The Northeast Corridor Report is Out
Here is the link. If people have questions, please post them in comments and I’ll address; see also Bluesky thread (and Mastodon but there are no questions there yet).
Especial thanks go to everyone who helped with it – most of all Devin Wilkins for the tools, analysis, and coding work that produced the timetables, which, as the scheduling section says, are the final product as perceived by the passenger. Other than Devin, the other members of the TCP/TLU program at Marron gave invaluable feedback, and Elif has done extensive work with both typesetting and managing the still under-construction graphical narrative we’re about to do (expected delivery: mid-June). Members of ETA have looked over as well, and Madison and Khyber nitpicked the overhead electrification section in infrastructure investment until it was good. And finally, Cid was always helpful, whether with personal support, or with looking over the overview as a layperson.
Pedestrian Observations 
The real question: is there any hope of this getting implemented, or is just another binder to put in the library next to all the other proposals done over the decades since Amtrak started?
I don’t know yet.
How easy is it to adapt off-the-shelf EMUs like FLIRTs, Mireos, Coradias, and Civitys to NE US contexts?
(Or here’s an alternate question: should the scope of regional rail stock design and acquisition only include the NEC, or should the scope consider any possible rail line—such as ESA—and optimize for one design?)
All those EMUs are built for low-floor, which particularly means all the running and electrical gear is mounted on the roof. If the entire carbody, designed as-is, is raised to a floor height of 51 inches (1295 mm), will that pose clearance problems? I also presume it would mess with the center of gravity those railcars were originally designed for.
I’m concerned that the only EMUs we can actually import from Europe that would fit our loading gauge are British stock, given that carbodies can be lifted slightly and widened with minimal impacts to cost. German S-bahn stock like the Class 423s and Class 430s (not the 425s) could also work since they’re high-floor and appropriately wide for the US, but those too have roof-mounted equipment, and aren’t off-the-shelf.
I am pretty sure British class 745/755 are high floor
The floor height on those is still lower than the top of the bogies; the interior is not 100% step-free.
In addition, equipment is still roof-mounted—though with FLIRTs, that second fact is not so relevant, since they actually put most of the electrical equipment inside the passenger compartment, taking up space.
All Japanese-built EMU/DMU train sets are high floor with equipment underneath. Most are single deck, a few double deck trailer (non-powered) cars are used on regional trains. Both Hitachi and Nippon Sharyo have built EMUs in US factories in the recent past.
Kawasaki builds a few
https://en.wikipedia.org/wiki/M8_(railcar)
https://en.wikipedia.org/wiki/R211_(New_York_City_Subway_car)
Assuming the numbers are correct in Wikipedia’s rolling stock articles, there are 1,162 MTA M7 cars in service. What’s the biggest fleet of single kind of Flirt?
You really don’t want to examine this
https://en.wikipedia.org/wiki/New_York_City_Subway_rolling_stock#Current_fleet
Or Japanese or Chinese fleets.
First, when modifying an existing design, going to a higher floor level is easier in every single way. You’re removing constraints, not adding any (recall we’re talking only about single-level cars here.)
Secondly there may be less roof-mounted equipment than you think. (The “usual” FLIRTs for example put everything in large cabinets that steal space from the passenger level.)
Third, exactly because under-floor equipment mounting has advantages, there’s a mini-trend of single-level traction-equipped cars sandwiching bi-level trailers in EU-land.
Generally, accomodating different floor heights (because it’s almost an exception when there’s just one platform height “standard” in a single European county!) is the sort of thing train designers do in their sleep, because they don’t get to sell trains unless they do.
Relocating any traction equipment from roof or above-floor level to below-floor is likewise almost a triviality: all the manufacturers have direct in-house experience because they also build high-floor metro trains and some of high-floor high speed trains.
Lower centre of mass can only make things simpler. What’s with all the presumptions?
Your concerns are misplaced. They’re also free-floating in a void utterly free of facts.
The Amtrak NEC “1355” static loading gauge is comparable to and generally larger (above 1320mm, it’s very slightly narrower below) than any of the various (because why just one?) continental European national gauges.
Why imagine problems? Why speculate without data?
Why type comments?
Anyway, very slightly altering the dimensions of passenger car body extrusions is the sort of thing that competent manufacturers do in their sleep. It’s the sort of thing that Stadler engineers do just for fun, in their sleep, while juggling five cog-wheels behind their backs.
As Richard said, it’s easier to raise the floor than to lower it.
There is a snag in that the latest Euro-EMUs are designed with lower platforms from the ground up. That said, modifications for the British market don’t cost much more than the standard products. When Patrick O’Hara asked Siemens and Stadler at InnoTrans, they said the modifications were doable but nontrivial; a large order like a future M10 should be able to do it with little cost premium (provided the trains are imported, the RFP is written per European norms, etc.). A small order, like just the Providence Line, would probably face a larger premium, but metro New York makes big orders and they’re ripe for being dragged from the 1990s into the 2020s.
New York should still have trains from 1990 in use (well there is some debate: most keep trains for 40 years, but some only 15 – we can discuss that another time so for this discussion lets accept that 40 years has won). How ever they should not be talking about updating to the 2020s, they should have some trains from the 2020s, some from 2010s.. Their contracts should be 10 (50, 100 – I don’t know the correct number) trains a year every year – update the design every year (but not more often except for critical issues that are retrofitted at your cost to existing trains) to the latest technology. Predictable orders are one of the keys to low costs, that means companies can go to their suppliers and make orders years in advance and so everyone can plan so that things arrive just in time for manufacturing.
Predictable orders mean it is worth investing in things that lower costs long term.
Now there is the valid question of how much trains are predictable because manufactures get orders from a lot of different transit agencies. The common variations in trains are things they already design around. However you should still work with them to minimize variations that are expensive – some variations are things they could save money on if everyone was consistent and thus they could use a custom jig, but that jig is too expensive to create when it needs to handle all variations.
The oldest cars on the NY City subways, the R46s, are from the 70s and are currently being retired. There are two large groups from the 80s, R62 and R68. Some will probably be retired with the next rolling stock order.
Right now Stadler is building long-distance FLIRTs for Norway with train-length 1200mm floor height! You know, FLIRTs, the low-floor Swiss trains designed around 550mm level boarding. It does seem that raising floor height simply isn’t a real-world engineering issue. And as I said, there’s a ton of existing expertise in building high-floor metro and high-floor HS trains; in the end they’re all just boxes with wheels under them and a bunch of doors and seats and stuff, just configured differently.
Also, Germany is full of hundreds of suburban EMU trains (DBAG Baureihe 423/424/425/430/etc, and grandfather 420) that are barely low-floor in function or in design, featuring train-long level floors and targeted at 960mm platform height. Class 430 for example has a 1030mm floor height. Getting down to 960mm step-free level boarding has been the ongoing engineering challenge. These are high-performance high-floor Euro-EMUs, and they’re everywhere. It’s getting the floor height down to 960mm which is the ongoing challenge. (DB has put out to bid for 400 3-car EMUs to partially replace these guys, and level boarding is a requirement.)
It’s too bad Siemens isn’t ( insert a long sigh here ) Swiss. Wikipedia says they have made or have orders for 937 “Venture” cars. https://en.wikipedia.org/wiki/Siemens_Venture#Operators
Interesting report! One thing I was wondering when reading the timetable, not serving all current Northeast Regional stops would undoubtedly lead to controversy. Even if the complete picture is clearly an improvement on the status quo.
Have you investigated more complex timetables than just the 10 to 15 minute pattern switch? Arguably you can “afford” more complexity on the 1D system that is the Northeast Corridor. You could alternate between hourly local stops on the NYC-Trenton segment (like on the southern half of the UK East Coast Main Line) to provide fast, direct trips to/from destinations beyond Trenton and NYC. And still stay close to a 10-10-10 spread at Philadelphia. You could have only one hourly through service to Harrisburg and Springfield, to keep an extra train on the NEC, with a timed transfer. This way you’d still follow the “frequency half the trip time” rule. This remaining extra train per hour could be used to serve additional stops on New Haven – Boston and Philadelphia – Washington DC. You just need to make it work around commuter trains, for instance by running the local intercity train just before/after an express intercity train in crucial locations, creating a 13 minute gap for commuter rail to fill with their stops, or a slightly longer overtake. It might also require additional overtaking tracks, which local authorities might consider worth it to keep their intercity stops.
This added complexity would perform well in ridership modelling, since even in relatively punctual systems like the Netherlands, a rider is considered to perceive a 5-minute cross-platform transfer as 15 minutes extra in-vehicle time.
On the timetabling assumptions, I would be curious to know if timetables are padded ~7% consistently for ~4 hour intercity trains in Switzerland and Sweden. At least in the Netherlands, the takt structure is tight for some segments (like Zwolle-Groningen/Leeuwarden), but on other segments (like Utrecht-Amersfoort), it leads to significantly extra padding and/or longer dwells.
Agree that reducing regional stop frequency is going to be politically controversial and therefore isn’t worth attempting.
Like sure you can do politically controversial stuff but you also lose elections over it.
The intercity express trains can be on the express tracks and the commuter trains can be on the local tracks. The constraints are five miles/eight kilometers on either side of Penn Station New York, where trains beging to diverge off. The busiest part is between New York and Rahway New Jersey where someday far far in the future there might be 20-ish trains an hour sharing 4 tracks. The ones going to The North Jersey Coast Line diverge off in Rahway. And there is enough space for six or more tracks allllllllllllll the way to Philadelphia.
Railfans seem to think the intercity trains will catch cooties from the express commuter trains or get the heebie jeebies if there is a local across the platform. Even more astounding to railfans, the local intercity train can share tracks with the express commuter train!! !!
There’s already complexity in the schedule, is the issue – the overtakes are complex, and don’t have much margin for error. The Providence Line has two overtakes between intercity and commuter trains, and the northern overtake is right next to a timed meet on the express track since there’s only triple-tracking and not quad-tracking planned and so during the overtake the intercities are effectively on single track. At one point we looked at even simpler schedules, with rigid 10-minute frequencies south of New York (scenario 1 in the scheduling section), with the Keystones cut to a forced transfer at Philadelphia, but between the costs of additional infrastructure in Maryland (nearly the entire Baltimore-Washington line would need to be quad-tracked) and the transfer penalties, we went with this scenario instead (2C).
The reason for reducing a lot of stops to transfers is that these stops are already incredibly weak. Other than Trenton, which is a judgment call, and Back Bay, which would be an express stop if it were possible to lengthen the platforms there, these are not high-ridership stops. Everything else is stops like Newark, Delaware (29,344 ons and offs in FY 2023), Aberdeen (42,671), and Old Saybrook (67,076), plus the main park-and-ride stops (Metropark, Route 128, New Carrollton), which exist only because of weak commuter rail-intercity rail interface to begin with – those stops were opened after WW2 as park-and-rides and never really worked for rail revival, and none of them is much of a destination, e.g. the Route 128 stop is in about the least developed, job-poorest part of the Route 128 corridor.
The padding was said to be just 7% in the papers I consulted for this, but you’re right that if trains wait for connections then it can be more. The modification of the Swiss or Dutch system that I’m using is that because the primary corridor is much stronger than any connection, the trains don’t work as a multi-node integrated timed transfer network, and timed transfers are mostly sporadic things like transfers from commuter trains from Perryville to Baltimore to intercity trains to Washington.
Take two deep breaths. If there is a high speed bypass of the squiggly bits through the middle of all the former fishing villages along the Connecticut, coast how many commuter trains on those tracks are in the way of the high speed trains miles away on the bypass? The clever clever people of Greater Old Saybrook are aware of this thing called a timetable and almost all of them know the many different ways one can consult one. They can look at them and see that an intercity train diverts off the bypass once every two hours, give or take a bit, and check the timetable when they want to be in Philadelphia at 8:45.
Why wouldn’t the trains from Perryville ruuuuuuuuuun throooooooooooooooouuggg!! Baltimore and go all the way to Washington D.C. I’m sure there are people in Aberdeen just dying to take the train to Odenton to hike to Walgreens and perhaps have lunch at the 7-11
Alon, it isn’t possible for the Americans to manage timed overtakes at a single station with no extra padding or backup.
The British cannot manage this.
To expand, good for America would be achieving a European level of service, I.e broadly similar to Brightline – or France/Germany/Britain.
I agree with most of the stations dropped from intercity service, but Trenton is a bit harsh. It’s a useful interchange station since it’s where both SEPTA and NJT service terminates. And surely it wouldn’t be too hard for the Harrisburg trains to stop there, giving it half-hourly trains to Philly, NY and points north?
Also: what turnaround times are you factoring in at the termini (DC and Boston)? It was unclear from the report unless I missed something. If both NY-BOS and NY-DC come in at 1:56 and the dwell at Penn is 0:03, are you leaving only 0:05 for turning around? For a nearly 4-hour train trip this seems like a level of punctuality that would even challenge the Swiss.
The Japanese have like 15 minutes at Tokyo from the north, and that is combined with large amounts of padding both into and out of Tokyo – I believe a total of 3 minutes each way.
But you don’t understand ( the trainspotter navel gazing) . Having a passenger railroad accommodate people would ruin the beauty and symmetry the of majestic trains gliding gracefully past the station. Where there are people. Who want to use the passenger trains.
There will be four or more tracks between New York and Philadelphia. The intercity trains making many stops can share the tracks with the commuter trains. Like they do now.
Also, while I’m sensitive to the reliability issues caused by long-distance trains running north of DC, what about trains to Richmond and then Norfolk/Newport News? Surely an hourly NEC train should be extended southwards given the size of Richmond as a market.
And perhaps the long-distance trains could be retained, running non-stop between Washington and NY Penn, but with enough schedule padding that they can be deprioritised and threaded in the gaps between NEC trains (since there’s only a few per day, they can be scheduled outside of peak hour where some of the commuter slots are freed up).
They could use the commuter LOCAL tracks between Washington D.C. and West Baltimore and the LOCAL tracks the commuter trains use between Wilmington and New York.
I don’t know why anybody other than nostalgic railfans with steamer trunks would want to take a four hour trip from New York to Washington D.C. when they could take a two hour trip to D.C. and walk across the platform to the slow train. Doing that in Richmond would save another half hour.
Though if the Southeast gets high speed rail, the trip to Miami from New York would be 10,12 hours and they don’t need a sleeping car. Amtrak could dedicate a car for the twice a day to Florida with high baggage fees. It’s surprising how little people can manage with if there are baggage fees.
If you’re on an overnight train you would probably prefer to arrive in NY Penn at 7:30am, rather than have to get off at Washington Union at 4:30am to change to an Acela that gets you into NY at 6:30am.
Even long-distance travellers on day-trains are probably not too fussed that their trip takes a little longer if it means less interchanging, particularly if they have luggage or are travelling as a family. I could definitely see a forced interchange at DC killing off patronage on the long-distance trains (as small as that is present). I’ll guess we’ll get a sense for this with the introduction of the Floridian.
Schedule the overnight train – singular – to arrive and depart Atlanta in the middle of the day. Though the Southeasterners have fantasies of high speed all the way to Birmingham.
The turnaround times at the terminals are around 20 minutes, not five. The express trains depart South Station :25 and :55, which means :05 and :35 arrival. They arrive WUS :19.75 and :49.75, which means :10.25 and :40.25 departure.
So the cycle time for the express service (i.e. BOS-DC-BOS) is 8.5h rather than 8h? And the other two half-hourly services run Boston-Harrisburg and DC-Springfield, right? What are your timings for the Philly-Harrisburg and New Haven-Springfield legs (the report seems to leave this open)?
There was mention of changing trains. Because sending a train from Springfield to Harrisburg without stopping in Philadelphia would ruin the symmetry of sending the trains regularly.
Yeah, both of these are right.
The report does leave off-NEC trip times open, except the Grand Central reverse-branch; only one I explicitly computed is Port Washington, for the decision of which branch through-runs beyond Great Neck. (We did compute Boston-Worcester when we thought the report would cover more commuter lines but that work got orphaned.) On the intercity branches, I was vaguely assuming similar trip times to the parallel local options, but it’s a first-order estimate, too optimistic on Keystone and possibly too pessimistic on a well-run Springfield branch (it’s 100 km and straight, it should be doable in less than an hour).
Is half-hourly 16-car trains on the Keystone corridor overkill in terms of capacity? It gets roughly hourly frequencies at present, but with much shorter trains. Given that it’s a metro area of about 600k and there’s little between it and Philadelphia, is there that much potential for ridership growth? And with a 1:40-1:55 trip to get to Philadelphia, you would be eating up 8 x 16-car sets to maintain the half-hourly service pattern. Perhaps either going down to hourly, or splitting the trains at Phialdelphia and only running 8 cars to Harrisburg would be feasible alternatives (there has to be some dwell time at Phily anyway due to the need to turn the train around).
Lancaster is between Harrisburg and Philadelphia. They will let people in Harrisburg’s Combined Statitcal Area use the trains. Having an 8 car train go to Springfield “even” hours and an 8 car train go to Boston “odd” hours and the train from Detroit provide express service to New York and a different train go to Philadelphia and beyond is too complicated.
Lancaster metro population is around half a million, so yeah, also not much. The Harrisburg CSA covers a huge chunk of the state, some parts of which are nearly as close to downtown Philly as they are downtown Harrisburg (and nowhere near the train line), so also not the most useful measure to use.
And I mean sure, if there was a high-speed link to Pittsburgh then that might justify the frequencies but we’re not quite there yet, are we?
Lancaster is a bit smaller than the Center of the Universe, Stamford. Or almost the same size as metro New Haven.
Parts of Harrisburg’s CSA might be closer to Philadelphia than they are to Harrisburg. That doesn’t stop people closer to Harrisburg using Harrisburg or other stations along the line. Like Lancaster.
Stamford and New Haven are ON the NEC. Trains pass through them anyway, the question is whether they should stop or not. Neither would justify the capacity of 16-car trains six times an hour by themselves, that capacity is needed for the major metro areas along the route (NY, Boston, Philly, Baltimore, DC). Alon has made the justifiable decision to minimise stopping patterns for their proposed service plan, so both get the full 6tph, unlike the slightly smaller New London, which would only get 2tph.
Lancaster and Harrisburg are NOT ON the NEC. They have to justify the level of service on the Keystone corridor BY THEMSELVES (save for the small number of passengers continuing onto Pittsburgh and points west along an at present snail-paced corridor).
Since they have a combined catchment population of just over a million people, the question is whether that is enough to justify 2tph of 16-car trains running across the full 104 miles of track, or whether those trains would mostly be carrying air and hence dragging down revenue.
The options I proposed were to run hourly instead of half-hourly (a service reduction, although roughly in line with what the line currently gets), or to have the trains split at Philly, with only 8-cars proceeding along the Keystone (this adds operational complexity, although there is a prolonged dwell at Philly anyway due to the need to turn the train around).
There can be more than one kind of train. There can be long trains and short trains. There can be trains that don’t stop at all the stations and ones that do. There can be trains that don’t go to New England and there can be trains that don’t go to New York. Harrisburg can have express service to New York on the train from Detroit – that terminates in New York. And local service on a train that goes allllllllllllll the way to Boston via Springfield. Every other hour that alternates with a train that goes to Boston via Providence. And yet another kind of train that goes to Washington D.C. that also provides service to Philadelphia. Because Philadelphians who want to go to Baltimore don’t give a shit if the train started out in Boston or New York.
:25 to :19 is 54 minutes. So it’s going to take 1:40 to get from Boston to New York but 2:14 to get from New York to D.C.? But then I’m assuming :05 arrival is in Manhattan and not Mamaroneck.
20 minute turnarounds are still getting towards a Tokyo from the north level. St Pancras manages around 30 minutes at best, Euston does some quicker for some reason, but it is enormously controversial. The average is definitely worse than 20 minutes at Euston.
Also at Euston and St Pancras only the Glasgow and Holyhead trains take 4 hours, most are 2 and a bit.
First train of the day into Tokyo from the south that doesn’t go to the depot has a 24 minute turnaround
I think you’re missing the potential for Philly service in NJ. Right now, Philly NEC ridership is low because the walk shed for PA stations are low-density housing, many stops lack sidewalk networks, and parking is limited. However, there is a large population there. Folks in the area are comfortable driving to Hamilton to take a train into NYC, but the SEPTA option is parking in Trenton.
Problem is, SEPTA charges a penalty fare for Trenton, and suburbanites are scared of Trenton and don’t feel super safe parking there. So they drive into the city. IMO, if SEPTA extended to Hamilton (on a dedicated track?) it would see a significant ridership boost due to the “safe” park and ride.
But more relevant to your scheduling, you have the sin of mistiming the NJT/SEPTA transfer. It was a 3-minute transfer before, but was abandoned with COVID. There is a large population in Mercer and Middlesex that IMO would ride trains into Philly if the connection didn’t suck. Today, that’s a little possible (but sometimes way overpriced) with Amtrak trains that stop in Trenton and Metropark, with some stopping in NBK and Princeton, but you remove that.
Having the NJT express run the clocker schedule (stopping at Cornwells Heights, North Philly and 30th) would solve this gap I think. The AC train should also stop at North Philly to allow transfers, but that’s less relevant here.
Many many many years ago Amtrak and NJTransit polled the people using Trenton and Hamilton. Roughly half of them are Pennsylvanians going to Manhattan. Moving Pennsylvanians is SEPTA’s job, not NJTransit’s.
It’s not New Jersey’s problem that SEPTA sucks. And it would be really really stupid to drive from Pennsylvania to New Jersey to get to Philadelphia when there are train stations in Pennsylvania that can do the same thing.
Why should Atlantic City trains stop at North Philadelphia? So both people who want to use it, every year, can?
For Connecticut residents who care about the Connecticut portions of this plan (there are so many), what’s the best way for a private citizen to advocate?
I attended a virtual information session on the CT Long Range Transportation Plan 2055 and brought up passenger rail improvements, and was basically ignored.
CTDOT is holding in person sessions. Should I show up with a printed copy of this? Yell and scream? Curse my state’s poor practices?
Or maybe do town halls with legislators?
Where abouts in the state are you? I am near Vancouver Canada but the whole infrasture efficiency challenges that Metro-North and CT Rail are fascinating to me (but understandingly frustrating for you). What I would do is find a member of the minority party of the State Legislature (e.g. Tony Hwang) on the Transport Committee. Passenger rail is discussed on slide 27 of “2018 Connecticut Statewide Long-Range Transportation Plan” presentation on the CT Long Range Transportation Plan 2055. I am happy to help with a bit of research as I have done similar volunteer activity here in British Columbia including getting to meeting with the transport minister which was quite a challenge.
I would also talk to staff at your planning region to see what their positions are and then contact your regional representatives. For example if you were in New London I would refer to the following document: SOUTHEASTERN CONNECTICUT METROPOLITAN TRANSPORTATION PLAN. Than I would first write to the transport planner and local represtative involved with the plan an ask for a meeting with to discuss what is being planned and how the City/Region is involved with promoting passenger rail as part of the state’s 2055 Long Range Transportation Plan. In New London I would contact
Katherine D. Rattan, AICP, Planner III, Transportation Program Manager at the Region
The City Mayor Michael Passero
Councilor John D. Satti who heads public works committee
If you are a regular transit user, you might want to talk to other riders you know and form a public transit riders group.
The Chambers of Commerce might be a useful ally if tourism is important locally.
Let me know if there is anything else I might be able to help you with, you can email me at thomas.cheney at gmail.com
It seems like capacity is highly limited in the New York City Region. If the long-term desire is frequent service to many destinations with limited new , would it make sense to use dividing trains to squeeze more capacity through the tunnels? For example, why not:
For MNR/ New Haven Line have
For the LIRR
For NJT
I know there are some delays with dividing trains but they seem to be a lot smaller compared to locomotive switching. Has such an approach been analyzed?
Your section on junctions says that Danbury and Waterbury branch trains stay within Connecticut with transfers to GCT, but your Takt diagram shows Danbury branch trains going all of the way to GCT, and doesn’t show Waterbury at all.
You need to include the cost of commuter rail rolling stock in your budget. Your entire plan revolves around getting HSR speeds out of the NEC without major infrastructure work to get passenger/HSR dedicated tracks as are found in Japan, China and France (mostly). This makes the commuter rail improvements non-optional in order to get your schedule to work and allow the intercity trains to make speed. If you wait for rolling stock to be replaced as part of normal recapitalization that means this plan cannot happen for 25-30 years when every commuter rail agency finally makes it around to replacing their newest equipment. If they even buy modern EMUs then, because the EMUs are useless without electrification, but no-one will spend money electrifying a line that doesn’t have EMUs to use it and leaving catenary hanging there for no reason. Every electrification project in the world buys new electric rolling stock as part of the project that hangs the wire, this one can be no different. Include the cost of the EMUs in your plan. Yes it will drive the cost above $20B, its still a good deal.
You need to provide a full day’s schedule, laid out like an Amtrak or appropriate commuter agency schedule. It’s fine to explain how takt works, but only giving southbound departures and then saying that northbound departures are symmetric about the hour is not comprehensible to almost anyone outside of the kind of people reading this blog. If you want policy makers to push this or citizens to vote for it you have to give them something a layperson is familiar with and can understand. Make it an appendix if you want, but provide a full and complete schedule with all stop times.
You also need a route diagram showing the service patterns, like a traditional transit map. The network diagrams are good for their purpose, but they don’t, as the biography of George Dow was titled, “tell the passenger where to get off.” Or in this case, tell people where they can go with this plan, in a way they are familiar with and grasp the improvements, not an abstract “TPH here is double than before.” This is especially because you are expressly planning for through running of certain NJT and MN or LIRR lines, which is a HUGE deal, as significant as providing HSR, maybe more so given the larger passenger volumes of intra vs inter-city traffic. You could miss this fact if only reading, you need a visual that makes it clear which services are running from where to where and what stops they call at.
You should give names to the various service patterns. People can understand, and more importantly get behind and advocate for, things like the “Keystone Express” or the “Hartford Flyer”. Give them the hook that makes these ideas sticky.
Am I understanding correctly that you are eliminating all intercity rail stops on the NEC except for the 10 in your Base Stopping plan? I’m sorry, that seems like a non-starter. Even the Tokaido Shinkansen runs the Kodama despite there being Tokaido Main Line exists and also provides regional service. The stops along the NH-Kingston bypass are one thing, but to say you are bringing HSR to the NEC then telling 13-14 stops they no longer get any intercity trains at all makes this plan politically and socially impossible to get adopted, no matter what the technical or economic merits.
The tracks for Shore Line East will still be there. Unlike Japan the stations, rail gauge etc for the high speed trains are the same for the low speed trains. Once every two hours a train can wander back onto the old tracks. The other every other hour it can wander to Vermont.
You don’t understand. Without it it ruins the simplicity and symmetry of having the trains glide smoothly past the stations in blissful taktness.
I suppose the masochist in Aberdeen could go to Seabrook instead of Odenton and go few blocks to CVS and have lunch in the Seabrook 7-11.
I’m not quite clear why the commuter trains have to stop at every other express subway stop but it’s just peachy that the suburbanites have to change trains to get someplace other than places with express subway stops.
Is 7% pad a reasonable request of the NEC at the start (really, a request of Amtrak, Metro North, etc.)? Would 10% be a lower risk start to plan around? Swiss and Dutch run probably the two best rail networks in the world, as least as far as network performance goes. Can US rail operators match that from a cold start? Did the Swiss and Dutch even pad at 7% when they first introduced takt? If the Japanese (the other candidate for best rail ops) does 3-5% pad on totally segregated tracks, and the TGV does 10-14% on tracks that are mostly segregated with a lot of point-point patterns that don’t require overtakes, can Amtrak and others really run the NEC with its multiple branches and service patterns at 7%?
The cost of high platforms seems large vs electrification. It doesn’t seem right that you can renew 363km of catenary (including 4-6 track sections) for just $1000M and new build 400km of catenary for $820M, but spend $2200M on just high platforms for 80 stations.
It appears that you are using international costs for electrification at ~$2M per km of track, but using real world costs from the NEC today for platform raising at ~$25-30M per station. You can’t mix the two. Either US transportation agencies reform themselves to be able to build things at international costs, both catenary and platforms, or they don’t. If the former, I found a reference to building a high platform in Britain for GBP900k in 2012, which would be ~$1.8M dollars today, or say $2-5M for a station with one or two platforms plus ancillary costs. If the later, you need to factor $10M per double track km and increase your electrification costs by some two and a half times.
Also, why are you including electrification and high platform costs for the Morris & Essex and Montclair/Boonton lines? Doesn’t your plan say that all NJT lines except for NEC and Coast lines are using the new Gateway Tunnels, so there will be no interlining in the North River Tunnels with the intercity trains?
NJtransit and Amtrak rebuilt 100 track miles of catenary between New Brunswick and Trenton in 2012? and apparently started to use it at higher speeds in 2022. If I remember correctly it was $400 million but it covered work in Penn Station New York and substation upgrades outside of Newark? ….It wasn’t all spent on catenary.
The Morris and Essex is electrified, has been since Thomas Edison himself ran the inaugural train in 1931. Connecting the Morris and Essex to the Boonton Branch, in Montclair, included electrifying to Montclair State University. So the only thing left is Montclair to Dover. Past Dover is interesting. NJTransit is gonna restore service to Andover any decade now. Pennsylvanians who get stuck in traffic crossing the Delaware within sight of the DL&W bridge are hoping for trains someday. Fanboys didn’t consider service to Allentown or Scranton. Sending a train once every other hour to Scranton and once an hour to Allentown screws up the beauty of keeping things simple.
The stations along the NEC have high platforms. If the train is loitering excessively in Maplewood because they can’t figure out how to put in level boarding it’s not going to delay trains on the express tracks in Princeton.
Re BEMUs vs electrification. I was surprised to learn over at Caltrain-HSR Compatibility Blog how good the recharge performance is for the Stalder KISS BEMUs Caltrain is planning to get for the service south of San Jose to Gilroy. Apparently they can do a 48 km trip with just 2.5km under wire, plus a short stretch of catenary at the terminal station platform for recharge during dwell. The Danbury/Waterbury branches are 38/45km respectively and will spend a lot more than 2.5km on the NEC mainline. If they are only going to get 1 train per half hour, is it worth spending $190M to electrify them versus buying a few BEMU sets? Even if either branch operates as a shuttle part of the time, wouldn’t the cost of wire just at the layovers be more cost effective?
I absolutely agree with you when it comes to long distance trains, but is it really a good idea to not have at least some of the Virginia services run north as they do today. The line to Newport News sees 1.21M pax per year over 287km, compared to the Keystone 1.27M on the Keystone service over 155km west of Phila. That is a little more than half the pax/km despite the Newport News branch having only 2-6 tpd with rides of 3-5 hours to DC and 5-8 hours to New York while Harrisburg sees 13 tpd and is only 2 hours to Philly and 3-4 hours to NYC. Richmond Staples mill sees more pax than any Keystone station (and more than Stamford!), and Alexandria almost as many as Harrisburg. It would seem those numbers would jump if those trains were much more frequent and continuing on the NEC as HSR. Also see above my comment on restricting service to only 10-11 stations.
My understanding is that this report is only permitting the use of Zefiro-type EMUs on NEC tracks, and extending trains past the dedicated NEC will disrupt the already tight schedule. 16-car trainsets are also overkill south of DC under current ridership and track conditions.
In the medium-term, Amtrak can run an hourly Airo relay service to Richmond and onwards to Newport News, Norfolk, and Raleigh after the S-Line is refurbished. Even with a 20-minute transfer, passengers save 1 hour when traveling to NYC compared to the existing NER.
In the long-term, SEHSR is built and all trains continue to Richmond. There, each 16-car trainset is split into two halves: one going to Norfolk / Newport News and the other continuing to Raleigh / Charlotte / Atlanta.
Right now, the line to Richmond is unelectrified and has an average speed of only 38 MPH. High speed trains are expensive, and according to Alon they depreciate mostly based on their age rather than mileage, so having high speed trains crawl along behind a diesel locomotive at 38MPH when they could be cruising along on the NEC at 100mph is a serious waste of money, especially when the Northeast railroads will have a large collection of unused trailer cars after they convert their commuter services to multiple units. If the line was electrified and upgraded to a more reasonable average speed like 60mph, then through running might make sense.
Electrification should be on the cards for Virginia, and proper high-speed service on the NEC would obviously boost the cost-benefit ratio there considerably. If you electrify, you would probably also want to improve the alignment, e.g. along I95 between Fredericksburg and Ashland, so it starts becoming a massive project.
I can see why Alon’s report would refrain from committing to it to avoid additional scope, but it could mention it as a future possibility.
In the meantime, would using bi-modal units for hourly trains to Richmond/Norfolk/Newport News be a viable option? Or would their performance characteristics impair the functioning of the NEC too much?
A point of discussion more than a criticism, but I feel you are using “schematic” and “diagram” backwards, at least as I understand their common use. What you label “schematics” when discussing 1 and 2 dimensional networks I have always heard defined as diagrams, because they maintain correct – if abstracted – spatial relationships and general features. For instance, there is a coastline with blue where the water is, and the line from Rochester to Schenectady is E-W while the line from Albany to NY turns N-S instead of the two segments being a single straight line. (On the other hand, these are not maps because the special relationships are not to scale or exact in their locations.
What you label “takt diagrams” on the other hand should be called schematics. Unlike the network drawings, the takt layouts only show the abstract schema of the service, without any special relationships. For instance, the size and orientation of station boxes is solely based on the number of lines serving them and where they come from on the page, not due to actual size or even volume of trains, for instance the size of the Rahway box because of the choice to have certain lines make a 90 degree turn there, or the fact that Harlem/125th is due “West” of Stamford when it is really more N-S.
I’m aware there is variability in use here (most “circuit diagrams” are schematics) and also that some consider any graphic representation to be a “diagram” (and thus would speak of a “schematic diagram” or a “scale diagram”).
Sould I be concerned about platform crowding in NYC? Since IC trains going both directions berth at the same platform. If they arrive at the same time and if we assume 50% turnover, you could have 1000 passengers waiting on the platform while 1000 passengers disembark.
I’m not that familiar with NY Penn, but I keep hearing the platforms are narrow.
Penn station has 22 tracks. With better operations (commuter rail through running, european-style switches that allow faster speeds through the station, not refilling Amtrak’s cafe cars in Penn Station, not requiring brake tests to be performed in Penn Station) there should be plenty of track capacity and you could avoid using tracks on the same platform at the same time or, even better, pave over some of the tracks to create wider platforms.
East of Penn Station there are 8 million people on Long Island and 15 million in New England. West of Penn Station is the rest of North America. I suspect there will be trains, from the west, terminating in New York because “rest of North America” will have more demand.
If they don’t test the brakes when they change ends of the train how do they know the brakes are still working, using the other end? It’s very very rare but sometimes the brakes don’t work when they change ends.
I suppose they could close the cafe and tell passengers it offended railfans that passengers get thirsty and hungry on long trips.
There is no reason the cafe car needs to be filled in New York. It is important for long distance trains, so fill it in Ohio (farm country) or something. Or for NEC trains, any city along the route will do. The place to fill the cafe car should be picked by logistical needs, a constrained passenger station in a large city is not a good place to do this work. It might even be possible to not have the Cafe car at a platform (there are many considerations in play here – don’t read this as an argument you should, only that you don’t have to if the other considerations don’t force it).
Different fills have different needs. I don’t want someone with a box of frozen burgers in Penn station. That box needs to be shipped (via truck!!!) from the butcher outside of the city and will be taken back out needless adding to traffic in the city. However water tank fills can be done via a hose on top of the car and wouldn’t impend the platform, or add traffic to the city.
The long distance trains get serviced in Queens. Those wily wily civil engineers had been fooling around with railroads for 75 years when plans were made for Pennsylvania Station New York. They bought up swampland in Queens – someplace outside, in the sun… the Sunnyside… to service the trains. Other Amtrak trains and commuter trains get serviced out there too.
They could either not use cafe cars on the Northeast Corridor trains– once they are HSR even a full Boston-Washington trip will only take 4 hours and all of the stations will have places to buy food– or refill them in Boston, Washington, and the other terminals. Boston south station has 12 tracks. If you give two tracks to each of the 4 commuter trunk lines that come out of the station, (Boston doesn’t have demand for more service than 1 train every 7.5 minutes on each trunk) that leaves 4 for Amtrak. The initial HSR will only serve South Station every 15 minutes so Amtrak can have extremely slow 30 minute turnaround times and still be fine. The other HSR terminal stations are similarly oversized– Washington Union Station has 30 tracks, more than Penn Station! The initial plan actually does not have trains terminating in New York– trains either run from Washington to Boston or from Harrisburg to Springfield, MA. Anyway, if trains terminate in NYC, then they can be serviced in sunnyside yard.
As for the brake tests, for one thing, once they adopt through running, most trains will leave Penn Station in the same direction that they arrive, so they wont be changing ends. For another thing, if they test that the brakes are working properly in both directions in the beginning of the day when they bring the train into service, there is no reason to expect that they would stop working just because they change directions. FRA regulations allow trains that don’t switch directions to go up to 1500 miles between brake tests. Even if a commuter train traveled at an average speed of 60mph for 20 hours without stopping it would only go 1200 miles, so there is no way a commuter train could exceed that milage in one day.
I’ll bite. I’m taking the train from Boston to D.C. How do I buy food in Philadelphia?
No there won’t. There are 27 million people east of Penn Station and the rest of North America west of it. There will be trains terminating in New York because there are more people west of New York. Unless you think running empty trains through the East River Tunnels is a better use for them than running full LIRR trains through them.
Are we talking about the reality today, or what it should be.
What is should be is frequent all day service – in https://pedestrianobservations.com/2020/02/13/metcalfes-law-for-high-speed-rail/ Alon calculates Boston to D.C should generate 400-500 passengers per hour. (some segments of the trip generate far more). That is enough to run at least 3 trains per hour (generally calculations for the whole route suggest 10-15 trains per hour). Except in the peak times you just get off at Philadelphia, find some food either at a stand in the station food court (which should exist as that rent is a good money maker for the station), or walk to a nearby restaurant. Either way, when you are done eating you just go back to the station where it is at most 20 minutes (6 minutes when considering all the other segments) for the next train – and since the train is running a schedule you can look at the clock while eating and so make a decision of eating fast or lingering thus making your wait much less.
The above requires not running like an airline where you try to fill all seats all the time. Trains should be less than 70% full most of the time. This means your ticket can be for the full trip without concern for which exact trains you are riding. Amtrak is not running like that – but they should.
500 people an hour is half of 16 car train.
What part of “Boston to Washington” was unclear? Pissing away a half hour in Phildelphia doesn’t help me achieve that goal.
500 people per hour is not enough to justify non-stop Boston to DC service. That route works out as worth doing because it stops in places like Philadelphia that the train is padding anyway, thus allowing someone to go DC to Philadelphia, or Philadelphia – not to mention all the other cities. There is of course trade offs – a non-stop train would be faster for you but we lose anyone on the other cities. Thus debate elsewhere on if the train should stop in smaller cities like Stamford.
Remember that I specified the NEC as it should be not as it is today. The train should not stop for half an hour anywhere – but if you are hungry get off at whatever station you happen to be at, eat, then catch the next train. This is easy enough that anyone can figure it out. It isn’t pissing away time, it is eating where you feel like it. People are smart enough to allow time in their schedule to eat. Or if you don’t like it pack a lunch (no peanuts) and eat on the train.
Because the report calls for the use of the same platform, and I believe with the symmetric timetable they’re scheduled to arrive at more or less the same time.
Take two deep breaths. In nice round numbers someday far far in the future, assuming no more tracks or platforms get built, there will be 40 west bound trains using 10 tracks and 6 island platforms. And 40 eastbound trains using 10 tracks and 5 island platforms. 40 divided by 10 is 4. Four trains an hour will be using each platform face/track. They can do something magical where the train heading east uses a different island platform than the train heading west. I’m sure the report has lots of other clueless railfannery in it too.
There are 21 tracks in Penn Station. Why would a southbound train and northbound train be at the same island at the same time?
They should berth at the same platform, but if nothing else uses the same platform then there is less crowding.
Why? There are 21 tracks and 11 islands to put trains on. Unless you are worried the intercity trains will get cooties from the commuter trains.
Birmingham new street loads the cafe car quite sensibly with a 4 minute stop when there is a delay. It also has ~40 trains an hour on 13 platforms with a fair number of terminating services.
That sort of thing might at best be achievable in America, assuming you accept some of the British disadvantages as well.
One think about Japan also is that it doesn’t seem like trucks and buses have speed limiters like Europe. And that they are at times also going much faster than the speed limits.
it appears the railways are at times also driving extremely aggressively to keep them on schedule. I wouldn’t be at all surprised if on the Tokadio Shinkansen they did up to 320 km/h or even more to make up delays. 230I will d
320 would be 20% padding
The Tokaido Shinkansen services are limited to 285km/h, due to current braking performance of N700A stock as well as curvature restrictions (2500m radius in some sections). Timetable recovery is performed not by speeding up a service, but by merging services or outright shortening/canceling them- the goal is to restore stability to the overall operating timetable, not to get individual trains closer to their timetabled running at the expense of propagating knock-on delays- this is an extremely dense hsr operation with 3 min. peak headways after all. On legacy lines- ATS-P and ATC signal systems have overspeed control so going over timetabled speed limits is impossible.
When I recently stayed in Tokyo I stayed in a hotel between Shimbashi and Yurakucho. I saw several Shinkansen leaving Tokyo and taking 2-3 minutes to reach the hotel which points to them departing maybe 1-2 minutes late. Now sure these are evening peak trains but even so you aren’t hitting the Shinkansen’s legendary sub one minute average delays with that sort of relaxed departure from Tokyo without a significantly larger padding than 4%.
As commercial road vehicles are clearly not speed limited in Japan like they have been in Europe since the 1990s it seems highly unlikely that Japanese trains are speed limited and that the most plausible explanation for how the Shinkansen meets it’s legendary on time performance is because the padding is in reality significantly larger than 4%, presumably due to drivers meaningfully exceeding the speed limits, just like Japanese truck drivers clearly do.
There are countless YouTube videos from a myriad of users who pair cab/window views from conventional and Shinkansen trains with GPS speedometer traces (some search keywords: 車窓 and 速度計). Not once does a train meaningfully exceed the ‘speed limit’ on the line, barring the sub-5 km/h fluctuations of the GPS measurement.
Schedule padding and operational discipline can easily explain how delays are kept to a minimum. Indeed, the same videos regularly show trains cruising 10 to 20 km/h below the nominal speed limit and arriving at the next stop precisely on time.
OK so the existing journey padding is significantly larger than 4%. That would be a plausible explanation to me!
Matthew, the correct response to information correcting free-floating hallucinatory supposition is “sorry, I was wrong”, not to double down with more misinterpretation Give it a go some time! It might be cathartic! Or at least provide a fresh experience.
In the schedule section, you seem to assume that NJ Transit trains are turning at Trenton. This is not the case. Most of those trains run through to/from Morrisville. Because of the configuration of Morrisville interlocking, they are either using the express tracks between Trenton and Morris, or they are crossing over the express tracks at Morris.
I’m not sure if that changes anything in the grand scheme of things.
The trains to the former “Erie” lines will be diverging off at in Secaucus. The trains to the former Delaware Lackawanna and Western, the Morris and Essex lines, will be diverging off in Harrison. The former Central of New Jersey trains diverge off to the Raritan Valley between Newark and Newark Airport. The North Jersey Coast trains diverge off in Rahway and the Jersey Ave. locals will be extended to North Brunswick and have a loop track to turn around on. So in some far off future it’s five or six NJTransit trains and five or six Amtrak trains. And a SEPTA train toddling in and out twice an hour during rush hour. Much less busy than it was in the heyday. If it’s ever a problem there’s plenty of space to do something.
Would you be able to expand on how you came up with the timetables? As in, how did you use Devin Wilkins’s code, how did you figure out running times between stations, etc?