and Japan – brim with their own NIH problems, toward each other. Their rolling stock markets are almost entirely distinct, due to a combination of protectionism and regulations. Japan is outside the European Train Control System umbrella and keeps developing its own signal systems, while ETCS in turn is based on the features of older systems in the major European countries and not in Japan. Japan lags Europe in automation (driverless metros are less common there), track capacity in trains per hour, and small-city cost-cutting innovations such as proof of payment. Europe in turn has higher big-city operating costs, more accidents, less punctual trains, and usually heavier trains. Both of those centers would benefit from adopting each other’s ideas.

And yet, things work. There is enough indigenous transit expertise in Japan that despite missing out on European innovations, Japanese transit systems work well. There has to be; Tokyo has comfortably not only the highest rail ridership in the world, but also the highest rail ridership per capita, about 400 annual trips, versus 200-250 in the most transit-oriented European cities. Of course, Europe’s own indigenous expertise is nothing to scoff at, either.

The US is not in the center of public transportation. I am going to develop this center-periphery dynamic in a later post, tentatively called Unbroken Country: The Periphery’s Manifesto, which will also go specifically into Israel’s domestic problems with public services. But, in short, the US acts as if it is in the center, since it is one of the centers of the global economy, and is the undisputed center of global society and culture. This is what leads to NIH syndrome and denigration – Americans think they’re doing well because, in most aspects, they are. But when it comes to transportation, the US is a peripheral region (even road construction techniques lag Europe’s), and thus, its NIH problems deprive the public discourse of much-needed knowledge.

For a concrete example, let us consider rail signaling. ETCS Level 2 is designed around the needs of the biggest European countries, where the main lines are at least double-track, there is much more passenger traffic than freight traffic, freight trains are light because bulk freight goes by sea, and the population density near the main lines is high. Neither the US nor Sweden fits this. Most importantly, both countries have highly-trafficked freight lines passing through remote territory – Norrland in Sweden, the Interior West in the US. Sweden, which does not have NIH syndrome with respect to the rest of Europe, worked on developing a lower-cost version of ETCS, called ERTMS Regional; but in the US, the freight railroads as well as the commuter railroads (even in the Northeast, where ETCS Level 2 is appropriate) ignored ETCS entirely and developed their own incompatible systems, on the grounds that ETCS doesn’t meet unique American needs.

The Mystery of the Foreign

People who don’t know something often consider people who do know it to be mysterious, almost magical. It is a commonplace that, in low-literacy cultures, illiterates viewed the written word as magical; see this account of Early Modern Italy, but also a counterpoint from Ancient Greece. Of course, literacy in the first world today is universal, but two to three more modern examples persist, of relevance to US transit advocacy:

  1. Math, among people who are not mathematically- or technically-minded. I was asked recently whether my background as a mathematician influences my blogging, and explained that I use fairly basic math, but I am not afraid of numbers, which means I am not afraid of trying to compute cost figures, train speeds, and so on. I am also secure enough in my mathematical knowledge that I am not afraid of nitpicking technical points, or of being nitpicked.
  2. Foreign languages, among monolinguals. I do not know enough monolingual Hebrew speakers to confirm this in generality, but monolingual Anglophones seem to treat foreign-language information as somewhat magical. For example, the vibe I have been getting both here and on Twitter is that if I cite a foreign-language Wikipedia this gets more respect than if I cite English Wikipedia. The monolingual Anglophone can verify an English Wikipedia citation, and maybe notice small mistakes in the article, but not a foreign-language citation.
  3. This may be the same as 2, but, foreign experience. Relatively few native-born Americans have lived outside the US, so people who do are treated as having unique expertise about the country or continent they were in.

Point 3 applies even to knowledge obtained by other means than living in a country. In 2005, when progressive pundits were talking about how to implement universal health care, there was so little knowledge about how universal health care systems worked that Ezra Klein became an A-list pundit out of a few short profiles of various countries’ health care systems, The Health of Nations (see 2007 version here). I, of course, have gotten a lot of mileage out of Googling various cities’ subway construction costs and putting them together.

The problem with viewing the foreign as mysterious is that it leads to wholesale import of ideas that may not work, or may require significant tweaks before working. Bus rapid transit, an efficient mode of travel in middle-income Bogota and Curitiba, does not port to high-income cities well: paying six bus drivers rather than one train driver to avoid spending money on rail construction is a bigger problem in a country with a GDP per capita of $40,000 than in a country with a GDP per capita of $13,000. There are successful tweaks, such as open BRT (see description here), but Jaime Lerner and ITDP have pushed Curitiba-style closed BRT. Here, the lack of detailed knowledge about what exactly makes BRT work leads American cities – and no European or Japanese cities – to propose ill-thought closed BRT.

Another example of a bad import caused by this kind of magical thinking is the mixed-traffic streetcar. Here, American transit advocates don’t just think in terms of “Europe has trams” but also in terms of “the US used to have trams but we ripped them all up in the 1950s.” Here, US cities import a mode of transportation that exists as a legacy around Europe, but is uncommon on new-build lines, and is used mainly as a compromise when the streets are so narrow it’s impossible to give streetcars dedicated lanes without closing the streets to car traffic. As the US does not have cities with such narrow streets, outside a few old neighborhoods in Boston and New York with good subway service, its import of mixed-traffic streetcars is bad transit.

This relates to the point I made above, about local knowledge and bullshit. I know that many people view me as somewhat mysterious for having such a different knowledge base from Americans. It means I can comfortably bullshit about many points. I don’t bullshit, but it’s likely I’m making some mistakes, and I try to encourage my commenters to check me on them. But this requires commenters who are also very technically-minded and don’t think that just because I say something, it must be true.


In a sociopolitical environment in which the public and the activists have very little knowledge about imported ideas, whether they support them (usually viewing them as magical) or oppose them (usually denigrating them) is based on whether they identify with and trust the people proposing the import. Adonia does not trust the people who promote Vision Zero, since she views them as too white and male and too insensitive to the concerns of nonwhites, for example regarding police enforcement of speed limits. Conservatives, in turn, do not trust those people because they view them as cosmopolitan liberal urbanites, whence Tea Party opposition to various commuter and intercity rail expansions.

Consider high-speed rail, which is entirely a foreign import. The political coalitions for and against HSR in the US are based entirely on cultural identification with the proposition that Europe is better at something than the US. In particular, business-class small-government conservatives, who tend to be big fans of HSR in the countries where it exists, hate it in the US; George Will claimed it would make Americans more amenable to collectivism, and in Texas, right-wing populists have tried blocking an entirely private HSR scheme and possibly connecting it to the Democrats. In contrast, the populist left in the US (for example, Robert Cruickshank) supports public transportation infrastructure because of the environmentalist tie-in; in contrast, in the UK, Jeremy Corbyn, who Robert is otherwise a big fan of, is at best lukewarm toward HS2. In Europe, the left is more pro-rail than the right, but the populists on both sides are more anti-rail, and the overall left-right gradient is small; in the US, the left-right gradient is large, and this comes from the issue of trusting the transportation program of countries that Americans associate with welfare-state social democracy.

The result is that any dialog based on foreign transit has to involve a certain amount of mystery and trust in the planners. I have no trust in the planners, because of various wheel reinventions proposed even by reformists, but I know enough to discuss technical items and not just the people. Generally, other people with this technical background come from a similar social background: educated, geographically mobile, white, male. The result is that, as with HSR, people’s opinions on these projects track their opinions of the tribe in the US that can talk at this level of technical detail. Usually it’s not even racial, not when it comes to transit – it’s mostly suburbanites looking for ways to screw the urbanites. If anything, nonwhite neighborhoods in the US are underserved relative to best practices, and agencies sometimes sandbag the idea of more service there. Adonia just weaponizes this in a different direction from the usual.

Local and Global Knowledge

I’m not going to rule out the possibility that there is valuable local knowledge in the US about cycling, but I know that there is very little such knowledge about transit, and given where the high cycling use is, I doubt cycling is much different. This means that American knowledge alone is worth approximately nothing. Jarrett Walker is of course American and has a lot to contribute, but as a consultant, he has extensive Australian and Canadian experience and some East Asian experience. The bus grid as an idea predates Jarrett – Jarrett attributes it to (at least) 1980s-era reforms in Portland – but by itself it’s not a game changer.

The problem here is that to implement something successfully, the people who run it need both local knowledge and knowledge about places that work, i.e. global knowledge. If there aren’t enough people with both, the solutions will not work, because people who can’t contradict what the planners are saying can’t exercise democratic accountability over them. One of the reasons Europe does transit better is that there’s more foreign knowledge here; see above for the contrast between how the US and Sweden handle rail signaling.

In fact, if you look at the examples above in the Mystery of the Foreign section, they both come from failure to adapt a successful foreign system to local conditions – namely, high wages in the case of BRT, and wide streets in the case of trams. I presume that the people who build mixed-traffic streetcars and BRT lines in the US have plenty of local knowledge, but they lack the global knowledge to appreciate what exactly makes those systems successes abroad. Conversely, international consultants don’t have the local knowledge to say “no, this is not a good fit for your needs” (besides, usually that’s not what they’re paid to say). This problem is especially acute with innovations in developing countries, such as BRT, since the large gap in incomes leads to different situations requiring major changes in adaptation, much more so than the relatively minor differences within the developed world.

Now, consultants can pick up local knowledge. Their trade is not just to possess global knowledge, but also to know how to acquire local knowledge rapidly when they’re working in a city. They run surveys, look at detailed breakdowns of costs and ridership to tease out patterns, quiz the in-house planners, and travel all over the city to gain ground-level impressions. The problem is that if the consultants are the only people who have both local and global knowledge, then there is no democratic accountability, and they have an incentive to bullshit. Of course, Jarrett specifically does not bullshit, but he has occasionally made mistakes (the main one, anchoring, I’ve been meaning to write about for two years), and of course my personal trust in one consultant is no substitute to systemic, institutional trust in the ability of the technocrats to response to what is essentially peer-review practiced by the community of local advocates.

It means the only way forward is for activists in US cities to pick up global knowledge and engage with such plans on the details. In the case of cycling, I could think of any number of reasons why US cities cannot emulate the success of Amsterdam and Copenhagen; but then again, it’s possible these reasons are all irrelevant, in the same manner many reasons Americans offer for why they cannot have the same per capita transit usage of Sweden are irrelevant.

I am also suspicious of the fact that, per Adonia, bike advocates look to Northern Europe as a source of examples of success. The biggest bike share systems in the world are all in China, and in Japan, bikes have largely replaced buses as the preferred mode of access to the train station in the suburbs. US bike advocates owe it to themselves and to their cities to be informed about Chinese and Japanese practices, and, if they clash with Dutch and Scandinavian practices, then to have opinions about which ones to pick and how to synthesize them in a local context. The only way forward for people in the periphery, by which I mean all of the US when it comes to any non-car transportation, is to know how the core works well enough that they can adapt its innovation without being so reliant on outside experts.

LIRR Scheduling

The Long Island Railroad’s timetable is a mess. There is too little off-peak service, especially at the urban stations. At the peak, there is more service, but the service pattern is inscrutable. The Babylon Branch runs a skip-stop pattern in which trains make three stops, skip the next three, and then make the three after them. The pattern of which branch east of Jamaica is sent to which city terminal (Penn Station, Flatbush/Atlantic, or occasionally Hunterspoint) is inconsistent; passengers generally get timed cross-platform transfers at Jamaica, but the frequent interlacing of trains introduces a lot of dependency between different branches in the schedule, reducing reliability. Worst, the Main Line runs trains one-way, so for an hour in the peak, there is no off-peak service. As expected, reverse-peak ridership is minimal, even though there’s a fair number of jobs within a comfortable walk of Mineola. In this post, I am going to discuss how to improve the schedules.

The main tool I will use is a map of LIRR line speed zones. This was made by Patrick O’Hara, of the invaluable but now taken-offline blog The LIRR Today. I emphasize that Patrick does not endorse my plan to eliminate one-way service, on the grounds that it would unacceptably add to the travel time for conventional peak trips from Hicksville and points east to Penn Station. However, using the map and some data about rolling stock performance, I am going to show that LIRR schedules are so padded that improvements to reliability via simpler scheduling can reduce trip times significantly, more than making up for additional trip times to the elimination of most express runs.

First, let us compute technical trip times. In Boston, I compute these by looking at the acceleration rate of the FLIRT, but New York has passable rolling stock already, which means that modernization does not require full replacement of the fleet. This means we should use the specs of the M7: 13.9 kilowatts per ton (FLIRT: 21.7 maximum, 16.7 continuous), and an initial acceleration rate of 0.9 m/s^2 (FLIRT: 1.2). Assuming no air resistance, this means the theoretical acceleration penalty to 130 km/h, the speed over most of the electrified LIRR main lines, is 23 seconds. Judging by the difference between theoretical and actual FLIRT acceleration performance, the actual penalty is about 26 seconds. The deceleration penalty is 19 seconds, for a total of 45. Up to a speed of 100 km/h, the acceleration penalty is 17 seconds and the deceleration penalty is 13 seconds, for a total of 30.

Let us take dwell times to be 30 seconds. With reasonably wide doors at the quarter points and level boarding, it should not be difficult for the LIRR to hold to this standard. Actual dwells appear to be about 40-50 seconds, but are in the context of considerable schedule padding, as we will see. I am going to round speeds up from mph to km/h, so 80 mph will be rounded to 130 km/h, and 60 mph to 100 km/h; the numbers are close, and when I compute curve speeds, the total equivalent cant seems very low, such that large speed increases are possible. However, I am going to stick to the speed map, only changing to km/h for ease of calculation. Including dwell time, the stop penalty in 130 km/h territory is 75 seconds, and the stop penalty in 100 km/h territory is 60 seconds.

Of note, the actual stop penalties we see on LIRR schedules are larger, on the order of 100 seconds. Part of it is the padding again, but part of it is that LIRR trains do not accelerate as fast as they can; the LIRR derated its trains, limiting their acceleration to about 0.45 m/s^2 to reduce the electric current. This can and should be reversed. If it is not, the acceleration penalty is 40 seconds to 130 km/h and 31 seconds to 100 km/h, while the deceleration penalty, unaffected by the change to maximum acceleration, remains the same; overall, this slows trains by about 15 seconds per stop.

East of Jamaica, there are almost no slow zones on either the Main Line or the Babylon Branch. Hicksville’s 65 km/h zone slows trains that stop at Hicksville by about 30 seconds (even a few hundred meters from the station, trains could go faster if the line speed were higher). The curve between Bethpage and Farmingdale is worth 15 seconds. The slowdown in the interlocking at the junction with the Hempstead Line adds 5 seconds. The slowdowns in Jamaica add 35 seconds east of Jamaica, and 55 west of Jamaica, both for stopping trains. On the Babylon Branch, there are a few restrictions in the 80-110 km/h range, worth in total about 70 seconds; Babylon itself is in 100 km/h territory, adding another 10 seconds.

It is 63.6 km from Jamaica to Ronkonkoma. An express train from Jamaica to Ronkonkoma stopping only at Hicksville would do the trip in 33 minutes. A limited-stop train that stopped at Floral Park, Mineola, Hicksville, and then all stops to Ronkonkoma would do the trip in 44.5 minutes. A train that made every LIRR stop, even ones that Ronkonkoma trains never stop at today, would do it in 53 minutes. Under the current schedule, limited-stop trains, not stopping at Floral Park (with technical travel time of 43.5 minutes), do the trip in an hour, for a pad factor of 38%. After accounting for the fact that LIRR trains don’t accelerate this quickly because of the derating, we obtain a technical travel time of around 45.5 minutes, for a pad factor of 32%, still immense.

In Zurich, schedules are padded 7%. Rerating the trains to allow faster acceleration, and reducing the pad to 7%, would cut the trip time under the current off-peak stopping pattern from an hour to 47 minutes, which can be taken as either a material speed boost or as an opportunity to make more local stops. As I will argue later, trains should make more local stops – specifically, all from Floral Park east. This is five more stops than trains currently make; taking the 7% pad into account, we get 54 minutes, still a noticeable improvement over the current situation.

It is 17.4 km from Penn Station to Jamaica. Rather than detail the slow zones, I will just give the technical travel time, for a full-acceleration M7 making no intermediate stops: 13 minutes, or 14 with a 7% pad; 1 of those 13 minutes comes from the Penn Station throat and its 25 km/h speed limit, which is one of the reasons I have emphasized the need for simpler interlockings in station reconstruction. The schedule has 19 minutes, which is a 45% pad relative to full-acceleration travel time, and around 40% relative to the derated travel time. This is even worse, which I believe comes from a combination of congestion in the Penn Station area and the timed transfer at Jamaica; these mean that delays on one branch propagate to the others, requiring more slack in the schedule to maintain reliability. However, I will note that Zurich’s 7% pad is in the context of an environment with even more branches sharing a trunk line, and a plethora of timed transfers and overtakes.

It is 44.4 km from Jamaica to Babylon. An all-stop train – counting Saint Albans but not Atlantic Branch-only Rosedale and Valley Stream – would do the trip in 41 minutes. As I’ve argued years ago, the Babylon Branch’s stations all have relatively equal ridership, unlike the Main Line, where a few stations dominate, and therefore, we shouldn’t plan around express trains. The current schedule‘s travel time on all-stop off-peak trains is 53 minutes, a pad of 29% relative to full-acceleration performance and 19% relative to the derated performance. I believe the reason there is much less padding here than on the Ronkonkoma Branch is that the service pattern is simpler: off-peak, all trains make all stops, whereas the Main Line mixes skip-stop and express trains between the Ronkonkoma and Port Jefferson Branches. If all trains make the same stops and there are no overtakes, it’s easier to recover from delays, so there is less need for padding. (A similar principle is that you need less padding on double-track lines than on single-track lines.)

As mentioned before, at Swiss 7% padding, making all Main Line trains all-local from Floral Park east allows 54-minute service from Ronkonkoma to Jamaica. It also allows 69-minute service from Ronkonkoma to Penn Station, with a minute-long dwell at Jamaica. This is two minutes less than the fastest daily train on the current schedule, a nonstop that runs once a day and arrives at Penn Station at 7:30 am, before the greatest rush. Even at the Babylon Branch’s 19% padding, we get 60-minute service from Ronkonkoma to Jamaica and 76-minute service to Penn Station, which compares with 75 minutes for two peak trains with a few intermediate stops, and 82 minutes for off-peak trains with the above-mentioned pattern.

As for the Babylon Branch, going down to 7% padding and rerating the trains at higher speed means all-stop trains, including the three current local stops between Jamaica and Penn Station, would do the trip in 62 minutes. This is competitive with most peak trains: one train stopping only at Jamaica does the trip in 53 minutes, arriving at 7:02 am, but the other morning express trains, with pads varying based on how close to the peak of peak it is, do the trip in 62-65 minutes.

I claim that the solution to the problems of the Main Line is to indeed abolish all express runs. At the peak, there is no excuse for them: current traffic between the Ronkonkoma, Port Jefferson, and Oyster Bay Branches is about 23 trains per hour at the peak, and this means that either all peak-direction trains run local, or trains run one way, with local trains on one track and express trains on the other. The LIRR chooses to sacrifice reverse-peak service, because frankly providing a coherent network is not a priority; the priority is connecting peak-hour suburban travelers to Manhattan, and saving them a few minutes at any cost. This is despite the fact that peak travelers are the most expensive to serve – the peak is what drives capital investment, to say nothing of the crew utilization problems. But in this case, the peak-focused service may be self-defeating, as the above computation of pad ratios shows.

In the morning peak, west of Hicksville, the service pattern should thus be the same for every Ronkonkoma or Port Jefferson Branch train: all stops to Floral Park (where passengers could transfer to the Hempstead Branch), then express to Jamaica and then Penn Station. All trains should be as identical as possible, which means cutting the diesels to shuttles and, in the medium term, electrifying the Port Jefferson Branch to the end, since there is high ridership the entire way, whereas the Oyster Bay Branch and the Main Line beyond Ronkonkoma have low ridership. The dispatching should emphasize headway management rather than the schedule. Since all trains are functionally identical from Hicksville west, it does not matter to passengers if their favorite train left early – the next one will show up in at most 3 minutes. For the same reason, the transfer at Jamaica should not be timed at the peak.

The highest rapid transit capacity in the world is on subway lines that use headway management rather than fixed schedules, including the Moscow Metro and many modern driverless lines, where the limit is 39 tph. I do not expect 39 tph on the LIRR, but there is no demand for that on the Main Line right now; the point is to maintain 24 tph without excessive schedule padding. Off-peak, trains should keep a schedule because the frequency is lower, but the lower frequency is precisely what makes delays not propagate so fast; similarly, off-peak, the Jamaica transfer should be timed. The greatest problem is in the afternoon off-peak, but there, the bulk of boardings are at Penn Station, where delays are less likely since it’s the start of the line.

This pattern also suggests which capital investments the LIRR needs to make: it needs to construct interlockings such that there are no conflicts between Main Line trains and other trains. This means two things. First, grade-separating Queens Interlocking, between the Main Line and the Hempstead Branch, which currently has an at-grade conflict between opposing trains (eastbound Hempstead Branch, westbound Main Line). And second, reconstructing Jamaica’s access tracks from the east in a way that allows the Main Line from the east to continue on the Main Line’s express tracks to the west without interference from other lines. Right now, there’s an at-grade conflict with the Babylon Branch, but only in the same direction, which is less problematic.

This means kicking other branches off the express tracks from Jamaica to Penn Station, the most desirable track pair heading west of Jamaica. This is fine. Passengers on branches that connect to Flatbush, or to the local tracks to Penn Station, could still transfer cross-platform at Jamaica, even if at the peak the connecting train does not wait for them. Besides, as noted above, 7%-padded local trains from Babylon to Penn Station would have the same trip time as all but the single fastest express Babylon Branch train today.

Jamaica’s current track layout is 8 platform tracks, numbered 1-8, north to south. There are platforms between tracks 1-2, 2-3, 4-5, 6-7, and 7-8. This platform configuration allows three-way timed transfers: when a train platforms on track 2, passengers can walk from track 1 to track 3 via the train. Right now, to the west, the Atlantic Branch connects to tracks 3-6, and the four tracks of the Main Line each connects to two Jamaica tracks. But track connections exist to persistently connect tracks 2 and 7 to the express Main Line tracks, making 1 and 8 the local tracks and 3 and 6 the tracks to Flatbush. To the east, the Far Rockaway and Long Beach Branches connect to the Atlantic Branch without conflicting with other trains. Local Main Line tracks connect to tracks 1 and 8 without conflict. The only conflict involves the Babylon Branch, which runs in the middle between the eastbound and westbound Main Line tracks before diverging, and points at tracks 2 and 7. The current service pattern is that most Babylon Branch trains run express from Jamaica to Penn Station, making this track layout desirable. However, if they are switched to the local, single-track flyovers to connect them to tracks 1 and 8 are required, or alternatively a connection to tracks 3 and 6, which can be done without flyovers. In either case, three-way timed transfers would be retained, except at the peak.

Under my through-running proposal, the Atlantic Branch would continue to Lower Manhattan, so its demand would be much greater than today, encouraging a layout in which the Babylon Branch connected to tracks 3 and 6 and went to Brooklyn and Lower Manhattan. The Main Line trains would express to East Side Access and Grand Central, with an additional stop at Sunnyside Junction. The Hempstead Branch, connected to Penn Station and the Empire Connection, would have service increased, with mode-neutral fares encouraging more travel from within New York and Hempstead. I would also propose a new branch of the Hempstead Branch, using the inner Central Branch, going to the East Garden City job cluster. The Oyster Bay Branch would be electrified and its junction with the Main Line grade-separated.

However, I emphasize that none of my proposed schedule changes requires the intensive capital investment associated with connecting Flatbush with Lower Manhattan. Even East Side Access is not required. Queens Interlocking would be grade-separated, and the Oyster Bay Branch would be reduced to a shuttle with an additional track at Mineola (unless electrifying the entire line and grade-separating the junction is cheaper in the short run, which I doubt). Initially, I am not sure the at-grade conflict with the Babylon Branch on the approach to Jamaica would be deadly. The subway has a same-direction at-grade conflict at Rogers Avenue Junction, between the 2, 3, and 5 trains, whose combined peak frequency is higher than that of the Main Line and Babylon Branch’s. Rogers Avenue Junction is a key bottleneck on the numbered lines in New York, which is why the LIRR should not replicate it in the long run, but in the short run, it is fine.

To conclude, here are proposed westbound timetables for Ronkonkoma, Babylon, and Hempstead trains. These assume no new stations and only the minimally required physical infrastructure (that is, grade-separating Queens Interlocking).

Main Line:

Ronkonkoma 7:00
Central Islip 7:05
Brentwood 7:09
Deer Park 7:12
Wyandanch 7:16
Pinelawn 7:19
Farmingdale 7:23
Bethpage 7:27
Hicksville 7:31
Westbury 7:35
Carle Place 7:37
Mineola 7:40
Merillon Avenue 7:42
New Hyde Park 7:44
Floral Park 7:47
Jamaica 7:53
New York Penn 8:08

This is a total travel time of 68 minutes, and not 69 as advertised above. This is because of rounding artifacts.

Hempstead Branch:

Hempstead 7:31
Country Life Press 7:33
Garden City 7:36
Nassau Boulevard 7:38
Stewart Manor 7:40
Floral Park 7:43
Bellerose 7:34
Queens Village 7:46
Hollis 7:49
Jamaica 7:53
Kew Gardens 7:57
Forest Hills 7:59
Woodside 8:04
New York Penn 8:12

The 4-minute difference between local and express travel time between Jamaica and Penn Station comes from the fact that the intermediate stations are for the most part in slower zones than 130 – only at Forest Hills is there enough of a distance to get up to 130, and only west of the station, not east. Erratum: although it is true the stations are in slow zones, I wrote this paragraph thinking there are four intermediate stations, where of course there are only three; 4/3 = 80 seconds per stop, which comes from rounding artifacts.

The Hempstead Branch has a 1.5-km single-track segment starting west of Hempstead and ending east of Garden City. It is quite slow; the 25 km/h curve just north (west) of Country Life Press has geometry good enough for 50 km/h without any superelevation (cant deficiency would be 150 mm), and with 150 mm superelevation would be good for 70. Replacing that entire 25-50 km/h segment with 70 km/h saves about a minute of travel time.

Babylon Branch:

Babylon 7:04
Lindenhurst 7:08
Copiague 7:10
Amityville 7:12
Massapequa Park 7:15
Massapequa 7:17
Seaford 7:19
Wantagh 7:21
Bellmore 7:24
Merrick 7:26
Freeport 7:29
Baldwin 7:31
Rockville Centre 7:34
Lynbrook 7:37
St. Albans 7:43
Jamaica 7:48
Kew Gardens 7:52
Forest Hills 7:54
Woodside 7:59
New York Penn 8:07

I arbitrarily chose the Ronkonkoma departure time to be 7:00, and then chose the Hempstead Branch schedule to allow a timed transfer at Jamaica. The five-minute offset for the Babylon Branch should be suggestive of the proposed frequency: off-peak, every ten minutes on the Babylon Branch (possibly every twenty but also every twenty on the West Hempstead Branch), every ten minutes on the Hempstead Branch (possibly every twenty but also every twenty on the Central Branch to East Garden City), and every ten minutes on the Main Line, with each of the Ronkonkoma and Port Jefferson Branches getting a train every twenty minutes. The Atlantic Branch trains should run every twenty minutes per branch, with a three-way timed transfer with the Main Line and Hempstead Branch. Off-peak, the Babylon Branch doesn’t transfer to anything else, so there is no need to worry about its at-grade conflict at Jamaica.

Amtrak Pays More Than Double for High-Speed Trains

Update 9/24: as Alex Block notes in comments, sources at Amtrak deny the story, saying that Schumer spoke too soon, and there are still two bidders and Amtrak has not yet made its choice. If the cost turns out to be $1-1.25 billion rather than $2.5 billion, I will withdraw any and all criticism of the procurement process.

A press release from Senator Charles Schumer’s office is abuzz: Amtrak chose Alstom’s bid for its next order of high-speed trainsets, the Next-Generation Acelas. The press release mentions the size of the contract, $2.5 billion, and the number of jobs it would create, 750; it did not include any information relevant to passengers, such as the number of trains, the expected schedule of delivery, the expected frequency, and the expected travel time. Various media outlets have reprinted Schumer’s press release without such additional information, or indeed any analysis. Let me rectify this and provide some background as to why this order is a fleece.

The order is for 28 trainsets with 425 seats each. This can be seen here and here. Of those 28 sets, 25 should be available for maximum service, well below the 98% peak availability achieved by the TGV, but an improvement over the Acela’s current 16 trains available out of 20. There is no mention of the number of cars, which is how orders are usually priced. However, on page 30 of the technical specs, it is mentioned that the maximum length is 200 meters, equivalent to 8 cars. The capacity is equivalent to about six cars’ worth of seating at the normal seat density of economy-class HSR (including the Amtrak Regional coach), or about seven cars’ worth averaged over all occupied Acela cars. The RFP mentions half a bistro car with an option for a full car (page 21 of instructions to offerors), so eight cars per train is a reasonable assumption. I have seen references to ten cars per set, which I believe come from the option for two additional cars per train (the instructions phrase this as “an extra 33.33% capacity”). From Schumer’s press release it’s difficult to know whether the $2.5 billion figure is the base order or also the option.

Eight cars per train times 28 trains equals 224 cars. $2.5 billion divided by 224 equals $11.2 million per car; if I am wrong and these are ten-car trains, then it is $8.9 million per car. In China, a very high-speed train, capable of 350-380 km/h, costs $4 million per car; this is $900 million at the size of Amtrak’s order. In Europe, the new Eurostar order cost a total of €600-700 million for ten 16-car Velaro trainsets, about $4.7-5.5 million per car in PPP terms (see here and here); the uncertainty comes from euro:pound conversion rates and from the fact that a portion of the order is for refurbishment of the older trainsets. Siemens also sold 8-car Velaros to Deutsche Bahn for $5.2 million per car, again in PPP terms. Japanese trains are even cheaper, about $3 million per car in a recent N700 order, but only last 20 years, whereas European HSR trainsets last 40 and Amtrak specified a 30-year shelf life. The only non-US trainset order that I’ve seen that approaches the $10 million per car mark is the Velaro RUS, which is €600 million for eight 10-car trains, and this includes substantial modifications, such as winterization.

There is no excuse for such high costs. The technical specs are not particularly innovative: on page 22 of the document linked above, it is mentioned that cant deficiency should be 127 mm if the trains don’t tilt and 229 if they do, both of which figures are unimpressive by the respective standards of non-tilting and tilting trains. There is no explicit requirement for tilt. There is a requirement that trains be capable of traveling between New York and Washington in 2:21 (current trip time is 2:48) and between New Haven and Boston in 1:51 (current trip time is about 2 hours, skipping New London, which the specs require trains to stop at); there is no mention of which track upgrades are forthcoming, but given Amtrak’s heavy schedule padding, it is not difficult for a good train to meet the requirements. I do not bring these specs up to attack Amtrak for not demanding more of the trains, but to note that what Amtrak is asking is standard, so there is no reason for trains to be unusually expensive.

I will note that due to Buy America provisions, the trains will be manufactured in the US, at Alstom’s factory in Hornell. This has not caused cost blowouts for the large orders made by the New York subway, the LIRR, and Metro-North, but perhaps this order is small enough that requiring Alstom to build it at a new factory leads to major cost increases. It is also possible that due to difficulties in the bidding process, there are fewer bidders than is normal – Bombardier dropped out of the process last year, and in general, some US contracts have just one bid, with correspondingly elevated prices. But regardless of the reason, Amtrak’s order comes at a factor-of-two cost premium, and Schumer just expressed pride at the few hundred jobs that this waste would create.

NEC Future: Moving Sideways

The Northeast Corridor high-speed rail investment studies are moving forward, and four days ago the FRA released an early environmental impact study on the subject, as part of the NEC Future program. The study moves in part in the right direction, in that it considers many different segment-level improvements (for example, specific bypasses of curvy segments), but it still isn’t quite going in the right direction. It’s not a bad study in itself, but it does have a lot of drawbacks, and I would like to discuss the ultimate problems with its approach.


The EIS studies three alternatives, as well as an obligatory No Build option.

Alternative 1 includes minimal investment: capacity improvements already under consideration, including new Hudson tunnels; grade-separation of at-grade rail junctions, including Shell interlocking between the Metro-North New Haven Line to Grand Central and the NEC, which imposes a severe speed limit (30 mph, the worst outside major city stations) and a capacity constraint; and a limited I-95 bypass of the legacy NEC route in eastern Connecticut, to avoid the existing movable bridges. The bulk of the expense under this alternative, excluding the predominantly commuter-oriented new Hudson tunnels, involves replacing or bypassing obsolete or slow bridges with faster segments. I have advocated such an approach in certain cases for years, such as the Cos Cob Bridge; if anything, Alternative 1 does not do this enough, but I do appreciate that it uses this solution.

Alternative 2 constructs HSR along the NEC route, except for a major deviation to serve Hartford. It is also bundled with various bypasses and new stations elsewhere: under this alternative, Philadelphia and Baltimore get new stations, with extensive urban tunneling to reach those stations. Alternative 3 does the same, but considers more deviations, including a tunnel between Long Island and New Haven, and an inland route through Connecticut, closer to I-84 than to I-95 and the legacy NEC; it also constructs dedicated HSR tracks between New York and Washington.

The EIS does not include cost figures. It includes travel time figures on PDF-p. 51, which seem to be based on unfavorable assumptions: Alternative 2, called Run 5, does New York-Boston in 2:17 for trains making a few major-city intermediate stops; the Alternative 3 proposals vary widely depending on alignment, of which the fastest, the I-84 inland route, takes 1:51, again making intermediate stops.

The Good

First, the EIS includes service plan elements, stating the projected frequency of regional and express trains using the tracks. It also talks about clockface scheduling and proposes a pulse in Philadelphia, allowing timed transfers in all directions between local and express intercity trains as well as trains on the Keystone corridor. It goes further and discusses regional rail on the intercity tracks in the alternatives that include extensive new construction. In these ways, it focuses on regionwide rail integration far more than previous plans.

Second, in general, the correct way to think about NEC investment is component by component. The EIS gets closer to this ideal, by considering many different route combinations north of New York, and advancing several of them under the Alternative 3 umbrella.

And third, the concept of Alternative 1 is solid. In many cases, it is possible to bundle a trip time or capacity improvement into the replacement of an obsolete structure at very low additional cost. The example I keep coming back to is the Cos Cob Bridge, but it is equally true of the movable bridges east of New Haven. I also greatly appreciate that Alternative 1 recognizes the importance of grade-separating railroad junctions.

The Bad

Ultimately, the EIS does not take the three good concepts – integrated service planning, component-by-component thinking, and bundling trip-time improvements when the marginal cost of doing so is low – to their full conclusion. Thus, there is no attempt at running intercity trains at high speed on shared track with commuter rail with timed overtakes, as I have proposed for both the inner New Haven Line and the Providence Line. On the contrary, the plan for capacity investment on the Providence Line includes extensive three-tracking, rather than limited, strategic four-track bypass segments. This cascades to the trip times, which are quite slow between New York and New Haven (1:08, for an average speed of 103 km/h), and a bit slower than they could be between Providence and Boston (24 minutes, whereas about 21 is possible with about zero investment into concrete).

The concepts of Alternatives 1, 2, and 3 represent bundles of levels of investment. This is the wrong approach. Alternatives 2 and 3 include new tunneled city-center stations in Baltimore and Philadelphia; but wouldn’t we want to consider city-center station tunnels in those two cities separately? It’s possible for one to turn out to be cost-effective but not the other. It’s possible for neither to be cost-effective, but for other improvements included in Alternative 2, such as curve modification around Metropark and Metuchen, to pencil out.

There’s far more interaction between different macro-level alignments, by which I mean such questions as “inland route or coastal route?” and “serve Hartford on the mainline or put it on a branch?”, than between such micro-level investments as individual curve modifications and urban tunnels. This means that instead of discrete alternatives, there should be one umbrella, taking in Alternative 2 and 3 variants, proposing all of those options as possibilities. A future study, with detailed cost figures, could then rank those options in terms of trip time saving per unit of cost, or in terms of social and financial ROI. This way, there would be concrete proposals for what a $5 billion plan, a $10 billion plan, a $20 billion plan, and so on would be.

The Ugly

Two elements in the study are inexcusable. First, the service plan description explicitly keeps Amtrak’s current separation of premium-fare Regionals and even-more-premium-fare Acelas. This is not how the rest of the world structures HSR: even when the HSR fares are substantially higher than the legacy rail fares, as in Spain, the fare per passenger-km is not very high, and is not targeted exclusively at business travelers. In France, the intercity fare (including TGVs, which are the bulk of French intercity traffic) was on average €0.112 per passenger-km in 2011. Premium service is provided on the same TGVs as standard service, in first-class cars. In contrast, Amtrak charges about $0.29 per passenger-km on the Regional and $0.53 on the Acela.

And second, the investment alternatives appear to include more tunneling than is necessary. I will focus on the Hartford-Providence-Boston segment in Alternative 2, since it is less sensitive to assumptions on commuter rail track-sharing than the segments overlapping the New Haven Line. It is possible to go all the way from Hartford to the western margin of the Providence built-up area without any tunneling, and without outrageous bridging; see a past post of mine on the subject here, which concludes that it’s better to just go parallel to I-95 for trip time reasons. In Providence, tunnels are unavoidable, but can still be limited to short segments, mixed with elevated routes along pre-impacted freeway corridors. When I looked at it two years ago, I saw an alignment with just 2 km of tunnel, in Providence itself. In contrast, run A in figure 9 on PDF-p. 56 says that tunnels are about 27% of new construction between Hartford and Boston, which consists of, at a minimum, about 100 km of track between Hartford and Providence.


The EIS is a step in the right direction, insofar as it does consider issues of integrated service planning and prioritizing construction based on where it can be cheaply bundled into bridge replacement. However, it fails to consider cost limitations, as seen in the excessive tunneling proposed even in areas where high-speed tracks can run entirely above ground. It’s considering more options, which is good, but, Alternative 1, while representing a golden concept, is not sufficiently developed.

What I would like to see from a study in this direction is a mixture of the following:

  • Discussion of how to avoid tunnels, including various tradeoffs that have to be made (for example, above-ground construction may require more takings). Generally, I want to see much less tunneling than is currently proposed.
  • A well-developed incremental option, similar to Alternative 1 but more extensive, including for example I-95 bypasses all the way from New Haven to Kingston and along strategic segments of the New Haven Line, such as in Port Chester and Greenwich.
  • Greater integration with regional rail; one litmus test is whether the Providence Line is proposed to be three-tracked for long stretches, or four-tracked at a key bypass station (the options are Sharon and the Route 128-Readville segment), and another is discussion of high-acceleration electric multiple units on the Providence Line and the Penn Line.
  • Unbundling of projects within each alignment – there is no need to, for example, consider the Philadelphia and Baltimore tunnels together (I also think neither is a good idea, but that’s a separate discussion). The view should be toward an optimal set of projects within each alignment, since macro-level decisions such as whether to serve Hartford are more political than micro-level ones of which curves to fix. This permits explicit discussions such as “would you be willing to spend $2 billion and slow through-trains by 9 minutes to serve Hartford?”.

Except for the first, all are kind of present in this study, but in insufficient amount for me to view it as truly a step forward. The ultimate goal must be HSR in the Northeast on a reasonable budget – closer to $10 or even $20 billion than to the Amtrak Vision’s proposed $150 billion – and this requires carefully looking at which scope is required and which is not. The EIS has elements that can be used toward that goal, but ultimately it is a step sideways, not forward or in the wrong direction.

Third-World Train Speeds

Developing countries often have no railroads, or poorly developed ones. In India, the colonial railway network and more recent additions form a tight mesh, and train speeds today are in the 80-90 km/h range, respectable by the standards of legacy rail lines in Europe and the US that have not received significant upgrades. But in nearly all other ex-colonies in Africa and Asia, the situation is far worse – some countries have no railways, and others do but have slow lines from port to the interior. This is an attempt to catalog the average speeds of various countries’ fastest intercity trains; only in few cases are they even as fast as many Amtrak trains. Most of the information is taken from Seat61.

Syria: before the war, the fastest trains traveled the 361 km between Damascus and Aleppo in 4 hours, an average speed of 90 km/h.

Egypt: the fastest trains between Cairo and Alexandria (called Specials) do it in 2.5 hours, a distance of 208 km. This is 83 km/h.

Indonesia: the express trains between Jakarta and Surabaya do the 725-km trip in 9 hours. This is 81 km/h.

Pakistan: the Shalimar Express and Karakoram Express go between Karachi and Lahore in about 16 hours, a distance of about 1,250 km, according to Wikipedia. This is about 78 km/h.

Thailand: from Bangkok, trains go to Chiang Mai in (at the fastest) 10.5-11 hours for 751 km, Nong Khai in about 10 hours for 621 km, Ubon Ratchathani in 8 hours for 575 km, and Surat Thani in 8:40-9 hours for 651 km. At the fastest, it’s about 73 km/h.

South Africa: from Johannesburg, the long-distance trains average in the 50s. The fastest is Johannesburg-Cape Town, which does 1,530 km in 26 hours, an average of 59 km/h.

Zambia: the Golden Jubilee Express trains travel the 467 km between Livingstone and Lusaka in 12:27, for an average speed of 38 km/h. The through-trains to Tanzania go between the border and Kapiri Mposhi, a distance of 882 km, in 16:39 (and the source says they’re often late by many hours), an average speed of 52 km/h. Tanzania, in the same link, has an average speed from Dar es Salaam to the border of about 38 km/h.

Bangladesh: the fastest trains between Dhaka and Chittagong do the trip in just under 7 hours; on Google Earth, the route seems to be about 310 km. This is 45 km/h on average.

Burma: the fastest trains between Yangon and Mandalay, a distance of 622 km, do the trip in 14 hours, for an average of 44 km/h.

Kenya: Nairobi-Mombasa trains take about 15 hours. The line is 610 km long, for an average speed of 41 km/h.

Vietnam: trains go between Hanoi and Ho Chi Minh City in about 56 hours, a distance of 1,726 km. This is 31 km/h.

What’s Going on with Hudson Tunnel Cost Overruns?

Twenty-five billion dollars. The New York region’s political heavyweights – Andrew Cuomo, Chris Christie, Chuck Schumer, Cory Booker, Bill de Blasio – all want new Hudson tunnels, without any state funding for them; Schumer is proposing federal funding and a new interstate agency, parallel to the existing Port Authority, and a total budget of $25 billion. This is the highest figure I have seen so far; Amtrak still says $16 billion and Cuomo says $14 billion, and it’s likely the Gateway tunnels are indeed about $16 billion, while the remainder is for associated projects, such as fully four-tracking the line from Newark to the tunnel portal, a distance of about 11 kilometers. It is not my intention to criticize the cost; I’ve done that before.

Instead, I would like to point out that each time Gateway is the news, there usually seems to be a fresh cost escalation. Is it a $10 billion project? A $14 billion project? A $16 billion project? Or a $25 billion project? And what is included exactly? Amtrak does not make it clear what the various items are and how much they cost; I have not seen a single cost estimate that attempts to establish a baseline for new Hudson tunnels without the Penn Station South component, which would provide a moderate short-term boost to capacity but is not necessary for the project. The articles I’ve seen do not explain the origin of the $25 billion figure, either; it may include the tunnel and full four-tracking of Newark-New York, or it may include additional scope, for example Amtrak’s planned vertical circulation for a future (unnecessary) deep cavern for high-speed rail (see picture here).

The main issue here, the way I see it, is the interaction between public trust and political self-aggrandizement. It is common in all aspects of Israeli governance for new ministers to announce sweeping changes and reorganizations, just to remind the country that they exist and are doing something; this generally makes it harder to implement gradual reforms, and makes it completely impossible to do anything by consensus. Implementing a plan that was developed by consensus over many years makes one a bureaucrat; leaders change everything. In the US, this is the case not everywhere in government, but at least within public transportation infrastructure.

As we see in the case of Schumer’s call for a new interstate authority, the changes a heavyweight politician makes in order to appear as a leader have nothing to do with real problems that the project may have. Solving those problems requires detailed knowledge of the project at hand, which is the domain of bureaucrats and technocrats, and not of heavyweight politicians. Even a heavyweight who understands that there is a problem may not know or care about how to fix it: for example, Christie used the expression “tunnel to Macy’s basement,” invoking the deep cavern, to explain why ARC was wasteful, but chose to cancel the project rather than to remove the cavern and restore a track connection from the tunnel to Penn Station, which was in the official ARC Alt P plan until it was cut to limit the cost overruns. Managing a project is hard, and is, again, the domain of technocrats. The heavyweight will grandstand instead, regardless of whether it means canceling the project, or proposing an entirely new layer of government to build it.

As for trust, let us look at the benefits of new Hudson tunnels. The traditional, and least objectionable, is added capacity: the existing tunnels are currently at capacity during rush hour, and there’s much more demand for rail travel from New Jersey to Manhattan than they can accommodate. We can measure this benefit in terms of the combination of increased ridership from more service from more suburban areas, reduced crowding, and possibly slightly higher speeds. As a crude estimate of this benefit, current New Jersey Transit ridership at Penn Station is 87,000 per weekday in each direction. Doubling capacity means roughly doubling ridership, which would come from a combination of induced demand and diversion of traffic from cars, Port Authority buses, and commuter rail-PATH connections. This means the new tunnel can expect about 175,000 new commuter rail trips per weekday. At $10,000 per weekday trip, which is about average for very large non-US cities’ subway extensions, this justifies $1.75 billion. At $20,000, about the same as the projection for Grand Paris Express, Crossrail, and Second Avenue Subway Phase 1, all of which are justified on grounds of ridership and capacity on parallel lines, this is $3.5 billion. At $40,000, about the same as old projections for Second Avenue Subway Phase 2, which I used to analyze de Blasio’s Utica subway proposal, this is $7 billion. A $25 billion budget corresponds to a cost per rider well into the range of airport connectors.

Now, I’d like to think that informed citizens can look at these costs and benefits. At least, the fact that public transit projects only cost as much per rider as Gateway if they’re airport connectors (thus, of especial interest to the elites) or if something very wrong happened with the ridership projections, suggests that there is, normally, a ceiling to what the political system will fund. Even at $14-16 billion, the two states involved and the federal government groaned at funding Gateway, speaking to the fact that it’s not, in fact, worth this much money. In contrast, a bigger project, with bigger benefits, would be funded enthusiastically if it cost this much – for example, California already has almost this much money for high-speed rail, counting Prop 1A funds that are yet inaccessible due to the requirement of a 50/50 match from other sources.

Against this background, we see scare stories that Gateway must be built for reasons other than capacity and ridership. The old tunnels are falling apart, and Amtrak would like to shut them down one track at the time for long-term repairs. The more mundane reality is that the tunnels have higher maintenance costs than Amtrak would like since each track can only be shut down for short periods, on weekends and at night. This is buried in technical documents that don’t give the full picture, and don’t give differential costs for continuing the present regime of weekend single-tracking versus the recommended long-term closures. The given cost for Sandy-related North River Tunnel repairs is $350 million, assuming long-term closures, and it’s unlikely the present regime is billions of dollars more expensive.

I am reminded of the Tappan Zee Bridge replacement: the existing bridge has high maintenance costs due to its age and poor state, but the net present value of the maintenance cost is $2.5 billion and that of the excess maintenance cost is less, both figures well below the replacement cost. The bridge itself is structurally sound, but in popular media it is portrayed as structurally deficient. This relates to the problem of heavyweight politicians, for the Tappan Zee Bridge replacement is Cuomo’s pet project.

More fundamentally, who can trust any claim Amtrak makes about the structural soundness of tunnels? It says a lot that, when I asked on Twitter why transportation authorities do not immediately shut down unsafe pieces of infrastructure, various commenters answered “politics,” and on one (I believe James Sinclair) suggested that Amtrak order an emergency closure of one of the Hudson tunnel tracks just to drive home the point that new tunnels are necessary. I would like to stress that this is not Amtrak or a heavyweight proposing that, but the mere fact that commenters can seriously talk about it is telling. Most of the writers and commenters on the US transit blogosphere are very progressive and hate the Republicans; I have not seen a single comment recommending that the Democrats steal elections, fudge official statistics to make the party look more successful, or arrest Republican politicians on trumped-up charges, because in the US (and other first-world democracies), this is simply not done, and everyone except conspiracy theorists recognizes it. But politicizing the process of deciding which infrastructure projects are necessary for safety purposes and which are simply service expansions is normal enough that people can propose it half-seriously.

This brings me back to the issue of what I want the politicians to do, and what I expect them to do. What I want them to do is to be honest about costs and benefits, mediate between opposing interests (including different agencies that fight turf battles), and make decisions based on the best available information. This would necessarily limit costs, since, from the point of view of a member of Congress, if they get $25 billion for a piece of infrastructure then they cannot get $25 billion for another priority of theirs. They don’t do that, not in the US, and I’ve learned not to expect any better, as have the voters. Instead of working to make $25 billion go a longer way (to put things in perspective, I expect my regional rail tunnel proposal to cost $15-20 billion, at Crossrail 2 costs), Schumer is working to make $25 billion to sound like it’s going to a bigger deal than the new Hudson tunnels actually are.

None of this is a secret. American voters have learned to expect some kind of machine-greasing and politicking, to the point of losing the ability to trust either the politicians or the agencies, even in those cases when they are right. The result is that it’s possible to stretch the truth about how necessary a piece of infrastructure is, since people would believe or disbelieve it based on prior political beliefs anyway, and there is no expectation that the politicians or public authorities making those claims will have to justify them to the public in any detail. Lying to the public becomes trivially easy in this circumstance, and thus, costs can rise indefinitely, since everyone involved can pretend the benefits will rise to match them.

Why Labor Efficiency is Important

In North America, commuter trains run with conductors, often several per train. On most systems they walk the entire length of the train to check every passenger’s ticket, whereas on a few, namely in California, they do not do that anymore, but there are nonetheless multiple conductors per train. In addition, the scheduling is quite inefficient, in that train drivers do not work many revenue hours. I investigated what effect this has on operating costs, and it turns out that the effect on the marginal operating costs, which are important for off-peak service, is large: on the LIRR and Metro-North, nearly fivefold improvements in revenue train-hours per on-board employee (driver or conductor) are possible, which would halve the marginal operating cost per train-km. The bulk of this post is dedicated to explaining the following breakdown of variable operating costs:

train costs

The National Transit Database has figures for service in car-km and car-hours for a variety of US transit agencies. In New York State, the Empire Center has lists of every public employee’s position and pay, which we can use to figure out the average pay of a train driver and conductor and the productivity of their labor. The NTD numbers are as of 2011, so I will use the number of employees of 2011, but the pay per employee of 2014 (at any rate, there have been no major service changes since 2011, so numbers are similar). In 2011, the LIRR averaged 5,000 car-hours per driver-year, and Metro-North averaged 4,000; the LIRR runs longer trains than Metro-North, so the figure for both railroads appear to be about 500 train-hours per driver-year. Both railroads had a little bit more than 2 conductors per driver on average (2.14 Metro-North, 2.47 LIRR). The average pay of a driver, as of 2014, is $109,000 on the LIRR and $120,000 on Metro-North, whereas the average pay of a conductor is $112,000 on the LIRR and $96,000 on Metro-North.

From this, we can piece together the average operating cost of commuter rail derived from on-board labor, per train-hour: $771 on the LIRR, $714 on Metro-North. Assuming 8 cars per train (and again, the LIRR tends to run longer trains), this is around $90-95 per car-hour. According to the NTD, the average operating cost of both was about $550 per car-hour in 2011, but this includes fixed costs, such as management and rolling stock. As we will see, variable operating costs are much lower.

As a digression, I’d like to point out that the peaky schedule of commuter rail contributes to the low productivity of the drivers. Crew schedules include substantial gap time between trips, and occasionally, especially on low-frequency diesel branches, they deadhead. That said, the subway’s number of revenue train-hours per driver is not materially different. For higher figures, one must leave New York. Toei got about 700 revenue hours per driver when I last checked, but I can no longer find the reference. On the London Underground, I do have fresh references, pointing in the same range: 76.2 million train-km per year at 33 km/h average speed (from TfL’s facts and figures), and a bit more than 3,000 train operators. In 2012, the last year for which there’s actual rather than predicted data (see also PDF-p. 7 of the TfL Annual Report), there were 720 revenue hours per train driver. This is in tandem with a less peaky schedule than in New York: although the average speed is barely higher than that of the New York subway, as reported in the NTD, the trains travel about 180,000 km per year (see fact 149 here), twice as long as in New York. In Helsinki, metro trains run every 10 minutes all day on each branch, every day, without any extra peak service, contributing to even higher utilization: the schedules show 65,000 revenue-hours per year, whereas a factsheet from 2010 shows 75 metro drivers, for a total of 867 revenue hours per driver. In both the UK and Finland, average hours per employee are marginally shorter than in the US; London Underground drivers have 36-hour workweeks.

The importance of this computation is not just to highlight that 44-73% improvement in revenue-hours per employee is possible, but to point out that, on the margins, adding off-peak service would make crew schedules more efficient, since higher frequency would reduce the need to deadhead and to wait between trains. This means that, although the average operating cost may be about $750 per train-hour, the marginal cost is lower, even without changes to work rules.

Suppose now that trains run without conductors, using proof-of-payment as on light rail lines, even ones in North America, and on countless commuter rail systems in Continental Europe. Suppose also that there are 720 revenue-hours per driver, and that a driver is paid $115,000 per year. This means that running extra trains would not cost $90-95 in on-board labor per car-hour, but only $20, a nearly fivefold improvement. At Helsinki’s level of productivity, a nearly sixfold improvement to $16.60 is possible. At the LIRR’s present average speed of 50 km/h (compared with 53 on New Jersey Transit and 59 on Metro-North), the fivefold improvement based on London Underground productivity would cut the average cost per car-km from $1.80-1.90 to $0.40; at a higher but still realistic 67 km/h, it’s a cut from $1.35 to $0.30. A large majority of this cut comes from eliminating conductors, which, by itself, would cut costs threefold, but raising driver productivity would allow an additional cut of 30-40%. I again stress that the marginal cost is lower than the average cost computed here, since less peaky schedules come with simpler crew scheduling; more off-peak service would by itself cut the average cost, which means its marginal cost would be quite low.

Let us now look at other variable costs than on-board labor. Two years ago, I did this computation for high-speed rail, and found that, provided the schedules did not have extra rush hour service, operating expenses would be very low. We can do the same computation for commuter rail, and note that the lower speeds imply that operating and maintenance costs are spread across less passenger-km, raising costs. Let us consider train maintenance, cleaning, and energy.

I do not have information about train maintenance costs on commuter rail. Instead, I will use those of high-speed rail, for which standards are higher. As I noted in my computation from two years ago, the reference here is California HSR’s 2012 Business Plan, which aggregates these figures from around the world on PDF-p. 136. Maintenance costs per train-km are $4.47 for the Tokaido Shinkansen (with 16-car trains) and $2.58 per the UIC (with what I assume are 8-car trains), both in 2009 dollars. These figures cluster around $0.30 per car-km in 2009 dollars, or $0.30-35 per car-km in 2014 dollars.

With cleaning, there is some information about commuter rail: the Empire Center has lists of coach cleaners on Metro-North (there are 314) and their pay (on average, a little less than $50,000 a year). This seems high given the amount of service Metro-North runs – about $0.15 per car-km. Shinkansen trains are cleaned on a seven-minute turnaround in Tokyo, using one cleaner per standard-class car; this includes tasks that are not required on commuter rail, such as flipping seats to face forward. A cleaner making $30 per hour cleaning a single car per 15 minutes, with each train cleaned once per 150 km roundtrip, would cost $0.05 per car-km. I suspect that part of the low productivity of Metro-North cleaners is again a matter of low off-peak frequency – Shinkansen cleaners work almost continuously – but I don’t have comparative data to back this up; New York City Transit pays even more per cleaner per car- or bus-km, but this is on much lower average speed, and per car- or bus-hour, it pays about $6.40, vs. about $8.90 for Metro-North. I’m going to pencil in $0.10 per car-km as the cost of cleaning.

Energy costs we can compute from first principles. This is easier than for HSR, since commuter trains travel at such speed that a large majority of their energy consumption is in acceleration, rather than cruising. The explicit assumptions I am making is that the top speed is 130 km/h (the two main LIRR lines are mostly 80 mph territory), each car weighs 54 metric tons (the LIRR M7s weigh 57.5 and the Metro-North M8s even more, but this is very high by international EMU standards, thanks to FRA regulations), the average distance between stations is 4 km (the LIRR’s average is less than that if all trains make all stops and more if there are some express trains), and the track resistance per unit of train mass is the same as for the X 2000, for which data exists on PDF-p. 64 of a thesis on tilting trains. Regenerative braking is assumed to exactly cancel out with losses in transmission. Train acceleration performance is assumed to be like that of the FLIRT, which would take about a kilometer to accelerate to line speed and have about 2 km of cruising before slowing down for the stop; the M7 has inferior performance, but this would reduce energy consumption since trains would spend more time at lower speed.

With the above assumptions, each acceleration, cruise, and deceleration cycle between stations consumes about 13 kWh, of which 10 kWh is required to accelerate the train to top speed, and the other 3 are for overcoming track resistance. See rough computations in a subthread on California HSR Blog starting with this comment, and bear in mind the initial comment made a large computational error. As for April of this year, transportation electricity costs in the state are $0.1245 per kWh, giving us about $1.60 per 4-km interstation, or $0.40 per car-km.

Overall, those three items are $0.80 per car-km. This means that going from paying train crew $1.35 per car-km to paying them $0.30 per car-km represents halving of direct marginal operating expenses: it means going from $2.15 to $1.10 per car-km. Finally, let us add management costs, which are not exactly marginal costs, but do grow as the workforce grows, since more employees require supervisors. At RENFE, we can extract 0.27 support and management employees per operations employee from the data on PDF-p. 46 of its 2010 executive summary. On the Helsinki urban rail network, the corresponding figure is 0.34 as per the factsheet referenced above. This affects train crew, cleaning, and maintenance staff, but not energy. If this means 30% extra costs, this means going from $2.675 to $1.31 per car-km – again, we see costs are halved.

The off-peak LIRR fare is 15 cents per kilometer at long distances (14 to Ronkonkoma, but much more at shorter distances, for example 21 to Hicksville). If the marginal cost of running off-peak service is $1.31 per car-km, it means a car needs to have 9 passengers without season passes on it paying 15 cents per km for the trip to break even. If it’s $2.675, it needs 18. Passengers who commute off-peak and get season passes for those commutes also contribute, but less – a monthly pass for Ronkonkoma is $377, which at 46 trips a month is 10 cents per kilometer. It is not hard to have 9 passengers even on a long train, or even 13 (at the lower rate of season passes); Ronkonkoma itself is a park-and-ride, where this is less likely, but high enough passenger volumes as far as Mineola and Hicksville and all over the Babylon Branch are quite likely. If the required minimum is 18, let alone 26, this is substantially harder.

I harp on North American mainline rail operations for a variety of antiquated practices, but the on-board overstaffing is by far the worst. While improvement in train driver productivity can occur as a natural byproduct of improvement in off-peak frequency, getting rid of conductors is not so easy. It means a fight with the unions over job losses. Some of the required layoffs can be mitigated by retraining conductors as train drivers and running more service, but this would not boost service hours by a factor of 5; on the Ronkonkoma Branch, the peakiest of the three long LIRR lines, boosting off- and reverse-peak frequency to half the peak frequency would only increase train service by a factor of about 1.8.

I am not an expert on labor relations, so I do not know if any solution barring a prolonged SEPTA-style strike could work, alone or in combination. One possibility would be to commit to reducing working hours in the next five or ten years instead of hiking pay; working hours would be gradually reduced to core Western European levels, with 35-hour workweeks and 6 weeks of paid vacation, and hourly pay would rise as scheduled while annual pay would be frozen. Another possibility is that the MTA would help laid off employees find private-sector work, as happened in the 1980s with Japan National Railways (see PDF-pp. 103-4 of a handbook on rail privatization). This possibility requires implementing the reform at a time of wage growth and low unemployment, when private-sector work is easier to find, but the US is posting strong job growth numbers nowadays and is projected to keep doing so for at least another year.

But whatever happens, the most important reform from the point of view of reducing marginal off-peak service provision costs is letting go of redundant train crew. Halving the variable operating costs is exactly what is required to convert the nearly empty off-peak trains from financial drains to an extra source of revenues, balancing low ridership with even lower expenses. This would of course compound with other operating efficiencies, limiting the losses of branch lines and turning the busier main line trains into profit centers. But nowhere else is there the possibility of cutting costs so much with one single policy change as with removing conductors and changing the fare enforcement system to proof-of-payment.

Update 7/31: first, check comments below about maintenance costs: as far as I can tell from poorly presented Empire Center data, they are about 2.5 times higher, for both trains and the infrastructure, than the maintenance costs of high-speed rail. Although the effect of reducing those costs to conventional HSR level is larger than the effect of eliminating conductors, the details of reducing maintenance costs are far more delicate than those of eliminating conductors and running trains more often so that train drivers have less downtime.

Second, there is a small error in the above calculations: the figure of $90-95 in crew salary per car-hour is based on two conflicting assumptions. To get to $771 per train-hour on the LIRR, I assumed the LIRR ran 10-car trains. To get down to the $90-95 range, I assumed 8-car trains; 10-car trains would make this $77/hour. If we redo the entire calculation with 10-car trains, still with HSR maintenance costs, then instead of a cut from $2.675/car-km to $1.31/car-km, improved labor efficiency would cut costs from $2.415/car-km to $1.21/car-km. This is based on exact LIRR salaries, whereas the original calculation assumes hybrid LIRR/Metro-North salaries, and Metro-North pays drivers better than the LIRR.

Now, trains are somewhat longer at the peak than off-peak. If off-peak service is already with 8-car trains, and the average number of conductors is constant, then the original calculation (a cut from $2.675 to $1.31) still holds. After all, the salaries of train drivers and conductors are the same no matter how long the train is. But the number of conductors is not constant – let’s say it is proportional to train length, so 8-car LIRR trains have 2 conductors instead of 2.47, just as Metro-North’s average number of conductors per train is shorter than the LIRR’s, in tandem with its shorter consists. This changes the calculation to a cut from $2.535 (reflecting fewer conductors than in the original calculation) to $1.31. Observe that no matter what assumption we use, the operating cost cut coming from removing conductors and using drivers more efficiently is about 50%, give or take 1-2%.

Penn Station Elimination Followup

Several commenters, both here and on Streetsblog, have raised a number of points about my proposal to eliminate above-ground Penn Station and reduce the station to a hole in the ground. A few of those points are things I’d already thought about when I wrote that post and didn’t want to clutter; others are new ideas that I’ve had to wrestle with.


On Streetsblog, Mark Walker says, “Getting on a train at Penn is not like using the subway. Instead of a train that runs every five minutes, you’re waiting for a train that runs once per hour (more or less),” implying nicer waiting areas and lounges are needed. My proposal, of course, does not have dedicated waiting areas. (That said, there’s an immense amount of space on the platforms under the escalators, which could be equipped with chairs, tables, and newsstands.)

However, I take exception to the notion that when the train runs every hour, passengers wait an hour. When I lived in Providence, a few trips to Boston, New Haven, and New York taught me the exact amount of time it’d take me to walk from my apartment to the train station: 21 minutes. I learned to time myself to get to the station 2 minutes before the train would leave, and as I recall, I missed the train twice out of maybe 30 trips, and one of those was when I had a lot of luggage and was in a taxi and couldn’t precisely gauge the extra travel time. Walking is that reliable. People who get to Penn Station by subway have to budget some extra time to account for missed subway trains, but from much of the city, including the parts of the CBD not within walking distance from Penn, the required spare time is less than 10 minutes. Moreover, Penn is at its most crowded at rush hour, which is precisely when subway frequency is the highest, and people can reliably time themselves to within less than 5 minutes.

Outlying train stations in Switzerland are deserted except a few minutes before a train shows up, because the connecting transit is all timed to meet the train. This is of course inapplicable at very large stations with many lines, but the modes of transportation that most Penn Station users take to the station are reliable and frequent, if you can even talk of frequency for walking. The result is that the amenities do not need to be extravagant on account of waiting passengers, and do not need to be more than those of a busy subway station in a busy area.


Several commenters raised the idea of shelter. One option, raised by James Sinclair, is an arched glass roof over the station, on the model of Milan. This involves above-ground infrastructure, but the arched structure is only supported at the margins of the compound, which means that the primary feature of a hole-in-the-ground station, the lack of anything that the track area must support the weight of, is still true. I do not think it’s a bad idea; I do, however, want to raise three additional options:

Do nothing. A large proportion of the usable area of the platforms would be located under the walkways above, or under the escalators and staircases. Having measured the depth more precisely, through Plate 14 here, I found it is 13 meters from street level to top of rail, or 12 from street level to platform level, translating to 21 meters of escalator length, plus 2.2-2.5 meters on each side for approach (see page 23 here). About 16 of those 21 (18.5 out of 25.7, counting approaches) meters offer enough space for passengers to stand below the escalators, leading to large areas that could be used for shelter, as noted in the waiting section above.

Build a simple shelter. Stockholm-area train stations have cheap corrugated metal roofs over most of the length of their platforms. These provide protection from rain. Of course those roofs require some structural support at the platform, but because they’re not supposed to hold anything except rainwater, those supports are narrow poles, easy to move around if the station is reconfigured.

Build a street-level glass pane. This may be structurally intricate, but if not, it would provide complete shelter from the elements on the track level, greatly improve passenger circulation, and create a new public plaza. But in summer, the station would be a greenhouse, requiring additional air conditioning.

Note that doing nothing or building a simple shelter would not protect any of the track level from heat or cold. This is fine: evidently, open-air stations are the norm both in cities with hotter summers than New York (Milan is one example, and Tokyo is another) and in cities with colder winters (for example, Stockholm). Passengers are usually dressed for the weather anyway, especially if they’re planning on walking to work from Penn or from the subway station they’re connecting to.


Multiple commenters have said that public art and architecture matter, and building spartan train stations is unaesthetic, representing public squalor. I agree! I don’t think a hole-in-the-wall Penn Station has to be drab or brutalist. It can showcase art, on the model of the mosaics on the subway, or the sculptures on the T-Bana. It can use color to create a more welcoming environment than the monotonous gray of many postwar creations, such as the Washington Metro. The natural sunlight would help a lot.

But more than that, the walkways themselves could be architectural signatures. The best way to build them without supporting them on the track level is some variant on the arch bridge – either the classical arch bridge (which would require three or four spans), or a through-arch. This gives a lot of room to turn the bridges into signature spans. The design work would raise their cost, but short pedestrian bridges tend not to display the same cost structure as massive vehicular ones; the Bridge of Strings, a Calatrava-designed light rail bridge on a line that cost far more to build than light rail should cost, was $70 million for 360 meters. The walkways would not carry light rail, and would be about 140 or 150 meters in span.


Commenters both here (Caelestor) and on Streetsblog (Bolwerk, Matthias, C2check) have brought up transit-oriented development as a reason to allow a tall building on top of the station. With respect, I think on top of a train station is exactly the wrong place to build a tower. Let’s Go LA has an explanation for why the engineering for air rights is so complicated, although he stresses that Penn Station and Grand Central, which were built with the expectation of future high-rise air rights, are exceptions. I’ll add that Penn Station track simplification would also remove many crossovers and switches, making it easier to build air rights. That said, the track spacing is not friendly to the column spacing he proposes.

In New York, the tallest and most expensive recent private-sector office tower on solid ground, the Bank of America Tower, cost around $6,000 per square meter of floor space, in today’s money. Some of the luxury residential towers are more expensive; so are the new World Trade Center buildings, e.g. One World Trade Center was $12,000 per m^2. But the office towers cluster in a specific band of cost, around $2,500 to $5,000 per square meter, with taller towers generally more expensive. The Hudson Yards air rights towers cost in the $10,000-14,000 per square meter range, as much as One World Trade Center. Contrary to Bloomberg’s promises of windfall property tax revenues as his justification for the 7 extension, the city has had to offer tax abatement to encourage developers to build at those prices. Amtrak’s plan for Penn Station South assumes the block immediately south of Penn Station would cost $769 million to $1.3 billion to acquire; when I roughly computed its floor area by counting floors per building, I got 100,000 m^2, which means the price of real estate in that area, $7,700-13,000/m^2, is no higher and may be lower than the construction cost of air rights towers.

In contrast, some sites on firm ground immediately surrounding Penn Station are ripe for redevelopment. The block south of Penn Station, as noted above, has about 100,000 m^2, for a block-wide floor area ratio of 6.7. The Empire State Building’s floor area ratio is 33, so replacing the block with closely spaced supertall towers would require developers to burn just 20% of their profit on acquiring preexisting buildings. To the north of Penn Station, the two sites at 7th and 8th Avenues, flanking One Penn Plaza, are flat; so is nearly all of the western part of the block northeast of Penn, between 33rd and 34th Streets and 6th and 7th Avenues. Eighth Avenue is not developed intensely at all in that latitude – it only becomes important near Times Square. Supertall buildings surrounding Penn Station could even be incorporated into the station complex: railroads using the station might decide to lease offices in some of them, and the exteriors of some of those buildings could incorporate large clocks, some signage, and even train departure boards.


TheEconomist, who has had some truly out-of-the-box ideas, raises a very good point: how to phase the deconstruction of Penn Station in ways that allow service to continue. I don’t have a complete answer to that. Arch bridges, in particular, require extensive falsework, which may complicate matters. However, a general phase plan could consist of knocking down the above-ground buildings, then removing the upper concourse (leaving only the lower), and then removing arms of the lower concourse one by one as the walkways above them are built.

Passenger Throughput

In comments here, people have suggested several alternatives to my proposal to reconfigure Penn Station to have 12 tracks and 6 island platforms between them. There should be 6 approach tracks, as I outlined here: southern approach tracks, combining new Hudson tunnels with a link to Grand Central (which I call Line 2); central tracks, combining the preexisting Hudson tunnels with the southern East River Tunnels (Line 1); and northern tracks, combining the realigned Empire Connection and West Side Yard with the northern East River Tunnels (Line 3).

In my view, each approach track should split into two platform tracks, flanking the same platform. In this situation, there is no need to announce track numbers in advance, as long as the platform is known. Stockholm does this on the commuter lines at Stockholm Central: the northbound lines use tracks 15 and 16 and the southbound lines use tracks 13 and 14, with a platform between each of these track pairs, and until a few minutes before a train arrives, it’s signed on the board as “track 13/14” or “track 15/16.”

The compound looks 140 or 150 meters wide; the maps are unclear about to what extent Penn extends under 31st and 33rd, but according to a diagram Joey shared in comments, it extends quite far, giving 150 meters or even a bit more. Under my proposal, this is enough for 6 platforms of 17 or 18 meters. It sounds like a lot, but it isn’t, especially on Line 3, where Penn Station is the only CBD train station, which implies entire trains would empty at Penn in the morning rush hour. (Line 2, which I expect to be the busiest overall because it’d serve both Penn and Grand Central, is the one I expect to have the least platform crowding problems, precisely because it’d serve both Penn and Grand Central.)

Staircases should be 3 meters wide. Escalators with 1-meter steps have 1.6-meter pits; their capacity is theoretically 9,000 passengers per hour, but practically only 6,000-7,000. Clearing 30 entire trains per hour, filled to seating capacity with 4 standees per square meter of standing space, requires moving about 75,000 passengers per hour. (Per meter of train length, this is comparable to the 4/5 trains and the RER A at their peaks.) With 6 access points, this requires 2 up escalators per access point. The minimum is then 3 escalators, running 2-and-1 at the peak; 4 is better.

In comments, Ari Ofesvit proposes the Spanish solution, which I’ve discussed in previous posts. I’m now convinced it is not the right solution, simply because it compels platforms to be too narrow (about 8.6 meters), which has room for exactly half of what a standard platform twice the width would have, without the possibility of running 4 escalators 3-and-1 at the peak. My comment in that post has more detail, albeit with the assumption that the compound is 140 meters wide.

Fbfree proposes something else: more platforms for intercity trains. Giving intercity trains more platforms (as is done in Stockholm, which has just two approach tracks to the south) gives them more time to dwell; unfortunately, it also narrows the platforms for the regional trains, precisely the ones that can expect the most crowding. Even a single-track platform would take up space out of proportion to the number of passengers it would serve.

Pedestrian throughput is, at the maximum, 81 people per meter of walkway width per minute; this assumes two-way flow, but the numbers for one-way and multiway flow aren’t too different. This is a little less than 5,000 per meter-hour. An escalator bank with two up escalators then needs almost 3 meters of unobstructed platform width on one side (the other side can be used as overflow, but most passengers would use the side of the platform the train discharged them on). This is easy to supply with a 4-escalator bank on a 17-meter platform (there would be 3.8 meters); on an 8.6-meter Spanish platform, there’s only one up escalator per bank, so half the width is required, and is indeed obtainable. But if there are extra platforms for intercity trains, this becomes more strained.

For maximum throughput, it is necessary to minimize separation between escalators on the platform, down to about 6 meters plus approaches, in order to allow wider walkways, which in this case would make the walkways about 25 meters wide. The point here is that the walkways have to have very high pedestrian capacity, since each of them is fed by escalators from all platforms. At 25 meters, the capacity is about 15% less than that of two up escalators per access point (121,500 vs. 144,000), which is fine since some platforms (Line 2 in both directions, Line 3 eastbound in the morning and westbound in the afternoon) would not have so much traffic. But putting in elevators would disrupt this flow somewhat.

I see two ways to increase capacity in the future, if train traffic warrants it: first, build the glass floor/ceiling I outlined above, in the shelter section. This is the simplest possibility. Second, build three more walkways, midway between 7th and 8th Avenues and the two walkways already discussed, and have each walkway or avenue serve only half the platforms – one serving eastbound platforms, one serving westbound platforms. At this point the station would be half-covered by walkways, if they are all about 24 meters wide, but the walkways could be narrowed; as long as they are longer than 15 meters, any passenger arriving on a platform by any of the included access points would be sheltered by the walkway serving platforms in the opposite direction. Elevators should go from each walkway to each platform still, which would facilitate transfers, but the workhorse escalators would spread the load among different walkways.


I’d originally thought that the walkways could host retail and food concessions. The calculation in the preceding section suggests that this wouldn’t be possible, unless the walkways are widened beyond the escalators, with concessions on the outside. Every meter of walkway width would be required for passenger circulation. Even information pamphlets might be restricted to the very edges of the walkways; train departure boards would have to be mounted in the air, for example on the support cables if the through-arch option were chosen for the walkways.

However, there is ample room directly beneath the escalators, staircases, and walkways. With the caveat that escalators of such length need an extra midway support point, they would still have a lot of space underneath: 15-16 meters with sufficient clearance for people to stand comfortably (say, at least 2.5 meters of clearance above); with the upper approaches and the walkways, this is 60-62 meters of largely unobstructed space, for a 60*10 space that could be used in almost any way. Even in the 5-6 meters with less clearance above to the escalator, it’d be possible to use the space at least partly – for example, for sitting, or for bathrooms, the minimum clearance is reduced (I’m writing this post from my apartment, where the ceilings slope down, and the ceiling height above my couch is about 1.5 meters).

There would be two such 60*10 spaces per platform, plus two smaller spaces, near 7th and 8th Avenues, depending on exact placement of access points to the subway. This gives us twelve 60*10 spaces. I doubt that they could ever host high-end concessions, such as full-service restaurants: passengers would probably not go out of their way, to a platform that they weren’t planning on using. This means newsstands could succeed, but not much else; food would have to be shunted to the streets, and presumably restaurants would pay extra to locate right outside the compound. In lieu of concessions, those spaces could host sundry uses, including additional circulation space, information pamphlets, busker performance space, waiting areas for passengers, public art displays, and waiting areas for train crew and cleaners.