Quick Preliminary Notes on Rolling Stock Costs
At the Transit Costs Project, we’re starting to build a database not just of infrastructure construction costs but also rolling stock acquisition costs. The database is extremely incomplete and not at all ready for public consumption, so please read the following preliminaries with the understanding that more data may reverse some conclusions. All costs below are in 2023 PPP dollars.
- The $100,000/linear meter cost for single-deck rolling stock, which I mentioned in previous blog posts, looks like it holds for European regional trains and Chinese metros. Smaller trains, such as trams, may just be more expensive per linear meter.
- The variation in costs is, as expected, much narrower than in the costs of infrastructure. $200,000/meter exists but is unusual and generally indicates something wrong happened, such as the Berlin S-Bahn stock, which suffered from a lawsuit by Alstom over losing the bid to Stadler.
- There is a rolling stock premium for rubber-tired metros in Paris, on the order of 40% over steel-wheeled metros.
- Berlin and London have expensive metro stock as well, and reading some about their production, I am left to wonder if there’s somewhat of a prima donna premium for cities of such size.
- To my surprise, in cases where the costs of a base order and options can be disaggregated, the options aren’t consistently cheaper than the base.
- There doesn’t seem to be any secular increase in real rolling stock acquisition stock going back to the early 2000s.
- The PPP conversion is dicey within Europe; there’s an integrated European supply chain, and then the question in non-euro countries is which PPP rate to delfate to, which makes a big difference where the PPP conversion differs substantially from the euro’s, in one direction in Switzerland (with its cross-border lines, to add a complication) and in the other in Eastern Europe. In contrast, it’s easy in China, with its domestic supply chain, and in the US, with its pretense of one.
- American costs are near the high end, and rising gently, but even the New York City Subway orders of the 1990s and 2000s had somewhat of a premium – I’m getting $148,000/meter for the base R142/R142A (source) and $120,000/meter for the base R160 (source) with higher costs for the options. Nonetheless, at worst the cost premium for the R211 ($160,000 base, $145,000 option) is a factor of 1.6, not the factor of around 10 we see for infrastructure.
- Some orders come with maintenance and some don’t, and we try to disaggregate this when we can, but sources are inconsistent on whether they mention maintenance bundles in the contract; in at least two cases, the maintenance contract drives up the cost premium, but in some others (like rubber-tired Parisian trains), the premium is not about maintenance.
Pedestrian Observations 
Yep – for the times they are a changin (Paul Simon song) !!! from the very simple wooden horse or mule or donkey pulled coaches to the high-tech rolling i-pads of 2024 that we all see today, and lots of advances features started to appear beginning with the R10 cars of 1948.
So the usual London narrative “we have to buy premium because of our smol tunnels and loading guage*” is just cope/excuse as usual**?
*weird 3rd rail.
** add the assembly plant in Goole. Which is one of those tells about how London is exploited by stealth so that Outer Britain can pretend they aren’t anything but mediocrities.
Berlin doesn’t have particularly standard requirements either from my understanding. The question would be how do these compare to other non-standard systems like Hamburg.
It seems to me that it’d be more useful to know the cost (and benefit) per rider (or unit of capacity/yield) rather than equipment length. Do you assume that the variation in capacity per unit of length is small enough as to be negligible, is the data to hard to parse that way, or is there another reason to normalize by length?
The variation in capacity per unit of length depends on interior configuration decisions that don’t really matter to cost, namely optimizing for seating vs. standing space.
Width matters for capacity too but seems to have a negative relationship with cost so far, if only because the costs of London, Paris, and Berlin subway equipment are high.
Other than a small amount of shell material, railcar width differences (within normal range) probably have no significant effect on manufacturing cost. Also, the shell probably is made in a low cost location, whereas customization is done in a higher cost location.
It think it depends on the scale of the production. Rolling stock manufacturers clearly want to limit dimensions/guage variety as much as possible, hence lots of the turnkey systems being built with foreign capital have incompatible gauges with legacy networks, this is most obvious in India with the new Shinkansen-HSR line being standard rather than broad (ditto metros). Any variety that can’t be easily mechanized (paint job) imposes costs. The width material itself is the not the main problem is how much you have to change the machine tools/production process.
Anybody who reads about other vehicle production finds similar issues, variety is cost, hence economies of scale are huge. But rail has the added complication of being very demanding in its specifications for dimensions due to track/platforms/tunnels etc.
The sad thing is that gauge is modular, because the European vendors are used to having to sell to Spain and Russia. India just has really bad cultural cringe.
Hey, JICA offered really generous loan terms.
That caste/religious group exclusion are the two most powerful forces in Indian society.
The worst part is that for HSR broad gauge is preferable. For a given curvature the broader the gauge the higher speed you can achieve, conversely for a given speed you can use a tighter curve (and thus lower construction cost). This is why the Shinkansen is standard gauge while the legacy Japanese network was narrow gauge-the top speed of narrow gauge trains on the expected curves for the Tokaido route would have been too low. The Indians are not only giving up the possibility of extending HSR trains onto legacy track for better access, but also making their HSR slower/more expensive with standard gauge.
Brunel was right etc etc.
It is probably going to blow up in Japan’s faces given the potential scale of the Indian market. Given that rail systems are necessarily bespoke to the country at some level, learning how to adapt to their needs is actually important, than simply trying to make everybody Japan. I think a big problem is that because so much of the Japanese railway system is vernacular/tacit they can’t really imagine other operational philosophies. A mix of Galapagos aging established corporatism and engineer-narrow-brain.
Also don’t underestimate how much the less capacity is part of the point. The Gujarat Model is “pogrom then ghettoise Muslims/others in order to make granting planning permission to business palatable”, and adapted to the national level by Modi is about leveraging Geopolitics and the shortage of demand globally to get foreigners to lend you money/tech on the cheap. And making sure the Pogrom-PM has photo-ops with Foriegn leaders.
India has also a Westminster style generalist civil service plus massive political discretion in every decision level. Agency staff are low in the hierarchy compared to generic IAS officers let alone politicians, even these agencies often have had India’s best institutions like the Electoral Commission, the Indian armed services and Indian Railways. And its slowly getting worse as Modi and the BJP attack anything that isn’t like them.
I honestly don’t think they’re deliberately reducing capacity…
No of course not.
They just want the “international standard” the same way they did with the metros that massively underperform on ridership. The main critics within India usually say “the current system of slow railways for the poor, planes for us is fine”. They don’t think about capacity at all which pretty much the same thing, given India’s scale.
Cultural cringe is deeply tied to Indian social hierarchies and political ideologies. See classic Congress Non-aligned “socialism” which suppressed social change for a generation, or the BJP’s “lets turn Hinduism into something more like Islam because Hinduism is the bestest ever, but also weak”.
Unless India really thinks they are going to do a high speed rail tunnel under the Himalayas to China broad gauge is the right choice for their high speed rail network.
Also I do wonder how quick Brunel’s broad gauge line would have let you go. I wonder if you could have managed 300km/h on it.
“Brunel was right etc etc.”
Transit types love to hate on BART for using Indian Gauge, but the short answer is yes, Brunel was right, to a point. The broader your gauge the faster you can go on a given curve and you gain greater stability/cargo capacity. The disadvantage is that broader gauge imposes a higher minimum curve radius (this is why mountain railways from Switzerland to Nepal have traditionally used narrow gauge, to twist and turn through curvy landscapes.)
Given that modern railway development is 300+kph HSR lines and containerized intermodal freight (which can be double stacked on flat cars with broad gauge but require well cars with standard gauge) the world would undoubtedly be better off if Indian Gauge was the default railway standard (although Brunel’s 7 foot gauge might be too wide).
“Also I do wonder how quick Brunel’s broad gauge line would have let you go. I wonder if you could have managed 300km/h on it.”
If you are asking if the Great Western Railway as built in the 1830s/40s could support 300kph, the answer is no, certainly not. Several technical factors go into supporting HSR speeds, principally curve radius and cant, vertical curve radius, and rail tolerance. Broad gauge allows higher speed on a given curve by allowing for greater cant (because the wheels are set wider apart, you can cant/tilt the track more than with standard gauge.) But even with greater cant HSR still requires very large curve radii, thousands of meters. Vertical curves must be even larger (tens of thousands of meters – taking a horizontal curve too fast causes one to be shifted to the side, taking a vertical curve too fast can throw you into the ceiling). The rails for HSR have to be very precise, continuous welded track with no joints, no vertical misalignment, etc. – a small defect can cause a train to derail at hundreds of kph. The GWR in the 19th century would never have been built with these standards in mind: curves would have been much tighter, cant much less (thus negating/ignoring the speed advantage of broad gauge), vertical curves/acceleration perhaps not considered at all, manufacturing technology couldn’t produce the kind of precision required.
Note that Brunel’s reasoning for his extra-broad 7’ gauge was to place the wheels outside of the carriages (this is back when railway carriages were the size of horse drawn carriages, not modern railway cars). This allowed the wheels to be much larger because they did not have to fit under the carriages. His thought was that the larger diameter wheels would provide a smoother ride. The speed/cargo advantages of broad gauge we know today were not on his mind.
The limiting factor to cant is what happens if a train sits still on the tracks, which is why cant limits are pretty consistently slightly higher than the maximum cant deficiency. On passenger rail, the cant deficiency limit comes purely from lateral acceleration’s effect on passengers within the carriage; the train has enough stability to run at a lateral acceleration approaching 2 m/s^2, or 300 mm cant deficiency on standard gauge, if it tilts enough that the passengers don’t feel such high forces. Thus, track gauge does not matter to maximum lateral acceleration, and therefore it doesn’t matter to maximum cant either if it is expressed in units of acceleration; the physical cant is higher with a broader gauge purely because the same cant angle corresponds to a larger cant measured in units of distance, scaled with the gauge.
On freight rail, the limits to cant deficiency are about the train’s stability on curves. The center of gravity on a diesel locomotive is much higher than on an EMU; the center of gravity on a double-stacked freight car is also higher. This means that double-stacked freight in India can run more easily than in the US or China – it can run on flatcars and not just well cars, because the broader gauge allows more crosswind stability as well as higher lateral acceleration on curves.
The curve radius on the great western mainline is pretty gentle to be fair.
All the other points are interesting for sure and quite right!
”The limiting factor to cant is what happens if a train sits still on the tracks,”
”the physical cant is higher with a broader gauge purely because the same cant angle corresponds to a larger cant measured in units of distance,”
It isn’t just that physical cant is higher because with a wider gauge you get a larger cant relative to cant angle. Cant angle is also higher for the reason you state above and then below for freight: the wider gauge allows for more lateral stability and a higher cant angle before ill effects set it (just as a wide book is more stable and resistant to tipping over than a narrow book). Since it is cant angle that determines limiting speed (referring to cant as a distance is a convenience for railroad engineers who once upon a time could more easily measure distance than a shallow angle*) broader gauge does in fact allow for higher speed on a given radius curve, just as a highly banked racetrack allows for higher speed than a less banked highway curve.
The limiting factor to cant deficiency on passenger trains is not the risk of tipping, though – it’s passenger safety on a moving train and to some extent passenger comfort, and those don’t depend on the gauge.
The limiting factor to cant deficiency isn’t tipping, but that the limiting factor to cant itself is (more properly to theta, the angle of the tilt). If you can increase the cant/theta/angle of tilt due to a wider track gauge, then you have reduced the cant deficiency at a given speed. If you keep the same deficiency (which as you note you can because it is effectively a constant due to human physiology) then you can run at a faster speed through a given turn, because the effective total cant (theta plus deficiency) is greater than with a narrower gauge train.
Bear in mind, the Métro and Berlin U-Bahn trains are smol too, and that could be the issue; we’ll know more when we have a more complete database. It is kind of weird that the cost per meter of length looks lower on wider trains than on narrower ones…
How much of that is wider trains typically being more standard and having much higher order volumes?
Within a country, mainline/mainline-ish trains are used all over with relatively little variation, but trains for small loading gauge metros are bespoke for a single city, and it’s not uncommon for a city to have several different incompatible small loading gauge metro types.
Globally, the most high speed trains are Shinkansen loading gauge, and Shinkansen loading gauge trains are split across fewer models. There were more N700 trains built than every non-Chinese Velaro variant, and CR400AF and CR400BF order volumes must certainly blow even N700 out of the water. Since China also started using Shinkansen loading gauge for low speed mainline as well, the most common loading gauge of any train worldwide might be Shinkansen loading gauge.
Big-city metro orders are huge; there are orders clocking in at well over 1,000 cars, like the J and JK Class order here, or the R160 in New York, or the S Stock in London.
Now they are taking delivery of R211’s and they have very few seats, and 4 sets of 58″ wide doors on each side, and the passengers are still BLOCKING the doors, making entrance and existing these cars difficult as well. Also, with the heatwave, the cooled air escapes from these larger doors much more quickly, and they feel HOTTER as a result. I rode them a few times this week on the (A) & (C) lines, and the air conditioning system works so much harder due to a term called “Heat Loss” because the cooled air gets dissipated much quicker than on cars with 54″ or 50″ wide doors, so therefore the R211’s will feel warmer and stickier as a result.
1000 is not huge. Automobile factories regularly put out that many cars in a single day (over 2 shifts). Those factories are only putting out one model (with limited variations) for the entire year.
Which is one more reason we need to standardize – the more trains are the same the more manufactures can build custom factories with high up front costs for special jigs. Or as I proposed a few weeks ago, you shouldn’t have a request for proposals to get a train – instead you should order from the catalog of models the manufacturer provides. Of course the final price will have some custom charges to program the digital signage and the color scheme will also require some custom work. You can only choose the seat layout the manufactures chooses to offer in the catalog, if you want something different you are not allowed to ask for it (a different manufacture may have something closer to your liking in their catalog). Your doors will be in predefined positions that are the same across all manufactures for that train style (height and width) – which will in turn fit your platform screen doors. Some manufactures will offer a restroom option some will not.
It is the other way around. China (and Korea) low speed lines have been using the Asian loading gauge for decades before the Shinkansen was launched.
Maybe China, but that’s really hard to believe for Korea, considering all their urban, suburban, and high speed trains can or almost can squeeze into UIC loading gauge.
Well, mini-skirts are not particularly cheaper than long ones…
If smaller non standard trains cost more – and I believe the classic compatible HS2 trains are more expensive than the non classic compatible ones then it’s difficult to believe the non standard tube trains with non standard 4 rail electricity supply aren’t more expensive due to that.
If the S stock were particularly expensive too then perhaps there is a special premium, but they are also have a non standard 4 rail electricity supply.
OK so it looks like as per https://en.wikipedia.org/wiki/London_Underground_S7_and_S8_Stock that there are 23500 linear metres of S stock and that they cost £1.5 billion, adjusting for inflation and using the standard 0.7 conversion means they come out at $100,000 PPP 2017.
So the tube trains may be more expensive but that is size and the sub-surface stock costs the same as everyone else’s.
The 2024 Stock comes with 40 years of maintenance, whereas the S Stock doesn’t, I don’t think. That could be the difference.
Crudely with cars the lifetime cost is 1/3rd fuel, 1/3rd the purchase cost and 1/3rd maintenance and insurance.
So first approximation would be that the 40 years of maintenance would double costs. I am sure you have access to better data than that pub level approximation though.
How does that 40 years adjust for inflation? I’d consider anyone a fool to accept a 40 year fixed price maintenance contract unless they use the market to hedge the inflation risks. I’m not sure what options the markets even provide for a 40 year inflation hedge, but there is enough risk there that is a good deal.
Or in other words I’d expect trains to come with 10 years maintenance/warranty, and a contract that all molds will be kept for 40 years and parts can be bought on a time+materials basis. (someone who works in supply management will be able to but that into better legal terms and know details I don’t) The important part is inflation style risks will not be one the private sector.
Rubber tires don’t result in a 40% premium. The MP14 were often cheaper than the steel wheeled MF19. The first batch of MF19 came at 171521 Euros/metre in 2019 and 103 five car trains were ordered in 2024 for 103 560 Euros/m. The line 14 driverless MP14 cost 123 333 E/m in 2014, and line 4 trains 90 555 E/m in 2017. Ligne 11 MP14 cost 104 667 E/m in 2018 and 92 632 E/m in 2021.
A good point, though I assume the analysis looks at lifecycle costs, and as I understand it, wheel replacement is a much more common occurrence on rubber-tired vehicles.
A steel wheel may only cost about $1000 but its installation is a difficult and lengthy process. WMATA is expected to spend $55 million and 3 years to re-press the wheels of its 7000-series cars. That’s $9162/wheel. Metro tires should be put in a cage when inflating them at the standard high pressure but are otherwise easy to change.
Specific circumstances like track curvature, cleanliness and maintenance, and performance of the metro will determine how often the wheels will have to be changed but it may be sooner than expected. In the 1970’s, WMATA discovered that its steel wheels lasted only 150 000 miles in average, half as much as expected, and less than the 200 000 miles of Montreal’s metro tires. The Ansaldo-Breda driverless metros used in Copenhagen and Italy ran into the same problem. In summary, steel wheels last longer but only if you don’t ask much from them.
Don’t forget that rubber-tyred Metros also have standard steel wheels and tracks. The wear on the steel wheels should be less compared to standard train (so partly compensating for the extra maintenance of rubber tyres?). Some rubber-tyres systems apparently have the steel wheels slightly above the track so that they only engage in switchings and crossings. I don’t think that is the case for the Paris trains because you can hear the squealing (of steel) on all those tight turns …
I was comparing with the MF01s… the MF19 option order looks like 1.1G€ for 103 trains of 76-77 m, so around 140,000€/m.
Sapporo metro which also uses rubber wheels also consistently reports elevated rolling stock costs (and operations costs).
Yes, but in a developed country, rolling stock and its power needs represent a small portion of a subway’s assets and costs. The question is whether spending more on the moving assets by running faster tyred trains makes the main fixed assets work harder i.e. increase the annual ridership per km of line. The answer is yes. Pre-Covid, in the Americas, tyred subways carried about 7.888 million/km and steel wheeled 3.186 million/km. In Europe, the averages were 5.606 million/km for tyres and 3.882 million for steel.
This phenomenon was exploited at both ends of the range in France. Paris Ligne 1 (11 million/km) was converted because the steel wheel trains could not cope with the demand, and 26 meter long VAL driverless mini-metros were successful in small cities.
Some of this is presumably the extremely high usage of the Paris Metro in Europe (especially per kilometre). Plus in the Americas the Santiago, Montreal and Mexico City systems are pretty good in general – and only the New York Subway is otherwise similarly good.
No, the subways with highest density ridership do not explain these numbers. On the contrary, they actually run on steel wheels with the notoriously undersized Sao Paulo and Cairo metros on top of the world in that respect and, in Europe, Praha and Budapest slightly above Paris metros sur pneus.
The lower average of conventional subways is caused by the large number of networks with a middling ridership or, if one looks at post-covid US numbers, failing ridership. These networks respond to that challenge by cutting frequency resulting in slower service to their shrinking captive clientele.
A lot of that is probably them not understanding the commercials of a rail service allowing extremely high frequency even with relatively low usage.
I mean SNCF doesn’t get that. So it isn’t a surprise that smaller cities don’t either.
it’s because they have no money, and nobody lives near most American rapid transit stations. Making statements about technology based on the simplistic method outlined here is absurd.
@ Nilo. You are right about the situation in the US and this is why I am not suggesting that adding rubber wheels will fix it. As for the “simplistic method” it is actually standard. That’s how transit agencies evaluate alternative bus or rail routes before deciding where to invest. It beats having no method at all, for example saying rubber wheels add costs and therefore are bad without trying to measure eventual benefits. I used averages over continents to minimize the effect of local peculiarities. If you have a better method I will be happy to try it.
The quick cheap win would be to run service every 15-20 minutes on existing railway lines using standard EMUs or even DMUs.
Matthew is as usual more polite than me. Those figures are junk considering the selection bias i.e. ratio of megacities to small cities. That most of the world that doesn’t have cultural cringe to Paris hasn’t chosen the technology is probably a big red light. Maybe it has a niche places with climbing steeper gradients and that’s about it.
Ginza line with steel wheels manages 25.7 million annual passengers per kilometer*. It has lower frequency but that’s mostly because its so busy they’d have to rebuild all the stations to deal with more passengers.
Considering how effort Paris has had to make to fix the problems of the original pre-RER Paris metro of which rubber wheels is hardly the worst, I’m not sure it deserves your praise.
*Use daily please, annual is used to deliberately mislead with biggest sexiest numbers.
Using passengers per route kilometer rather than ridership density (passenger kilometer per route kilometer) also favors lines that are used for shorter trips, which tend to be lines with tighter stop spacing and just physically shorter lines. 42nd Street Shuttle in NYC has like 50 million passengers per route kilometer per year.
Passenger-km figures would indeed be better but they are only publicly available in rare cases like the 42nd St shuttle. The metric I used indeed favors lines with tighter spacing between stations, a domain where rubber wheels shine, but note that even with fairly distant stations like in Montreal (average interval 896 m) one can still find benefits in terms of average speed (36.76 km/h) in comparison with Toronto (interval 950 m, average speed 31 km/h).
I like yearly numbers because there is no ambiguity. With daily numbers one may be quoting a yearly average or peak/working days. I was just trying to dispel the common stereotype, the belief that rubber metros are inherently wasteful “gadgetbahns”. The only wasteful subways are the ones with poor patronage.
You suggest that the lower ridership figures of steel wheel subways reflect a selection bias in the ratio of megacities and small cities because you believe in another dated stereotype, namely that metros only work in large cities. The stereotype is not baseless. Rubber metros are busier in Paris and Mexico City but they also pick up more passengers, per km of line, than the average London Underground, NYC subway, Boston T or Chicago El in Lausanne, 140 000 inhabitants, and Rennes, 300 000. This was unexpected. Both cities and Lille, where the mini metro phenomenon started, are trying to add capacity. Lille is particularly interesting because the VAL serves a large low density area where buses and tram had a low ridership, even by French standards of the time. It may be too French for your taste but there is nothing niche in a technology that works in settings as diverse as Lausanne, Lille and Mexico City, and the automation part has spread all over the world.
Lille’s passenger numbers are certainly very good compared to the Manchester Metrolink for example, but compared to the transport systems in Paris or London the ridership isn’t that strong.
Lille public transport figures went from 48.5 trips/inhabitant in 1982, before the VAL opening, to 88 trips in 1991 (partial opening of the second VAL line). This is still modest, not only in comparison with London and Paris, but also with Marseille (201) and Lyon (275), agglomerations of similar population, around 1 million inhabitants, but concentrated on 1/3rd the area served by Lille’s transit agency.
Annual Figures have tons of ambiguity, that’s why crappy agencies love them.
You continue to gerrymander. Tube lines go into the suburbs because people live there which kills relative passenger density. Paris metro is designed not because the 3rd-4th Republics were less democratic and more unequal than Britain in those years (not true of the 5th Republic). It took the the first 30 years for the RER system to surpass the classic tube despite London being subject to decades of deliberate destruction by the state 1945-?.
I guess we’ll see when Alon puts the figures together, I have my prejudices you have yours. And I have plenty of respect for the post-1970 French transit system over the Anglo-saxon ones. Investing in RER, the TGV, orbital Paris tramways, subway line automation etc etc. That doesn’t stop the rubber fetish, the locomotive fetish, the non-boarding level platforms, the airline timetabling and the inability to use the TER system for s-bahn services from being stupid.
Or that Marseilles makes the North of England look good given it has good basic economic geography ruined by the weaknesses of French urban and transit planning plus a big dose of classism and racism. North of England never had a chance given its crappy geography. Provence should be as wealthy as Lyon or Ile de France given its position along major communications systems plus nice weather. And don’t get me started on Paris and Lyon being Nimby softcore apartheid cities.
Rubber wheels for new build systems aren’t gadgetbahn so much as gimmick-bahn. Like standard-guage Indian metros, or the Tube’s stupid 4 rail electrification systems. Sure you can make it work well, but you cause trouble for no apparent reason other than you can’t admit the original WW2 Paris Metro wasn’t especially good.
@Borners : My prejudice is that London is uniquely The Railway Metropolis, the world city who said no to urban expressways and whose daily life has been entwined with the railways for almost 200 years. I am not going to criticize this or that because nothing comes to mind and Londoners can do it better.
My intention was never to compare specific metros such as London and Paris. They are different animals living in different ecosystems. While London Underground often merge with the railways and has some of the functions of the RER, Paris built small metro tunnels to keep the railroads at bay and reaching out to the suburbs was not a priority. This is why I presented my data as averages across continents.
The 4 track electrification is because London has metal water and gas pipes that aren’t electrically bonded properly because they pre-date electricity and there was leaking of current from the running rails.
But yeah obviously in an ideal world you wouldn’t use a 4 track electrification system and you would build the tunnels big enough to carry a larger train than a tube train.
What is the ambiguity of annual figures? If anything it’s less ambiguous that daily, which can in the fine print mean daily (weekday) figures or daily (in a particular month, which due to a major event made the system busier than normal) figure.
Annual is more clear about what is being measured, is less affected by one-off events, and is inherently unaffected by seasonal trends. And isn’t the gold standard of daily figures, the daily average over the year, effectively annual/365 anyways? There is error introduced by leap years, which a proper daily average would account for, but that error gets lost in actual long term trends, and even just differing numbers of weekdays each year, anyways.
Daily is an easier number to think about though. It’s easier to turn a daily number into intuitions like “on average there is 0.xyz round trips per person living here per year” but the good daily number is almost perfectly correlated with the annual number.
@ Sassy. Why would you start from daily numbers to evaluate the number of trips per inhabitant per year ? Dividing the yearly ridership by the local population is easier.
@ Borners. One of the problems facing the transit agencies in the post-Covid/hybrid work from home and office era is that the 5 week days are gone. We now have 3 week days separated from the week-end by shoulder days. In the spring 2024, ridership on Washington Metro was roughly as follows : Mondays 380 000 ; Tuesdays 460 000 ; Wednesdays 460 000 ; Thursdays 450 000; Fridays 380 000 ; Saturdays variable between 240 000 and 410 000 ; Sundays variable between 185 000 and 311 000. What was the daily ridership?
Passenger railcar manufacturing is pretty primitive. One of the RM Transit videos shows workers hand drilling holes. Manufacturers promote customization as do consultants. It would be interesting if US DOT were to adopt some standard designs and say, “We’ll pay for this.”
Connecticut DOT (CTDOT) has ordered 60 unpowered passenger railcars at a cost of $315 million. ($5.25 M/car). (That cost-per-car is why I suggested creating a railcar cost database).
CTDOT has told me that “contract includes design, construction of the cars and warranty coverage. It does not include maintenance.” At a cost of $100K per meter, each car in that order would be 52.5 meters long. :>)
Please include in the database:
We’ll see more as we include more trams and more small commuter rail car contracts. So far the high-floor Berlin and Munich S-Bahn trains are not cheaper than the low-floor commuter trains, and the Berlin ones are considerably more expensive (and also unique in a number of ways, some justifiable, some not).
So far everything in our database is an EMU; we’ll have to figure out what to do about trainsets with separate power cars later.
When I took a quick look into rolling stock costs, I didn’t see much difference between low floor and high floor in Europe. However, in Japan, I think high floor is like half the cost per meter of low floor.
Could be an economies of scale issue- low floor rolling stock being only used on trams/light rail systems domestically, and the manufacturers are the smaller firms Niigata Transys and Alna Sharyo (née Alna Koki), rather than the “big 5”.
High floor railcars are simply a box on bogies/trucks. Low floor railcars are more complex in both the body and the bogies/trucks. See https://bqrail.substack.com/p/no-low-floor-railcars-for-the-interborough .
@bqrail
That’s the theory, and from your link, it seems based on that it is reality for maintenance costs including in Europe. However, it seems like for purchasing trains, European manufacturers are either not realizing the potential for cheaper high floor trains, or not passing along the cost savings.
In Canada orders seem to cluster around C$200,000 (2023) per metre. At nominal conversion this seems high but not exceptional; at PPP it is alarming. Toronto’s wide, long TR, Montreal’s rubber-tired MPM-10, Vancouver’s automated Mark V are all very close, and several tram orders are very close. I was surprised to see such consistency, although an in-progress Toronto procurement estimates about C$300,000 per metre, by any means very expensive.
PPP comparisons don’t seem to make sense, as countries frequently export rolling stock and many importers are not demanding on customization. India, Japan, and China in particular export rolling stock often. It wouldn’t make sense to compare car or airplane prices by PPP either.
It seems that rolling stock prices should be more comparable at nominal prices than PPP. Countries should import rolling stock if they can’t make it efficiently themselves, as having more rail infrastructure will be a benefit the industries they have a comparative advantage in.
China, Europe, Japan, and the US have domestic supply chains for this, is the main issue. But you’re right, the PPP comparisons within Europe get hairy, yeah…
India has a much lower PPP level than these countries/regions, so their rolling stock (and the ones they export to Quebec, Bahrain, and Australia) will seem very expensive at PPP prices but extremely cheap at nominal prices.
https://themetrorailguy.com/2023/05/19/beml-wins-bangalore-metros-318-coach-rolling-stock-contract-5rs-dm/
These would be great nominal prices but terrible PPP prices.
Turkey also manufactures some rolling stock, it would be interesting to see what their costs are and if they are similar to the rest of the EU customs union.
I think whether PPP or nominal matters more depends on whether you are looking at it from the perspective of a rolling stock manufacturer looking to reduce cost of production, or from the perspective of a rolling stock buyer looking to reduce cost of procurement.
A lot of rolling stock buyers have no domestic supply chain, and some countries with domestic supply chains, e.g., Indonesia, still import a decently large portion of their rolling stock. And even with domestic supply chain, a lot of the cost, especially in poorer countries might be in importing components that cannot be made domestically.
And since rolling stock is relatively easy to ship, nominal costs can be cross shopped. If it’s cheaper in nominal terms for NYC Subway to order railcars from Kawasaki Kobe than Kawasaki Yonkers, they would still be saving a ton of money by doing so, regardless of whether the price of Kawasaki Kobe is cheaper in PPP terms. This is basically just the implication of your “free trade in rolling stock is good” argument in how to look at price.
And since rolling stock is relatively easy to ship, a look into rolling stock costs should also at least look at used rolling stock. While the price of new rolling stock is probably more important to understand, the price of used rolling stock is quite relevant for developing countries and EU open access operators. And used rolling stock should definitely be looked at in nominal terms.
Very True statement. Kawasaki Railcar has to comply with these 1983 Buy America Act so that some work on these cars are done at Lincoln Nebraska, and Yonkers New York as per agreement, since UMTA (now FTA) funds are involved. However the body shells are still produced in Kobe Japan and Nantong China to expedite delivery of these railcars for their US clients. The same can be said for Alstom which produces their body shells in Lapa Brazil and finishing work is done at Hornell New York, which I worked at when it was owned by the defunct Morrison Knudsen Company rebuilding many NYC subway cars there as well.
Can you see the cost difference between very common products (say Flirts) vs. more specialised rolling stocks (the RhB’s Capricorns) or even smaller series (the SZU Be 510).
Our database isn’t there yet.
Regarding what you said in the comments last post about land costs, I wasn’t just talking about land costs in the US, I was talking about land costs among all the countries studied in the project. If gaining planning permission etc. is considered a “land cost”, are land costs still a minor portion of costs worldwide? Does this apply to countries like the UK?
Planning costs are included if they’re contracted out, or if they’re in the force account; if there’s a permanent civil service then they might not be, but the numbers I’ve been quoted in some low-cost environments are very low. In Spain, I was told that ADIF has permanent in-house staff for planning and that’s around 5% of project costs with some additional costs contracted out – and that’s for the entire planning process, not just the permission to build. It could just be that the UK makes that one step impossibly difficult. Then again, the UK also has laws favoring the state when it comes to safeguarding routes – in Italy safeguarding is for five years, not decades, and in the US it’s considered a taking so the agencies in theory have to pay for it and in practice don’t really do it, leading to private encroachments that sometimes scuttle projects (like the Downtown Extension in San Francisco).
UK is unusual because the “development control” planning system is so discretionary and politicised at every detail. Government can ignore it but it requires constant political will and cost, and the trajectory since 1947 is overload it with more and more requirements because it was cheaper than taxing and spending and learning. Its a para-legal system with pdf trench warfare not so much a “planning system”.