The article doesn’t understand dwell times at Penn Station or the capacity limitations there.
For example:
* Penn Station Dwells are not limited by passenger movement at doors. It’s not a metro, go and watch trains being loaded.
* Escalators constrain the rate at which people can actually reach the platforms
* These escalators need to switch direction when a train arrives, meaning multiple arrivals and departures are not simultaneously supported.
* The tracks do not support all operators and the station design divides Amtrak from LIRR
* Some tracks do not support full train lengths
* Not all tracks in the station are through tracks
* Trains are terminating at NYP which requires a longer process of ensuring no passengers are still on on board before leaving for the yard.
* It’s not only the north and east river tunnels, there is an entrance from the Empire line for Amtrak and separate paths to the WSSY for LIRR, not to mention train storage within Penn in the ACDE yards.
* Engineers walk the entire length of the train when a trip reverses direction. This takes a long time. The OP mentions through-running solution, but doesn’t point out the problems with it, namely incompatible equipment and giving an entirely different agency access to your multimillion dollar trains.
> Escalators constrain the rate at which people can actually reach the platforms
Let people wait on the platform as the train arrives. This is something that basically every train operator but Amtrak does.
> The tracks do not support all operators and the station design divides Amtrak from LIRR
This isn't a problem... trains in most stations are already basically scheduled to an individual platform anyways.
> Trains are terminating at NYP which requires a longer process of ensuring no passengers are still on on board before leaving for the yard.
Through-run trains. Yes, this requires cracking some heads at NJT and LIRR (and MTA before much longer). Is it really worth spending billions of dollars to avoid having to resolve agency turf fights?
The platforms are not large enough to support people waiting at the same time as the train is unloading and are full of columns. There are safety concerns in not being able to evacuate passengers quickly enough in the event of an emergency via narrow escalators.
Through running NJ Transit onto LIRR is through running AC trains onto a DC network. That is billions. In addition, nobody from NJ wants to go to Long Island and vice versa, and most NJ Trains run-through the station in the same direction already to be stored at Sunnyside. The same is true of LIRR, they go into the west side yard as a straight move. The NJ trains that don’t run to SSY are usually making use of the stub end tracks at Penn where through running isn’t an option.
> Let people wait on the platform as the train arrives. This is something that basically every train operator but Amtrak does.
I haven't been to penn station in a while but I know the LIRR track numbers aren't available until the train arrives or right before. This dictates which platform to stand on.
Basically you end up standing in a huge open lobby and eventually a big display says to goto track 21, then there's a mad rush of people making their way to track 21 which involves leaving the lobby, going down stairs, and making your way to the track to board the train.
Los Angeles Union station turns around the Pacific Surfliner because it's a end stop - and they've now been working for 10+ years on adding a through track which would shave 20+ minutes off the trip.
And that one doesn't need anything but a bridge over a freeway.
> Let people wait on the platform as the train arrives. This is something that basically every train operator but Amtrak does.
In no small part because the platforms are too narrow — something that getting rid of several tracks would readily fix. (Just fill in the space between two platforms and you have a nice, very wide platform.)
Not directly in this, but Alon (the author) has consistently harped on how bad US train operations are.
> * Escalators constrain the rate at which people can actually reach the platforms
> * These escalators need to switch direction when a train arrives, meaning multiple arrivals and departures are not simultaneously supported.
> * Some tracks do not support full train lengths
> * Not all tracks in the station are through tracks
Alon isn't against a station remodel. He is against this one because it is a bad plan. Platforms should be full length for the train that will use them. Platforms should be wide enough for people to wait on them. Penn station needs less platforms that are used better, not more of them that cannot be used very well.
Just a personal anecdote, but I was waiting for a train in Osaka with a friend. It arrived 2 minutes before our departure time so we got on. Then it announced departure 1 minute before our departure time... we looked at each other and jumped at the door with it closing on us and our luggage. With a nasty look from the other people on the train we got off... and waited another minute until our train arrived.
The trains in Tokyo all seem to have nice new signs with upcoming train info next to each of the tracks. That makes it much harder to get on the wrong one. Also, google maps works disturbingly well for Tokyo subways.
In Japan, is the "departure time" understood to be the time at which the doors close, or the time at which the train starts moving? This is a common debate in Britain:
Personally I think that for the convenience of passengers the doors should not close before the published time, which is traditionally called "departure time" but I don't really care what they call it; the point is that people need to know the time at which they need to be on the train and the easiest way to tell them is to tell them directly without any bizarre offsets.
This is an unbelievable debate for me as a citizen of a smaller city with a much less developed system. Our system is so far away from being able to accurately time trains like this that asking whether it should be door close or train moving is like debating how many angels can dance on the head of a pin. I travelled to London a few weeks ago and was shocked that the trains would come when the schedule said it would. I'm used to +/-5 minutes on a good day, but with the max delay much worse than that.
Trains should run down to the second in the majority of cases. There should be enough padding in the system that they can finish any segment in 10 seconds less than scheduled time if someone holds a door open. Computers are very good at driving a train at exactly the planned speed, and will also stop with the doors at exactly the same place on the platform. Computers can open doors at exactly the right time. In some old stations you might need a human to verify the doors are safe to close (some stations have curved platforms which means there is a bit of a gap and someone can get stuck in an unsafe place), and hit the ready to go button. You might have a human up front to hit the emergency stop (ie someone on track), but there is no other use for a human driver these days.
The above is about trains. Buses and trams are more complex and cannot run this way.
The problem is with the word "departure". Just call it allaboard time. Its the time when the old timey fat conductor yells "ALL ABOARD" and then the steam starts to whistle out the smoke stack.
If the schedule depends on the train actually beginning to move at "departure time" then the doors need to start closing before that, because if someone is in the door it will reopen, beep, and reclose.
However, you could have the doors close at the beginning of the minute, and the train leave at the end of the minute and have 59 second leeway.
Airplane tickets have both boarding time and departure time listed.
Things I walk up to momentarily before departure: every mode of transportation except planes. For trains and buses, the two times should be almost simultaneous (as anything else is just adding additional linger time)
Which varies in my experience. If there's an available window 5-10 minutes early, I've absolutely seen planes take it. And, of course, conversely delays are not infrequent. They presumably won't leave before boarding time (and don't think I've ever observed that) but I have seen boarding time be very close to door closed time. I've had connections that seemed pretty comfortable and I've ended up being pretty much the last person on the plane.
I once saw a plane leave before boarding time. Everyone was in the terminal already and bad weather was coming so the pilots rushed boarding to get in the air before the storm (well they tried: we spent 20 minutes on the taxiway waiting out the storm, but if they had been 1 minute faster they would have been able to get above the storm before it hit)
It is probably the anchor for everything else. The plane had scheduled a flight plan based on that departure time. Check in might close consistently X minutes before that, but sometimes not. I have been at an airport (pre covid) 3 hours early and only just made it on the plane due to airport chaos.
The reality is that transportation operations involve multiple events, each of which have a specific timing factor and variance.
There's arrival and departure (to the gate, platform, or quay) of the vehicle / vessel itself.
There's ticketing and check-in time, if those are required. Possibly security screening for air travel. And for aircraft, gate arrival time vs. wheels-down / wheels-up, where the latter includes taxi and takeoff queue time. Add to this the reliability or unreliability of actual travel times --- busses and shared-route transit must deal with traffic and delays, aircraft with weather and airport conjestion, ships with dock and pilot availability as well as weather factors, for example.
Schedules are written from several points of view --- the operator (when must vehicles be at specific points or departing from them), passengers, port or terminal operators, etc. Each has different needs. Learning how to interpret the published schedule to your own planning is key to effective use of such systems.
When the doors open, depending on traffic, they easily spend less than 20 seconds open before closing. I've run for a tube and dove in just in time _so_ many times now.
Yeah, the upcoming trains are usually shown on these. There is a difference though: In Germany the train will usually be 10++ minutes late or straight up never show up (at least for long distance travel).
That heavily depends. e.g., ever since the IC208/209 has been replaced with a Br 412, it’s been far more reliable. Which is why e.g., in Düsseldorf 1038 and 208 can now be scheduled on the same platform, so in the morning you’ll have the 208 at 6:32 and the 1038 at 6:38.
Where I used to live, the digital signage only showed the schedule, not the actual train ETA. The schedule was almost completely useless, however, the online website (at the time) didn't show the schedule for every station (edit: they still don't). So there was some use, at least. Not that it mattered, since the train never met that schedule, outside of the first run of the day.
Then covid hit, the city paused train service, I moved out, and I really don't care to commute to work (or school, at the time) again.
Amtrak had added the signage when I was using it in Southern California, and it could be fun to watch it continually update the "expected arrival time" which seemed to only jump by five minutes even when the train was physically thirty minutes away. The website was much more accurate about where the train was and when it would arrive.
In Hamburg there are similar, but slightly smaller things at almost every bus stop, indicating which line to which destination arrives next in minutes.
In 4 rows for a single panel, where larger bus stops have more panels. Or also smaller ones, single lined under the freestanding roof in some. Also delays, if they happen. Seems like they are all over Germany, too.
Not exactly 'comfortable', but informative. Don't have to use an app, don't need to look at the timetable, don't have to wait in uncertainty. Just go, see, ride, done.
I think the commenter was making the point that, whilst blindly getting on a train arriving at the correct platform 2 minutes early would work in most parts of the world, it doesn't in Japan.
I once boarded a train that had arrived at the correct platform, on time, and to the correct destination. It turned out, though, that I had a ticket for a different train that was late.
In NL, trains indeed do not have route numbers, and the vehicle (carriage) numbers are usually not for public consumption. The one exception is for when the train splits halfway through the route: it'll be announced as "after the next stop, carriages A and B will continue on to X; carriages C and D will go to Y".
That said, we have had dynamic train information displays on all platforms for as long as I can remember (wikipedia says since 1961). When I was young, they used to be mechanical rotary displays [0] (klapperborden, aka "clacking displays", referring to the sound they make when the system was leafing through the pre-printed sheets), but they have been replaced with TV screens or LED matrix displays in most places now.
The group I work in (Amey Consulting) have built a couple of systems for Network Rail here in the UK, called Quartz and Cosmo. These are analytic systems examine rail delays and delay propagation, especially sub-threshold delays (in the UK delays over a threshold - 2 mins IIRC) have to be reported, but those under 2 mins don't. We started looking at sub-threshold delays a few years ago, and have understood how these small delays build and propagate through the network, affecting multiple services.
We are always recruiting (UK - main sites London and Manchester, some folk in Birmingham too) - so if you have an interest in applying data science to infrastructure have a look here: https://www.amey.co.uk/what-we-do/amey-consulting/careers-wi..., or DM me
Out of interest, what kind of state is software engineering in the rail industry in?
Watching a documentary about Crossrail the other day their testing seemed to consist of a driver manually driving a simulator in real-time through signals and recording when buttons don't work etc. I was kind of hoping that was just a misrepresentation.
I assume part of the reason for doing it like that is that it gives user feedback for the systems rather than just following what ever assumptions they put in place for their automated testing. It also servers to be driver training, which is very important and takes a long time, so if they can start sooner its probably a good thing.
I saw the same program and was cringing. "The train has crashed" now meaning its software has crashed and it needs a reboot. It is not frightening that it might physically crash the train, which I assume has crazy numbers of failsafes but more about the terrible reliability of some software which has unexpectedly crashed despite whatever modelling/testing was already carried out. If it is that unreliable, it has to have loads of failsafes which means lots of unexpected failures.
At least with older trains, you could usually move the train in emergency, even if slowly, but having the software tell you that you can't close the doors or pull away is really bad.
There were also issues with trains missing from the passenger information systems and even worse, a broken down train on the boundary between TfL-controlled and Network Rail-controlled lines and it sounded like no-one knew what to do and it caused 2 hours of messing around.
Indian railways is really good at managing much higher traffic at lower number of platforms and tracks in general. New infra upgrades are much slow and expensive.
So at a majority of high traffic stations, track sections IR has to do a really tight exercise of managing schedules and maintenance.
They also have recently improved on time performance. It is not Japan like or anything, but it is getting better.
Indian Railway must have some fascinating stories. It's absolutely ginormous and historic. Isn't it also the largest employer in India, which must put it in the millions? The fact it works at all shows there must be something special about its structure.
Yes, IR is largest employer employing 1.54 million people.
There has been a lot of automation in signalling, switching recently, because of which we will see number of employees going down in near future.
But beyond this, yes it is a humungous project that keeps humming without major glitches. Supported by a lot of workers on the ground doing tasks like track inspections, cleaning, maintenance day and night.
I am not sure about that factoid, but before pandemic in FY 2019-20 IR did 8.08 Billion trips. IR runs in losses for passenger service, it makes up for those losses in freight service.
> The biggest required investments for this are electrification and level boarding. Both have many benefits other than schedule reliability, and are underrated in Europe and even more underrated in the United States.
Neither is underrated in Europe. Both are the norm on new projects, and almost all heavy duty old lines have been upgraded to at least be electrified. However the costs associated are often astronomical ( and might require closing the line for extended periods of time) so for some of the rest it's done piecemeal.
In Scotland the electrification of part of the (200 year old) network involved rebuilding every car and foot bridge and tunnel along the route to add enough spare height for the overhead wires. It's one downside to being "first" to build widespread railways - with the initial huge investment made in Victorian era when trains were single decked and run on coal, the room for expansion was limited. The good fortune Britain has had in avoiding real destruction of infrastructure in war has also meant there hasn't been much opportunity to rebuild having learned lessons since the Victorian era (separate cycle lanes? Hah). Meanwhile, in Hannover for instance, U-Bahn stations were built in the bomb craters in the city center - making use of the "free" excavation.
Obviously not saying war or coming later to the party are good things, but it's one of those ironies of humanity.
There is an interesting development going on to electrify some of the UK rail lines - an OverHead Line (OHL) - battery hybrid system.
The idea is that you have a standard electrified OHL system where it is easy and cheap (!!!) to install. Places whee you can't put it easily, such as under bridges, through tunnels etc, you don't. You have some (small-ish) battery capability in the train which recharges from the OHL system, and use that in the areas where the OHL hasn't been installed.
If somebody wonders why costs are high: It's not only about putting a few masts and a cable up. Aside from sub stations etc to actually provide the power the big things are bridges and tunnels.
If a train goes below a highway bridge it's probably too low for the wire, thus the highway bridge has to be replaced (or tracks be lowered) If there happens to be a tunnel the diameter has to be increased.
.. and incompatible with existing train systems in most regions.
In many systems you currently have a central hub which is overhead wired. If you then have a unwired branch line installing a third-rail-system there would require vehicles which can use both. Such systems exist, like the underground in Nuremberg, Germany, which operates with a third rail on tracks, but in the maintenance area has overhead power to make work safer. However that increases cost by requiring custom vehicles etc.
Considering DB just ordered ICE 3 neo of the Velaro D series without level boarding, ICE 4 is also without level boarding, and the only new train will level boarding will be the Talgo ECx, at least in Germany it’s quite underrated.
The problem is that for high-speed trains you're limited in how small you can make the wheels (that is, even more limited than with other trains – even urban railways with more modest top speeds need a floor height of approximately 0.9 - 1.0 m for level boarding without having to compromise on the number and location of doors, though with some compromises you can nowadays get down to roughly 0.6 m), and with conventional bogies (and both wheels linked by a fixed axle, which is you how you classically get the wheels to self-steer themselves around curves) it means you need a floor height > 1.1 m for a level interior and level boarding.
Conversely for historical reasons however in Germany 0.76 m is the highest platform height that doesn't conflict with the loading gauge of "regular" exceptional loads on freight trains (on a very few selected routes intended for the transport of extra large transformators and similar loads actually anything higher than 0.55 m already causes problems).
If you're building a pure high-speed network from scratch without any compatibility constraints (like Japan did with the Shinkansen) you can just choose a platform height of something around 1.1 – 1.25 m and be done with it, but if you're starting with an incrementally upgraded mixed-traffic network as is more common in Europe, you now have a problem…
(Talgo sidesteps this problem because of the unconventional bogie design, where the two wheels aren't linked by a fixed axle and steering is assured by the interaction of neighbouring carriages instead, so even with a lower floor height the floor can simply pass between the wheels without necessitating any steps.)
That'd make sense, if trains like the RRX didn't support 220km/h at an entry height of 520mm and long-distance platforms didn't have a height of 960mm already.
RRX? The double deck EMUs I'm aware of go up to 200 km/h, though maybe some designs could be pushed to 220 km/h. In some way yes, touché, although that's still not quite the same league as the class of 250 - 320+ km/h trains.
Another counterexample might be the Giruno (250 km/h), but that in turn needs to compromise on its door location (only a single door per coach, and placed centrally in the middle of the coach).
> long-distance platforms didn't have a height of 960mm already
They don't. 960 mm is only used in various S-Bahn systems and not compatible with the structure gauge of the general railway network.
> “It is bad enough that Germany is keeping some outer regional rail branches in the exurbs of Berlin and Munich unwired; that New York has not fully electrified is unconscionable.”
There is a strong argument that battery-electric trains will replace the need for overhead electrification on shorter routes and branch lines.
Track electrification can be very expensive, and hard to justify on less-busy lines. But with batteries, a train can charge while running on an electrified trunk line, then continue onto smaller branch lines.
Battery trains aren't a new thing. Prussia had been using them since 1907 and German federal rail later used the newer Class 515 [1] for decades. They had a reach of about 300 km (theoretically and in level areas).
For the 515, costs were a big factor. The lead based batteries had to be rented from the manufacturer.
But even if you take current technology into account: A train isn't a Tesla. It needs much more energy to accelerate. Currently, an engine and a diesel tank are much cheaper to buy and especially maintain. They will give you decades of use. How long are your batteries going to last?
Also keep in mind that batteries add tremendous weight, reducing reach again. More weight is especially painful in hilly areas. All of this needs to be factored in.
Bottom line: It's feasible and has been done. Operating costs, however, are a big factor, especially compared to what other technologies get you.
edit:
There's a battery version of the Stadler FLIRT [2]. It went 224 km (at about 80 km/h) on one battery charge recently [3]. They will be in operation for DB from the end of 2022 on.
> "Currently, an engine and a diesel tank are much cheaper to buy and especially maintain."
Yes, diesel is cheaper to buy than battery trains. But pretty much everyone, including the main article here, agrees that maintenance costs are much lower with electric trains compared to diesel. I'd expect that (at least once the technology matures) this will also be true of battery electrics.
The point is, if you start with the assumption that diesel is unacceptable and routes must be electrified - for reasons like climate change, toxic air pollution, route reliability, passenger comfort, etc, then there are many routes where battery electrics are going to greatly reduce costs compared to installing overhead wires on every section of track.
> "There's a battery version of the Stadler FLIRT [2]. It went 224 km (at about 80 km/h) on one battery charge recently [3]. They will be in operation for DB from the end of 2022 on."
Yes, I expect we'll see a lot more battery electric units introduced in Europe in the coming years. There are so many routes where they make perfect sense.
The big acceleration points are predictable. Electrify those areas. Then you only need enough batteries to maintain speed and get through the lesser acceleration points.
They would still want enough batteries to accelerate from a dead stop to prevent a 5 minute delay from becoming a 5 hour delay. Energizing the acceleration points would be more about reducing wear on the batteries and prolonging their life.
I don't think constantly raising and lowering the pantograph is a practical solution. Having lines with spotty overhead is very error prone and reliability is a big factor in rail. If we're talking about longer sections that are either electrified or not, things are different. E.g., extending the line for 100 km into an unelectrified part. Trains like the Stadler LINT (see my edit above), could make that happen.
> "I don't think constantly raising and lowering the pantograph is a practical solution."
I don't see why not. There are actually urban routes in London (London Overground) that do this multiple times on every journey! Switching between overhead AC electrification used in North London and the older third-rail DC electrification used in South London.
> "Having lines with spotty overhead is very error prone and reliability is a big factor in rail. "
But that's where batteries come in. Being able to keep moving under your own power if there's a problem with the overhead wires adds a lot of robustness/redundancy.
It depends on how long these sections are. We're probably of the same mind but thinking of different dimensions. If we're talking just for electrified sections of up to 2 km, alternating a dozen times or more, I'm not sure if it's worth it. For longer sections, yes, it's an absolute killer feature.
The cost of raising the panto slightly too early even once is tremendous. You would have to close the track for maintenance. So you need a safety margin of a few seconds. That's several 100 meters. Frequent changeovers would necessitate the system to be fully automated in order to factor out human error.
The changeover on London Overground is usually done while standing at a station (by the driver). According to [1] it's possible to switch while moving. The speed limit is given as 25mph, however.
Yes, you'd certainly make sure everything was automated or at least have a software lock-out to ensure the pantograph is only raised on the correct sections. But having a battery makes this a lot simpler, there'd be no need to stop or coast during the changeover.
It's a special operation, requiring special signage, and awareness of the driver. Here in Hamburg, where the S-Bahn using third-rail got extended to service Stade, is using multi-system rolling stock now. Which traverses the gaps by rolling without power through that gap. I also remember standing on some platform above ground, where you could see them coming over a ramp, with the pantograph up, on a piece of track using third rail. Wondering how that happened, and what would happen if it goes back into a tunnel (with third rail),
like that. Or under a bridge, overpass, all built with low clearance, because: hey, third rail!
Nerd that I am, I walked up to the driver, knocking on the glass and asked: "Moin, do you know your pantgraph is up?" Turned out he didn't, and there wasn't any indication. He got it down by repeatedly switching up/down, and finally it stayed down. Two minutes late!1!! ;-)
Indeed, and neither are electric cars. But battery chemistry, electric motors and inverters, and controller software and hardware have all come a long way in that time. The technology is far more viable and efficient today than it was back then!
They would also allow 95% of a route to be electrified, leaving tunnels and sections under pedestrian and other bridges un-electrified and covered by small capacity batteries. Especially since the train is unlikely to be accelerating during those sections.
When Network Rail in the UK looked at electrifying the Great Western Mainline, the majority of the cost is actually in a very small % of track length which is where the line goes through Victorian tunnels and under old bridges.
If your battery only needs to be sized to maintain speed for sections a few miles long at the most, you can get away with much smaller batteries.
I know they are running trials of what are effectively "tri-mode" trains already which are diesel/battery/OHL.
There's a strong counter-argument that batteries large enough to power entire trains are also expensive to produce and maintain, especially when battery prices shoot up due to lithium scarcity, and that's not even mentioning the massive carbon footprint of producing those batteries. Diesel-electric trains are probably better emissions-wise.
> “battery prices shoot up due to lithium scarcity”
There is really no scarcity of lithium in the world. If more is needed then it’s easy to just tap new supplies because it’s pretty much everywhere if you go looking.
Battery producers are more constrained by things like nickel and cobalt, but there are alternative, cheaper li-ion chemistries (such as LFP) that do without these.
Probably safer in the event of a crash or mechanical failure. I would not like to see what electrical fire would look like from a lithium battery big enough to power s train.
Electric buses in Europe and China usually use LFP chemistry, because it's cheap(er) and much safer in a crash or fire because it's not prone to thermal runaway. Electric passenger trains will likely be using the same, or other stable li-ion chemistries rather than the more volatile nickel-based ones.
I think the point the article is making is that you shouldn't be looking at the cost of electrifying based on how busy the lines are, because the introduction of delays in those sections then propagate into the busier sections causing a ripple effect. Quiet sections with low reliability cause delays in the busy sections. So it's not about how busy that line itself is, it's about the downstream impact.
Capacitors make much more sense than batteries, since you need to discharge at like 2-4C average, probably 20C peaks, and charge at much higher rate. Some buses are already running with this, for instance: https://www.youtube.com/watch?v=nlJlQprrEmg . So far it doesn't look like there is much interest in it, since I'm not aware of any further development since that 2014 line.
Minsk in Belarus runs a network of super-capacitor powered buses. I travelled on one when there in 2020. Just like a normal bus from the passenger's point of view.
"The bus is fitted with a system of super capacitors, which get fully charged within 5-10 minutes while the bus is at an end stop. The bus can travel along the entire distance of the line for up to 15 times per day. On the first day it managed to travel for over 40km on one charge."
Capacitors are power dense but not energy dense, compared to batteries. So for a given size/weight, you get a lot more range with batteries.
There has been some work, a few years back, using super-capacitors (alongside storage batteries) in high-performance electric cars to increase peak power output. But really, battery technology has improved so much that it's made that sort of thing unnecessary. Even your standard Tesla is pulling 8-10C if you really put your foot down.
Hah! Nobody sane is doing that at every stop. For one thing, it shreds your tires.
But for the ones who are doing it repeatedly (at the track), the answer is: very effective battery cooling! Tesla's software in their performance models actually offers a special "track mode" which, amongst other things, applies max cooling.
The reason people look to battery trains is to avoid the infrastructure costs of the overhead lines / third rails, substations etc. to get electricity from trackside to the trains. The costs can be difficult to justify for smaller branch lines. A train that can use overhead lines on the main lines, then switch to battery for a branch is (hopefully) greener / more efficient than say running a diesel under the wires all the way[1].
It is a bit of a case of shuffling costs from infrastructure to rolling stocks, trains with batteries would still cost more than a pure overhead line electric, but I guess it comes from a different budget...
I'm not sure of any real reason why you'd use a battery off-track instead of a connection to the electricity grid except as some sort of backup for a power failure?
[1] Bi-mode diesel / electrics also exist of course, which solve the problem in a slightly different way (insert NOX / CO2 v Lithium mining debate here).
Running a trainset four times along a 100 mile route, or 40 times along a 10 mile route, you're still travelling 400 miles.
And the stop frequency is likely far higher on the urban route. It's the acceleration that costs, and batteries only save you so much there, even with regenerative braking.
Add in heating/cooling loads, air handling, etc., and there's a lot to consider.
About the only place batteries make sense on rail is where you're moving a heavy load one-way down-grade. There are operating quarry and logging operations which effectively work as battery storage. Laden cars (or trucks) run downhill, whilst regenerative braking charges the batteries. Empties return uphill, slowly, on that charge.
Other than a Soylent Green operation from a mountain resort, few commute rail systems have a suitably matching travel pattern.
> "Running a trainset four times along a 100 mile route, or 40 times along a 10 mile route, you're still travelling 400 miles."
We're not talking about 100 miles on battery power here - although there are already battery electric transits that are capable of that and more. For the most part, inter-urban trunk lines in Europe are already electrified.
There are many outer-suburban/regional routes that are, for example, 50 miles along an electrified trunk line then 25 miles on an un-electrified branch line. Currently that means whole route needs to be operated by diesel trains. But a battery train only needs to be capable of doing (eg) 50 miles on battery, then it can recharge when it's running under the wires.
Not just proposals: orders have been placed and trains should be in passenger service this year. Google "FLIRT Akku" which is the name of Stadler's battery-electric train.
Could make sense for short distance, low frequency. I know of a few services on branch lines that have long linger times at stations. They will do a 10 minute journey twice an hour and stay in the station between trips.
For NYC diesel trains, you're talking about stretches like the one from Croton-Harmon to Poughkeepsie (by car, it's about 40 miles). I'm not sure whether this counts as "short."
Batteries will add weight, but you also get rid of diesel engines, traction generator/alternators, fuel tanks, etc which are all pretty heavy too.
On long routes the required size of the batteries might become prohibitive. But we’re talking about outer-suburban branch lines here which are usually pretty short, often only a few miles.
You don't have diesel engine and fuel tank on electric wired train ;) Also, looking at cars, batteries are much heavier than engines and fuel tanks giving same kilometrage. On top of that, fuel tank is heavy only in the beginning of the trip. You can fill up only as much as you need for the trip for better fuel economy. While batteries stay same weight for lifetime.
Batteries make sense for manoeuvre trains working ports, cargo train depots etc. Small batteries and recharge frequently.
> "You don't have diesel engine and fuel tank on electric wired train ;)"
In some cases you do because hybrid (bi-mode) trains exist too! [1] Although as another commenter points out, they are also working on tri-mode diesel/battery trains.
Batteries are still heavier than your typical commuter train diesel engine and tank. The diesel commuter trains are usually diesel hydraulic in Europe, so much less weight than you might expect, e.g. https://en.m.wikipedia.org/wiki/DBAG_Class_612
> Good operations can substitute for station size, and it’s always cheaper to get the system to be more reliable than to build more tracks in city center.
The main problem here is one of politics and human attention span.
It’s easier to drum up a bunch of support towards a construction project. It’s visible, some politicians get credit for it, and you can have a campaign that builds excitement to a crescendo before people move on to something else.
Good operations are invisible because everything operates as the margins and gets small wins. No one gets credit. If they slip just a little then no one notices, so you can borrow a little from operations to find something more visible, and it’s a boil the frog situation. And, finally, people just don’t have the attention span for caring about long-term operations.
I don’t know the solution and I’m not saying the above is good, but any solutions that don’t attempt to address the human and political aspects of public works are going to have trouble.
That was THE reason imo that Andy Byford was so promising as head of NYCTA - he seemed to be someone who would focus on the pragmatic human aspects. Sadly Cuomo made sure Byford had no chance of success.
Maybe I'm misunderstanding, but as far as I've gathered, the main source of unreliability is old and shitty signaling systems. Found it odd that signaling wasn't mentioned.
I don't know about "shitty" but old signalling system isn't a problem if well maintained.
Japan Chuou line mentioned in the article is still on the old ATS-P system and it runs 28 train per hour. Even older ATS-S system in Osaka runs 16 train per hour on the inner track of the Tokaido line.
The NY Subway made a big fuss of CBTC and that it would allow them to run more train more reliably than their current old ATS system, which I find funny looking at various line in Japan.
In fact, CBTC in Japan is mostly planned for rural line with infrequent service so the amount of infrastructure (track circuit, etc) are reduced
> Japan Chuou line mentioned in the article is still on the old ATS-P system and it runs 28 train per hour. Even older ATS-S system in Osaka runs 16 train per hour on the inner track of the Tokaido line.
Meanwhile in Germany: a quarter [1] of the network's signal controller stations still runs on levers, wires and semaphore signals - some of these are from the 19th century and I wish I were joking. [2]
And yet it’s still more reliable and has higher efficiency than whatever Amtrak and the MTA are doing. It’s honestly kind of incredible just how much they have fucked up that even DB is ahead of them in efficiency.
The problem with Amtrak is that the US railway network belongs to freight train operators, and these prioritize their own freight trains over Amtrak despite the law prohibiting that [1] [2].
In Germany, most freight trains run at night because at daytime many tracks are at capacity with passenger trains.
>The problem with Amtrak is that the US railway network belongs to freight train operators
That doesn't generally apply to the Northeast Corridor though which I assume was being referenced by the mention of the MTA. That said, while it's certainly imperfect, Amtrak is actually pretty reasonable for the most part on that route. Certainly the epic delays that are fairly common on Amtrak long distance trains elsewhere are rare on that corridor outside of major infrastructure problems, such as power failures.
The current signals being replaced by CBTC date from the 1930s, and are mechanically operated. They are literally historical relics. https://www.youtube.com/watch?v=Mjx3S3UjmnA
Old signals, as installed, usually have very coarse resolution - enough so to stop trains running as close together as they otherwise would, which limits throughput
On very busy urban systems, one major advantage of CBTC is that the signalling blocks (the section of line that only one train can be in at a time) can be smaller, since the possibility of a driver failing to notice a red signal can be discounted. This increases the number of trains that can be run on the same line.
Does Paris runs or just the CBTC capable of running? You need more than CBTC to run at 85 second headway. Loading/unloading time became huge problem at that interval.
I just checked Line 14 schedule and the most I can find is 34 train per hour, which is approx 105 second headway.
I am not saying that CBTC isn't capable of running higher interval than regular ATS and wayside signal (which is actually a very old technology), but in most case it doesn't matter. Even without moving block of CBTC, a very short static block (say, 100m) and in-cap signally should theoretically have the came capacity as CBTC.
For local trains you need your loading and unloading time to be as fast as possible anyway. People have places to be, waiting around on your trains and platforms is not on the list of things they want to do. It is even worse when they are waiting around not moving at stations they don't want to get off at. 60 seconds is the absolute longest any train door should be open for, and 45 is reasonable. (it can be pushed to 30 or even 15, but that starts to be impossible for the disabled and so not worth it)
Any mechanism which might introduce perturbations into the schedule.
For light-rail transit, shared routes (LRV + automobiles) and traffic signals mean that arrivals are not predictable. Where shared-roadway and dedicated-route segments exist (say, with San Francisco's Muni Metro), hold times must be built in for the dedicated-route travel segments. Train arrivals are highly unpredictable. A level below Muni, on Bart, rush-hour headways through the San Francisco subway are 4 minutes. That's close to the theoretical maximum Alon gives of two minutes.
For conventional heavy rail, boarding / debarking delays, equipment failures, medical emergencies, police activity, level-grade collisions, and other factors, may all lead to service delays.
Signalling matters mostly in that it both doesn't introduce new delays into the system, and that it can compensate for delays originating elsewhere.
Reliability overall, as discussed in the article, is largely based around signalling to manage flows.
Looking at german Stadtbahn systems (trams with a evtensive dedicated RoW & central tunnels, but still a lot of street-running), they archieve headways down to 2 minutes, e.g. in on of the central sections of the Stuttgart Stadtbahn.
That should probably be doable on some american light rail systems, with signal priority and avoiding street running on roads with a lot of traffic congestion.
For urban railways, stations are usually the capacity-limiting factor, so the top speed doesn't really enter into it, and besides
> though I'm not sure it hits those speeds within San Francisco.
due to the shorter station spacing I'd assume that, too.
As for the stations, the limiting factor in turn consists of the sum of dwell time (wheel stop to wheel start) and platform reoccupation time (governed by the signalling system).
The London Bridge Station (UK) rebuild addressed two main reasons for unreliability - the signalling as you said (some of it was almost a century old, and that's not an exaggeration), and the other things was removing points. Points are tracks crossing tracks and the rails mechanically switching/
Arguably fixing that by building a fly-under (a tunnel under the existing tracks) fixed a huge source of problems for that part of the UK rail network, probably more so than the signalling.
The article alludes to the need for through tracks although it is perhaps not immediately obvious that this can also be used for main "terminal" stations. In Germany, very few trains terminate in the Hauptbahnhofs (main stations) but instead will then move out, often to another large station or depot outside of the city centre. This allows high traffic into the main stations without blocking them up while cleaning/turning round.
Another issue about level loading is made but it sounds like this is mainly related to the difference between a "road" at ground level needing steps vs a platform at carriage level. However, even in the UK where everywhere uses platforms, there are still issues with level due to the variation in track height related to ballast working so even "mostly level" can be a problem, particularly for wheelchairs and even done efficiently, loading these can easily take a minute or more.
Thirdly, at least in the UK, it isn't just the build up of subtle delays that can be an issue but small problems creating large delays. The other week, trains were delayed and cancelled for hours (for some reason) because a trespasser was seen on the main line and needed to be found by the police. One relatively small and common occurance affected 1000s of travellers. A single signal problem, a broken rail, failed point etc. This means that you could in theory run e.g. Kings Cross Station with fewer platforms but as soon as there is a problem, people don't want to sit on a train for 2 hours somewhere on the line. Having more platforms gives you that flexibility.
Also, "equal level" can have issues if you're serving multiple trainsets - they may all be at platform height but some are narrower than others, resulting in a platform gap that can be difficult to navigate.
If the trainsets are all identical and the platforms are straight, you can get a very, very narrow gap (easily navigated by a wheelchair alone).
> It is bad enough that Germany is keeping some outer regional rail branches in the exurbs of Berlin and Munich unwired; that New York has not fully electrified is unconscionable.
I think about this a lot every time I see the glacial pace of the ongoing electrification work for Caltrain on the SF peninsula — "studied" since 1992, groundbreaking in 2017, and hopefully in service by 2025 if there are no additional delays: https://en.wikipedia.org/wiki/Electrification_of_Caltrain
Electric motive power was novel in 1895 when the B&O electrified part of the Baltimore Belt Line, but we should be good at it by now! For comparison, the PRR was able to electrify their line from New York City to Washington D.C (225 track-miles, 1928–1935) in roughly the same amount of time it will take Caltrain to build out their 77 track-miles, at which time 90 years will have passed since GG1s were zooming up and down the NEC at 100mph:
I don't know what physical issues exist but there are immense challenges with these projects, not least of which is that you often have to do most or all of the work before you can get any benefit. You then have old trains that need replacing as well, driver training, maintenance training etc. Oh, and you also need to find a workforce of several thousand skilled workers and mostly do the work at night so that you can still run a train service.
In the UK, they recently electrified a lot of the Great Western mainline and decided to go with a cheaper/quieter type of installation method - auger drills iirc - for the pillars (why not), but at certain points, they started finding large granite boulders from the original railway construction which wouldn't be drilled, which meant new plans, new equipment, new training and loads of cap-ex when they were still years away from service. They also found tonnes of electrical system conflicts which meant whole signalling systems being replaced even before they had life-expired.
I love trains and I love modernisation but most of us underestimate the enormous amount of work involved in an upgrade like this.
> The quickest it will ever slow down is with its emergency brakes.
What about derailments? Can a derailment lead to a higher deceleration than the emergency brakes are able to produce? (I don’t know the answer, genuinely curious.)
> So the minimum headway is the reaction time (milliseconds if automatic)
How do the following train detects that the leader train is emergency breaking? Is it based on trackside sensors, or some train-to-train communication?
> Can a derailment lead to a higher deceleration than the emergency brakes are able to produce?
Yes. But a derailment of one carriage of a 5 carriage train will already likely cause the derailment of the other carriages, and in my proposal the carriages of the train behind. Derailments at speed are already typically deadly.
> How do the following train detects that the leader train is emergency breaking?
I would use train 2 train communication to say "I am not emergency braking" every 1 millisecond. If more than a few of these messages aren't received, then you have to assume the train in front is emergency braking, and act accordingly.
A design like that is important because infrastructure failures are frequently coincident - eg. power loss causes signalling and comms failure due to an untested UPS.
In a more detailed design, you would have every train publish a "movement plan" every few milliseconds. The plan contains the current location (with uncertainty bounds), and future locations. The plan also contains alternate future locations that could happen in various emergency or failure situations - and one of those is to apply the emergency brakes.
A system checks each plan against the plans of all other trains, and approves them if they are non-colliding. The 'alternate' plans can overlap only if to do so indicates independant failures must have occurred - for example the brakes of train 1 fail at the same time as the traction motor of train 2 gets stuck on max, and comms fail.
Track infrastructure such as signals and points would also submit plans - so that for example a set of points can prevent any trains passing over while it isn't in a safe locked state.
All trains must then follow one of the plans they have approved. If comms fail and they can't get a new plan approved, they must still follow an approved plan - which means they are guaranteed not to collide with any other train.
All plans would be signed with timestamps (to prevent someone maliciously changing them to cause an accident), and would normally be delivered by dedicated rail infrastructure, but could be delivered by 3g/4g/wifi in an emergency. Manual plans could also be entered and distributed, for example via USB stick, for maintenance - such plans would say "I have access to this huge segment of track, and nobody else may enter".
Plan A is used if all systems remain operational. It says that the train will continue to move at 60 +- 5 mph for the next 10 seconds, then slow gradually to 45 mph. The plan could consist of a section of "reserved" track for each second in the future, and the train is sticking to the plan if it remains within the reservation.
Plan B is used if service brake failure happens. It says the train will apply its emergency brake and slow to zero at a given rate.
Plan C is used if both the service brake and the emergency brake fail. It says the train will continue at 60 mph, and may speed up or slow down a little, before finally guaranteeing to stop within 30 miles.
All the plans would be written automatically by software, and each would be 'conservative' - ie. if there are any uncertainties, the plan must include them. Less advanced trains would have wider bigger plans, and must therefore keep wider separation distances.
Plan C is obviously very unlikely - and is therefore allowed to overlap with similar plans B and C of other trains - the reasoning being it is very unlikely for the service and emergency brakes of this train to fail, at the same exact time as the emergency and service brakes of another train.
> I would use train 2 train communication to say "I am not emergency braking" every 1 millisecond.
I like that. How would that work when there is no lead train? In other words how would a train know that it is not hearing from the train before it because there is no train, vs not hearing because the lead train is damaged/emergency braking.
See my other response in this thread - an actual deployment would require every train on the tracks to automatically submit a 'plan' for it's location, and if it cannot submit a plan, then it isn't allowed on the tracks. Plans last until replaced by a new plan, so a train will always have at least a place on the track it can be.
It doesn't solve the problem of 'there is a tree on the tracks' or 'someone put a homemade train on the tracks' - you need a different solution to that.
The second train in this example could be heavier, requiring more distance to stop, but that can probably accounted for by adjusting minimum following distance based total mass of the train.
But when if the first train is experiencing something it shouldn't, like a big accident/derailment, the landscape could be a way better brake than the train's own brakes. When does the second train start its emergency braking?
Forgive me if this is a dumb question, but why would a heavier train take longer to stop? Assuming the brakes are strong enough to lock the wheels, shouldn't the increase in braking force required be exactly matched by the increased friction from the additional weight?
I understand that that's the conventional wisdom and that I am almost certainly wrong here.
I thought train emergency brakes were capable of fully locking the wheels (to the point where they get significantly damaged/deformed after an emergency stop), so why would the braking pad heat dispersion matter?
Nagoya station in Japan has a Track 0 that runs to Kariya. Tracks in most of the world's train stations are numbered starting from Track 1. Has anyone seen another Track 0? Where?
Thanks for the link. All these replies are great, but the video from Geoff and Matt is an epitome of modern British communication. They really did stop at nothing!
Tølløse station in Denmark. Which is for the local train, as tracks 1 and 2 are for the regional trains. I assume the local train track was added later, and was next to track 1, so rather than naming it track 3 or 11 (as they sometimes do to highlight an "in between track"), they went for track 0.
There is a bunch of pictures here[0], and you can spot in some of them where it says "Spor 0". But I assume "track 0" isn't _that_ uncommon after all.
Which leads me to wonder whether a "track -1" exists anywhere.
I don't know about track, but the Northern river terminal in Moscow has a negative pier. The main piers are numbered 1 to 17, but there are two discount piers outside the main area, which are numbered -1 and 0.
> Tracks in most of the world's train stations are numbered starting from Track 1
In France sometimes tracks aren't numbered but lettered. Sometimes they're both - e.g. the same train station can have tracks A, B, 1, 2, 3, 4 ( e.g. Avignon TGV, where 1 and 4 are for stopping high speed trains, 2,3 are for passing through high speed trains, and A(maybe also B, don't remember) is a separate track terminating at that station for the commuter rail link to Avignon's main railway station). Sometimes there are weird likes like only two tracks, one is 1bis, the other A. Or 1 and 1bis. Usually it's related to the historical evolution of the train station ( e.g. in the 1/1bis case it was a single track station which was expanded).
At Nijmegen central station in the Netherlands , track 1 is preceded by 35, even though the station only has a handful of tracks. I’m not making that up.
In Sittard, The Netherlands track 1 was preceeded by track 20 (kopspoor with buffer stop). Track 20 being the train to Kerkrade and the other tracks being for other trains (mainly but not solely ICs).
One such (obvious) exception is OK with me. Its annoying when every train station has random track numbers though.
It’s not random. The numbers aren’t only used for passengers, but also for train personnel, and because of that not only cover the train station, but also tracks in the neighborhood.
Tracks are numbered logically or at least start out that way, but in some stations, many tracks don’t border a passenger platform (which freight trains don’t need)
I don’t know what rules they use for renumbering tracks, but they must weigh the cost of redocumenting things and retraining personnel (likely the trickier part. Some of them may have known the track numbers for decades, so reassigning number to make them easier to learn can be risky. From a safety viewpoint, moving from 1,2,3… to A,B,C… may be a better option) against the training of future employees.
Interestingly, tracks that were turned over from ProRail to HTM (for RandstadRail operations) were never renumbered. This means that Zoetermeer station has 2 tracks numbered 3 and 4, and Laan v. NOI station has 4 tracks numbered 3-6.
Lidcombe station in NSW Australia has a platform 0. It happened when a new platform for an Olympic Park shuttle was added on the “up” side of the station (which gets numbered starting from 1) and for one reason or another the existing platforms were not renumbered.
For once, you‘d need to synchronize the actual track renumbering with the (yearly?) train schedule update as track numbers usually are part of the schedule.
Bologna Centrale has numbered tracks 1,2,3.. and different number series for both "West" and "East" tracks, with "1 West" being the leftmost of a number of tracks side-by-side.
When I was there the first time, of course we waited at the wrong platform and missed our train.
Haymarket Station in Edinburgh has a platform 0 that was added to the side of platform 1 during an expansion. It's the only platform/track that terminates at the station, all the others go through the tunnel to Waverley Station.
Shinkansen isn't really designed for short dwell time. And since it's Tokyo terminal, the train also need to be cleaned. For commuter train, 15 train per hour is easily doable with 2 tracks.
2- one for the train to pass through and another for wagons with arriving to decelerate and another wagon with departing to accelerate and catch up with the passing through train..
I just want to say that trains are I've of the great amazing features of civilization. I look at them like Plato might look at electricity. It is a wonder.
For example:
* Penn Station Dwells are not limited by passenger movement at doors. It’s not a metro, go and watch trains being loaded.
* Escalators constrain the rate at which people can actually reach the platforms
* These escalators need to switch direction when a train arrives, meaning multiple arrivals and departures are not simultaneously supported.
* The tracks do not support all operators and the station design divides Amtrak from LIRR
* Some tracks do not support full train lengths
* Not all tracks in the station are through tracks
* Trains are terminating at NYP which requires a longer process of ensuring no passengers are still on on board before leaving for the yard.
* It’s not only the north and east river tunnels, there is an entrance from the Empire line for Amtrak and separate paths to the WSSY for LIRR, not to mention train storage within Penn in the ACDE yards.
* Engineers walk the entire length of the train when a trip reverses direction. This takes a long time. The OP mentions through-running solution, but doesn’t point out the problems with it, namely incompatible equipment and giving an entirely different agency access to your multimillion dollar trains.
I could go on and on.