Governor Ned Lamont’s plan for speeding up trains between New York, New Haven, and Hartford seems to have fallen by the wayside, but Metro-North and the Connecticut Department of Transportation are still planning for future investments. Several high-level officials met with the advocates from the Connecticut Commuter Rail Council, and the results are unimpressive – they have made false statements out of ignorance of not just best practices outside North America but also current federal regulations, including the recent FRA reform.
The meeting link is a video and does not have a searchable transcript, so I’m going to give approximate timestamps and ask that people bear with me. At several points, highly-paid officials make statements that are behind the times, unimaginative, or just plain incorrect. The offenders are Richard Andreski, the bureau chief of public transportation for CDOT, who according to Transparency.CT earns a total of $192,000 a year including fringe benefits, and Glen Hayden, Metro-North’s vice president of engineering, who according to See Through NY earns an annual base salary of $219,000.
20-25 minutes: there’s a discussion, starting a few minutes before this timestamp, about Metro-North’s future rolling stock procurement. In addition to 66 M8 electric multiple units (EMUs), the railroad is planning to buy 60 unpowered railcars. Grilled about why buy unpowered railcars rather than multiple units, such as diesel multiple units (DMUs), Andreski said a few questionable things. He acknowledged that multiple units accelerate faster than locomotive-hauled trains, but said that this was not needed on the lines in question, that is the unpowered Metro-North branch lines, Shore Line East, and the New Haven-Hartford line. In reality, the difference, on the order of 45 seconds per stop at a top speed of 120 km/h (55 seconds if the top speed is 144 km/h), and electrification both massively increases reliability and saves an additional 10 seconds per stop (or 30 if the top speed is 144).
More worryingly, Andreski talks about the need for flexibility and the installed base of diesel locomotives. He suggests unpowered cars are more compatible with what he calls the train of the future, which runs dual-mode. Dual-mode trains today are of low quality, and the innovation in the world focuses on single-mode electric trains, with a growing number of railroads electrifying as well as transitioning to multiple units. Metro-North itself is a predominantly EMU-based railroad – running more EMUs, especially on the already-wired Shore Line East, is more compatible with its existing infrastructure and maintenance regime than keeping low-performing diesel branches and running diesel under catenary on the trunk line.
1:14-1:17: Andreski states that the 60 unpowered single-level cars should cost about $250 million, slightly more than $4 million per car. When a reader of this blog noted that in the rest of the world, a 25-meter multiple-unit costs $2.5 million, Andreski responded, “this is not accurate.” The only trouble is, it is in fact accurate; follow links to contracts reported in Railway Gazette in the rolling stock cost section of this post. It is not clear whether Andreski is lying, ignorant, or in a way both, that is making a statement with reckless disregard for whether it is true.
Hayden then chimes in, talking about FRA regulations, saying that they’re different from American ones, so European and Asian prices differ from American ones, seemingly indifferent to the fact that he just threw Andreski under the bus – Andreski said that multiple-units do not cost $2.5 million per car and if a public contract says they do then it’s omitting some extra costs. The only problem is, FRA regulations were recently revised to be in line with European ones, with specific eye toward permitting European trains to run on American tracks with minimal modifications, measured in tens of thousands of dollars of extra cost per car. In a followup conversation off-video, Hayden reiterated that position to longtime reader Roger Senserrich – he had no idea FRA regulations had been revised.
Hayden’s response also includes accessibility requirements. Those, too, are an excuse, albeit a slightly defensible one: European intercity trains, which are what American tourists are most likely to have experience with, are generally inaccessible without the aid of conductors and manual boarding plates. However, regional trains are increasingly fully accessible, at a variety of floor heights, and it’s always easier to raise the floor height to match the high platforms of the Northeast Corridor than to lower it to match those of low-platform networks like Switzerland’s.
1:45: asked about why Metro-North does not run EMUs on the wired Shore Line East, a third official passes the buck to Amtrak, saying that Amtrak is demanding additional tests and the line is Amtrak’s rather than Metro-North’s property. This is puzzling, as 1990s’ Amtrak planned around electrification of commuter rail service east of New Haven, to the point of constructing its substations with room for expansion if the MBTA were ever interested in running electric service on the Providence Line. It’s possible that Amtrak today is stalling for the sake of stalling, never mind that commuter rail electrification would reduce the speed difference with its intercity trains and thus make them easier to schedule and thus more reliable. But it’s equally possible that CDOT is being unreasonable; at this point I would not trust either side of any Amtrak-commuter rail dispute.
The largest single transportation project in Germany today is a new underground main station for Stuttgart, dubbed Stuttgart 21. Built at a cost of €8.2 billion, it will soon replace Stuttgart’s surface terminal with a through-station, fed in four directions by separate tunnels. The project attracted considerable controversy at the beginning of this decade due to its cost overruns and surface disruption. It’s had a long-term effect on German politics as well: it catapulted the Green Party into its first ever premiership of a German state, and the Green minister-president of the state, Winfried Krestchmann, has remained very popular and played a role in mainstreaming the party and moving it in a more moderate direction.
But the interesting thing about Stuttgart 21 now is not the high cost, but a new problem: capacity. The new station will face capacity constraints worse than those of the surface station, particularly because Germany is transitioning toward timed connections (“Deutschlandtakt”) on the model of Switzerland. Since Stuttgart is closing the surface station and selling the land for redevelopment, a second underground station will need to be built just to add enough capacity. It’s a good example of how different models of train scheduling require radically different kinds of infrastructure, and how even when all the technical details are right, the big picture may still go wrong.
What is the Stuttgart 21 infrastructure?
The following diagram (via Wikipedia) shows what the project entails.
The existing tunnel, oriented in a northeast-southwest direction, is used exclusively by S-Bahn trains. Longer-distance regional trains (“RegionalBahn“) and intercity trains terminate on the surface, and if they continue onward, they must reverse direction.
The new tunnel infrastructure consists of four independent two-track tunnels, two coming in from the northwest and two from the southeast, with full through-service. In addition, an underground loop is to be constructed on the south in order to let trains from points south (Singen) enter Stuttgart via the Filder tunnel while serving the airport at Filder Station without reversing direction. The total double-track tunnel length is 30 kilometers.
Stuttgart 21’s station infrastructure will consist of eight tracks, four in each direction:
The two tracks facing each platform are generally paired with the same approach track, so that in case of service changes, passengers will not be inconvenienced by having to go to a different platform. The interlocking permits trains from each of the two eastern approaches to go to either of the western ones without conflict and vice versa, and the switches are constructed to modern standards, with none of the onerous speed restrictions of American station throats.
So what is the problem?
First of all, the four approach tunnels are not symmetric. The Feuerbach tunnel leads to Mannheim, Frankfurt, Würzburg, and points north, and the Filder tunnel leads to Ulm and points east, including Munich; both are planned to be heavily used by intercity trains. In contrast, the other two tunnels lead to nothing in particular. The Obertürkheim tunnel leads to the current line toward Ulm, but the under-construction high-speed line to Ulm feeds Filder instead, leaving Obertürkheim with just a handful of suburbs.
On the Deutschlandtakt diagram for Baden-Württemberg, every hour there are planned to be 12 trains entering Stuttgart from the Feuerbach tunnel, 10.5 from the Filder tunnel, 5.5 from the Bad Cannstatt tunnel, and 6 from the Obertürkheim tunnel. For the most part, they’re arranged to match the two busier approaches with each other – the track layout permits a pair of trains in either matching to cross with no at-grade conflict, but only if trains from Feuerbach match with Filder and trains from Bad Cannstatt match with Obertürkheim are both station tracks facing the same platform available without conflict.
A train every five minutes through a single approach tunnel feeding two station tracks is not normally a problem. The S-Bahn, depicted on the same map in black, runs 18 trains per hour in each direction through the tunnel; bigger cities, including Paris and Munich, run even more frequent trains on the RER or S-Bahn with just a single station platform per approach track, as on any metro network.
However, the high single-track, single-direction frequency is more suitable on urban rail than on intercity rail. On a metro, trains rarely have their own identity – they run on the same line as a closed system, perhaps with some branching – so if a train is delayed, it’s possible to space trains slightly further apart, so the nominal 30 trains per hour system ends up running 28 trains if need be. On an S-Bahn this is more complicated, but there is still generally a high degree of separation between the system and other trains, and it’s usually plausible to rearrange trains through the central tunnel. On intercity rail, trains have their own identity, so rearrangement is possible but more difficult if for example two trains on the same line, one express and one local, arrive in quick succession. As a result, one platform track per approach track is unsuitable – two is a minimum, and if more tracks are affordable then they should be built.
How do you intend to run the trains?
If the paradigm for intercity rail service is to imitate shorter-range regional trains, then through-tunnels are both easier and more desirable. A relatively closed system with very high frequency between a pair of stations calls for infrastructure that minimizes turnarounds and lets trains just run in the same sequence.
The Shinkansen works this way, leveraging three key features: its near-total isolation from the legacy train network, running on a different gauge; the very high demand for trains along individual corridors on specific city pairs; and the generally high punctuality of Japanese trains even on more complex systems. As it happens, Tokyo is a terminal, with trains going north and south but not through, as a legacy of the history of breaking up Japan National Railway before the Shinkansen reached Tokyo from the north, with different daughter companies running in each direction. However, Shin-Osaka is a through-station, fitting through-trains as well as terminating trains on just eight tracks.
In the developed world’s second busiest intercity rail network, that of Switzerland, the paradigm is different. In a country whose entire population is somewhat less than that of Tokyo without any of its suburbs, no single corridor is as strong as the Shinkansen corridors. Trains form a mesh with timed connections every hour, sometimes every half hour. Intercity trains are arranged to arrive at Zurich, Bern, and Basel a few minutes before the hour every 30 minutes and depart a few minutes later. In that case, more approach tracks and more platform tracks are needed. Conversely, the value of through-tracks is diminished, since passengers can transfer between trains more easily if they can walk between platforms without changing grade.
Germany aims to integrate the infrastructure and timetable, as Switzerland does. However, Stuttgart 21 is a failure of such integration. The Deutschlandtakt service paradigm calls for many trains entering and leaving the station within the span of a few minutes. Today there are four effective approaches with two tracks each, same as under the Stuttgart 21 plan, but they are better-distributed.
The idea of Stuttgart 21, and similar proposals for Frankfurt and Munich, is solid provided that the intention is to run trains the Japanese way. It Stuttgart were designed to be the junction of two consistently high-intensity lines, then it would work without additional infrastructure. But it is not: its approach tunnels are supposed to support such design, but the service pattern will not look this way because of how the tunnels are placed relative to Germany’s population distribution. Even highly competent engineering can produce incompetent results if the details do not match the big picture.
The American rail activist term regional rail refers to any mainline rail service short of intercity, which lumps two distinct service patterns. In some German cities, these patterns are called S-Bahn and RegionalBahn, with S-Bahn referring to urban rail running on mainline tracks and RegionalBahn to longer-range service in the 50-100 km range and sometimes even beyond. It’s useful to distinguish the two whenever a city wishes to invest in its regional rail network, because the key infrastructure for the two patterns is different.
As with many this-or-that posts of mine, the distinction is not always clear in practice. For one, in smaller cities, systems that are labeled S-Bahns often work more like RegionalBahn, for example in Hanover. Moreover, some systems have hybrid features, like the Zurich S-Bahn – and what I’ve advocated in American contexts is a hybrid as well. That said, it’s worth understanding the two different ends of this spectrum to figure out what the priority for rail service should be in each given city.
S-Bahn as urban rail
The key feature of the S-Bahn (or the Paris RER) is that it has a trunk that acts like a conventional urban rapid transit line. There are 6-14 stations on the trunks in the examples to keep in mind, often spaced toward the high end for rapid transit so as to provide express service through city center, and all trains make all stops, running every 3-5 minutes all day. Even if the individual branches run on a clockface schedule, people do not use the trunk as a scheduled railroad but rather show up and go continuously.
Moreover, the network layout is usually complementary with existing urban rail. The Munich S-Bahn was built simultaneously with the U-Bahn, and there is only one missed connection between them, The Berlin S-Bahn and U-Bahn were built separately as patchworks, but they too have one true missed connection and one possible miss that depends on which side of the station one considers the crossing point to be on. The RER has more missed connections with the Metro, especially on the RER B, but the RER A’s station choice was designed to maximize connections to the most important lines while maintaining the desired express stop spacing.
Urban rail lines rarely terminate at city center, and the same is true for S-Bahn lines. In cities whose rail stations are terminals, such as Paris, Munich, Frankfurt, and Stuttgart, there are dedicated tunnels for through-service; London is building such a tunnel in Crossrail, and built one for Thameslink, which has the characteristics of a hybrid. In Japan, too, the first priority for through-running is the most local S-Bahn-like lines – when there were only six tracks between Tokyo and Ueno, the Yamanote and Keihin-Tohoku Lines ran through, as did the Shinkansen, whereas the longer-range regional lines terminated at the two ends until the recent through-line opened.
The difference between an S-Bahn and a subway is merely that the subway is self-contained, whereas the S-Bahn connects to suburban branches. In Tokyo even this distinction is blurred, as most subway lines connect to commuter rail lines at their ends, often branching out.
RegionalBahn as intercity rail
Many regional lines descend from intercity lines that retooled to serve local traffic. Nearly every trunk line entering London from the north was built as a long-range intercity line, most commuter rail mainlines in New York are inner segments of lines that go to other cities or used to (even the LIRR was originally built to go to Boston, with a ferry connection), and so on.
In Germany, it’s quite common for such lines to maintain an intercity characteristic. The metropolitan layout of Germany is different from that of the English-speaking world or France. Single-core metro regions are rather small, except for Berlin. Instead, there are networks of independent metropolitan cores, of which the largest, the Rhine-Ruhr, forms an urban complex almost as large as the built-up areas of Paris and London. Even nominally single-core metro regions often have significant independent centers with long separate histories. I blogged about the Rhine-Neckar six months ago as one such example; Frankfurt is another, as the city is ringed by old cities including Darmstadt and Mainz.
But this is not a purely German situation. Caltrain connects what used to be two independent urban areas in San Francisco and San Jose, and many outer ends of Northeastern American commuter lines are sizable cities, such as New Haven, Trenton, Providence, and Worcester.
The intercity characteristic of such lines means that there is less need to make them into useful urban rail; going express within the city is more justifiable if people are traveling from 100 km away, and through-running is a lower priority. Frequency can be lower as well, since the impact of frequency is less if the in-vehicle travel time is longer; an hourly or half-hourly takt can work.
S-Bahn and RegionalBahn combinations
The S-Bahn and RegionalBahn concepts are distinct in history and service plan, but they do not have to be distinct in branding. In Paris, the distinction between Transilien and the RER is about whether there is through-running, and thus some lines that are RegionalBahn-like are branded as RER, for example the entire RER C. Moreover, with future extension plans, the RER brand will eventually take over increasingly long-distance regional service, for example going east to Meaux. Building additional tunnels to relieve the worst bottlenecks in the city’s transport network could open the door to connecting every Transilien line to the RER.
Zurich maintains separate brands for the S-Bahn and longer-distance regional trains, but as in Paris, the distinction is largely about whether trains terminate on the surface or run through either of the tunnels underneath Hauptbahnhof. Individual S-Bahn branches run every half hour, making extensive use of interlining to provide high frequency to urban stations like Oerlikon, and many of these branches go quite far out of the city. It’s not the same as the RER A and B or most of the Berlin S-Bahn, with their 10- and 15-minute branch frequencies and focus on the city and innermost suburbs.
But perhaps the best example of a regional rail network that really takes on lines of both types is that of Tokyo. In branding, the JR East network is considered a single Kanto-area commuter rail network, without distinctions between shorter- and longer-range lines. And yet, the rapid transit services running on the Yamanote, Keihin-Tohoku, and Chuo-Sobu Lines are not the same as the highly-branched network of faster, longer-range lines like Chuo Rapid, Yokosuka, Sobu Rapid, and so on.
The upshot is that cities do not need to neatly separate their commuter rail networks into two separate brands as Berlin does. The distinction is not one of branding for passengers, but one of planning: should a specific piece of infrastructure be S-Bahn or RegionalBahn?
Highest and best use for infrastructure
Ordinarily, the two sides of the spectrum – an S-Bahn stopping every kilometer within the city, and a RegionalBahn connecting Berlin with Magdeburg or New York with New Haven – are so different that there’s no real tradeoff between them, just as there is no tradeoff between building subways and light rail in a city and building intercity rail. However, they have one key characteristic leading to conflict: they run on mainline track. This means that transportation planners have to decide whether to use existing mainline tracks for S-Bahn or RegionalBahn service.
Using different language, I talked about this dilemma in Boston’s context in 2012. The situation of Boston is instructive even in other cities, even outside the United States, purely because its commuter rail service is so bad that it can almost be viewed as blank slate service on existing infrastructure. On each of the different lines in Boston, it’s worth asking what the highest and best use for the line is. This really boils down to two questions:
- Would the line fill a service need for intra-urban travel?
- Does the line connect to important outlying destinations for which high speed would be especially beneficial?
In Boston, the answer to question 1 is for the most part no. Thirty to forty years ago the answer would have been yes for a number of lines, but since then the state has built subway lines in the same rights-of-way, ignorant of the development of the S-Bahn concept across the Pond. The biggest exceptions are the Fairmount Line through Dorchester and the inner Fitchburg Line through suburbs of Cambridge toward Brandeis.
On the Fairmount Line the answer to question 2 is negative as well, as the line terminates within Boston, which helps explain why the state is trying to invest in making it a useful S-Bahn with more stops, just without electrification, high frequency, fare integration, or through-service north of Downtown Boston. But on the Fitchburg Line the answer to question 2 is positive, as there is quite a lot of demand from suburbs farther northwest and a decent anchor in Fitchburg itself.
The opposite situation to that of Fairmount is that of the Providence Line. Downtown Providence is the largest job center served by the MBTA outside Boston; the city ranks third in New England in number of jobs, behind Boston and Cambridge and ahead of Worcester and Hartford. Fast service between Providence and Boston is obligatory. However, Providence benefits from lying on the Northeast Corridor, which can provide such service if the regional trains are somewhat slower; this is the main justification for adding a handful of infill stops on the Providence Line.
In New York, the situation is the most complicated, befitting the city’s large size and constrained location. On most lines, the answers to both questions is yes: there is an urban rail service need, either because there is no subway service (as in New Jersey) or because there is subway service and it’s overcrowded (as on the 4/5 trains paralleling the Metro-North trunk and on the Queens Boulevard trains paralleling the LIRR trunk); but at the same time, there are key stations located quite far from the dense city, which can be either suburban centers 40 km out or, in the case of New Haven, an independent city more than 100 km out.
Normally, in a situation like New York’s, the solution should be to interline the local lines and keep the express lines at surface terminals; London is implementing this approach line by line with the Crossrail concept. Unfortunately, New York’s surface terminals are all outside Manhattan, with the exception of Grand Central. Penn Station has the infrastructure for through-running because already in the 1880s and 90s, the ferry transfers out of New Jersey and Brooklyn were onerous, so the Pennsylvania Railroad invested in building a Manhattan station fed by east-west tunnels.
I call for complete through-running in New York, sometimes with the exception of East Side Access, because of the island geography, which makes terminating at the equivalent of Gare du Nord or Gare de Lyon too inconvenient. In other cities, I might come to different conclusions – for example, I don’t think through-running intercity trains in Chicago is a priority. But in New York, this is the only way to guarantee good regional rail service; anything else would involve short- and long-range trains getting in each other’s way at Penn Station.
I’ve argued in two previous posts that Germany needs to build a complete high-speed rail network, akin to what China, Japan, France, South Korea, and Spain have built. Here is the network that Germany should build in more detail:
The red lines denote high-speed lines, some legacy 250-280 km/h lines but most built to support 300-320 km/h, that are justifiable within the context of domestic travel. Some of these already exist, such as the Frankfurt-Cologne line and the majority of the Berlin-Munich line; Berlin-Hamburg is a legacy line upgraded to 230, currently tied with Frankfurt-Cologne for fastest average speed between two major cities in Germany. A handful of red lines are key legacy connections, i.e. Dresden-Leipzig and Dortmund-Duisburg. Some more detail on the red lines is available in Google Maps.
The blue lines denote high-speed lines, generally built to 300, that only make sense in an international context. The lines in France are the LGV Est and its short low-speed branch across the border to Saarbrücken. In Belgium the line preexists as well as HSL 3 and HSL 4, but is quite slow, averaging only 140 km/h from Brussels to Aachen thanks to a combination of a slow segment to Leuven and a speed-restricted western approach to Liege. In the Netherlands, Switzerland, Czechia, Austria, and Poland the lines are completely speculative, though in Czechia a high-speed line from Prague to Dresden is under study.
Update 8/19: here is another map of the same network, color-coded differently – red is proposed lines (most by me, a few officially), yellow is lines under construction, blue is existing lines, black is low-speed connections. Note that outside Berlin’s northern approaches, urban approaches are not colored black even if they’re slow.
To compute trip times, I dusted off my train performance calculator, linked here. The parameters I used are those planned for the next-generation Velaro (“Velaro Novo“), i.e. a power-to-weight ratio of 20.7 kW/t and an initial acceleration rate of 0.65 m/s^2; the quadratic air resistance term is 0.000012, as any higher term would make it impossible to reach speeds already achieved in tests. On curves, the lateral acceleration in the horizontal plane is set at 2.09 m/s^2 on passenger-priority lines, mirroring what is achieved on Frankfurt-Cologne, and 1.7 elsewhere, accounting for lower superelevation.
These are aggressive assumptions and before running the code, I did not expect Berlin-Munich to be so fast. With intermediate stops at Erfurt, Nuremberg, and maybe also Ingolstadt, this city pair could be connected in 2.5 hours minus a few minutes for interchange time at the terminals. In general, all trip times printed on the map are a few minutes slower than what is achievable even with some schedule padding, corresponding to dwell times at major through-stations plus interchange at terminals. The upshot is that among the largest metro areas in Germany, the longest trips are Hamburg-Stuttgart at 3:30 minus change and Hamburg-Munich at 3:15 minus change; nothing else is longer than 3 hours.
The stopping pattern should be uniform. That is, every 320 km/h train between Berlin and Munich should stop exactly at Berlin Südkreuz, Erfurt, Nuremberg, and maybe Ingolstadt. If these trains skip Ingolstadt, it’s fine to run some 250 km/h trains part of the way, for example between Munich and Nuremberg and then northwest on legacy track to Würzburg and Frankfurt, with the Ingolstadt station added back. Similarly, from Hamburg south, every train should stop at Hanover, Göttingen, Kassel, and Fulda.
In certain cases, the stopping pattern should be decided based on whether trains can make a schedule in an exact number of quarter-hours. That is, if it turns out that Munich-Nuremberg with an intermediate stop in Ingolstadt takes around 42 minutes then the Ingolstadt stop should be kept; but if it takes 46 minutes, then Ingolstadt should be skipped, and instead of running in the depicted alignment, the line should stay near the Autobahn and bypass the city in order to be able to make it in less than 45 minutes. I think Ingolstadt can still be kept, but one place where the map is likely to be too optimistic is Stuttgart-Munich; Ulm may need to be skipped on the fastest trains, and slower trains should pick up extra stops so as to be 15 minutes slower.
Frequency and service planning
Today, the frequency on the major city pairs is hourly. Under the above map, it should be half-hourly, since the faster trip times will induce more ridership. As a sanity check, TGVs connect Paris with each of Lyon’s two stations hourly off-peak and twice an hour at the peak. Paris is somewhat larger than the entire Rhine-Ruhr, Lyon somewhat smaller than Stuttgart or Munich and somewhat larger than the Rhine-Neckar. But the ICE runs somewhat smaller trains and has lower occupancy as it runs trains on a consistent schedule all day, so matching the peak schedule on the TGV is defensible.
The upshot is that Berlin can probably be connected every 30 minutes to each of Hamburg, Munich, Frankfurt, Cologne, Düsseldorf, and the Ruhr proper. Frankfurt-Munich is likely to be every 30 minutes, as are Hamburg-Frankfurt and Hamburg-Munich. To further improve network connectivity, the schedule at Erfurt should be set in such a way that Hamburg-Munich and Berlin-Frankfurt trains are timed with a cross-platform transfer, regardless of the pulse anywhere else. A few connections to smaller cities should be hourly, like Berlin-Bremen (with a timed transfer at Hanover to Hamburg-Frankfurt or Hamburg-Munich), Leipzig-Munich, Leipzig-Frankfurt, and Frankfurt-Basel.
The loop track around Frankfurt is based on a real plan for mainline through-tracks at the station, currently in the early stages of construction. The near-Autobahn loop is not included, but such a connection, if done at-grade, could provide value by letting trains from Munich enter the station from the east and then continue northwest toward Cologne without reversing direction.
If the international connections are built as planned, then additional hourly and even more frequent connections can be attractive. Zurich-Stuttgart might well even support a train every half hour, going all the way to Frankfurt and thence to either Cologne or Berlin. Similarly, Berlin-Frankfurt-Paris could plausibly fill an hourly train if Frankfurt-Paris is cut to 2:30 via Saarbrücken, and maybe even if it takes three hours via Karlsruhe.
The one exception to this interconnected mesh is Fulda-Würzburg. The Hanover-Würzburg line was built as a single 280 km/h spine through West Germany with low-speed branches down to Frankfurt and Munich. Unfortunately, completing the Würzburg-Nuremberg segment has little value: Munich-Frankfurt would be almost as fast via Stuttgart, and Hamburg-Munich would be half an hour faster via Erfurt with not much more construction difficulty on Göttingen-Erfurt. Fulda-Würzburg should thus be a shuttle with timed transfers at Fulda, potentially continuing further south at lower speed to serve smaller markets in Bavaria.
The domestic network depicted on the map is 1,300 km long, not counting existing or under-construction lines. Some lines require tunneling, like Erfurt-Fulda-Frankfurt, but most do not; the heaviest lifting has already been done, including between Erfurt and Nuremberg and around Stuttgart for Stuttgart 21 and the under-construction high-speed line to Ulm. I doubt 100 km of tunnel are necessary for this network; for comparison, Hanover-Würzburg alone has 120 km of tunnel, as the line has very wide curve radii to support both high-speed passenger rail and low-speed freight without too much superelevation. The cost should be on the order of 30-40 billion euros.
The international network is more complex. Berlin-Prague is easy on the German side and even across the border, and the only real problems are on the Czech side, especially as Czech planners insist on serving Usti on the way with a city center station. But Stuttgart-Zurich is a world of pain, and Frankfurt-Saarbrücken may require some tunneling through rolling terrain as well, especially around Saarbrücken itself.
Even with the international lines added in, the German share of the cost should not be too onerous. Getting everything in less than 50 billion euros should not be hard, even with some compromises with local NIMBYs. Even on an aggressive schedule aiming for completion by 2030, it’s affordable in a country where the budget surplus in 2018 was €58 billion across all levels of government and where there are signs of impending recession rather than inflation.
With its mesh of medium-size cities all over the country following plausible lines, Germany is well-placed to have the largest high-speed rail network in Europe. It has the ability to combine the precise scheduling and connections of Switzerland and the Netherlands with the high point-to-point speeds of France and Spain, creating a system that obsoletes domestic flights and competes well with cars and intercity buses. The government can implement this; all it takes is the political will to invest in a green future.
I’ve been asked from time to time, Alon, you write about comparative rail costs all the time, but what about roads? Sometimes the question expresses curiosity about whether roads display the same American construction cost premium as urban rail does; sometimes it expresses frustration that The Discourse doesn’t complain about road costs. Regardless of why people ask, I’d like to explain my reasoning in depth, especially now that serious people are asking why this is the focus of my comparative research.
There’s an easy answer and a hard answer. The easy answer is that I’m a railfan. I got into this because I was living in Morningside Heights and taking the subway to social events in Brooklyn and Queens, which involved 3- and sometimes 4-seat rides. It got me interested in coverage gaps and subway extensions, which got me interested in the construction costs of such extensions.
But that’s not really it. From my original purpose of comparing a few urban infill subways in large global cities I got into operating costs, and high-speed rail, and light rail, and electrification, and even road tunnels (here is my comparison of urban road tunnel projects). What’s more, other people have looked at comparative costs, and even without sharing my not-knowing-how-to-drive origin story, they don’t compare individual road projects much. The Brookings study about the Interstates looked at the entire cost of the US Interstate program rather than teasing it out project by project.
What’s really going on is that subways are megaprojects. Megaprojects are visible, and I don’t just mean physically – they’re widely discussed in the media and politics, and cost overruns invite intense criticism by the opposition and by investigative reporters. Everybody in New York knows about Second Avenue Subway, and everybody in New Jersey knows about the Gateway tunnel, and everybody in London knows about Crossrail.
The upshot is that megaproject cost estimates are just more reliable than those of anything else. What I mean is not that cost overruns are unlikely. Rather, what I mean is that cost overruns are difficult to hide, unless the agency goes the Canadian route of fluffing the budget with very high contingencies. The current budget for Grand Paris Express is around €35 billion, up from €25 billion when it was first announced. If it actually ends up at €36 billion and not €35 billion then it may be possible to scrounge extra funds from a few sources sub rosa, but not if it ends up at €45 billion.
The largest source of wasteful spending in the world is the American military. It has a budget of $700 billion a year, debated largely behind the scenes, with boisterous generals and their lackeys ready to publicly defend every $600 toilet seat and every procurement item in the district of any member of Congress who dares object. There is a shroud of secrecy around everything that can be justified as national security. There is no exit threat – the military can’t be shut down the way an underperforming state railroad can be privatized. Hidden costs are rampant, and as far as I understand, they are on the order of a few billion dollars at a time.
I bring up American military waste not to justify civilian waste on infrastructure, but to compare which costs can be plausibly hidden. If the US military can miss a few billion dollars, the transport planners of Ile-de-France can miss tens to hundreds of millions of euros on a 15-year, 200-kilometer project. Those of Madrid can probably miss an amount of money on the same order of magnitude as those of Paris. The low construction costs in Madrid have been plugged into additional construction, giving Madrid Europe’s third longest metro network after London and Moscow; those hundreds of kilometers built in the last 25 years could not have cost the same as in France, let alone the US, because this would have been too big of a difference, and the media would have noticed.
The same situation equally occurs for road megaprojects, such as tunnels or big urban reconstruction projects, such as the lane additions in Los Angeles. But it does not occur for run-of-the-mill road widening outside urban areas or for small projects to increase the capacity of a junction from a cloverleaf to a four-level interchange. These are not sufficiently visible for me to be able to trust that there is full cost accounting in the trade and popular press.
I’m happy to compare the costs of road tunnels between different cities; the few examples I have found paint the same picture as the subway cost comparison. But above-ground road construction is harder, just because “above-ground” can mean anything from a complex viaduct-over-viaduct to simple at-grade construction. Even then, ancillary costs like unnecessary street reconstruction may be bundled into the overall budget, and since above-ground construction isn’t so expensive, these extras may be a sizable fraction of the cost.
For a similar reason, I don’t look at airports so much: they’re just harder to compare. I do not know how big the Berlin-Brandenburg disaster is compared with other airports under construction, so I do not know how much it should cost; I don’t even know what the equivalent metric of cost per km or cost per new station excavated is. In contrast, to take another well-known German infrastructure disaster, Stuttgart21 has a definite tunnel length – 30 kilometers, as well as another 25 above ground – so I can compare with other regional rail projects and say that actually the cost of Stuttgart21 (€6.5 billion) is not so high relative to how much urban mainline rail tunneling costs elsewhere in the world.
For the exact same reason, when I look at above-ground urban rail I try to separate out truly at-grade light rail from elevated lines. The only times I try to do a deep dive are when these projects encroach on the cost range of subways, like the Boston Green Line Extension. Elsewhere, ancillary costs can be substantial, as with the Nice tramway: 70% of the budget was the tramway itself and 30% was stormwater drainage, rebuilding a public plaza, tree planting, and other extras. Extras introduce an error term into comparisons that are harder to ignore when the cost is $50 million per kilometer than when it is $300 million per kilometer.
Road costs remain a powerful sanity check. All of the reasons I (and others) believe are behind the American construction cost premium are equally applicable to roads and urban rail. So far, looking at road tunnels confirms the subway pattern, but there just aren’t a lot of road tunnels built around the world – they’re expensive for the capacity they provide. And if it’s possible to carefully tease out above-ground road megaproject costs then a comparison is welcome as well. But they are unlikely to form the backbone of any comparison.
Metro tunnels, for all the handwringing about special circumstances, are pretty consistent. Some places have easier rock and some have harder rock, but usually this will be noted in the trade and popular press; the most fundamental quantities, length and the number of stations, are if anything easier to find than the headline costs; ancillary extra costs are usually not significant, and when they are, they tend to be bundled into quantifiable metrics like station size and depth. The only big difference in reporting regimes is that some places (like Spain) bundle together infrastructure and rolling stock costs whereas most don’t.
The main approach to project-level comparison of infrastructure costs across countries has to be about urban rail, because that’s by far what’s most common across the world. The error bars around ex post costs are small enough that even a relatively restricted sample is suggestive of the real global effect as I’m learning when adding more and more projects to my database (currently about 130 projects totaling 2,000 km). This is the most comparable list of public infrastructure projects, and what we may learn about why various American urban rail lines cost so much and why Spanish and Korean and Nordic ones cost so little is likely to generalize.
Some countries build complete high-speed rail networks, on which one can travel between cities almost entirely at high speed, such as France, Japan, and China. Others build partial networks, mixing low- and high-speed travel, such as Germany. The planning lingo in the latter is “strategic bypass” or “strategic connection.” And yet, there is nothing strategic about most mixed lines. If a line between two cities is partly high-speed and partly low-speed, it is usually strategic to complete the high-speed line and provide fast travel – the benefits will exceed those of having built the original high-speed partial segment. Since Germany’s rail network largely consists of such mixed lines, the benefits of transitioning to full high-speed rail here are large.
The arguments I’m about to present are not entirely new. To some extent, I discussed an analog years ago when arguing that in the presence of a complete high-speed line, the benefits of building further extensions are large; this post is a generalization of what I wrote in 2013. Then, a few months ago, I blogged about positive and negative interactions. I didn’t discuss high-speed rail, but the effect of travel time on ridership is such that different segments of the same line positively interact.
The upshot is that once the basics of a high-speed rail networks are in place, the benefit-cost ratio of further extensions is high. In a country with no such network, the first line or segments may look daunting, such as India or the UK, but once it’s there, the economics of the rest tend to fall into place. It takes a while for returns to diminish below the point of economic viability.
A toy model
Take a low-speed rail line:
Now build a high-speed line parallel to half of it and connect it with the remaining half:
You will have reduced trip time from 4 hours to 3 hours. This has substantial benefits in ridership and convenience. But then you can go all the way and make the entire line fast:
Are there diminishing returns?
The benefits of reducing travel time per unit of absolute amount of time saved always increase in speed; they never decrease. The gravity model holds that ridership follows an inverse square law in total cost, including ticket fare and the passengers’ value of time, which time includes access and egress time. Reducing in-vehicle travel time by a fixed amount, say an hour, increases ridership more if the initial travel time is already lower.
This is on top of reductions in operating costs coming from higher speed. Trains on high-speed track consume less electricity than on legacy track, because they cruise at a constant speed, and because head-end power demand scales with time rather than distance traveled. Crew wages per kilometer are lower on faster trains. And the cost of rolling stock procurement and maintenance is spread across a longer distance if the same train is run more kilometers per year. In the toy model, there are actually increasing returns coming from rolling stock costs: upgrading half the line to high speed requires running an expensive high-speed train on the entire line, whereas completing the high-speed line does not require increasing the cost per unit of rolling stock.
Diminishing returns do occur, but only in the context of an increase in top speed, not in that of speeding up slow segments to match the top speed of faster segments. In that context, benefits do diminish and costs do rise, but that is not the same as completing high-speed lines.
As the maximum speed is increased from 160 to 200 km/h, the train speeds up from 22.5 seconds per kilometer to 18. To provide the same increase further, that is to reduce the time taken to traverse a kilometer by a further 4.5 seconds to 13.5, the speed must increase to 266.67 km/h. To provide the same 4.5-second increase once more, the speed must increase to 400. Curve radius is proportional to the square of speed, so these increases in speed must be accompanied by much more exacting track geometry. Tunnels may well be unavoidable at the higher speeds in topography that could accommodate 200-250 entirely at-grade.
What’s more, operating costs rise too as top speed increases. The electricity consumption on a 300 km/h cruise is lower than on a legacy line on which trains transition back and forth between 200 and 100 and all speeds in between, but the electricity consumption on a 350 km/h cruise is definitely higher than on a 250 km/h cruise.
However, what is relevant to the decision of what standards to build a line to is not relevant to the decision of how far to extend this standard. Once a 300 km/h segment has been built, with a dedicated fleet of trains that cost €30 million per 200-meter set, the returns to upgrading the entire segment the train runs on are higher than those of just building the initial segment.
Can some strategic segments be easier to build than others?
Yes, but only in one specific situation: that of an urban area. The toy model says nothing of construction costs – in effect, it assumes the cost of making the first 200 km fast is the same as that of making the next 200 km fast. In reality, different areas may have different construction challenges, making some parts easier to build than others.
However, if the construction challenge is mountainous topography, then the higher cost of mountain tunnels balance out the greater benefit of fast trains across mountains. The reason is that in practice, legacy rail lines are faster in flat terrain than in the mountains, where past construction compromises led to sharp curves.
This situation is different in urban areas. In urban areas as in the mountains, costs are higher – land acquisition is difficult, and tunnels may be required in areas where the alternative is buying out entire city blocks. But unlike in the mountains, the existing rail line may well be reasonably straight, permitting average speeds in the 120 km/h area rather than the 70 km/h area. In that case, it may be advisable to postpone construction until later, or even keep the legacy alignment.
One example is the Ruhr area. The tracks between Dortmund and Duisburg are not high-speed rail – the fastest trains do the trip in about 34 minutes, an average speed of about 95 km/h. Speeding them up by a few minutes is feasible, but going much below 30 minutes is not. Thus, even if there is a 300 km/h line from Dortmund to points east, the returns to the same speedup between Dortmund and Duisburg are low. (Besides which, Dortmund is the largest city in the Ruhr, and the second largest, Essen, in the middle between Dortmund and Duisburg.)
Another is Connecticut. East of New Haven, there is relatively little urban development, and constructing a 300-360 km/h line roughly along the right-of-way of I-95 poses few challenges. West of New Haven, such construction would require extensive tunneling and elevated construction – and the legacy line is actually somewhat less curvy, it’s just slower because of poor timetable coordination between Amtrak’s intercity trains and Metro-North’s regional trains. While the returns to building 250-300 km/h bypasses around the line’s slowest points in southwestern Connecticut remain high enough to justify the project, they’re lower than those in southeastern Connecticut.
The situation in Germany
On the following map, black denotes legacy lines and red denotes purpose-built 300 km/h high-speed lines:
The longer red segment, through Erfurt, is the more challenging one, including long tunnels through the mountains between Thuringia and Bavaria. The complexity and cost of construction led to extensive media controversy. In particular, the choice of the route through Erfurt came about due to Thuringia’s demands that it serve its capital rather than smaller cities; DB’s preference would have been to build a more direct Leipzig-Nuremberg route, which would have had shorter tunnels as the mountains in eastern Thuringia are lower and thinner.
Since then, a lot of water has passed under the bridge. The route opened at the end of 2017 and cut travel time from 6 hours to 4, bypassing the slowest mountain segment, and is considered a success now. In the North German Plain, the trains mostly cruise at 200 km/h, and trains traverse the 163.6 km between Berlin and Halle in 1:09-1:11, an average speed of 140 km/h.
Nonetheless, the benefits of painting the entire map red, roughly from the city limits of Berlin to those of Munich, are considerable. The North German Plain’s flat topography enables trains to average 140 km/h, but also means that building a high-speed line would be cheap – around 137 km of new-build line would be needed, all at-grade, at a cost of about €2.5 billion, which would cut about half an hour from the trip time. In Bavaria, the topography is rougher and consequently the legacy trains’ average speed is lower, but nonetheless, high-speed rail can be built with cut-and-fill, using 4% grades as on the Cologne-Frankfurt line.
I’m uncertain about the exact travel time benefits of such a high-speed line. I put a route through my train performance calculator and got about 2.5 hours with intermediate stops at Südkreuz, Erfurt, Nuremberg, and possibly Ingolstadt (skipping Ingolstadt saves 3 minutes plus the dwell time), using the performance characteristics of the next-generation Velaro. But I’m worried that my speed zones are too aggressive and that the schedule should perhaps accommodate TGVs coming from Paris via Frankfurt, so I won’t commit to 2:30; however, 2:45-2:50 should be doable, even with some unforeseen political compromises.
But even with less optimistic assumptions about trip times, Germany should do it. If it was justifiable to spend €10 billion on reducing trip times from 6 hours to just under 4, it should be justifiable to spend around half that amount on reducing trip times by another hour and change.
Does the absolute size of a country matter for public transport planning? Usually it does not – construction costs do not seem to be sensitive to absolute size, and the basics of rail planning do not either. That Europe’s most intensely used mainline rail networks are those of Switzerland and the Netherlands, two geographically small countries, is not really about the inherent benefits of small size, but about the fact that most countries in Europe are small, so we should expect the very best as well as the very worst to be small.
But now Germany is copying Swiss and Dutch ideas of nationally integrated rail planning, in a way that showcases where size does matter. For decades Switzerland has had a national clockface schedule in which all trains are coordinated for maximum convenience of interchange between trains at key stations. For example, at Zurich, trains regularly arrive just before :00 and :30 every hour and leave just after, so passengers can connect with minimum wait. Germany is planning to implement the same scheme by 2030 but on a much bigger scale, dubbed Deutschlandtakt. This plan is for the most part good, but has some serious problems that come from overlearning from small countries rather than from similar-size France.
In accordance with best industry practices, there is integration of infrastructure and timetable planning. I encourage readers to go to the Ministry of Transport (BMVI) and look at some line maps – there are links to line maps by region as well as a national map for intercity trains. The intercity train map is especially instructive when it comes to scale-variance: it features multihour trips that would be a lot shorter if Germany made a serious attempt to build high-speed rail like France.
Before I go on and give details, I want to make a caveat: Germany is not the United States. BMVI makes a lot of errors in planning and Deutsche Bahn is plagued by delays; these are still basically professional organizations, unlike the American amateur hour of federal and state transportation departments, Amtrak, and sundry officials who are not even aware Germany has regional trains. As in London and Paris, the decisions here are defensible, just often incorrect.
Run as fast as necessary
Switzerland has no high-speed rail. It plans rail infrastructure using the maxim, run trains as fast as necessary, not as fast as possible. Zurich, Basel, and Bern are around 100 km from one another by rail, so the federal government invested in speeding up the trains so as to serve each city pair in just less than an hour. At the time of this writing, Zurich-Bern is 56 minutes one-way and the other two pairs are 53 each. Trains run twice an hour, leaving each of these three cities a little after :00 and :30 and and arriving a little before, enabling passengers to connect to onward trains nationwide.
There is little benefit in speeding up Switzerland’s domestic trains further. If SBB increases the average speed to 140 km/h, comparable to the fastest legacy lines in Sweden and Britain, it will be able to reduce trip times to about 42 minutes. Direct passengers would benefit from faster trips, but interchange passengers would simply trade 10 minutes on a moving train for 10 minutes waiting for a connection. Moreover, drivers would trade 10 minutes working on a moving train for 10 minutes of turnaround, and the equipment itself would simply idle 10 minutes longer as well, and thus there would not be any savings in operating costs. A speedup can only fit into the national takt schedule if trains connect each city pair in just less than half an hour, but that would require average speeds near the high end of European high-speed rail, which are only achieved with hundreds of kilometers of nonstop 300 km/h running.
Instead of investing in high-speed rail like France, Switzerland incrementally invests in various interregional and intercity rail connections in order to improve the national takt. To oversimplify a complex situation, if a city pair is connected in 1:10, Switzerland will invest in reducing it to 55 minutes, in order to allow trains to fit into the hourly takt. This may involve high average speeds, depending on the length of the link. Bern is farther from Zurich and Basel than Zurich and Basel are from each other, so in 1996-2004, SBB built a 200 km/h line between Bern and Olten; it has more than 200 trains per day of various speed classes, so in 2007 it became the first railroad in the world to be equipped with ETCS Level 2 signaling.
With this systemwide thinking, Switzerland has built Europe’s strongest rail network by passenger traffic density, punctuality, and mode share. It is this approach that Germany seeks to imitate. Thus, the Deutschlandtakt sets up control cities served by trains on a clockface schedule every 30 minutes or every hour. For example, Erfurt is to have four trains per hour, two arriving just before :30 and leaving just after and two arriving just before :00 and leaving just after; passengers can transfer in all directions, going north toward Berlin via either Leipzig or Halle, south toward Munich, or west toward Frankfurt.
Flight-level zero airlines
Richard Mlynarik likes to mock the idea of high-speed rail as conceived in California as a flight-level zero airline. The mockery is about a bunch of features that imitate airlines even when they are inappropriate for trains. The TGV network has many flight-level zero airline features: tickets are sold using an opaque yield management system; trains mostly run nonstop between cities, so for example Paris-Marseille trains do not stop at Lyon and Paris-Lyon trains do not continue to Marseille; frequency is haphazard; transfers to regional trains are sporadic, and occasionally (as at Nice) TGVs are timed to just miss regional connections.
And yet, with all of these bad features, SNCF has higher long-distance ridership than DB, because at the end of the day the TGVs connect most major French cities to Paris at an average speed in the 200-250 km/h range, whereas the fastest German intercity trains average about 170 and most are in the 120-150 range. The ICE network in Germany is not conceived as complete lines between pairs of cities, but rather as a series of bypasses around bottlenecks or slow sections, some with a maximum speed of 250 and some with a maximum speed of 300. For example, between Berlin and Munich, only the segments between Ingolstadt and Nuremberg and between Halle and north of Bamberg are on new 300 km/h lines, and the rest are on upgraded legacy track.
Even though the maximum speed on some connections in Germany is the same as in France, there are long slow segments on urban approaches, even in cities with ample space for bypass tracks, like Berlin. The LGV Sud-Est diverges from the classical line 9 kilometers outside Paris and permits 270 km/h 20 kilometers out; on its way between Paris and Lyon, the TGV spends practically the entire way running at 270-300 km/h. No high-speed lines get this close to Berlin or Munich, even though in both cities, the built-up urban area gives way to farms within 15-20 kilometers of the train station.
The importance of absolute size
Switzerland and the Netherlands make do with very little high-speed rail. Large-scale speedups are of limited use in both countries, Switzerland because of the difficulty of getting Zurich-Basel trip times below half an hour and the Netherlands because all of its major cities are within regional rail distance of one another.
But Germany is much bigger. Today, ICE trains go between Berlin and Munich, a distance of about 600 kilometers, in just less than four hours. The Deutschlandtakt plan calls for a few minutes’ speedup to 3:49. At TGV speed, trains would run about an hour faster, which would fit well with timed transfers at both ends. Erfurt is somewhat to the north of the midpoint, but could still keep a timed transfer between trains to Munich, Frankfurt, and Berlin if everything were sped up.
Elsewhere, DB is currently investing in improving the line between Stuttgart and Munich. Trains today run on curvy track, taking about 2:13 to do 250 km. There are plans to build 250 km/h high-speed rail for part of the way, targeting a trip time of 1:30; the Deutschlandtakt map is somewhat less ambitious, calling for 1:36, with much of the speedup coming from Stuttgart21 making the intercity approach to Stuttgart much easier. But with a straight line distance of 200 km, even passing via Ulm and Augsburg, trains could do this trip in less than an hour at TGV speeds, which would fit well into a national takt as well. No timed transfers are planned at Augsburg or Ulm. The Baden-Württemberg map even shows regional trains (in blue) at Ulm timed to just miss the intercity trains to Munich. Likewise, the Bavaria map shows regional trains at Augsburg timed to just miss the intercity trains to Stuttgart.
The same principle applies elsewhere in Germany. The Deutschlandtakt tightly fits trains between Munich and Frankfurt, doing the trip in 2:43 via Stuttgart or 2:46 via Nuremberg. But getting Munich-Stuttgart to just under an hour, together with Stuttgart21 and a planned bypass of the congested Frankfurt-Mannheim mainline, would get Munich-Frankfurt to around two hours flat. Via Nuremberg, a new line to Frankfurt could connect Munich and Frankfurt in about an hour and a half at TGV speed; even allowing for some loose scheduling and extra stops like Würzburg, it can be done in 1:46 instead of 2:46, which fits into the same integrated plan at the two ends.
The value of a tightly integrated schedule is at its highest on regional rail networks, on which trains run hourly or half-hourly and have one-way trip times of half an hour to two hours. On metro networks the value is much lower, partly because passengers can make untimed transfers if trains come every five minutes, and partly because when the trains come every five minutes and a one-way trip takes 40 minutes, there are so many trains circulating at once that the run-as-fast-as-necessary principle makes the difference between 17 and 18 trainsets rather than that between two and three. In a large country in which trains run hourly or half-hourly and take several hours to connect major cities, timed transfers remain valuable, but running as fast as necessary is less useful than in Switzerland.
The way forward for Germany
Germany needs to synthesize the two different rail paradigms of its neighbors – the integrated timetables of Switzerland and the Netherlands, and the high-speed rail network of France.
High investment levels in rail transport are of particular importance in Germany. For too long, planning in Germany has assumed the country would be demographically stagnant, even declining. There is less justification for investment in infrastructure in a country with the population growth rate of Italy or of last decade’s Germany than in one with the population growth rate of France, let alone one with that of Australia or Canada. However, the combination of refugee resettlement and a very strong economy attracting European and non-European work migration is changing this calculation. Even as the Ruhr and the former East Germany depopulate, we see strong population growth in the rich cities of the south and southwest as well as in Berlin.
The increased concentration of German population in the big cities also tilts the best planning in favor of the metropolitan-centric paradigm of France. Fast trains between Berlin, Frankfurt, and Munich gain value if these three cities grow in population whereas the smaller towns between them that the trains would bypass do not.
The Deutschlandtakt’s fundamental idea of a national integrated timed transfer schedule is good. However, a country the size and complexity of Germany needs to go beyond imitating what works in Switzerland and the Netherlands, and innovate in adapting best practices for its particular situation. People keep flying domestically since the trains take too long, or they take buses if the trains are too expensive and not much faster. Domestic flights are not a real factor in the Netherlands, and barely at all in Switzerland; in Germany they are, so trains must compete with them as well as with flexible but slow cars.
The fact that Germany already has a functional passenger rail network argues in favor of more aggressive investment in high-speed rail. The United States should probably do more than just copy Switzerland, but with nonexistent intercity rail outside the Northeast Corridor and planners who barely know that Switzerland has trains, it should imitate rather than innovating. Germany has professional planners who know exactly how Germany falls short of its neighbors, and will be leaving too many benefits on the table if it decides that an average speed of about 150 km/h is good enough.
Germany can and should demand more: BMVI should enact a program with a budget in the tens of billions of euros to develop high-speed rail averaging 200-250 km/h connecting all of its major cities, and redo the Deutschlandtakt plans in support of such a network. Wedding French success in high-speed rail and Swiss and Dutch success in systemwide rail integration requires some innovative planning, but Germany is capable of it and should lead in infrastructure construction.
I’ve been asked on Twitter about the differences between various kinds of urban rail transit. There is a lot of confusion about the term light rail in English, since it can be used for urban public transport typologies that have little to do with one another. The best way to think about urban rail (other than regional rail) is to use the following schema:
|Slow in center||Fast in center|
|Slow in outlying areas||Tramway||Subway-surface|
|Fast in outlying areas||Tram-train||Rapid transit|
In American parlance, all four have been called light rail: subway-surface and tram-train lines are always called light rail, and officially so are tramways; then one full rapid transit line, the Green Line in Los Angeles, is called light rail as it runs light rail vehicles (LRVs) rather than subways. Nonetheless, in this post I will ignore what things are called and focus on their speed.
In this context, fast and slow refer to right-of-way quality. A tramway in a low-density city with little traffic and widely separated stops may well be faster than a rapid transit line with many stops, such as most Paris Metro lines, but relative to the local urban typology the tramway is still slow while the metro is fast.
The two hybrid forms – subway-surface and tram-train – differ in where they focus higher-speed service. On a subway-surface line, the city center segment is in a subway and then the line branches farther out, for examples the Boston Green Line, San Francisco Muni Metro, Philadelphia Subway-Surface Lines, and Frankfurt and Cologne U-Bahn networks. On a tram-train, the train is fast outside city center, where it runs in a dedicated surface right-of-way, but then in city center it runs in tramway mode on the street at lower speed; the Karlsruhe tram-train is one such example, as are virtually all postwar light rail systems in the United States and Canada.
The 2*2 typology simplifies the situation somewhat. There exist lines that don’t fully obey it, and instead change between metro and streetcar mode haphazardly. Some of the Cologne lines go back and forth. Buffalo has a single light rail line without branches, dubbed the Buffalo Metro Rail, running on the surface in the center and in a greenfield tunnel farther out toward Amherst and the university campus. Frankfurt’s U1/2/3/8 trunk is the opposite of Buffalo, running in a tunnel in the center and on the surface farther out even downstream of the branch point. The Los Angeles Blue Line is underground at Metro Center but then runs on the surface, transitions to a grade-separated right-of-way later, and finally drops back to streetcar mode in Downtown Long Beach.
The most fascinating case is that of the Boston Green Line D branch. It is technically rapid transit, since the trunk line is in a tunnel alongside the other branches whereas the branch itself is a former commuter rail line; it is called light rail because it runs LRVs, like the Los Angeles Green Line, and shared the trunk with the B, C, and E branches, all of which have surface segments. But conceptually, it presages most proper American light rail lines: it was built in the 1950s as suburban-oriented rapid transit, with park-and-rides and downtown-focused service, creating a paradigm that postwar metros like BART and the Washington Metro would sometimes follow and that light rail systems from the 1980s onward (San Diego, Portland, etc.) always would.
Nonetheless, such aberrations are uncommon enough that the 2*2 simplification works when explaining what cities should be building.
Cities are more likely to build fast trains when there is preexisting right-of-way for them. The Karlsruhe Zweisystem is based on using the area’s extensive legacy mainline network, on which LRVs run in train mode, and then diverging toward city center in streetcar mode. Jarrett Walker has a good post about Karlsruhe specifically: there is no good right-of-way with which to drag the Stadtbahn into city center in train mode, and thus the alternative to a tram-train is an expensive tunnel; such a tunnel is under construction now, at the cost of about €1 billion, but as Karlsruhe is a small city, it comes a generation after the tram-train system was put into place.
North American light rail systems often use mainline rail corridors as well, but thanks to federal regulations as well as weak regional rail systems, they almost never use mainline tracks; the Blue Line in San Diego, the first tram-train in the United States, is one of very few exceptions, and even then it shares track with a very lightly-used freight line, rather than with a frequent S-Bahn as in Karlsruhe. It is more common for North American tram-trains to run in disused corridors, on new tracks parallel to the mainline, or even in highway medians.
Reusing legacy rail lines and running in freeway medians are not unique to tram-trains. Rapid transit does both outside city center; the first subway network in the world, the London Underground, makes extensive use of branches of former commuter lines, and even shares track with a still-active one on a portion of the Watford DC Line. New York, likewise, connected former excursion lines in Brooklyn to the subway, forming most of the Coney Island-bound system, and later did the same with the LIRR in the Rockaways, now carrying branches of the A train. It is usually easy to spot whether an urban rail line descends from a legacy branch line – if it does then it is very unlikely to follow a single street (none of the lines serving Coney Island does), whereas if it doesn’t then it is usually a subway or el on a major arterial (such as Fourth Avenue in Brooklyn).
The upshot is that cities are likelier to build tram-trains and rapid transit in preference to tramways and subway-surface lines if they have high-quality right-of-way. New York and London were unlikely to build subway-surface lines in the early 20th century either way, but the high density of their metro networks in Southern Brooklyn and West London respectively can be explained by the extent of preexisting legacy lines in these areas. Comparable areas that did not have such good connections, for example Queens, have much less rapid transit coverage.
While this issue in theory affects tram-trains and rapid transit equally, in practice it is especially relevant to tram-trains. Rapid transit is more expensive, so it is likely to be built in larger and denser cities, where it is more acceptable to just tunnel under difficult segments. Tram-trains are present in smaller cities – Calgary, Edmonton, Karlsruhe, and so on – as well as in American Sunbelt cities that are so auto-oriented that they have the public transport of European cities one third or even one tenth their size. In those cities, tunneling is harder to justify, so the train goes where it can go cheaply. Downtown transit malls like those of Portland and Calgary are the least bad solution for connecting fast lines from the suburbs to provide better city center coverage and connect to lines on the other side of the region.
Subway-surface lines are fast in city center and slow outside of it. Moreover, in city center their right-of-way segregation (in a tunnel in all of the American cases) means there is more capacity than on the surface. This makes branching especially attractive. Indeed, in all three American cases – Boston, Philadelphia, San Francisco – the subway-surface line has four to five branches.
Outside the United States, subway-surface branching is more complicated. In Frankfurt, the U4/5 and U6/7 lines work as in the United States, but with only two branches per trunk rather than four or five; but the U1/2/3 line has a surface segment on the mainline. In Cologne, there is extensive reverse-branching (see map), and while most of the system runs in subway-surface mode, one line runs in tramway mode through city center but then drops to a tunnel in Deutz and splits into two surface branches farther east.
Tel Aviv is building a subway-surface line from scratch, without any branching. The Red Line is to run underground in Central Tel Aviv, Ramat Gan, and Bnei Brak, and on the surface farther east in Petah Tikva as well as at the other end in Jaffa and Bat Yam. At the Petah Tikva end, an underground connection to the depot is to enable half the trains to terminate and go out of service without running on the surface; at the Jaffa/Bat Yam end, a loop near the portal is to enable half the trains to terminate and reverse direction without running on the surface.
The American way works better than the incipient Israeli way. The main advantage of branching is that the greater expanse of land in outer-urban neighborhoods and suburbs means more lines are needed than in the center to guarantee the same coverage. Thus Downtown San Francisco has just one line under Market Street, serving not just Muni Metro but also BART on separate tracks, but in the rest of the city, BART and the five branches of Muni serve an arc of neighborhoods from the Mission to the Sunset. The lack of branching on the Tel Aviv Red Lines means that it will not be able to serve Petah Tikva well: the city is not very dense or very central and has no hope of getting the multiline crisscross pattern eventually planned for Central Tel Aviv.
One implication of the fact that subway-surface lines should branch is that they are more appropriate for cities with natural branching than for cities without. Boston in particular is an excellent place for such branching. Its street network does not form a grid, but instead has arterials that are oriented around the historic city center; the Green Line makes use of two such streets, Commonwealth Avenue hosting the B branch and Beacon Avenue hosting the C branch. A light rail line following Washington Street could likewise branch to Warren Street and Blue Hill Avenue and potentially even branch farther out on Talbot Avenue to Ashmont, effectively railstituting the area’s busiest buses.
In contrast, cities whose street networks don’t lend themselves well to branching should probably not build subway-surface lines. North American cities with gridded street networks have little reason to use this technology. If they are willing to build downtown tunnels and have the odd right-of-way running toward city center diagonally to the grid, they should go ahead and build full rapid transit, as Chicago did on the Blue Line of the L.
Speed and range
Tramways are the cheapest variety of urban rail and metro tunnels are the most expensive. The reason cities don’t just build tramways in lieu of any grade separation is that tramways are slow and therefore have limited range. Berlin’s tramways average around 16 km/h; they run partly in mixed traffic, but I don’t think they can cross 20 km/h even with dedicated lanes and signal priority.
What this means is that tramways are mainly a solution for city centers and near-center neighborhoods. The tramways in Berlin work okay within the Ring, especially in U-Bahn deserts like the segment of East Berlin between U2 and U5. But in suburban Paris, they’re too slow to provide the full trip and instead work as Metro and RER feeders, providing circumferential service whereas the faster modes provide radial rail transport.
Tramway-centric transit cities can work, but only in a constrained set of circumstances:
- They must be fairly small, like Karlsruhe, Strasbourg, or Geneva.
- They should have a network of sufficiently wide streets (minimum 20-25 meters including sidewalks, ideally 30-35) through city center as well as radiating out of it.
- They should have a supplementary regional rail network for longer trips.
Tramway-and-regional-rail is a powerful combination. Zurich is based on it, having rejected a subway network in two separate referendums. However, once the city grows beyond the size class of Strasbourg, the regional rail component begins to dominate, as there are extensive suburbs that are just too far away from city center for streetcars.
Upgrading to rapid transit
It’s common for cities to replace light rail with rapid transit by building new tunnels and burying the tracks. Historically, Boston and San Francisco both built their subway-surface networks by incrementally putting segments in tunnel, which would later protect these lines from replacement by diesel buses. Stockholm and Brussels both incrementally upgraded streetcars to metro standards, calling the intermediate phase pre-metro. Karlsruhe is building a tunnel for its Stadtbahn.
However, in the modern era, not all such tunneling projects are equally useful. Subway-surface lines stay subway-surface indefinitely: they have so much surface branching that the cost of putting everything underground would be prohibitive. San Francisco activists have flirted with a plan to replace one Muni Metro surface line with rapid transit and then reduce the rest to tramways with forced transfers; this plan is both terrible and unlikely to happen. Tel Aviv might eventually come to its senses and bury the entire Red Line, but this is possible only because the current branch-free layout is already more suited for a subway than for a subway-surface system.
Tram-trains are easier to convert to rapid transit. All that’s needed is a short tunnel segment in city center. Thus, in addition to the Karlsruhe tunnel project, there are serious discussions of city center tunneling in a variety of North American cities, including Portland and Calgary (in the near term) as well San Diego (on the 2050 horizon).
Finally, tramways can be upgraded to full rapid transit more easily than to either of the two intermediate forms. A good tramway is rarely a good subway-surface system, because the subway-surface system ideally branches and the pure tramway ideally does not. Moreover, a good tramway is unlikely to go very far into the suburbs because of its low speed, whereas a tram-train’s ability to leverage high speed in train mode allows it to go deep into the suburbs of Karlsruhe, Calgary, or San Diego. The optimal place for a tramway – dense city neighborhoods following a single line – is also the optimal one for a metro line, making the upgrade more attractive than upgrading the tramway to a hybrid.
In the last year, Massachusetts has been studying something called the Rail Vision, listing several alternatives for commuter rail modernization. This has been independent of the North-South Rail Link study, and one of the options that the Rail Vision considered was full electrification. Unfortunately, the report released yesterday severely sandbags electrification, positing absurdly high costs. The state may well understand how bad its report is – at least as of the time of this writing, it’s been scrubbed from the public Internet, forcing me to rely on screencaps.
In short: the alternative that recommends full system electrification was sandbagged so as to cost $23 billion. This is for electrification, systems, and new equipment; the NSRL tunnel is not included. All itemized costs cost a large multiple of their international cost. The Americans in my feed are even starting to make concessions to extremely expensive projects like the Caltrain electrification, since the proposed MBTA electrification is even costlier than that.
But the telltale sign is not the cost of the wires, but rolling stock. The report asserts that running electrified service requires 1,450 cars’ worth of electric multiple units (“EMUs”), to be procured at a cost of $10 billion. More reasonable figures are 800 and $2 billion respectively.
Why 1,450 cars?
The all-electric option assumes that every line in the system will get a train every 15 minutes, peak and off-peak. What counts as a line is not clear, since some of the MBTA’s commuter lines have branches – for example, the Providence and Stoughton Lines share a trunk for 24 km, up to Canton Junction. However, we can make reasonable assumptions about which branches are far enough out; overall rolling stock needs are not too sensitive to these assumptions, as most lines are more straightforward.
The MBTA is capable of turning trains in 10 minutes today. In making schedules, I’ve mostly stuck to this assumption rather than trying to go for 5-minute turnarounds, which happen in Germany all the time (and on some non-mainline American subways); occasionally trains steal 1-2 minutes’ worth of turnaround time, if there’s a longer turn at the other end. Thus, if the one-way trip time is up to 50 minutes, then 8 trainsets provide 15-minute service.
To me, high-frequency regional rail for Boston means the following peak frequencies:
Providence/Stoughton: a train every 15 minutes on each branch. Service south of Providence is spun off to a Rhode Island state service, making more stops and running shorter trains as demand is weaker than commuter volumes to Boston. With this assumption, the Providence Line requires 7-8 trainsets. The Stoughton Line, with the South Coast Rail expansion to New Bedford and Fall River, each served every half hour, requires around 9-10. Say 18 sets total.
Worcester: the big question is whether to exploit the fast acceleration of EMUs to run all-local service or mix local and express trains on tracks in Newton that will never be quadrupled unless cars are banned. The all-local option has trains doing Boston-Worcester in just under an hour, so 9-10 trainsets are required. The mixed option, with a train every 15 minutes in each pattern, and local trains only going as far as Framingham, requires 14 sets, 8 express and 6 local.
Franklin/Fairmount: a train every 15 minutes on the Franklin Line, entering city center via the Fairmount Line, would do the trip in around 50 minutes. It may be prudent to run another train every 15 minutes on the Fairmount Line to Readville, a roughly 17-minute trip by EMU (current scheduled time with diesel locomotives: 30 minutes). Overall this is around 12 trainsets.
Old Colony Lines: there are three lines, serving very low-density suburbs. The only destinations that are interesting for more than tidal commuter rail are Plymouth, Brockton, Bridgewater State, and maybe an extension to Cape Cod. Each branch should get a train every 30 minutes, interlining to a train every 10 from Quincy Center to the north. About 10-12 trainsets are needed (2 more if there’s an hourly train out to Cape Cod); this is inefficient because with three branches, it’s not possible to have all of them depart South Station at :05 and :35 and arrive :25 and :55, so even if there’s a train every 15 minutes per branch, the requirement doesn’t double.
Fitchburg Line: a local train to Wachusett every 15 minutes would require around 12 sets (75 minutes one-way). The number may change a little if there’s an overlay providing service every 7.5 minutes to Brandeis, or if trains beyond South Acton only run every half hour.
Lowell Line: an EMU to Lowell would take about 27 minutes, depending on the stop pattern; 5 trainsets provide 15-minute frequency.
Haverhill Line: an EMU to Haverhill running the current route (not via the Wildcat Branch) would take about 40 minutes, so 7 trainsets provide a train every 15 minutes.
Eastern Lines: like the Old Colony Lines, this system has very low-density outer branches, with only one semi-reasonable outer anchor in Newburyport. Trains should run to Beverly every 10 minutes, and then one third should turn, one third should go to Rockport, and one third should go to Newburyport. With the same inherent inefficiency in running this service on a symmetric schedule as the Old Colony, around 10-12 sets are needed.
This is about 90 sets total. At eight cars per set, and with a spare ratio of 11%, the actual requirement is 800 cars, and not 1,450. The difference with the state’s assumption is likely that I’m assuming trains can run at the acceleration rates of modern EMUs; perhaps the state thinks that EMUs are as slow and unreliable as diesel locomotives, so a larger fleet is necessary to provide the same service.
Rolling stock costs
Reducing the cost of infrastructure is complicated, because it depends on local factors. But reducing the cost of industrial equipment is easy, since there are international vendors that make modular products. Factories all over Europe, Japan, and South Korea make this kind of equipment, and the European factories barely require any modifications to produce for the American market under current federal regulations.
It is not hard to go to Railway Gazette and search for recent orders for EMUs; names of trainsets include Talent, FLIRT, Mireo (cost information here) and Coradia. The linked Coradia order is for €96,500 per meter of train length, the other three orders are for about €70,000. A US-length (that is, 25 meters) car would cost around $2.5 million at this rate. 800 cars times $2.5 million equals $2 billion, not the $10 billion the MBTA claims.
Railway Gazette also discusses a maintenance contract: “Vy has awarded Stadler a contract worth nearly SFr100m for the maintenance in 2020-24 of more than 100 five-car Flirt EMUs.” These trains are 105 meters long; scaled to US car length, this means the annual maintenance cost of an EMU car is around $50,000, or $40 million for the entire fleet necessary for electrified service.
The actual net cost is even lower, since the MBTA needs to replace its rolling stock very soon anyway. If the choice is between 800 EMUs and a larger diesel fleet, the EMUs are cheaper; in effect, the rolling stock cost of electrification is then negative.
Why are they like this?
I struggle to find a problem with Boston’s transportation network that would not be alleviated if Massachusetts’ secretary of transportation Stephanie Pollack and her coterie of hacks, apparatchiks, and political appointees were all simultaneously fired.
There is a chain of command in the executive branch of the Massachusetts state government. Governor Charlie Baker decides that he does not want to embark on any big project, such as NSRL or rail electrification, perhaps because he is too incompetent to manage it successfully. He then intimates that such a project is unaffordable. Secretary Pollack responds by looking for reasons why the project is indeed unaffordable. Under pressure to deliver the required results, the planners make up outrageously high figures: they include fleet replacement in the electrified alternative but not in the unelectrified one (“incremental cost”), and then they lie about the costs by a factor of five.
Good transit activists can pressure the state, but the state has no interest in building good transit. The do-nothing governor enjoys no-build options and multi-billion dollar tweaks – anything that isn’t transformative is good to him. The do-nothing state legislature enjoys this situation, since it is no more capable of managing such a project, and having a governor who says no to everything enables it to avoid taking responsibility.
In the last few years, ever more serious and powerful actors have begun investigating the fact of high American infrastructure construction costs. First it was Brian Rosenthal’s excellent New York Times exposé, and then it was the Regional Plan Association’s flop of a study. At the same time, I was aware that the congressional Government Accountability Office, or GAO, was investigating the same question, planning to talk to sources in the academic world as well as industry in order to make recommendations.
The GAO report is out now, and unfortunately it is a total miss, for essentially the same reason the RPA’s report was a miss: it did not go outside the American (and to some extent rest-of-Anglosphere) comfort zone. Its literature review is if anything weaker than the RPA’s. Its interviews with experts are telling: out of nine mentioned on PDF-p. 47, eight live in English-speaking countries. Even when more detailed information about non-English-speaking countries is readily available, even in English, the GAO report makes little use of it. It is a lazy study, and people who ideologically believe the American federal government does not work should feel confident citing this as an example.
Brian himself already notes one of the reasons the report is so weak: Congress mandated a comparative study, but the report made no international comparisons at all. Instead, the report offered this excuse (PDF-p. 27):
The complexity of rail transit construction projects and data limitations, among other things, limits the ability to compare the costs of these projects, according to the stakeholders we interviewed. As highlighted above, each project has a unique collection of specific factors that drive its costs. According to FTA officials, each proposed transit project has its own unique characteristics, physical operating environment, and challenges. Some stakeholders said that the wide disparity in the relative effect of different cost factors renders cost comparisons between projects difficult. For example, representatives of an international transit organization said that because of the large number of elements that can affect a project’s costs and the differences in what costs are included in different projects’ data, projects should be compared only at a very granular level and that aggregate cost comparisons, such as between the costs per mile or costs per kilometer of different projects, are likely flawed. Some stakeholders also said that project costs should not be compared without considering the projects’ contexts, such as their complexity. For example, one academic expert contended that project costs cannot be compared without considering the context of each project, and that analysis of projects should focus on leading practices and lessons learned instead.
There is a big problem with the above statement: disaggregated costs for many aspects of urban rail construction do exist. The Manhattan Institute’s Connor Harris has done a lot of legwork comparing tunnel boring machine staffing levels and wages in New York and in Germany, and found that New York pays much higher wages but also has much higher staffing levels, 25-26 workers compared with 12. I have done some work looking at station costs specifically, and at the cost of installing elevators for wheelchair accessibility.
There is a lot of detailed comparative research about the costs of high-speed rail; the report even references one such meta-study undertaken within Europe, but omits the study’s analysis of causes of cost differences and instead asserts that it shows that comparing different projects is hard. In the interim, California contracted Deutsche Bahn to do a post-mortem of its elevated high-speed rail costs, which found that California needlessly built larger structures than necessary, explaining its cost premium over Germany.
Instead of probing these disaggregated estimates, the GAO preferred to say that they are too hard and move on.
Even without disaggregation, there are some good sanity checks one can make about construction costs. The most important is that big projects – major subway expansions, regional rail tunnels, high-speed rail – cost an appreciable amount of the government’s budget. The budget for the 200-kilometer Grand Paris Express project is €35 billion, plus another €3 billion in contributions for related suburban rail extensions such as that of the RER E. There may be future cost overruns, but they will be reported in the media, just as the current overrun has been; it is extremely difficult to hide cost overruns measured in tens of billions in a Paris-size city, and even in a China-size country it may not be easy.
Is it plausible that GPX is inherently easier to build than New York’s $1+ billion/km subway tunnels? Yes. It’s equally plausible that it is inherently harder. Second Avenue Subway runs under a wide, straight throughfare, a situation that simplifies construction. In Israel, the ministry of transportation has long mentioned the ease of tunneling under wide, straight boulevards in connection with plans to extend the second line of the Tel Aviv subway to North Tel Aviv under Ibn Gabirol, and admitted this even when it opposed the extension on land use grounds.
The most important sanity check is that in a world with several dozens of cities with a wide variety of wealth levels, land use patterns, geologies, and topographies, no city has managed to match or even come close to New York’s construction costs. New York is not special enough to be an edge case in all or even most relevant geographic variables – it is dense but no denser than Seoul or Paris, it is wealthy but no wealthier than London or Paris or Munich and barely wealthier than Stockholm, it has hard rock but less hard than Stockholm (and in Stockholm the gneiss is cited as a cost saver – bored tunnels do not require concrete lining), etc.
Moreover, the cities that have the highest construction costs outside New York are almost without exception in the same set of countries: the US, Canada, Britain, Singapore, Australia. What’s likelier – that there is some special geographic feature common to the entire Anglosphere (including Quebec) but absent from all other developed countries, or that there is a shared set of legal and political traditions that developed in the last 50 years that impede cost-effective construction? Instead of probing this pattern, the GAO preferred to wash its hands and refuse to compare projects across countries.
In lieu of making international comparisons, the GAO has engaged in extensive internal comparison. It cites aspects that have raised the costs of Second Avenue Subway above other American subway projects, such as overdesign for stations. Apparently, it’s completely legit to compare two different cities’ construction if they’re in the same country.
Over and over again, it references its own domestic standards. The GAO has 12 design standards, e.g. on PDF-pp. 51-52 and 56-60; the report mentions that existing cost estimation methodologies by the Federal Transit Administration, or FTA, meet 7 of them; thus, it exhorts transit agencies to meet the other 5 standards.
The only problem is that there is no evidence supplied that those design standards are really useful. After all, the United States has very high costs, so why should anyone trust its standards? Even domestically, the report makes no effort to bring up successful examples of low overall costs coming from following prescribed standards. Seattle recently opened a light rail tunnel built for around $400 million per kilometer, a cost that would get most European project managers fired but that is still the lowest for an American urban rail tunnel built in this century. But the report never brings up Seattle at all, never mind that New York would salivate over the prospect of tunneling at Seattle’s cost.
The real internal comparisons then are not between different cities in the United States. Rather, they’re between different stages of cost estimation for the same project. There is published literature on cost overruns, most famously by Bent Flyvbjerg and his research group. The report cites Flyvbjerg. Moreover, one of the nine academic experts it consulted is Don Pickrell, who published a seminal paper on American cost overruns and ridership shortfalls in 1990. Pickrell was influential enough that a 2009 review found that not only had cost and ridership projections improved greatly in the intervening two decades, but also there was an improvement in ridership estimate quality attributable to Pickrell’s paper.
The GAO report is not the best source on cost overruns, but it is not completely useless there. Unfortunately, it remains useless when it comes to discussing absolute costs, a different topic from relative increases. Flyvbjerg’s original paper found that the US did not have higher cost overruns than Europe; but absolute costs in the US are several times as high. Flyvbjerg’s paper found that urban rail has higher cost overruns than road projects; but when a rail tunnel and a road tunnel are built in the same city, the road tunnel is more expensive by a factor of 1.5-2.5, at least in the four-city pilot I reported in 2017, owing to the need to build bigger bores with ventilation to carry heavy car traffic.
Lazy analysis, lazy synthesis
Americans who think of themselves as reformers like to point out real problems to solve, but then propose solutions that they made up without any connection with their analysis. The RPA study is one such example: even though one of its sources (namely, former Madrid Metro CEO Manuel Melis Maynar’s writeup about low Spanish costs) explicitly calls for separation of design and construction, its recommendations include a greater reliance on design-build. The same design-build recommendation appeared in a 2008 report in Toronto comparing the costs of the Sheppard subway, opened in 2002, with those of subways in Madrid; construction costs in Toronto have since tripled, while those of Madrid have barely risen.
To the GAO report’s credit, it does not recommend design-build. It even mentions the biggest drawback of design-build: it shifts cost risk to the private contractor, who compensates by demanding more public money up front. Nonetheless, it does not follow through and does not make the correct recommendation on this subject – namely, that cities and states should cease using this approach. It buries a recommendation for in-house expertise alongside a fad for peer review of projects.
Instead of lazily proposing design-build, the GAO lazily proposes two barely relevant tweaks (PDF-p. 43):
- The FTA administrator should ensure that FTA’s cost estimating information for project sponsors is consistent with all 12 steps found in GAO’s Cost Estimating and Assessment Guide and needed for developing reliable cost estimates.
- The FTA Administrator should provide a central, easily accessible source with all of FTA’s cost estimating information to help project sponsors improve the reliability of their cost estimates.
In other words, the report makes no recommendation about how to reduce costs, only about how to tell the public in advance that costs will be unaffordably high.
Why are these reports so bad?
This is not the first time a serious group releases an incurious study of American construction costs. What gives?
I suspect the answer has to be a combination of the following problems:
- Reform factions often have a lot of internal ideas about how to improve things based on what they already know. They will cite new information if they feel like they must do so to save face, but they will not let new evidence change their conclusions. A little knowledge can be dangerous.
- Finding information from outside the US, especially outside the English-speaking world, puts Americans (and Canadians) at a disadvantage. They know few to no foreigners, have little experience with cities abroad except as tourists, and do not speak foreign languages. Even when machine translation is decently accurate, which it is in the engineering literature in European languages, they are intimidated by the idea of dealing with non-English material. The process of learning is humbling, and some people prefer to remain proud and ignorant.
- Open-ended analysis does not always lend itself to easy explanations or easy solutions. Even when solutions do present themselves, they may not flatter the people in power. Ten years ago I did not think senior management at American transit organs should be fired; today I think mass layoffs of the top brass, especially the political appointees, are somewhere between very useful and essential.
All three problems interact. For example, senior management is even less likely to be multilingual than junior staffers, who may be second-generation immigrant heritage speakers of a foreign language; thus, anything relying on foreign material disempowers the high-ups in favor of up-and-comers. The quick-and-easy-and-wrong solutions reformists seize upon if they find a little bit of knowledge let outfits like the GAO feel more powerful without actually challenging any obstructive politician or interest group, and if those solutions fail, they can always keep churning reports about implementation.
Last year, I did not know whether the GAO was capable of providing a blueprint for improving American infrastructure at lower cost. I assumed good faith because I had no reason not to. With this report, it is clear to me as well as to other observers of American public transit that the GAO is not so capable. Instead of doing what was in the country’s best interest, the people who commissioned and wrote the report delivered the minimal product that would get them kudos from superiors who do not know any better. They could have learned, or made a serious effort to learn, but that might challenge their assumptions or those of the high political echelons, and thus they preferred to say nothing and propose to do nothing.