Premium Cost, Substandard Quality Locomotives

I’m a little late to the game here, but let me just say that Amtrak’s just-funded contract for new electric locomotives is supremely expensive: $560 million for 70 locos, or $8 million each, $466 million for 70 locos, or $6.7 million each (see comment by aw with this link). The locomotives are an FRA-compliant version of Siemens’ EuroSprinter product, which has recently been sold in Europe for €3.74 million per unit, as has Bombardier’s competing TRAXX locomotive (in fact, the TRAXX even sold for €3.2 million). Amtrak is paying a premium of about 60-80% 35-50% for these locomotives, depending on exchange rates.

It gets worse. The new locos will enter service in 2013, just two years before the national mandate for positive train control goes in effect, allowing trains to be lighter and avoid the most onerous FRA regulations (in fact, the Northeast Corridor, where most of the locos are to run, already has a PTC system). The special modifications and design are what caused an increase in both weight, from 86-87 metric tons for the standard EuroSprinter to 97 for the Amtrak Cities Sprinter, and cost.

To put things in perspective, Sweden recently bought 180 km/h EMUs for €1.6 million per car. And the 700 Series Shinkansen cost $2.5 million per car. In other words, Amtrak could have gotten 3 EMUs for the price for one locomotive. (Amtrak’s new single-deck coaches cost $2.3 million per car, the same as EMUs abroad.)

The US Department of Transportation is announcing that “Siemens Industry USA is adding 250 new manufacturing jobs in order to design and build 70 new energy-efficient locomotives for Amtrak.” The cost premium works out to about $200-250 million $100-150 million, or $800,000-1,000,000 $400,000-600,000 per job added; the total cost is $2.2 million $1.9 million per US manufacturing job. Needless to say, most of this money is not going to American manufacturing workers, but to consultants and Siemens’s train designers.


  1. Danny

    A year ago the cost was supposed to be $466 million…I wonder what caused the spike?

    I know that the FRA shares a large part of the blame, but this is exactly the kind of madness that makes me never want to support any subsidies to Amtrak. Given that operating costs are miniscule compared to capital costs for rail, this contract alone could very well mean that we are getting half the rail transportation that we would be getting if we demanded more accountability from our government’s transportation agencies.

  2. Alon Levy

    Stephen: the NEC’s PTC system was buggy and only started to come into service in 2000-1, after the Acela had been manufactured. In addition, it’s still only installed on about half the NEC, though Amtrak wants to extend it to the other half (link). Why it’s not speeding this up so that it can run lightweight trains… that’s another question.

    Also, there’s a huge amount of inertia in the rolling stock, since coaches and locos aren’t replaced at the same time: right now Amtrak needs to replace the AEM-7 locos but can keep the Amfleets, and presumably in 10 years it’ll need to replace the Amfleets but have the new Sprinters by then. The TGV uses locomotives and so do Talgo products, for more valid reasons – Talgo because its passive tilt system only works with unpowered axles, and the TGV because on the one hand EMUs require more maintenance, and on the other the advantages of EMUs (acceleration, lower axle load) are less applicable to the TGV’s operating model (few stops, uniform 17 t axle load).

    • Stephen Smith

      Haha, I love the note at the end of the Powerpoint that says “It takes time to do these things right, and to integrate them into the fabric of a very highly developed, safety sensitive, RR culture.”

      So anyway, this is what I understand from what you’re saying – electronics should be rushed and, wherever possible, new rolling stock purchases should be postponed until that happens. I guess the question is, if all the policy stars aligned, would there be enough time to install PTC up and down the NEC before Amtrak has to make these purchases, or is it already too late?

      • Alon Levy

        It’s not really too late – Amtrak needs full PTC by the end of 2015 and seems to be on track to make good on it. The AEM-7s’ problems are not so awful that Amtrak can’t postpone rolling stock purchases for two more years, and spend the extra time soliciting bids for high-speed EMUs with tilting capability for future NEC service.

      • Stephen Smith

        So by 2015, once PTC is everywhere, will this then allow us to use European or Japanese designs off the shelf (or close to it), or will there still be a regulatory gap?

        • Alon Levy

          Unclear… First, in California, the FRA is still harassing Japanese vendors about safety, so the question is more about European designs. It’s a pity – Japanese trains are lighter, and for high-speed service have better performance. But at least in California the FRA seems to be okay about off-the-shelf or barely modified European imports. The 900 ton buff strength rule is almost certainly going to go, but the other regulations seem to be here to stay.

    • jim

      Even if Amtrak converted “within the NEC” Regionals to EMUs, it would still need a whole bunch of electric locomotives to haul trains that crossed from the NEC to unpowered tracks (but, yes, less than 70). The five long distance trains that traverse the NEC, the three “long distance” corridors — the Vermonter, Pennsylvanian and Carolinian — and the six Virgina service Regionals (and the odd NEC trains that continue to Springfield) would all need to be hauled by electric locomotives on the NEC and diesels off it. That’s actually more trains than the 14 Regionals that just run between Washington and Boston. But I suppose the Keystones could be converted to EMUs, too.

      It’s a mistake to think that the NEC can be separated from the rest of Amtrak’s northeastern operations.

      • Alon Levy

        Some of those trains can be split – the Springfield trains sometimes didn’t run all the way through, and the Vermonter train Amtrak and Vermont sometimes want to convert to DMUs. (This is independent of the fact that the Springfield line should be electrified anyway.)

        Since Amtrak’s coaches cost about as much as EMUs, my cheeky suggestion would be to just use the EMUs as coaches off-NEC, much like how SNCF stuck a diesel locomotive in front of a TGV to get to Sables d’Olonne before the line to it was electrified. If there’s a reasonable chance of electrifying more of those lines (or replacing them with HSR) in the life of the cars, i.e. in the next 40 years, then it’s cost-effective and will prevent the electrification from requiring additional rolling stock when it happens. It makes it easier for Amtrak to modernize in the next few decades, whereas staying with unpowered coaches makes it harder.

  3. Beta_Magellan

    Might Amtrak also be betting on a delay in the PTC deadline, perhaps an indefinite one? It’s still a largely unfunded mandate, and one that’s drawn some criticism from the Class I’s.

    • Alon Levy

      Maybe, but I doubt it – Amtrak’s said nothing of the sort. Also, extending an already debugged PTC system for the remaining half of the existing NEC is, I’m sorry, a trivial undertaking.

  4. BBnet3000

    This is incredibly frustrating considering the life of the rolling stock as well. A poor purchase can keep hurting service for decades…..

    • Alon Levy

      Yes, exactly. Amtrak’s going to build and maintain new track in the NEC to the standards of the overweight Cities Sprinters, and when the Amfleets become obsolete obtain new unpowered coaches for the price of high-speed EMUs, and so on. In principle the fleet is small enough that moving it all to less important lines or even reefing the locos is more cost-effective, but as with the Acelas, it won’t be done because it would be a political admission of defeat.

      • EngineerScotty

        I think you mean “political admission of error”.

        Doing the same wrong thing to save face rather than correcting course in the face of evidence is an all too common bureaucratic habit, both in private sector as well as in public. (Though it’s especially prevalent among elected officials, for whom admitting error is often politically fatal).

      • Alon Levy

        I suppose so.

        But in a way it’s both defeat and error. Bureaucracies sometimes admit errors, decades after the fact. Usually this comes when a dead-end product’s useful life is over, and the replacement is not based on the product’s platform but on previous designs or even brand-new designs (for examples, Windows Millennium and the Space Shuttle). Such admissions are usually tacit and discreet; to admit error publicly and scrap the program early, as Microsoft more or less did when it replaced Windows Me with Windows XP, is rare.

  5. Matt Mitchell

    >>According to this press release, the extra $97M is for “maintenance facility upgrades and spare parts”.
    Could well be that they wanted to finance necessary improvements to maintenance facilities (many of them are old and deteriorated) as part of the equipment deal. If they’re getting a below-market rate on the financing (think 0% financing on autos), it could be advantageous for Amtrak.

    Matt Mitchell

    • Alon Levy

      Rolling stock purchase deals sometimes include maintenance. I’m not sure how much the maintenance part is supposed to cost; generally rolling stock plus maintenance is much less than what Amtrak is paying, but there aren’t enough electric locomotive orders for me to compare it to.

      • Richard Mlynarik

        One data point: here’s a 2007 summary from SBB of FLIRT (ie a modern, lightweight, service-oriented design, maintained by professionals) lifecycle costs on page 11 of

        The breakdown for over 25 years is given as:

        * Purchase price: 56%
        * Vehicle maintenance: 25%
        * Power (ie electricity): 11%
        * Cleaning(!): 8%

        Up-front capital cost is little over half.

        Maintenance (by professionals, not shade-tree; scheduled, not wait-until-in-breaks; 95% availability, not sub-70%) is a quarter of the total.

        Meanwhile energy is becoming a bigger and bigger deal. A number of recent tenders have been won based on the lifecycle energy savings of lighter and more efficient vehicles. None of this — lightness, efficiency, competitive bidding — is relevant to Amtrak.

  6. Pingback: Streetsblog Capitol Hill » Amtrak’s Loco Locomotive Purchase for the Northeast Corridor
  7. Walter

    It is strange Amtrak hasn’t tried to piggy-back on other purchases; for example, it could easily piggy-back on Metro-North’s M8 purchase (which I believe is under $3 million per car) and use them as either Was-Bos Regionals or on the Keystone Corridor (and on the Springfield and NYP-Albany lines if and when those are ever electrified), and just configure the interior however they want.

    Perhaps they don’t want to use EMUs because they can use the Amfleets anywhere in the country, and will try to bundle an extremely huge order to (hopefully) keep down the costs?

    • Alon Levy

      Yeah, this is possible. Well, another reason is that the EMUs used by Metro-North and the LIRR have a top speed of 160 km/h, though in general the modifications required to rate a 160 km/h train at 200 are not difficult. (Plus, the actual time saving from 200 vs. 160 on the NEC is fairly small, because of all the slow zones.)

      Bear in mind, Amtrak’s coach orders aren’t very large, and tend to be pretty expensive as well. I’d guess the lack of piggybacking is standard agency turf, and the avoidance of EMUs comes from the fact that the locos and coaches need to be replaced at different times, like this.

    • Alon Levy

      Ugh. So I just double-checked and the M8s are only rated at 90 mph/145 km/h, though they still have a higher power-to-weight ratio than an AEM-7 leading 6 Amfleets. The M7s are lighter and rated at 160 km/h, but they’re not configured for high voltage, and making them multi-voltage could raise weight somewhat.

  8. Pingback: Amtrak’s Expensive Trains | 1FENS
  9. eldondre

    I suspect that the number of trains operating on/off the corridor will continue to increase over time. Some will be extended regionals, others will require added capacity. there are a few wild cars that could lead to the sprinters finding use elsehwere…electrification of the springfield line (where engine changes might make it to springfield rather than new haven) and possible electrification of the MI line.

  10. Pingback: Followup on the FRA and Amtrak | Pedestrian Observations
  11. James

    The devil is in the details.

    The link for the 3.2 million Euro TRAXX units was MAV’s order for P160 type units – maximum speed 160 kph. The fact that they were ordered in early 2010 with delivery to start about one year later shows that these are very conventional units. Amtrak needs a 200 kph unit. While an increase of 40 kph doesn’t sound like much, it is since Bombardier has not designed or built such a unit. The locomotive will need an all new suspension design with lots of design and testing work involved. TRAXX locomotives (both F140 and P160) have been essentially freight locomotives from the beginning.

    The 3.7 million Euro Siemens units are SNCB class 18’s (200 kph). The order for the first 60 was in 2006, the second 60 were an option exercised in 2008. This was a new model for Siemens with significant redesign required. There were lengthy delays in delivery and acceptance, so much so that Siemens had to pay 21.2 million Euros in penalties, or in other words, underpriced this contract by that same amount.

    The 4 million Euro Siemens units are model ES64U4 multisystem high speed (230 kph) units developed for Austria (1216 class). Like the TRAXX units above, they are already accepted for cross border operations in a number of European countries particularly between Austria and Italy over the Brenner pass. This is why an Italian freight operator decided to buy a high speed multi-purpose locomotive even if they won’t use it to its full capabilities. That plus the fact that these particular units were built by Siemens on spec in 2008 and had been sitting unused in Linz Austria since then. Siemens finally unloaded some of them in late 2010, early 2011 at a very good price. There are still at least three units built at the same time sitting in the shop in Linz waiting to be sold. I’m sure Siemens will be willing to unload these too at an attractive price if they can only find someone willing to buy them.

    Add in suspension redesign/testing for the Bombardier unit, inflation and/or true cost/actual prices for a unit built to Amtrak’s spec (even if it is built to the same structural standards as defined by the FRA) and you’ll see that 30-50% premium disappear. You might also consider what NJT paid to Bombardier for their ALP46’s and then tell us if Amtrak got a good deal.

    As for the weight, your implication is that the heavier overall weight of the Amtrak unit will only be due to FRA standards. I’m not sure if that’s the case since the reality is, aside from buff/draft loads, the FRA Tier I and corresponding UIC standards are very similar; significant structural redesign is not necessary. This is why the FRA has been willing to grant waivers for UIC compliant equipment in the first place. FRA Tier II is where the problems are, not Tier I. Personally, I’d look at the details of the Amtrak spec to see what else may be driving the weight before I take the easy road and implicate the FRA.

    • Alon Levy

      Fair enough. But:

      1. Amtrak’s locos are from Siemens, not Bombardier. Siemens does have experience building 200 km/h units, as you’ve said.

      2. I’m not sure how big the difference between 160 and 200 actually is. The main example I can think of of a 160 km/h train that’s sometimes clocked to 200 is the FLIRT in Norway, where it cost 2.56 million CHF per car, on the high side. On the other hand, there are other reasons for such an order to be expensive: the need for extensive winter-proofing, the overvalued currency in both Norway and Switzerland, and the use of three powered bogies versus two for a normal FLIRT. Stadler’s other orders, for 160 km/h equipment, aren’t much cheaper – e.g. see SBB’s order for 2.46 million CHF per car, and Belarus’s order for €1.5 million per car.

      3. There are some differences between FRA Tier I and UIC rules. That’s why Caltrain asked out. This also applied to the ALP-46, whose weight penalty was 4.5 tons; it’s plausible the penalty has gone up recently due to European weight-saving innovations such as crumple zones. And, correct me if I’m wrong, but aren’t all FRA-compliant train cars required to withstand 800,000 pounds of force without deformation?

  12. James

    As I noted, Bombardier’s TRAXX units have been essentially freight units from the start. They are not ‘universal’ units that can either be used for freight or high speed passenger as the Eurosprinters are. The TRAXX units that do run in passenger service run at lower speeds (160 kph). The trucks are not designed to run any faster. Offering the price of a TRAXX unit as a comparison to the Amtrak order is comparing apples and oranges.

    There’s a long history behind the development of the Bombardier and Siemens designs but a major part of it has to do with two orders: the DB wanted a low cost freight unit while OBB wanted a high speed universal unit. Universality has a cost which is why Siemens has had a very difficult time winning orders outside of operators that run into Austria (the OBB has been pretty much off limits to TRAXX units thanks to their more track damaging axle hung traction motors). They cannot compete with Bombardier on price. That tells me 200 kph is a significant cost add.

    The 800,000 pound requirement is the buff/draft requirement I mentioned. Outside of that, the differences between UIC and FRA are pretty minor. The FRA waivers that have been recently granted are pretty much limited to an exception from the buff/draft requirement.

    • Alon Levy

      Alright. Well, the cost premium is still there; I deliberately avoided using the lower-cost TRAXX in my 35-50% premium number. The Belgian order would be €3.85 million per unit with the extra penalty, and apparently the Prima II, which is rated at 160 km/h but universal, costs €3.75 million per unit.

      The buff strength requirement is what increases the weight, and why I’m bringing up the impending PTC deadline. Getting the FRA to let go of its entire body of regulations is a long and arduous process; getting the FRA to let go of buff strength seems to be a problem that will solve itself by 2015, or at least that’s what Amtrak believed in 2009.

  13. James

    The Prima II is as universal as a TRAXX P160 – they can haul passenger trains at up to 160 kph. In other words they’re commuter locomotives at best. They’re not in the same league as the Siemens units.

    As for the SNCB units, Siemens took a bath on those since they effectively bought the order (not that they had to since Bombardier doesn’t have a competing design). As I noted at the start, once you add in true cost, inflation (since 2006) and currency exchange rates, you’ll see most if not all of that so-called premium disappear.

    • Alon Levy

      I checked the Prima II simply because it’s newer and higher-powered than the original Prima, which is universal – I don’t think it can do more than 160, but it can haul freight trains.

      Inflation in the cases of both Belgium and Italy is quite low. The units were delivered to Belgium in 2009-11 and to Italy in 2011, and will be delivered to the US in 2013; it’s a period of very low inflation. Currency exchange rates, maybe – though, at least in theory (clearly not in practice due to economies of scale, foreign design, etc.), they shouldn’t be an issue when the manufacturing is done locally.

  14. Pingback: Buy America is a Scam | Pedestrian Observations
  15. peettheengineer

    Well, I guess the appropriate Bombardier product for the Amtrak contract would not have been TRAXX, but ALP-46. These engines are mechanically based on German BR 101, which is good for 220 km/h. I think there should be more focus on getting a working product than just looking on the price tag. The Siemens locomotives are still not wokring in Belgium and so that the trains are holed by rented Bombardier locomotives. The products of Stadler may have a high price tag, but they get the things working even in harsh environment like Algeria or Belarus.

    • Alon Levy

      The ALP-46 in New Jersey is also substandard, due to the FRA weight penalty, and doesn’t even have the excuse that it needs to run at 200 km/h, since the top speed in use is 160.

      • peettheengineer

        @Alon I do not understand, what you want to say with “substandard”. The ALP-46 is a prooven product for the American marked. The heigher weight compared to European locomotives can be used to give more traction force and allows the use of cheaper steal than aluminum and fibre glas for the production of the locomotive. From my side it is not understandable why Amtrak selected a new product from a supplzer not familiar with American Railroads.

        • Alon Levy

          First, going by the experience of the Acela, higher weight raises maintenance requirements. It also raises track maintenance requirements, by a large margin; at the axle loads in question, track wear grows as a large power of axle load (empirically, >3), and if the power is 4, as with trucks, then the extra weight raises track wear by 50%.

          Second, the ALP-46s actually have less tractive power than the DB Class 101s from which they’re derived, 5,600 kW vs. 6,400, and are rated at a lower maximum speed, 200 km/h vs. 220. The Class 101’s current European descendant is the TRAXX, which costs about $4.5 million per unit; in contrast, the newest ALP-46 version, the ALP-46A, costs $8.5 million (link), even more than the Amtrak locos.

  16. Pingback: EMUs Versus Locomotives | Pedestrian Observations
  17. Todd

    Check your math.

    There is a difference between tons and tons. The UK uses what is also called a ‘long ton’ which is 2240 lbs compared to the US which uses a ‘short ton’ or 2000 lbs.

    So the ’97 ton’ US locomotive is about 87 metric tons, the same as the UK locomotive in your comparison.

    Also, I don’t see where you factor in the cost of transporting them across the ocean i.e. fuel cost, insurance for if your multi-million dollar locomotive falls off the ship, import cots etc.

    When comparing US to UK railway standards keep a few things in mind. The safety standards are completely different. The axle loads of rolling stock and passenger cars are significantly higher in the US and ever increasing. A locomotive in the UK has to operate in the UK climate/weather in the US locomotives have to operate in greater varying conditions. Look at BNSF, the different states in which they operate. There is a huge difference between Arizona, Louisiana and Colorado. But these companies want a single solution for all their rail. Keeps the maintenance costs down, which in the long run is the largest cost.

    I think you touch on an important topic of discussion, politics CAN foul up any business. However a discussion without all the facts is just someone yelling ignorantly.

    • Alon Levy

      No, the locomotive is 107 short tons, 97 metric tons. See here for a reference, or just Google Amtrak Cities Sprinter weight.

      The higher weights of the Amfleets are not an excuse for higher-weight locomotives. A locomotive needs to be powerful enough to pull them at speed, which every version of the Sprinter is, including the lighter ones used in Europe. It doesn’t need to operate in the same environment as BNSF’s equipment; Amtrak has a grand total of two electrified lines and no plans to electrify further, and any freight compatibility is out of the question since US freight railroads only run diesel locos. The weight increase comes entirely from the higher buff strength requirements, which could be euphemistically described as “different safety standards” and more correctly described as “random numbers pulled out of a hat that end up making trains less safe for their passengers.”

  18. Pingback: Free Trade In Trains | ThinkProgress « mawillits
  19. Paulus Magnus

    One oddity on that pricing that I just noticed: BNSF announced earlier in the year that they would be purchasing 227 locomotives for $450 million this year, just under two million per locomotive. Not only is that greatly lower than Amtrak’s locomotive cost’s, it’s also cheaper than buying new passenger cars for Amtrak. Surely passenger rail doesn’t have that great of a difference in requirements to justify such a marked premium, does it?

    • Alon Levy

      Probably the biggest differences are that BNSF is a huge buyer of diesel locos, and that BNSF is private. Looking at EMU costs in Europe, what I see is that private operators in Germany pay €1 million per FLIRT car and public ones €1.5 million or higher. I don’t think those together are enough to explain the difference, though.

      • alki

        I think you and Paul are ignoring the differences in the size of the orders. Typically, the larger the order the lower the price for each unit. Europeans order many more trains than we do in the States. They may get considerable price breaks for those larger orders.

        And speaking of price differentials. I can’t figure out how you came up with the ratio of 35-50% in your article. After converting euros to dollars, I come up with a cost differential ratio of 19-26%. Amtrak still looks like its getting gouged but not nearly as badly.

        • Alon Levy

          The most expensive order per train that I’m comparing this to is for two locos.

          The 35-50% figure is if you do a PPP conversion. Those locos are manufactured in the US, right? Importing them would involve a small premium over PPP rates, coming not from today’s exchange rate but from either the actual exchange rate in 2010 or from the perceived rate of 2011-13.

  20. peettheengineer

    You can not compare the BNSF contract with the Amtrak contract. The BNSF locomotives are series produced in several hundreds when not thousands, the Amtrak locomotives are only a very small series. The most cost driving thing are the one-off costs, which are very high for Amtrak since they ask for many additional tests, which are not standard in the rest of the world. Furthermore do electrics anyway need more testing than diesel engines due to additional comaptibility tests with the overhead wire system.

    • Richard Mlynarik

      You can not compare the BNSF contract with the Amtrak contract.

      Of course one can compare. That’s the point. The comparison is extremely unflattering.

      The most cost driving thing are the one-off costs …

      That’s the point.
      Why do we have to put up with and pay for this nonsense?

  21. Pingback: Materias Grises » Blog Archive » Estados Unidos, infraestructuras y los trenes de alta velocidad
  22. Pingback: SMART Money Part II: The Myth and Allure of Caltrain North « The Greater Marin

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s