Quick Note: More on Urban Absolutism

In previous posts, I brought up the theory that American cities are run in a feudal fashion, despite the nominally democratic system, and that the failings of feudalism are leading proponents of livable streets and public transit to demand elected absolute monarchs instead. The recent 125th Street bus debacle, and the online livable streets community’s response to it, represent another example of this trend.

To recap: New York City’s Department of Transportation proposed a bundle of bus upgrades along 125th Street: dedicated bus lanes on most of the street for the use of all four bus routes running along 125th, and Select Bus Service on the M60, which connects Morningside Heights with Astoria and LaGuardia. The M60 is by a small margin the top route for boardings along 125th (not necessarily for boardings elsewhere for trips ending on 125th), but it’s third in overall ridership among the four routes. Because of its Morningside Heights bend at the west end and its LaGuardia service it’s perceived as a whiter route than the other three routes: the Bx15, connecting to Third Avenue in the Bronx; the M100, connecting to Washington Heights and Inwood; and the M101, connecting to Washington Heights at one end and going along Third and Lexington at the other. Harlem politicians were livid that DOT were only giving SBS upgrades to the whiter route. State Senator Bill Perkins opposed the plan’s ban on double parking and got the bus lanes truncated from Central and West Harlem to just Central Harlem; he and City Councillor Robert Jackson continued to oppose the plan, Jackson doing so explicitly on the grounds of privileging the M60, and DOT just dropped it.

It is not my intention here to rehash my argument for why Jackson was right and DOT should have proposed SBS upgrades for all four routes, or if it had to pick one then the M101. I have said this on Streetsblog and Second Avenue Sagas in comments. Rather, I bring this up because while many commenters said “we lost, let’s try again” or “we lost, let’s defeat Perkins and Jackson for opposing our interests,” other responded with fantasies of absolute power: fantasies of the city cutting bus routes to West Harlem to punish Jackson and Perkins, fantasies of the city making the Harlem communities beg for any further livable streets improvements (as already happens with bike lanes in East Harlem), fantasies of a Robert Moses for livable streets, fantasies of Bloomberg buying election campaigns to defeat all livable streets opponents.

An absolute ruler is not going to do anything positive. He doesn’t have to – either his rule is secure and he doesn’t need to care, or it’s not and he needs to spend his effort shoring it up with patronage and attacks on opponents. A city government strong enough to do things over the objections of black politicians who are concerned with racial inequality, or over those of pro-car NIMBYs, will also be strong enough to do things over the objections of the livable streets community. Robert Moses’s problem wasn’t just that he was pro-car; it’s that he was authoritarian and didn’t need to care too much about what people thought, so that his own biases for segregation could become city policy.

Are Express Trains Worth It?

So, you have your urban rail line. It’s mostly above ground, so constructing new express overtakes is feasible. It has decent frequency, and carries trains to destinations at a variety of distances from city center. But it’s not an overcrowded subway line that brushes up against line capacity, requiring all trains to run at the same speed. Do you run express trains?

I’m going to focus on regional rail in this post, since with two Tokyo-area exceptions, proper subways are incapable of running express trains without dedicated express tracks due to their high frequency. On a line with a train every 10 minutes it’s feasible to mix trains of different speeds with timed overtakes; on a line with a train every 2 minutes, it’s not. I’m going to use the LIRR and Caltrain as examples, and then apply the derived general principles to other cases in the US, including future regional rail schemes.

The basic tradeoff of express service is that it provides faster service to the express stations at the cost of frequency at the local ones. This can be done in two ways: expresses that stop once every few stations, and local-then-express patterns. Jarrett Walker calls this limited versus express, based on bus service patterns; with trains, both types are called express. The subway in New York, the Chuo Rapid Line, Seoul Subway Line 1, and Caltrain baby bullets are examples of the first kind; the Caltrain limited-stop trains and the peak-hour trains on some LIRR lines are examples of the second kind.

Express trains of either kind but especially the first reduce line capacity, even with very long overtake segments. If train X overtakes train L, then there needs to be an available slot ahead of train L, and after the overtake there’s a slot opening up behind L. The Chuo Rapid Line runs a mixture of local (“rapid”) and express (“special rapid”) trains for most of the day, but at rush hour, there are only local trains, peaking at 28 trains per hour; on the shoulders of rush hour, there are some express trains, with total traffic of about 20 tph. The LIRR runs 23 tph on the Main Line at the peak, so this is an issue, which the LIRR unsatisfyingly resolves by running trains one-way at rush hour. It’s less an issue on Caltrain given constructable overtake locations, but right now the overtake locations are inconvenient and the trains are pulled by diesel locomotives, increasing the stop penalty and reducing the capacity of a mixed local-express line.

The second kind of express service is bad industry practice and should not be used. It avoids the capacity problems of the first kind at low traffic levels, but at high traffic levels the speed difference is still too large. It is used when the trains are a special CBD shuttle and makes it impossible to serve passengers who are cheap to serve, i.e. those getting off short of city center. Caltrain’s limited-stop trains do this because of capacity problems during rush hour, when they need to get out of the baby bullets’ way. The LIRR does this because of a cultural belief that trains exist only to shuttle people from Long Island to Manhattan and back; due to the same belief, it runs trains one-way at rush hour rather than giving up on rush hour express runs as JR East does.

The first kind of express service may or may not be warranted. It depends on the following questions:

1. What is the line’s expected traffic level? Low traffic, up to about 4 tph for a regional line, favors an all-local configuration to prevent cutting local stations’ frequency unacceptably. Very high traffic favors all-local configuration for capacity reasons, or else investment into long overtakes or even full four-tracking. Intermediate traffic, in the 6-12 tph range, is the best zone for express trains.

2. Have local trains already been sped up by use of good industry practices? Level boarding, high-acceleration EMUs, better track maintenance allowing higher speeds between stations, good timetable adherence allowing less schedule padding, and infrastructure preventing delays on one train from cascading to others allowing even less padding can all significantly reduce the speed difference between local and express trains. In some extreme cases, a local train can end up not much slower than an express train hauled by a diesel locomotive.

3. How long is the line, and how many stations does it have? Longer lines and shorter interstations both favor express trains, all else being equal. Intercity rail, which also has higher stop penalties because of the higher line speed, deserves more than one stopping pattern even at low frequencies.

4. How big is the difference between minor and major stations? It is crucial not to confuse current ridership with ridership potential, since lines with express service often pick winners and losers, after which the better-served express stations steal riders who live closer to bypassed minor stops. This is common on Caltrain, where some but not all express stops are major job centers.

5. Can intercity trains plausible substitute for express service?

It is question 4 that makes the difference in many cases. On the LIRR, the Main Line has a clear distinction between major stops (Mineola, Hicksville) and minor ones (all the rest). The Montauk Line does not. Note the ridership levels of the stations, going eastward from Jamaica to the end of electrification:

Main Line:

Hollis: 114
Queens Village: 791
Floral Park: 1495.5
New Hyde Park: 1725.5
Merillon Avenue: 766.5
Mineola: 5174
Carle Place: 386
Westbury: 1951.5
Hicksville: 8107.5
Syosset: 2748.5
Cold Spring Harbor: 2083
Huntington: 5556.5
Bethpage: 2481.5
Farmingdale: 2312.5
Pinelawn: 25
Wyandach: 1758.5
Deer Park: 2708.5
Brentwood: 1375
Central Islip: 1787
Ronkonkoma: 8639

Montauk Line:

St. Albans: 93.5
Lynbrook: 2738
Rockville Centre: 3425
Baldwin: 3371.5
Freeport: 2514.5
Merrick: 3383.5
Bellmore: 3267.5
Wantagh: 2890.5
Seaford: 1804
Massapequa: 2959.5
Massapequa Park: 1672.5
Amityville: 1542.5
Copiague: 1430.5
Lindenhurst: 1791.5
Babylon: 3293

There are three ends of electrification: Babylon, Huntington, and Ronkonkoma. All have markedly more ridership than nearby stations, especially Ronkonkoma, though in all cases it’s an artifact of their being the ends of electrification, with many people driving in from farther east. Ronkonkoma has nothing nearby that justifies its ridership level, the highest of any suburban LIRR station; it’s a park-and-ride that has a lot of ridership because it’s the end of electrification and has express service.

In contrast, in Mineola and Hicksville, there really is a concentration of activity justifying their status. Both have trivial transit usage as job centers, but there’s enough of a core, especially around Mineola, to justify higher service, and Hicksville is also the junction of the Main Line with the Port Jefferson Branch: see the census bureau’s OnTheMap tool.

But there are no special stations on the Montauk Line. Excluding St. Albans, which is in New York itself and has to compete with cheaper and more frequent if slower bus-to-subway options, the ratio between the busiest and least busy stations is 2.4:1. A similarly flat situation occurs east of Hicksville, excluding the two end-of-electrification stations.

What this means is that the LIRR should only run local trains on the Babylon Branch and east of Hicksville, while maintaining express service on the Main Line west of Hicksville when there’s enough capacity for it. A similar analysis of other lines in the New York area should give the following answers:

Hempstead, West Hempstead, Long Beach, and Far Rockaway Branches: all local due to short length.

Port Washington Branch: probably all local due to short length, but if additional local stations are added in Queens, then some express trains to Great Neck may be warranted.

New Haven Line: very long, sharp distinction between major and minor stops all the way but especially west of Stamford, high frequency, four tracks give enough capacity for everything. The current configuration of nonstop trains to Stamford continuing as local to New Haven and local trains turning at Stamford is fine, except that the express trains should also stop at New Rochelle (a junction with the Hell Gate Line, which deserves service, but also a major stop in and of itself, with the third highest weekday ridership of Metro-North’s suburban stations) and maybe also Greenwich; HSR overtake considerations may require stopping also at Rye and Port Chester.

Harlem Line: generally favors local trains, except that White Plains is a major job center and thus a far more important stop than all others, independently of its better service. There are four tracks south of Wakefield, favoring express trains, but conversely charging subway fares and allowing free transfers to the subway would lead to a ridership spike as people switch from the overcrowded 4 and 5 trains. There’s a big dropoff in ridership north of North White Plains, so the current configuration of locals that turn at North White Plains and expresses that go nonstop south of White Plains is fine, as long as off-peak frequency is raised.

Hudson Line: favors express trains because of length and four-tracking. Although on paper there are more and less important stations, this is an artifact of service patterns. The secondary stations in Yonkers serve higher density than the busier stations in the proper suburbs, and the dense parts near Tarrytown are actually in Sleepy Hollow, about equidistant from the Tarrytown and Philipse Manor stations: see the New York Times’ population density map.

Erie Lines and West Shore Line: probably all local since the population density thins too uniformly going north, with Paterson as the major exception. There are somewhat denser anchors at the outer ends of some lines – Spring Valley and Nyack – but Harlem Line-style nonstops run against a capacity problem, coming from the fact that this part of the network is necessarily highly branched.

Rest of New Jersey Transit: the main lines (Northeast Corridor, Morristown) are very long and have some distinguished suburban job and population centers (Metropark, New Brunswick Morristown) deserving express service, but the branches (North Jersey Coast, Montclair, Gladstone) do not. However, the fare structure and off-peak frequency lead to much less ridership on the inner-urban segments in Newark, Orange, etc., than would be expected based on population density. In addition, the difference between major and minor stops is fairly small on all lines when taking electrification into account, sometimes as small as on the Babylon Branch: see ridership data per line and per station.

Although my initial decision in my regional rail plan to pair the Erie lines with the Atlantic and Babylon Branches of the LIRR was aesthetic, creating a northwest-to-southeast line, in reality the systems are fairly similar in their characteristics. More or less the same can be said about the Staten Island-Harlem system. There are no direct connections to intercity rail except at Jamaica and in the Metro-North tunnel to Grand Central, the lines pass through urban or dense-suburban areas, the interstations are fairly short, and there’s relatively little distinction between major and minor stops. (White Plains is the major exception, and Paterson is a secondary one.) This makes the Lower Manhattan-based system much more RER-like than the Penn Station-based one, which is longer-distance and practically intercity at places.

Finally, the same set of questions in the other three major Northeastern cities generally lead to the conclusion that no express trains are needed.

In Boston, there’s too little difference between major and minor stops on each line (see PDF-page 70) – somewhat more than on the Babylon Branch, but much less than on the LIRR Main Line. The most prominent major station is Salem, but the low-ridership stations farther in on the Rockport/Newburyport Line are in working-class suburbs; the ridership there is depressed because of fare and schedule issues coming from competition with buses, and good regional rail would get much more additional ridership from Lynn and Chelsea than from Salem and the suburbs farther out.

In Washington, current traffic demand is so low that express service would seriously eat away at the frequency offered to local stations. MARC and VRE ridership is so low that any analysis of travel demand has to start from geographic and demographic information rather than from preexisting ridership; the only major outlying destination on any of the lines is Baltimore, which can be connected to Washington by intercity rail, and which conversely has much less Washington-bound commuter traffic than the Washington suburbs. The closest thing to justifiable express service is that when the commuter lines closely parallel Metro, they should have wider stop spacing.

In Philadelphia, on most lines, express service eats away at frequency too much. The one exception is the PRR Main Line, with the SEPTA Main Line a possibility. Many lines have sharp differences between local and express stations: for example, Cornwells Heights on the Trenton Line is much busier than the rest. But a combination of low frequency and lack of easy overtakes (on the Trenton Line, the inner tracks should be mainly used by intercity trains, with only the occasional regional rail overtake if required) makes this not useful. The PRR Main Line actually has less difference between major and minor stops than many others, but it is longer and has short interstations and higher frequency. The SEPTA Main Line has the frequency to support multiple stopping patterns, though the population density near the minor stations is high and the problem, as in other Northeastern cities, is high fares and lack of integration with urban transit.

Large-Diameter TBMs

Deep-level subway tunnels are usually built with tunnel-boring machines (TBMs), which can dig and create their own lining even under other infrastructure, such as older intersecting tunnels. But then deep-level stations require larger caverns, which are expensive to dig from the surface. Three-quarters of the cost of Second Avenue Subway Phase 1 is the three stations. As commenters Jim and Anon256 noted a year and a half ago, to avoid this problem, such cities as Barcelona pioneered the use of large-diameter TBMs, which have enough space to accommodate tracks together with platforms by their sides. This is especially useful for construction in dense city centers, where surface disruption must be minimized and demolitions of buildings that are in the way are expensive. I claim that this is the optimal construction method for both regional rail to Lower Manhattan and the North-South Rail Link in Boston.

In Barcelona, the internal diameter of the TBM used for Line 9, 11.7 meters, is enough to have both directions of a two-track line use one tunnel. With an internal horizontal slab, trains can be stacked so that each direction gets one track and one platform at a station, which looks about 4.5 meters wide in diagrams. Between stations, there is enough space for each of the two levels to have two tracks, allowing crossovers. The only required construction outside the tunnel is access points, which can be drilled straight down for elevators or at an angle for escalators.

While the cost of Barcelona Metro Line 9 is about $170 million per kilometer, more than three times the original budget, compared with $40-60 million per kilometer for most Spanish tunneling projects, it is still much lower than the cost of comparable projects tunneling under preexisting subway systems that have stations built by blasting caverns or cut-and-cover construction. In addition, the standards are relatively easy to adapt to the standards of American mainline construction, since the Line 9 trains are powered by catenary and are only ten centimeters shorter than the LIRR’s M-7s. Mainline catenary is energized at 25 kV and requires more clearance than low-voltage rapid transit catenary, but this adds only about half a meter to the total diameter: German standards call for 27 centimeters of clearance from 25 kV.

To allow two lines to meet at cross-platform transfers, there are two possibilities, both used by narrower-diameter TBMs (or older tunneling shields). One, used by the London Underground’s tube lines, is to have two parallel circular tunnels with numerous passages drilled between them. Another, used by some subway lines in Shanghai and Tokyo as well as by the Harlem River tunnels of New York’s Lexington Avenue Line, is to overlap the two circular tunnels, using a tunneling shield with a double-O tube (DOT) design. The DOT design is more complex and would also require any access point to either obstruct the platforms or go at the platform edges, but would create a wider platform allowing easier cross-platform circulation.

In Boston, regardless of which design is used, the North-South Rail Link involves three central stations in which two tubes (one feeding the Worcester and Providence Lines, one feeding the Fairmount and Old Colony Lines) meet: South Station, Aquarium, and North Station. Each should have a cross-platform transfer, in the style of the Hong Kong MTR: at Aquarium northbound Providence and Worcester trains should face northbound Fairmount and Old Colony trains and likewise for southbound trains, whereas at South and North Stations, northbound trains should face southbound trains. This way, people transferring between two points south of the link could transfer cross-platform at South Station, and people transferring between two points north of the link could transfer cross-platform at North Station.

A large-diameter TBM has enough space not only for crossovers, but for trains to switch what levels they’re on. With a design speed of 100 km/h, a curve radius of 500 meters, and a superelevation ramp lasting 2 seconds, it takes about half a kilometer for the track on the lower level to swerve sideways so as to no longer be directly under the upper-level track, climb to the upper level while the upper-level track descends, and then swerve sideways again so that both tracks are on the correct side of the tunnel to allow a cross-platfom transfer. There is space to do this between both pairs of successive stations. The portals could be constructed where convenient on the approaches to South Station and immediately north of the Charles, and the infrastructure for pairing lines at the north end with the two tubes could be done above or below ground, based on local tradeoffs between disruption and cost.

In Lower Manhattan, the problem is capacity. The system would involve a line from Atlantic Terminal to Jersey City or Hoboken intersecting a line from Grand Central to Staten Island. There is room for only one station, and some configurations, notably any in which the New Jersey end is at Exchange Place, require a cruciform station, without cross-platform transfers. Moreover, this station is at a site with much more intensive development than Downtown Boston, and close attention must be paid to capacity. This is why I bring up DOTs in the first place: London-style passages may not allow sufficient circulation of transferring passengers. The platforms would be obstructed with many escalators between the upper and lower levels since there is no room for Hong Kong’s three-station cross-platform transfers, and peak demand for egress to both street level and intersecting subways is also likely to be very high.

The optimal solution seems to be to have no real Lower Manhattan station beyond the platforms and access points. Most ticket-vending machines should be placed at street level next to the escalator and elevator banks, and the blocks above the station should be pedestrianized to allow for access from the middle of the street, avoiding the need for a mezzanine. The width and pedestrian volume of Lower Manhattan streets are such that it would be at good human scale.

The remaining capacity issue is sufficient space for escalators. There are four tracks in total, each of which is inbound from some direction, and at the peak there could be a 12-car, 300-meter long train with 2,000 passengers every 2 minutes per track. If all passengers are discharged and the trains leave the station empty in the morning peak, then the required capacity is 240,000 people per hour. This is in fact quite unlikely, even though there is only one Lower Manhattan stop: many Staten Islanders work in Brooklyn or Midtown, people from points north of Grand Central are more likely to get off the train at Grand Central than to stay on until Lower Manhattan, and there is a substantial volume of commuters between Brooklyn and points west or north of Manhattan, who would benefit the most from through running.

Factsheets by Kone and ThyssenKrupp suggest each meter-wide escalator has a practical capacity of 6,000-7,000 passengers per hour. If we assume half of a full train capacity’s worth of passengers get off at the station, not including passengers who transfer, then we need 120,000 passengers per hour, i.e. seventeen to twenty escalators. This can be done quite easily with two parallel circular bores, at the cost of restricted capacity for connecting passengers. With a DOT design with 8-meter wide platforms, it’s still possible to have an escalator bank at each end of each platform; the large separation between the upper and lower levels, about 6 meters, allows independent escalators at the end, though not anywhere else. The widest standard escalator is a meter wide at the step and requires a 1.6-meter wide pit (see above ThyssenKrupp link as well as brochures by Kone and Otis), enough for a three-and-one or three-and-two escalator bank at each end, giving twelve peak-direction escalators. Eight additional escalator banks in a one-and-one configuration (or perhaps four in a two-and-one configuration, which is a wider platform obstruction) can be placed roughly evenly along the upper-level platform, along with elevator shafts, escalators that only connect the two platforms, and access points to intersecting subway lines.

The advantage in both New York and Boston is that there’s no need to construct a station beyond those shafts and bores. The station mezzanine in this configuration is a street, most likely Broadway in Lower Manhattan and (according to prior North-South Rail Link plans) the greenway above the Central Artery tunnel in Boston. The station retail is ordinary street retail. Fare control is roving inspectors riding the trains or patrolling the platforms. It’s still a multi-billion dollar undertaking due to all the underwater access tunnels, but the cost per kilometer could be held down to normal first-world levels even while crossing the difficult infrastructure of Lower Manhattan and Downtown Boston.

Who are the Opponents of Transportation Alternatives?

Streetsblog has traditionally lashed at multiple factions that oppose bicycle and transit infrastructure, but reserved the harshest criticism for entrenched community groups and NIMBYs, and their representatives including most of the high-profile Democratic mayoral candidates in New York. Early community board opposition to some of Janette Sadik-Khan’s bike lanes and pedestrian plazas turned into full-fledged attacks by the livable streets community on NIMBYs, of which some were justified but some were cases without any evidence of community opposition.

But now the Wall Street Journal has run an editorial video calling New York’s new bikeshare totalitarian, adding to a Front Page article from a month ago saying that bikeshare was a failure in Paris and Montreal and that Sadik-Khan’s grandfather was a Nazi. Paul Krugman chimed in with an explanation relating the opposition to upper-class politics, New York Magazine tried to explain how bikeshare goes against conservative ideology more broadly, and suddenly there’s supposed to be a partisan realignment on the matter. When I reminded Robert Cruickshank on Twitter that Charles Schumer and Anthony Weiner were against bike infrastructure, he responded, “no, that’s not driven by values or ideology but by a search for votes.”

There’s a real danger in reducing the world to a struggle between Us and Them, in which the bad aspects on the Us side show that people on the Us side need to be nudged in the right direction while the bad aspects on the Them side show that people on that side need to be defeated. People who spend too much time in national or even state-level partisan politics think in those terms even in places where they are completely inappropriate, such as local blue-city transportation matters. Streetsblog has occasionally engaged in this as well, with the factions being pro- and anti-Bloomberg: it has let the city’s DOT off the hook for the truncation of the 125th Street dedicated bus lanes, though in past years it did attack the city for not extending 1st and 2nd Avenue’s bike lanes into Harlem despite community support.

What both of those sides – Krugman and the Streetsblog crowd – miss is that there’s considerably diversity of opinion in both the Us camp and the Them camp. Although there is something like an Us camp comprising supporters of rail, urban density, and livable streets, there are still sharp internal disagreements that shouldn’t be papered over. On the Them side there isn’t even a recognizable camp: what do Michele Bachmann, Chris Christie, Andrew Cuomo, Charles Schumer, and Anthony Weiner have in common except their opposition to bikes or transit? Instead of a binary Manichean view it’s important to understand that politics, especially urban politics, has multiple factions, of which none can obtain a persistent majority, requiring some measures of negotiation and consensus.

First, the Them sides. The easiest segment to explain is right-wing populism: as a movement, it tends to be anti-urban and pro-road, even in Switzerland, whose Swiss People’s Party opposes rail investment and supports roads. The support base of right-wing populism is rarely urban, because as a movement it tends to be against what it views as cultural deviance of (mainly urban) immigrants; since transit ridership tends to be concentrated in the cities, populists have less reason to support it.

Non-populist conservatives sometimes borrow from right-wing populism and sometimes do not. Christie canceled ARC and transferred the state money for it toward roads, but he is quite influenced by populism in style even if his actual politics is mainline conservative. But the British Tories support high-speed rail, as did the Sarkozy administration. Contrary to popular belief, Thatcher never said that bus riders over the age of 30 are failures in life; the quote comes from a writer who, far from being a Thatcherist, worked for The New Statesman. However, with exceptions such as Sarkozy’s support for Arc Express, conservatives and right-wing liberals tend to be less supportive of urban transit and of taxing cars on environmental grounds. For examples, the Skyscraper Page posters believe the BC Liberal election victory is likely to make it harder to find money for SkyTrain extensions, Boris Johnson canceled proposals to extend London’s congestion charge to other parts of the city, and the Swedish right-wing parties originally opposed Stockholm’s congestion charge and eventually implemented it but with a caveat that the proceeds go to roads only.

Among centrist liberals, opinions are more split. Bloomberg is unabashedly neo-liberal; he’s also spent $2 billion of city money on a subway extension and championed congestion pricing and bike lanes. Andrew Cuomo is less explicitly neo-liberal but ran on such a platform; he’s championed the Tappan Zee Bridge replacement, opposed including transit on the new bridge, and spent money that was supposed to go to the MTA on other things. The opposition to transit and livable streets that exists in this group is less a matter of hating what cities stand for and more a matter of fiscal conservatism that views roads as normal service used by the upper middle class and transit as wasteful and serving the poor. Charles Schumer’s opposition to bike lanes in his neighborhood should be placed in the same category, as should Richard Brodsky’s claim that congestion pricing is unfair while he represented a rich Westchester district. It’s here that the faction Krugman describes belongs, but it includes board swaths of the upper middle class rather than the top 0.5% of the population that Krugman argues is pro-road because they’re chauffeured around Manhattan.

The community boards who oppose transit and livable streets, for examples in Washington Heights and Sheepshead Bay, are a more mixed bunch. I believe Weiner falls in this category too: instead of congestion pricing he proposed a commuter tax, which would not fall on his outer-urban district, a proposal that more recently the other mayoral hopefuls supported. In the forums I spent time in, mainly The Straphangers forums, the opposition seemed to be from the left and not just from the right. It’s probably best understood as a general populism as well as personal dislike for Bloomberg; while this populism may not be leftist, it is not really right-wing either, and often comes from minorities, which right-wing populists almost universally spurn. I believe it’s Cap’n Transit who noted the disconnect between the elite even in poor neighborhoods and the average residents, who rarely own cars, leading to a kind of populism in leftist areas that is not by itself really leftist.

Now, the Us side. There is, in fact, a coherent movement that calls for more investment in rail for intercity transportation, proposes local transit and bike and pedestrian projects, and supports taxes on driving when they are politically feasible. The arguments between various factions, such as more left-wing versus more right-wing transit supporters or supporters of restoration of pre-WW2 streetcars and interurbans versus supporters of more modern technologies like light rail and high-speed rail, really are internal to a movement.

However, there really are problems, coming from the cores of the movement in supporting more spending and in having leaders who are quite neo-liberal and indifferent to issues of racism and disinvestment. New York really did take its time to extend bike lanes into East Harlem despite community support; the same neighborhood is now not getting a 125th Street dedicated bus lane. While the first five bus routes to receive Select Bus Service upgrades were chosen as one per borough for trial, the newer lines so chosen, first on 34th Street and now the M60 to LaGuardia, are not very high-ridership; the M60 in particular is at least in perception the highest-income and whitest among the buses that use 125th Street while its ridership rank is third out of four routes on the corridor.

Likewise, the transit investment decisions made not only in New York but also in cities ranging from Boston and Providence to San Francisco are development-oriented and tend to serve residents of rich suburbs and inner-urban gentrification projects at the expense of high-productivity transit routes in low-income neighborhoods in between. Bloomberg spent $2 billion of city money on a subway extension, but it was the wrong one, a development-oriented project to Hudson Yards rather than an extension of Second Avenue Subway or a new subway line following Utica, which is currently in a near-tie with First and Second Avenues for highest bus ridership in the city.

While neo-liberalism as an ideology also supports efficient government and reducing red tape, the built-in bias for prestige projects makes it hard to support vanilla improvements in efficiency. This combines with a particular leftist opposition to anything that sounds like reduced spending; the fact that it’s Christie who began the wave of cancellations adds a partisan dimension. As a result, the people who are most sensitive to costs tend to be far outside political power: Stephen Smith is not a major libertarian pundit, Aaron Renn occasionally talks to city leaders but has no real power, I am a mathematician who writes about transit and urban issues. The (neo-)liberal centrists who’d be best placed to implement a program that would reduce transit construction costs are the ones with the least political incentive to do so.

That said, despite the above essentially multi-partisan and multi-factional picture, it could be that the Wall Street Journal’s video and Krugman’s response will lead to partisan realignment. High-speed rail used not to be a partisan issue either: in 2009, Newt Gingrich said he envisioned medium-speed rail together with maglev. But after Christie canceled ARC, canceling rail projects became a test of right-wing bona fides, and conversely, defending infrastructure spending became a test of left-wing bona fides even when infrastructure was a small component of the stimulus. It is possible that the American political world will soon become bipolar on matters of local transit and livable streets issues. It’s just not there now.

Comparative Subway Construction Costs, Revised

Here is a list of subway projects in the last 15-20 years, in both developed and developing countries. It’s in addition to my initial lists for developed and developing countries, but includes projects mentioned in past blog posts not on those two lists. This is still not an exhaustive list, due to some cities for which I couldn’t find any information (Moscow), cities for which the information from different sources contradicts itself (Bucharest), and cities for which I couldn’t source numbers beyond Wikipedia (Osaka). My rule is that Wikipedia is an acceptable source for construction timelines and route length but not cost.

While the list is meant to be for urban subways, urban rail projects that are predominantly elevated are also included. As far as possible I have tried using PPP dollars adjusted for inflation to give 2010 dollars (2010 and not 2013, because when I started comparing costs that’s what I used). For core developed countries, because inflation rates are similar, I use American inflation rates, using the CPI (not GDP deflator: the two measures have disagreed for a while, and the CPI points to higher inflation). For other ones, I’ve tried focusing on more recent projects, including even some that are under construction, but I use actual inflation rates.

Bear in mind the data is only as accurate as my sources for it and my PPP conversions. Errors of 10-20% in each direction are to be expected: sources disagree on conversion rates, sometimes the years of construction are not made clear so deflating to the midpoint is not reliable, etc. Even larger errors sometimes crop up, for example if old cost figures are not updated after a cost overrun.

Explicitly, the rates I use today are C$1.25 = S$1 = US$1 = 3.8 yuan = 100 yen = 800 won; £1 = $1.50; €1 = $1.25; CHF1 = $1.65.

Singapore Thomson MRT Line: not yet under construction, expected to open 2019-21, S$18 billion for 30 km. This is $600 million/km, all underground. Included only as a lower bound of costs; costs can rise beyond budget but rarely come significantly under it.

Hong Kong Sha Tin to Central Link: a 1-km segment underground (not underwater) is £270 million, under construction with opening expected in 2018. After converting to PPP using Hong Kong’s conversion rate this is $586 million/km.

Singapore Downtown MRT Line: under construction since around 2008, to be completed in 2017; S$20.7 billion for 42 km: $493 million/km. This line is fully underground. This represents a 70% cost overrun already, announced after I previously reported the original budget of S$12 billion.

Budapest Metro Line 4: under construction since 2006, completion expected in 2014, 400 billion forint for 7.4 km. This is $358 million per km. The line is fully underground.

Fukuoka, Nanakuma Line extension to Hakata: construction expected to begin 2014 with line opening expected in 2020, ¥45 billion for 1.4 km: $321 million/km. I do not know for certain that the extension is fully underground, but this is likely, as the preexisting line is underground and the extension follows busy CBD streets.

Cairo Metro Line 3, Phase 1: opened 2012 with construction since 2006, LE4.2 billion for 4.3 km. This is $310 million/km. The phase is fully underground.

Kawasaki Subway: under construction, opening expected in 2018, ¥433.6 billion for 16.7 km: $260 million/km. The line is fully underground. Update: people in comments explain that the line was actually canceled; the link in this paragraph is just a plan.

Stockholm City Line: to open in 2017, 16.8 billion kronor (2007 prices) for 6 km of tunnel and 1.4 km of bridge: $259 million/km.

Sao Paulo Metro Line 6: construction due to begin in 2014; 7.8 billion reais for 15.9 km: $250 million/km. The line is 84% underground.

Sao Paulo Metro Line 4: construction began in 2004, first phase opened in 2010, completion expected in 2014; 5.6 billion reais for 12.8 km: $223 million/km. The line is fully underground.

Dnipropetrovsk Metro extension: under construction since about 2008, opening expected in 2015, €367 million for 4 km. After PPP conversion this is $214 million/km. It appears to be fully underground.

Malmö City Tunnel: built 2005-10, 9 billion kronor for 4.65 km: $212 million/km. This is a fully underground project.

Bangalore Metro Phase 2: to be opened by 2017, 264 billion rupees for 72.1 km. This is $164 million/km. I do not know what proportion of the project is underground; it does not seem to be large, as the extension of the phase 1 lines are all outbound, and only line 4 seems to have significant tunneling, about 14 km by pure Wikipedia eyeballing.

San Juan Tren Urbano: built 1996-2004, $2.28 billion (2001 figures, see PDF-p. 145) for 17.2 km: $163 million/km. The line is only 7.5% underground by direct inspection on Google Earth.

Lucern Zentralbahn: built 2008-13, CHF250 million for 1.32 km of tunnel: $151 million/km.

Hangzhou: I can’t find any ex post numbers, but in both 2005 and this year (Chinese) officials pegged the cost of future construction as ¥550 million/km: $145 million/km.

Sofia Metro Line 2: built 2008-12, €952 million for 17 km. After PPP conversion, this is $148 million/km. The line appears to be almost fully underground: the numbers here do not fully add up but point to 1.3-2.9 km above ground (7.6-13% of total line length) in one segment while Wikipedia’s line map shows only that segment with above-ground segments.

Thessaloniki: I can’t find any ex post numbers, but in 2005 the budget for the first phase, under construction to be opened in 2016, was €798 million for 9.6 km: $104 million/km. The second phase received bids last year and is expected to open in 2017, with an estimated cost of €518 million for 4.78 km: $135 million/km. Both phases are fully underground.

Vancouver Evergreen Line: under construction since 2012, completion expected 2016; C$1.4 billion for 11 km: $103 million/km. Only 2 km of the system, 18%, is underground, but Vancouver seems to have an unusually low underground construction cost premium.

Dubai Metro (lines 1 and 2): built 2005-11, Dh28 billion ($6.9 billion in PPP2010US$) for 75 km: $92 million/km. Only 13 km of the system, 17%, is underground.

Mexico City Metro Line 12: built 2007-2012, $1.8 billion for 26.4 km. After PPP conversion, this is $90 million/km. From a Google Earth overlay map, this line is 49% underground.

Seoul Sin-Bundang Line: built 2005-11, 1,169 billion won for about 18 km (sources disagree on whether it’s 17.3 or 18.5): $87 million/km. The line is 100% underground according to YouTube videos.

Bangalore Metro, Phase 1: built 2006-11, 8,158 crore rupees for 42.3 km: $93 million/km. Only 8.82 km, or 21% of the project, is underground. See above for Indian construction costs in a heavier-tunneling setting.

Helsinki Westmetro: under construction since 2009 with completion expected in 2015, €714 million for 13.5 kilometers: $66 million/km. The line is fully underground.

Seoul Subway Line 9: opened 2009, 900 billion won for 27 km: $43 million/km. The line is almost fully underground by direct inspection on Google Maps.

Barcelona Sants-La Sagrera tunnel: built 2008-11, €179.3 million for 5.8 km: $39 million/km. This project is intercity but fully underground.

Just from eyeballing the data, spliced together with the two older lists, the biggest correlation of each country’s construction costs is with the construction costs of other lines in the same country. When there is more than one project listed separately in a city – e.g. Seoul, Singapore, Sao Paulo – the projects have similar costs. This persists across different cities in the same country, judging by the similarity between Bangalore Metro’s Phase 2 cost and the Delhi Metro’s cost from a previous list and by the similarity between Hangzhou and Beijing’s costs.

Quick Note: Why the Focus on Penn Station?

Penn Station is in the news again: the Municipal Art Society ran a public competition for a rebuilt station house, involving proposals by four different architectural firms. This does not include any track-level improvements at all: only the concourses and above-ground infrastructure are to be rebuilt, at a cost of $9.5 billion according to one of the four firms. The quotes from the architects and other backers of rebuilding use language like “great train station” and “gateway to the city,” and this is where the subtle hate of the city’s actual residents lies: why the focus on Penn Station? Why not a subway station?

The headline figure for the ridership at Penn Station is 600,000-650,000 a day, but this is a wild exaggeration. First, this includes both entries and exits, so the real number is half that. Second, about half of the number comes from subway riders, who these discussions always ignore. And third, there is a large number of passengers transferring between commuter rail and the subway who are doubled-counted; at subway stations, passengers transferring between lines are not even single-counted, since the subway counts entries at the turnstiles. Taking an average of boardings and alightings when both numbers are given or just boardings otherwise, Penn Station has 100,000 weekday LIRR riders, 80,000 weekday New Jersey Transit riders, and 170,000 weekday subway riders between the two stations. However, people transferring between the subway and commuter rail are double-counted.

In contrast, not counting any connecting passengers, there are 195,000 weekday Times Square subway riders. Without detailed data about transfer volumes at each station we can’t compare the two, but since the proposals for a better Penn Station focus only on the mainline station, the number of passengers served is certainly less than that of Times Square passengers.

Indeed, every single problem that the architects are trying to fix with Penn Station exists at Times Square. Times Square has low ceilings. The corridors between different lines and between the platforms and the exits are as labyrinthine as at Penn Station. In my experience rush hour passenger crowding levels within the station itself are comparable. Most platforms are wider, but nobody is proposing to widen platforms at Penn Station, and the 42nd Street Shuttle platforms are narrow and curvy and have been this way since 1918. The tickets are all integrated because the trains are all run by one operator, but again nobody who proposes to replace Penn Station is talking about the separate LIRR, NJ Transit, and Amtrak fiefs.

There are some legitimate changes that could be done if Penn Station is knocked down and rebuilt: instead of a hack involving paving over platforms to increase their width, the platform level could be rebuilt, two tracks at a time, with six approach tracks in each direction each splitting into two platform tracks, giving twelve tracks on six platforms; the train box appears about 140 meters wide, enough for 15-meter-wide platforms (compare 10 meters on the Chuo Line platform at Tokyo Station, where 28 trains per hour turn on two tracks).

However, the technical issues here are a lot less important than the fact that city leaders, architects, and even transit commentators assume that it is more important for New York to have a great train station used by 200,000 suburban commuters than for it to have a great subway station used by (at least) 200,000 city residents. It speaks to the utter hatred most city leaders have of the people who live in what they consider their fief or perhaps their playground. For most people in the city, there are more important transportation facilities, and even on a metro area level Penn Station isn’t unusually important.

This leaves the argument that Penn Station is a gateway to the city. But if the point is to impress a few thousand tourists, why not spend the same money on improving tourist amenities at Times Square, building more hotels? Or maybe building free housing for tens of thousands of homeless people (both the ones at Penn Station and the ones in the rest of the city) so that they stop being homeless and disturbing the rest of the population? If the point is to have great art, why not spend the money on employing artists to produce more work or to improve the aesthetics of the city’s ordinary spaces?

Of course, none of those options involves city leaders getting together and building important edifices with plaques with their names on them. So at the end the idea is to tax actual city inhabitants $10 billion to build a monument to the vision of city leaders. Large corporations pay their executives hundreds of millions a year in stock options and bonuses; governments cannot pay top political power brokers this way, so instead they spend large quantities of money on monuments that glorify them.

Can HSR Connect Hartford and Providence?

If new high-speed rail construction has to largely follow Interstate corridors, then a new line from New Haven to Boston can serve either Providence on I-95 or Hartford on I-91/84/90, but not both. However, there’s still the possibility of building a completely greenfield alignment between Hartford and Providence; the FRA is investigating this as option 13 of NEC Future and Amtrak is proposing this in the latest update of its Vision. Since the terrain is hillier than on the coast, it requires some investigation as to whether it’s possible to connect Hartford and Providence without excessive tunneling. The answer turns out to be yes, but only at the cost of slowdowns both north and south of Hartford that impose real costs relative to following I-95: construction is likely to be more expensive and travel time including a Hartford stop is 9 minutes longer.

I believe the alignments depicted in this map are near-optimal for New Haven-Providence via I-95 and via Hartford. The New Haven-Hartford alignment is similar to that of Penn Design with two major differences: Penn Design diverges to cut off some curves near Hartford, but to guarantee sufficient curve radius it has to slice a significant chunk of downtown New Britain; and Penn Design also straightens the route in New Haven with a tunnel, which is unnecessary as the time savings do not justify the expense. Amtrak prefers getting to Hartford from Danbury, but to get there from New Rochelle requires long suburban tunnels, which my alignment avoids. I have not seen a detailed Hartford-Providence alignment, and I drew a line based on Google Earth elevation with an eye toward avoiding tunneling, which means there may be some further optimization, for example a rigorous cost-benefit analysis of viaducts versus curve avoidance.

The Hartford-Providence greenfield route has no tunnels except in Providence itself, where the line tunnels under Olneyville for about 2 kilometers. In Connecticut the route has many viaducts, but does not need to tunnel through the inland hills. Rather than giving detailed cost estimates, which are possible but not with sufficient reliability or precision, I am going to qualitatively describe construction challenges for each route and then the differences in travel time, which favor not serving Hartford. The final decision should boil down to the question, what cost is it acceptable to impose on New York-Boston travelers to allow for service to Hartford?

Tunneling

The I-95 route is zero-tunnel. The Hartford route has no tunnels in Connecticut, but requires a tunnel of 1.5-2 km in Providence. There exists an old railroad alignment going around the river and connecting Providence to the west without a tunnel, but the right of way was given away and to restore it would require some urban building demolitions as well as configuring a flying junction under Route 6 while also slowing down trains further.

River crossings

The I-95 route has significant challenges in river crossings, since it is close to the coast. Three difficult crossings are needed, of the Quinnipiac, the Connecticut, and the Thames. The Connecticut only requires a span parallel to I-95. The Quinnipiac requires a new span parallel to US 1 and I-95 and a new approach from Union Station; there is space for this approach, and the curve radius can be kept to at least 500 meters, but it requires work on active track. The Quinnipiac span can be avoided by using the existing route around the bay, which crosses the river at a much narrower point, but this adds several slow kilometers to the route. Recent construction costs for parallel bridges are $125 million for the four-lane US 1 bridge and $554 million for a signature ten-lane I-95 span; I believe the lower cost is more indicative of the infrastructure required for a two-track rail bridge.

The Thames is the hardest, since the route of I-95 and the terrain make it hard to cross anywhere except downtown New London, a constrained urban location. There is just enough space for a station between the decks, and the alignment may impose further constraints on curve radius. There is more space north of both decks, or alternatively Connecticut could build a third I-95 deck and repurpose one of the existing decks for rail.

The Hartford route has one significant water challenge: crossing the Connecticut in downtown Hartford. There is an existing bridge, but it is single-track and would require a completely new span to be used by high-speed rail. It is also used by freight, but only by a short branch line that could be bought out.

The Hartford route also needs to cross the Scituate Reservoir, adding about 3 kilometers of viaduct. However, there is a choice of where exactly to cross it and not much development on its banks, making construction easier than on I-95 or across the Connecticut in Hartford.

Terrain and viaducts

I-95 is substantially flatter than the inland route. Only two short segments require significant overland viaducts and earthworks: the transition in southern Rhode Island from I-95 to the Shore Line, and the curve west of New London cutting off curvier parts of the Interstate. The transition is in total 16 km long but only about the western 10 km of it are difficult (of which about half require viaducts and half can be done cut-and-fill), and west of New London there are 6 difficult km requiring a viaduct north of the Niantic.

In contrast, the inland route needs to be on viaduct for a significant portion of the Hartford-Providence section. Of particular note is the Quinebaug River valley, about 13 kilometers of route of which most requires extensive grading and viaducts, as well as some takings in the built-up areas of the towns of Brooklyn and Killingly. The Willimantic River-Mansfield Hollow Lake-Natchaug River complex adds another 16 kilometers, some hard and some less hard; the Willimantic itself is in a deep valley requiring a tall viaduct of about 3 km, and the total viaduct length required appears to be about 8 km. The following 12 km, on the crest heading to I-84, require some earthworks, but probably no significant viaducts.

Urban construction challenges

I-95 has an existing route into Providence. Some curve modifications from East Greenwich northward are helpful for keeping speeds up, but the grade-separated route already exists. The main challenge is fitting regional trains if Rhode Island desires to run them: the right-of-way has room for four tracks but only if track centers are narrowed so much as to preclude tilting, reducing cant deficiency to about 125 mm. At the New Haven end, the main challenge is crossing the Quinnipiac, but once the tracks are east of the harbor, suburban development intensity drops rapidly, requiring only occasional grade separations with roads crossing I-95. Conversely, if intercity trains are all routed through Hartford then no new construction is required for any Rhode Island regional rail.

The major problem then is New London. The entire complex of crossing the city and the Thames is the biggest difficulty in the route, as outlined above in the water crossing section. In addition to geometric difficulties, there are also noise abatement issues, since the track geometry still allows very high speeds (the curve drawn above just west of New London looks like it can be eased to about 3 km, allowing 310 km/h). This is what favors putting the tracks between the two I-95 bridge decks instead of to the north.

The inland route has far greater difficulties. First, it needs to carve a partially new route into Providence, hence the Olneyville tunnel; however, it also leaves the Providence built-up area much faster, within about 6 kilometers vs. 24 for the Shore Line. In New Haven and Hartford it can for the most part transition between legacy rail routes and expressway corridors, but a substantial portion of the route is in the suburbs of those two cities, which requires more grade separations and makes curve modification harder. There are also noise abatement issues, though Shinkansen trains skip some urban stations at 300 km/h, so those issues are more about cost than about speed limits.

There are several alignment choices north of New Haven. The one I used on the map follows the Providence and Worcester’s Middletown Branch right-of-way and thence I-91, but it is equally feasible to take a more westerly route via the Amtrak line transitioning either to I-91 or Route 15; both options involve grade crossings and extensive suburban construction. In all cases, the trains are almost continuously in built-up area from New Haven until 19 kilometers east of Hartford. Grade separations have the full cost of urban or dense-suburban construction, and moreover, the transition to I-384 east of Hartford requires some additional takings.

Total new construction

This is the primary advantage of I-95, cost-wise: the track already exists from Kingston north and requires only minor facelifts. The New Haven-Kingston construction is just 124 km, whereas between the splits with the legacy Northeast Corridor in New Haven and Providence the Hartford route is 167 km.

Curves

With this in mind, nearly the entire I-95/Shore Line segment between East Greenwich and East Haven can be eased to a curve radius of 4 km. New London, where noise abatement prevents running at full speed anyway, can accommodate slightly lower radius, about 3 km on the western approach. At the New Haven end, the transition to the Quinnipiac bridge right next to the station has radius 500 meters, but the speed restriction is minor since it is so close to the station.

Hartford-Providence can also be eased to quite high curve radius. In Rhode Island, once out of the Providence built-up area, the tracks can maintain a 4-km standard, and until the transition to I-384, the worst radius is 3.1 km around Mansfield. However, from I-384 west, things become far worse: the transition to the east has a radius of 1.2 km and seems impossible to increase further, the transition to the west has a radius of at most about 1 km, and the curve west of the Connecticut bridge is 500 m and is slightly farther away from Hartford than the Quinnipiac bridge curve is from New Haven.

It is south of Hartford that things deteriorate. The worst curves on the legacy lines are in Meriden and can be bypassed, but there is a 1.3-km curve in New Britain, on an S with a 2.3-km curve just south in Kensington that makes it unfixable. At the New Haven end there’s a curve on the legacy line, bypassed on I-95 by the Quinnipiac bridge, with radius about 450 m about 2.5 km out of the station.

Overall travel time

The explicit assumptions on trains are aggressive, based as always on the need to keep speeds up in big cities and on the only partially fixable New York-New Haven segment. Trains accelerate like the N700-I (26.74 kW/t, more than any high-speed train that currently exists except the Talgo AVRIL), cant deficiency is 175 mm as on the E5/E6 and on the AVRIL, cant is 200 mm as on the Tokaido Shinkansen, and initial acceleration is 0.89 m/s^2 as on the N700-I. With these performance specs, the minimum curve radius required for a full speed of 360 km/h is 4 kilometers; the Tohoku Shinkansen has such radius and JR East intended to run trains on it at 360 km/h before deciding to reduce speeds to 320 for reasons that are not track geometry.

For simplicity of computation I’m going to ignore grades. Since the I-95 route is flat, with very few grades higher than about 1%, this is justifiable there; it’s a little less justifiable through Hartford because a few segments have 3% grades, but they are also quite limited.

Without any schedule padding, we can set the following speed zones for I-95, measuring from 0 km point in Providence and going southbound:

0-0.6 km: 90 km/h (curve around Providence Station)
0.6-4.5 km: 120 km/h (two 450-m curves)
4.5-7.5 km: 180 km/h (Mashapaug Pond curve is too close to 120 km/h to matter, curve into Cranston is about 1 km)
7.5-17 km: 250 km/h (no curves, trains can achieve 270 in between curves but this would only save 5 seconds)
17-22 km: 220 km/h (curves have radius about 1.4 km and the controlling curve at km-point 17 can be eased a bit)
22-92 km: 360 km/h (full speed to New London)
92-103 km: 310 km/h (speed restriction in New London and the curve north of the Niantic River)
103-162 km: 360 km/h (full speed to East Haven)
162-167 km: 250 km/h (curve around an East Haven hill, though trains can barely accelerate fast enough for it to matter going eastbound)
167-168 km: 100 km/h (New Haven approach)

The time taken to transition between speed zones is the average of acceleration and deceleration time penalty. This gives a technical travel time of 33:40 for nonstop trains. If trains have a top speed of 300 km/h, this raises the technical travel time to 37:28.

Now, let us set speed zones for the Hartford route:

0-0.6 km: 90 km/h (curve around Providence Station)
0.6-4.5 km: 180 km/h (curve north of Hartford)
4.5-6.5 km: 200 km/h (curve into Johnston)
6.5-10 km: 240 km/h (curve west of I-295)
10-57 km: 360 km/h (full speed to the Hampton-Mansfield area)
57-86 km: 310 km/h (Hampton and Mansfield impose a 310 km/h restriction to km-point 67, and trains going eastbound can’t accelerate to 360 before they have to slow down again anyway)
86-88 km: 220 km/h (gentler curve in the transition to I-384)
88-101 km: 200 km/h (transition curve to I-384, further curves on I-384 making speedup between transition curves pointless)
101-103 km: 160 km/h (transition curve)
103-109 km: 200 km/h (minor opportunity to make up time, saves 20 seconds over 160 km/h)
109-110 km: 130 km/h (curve on eastern approach to bridge)
110-112 km: 110 km/h (curve on western approach)
Hartford Station: all trains stop since curves limit time savings from not stopping, as at New Haven and Providence
112-127 km: 250 km/h (New Britain curve, speed increase to 270 km/h in between is possible but saves only about 8 seconds)
127-153.5 km: 270 km/h (Kensington and Berlin curves preclude higher speed)
153.5-155 km: 210 km/h (S-curve precludes easy straightening, and significant speed boost requires significant residential takings)
155-169 km: 250 km/h (this requires straightening the kink around and north of the I-91 underpass, otherwise 210 km/h to km-point 162, 160 km/h to km-point 164, and 200 km/h farther south)
169-172 km: 120 km/h (New Haven approach, legacy line curve)

The travel time is 25:30 for nonstop trains from Providence to Hartford and 16:10 from Hartford to New Haven. With a minute of dwell time at Hartford, this is exactly 9 minutes longer than I-95.

Compatibility with other plans

Although I-95 requires less construction overall than Hartford and the construction difficulties are about comparable, Hartford is more compatible with other intercity rail plans for New England, which reduces the advantage of I-95. Under an I-95 option, it is still useful to serve Hartford (and Springfield), which means the Amtrak Shuttle line needs to be electrified, double-tracked, and partially curve-modified anyway. Under the Hartford option this is not required except to provide regional service to Wallingford and Meriden, so the bypassed parts of the legacy line could be built to lower standards.

That said, 60 km of 160-200 km/h electrified track is still a lot cheaper than 60 km of 250-270 km/h track, which means that this reduces the cost advantage of I-95 but does not eliminate it. Of course 60 km of 250-270 km/h track is cheaper than 60 km of 360 km/h track, but I-95 still involves much less overall greenfield track construction.Hartford is also more compatible with any plans Rhode Island might make for southward commuter rail service. The current plans are too low-ridership to bother accommodating, but future plans might involve higher service levels.

Conversely, I-95 is useful for Shore Line East service, since regional trains could use the Quinnipiac bridge as a shortcut. The tracks cross in East Haven and a track connection could be built; it is likely that there will always be enough capacity for 5 km of track-sharing between intercity and regional trains. I-95 is also useful for the New London connection in case anyone wants to build a New London-Norwich regional train serving Mohegan Sun on the way.

Phasing

Neither route is particularly expensive by the standards of what both Penn Design and Amtrak think are appropriate budgets. At French construction costs, 124 km of high-speed track with no tunnels, few viaducts, and a mostly preexisting Interstate right-of-way should be about $2.5 billion. Likewise, the cost of 167 km with only 2 km of tunnel and a fair number of viaducts should be less than $4 billion, possibly down to $3.5 billion.

However, in case there’s only enough money for part of the route, construction has to be phased. The Hartford route has no track connections to usable passenger railroads between Hartford and Providence, so the only useful partial construction there is the entire Hartford-Providence segment at once plus electrification of New Haven-Hartford(-Springfield). The I-95 route comes sufficiently close to the legacy track in East Haven and Old Saybrook, giving three segments each of which can be built separately: across the Quinnipiac, from East Haven to Old Saybrook, and from Old Saybrook to Kingston.

Station-skipping decisions

Every possible train station on a route deserves an answer to two questions: what is the time advantage gained by skipping it?, and who is served by it?. Stations very close to urban terminals, for example Back Bay, have a very low stop penalty because of low approach speeds, but don’t add much service since people can just ride to the urban terminal. Suburban stations such as Route 128 and even Stamford given necessary track upgrades impose high enough a cost that they should also be skipped by express trains even if there’s a fair number of people who’d use them on the local trains.

Between New York and Boston, there are three stations where the answers to both questions favor express stops: New Haven, Hartford, Providence. With New Haven and Providence, the time cost of serving them is so low given urban curves that the only way to skip them at speed is to build new urban tunnels, which cost a lot of money relative to how much time they save. With Hartford, the situation is the same if all trains go via the inland route that serves it.

However, on some level, the time cost of serving Hartford is 9 minutes, compared with about 2 for Providence. But this is not really comparable, so we can’t just say “9 minutes is too much,” as it would be if a station on a running line imposed a 9-minute stop penalty. If we skip an intermediate station that imposes a time penalty of 4 minutes, the express trains gain 4 minutes but there are still local trains serving it. In contrast, if we go via I-95 we save 9 minutes but have no way of serving Hartford on local trains; trains can branch off north of New Haven and serve Hartford and Springfield at lower speed, but this only connects Hartford to New Haven and points south rather than to Providence or Boston. So we lose something more fundamental than stopping train frequency.

So it’s not enough to say that Hartford should be skipped because it saves the trains 9 minutes. That cost-benefit calculation depends on how important serving Hartford is to people. It’s up to the potential users of Northeast Corridor HSR and the politicians providing the funding to decide whether it’s worth it to connect Hartford with Providence and Boston.

The Different Cities We Inhabit

I do not know many people who live in Boston proper. I know about a hundred who live in the Greater Boston area, but only a small minority lives in the city proper, as is of course true in general. I know many people who live in Boston suburbs or in secondary cities like Worcester, but the largest concentration lives in the urban parts north of the river: Cambridge, Somerville, and Watertown. This is true even if I exclude everyone with Harvard or MIT affiliation. In the geek community, Boston proper is where Chinatown is and where the train station is; the social centers are around MIT and Harvard, the jobs seem to be centered in Cambridge as well, and Brandeis graduates often gravitate toward Cambridge and Somerville.

What this means is that I don’t know what people in Greater Boston think about things very reliably. I know the attitudes in Greater Cambridge, or at least the part of Greater Cambridge that goes to conventions. Just as the Providence I inhabited was really a Greater East Side, one in which more people know more residents of Back Bay than of Olneyville, the Greater Boston I inhabited is a specific subculture that’s very active in New England, with specific attitudes that aren’t found elsewhere. For example, support for public transit is quite high, while at the same time enough events don’t and can’t take place in the urban core that people still figure cars are needed, leading to a culture of carpooling.

I keep being reminded by this every time I read pieces by Aaron Renn about attitudes in a city. His latest piece about gentrification is a more subdued example since he talks mostly about the actual effects of gentrification, but the point about people’s attitudes toward it is still there. An earlier piece about Rhode Island mindsets is more indicative. To Aaron, people in and around Providence identify with the state or with their local town rather than with the region. The people I have met are not like that, and often live in southern Massachusetts while still identifying with Providence somehow. For example, one of the Waterfire performers performs in Plymouth and Providence (and Providence is the bigger draw) and lives just outside the census-defined Providence metro area. For another example, I know a recent Brown graduate who is from a Providence suburb of Massachusetts who identifies with Rhode Island and southern Massachusetts, having gone to another college in Providence and to grad school at Brown.

The upshot of this is that it is extremely hard to make any generalization about a city from our own social circles. I live in social circles that are well within Richard Florida’s creative class, but aren’t really what urban leaders seem to care about. I went to a Providence event called Geeking Out once, and it turned out to be about subsidizing smartphone app developers. It’s clearly geeky; it also has no overlap with the geeks I know who teach children how to build robots, or go to fandom conventions. When we talk about cities and urban politics, we never say things like “the city needs to attract more talent.”

But the same difficulty of generalization of course affects the elites as well as people who perceive themselves as normal. If Aaron’s experience talking to urban development leaders is indicative of what they too think, then their social circles also consist mainly of other urban development leaders and their immediate families. There is nothing inherently wrong with that, but there is a real danger of overgeneralizing from an unrepresentative social network. Aaron himself doesn’t do this, but people in positions of power do. The New England I inhabited was a bubble in which downtown Boston didn’t really exist; the Providence that the power brokers seem to inhabit is one in which it is more important to improve transit access to the Jewelry District than to South Providence and Olneyville. I say this as someone who in a year in Providence visited Olneyville once and South Providence never, but because I never saw myself as representative of transit riders, I formed opinions based on where preexisting ridership is and where usable rail infrastructure is and not on gut feeling about where service should go. We all have subcultures, but some subcultures think of themselves as more normal than others, and a few think of themselves as not subcultures at all but as representative of everyone.

Infrastructure and Democracy

Two stories, one recent and one older, have made me think about the undemocratic way the US builds infrastructure. The older story is California HSR’s cost overrun coming from scope creep; the biggest overruns were in the Bay Area, where power brokers from different agencies wanted separate territory at stations, leading to additional tunnels and viaducts. The newer one is Long Island’s reaction to the MTA’s developing proposals to add Metro-North service to Penn Station, sharing the East River Tunnels with the LIRR and Amtrak; the reaction is negative on misinformed grounds, but the misinformation often comes from official sources.

In both cases, there’s a democratic deficit in US local government that’s in play. Swiss infrastructure projects require a referendum, and involve detailed benefits announced to the public. In Lucern, a recent urban tunnel was sold to the public on the grounds that it would enable certain clockface frequencies toward the south and southeast, such as a train every 15 minutes to Hergiswil and an hourly express train to Engelberg; the full cost was included in the referendum. Even much larger projects, such as the Gotthard Base Tunnel, are funded by referendum. Nothing of that sort happens in the US, even when there are referendums on infrastructure.

I’ve begun to believe that California’s original sin with its HSR project is that it refused to do the same. Prop 1A was a referendum for what was billed as one third of the cost, $10 billion. In reality it was $9 billion and $1 billion in extra funds for connecting local transit; in year of expenditure dollars the estimated budget then was $43 billion, so barely a fifth of the project’s cost was voted on. The HSR Authority planned on getting the rest of the money from federal funding and private-sector funding. Prop 1A even required a 1:1 match from an external source, so confident the Authority was that it would get extra money.

In reality, at the time the proposition was approved to go to ballot, the financial crisis hadn’t happened yet, and there was no talk of a large fiscal stimulus. Although the stimulus bill gave California $3 billion, in 2008 the HSR Authority couldn’t know this source of money would be available, and yet it assumed it would get $17-19 billion in federal funding. Likewise, no private investor was identified back then, and promises of foreign funding have been inconclusive so far and again only come years after the referendum. Put another way, Californians voted without any information about where 79% of the budget for HSR would come from. The state is now scrambling for extra funding sources, such as cap-and-trade revenues. Since there is no real dividing line between on-budget and off-budget when 79% of the budget is undetermined, costs could rise without controls. An agency that had lined $43 billion in prior funding via referendum would be too embarrassed by any cost increase requiring it to ask for more money from any source; a large cost increase could make the difference between project and no project.

In the Long Island case, there was of course no referendum – East Side Access and Metro-North’s Penn Station Access were both decided by the commuter rail agencies and the state legislature. However, even subject to the legislative decisions, there has been very little transparency about what’s going on. The MTA has provided scant details about service planning for after East Side Access opens: total tph counts for each terminal, but nothing about off-peak frequencies, nothing about which LIRR lines would have service to which terminal, and nothing about the frequency of each individual LIRR line. A major change, the end of through-service from east of Jamaica to Flatbush Avenue, is not explicitly mentioned; one has to read between the lines to see that there’s no service planned to Flatbush Avenue, which is planned to be connected to Jamaica by shuttle service (and the shuttle service is still not going to offer urban rail frequencies or fare integration with buses and the subway).

In this climate, it’s easy for people to disbelieve that the agencies involved know what they’re doing, even when they are. Penn Station Access is unpopular among Long Island politicians, who view the East River Tunnels as their turf and do not want to share with Metro-North. The MTA and New Jersey Transit keep saying that Penn Station is at capacity without further explanation, and the MTA says it will add Metro-North trains to Penn; is it any wonder that state legislators see those two statements and, in the context of past cost overruns, oppose Penn Station Access?

When there is democracy – by which I mean not just periodic elections offering two parties to choose from, but a referendum process, transparency, and community consultations – people have an incentive to be informed. It’s possible to sway many people in one’s community and have a positive effect on local state services. Local politicians who are informed on the subject will be able to lead spending and planning efforts and can count on the support of informed voters. In contrast, when there is democratic deficit, being informed is far less useful, because decisions are made independently of what people think unless they are power brokers, or perhaps wealthy, power-brokering communities.

Alexis de Tocqueville observed as much when he visited the US two hundred years ago, when it was already far more democratic, for white men, than any European country: American farmers were more informed about politics than their European counterparts. Today, everyone in the first world has democracy and universal franchise, with a few exceptional countries that are worse-run than people give them credit for. But on the local level, some countries have done much more and get rewarded with a system of accountability to the voters, leading to better governance. The US is trading on an unreformed political system, in which the check on local officials’ power comes from neighboring fiefdoms rather than from the people.

The feudal character of local government in the US is leading to the usual exasperation with the system. But instead of turning toward democracy, transit supporters cheer as governments turn toward absolutism, increasing the power of the state at the expense of other stakeholders. California is reforming its environmental protection laws in response to abuse of the system by powerful communities; in reality, one of the state legislators involved in the effort recently left politics to work for Chevron. A reformer at Cornell recently proposed to improve transportation governance by “[putting] a bipartisan committee in a locked room.” Thomas Friedman cheers Chinese megaprojects as a way to achieve progress and sustainability; he says nothing about the more cost-effective projects done democratically in Europe, even though they involve some equally impressive edifices like the Alpine base tunnels. Throughout the transit activist community, including nearly every blogger and commenter but also the main activists on the ground, there’s a tendency to view any community opposition to a project as NIMBYism and to ask for changes that make it easier for the government to get its projects done, as in the Robert Moses era. Social democrats and neo-liberals are equally complicit in the march for not just centralization, which can be done with democratic checks, but also concentration of power in the hands of state officials.

Good infrastructure does not come from autocrats. Nothing comes from autocrats except more wealth and power for the autocrats, which may or may not involve infrastructure that is useful to the public. Undemocratic systems lead to a feedback loop in which the people have no incentive to be informed while the power brokers have no incentive to make sure anyone is informed, and this way it’s easy to spend $8 billion on a train station and approach tracks, without knowing or caring how many orders of magnitude this is more expensive than the average first-world rail tunnel. A good transit advocate has to advocate for more democracy, transparency, and simplicity in government operations, because decisions made behind closed doors are almost invariably made for the benefit of the elite that’s on the right side of those doors.

New Hudson Tunnels

Traditionally, proposals for new mainline rail tunnels across the Hudson enter Penn Station’s southern tracks, which are used by New Jersey Transit. This includes the ARC incarnations that connected to Penn Station’s preexisting tracks, Amtrak’s Gateway, and the Institute for Rational Urban Mobility’s through-running counter plan to ARC. To my knowledge, ARC-North proposals, entering the northern tracks used by the LIRR, have not been investigated. I am not ready to sign off on ARC-North yet and in fact called it a troll proposal, but I believe it deserves more study and is probably superior to southern proposals except at high levels of investment.

While this in principle only concerns the configuration of the tunnels across the Hudson, it has implications about the configuration of Penn itself and the service plan supported by the infrastructure. The one element of the various Penn Station redesign ideas that isn’t relevant is the look of the station itself; this is also the least important element for passenger throughput. Penn Station’s failure to look like a cathedral is a lesser problem in a city that is full of multi-billion dollar starchitect-designed ugliness.

First, some principles for tunnel design. A good infrastructure proposal should have the following elements:

1. A through-running service plan. It matters which line at the New Jersey (or Hudson Line) end connects to which line at the Long Island or Connecticut end, since this influences the interlockings. It also matters where people are expected to transfer, since this influences platform crowding; ideally, transfers should be handled at Secaucus and Sunnyside.

2. Simple interlockings. Complex interlockings limit train speed, and switches especially do. It should be possible for trains to enter and exit Penn Station at speed. Avoiding slow zones in station throats is an underrated way of improving line haul time at relatively low cost, and the fancier the trains and more upgraded the tracks elsewhere are, the heavier the time penalty of slow throats is.

3. Adequate platforms. Penn Station has less bad platforms than people think – when I timed rush-hour LIRR trains, they emptied in about 90 seconds or a little more – but they’re still not good, especially if we’re assuming large increases in ridership coming from better service. The LIRR has better platforms than New Jersey Transit because it has more access points per platform, but the platforms are still narrow. In the worst case, a plan should consider paving over some tracks to widen the platforms, since Penn’s 21 tracks are more than enough for its traffic.

4. Adequate speed for intercity trains. The current tunnels are limited to 60 miles per hour (97 km/h) because of air resistance generated by non-aerodynamic trains in narrow tunnels. The cross-sectional area of the trains going through the tunnels is about half the cross-sectional area of the tunnels; in new high-speed rail tunnels, the corresponding ratio, called the obstruction ratio, is about 15%. It is possible to squeeze more speed out of the existing tunnels with better aerodynamics and sealing, but 200 km/h is probably impossible, and even slightly lower speeds are a problem if the tunnels are very busy and there is a speed mismatch with unsealed, non-aerodynamic commuter trains. So ideally, the larger-diameter new tunnels should be used by intercity trains. We can plan around one minute of travel time difference; this figure can in reality be anywhere between zero and 2.5 minutes.

5. Separation between intercity trains and Grand Central trains. If there is a Penn-Grand Central tunnel, it should be used exclusively by commuter trains because of the high local travel demand, and because to reduce real estate acquisition cost the curve radius should be low, possibly too low for Shinkansen equipment. This means that if a Hudson tunnel points toward the Penn-Grand Central tunnel, intercity trains should use the other one.

6. Two platform tracks per tunnel track. This improves capacity in two ways. First, rapid transit capacity is a combination of tunnel capacity and station dwell, and splitting each tunnel track between two platform tracks allows slightly higher capacity by deemphasizing the dwell since successive trains use different platform tracks. Paris is limited to 30 tph on the RER A with moving-block signaling, whose central segment has one platform track per tunnel track, but the shared RER B+D tunnel between Gare du Nord and Chatelet-Les Halles gets 32 tph with fixed blocks, where the B and D serve separate platforms at each station. And second, because each train can dwell at the platform for longer, this reduces the need for wider platforms, allowing violations of #3. Ideally, the two platform tracks would face the same platform to improve wayfinding and allow unscheduled track changes in case of train delays.

Now, Penn’s tracks are numbered 1-21, from south to north. There are platforms between tracks 20-21, 18-19, 17-18, 15-16, 13-14, 11-12, 9-10, 7-8, 5-6, 3-4, and 1-2; the 18-19 platform is wider than the rest, as if there was supposed to be another track immediately north of 18. The Hudson tunnels connect to tracks 1-19, the southern pair of the East River tunnels connects to 5-15, the northern pair connects to 15-21. The Hudson tunnels and the southern pair face each other, and trains on tracks 11 and 12 go straight through every switch; the northern pair offers no switch-free option, since the eastbound track faces track 21 and the westbound curves into the interlocking.

Since the Hudson tunnels face the southern pair, a simple proposal for new Hudson tunnels should face the northern pair. This would give 4 tracks between New Jersey and Sunnyside. Each of the two track pairs could point toward either Long Island or Connecticut, because of the Harold Interlocking, flawed as it is. This means intercity trains would use the northern pair and go to Connecticut, other regional trains would use either pair and go to either Connecticut or Long Island, and Harold would be superfluous. New Penn-Grand Central tunnels could be constructed branching from the southern East River tunnel pair. It is possible to also construct the tunnels around the southern pair, with trains from the southern pair either merging heading into the Hudson tunnels or terminating and reversing at Penn Station.

Under this plan, all of the numbered principles could be satisfied except #3, and to satisfy #3 every plan requires track paving or other platform modification. It’s also simple to construct: it’s just a new tunnel pair, and ideally also some work on the preexisting southern East River tunnel pair to construct a connection to Grand Central. The one drawback is that, unlike in ARC Alt G, Gateway, and the IRUM plan, the Penn-Grand Central tunnel shares approach tracks to Penn Station with the southern tunnels, reducing capacity. This becomes a problem if ridership from Grand Central, the LIRR, and points east grows to the point of overwhelming three tunnel pairs (the two heading into Penn, and East Side Access). The alternative with the Penn-Grand Central tunnels going around the southern tunnel pair and only merging heading into the Hudson tunnels has more capacity, but interferes with principles #3 and #6.

Track-paving for any plan is hard, because many of the straightest, longest-platform tracks have to be removed. To widen the platforms and improve throughput, there are two ways to pave over tracks, each of which gets rid of about half the station tracks. One is to pave over every other track, guaranteeing each track access to two medium-width platforms; trains could open both doors then, improving egress. The other is to pave every other pair of adjacent tracks lying between platforms, giving each track access to just one very wide platform. The former option is difficult at Penn because there are support columns between adjacent tracks, and a look at New Jersey Transit rolling stock suggests that each train would have 1-2 doors facing a column. We are left with the latter option, paving over, for example, tracks 20, 17, 16, 13, 12, 9, and 8. Each tunnel track would get two platform tracks facing the same platform, except for the westbound northern tunnel track, whose two tracks (21 and 19) would be split because of the aforementioned columns. The lowest-numbered tracks would not be used; the LIRR’s West Side Yard would not be used regularly but instead trains would run on the tracks more often off-peak. The now enlarged platform between tracks 7 and 10 would be lengthened to allow 16-car trains at track 7, which is currently 13 cars long.

Since principles #3 and #6 are both satisfied, the capacity per track can be quite high. The RER B+D achieves 32 tph, but this is split as 20 tph B and 12 tph D, which is suboptimal since two successive trains could both be B, making the dwell a problem. With even alternation between each platform’s two tracks and moving-block signaling, even higher capacity may be possible, reducing the capacity disadvantage of having just two tunnels coming into Penn from the east rather than three.

The IRUM plan, building on ARC Alt G, is quire different. It has three tunnels in each direction: the northern East River tunnels are paired with the Hudson Line via a new short tunnel linking the Empire Connection with the northern Penn Station tracks, the new Hudson tunnels come into the southern station tracks and then continue to Grand Central via new tunnels, and no tracks are paved. Intercity trains have to keep using the old tunnels, necessitating Harold, but that’s money already spent. Principle #3 is violated, or alternatively #6 is if tracks are paved, but there is more track capacity. The risk there is insufficient demand from the Empire Connection, which would leave the northern tunnels underused, reducing the system’s capacity advantage over a two-tunnel option to just a few tph. This plan has to route intercity trains through the old tunnels. Conversely, the advantage is that it easily shoehorns Empire Connection service, which ARC-North does not.

The main difference between ARC-North and IRUM is investment levels. ARC-North is cheaper; it’s more comparable to the bare-bones proposals for ARC(-South) and Gateway that are hinted at but never formally published. At the low investment levels of a bare-bones proposal, ARC-North is superior because it provides better capacity because of the relatively straight train paths through the station. If there’s no connection to Grand Central, then it is not at a capacity disadvantage since Long Island and Connecticut don’t have the demand, and barely have the capacity, to overwhelm three track pairs (including East Side Access again).

Conversely, ARC-North is harder to retrofit for somewhat higher investment. The best that can be done is tunneling to connect the Empire Connection to the southern tracks, digging new tunnels to Long Island as in Alt S, and running trains to just one line, probably Lower Montauk, which goes through serviceable neighborhoods but lacks a direct connection to the existing tunnels to Penn. This provides much better service coverage because of the Montauk connection, but at higher cost since there’s an additional underwater tunnel. It avoids the expensive components of Gateway and ARC, but so does IRUM, which becomes an intermediate level of investment between ARC-North proper and ARC-North with a Montauk tunnel.