Standardizing the Right Way
Picking consistent standards in order to make use of economies of scale is an important part of good planning. In our construction costs report, we attribute a high cost premium on systems and finishes in New York to lack of standardization of station designs and parts, to the point that the three stations of Second Avenue Subway used two different escalator vendors. This point has appealed to a number of area activists, who reach to not just what we report cross-nationally but also American history. John Pegram, who comments here as BQRail and writes an excellent blog on Substack, gave the example of the PCC streetcars of interwar America a week ago, and I promised I’d follow up on this; the news of the cancellation of congestion pricing delayed this post somehow but it’s still important to discuss. The issue here is that good public transportation procurement requires not just consistent standards, but also good ones, which give international vendors a familiar environment and keep in touch with technological advances.
The starting point for me is that the rolling stock on American subways and commuter rail is fairly standardized. New York City Transit procured standard designs in the 1990s, dubbed the R110A and R110B, and for decades kept buying trains based on these designs. In the 1990s and 2000s, it worked, in the sense that the trains were of comparable quality and cost to rolling stock in other large cities (although they were on the heavy side). But over time, technology diverged, and by the 2010s, a cost premium started to appear. By now, NYCT subway car contracts have a noticeable premium over the European norm, even if this premium is far smaller than the infrastructure cost premium.
Commuter and intercity rail cars have a similar issue with what the standard is. American commuter rail cars follow a few standard designs – the EMU design (in either the LIRR/Metro-North version or the SEPTA one), and the unpowered car hauled by a diesel locomotive one. DMU designs are not at all standard, and do have cost premiums as a result, especially since these are also small orders. That said, nearly all American commuter rail ridership is on EMUs or locomotive-hauled trains (usually diesel, occasionally electric), and those, too, have their problems.
The most glaring problem is that those designs are not at all what the rest of the world does. A few of the changes are modular, including the platform height and the loading gauge. The others are not; the consultants who write the design specs do so without trying to fit themselves to common products made by the multinational vendors.
Then, those specs are extremely detailed; there’s little room for a vendor to try to pawn off a standard Coradia or FLIRT and make that fit with little modification. The RFPs run into the deep hundreds of pages; SEPTA had one with more than 500 pages, and Amtrak’s most recent one ran to, I believe, 1,000. They define even what a train is, as opposed to the looser RFPs common in Europe – Spanish RFPs are 50-70 pages and have single-digit summaries, detailing just how many cars are needed, what the loading gauge is, what electrification is required, and what the expected performance level is.
Designs exist that do dialog with the international vendors and aim at a comparable product – the FRA reform process that led to alt-compliance did exactly that. But then no American commuter rail operator has bothered to make use of alt-compliance; they still buy the heavy, low-performance, low-reliability equipment that they’re used to buying, even as technology marches on and vendors don’t specialize in making that anymore.
The original example of the PCC standard is well-taken in the sense that there need to be repeatable standards. However, it’s important to understand that technological advances in trains exist in East Asia and Europe, and not in North America. American standardization needs to be around what is sold on the other side of either the Atlantic or the Pacific, with no wheel reinvention, and no “we are familiar with this so we’ll keep buying this” excusemaking.
Pedestrian Observations 
Is there a database of passenger railcar purchase costs? That would be a good project for the DOT.
No. We should make one.
Apparently the new regulations went into effect in 2019. Who has ordered train since then?
I think the US should push for more DMUs. The old class 168s on the Chiltern line can do 150km/h-0-150km/h in 3 minutes with a 1 minute dwell.
While that isn’t as good as the ~100km/h a minute with a late model Shinkansen or better with a FLIRT train it certainly isn’t terrible.
I would’ve agreed a couple of years ago, but I think BEMUs are a better choice today (if you’re not able to electrify). The acceleration of the Class 168 is nothing to write home about and better performing DMUs (like the DB Class 650 which accelerates more like a good EMU) have been discontinued, to be replaced by more sluggish options. Manufacturers are moving away from DMUs and currently available DMUs are often worse than the options available in the 2000s, the heyday of DMUs (IMHO).
By using cheap, small-scale electrification at the terminals, most commuter rail will be able to operate with BEMUs available in the European market today (~150+ km range) and save on fuel and maintenance, while enjoying most of the benefits of an EMU (acceleration, noise, pollution), with the only downside being the increased purchasing cost (compared to an EMU). If they’re slowly electrifying, running a mixed EMU+BEMU fleet could be a great option, as they can often be coupled together.
Aren’t there two standardisations at play? There is internal standardisation, i.e. minimising the different types of material within the organisation, and external standardisations, i.e. ordering rolling stock that is a minor variation of stuff that is on the market. Large orders of “Universallokomotive” of the 70s were about the former, buying a Vectron is the latter. Ideally, you would want both, like the SBB’s huge order of Kisses.
There should not be a RFP to buy a train. You should instead look at manufactures catalogs, and when they can deliver to place an order for a train at list price. Of course some things are simple options (standard gauge or broad gauge; several different seat options; maybe an optional restroom), and you will need a custom work to get your way finding integrated (LCD screens might be included, but someone still needs to program them – though I would hope there is reuse and standards here I don’t expect it). It should go without saying that each manufacture has the ability to do some amount of custom painting/applying logos, but there are limits to what you can do. However in general you shouldn’t be buying a custom train, you should be buying a standard train. These trains should already be certified to run on standard system (FRA, whoever needs to apply certification)
This is a well meaning comment that doesn’t reflect reality. Trains are not commodities. There are, generally speaking, no ‘list prices’. Manufacturers do not have ‘catalogs’ in the sense that retail business or manufacturers do, where you can order a stock product ‘off the shelf.’ Unlike retail products, not a single train car is made without a confirmed order; rolling stock manufacturers do not make train cars worth millions each and just have them sitting around the warehouse hoping to make a sale. Those confirmed orders come as the result of RFPs and bids.
All of the “custom work” items you identify (seat options, restroom or not, integrated wayfinding, etc.) are all things that need to be specified in an RFP otherwise you won’t get what you want (or you will get non-comparable bids as each manufacturer interprets your ‘options’ differently if they are not consistently specified).
The answer is not to advocate for no RFPs, instead it is to advocate for a contracting environment that allows for shorter more efficient RFPs, which lead to less customization and lower costs. As Alon notes, every agency everywhere, even the best run ones, use RFPs, its just that they are 1/10th, to 1/20th as long as US RFPs. Some of this can come from the external standardization @wiesmann refers to above. If many agencies have the same standard specs, you can write a shorter RFP identifying the key elements knowing that you will get many more competitive bids (because they already produced rolling stock with those specs for another agency, so it costs little to respond to the RFP and they know what it costs to build), instead of getting a few higher bids from only the manufacturers who take the plunge to estimate a custom spec (with all of the overhead of designing the custom equipment and risk from mis-estimating how long/expensive it will be to manufacture built into the bid).
I would like to argue for a perspective flip. Trains should be seen as “slightly customized commodities” rather than “bespoke designs that curiously have a second-hand market anyway”. Which is to say, procurement offices shouldn’t think in terms of “we want this-and-that, by default/tradition we expect that manufacturers (ye olde times: the in-house works) would come up with a substantially new design for it, it is merely a convenient coincidence if they already have a standard design that almost fits our requirements and can be easily modified to fully do so”, instead they should think in terms of “these are the standard model-families (with their respective options) that manufacturers offer, we should pick the one that is closest to what we’d like to see”.
Which is to say, in most cases it shouldn’t be the designs that are customized around the requirements (which are treated as if set in stone) but the requirements should be, at least reviewed, in view of what designs exist already. This is a mature enough industry that if you ask for something that nobody is offering to build as “standard with options”, you are probably asking for something stupid.
rolling stock on American subways and commuter rail is fairly standardized
I actually disagree with this from a subway perspective. @weisman made the distinction between internal standardization (are all of your lines and rolling stock interoperable) and external standardization (is your rolling stock of the same specifications as other agencies). With the exception of Boston (where each of the 4 lines uses different rolling stock) and Philadelphia, internal standardization is basically a given for every US subway (NYC of course has the A-Division and B-Division with different specs, but each is so big it doesn’t matter, they each have thousands of cars in use.
External standardization is much worse. Much of what I write below I previously covered in comments at https://pedestrianobservations.com/2023/08/12/connecticut-pays-double-for-substandard-trains/, so head over there for a deeper treatment.
Basically, pre-1960’s there were two main standards for subway cars in the US: NYC A-Div (former IRT), NYC B-Div (former BMT and IND) and Chicago L. Chicago L was not relevant to future subways as its loading gauge was tied to the very narrow curves of the Loop. B-Division was by far the largest, and the one used to some extent outside of NY (Phila. Broad St line, Boston red line, etc.). When the “Great Society Subways” (DC, Atlanta, SF) were funded in the 1960s, it would have made sense to require them to adhere to NYC B-Div specs, making it the de facto national standard.
Instead, each system was designed slightly differently. DC Metro and Atlanta MARTA use the same voltage, DC and BART have the same platform height, BART and MARTA have the same width trains. As a result these three otherwise very similar systems cannot share rolling stock orders with each other or with almost any other system in North America. It has only gotten worse. When Miami and Baltimore opened their systems they choose vehicles that did match B-Division specifications – except for power, which is the same as DC and Atlanta. As a result almost every US subway rolling stock order has to be bespoke, instead of everyone just ordering a few more of whatever the NY Subway ordered most recently.
The US is currently in a position for intercity rail and HSR that it was in the 1960s for mass transit. Between Gateway, CAHSR, Brightline West, and Texas Central, more money is going to go into new intercity rail and HSR infrastructure than ever. This is the perfect time for Congress to act and make such funding contingent on buying equipment that meets the standards used on the NEC between Boston and Washington. Just as the NY subway is by far the largest in the country, the NEC will always have more rail passengers than anywhere else. If this were done, then rolling stock orders for other operators can take advantage of not spending money on R&D/customization by ordering whatever Amtrak already bought a ton of. Conversely if another operator does make the investment in new technology first, then Amtrak can specify it for their next order (indeed manufacturers would have an incentive to work or such a new order at a competitive price, knowing they could later sell it to a larger market). Same for commuter rail.
The US is already close in this regard, commuter rail/intercity rail equipment is already much more standardized, it would be a shame if balkanized decision making (Tx Central chooses a Japanese supplier, CAHSR a French one, each has slightly different specs for traditional reasons) hurt any future US HSR in the long run.
Great idea! Freeze the entire USA to whatever shit was in use in NYC in 1970 and whatever shit Amtrak was running on the NEC in 1980. Ignore what everybody else on the entire planet is doing, ignore anybody who has had any success at doing anything, ignore anything that’s ever been improved, because, um, National Standards, because economies of More More More Amtrak. Toe the line, Brightline, you are required to use national standard PCC cars, for great prosperity.
The Soviet Union really did win the Cold War, didn’t it?