What Washington Metro Should Build
I’ve been thinking intermittently about how to relieve the capacity crunch on the Washington Metro. The worst peak crowding is on the Orange Line heading eastbound from Arlington to Downtown Washington, and this led to proposals to build a parallel tunnel for the Blue Line. Already a year ago, I had an alternative proposal, borrowing liberally from the ideas of alert reader Devin Bunten, who proposed a separate Yellow Line tunnel instead. Matt Yglesias’s last post about it, using my ideas, made this a bigger topic of discussion, and I’d like to explain my reasoning here.
Here is the map of what I think Metro needs to do:
Existing stations have gray fill, new ones have white fill. The Yellow Line gets its own route to Union Station, either parallel to the Orange Line and then north via the Capitol (which is easier to build) or parallel to the Green Line (which passes closer to the CBD), and then takes over the route to Glenmont. The rump Red Line then gets a tunnel under H Street, hosting the busiest bus in the city, and then takes over the current Blue Line to Largo, with an infill station in Mayfair for a transfer to the Orange Line and another at Minnesota Avenue for bus connections.
The Blue Line no longer presents a reverse-branch. It is reduced to a shuttle between the Pentagon and Rosslyn. Matt mistakenly claims that reducing the Blue Line to a shuttle is cost-free; in fact, it would need dedicated tracks at Rosslyn (if only a single track, based on projected frequency), an expensive retrofit that has also been discussed as part of the separate Blue Line tunnel project. At the Pentagon, initially shared tracks would be okay, since the Yellow Line is still a branch combined with the Green Line today; but the separate Yellow Line tracks would then force dedicated turnback tracks for the Blue Line at the Pentagon as well. Frequency should be high all day, and at times of low frequency (worse than about a train every 6 minutes), the lines in Virginia should be scheduled to permit fast transfers between both the Yellow and Orange Lines and the Blue Line.
The reverse branch today limits train frequency at the peak, because delays on one line propagate to the others. Peak capacity on Metro today is 26 trains per hour. I don’t know of anywhere with reverse-branching and much higher capacity: the London Underground lines that reverse-branch, such as the Northern line, have similar peak traffic, whereas ones that only conventionally branch (Central) or don’t branch at all (Victoria) are capable of 35-36 peak trains per hour. This means that my (and Devin’s, and Matt’s) proposed system allows more capacity even in the tunnel from Rosslyn to Foggy Bottom, which gets no additional connections the way 14th Street Bridge gets to feed a new Yellow Line trunk.
The big drawback of the plan is that the job center of Washington is Farragut, well to the west of the Yellow and Green Lines. WMATA makes origin-and-destination data publicly available, broken down by period. In the morning peak, the top destination station for each of the shared Blue and Yellow Line stations in Virginia is either the Pentagon or Farragut; L’Enfant Plaza is also high, and some stations have strong links to Gallery Place-Chinatown. Metro Center is actually faster to reach by Yellow + Red Line than by taking the Blue Line the long way, but Farragut is not, especially when one factors in transfer time at Gallery Place. The saving grace is that eliminating reverse-branching, turning Metro into four core lines of which no two share tracks, allows running trains more frequently and reliably, so travel time including wait time may not increase much, if at all.
This is why I am proposing the second alternative for the route between L’Enfant Plaza and Union Station. Devin proposed roughly following the legacy rail line. In the 1970s, it would have been better for the region to electrify commuter rail and add infill stops and just run trains on the route, and today a parallel route is appealing; Matt even proposed using the actual rail tunnel, but, even handwaving FRA regulations, that would introduce schedule dependency with intercity trains, making both kinds of trains less reliable. This route, the southeastern option among the two depicted in dashed lines, is easier to build, in that there are multiple possible streets to dig under, including C and E Streets, and giant parking lots and parks where the tracks would turn north toward the Capitol and Union Station. It also offers members of Congress and their staffers a train right to the officeUnfortunately, it forces Farragut-bound riders to transfer to the Orange Line at L’Enfant Plaza, slowing them down even further.
The second alternative means the Yellow Line stays roughly where it is. Four-tracking the shared Yellow and Green Line trunk under 7th Street is possible, but likely expensive. Tunneling under 8th Street is cheaper, but still requires passing under the Smithsonian Art Museum and tunneling under private property (namely, a church) to turn toward H Street. Tunneling under 6th Street instead is much easier, but this is farther from 7th Street than 8th Street is, and is also on the wrong side for walking to Metro Center and points west; the turn to H Street also requires tunneling under a bigger building. By default, the best route within this alternative is most likely 8th Street, then.
A variant on this second alternative would keep the Red Line as is, and connect the Yellow Line to the subway under H Street and to Largo. This is easier to construct than what I depict on my map: the Yellow Line would just go under H Street, with a Union Station stop under the track and new access points from the tracks to a concourse at H Street. This would avoid constructing the turns from the Red Line to H Street next to active track. Unfortunately, the resulting service map would look like a mess, with a U-shaped Red Line and an L-shaped Yellow Line. People travel north-south and east-west, not north-north or south-east.
Under either alternative, H Street would provide subway service to most of the remaining rapid transit-deprived parts of the District west of the Anacostia River. Some remaining areas near the Penn and Camden Lines could benefit from infill on commuter rail, and do not need Metro service. The big gaps in coverage in the District would be east of the river, and Georgetown.
Georgetown is the main impetus for the Blue Line separation idea; unfortunately, there’s no real service need to the east, along K Street, so the separate Blue Line tunnel would be redundant. In the 1970s it would have been prudent to build a Georgetown station between Foggy Bottom and Rosslyn, but this wasn’t done, and fixing it now is too much money for too little extra ridership; Bostonian readers may notice that a similar situation arises at the Seaport and BCEC, which should be on the Red Line if it were built from scratch today, but are unserved since the Red Line did not go there in the 1900s and 10s, and attempting to fix it by giving them their own subway line is a waste of money.
East of the river, the Minnesota Avenue corridor would make a nice circumferential rapid bus. But there are no strong radial routes to be built through it; the strongest bus corridor, Pennsylvania Avenue, serves a small node at the intersection with Minnesota and thereafter peters out into low-frequency branches.
This means that if the Yellow Line separation I’m proposing is built, all parts of the District that could reasonably be served by Metro will be. If this happens, Metro will have trunk lines with frequent service, two not branching at all and two having two branches on one side each; with passengers from Alexandria riding the Yellow Line, the Orange crush will end. The main issue for Metro will then be encouraging TOD to promote more ridership, and upgrading systems incrementally to allow each trunk line to carry more trains, going from 26 peak trains per hour to 30 and thence 36. Washington could have a solid rapid transit skeleton, which it doesn’t today, and then work on shaping its systems and urban layout to maximize its use.
Pedestrian Observations 
H Street NE is finally reaping some benefits from more than a decade of rail construction, so folks near there would not likely look with favor on a proposal to dig it all up again; rather than advocate to build their stub line into a viable crosstown route this map says “let’s start fresh.”
If I understand the maps properly, a major system bottleneck–Potomac River crossings–is static in this plan. Both Potomac tunnels reach capacity at peak times now, so I fear Virginia riders could see this as not as a step forward, but as a rearrangement of their predicament.
Their stub line is a mixed-traffic streetcar. Not building it would leave an awkward gap between the rump Yellow Line at Union Station and the Largo branch.
The Potomac River crossings aren’t actually static. 14th Street Bridge is currently a branch, with only half service, since the other half of the trunk’s capacity has to go to Anacostia. This change makes it a full trunk line, moving the branch point to Alexandria. It also simplifies schedules, which would allow squeezing more trains per trunk – hence my comment about London Underground peak capacity.
H St. is junk; it has net negative transportation value to the entire transportation network. The streetcar has no dedicated lane, and degrades X2 service, which has far higher ridership. I think H St. would very much appreciate absolute rights of way as well as a one-seat ride to Gallery Place, Metro Center, and Farragut and an easier 2-seat ride to L’Enfant.
Good analysis Alon.
My view as a Metro-dependent DC resident:
Electronics before concrete. I am tired of seeing the “Rosslyn can only handle 26 TPH” trope repeated as if it justifies a separated blue line – or any other service changes for that matter. Currently, 8 min headways on Blue/Orange/Silver (at rush!) mean ~23 TPH. Let’s get some serious moving block signaling (36 TPH) a la the London Underground in place, and a restoration of ATO (reducing delays) before moving any earth.
Serving H Street is certainly worthwhile, and I applaud thinking beyond the “blue line separation” concept. But I don’t think your southeastern option offers more meaningful metro service to the capitol than the existing Capitol South stop, which is right next to most of the US House of Reps office buildings, the Republican National Committe HQ, and reasonably close to the capitol. The proposed stop east of the capitol should be nudged as far east as possible given the constraints of the track/curve profile so it can serve the Capitol Hill neighborhoods rather than the already served political element of the capitol.
The maximum frequency on a London Underground line with reverse-branching is 28 peak tph on the Northern line. There’s a hard limit to what electronics can do for you when the trains run on a highly-branched system, with uncertainties in scheduling coming from high passenger crowding. You need to leave extra capacity at the merge points in case a train gets delayed 30 seconds.
Also, while 26 tph isn’t great, I think it’s alright considering the complexity of the network around Rosslyn and in Virginia in general. I mean the Paris RER A, which is also quite a complex line but quite as complex as around Rosslyn, timetable just dropped their peak hour frequencies from 30 tph down to 25 tph in the morning and 24 tph in the evening rush (presumably because they couldn’t run double-decker trains reliability enough) so 26 tph isn’t bad at all.
If you’re going to run 36 tph, you’re almost certainly going to need new trains with more doors as well.
*not quite as complex
Are you sure the RER A went back down to 24 tph permanently? They run 30 peak, 24 reverse-peak normally, but then cut frequency around New Year’s because it’s a holiday period.
I’m pretty sure this is permanent. It doesn’t say that it’s a holiday timetable.
https://rera-leblog.fr/nouveaux-horaires-rer-a-2018/
Yes, the timetable changes appear to be permanent. But it is worth saying why. It is due to the relentless increase in numbers of passengers–this line has always outrun the planners expectations. This single line (RER-A) carries 300m pax p.a. which is a fair bit more than the entire Washington DC Metro system (seems to be various data out there but the highest appears to be 260m). So Washington won’t have to worry about the same issue for another century or so …
Most of the time these amazing trains (duplex, the 112m long, 5-carriage version carries 1,725 pax; can be run as 10-car set) can handle the 2 min headway but with the press of the crowds at peak times, despite the platform-aligned doors, any pax incident can cause a cascading delay/failure. And as I think Alon would point out, this is compounded by the branching.
To increase pax movement they decreased the number of doors from 4 to 3 per side.
You don’t need every train to be delayed; one rush hour delay propagates back to the rest of the line because of the tight headways. I don’t think the bilevel configuration is necessarily great – they’re talking about allowing 105-second peak dwells now. Paris has a unique aversion to bench seating, which increases standing capacity, so the idea of a single-level RER train with ample standing space (which the RER B doesn’t have) is foreign.
“To increase pax movement they decreased the number of doors from 4 to 3 per side.”
Can you explain this?
They went from single-level trains with 4 door pairs per car to bilevels with 3 wide door pairs.
>they’re talking about allowing 105-second peak dwells now.
At that rate, might as well give up, buy new trains and cascade the existing two-level ones to Transillien. Then they can bump the tph back up to 30 again. The new timetable will mean more people will need to stand anyway so might as well go back to single deck trains with 4 doors (and slightly wider doors than the old ones) and allow the higher frequencies to make up for the loss of seats per train while restoring the old total capacity in the form of standing room.
The current 950 seat RER A trains at 24 tph do 22800 seats per hour, a decent single deck will probably have 700 seats so at 30 tph, it will mean 21000 seats per hour, not too bad of a reduction.
Alon wrote:
These are suburban commuter trains. RER-A spans about 120km east-to-west across Paris. Forcing people to stand for these journeys is unacceptable and would produce a revolt; in turn it would reduce the number of commuters willing to put themselves thru this every day of their lives. I think the compromise that the French trains (both RER and Metro) is a good one. It is the same reason for the design of those gallery (duplex) commuter trains in the US. Would you insist on bench seating on the LIRR?
I am reminded of the London long-distance commuter trains from the likes of commuter towns of Brighton or Oxford (both places I have lived). A season ticket for London-based workers is subsidised but still costs a fortune, yet with the deterioration of the train system (along with everything else in the UK) more and more people were being forced to stand the whole ride. Only in the UK would the “mustn’t grumble” sheep put up with it; however it got so bad and so ridiculous that there were highly publicised revolts. Unbelievably (well, not if you know the UK) they still had First Class carriages (whose costs were even more eye-watering, you know, to keep the riff-raff away–only for the banksters and others who are not paying for their tickets). A few people with standard season tickets began occupying the first-class seats if there were none remaining in standard class, and BR kept trying to turf them out and some of these went all the way to the courts. I think BR had to rewrite their terms-of-service so as to specifically warn that there was zero obligation on them to provide a seat (among other things). I remember there was discussion about duplex trains but the cost of modifying all the tunnels was unthinkable (for the Brits). When the trains were privatised, it all got worse (Google “Southern” and read of the woe to this day). Many of the journeys on RER-A (and all of the 5 RER lines) will be longer than Oxford or Brighton to London.
So Alon, you really have to get over this criticism and the planner’s wanton desire to treat paying passengers as if they are not entitled to anything other than what you the planner deign (and will never experience). The fact that RER-A handles more pax than the entire DC Metro (and many others) removes any technical argument. It can be done.
I have read recently about the factors involved in cross-seating versus bench-seating (but naturally can’t find it now): I think I even mentioned the same thing in an earlier post here: pax have a much higher degree of comfort and feeling of security and privacy–even on crowded trains–with the former. Part of the train is divided functionally into “compartments”. This extends to even those who are standing in the aisles for the obvious reason that you are not going to be stampeded as a crowd in the open foyer area next to doors surge out and in; and trouble-makers (pickpockets, gropers) are less likely to be in these areas (they try to keep a quick exit at all times). And of course it is those pax who begin their journeys further out who get the seats or the more protected standing positions which is fair; equally if you only have a shorter ride you don’t object to standing. I have never liked the long bench seating in most city metro systems, particularly London’s.
Untangled 2017/12/31 – 22:31
? There is no evidence this would achieve what you think it would. Almost certainly there is no actual ideal solution to this problem which is as simple as a single line reaching its effective capacity. The only way to significantly improve egress & access times is to use Spanish platforms with unloading one side and loading the other side of the train. Though I suppose this could be done on the relatively few city stations which have the biggest problems, it would still be extremely expensive, probably impossible to maintain service during construction and may yet not be physically possible (eg. at Chatelet there wouldn’t be room).
But then, we’ve had this discussion before and you are perfectly happy to have more than half of all pax standing on those 50km runs from Sydney’s exurbs … yeah, that will tempt them out of their cars (the $50 toll roads will do that I suppose). Needless to say that Sydney takes its cues from London and the UK which has such a wonderful transit rail network and such a contented and happy bunch of commuters … not.
The only “solution” is what Paris is doing: Paris Grand Express, ie. building more circumferential Metro and RER lines, plus tramways, to take the pressure off trans-Paris journeys on those crowded lines.
No evidence? Unfortunately for you, there is a huge number of studies which show that the number of doors has an influence on dwell times and more doors do help in reducing dwell time since there are more points to leave and board the train (or bus or tram or ferry) at the same time. And no I’m not going to link to any because it can be easily searched. As you point out, the Spanish solution is one strategy to reduce dwell time but it is very expensive to rebuild platforms and it is not the only solution to reducing dwell time. It is far more cost effective to have more doors on the train instead.
Why can the MTR’s Tsuen Wan Line, which can have 75,000 per hour, has reliably run trains every 2 minutes yet RER A, with 55,000 per hour, can’t get 30 tph? The Tsuen Wan Line trains have more doors and it’s not double decker. If you want a regional service at metro intensity, then it then trains must also be metro-like as possible, there’s no way around it and the current trains don’t really fit that. And if still think that more doors have no effect, why do you think that Perth in Australia is ordering new trains with 4 doors instead of 2 doors per car? And Perth is hardly an intensive operation yet they want 4 doors instead of 2 doors (out to Mandurah as well). Why is BART adding more doors in its new trains compared to the old ones? If Washington is going to run more trains per hour, particularly at 30+, they’ll need more doors per carriage as well.
I know you addressed this to Alon but I would I like to add that the reduction in the number of seats between bench seating and 2+2 seating isn’t very dramatic. It’s usually around a 10% decrease in seating, bench seating generally takes up more room than you would think. Even the Washington Metro Wikipedia rolling stock page tells you that. Personally, I like transverse seating so I can look out the window but longitudinal isn’t a huge reduction in comfort.
7000-series seating:
A-Car: 64 (transverse), 58 (longitudinal) (And it really is all longitudinal in the drawings)
B-Car: 68 (transverse), 64 (longitudinal)
I would agree that Grand Paris Express Metro is the only long-term solution to relieve the RER though. The new express trams look interesting too. No one is standing at 50km out in Sydney, they will get a seat since it’s the start of the line at that distance. And the reduction in the number of seats isn’t huge it’s 15000 per hour for metro and 18000 per hour for the suburban line.
Untangled 2018/01/01 – 01:25
As usual you are not comparing like with like.
First, you are comparing Western with Asian crowds. For multiple reasons, nowhere has the west attempted to run their operations the way the Japanese, Chinese and Koreans do. Heck, even Londoners won’t board/exit their trains as efficiently as Parisians!
Second, Tsuen Wan Line is a metro that is 16 km long and terminates at Central. As with the DC Metro, RER-A is almost as long as all of the HK Metro. I don’t know where you get your figure of 75,000/h from but I suspect it is theoretical as the 1m pax per day doesn’t really seem to work out (I may be wrong but it seems to work out at approx. 25k per hour each direction but that is averaged so doubtless much higher at peak). Also I am pretty confident in thinking that even though it is only 16km the HK line has more distributed ridership while the RER is collecting people from all over its 108km and 46 stations but proportionately more of its riders are headed for the 5 stations in the city (plus La Defense) creating those wonderful crushes Alon so (secretly) loves at Chatelet …
Third, you again ignore the seating issue. Not an issue for the HK 16km metro compared to the 108km RER.
Fourth, you can talk about all these other western designs (DC, BART, Sydney, Perth!) but it is uncontestable that RER-A carries more pax than any other line in the western world (and I suspect a higher crush effect at the central city stations). In fact if those others you talk about are having trouble coping with their existing load such that they are reconfiguring their layout further, then why would you consider their design decisions are better than a system that actually outperforms them (while seating a lot more pax over much greater distances)? The only convincing proof will be in the pudding, ie. when/if those systems have to cope with the same pax load (especially at a few hot-spot stations like RER-A.)
Fifth, your seat calculations don’t convince me at all. Not least because you have not compared a duplex RER car*. As I see it, the transverse seating can hold 8 seats for every 6 in the longitudinal arrangement (ie. 33% more); but actually that is an underestimate because with 3 wide doors you can fit in more than with 4 doors. (If the 4 doors were as wide as the 3 doors you would have even less seating–of any kind.) In addition the wider foyers/vestibules of the 3-door arrangement hold another 24 strapontin (foldup) seats per car. (I have only seen these seats on a few of the world’s metros but they won’t be in those narrower vestibules because obviously they would inhibit egress. Note that these seats are very useful because they allow people to be seated from the outer-most stations even if people have to end their journey standing as it gets more crowded approaching the city. And of course can be used all the time off-peak.)
Sincerely, I reckon your argument falls down on any one of these single issues. However I will concede that some of this is a design/function decision, ie. a judgment rather than purely scientific/technical. Even you seem to agree with me on this, and in reality I believe anyone given the different experiences would make the same choice. I read in the Sato & Essig report on their examination of the Tokyo system (in 1970 I think; possibly their visit was earlier) they noted among the differences was that the Japanese trains (most like the RER which the French were trying to finalise designs on at that time) used longitudinal seating. Obviously they made an explicit consideration and rejection of that format. IMO, they were correct. It put passengers first, perhaps at the sacrifice of (marginal) performance (though this remains unproven despite everything you write). I see that on the new cars on the now-automated M1 they have adopted 2+1 seating that presumably provides more standing capacity; note that this line is more valid to compare to the Tsuen Wan Line in that it is 16.6 km with 25 stations, handles 214m pax p.a. and can have headways of 85 seconds. Note that M1 carries almost as much as the total DC Metro. (And Untangled, RER-A alone carries more pax than the entire Australian train system.) I may be notoriously Francophile and biased (but being a research scientist of course I don’t believe that; I reckon I am assessing the evidence) but I’d say to all these others: sure, adopt your various different design options but my advice, don’t teach grandma to suck eggs. I wonder what you would predict about what the American users of the NYC subway and DC Metro and BART, or Chicago El, would think after using the Paris Metro (& RER)? IMO, no contest. (I’ve used all of these but not as a commuter, except Paris.)
* Your figures and diagrams don’t make sense to me; neither show transverse seating and neither is duplex and I can’t see how you derive the figures you cite; I can’t find the data and the French websites won’t load … but I see that the original (1977, non-duplex) cars of RER-A, the MS-61 had 292 seats (200 fixed + 92 strapontin; I can’t really visualise this arrangement) but I can’t find the data for the modern MI-09 duplex cars that currently service the line. Nor the number that can be standing. There must be some trainspotter website out there with this kind of thing?
1. The fact that the Metro has mostly 2+2 (with some 2+1) seating strengthens my argument that RATP is just used to cross seating – after all, the Metro lines are short. M1 has more room because of 2+1 seating, yes, but some of the older trainsets with 2+2 get very crowded at the vestibules because there’s no standing space, for example M11.
2. The RER A is 108.5 km long and not 120, and this is counting all the branches. Boissy is 24 km and 32 minutes from Les Halles and 33 km and 42 minutes from La Defense, and Torcy is 26 km and 33 minutes from Les Halles (31 if you catch an express train). Marne-la-Vallee is way east, but that’s not a station where commuters board in the morning. The E train in New York is 33 minutes from Jamaica Center to Lex/53rd, and 41 minutes to Times Square – and unlike the RER A, the inbound peak E fills in immediately because the first station is already the second busiest in Queens, with the same ridership as the five easternmost stations on the Boissy branch combined. Epping is 40 minutes from Bank, and West Ruislip is 45 minutes.
3. Putting passengers first means not making passengers stand in overcrowded vestibules because the corridors between the seats are too narrow for people to squeeze through. The problem is that even when there is standing space, it’s difficult for passengers on e.g. the RER B to move between vestibules to get to the less crowded one, since the passageways are obstructed.
Alon 2018/01/01 – 08:39
1. But so what? In all mega-cities, Metro cars gets very crowded. Including those of the type you prefer. If I am standing I don’t like the long benches because your only means of stability is strap-hanging, so you end up getting buffeted around and into other pax. In the Paris system you can usually hang on to the rail as part of the back of the seats (and it has a loop for that purpose).
2. I wrote that RER-A is 108km. As to your other points, not sure of your point. Untangled wasn’t using any of those other examples when he was trying to make his point. Yes, there are some very long Metro lines and they are usually a bore to use: I have often taken the A from JFK all the way to Washington Heights (Columbia Med School) and it takes forever, probably double the time for same distance by RER (but then that is why Paris decided to create the RER).
Just making more of them stand in all positions in the car, for longer. Besides, crowded is crowded. Even in the longitudinal seating arrangement, it easily gets crowded enough to impede you from (wanting to) work your way from one vestibule to another. Incidentally in the old-style Metro cars the distance between vestibules is only two “compartments” (ie. 4 rows of seats) and so you can wade your way through (because standing pax can move into the “compartment” space to allow you to pass; no one will be happy with you but …). And >200m pax use M1 it clearly works. I don’t know why there are objections to something that so manifestly works, and works quite well. I mean all this nit-picking. Have you tried a crowded bus lately? In that case it is a serious bore to work your way down the aisle to either the rear exit or the front exit, but buses don’t go for longitudinal seating.
As I have said, I don’t believe the other arrangement has a particularly noticeable advantage or any special efficiency premium (even if theoretical considerations might suggest it), while most pax would prefer the transverse seating if you gave them the choice. In fact Untangled prefers it! How about you, as a pax not a planner, what is your pref?
The 75,000 was from the MTR website. The reason I compared Tseun Wan Line with RER A was not for distance or comfort, but because you earlier said that the reason RER A frequency droped to 24-25 tph was because of too many passengers. I’m using it to point out the justification that it was not possible to run 30 tph for RER A as junk. Clearly, it is possible when you have more doors and single-decker trains (which can still seat a lot of people).
The RER A doors are nice and wide which does go some way to making up for the loss of a door but some of the congestion issue remains, I’m hearing that 50 second RER A dwell times are normal, that’s on par with Sydney double-decker dwell times, not good, especially when Paris has more doors. So I suspect it has something to do with stairs (which I’ve said before) and maybe something else as well.
Also, the trains around Sydney now carries almost 390 million passengers per year so that’s busier than the RER A by itself (although still half of RER overall). Sydney’s busiest line, the T1 line does almost 150 million passengers per year (half of RER A), at 20 peak tph (plus a few more which are scheduled to finish at Central) and trains that are 1/3 shorter.
If it’s in a tunnel or a route is mostly in a tunnel, it doesn’t make much of a difference to me, especially since the seating sacrifice isn’t much. Most Paris Metro lines are mostly in a tunnel and I think longitudinal seating is better suited short and/or very intensive operations like the Paris Metro or most of the MTR.
Untangled 2018/01/01 – 19:31
You’re doing it again. Waving your hands around and pretending that wins an argument. So you read that figure on the MTR website; quite possibly the exact same thing I read but which was given as a design target (maximum) and no indication it has been broached (yet). And why you just keep waving your hands around. Naturally you ignore all the other issues I raised. If you had a poll of readers of Alon’s site, how many would believe you could just transpose conditions and designs from an Asian MTR to an American or European one? Despite everything you so confidently write/repeat, you have not shown that any Western train system outperforms either the RER or the Paris Metro. (Without even raising issues of comfort etc …)
Besides which, it comes down to the subjective issue of whether to strive to seat as many pax as feasible–balancing all the various factors–on trains that cover long distances (relative to short-hop Metros like the ones you keep citing. In fact I think it is clear you actually prefer transverse seating for such journeys, but now are switching your story so as to try to win an argument. You didn’t win it!
And by the way, clearly RER-A can and has achieved 60,000 pax ph and 30tph, but as pax load increases relentlessly the occasional impact of passenger events has caused too much disruption to the line’s service, obviously especially at peak hour. I don’t know how often that happens–possibly only once a week, but presumably too much for what may not be worth the “gain”. Your claims about the duplex train “failure” is nothing more than hyperbole.
……………………..
I note that BART and DC-Metro retain 2+2 seating and was originally designed to fully seat all pax. Wouldn’t you and Alon agree that RER is more like BART than it is like Metro (especially HK’s Tseun Wan Line or any others you keep citing)? A curious observation in Wiki on BART refurb: “At the time of their construction, the C2 cars also featured flip-up seats which could be folded to accommodate wheelchair users; these seats were later removed during refurbishment.” These are the identical strapontin seats in Paris–as the C1 & C2 cars were built/designed by Alstom. Apparently their use was just too much to be grasped by Americans! I note that my city’s modern buses have these seats at the front of the bus so that wheelchairs and mothers with prams can use them–and they do all the time.
If the RER A can achieve 30 tph, why can’t they do it now when it’s most needed? When a busy line needs more trains but can’t achieve it but somehow others can (even if it’s theoretical), there’s clearly something wrong. Somehow you don’t seem incapable of understanding that. I’m not saying that RER should be spartan in comfort like MTR (you might have grasped that on my point about 700 seats on single-decker trains or my point on 21000 seats per hour), that was not the point I was trying to make, but there are key features of heavily used trains that run at 30 tph+ and these features enable them to handle passenger exchange better.
If a line is carrying loads more comparable to Asian cities than Western cities then I feel that it is appropriate to compare it to Asian cities in some aspects since it’s in a different league. If you’re comparing it to Western cities, then it’ll never improve.
I’ve made this clear but it depends on the situation but generally, I do prefer transverse seating, but there are situations where longitudinal seating is wouldn’t be too bad (Paris Metro would be one of then, but not RER or BART, long-ish lines that are fully or mostly underground will be fine with longitudinal since you can’t really look out). I did point out in the first post that the loss of seats isn’t dramatic with longitudinal seating when I mentioned that I preferred transverse seating because I can look out, so no I didn’t switch my anything.
Personally I strongly prefer longitudinal seating, because I hate backward-facing seating, and in a longitudinal seat I can tilt my head facing in the direction of motion. In South Korea the preference against backward-facing seats is enshrined in the pricing system: the KTX gives 5% discounts for backward-facing seats.
Alon Levy 2018/01/02 – 05:00
Well, at least that is honest. I don’t seem to have the same problem but can understand that these motion issues affect different people in different ways. One reason I hate buses is that I can’t read or use a screen before quickly getting nauseous. Having said that I am not totally convinced by your explanation of how it doesn’t matter on longitudinal seats. And on a Paris train (Metro or RER) you would still have a 50% chance of getting your preferred seat, in reality higher if you took the backward facing seat and waited for one to be free. Don’t BART and DC-Metro have mostly forward-facing seats?
BART has a mix of longitudinal and transverse seats. I forget what the Washington Metro has. The trains turn back at each terminal, so they can’t have more forward- than backward-facing seats.
And what I’d do when I’d take the RER B to Bures-sur-Yvette was stand until a forward-facing seat opened, if need be. Off-peak I’d almost always be able to get one at Les Halles or Saint-Michel, but closer to the peak (say, at 3, certainly at 4) I’d wait until about Bourg-la-Reine.
So what happens to the Silver Line? Where will it go under this idea?
Branch of the Orange Line. The branch point is beyond the area depicted in the map.
The Silver Line still interlines with the Orange Line from East Fall Church to New Carrolton. Outbound of East Falls Church, riders will enjoy half the frequency as from East Falls Church to New Carrolton. But East Falls Church and outbound stations are all Park ‘n Ride stations anyway (well the Orange Branch, anyway), so half the frequency of Ballston/Rosslyn Corridor makes sense.
I do very much prefer the Yellow Line to follow Archives and Gallery-Place to Union Station over Federal Center SW and Capitol South. The former is option much closer to the CBD as plenty of people work at Archives (my dad did), Federal Triangle (the Commerce Dept. Building), Gallery Place, and Metro Center. Further, more Alexandria Yellow Line riders will be willing to transfer at Gallery Place to the Red to get to Farragut — effectively reducing the growth of demand on the Orange and the L’Enfant Station (which will have six tracks) — but won’t be willing to transfer to the Red Line at Union Station to get to Farragut (too circuitous).
Also, is Georgia Ave. a good corridor for Metrorail? Traffic is so bad on 14th and 16th Sts. and Georgia Ave., that it seems to be worth expansion there. Other potential good corridors: Columbia Pike in NoVA, Pennsylvania Ave. in SE DC, Good Hope/Naylor Rd. in SE DC, MLK in SE DC, no?
Thanks for all your hard work.
Georgia Avenue sort of already has Metro service…
I’m thinking around 16th St. Heights and further north. The 52/54, S2/S4, and 70/79 buses all have decent ridership such that rail is worth considering on 14th or Georgia.
This doesn’t solve the secondary goal of the expansion plan, which is to further address the trunk capacity imbalance between VA and MD. Building out a Blue Line tunnel and terminating trains at Pentagon also solves the branching issue, but preserves the ability of Metro to build, say, a Columbia Pike line in 2100.
Isn’t there still a reverse branch because the Silver Line terminates at Largo, or are you pushing that somewhere else?
Also, if you wanted to split the red line, and the job center is Farragut, isn’t the best place to split the red line down the middle between Metro Center and Gallery Place? Yellow line trains take the line Metro Center and east, Blue line trains from the new tunnel take the line Gallery Place and west.
I’m not sure I agree with this map or with the thinking that led to it.
There is only one case of reverse branching in the current network – the Yellow and Blue lines at Pentagon. Surely the easiest solution to that is to extend the Yellow Line along Columbia Pike, and reserve the southbound branches for the Blue Line. A Columbia Pike extension with TOD is a natural next step even without the interlining problem.
Once that is done, all that’s left are the “traditional” problems of decongesting the Orange-Silver-Blue tunnel and serving Georgetown. The “traditional” solution, a separated Blue Line, works fine for this.
Where the separated Blue Line should go east of Union station? WMATA suggests that it turn south and merge into a separated Yellow Line, forming a loop around downtown. In that case, H Avenue/Benning Road could be served by a light rail line (like at present, but *not mixed traffic*) extended west to the Farragut area. It’s a short distance, so this should be acceptable. It wouldn’t be so bad to keep 2 branches on the east end of the Orange-Silver line, paralleling the 2 branches on the west end.
Alternatively, the separated Blue Line could continue east and take over the Largo branch, like you suggested. In this case, the Yellow Line could continue north as at present, with the Green Line extended along the waterfront and then to the job center at Farragut.
The problem with what you’re proposing is that it makes the Orange Crush even worse than it already is, because it forces four different branches onto the same tunnel. The Blue Line is running 6 peak tph today, and adding another branch to it would require intolerable compromises on Orange Line capacity.
You obviously know a lot more about this than I do, but I was under the impression that the proposed Georgetown stop is a convenient landing spot for a separated Blue Line, and not a raison d’être by itself. The other proposed stops on the separated Blue Line are much more important, especially the connection at Union Station. Creating an entire third river crossing has to increase capacity more than maximizing the Yellow Line bridge, no?
There are a number of current and potential problems with the existing layout, especially on the Virginia side.
The Silver line is a big and still growing influx of new passengers on the Rosslyn side; it also would be reasonable to expect the Orange line to add some additional stations as they are already positioned in the middle of I66. There are enough potential riders out there to fill up the Rosslyn tunnel.
You could probably say the same thing about the 14th St. bridge side, although the powers that be don’t seem to be as interested in expanding that way yet.
Unfortunately, the Yellow line’s route does not hit so many desirable DC destinations; this is a part of the problem. The Blue-Orange line was designed to hit all the DC places the designers thought that people wanted to go to, the other lines just go “wherever”.
There’s no end of kludgey patches, but I think that what you really need is a few basic changes in the DC part of the network.
K St. Trunkline
Instead of having the Blue/Orange line taking a right on 12th St. NW, keep on going on K, heading toward Union Station switching to Massachusetts at Mt. Vernon Sq. or just sticking with K (could work out with future Union Station add-ons).
Georgetown Crossing
Build the new crossing for an express from East Falls Church (either Silver or Orange or – Gasp – half of each) hitting Georgetown, Dupont Circle (or a new Red line stop at Connecticut and M), Thomas Circle, then go south on 12 to pick up the orphan eastern segments of the Orange and Blue.
the All Virginia Line
Build a link west of Rosslyn to allow some Orange and/or Silver trains to go south to the Pentagon and Alexandria; there is sure to be some demand for Tysons Corner and Dulles Airport from down there. (Leaving consideration of routing some of those trains over the 14th St. Bridge ’til some other plan finds a decent route for some of the Yellow line).
the Pennsylvania Ave. Corridor
The Georgetown Crossing could follow the 30’s Bus lines from Georgetown down M to Pennsylvania Ave and by the White House, Federal Triangle and the Capitol to SE and onward, either interacting or not with the Blue Orange at Eastern Market.
the Massachusetts Ave. Corridor
Mass Ave is the longest street in the District and has its own set of highlights, like Northern Georgetown / the National Cathedral, Embassy Row, Dupont Circle, DC Convention Center, Union Station / Capitol Bldg., [available big lot formerly RFK Stadium / DC General Hospital / DC Jail up for redevelopment (i.e. Amazon East)].
Nothing to fear with interlining, after all the computer can run the whole thing on time.
“The Blue-Orange line was designed to hit all the DC places the designers thought that people wanted to go to, the other lines just go ‘wherever’.”
And therein lies the problem: you can’t just build a line with a one-seat ride to everything and connections to the line. You have to build a network. Cities evolve, major and minor job clusters change, and one line can’t do it all for an entire half to full century. You need a grid with linear paths like NYC (granted waaay too much reverse branching) — or at least Boston (which doesn’t connect the Red and Blue), Chicago (crazy transfers with some lines underground and some above ground), Philadelphia (only two lines), or San Francisco (only 1.5 lines). DC’s transit skeleton is so inefficient with reverse branching and circuitous paths, and even Alon’s corrections to it can only solve so much.
If your trains carried no passengers then reverse-branching wouldn’t be a problem. The issue is that real trains have somewhat unpredictable dwell times and there’s nothing a computer can do about that, other than let the delays propagate.
That was supposed to be a reply to johndmuller above
pesky passengers. Branching at the suburban ends doesn’t help either. Think of how much it would cut down on fleet size if everything turned around at the place where the suburban branches, branch! Pesky passengers.
Well, there’s only so much traditional branching you can handle too, but the capacity effect isn’t quite the same as reverse branching.
Why does the Yellow Line need to follow a rectangular street grid? Wouldn’t it be simpler to tunnel at a diagonal under Maryland Avenue and the various parks to get to Union Station? That would save the costs of constructing two transfer stations.
It could be cheaper, I’m not sure – the southeastern alternative can be done cut-and-cover to save money. The northwestern alternative is not about saving money but about maintaining service to Gallery Place, for easy CBD access.
In that case, it might be better to go north on 14th St and have transfers at Federal Triangle and Archives/NM/PQ. Then there are no overlapping lines, or 3-way transfer stations that are prone to overloading.
The problem is that this breaks the connection to L’Enfant Plaza, which is the #2 morning peak destination from the Pentagon and points south (and #1 is Farragut West, already broken in this plan).
“Some remaining areas near the Penn and Camden Lines could benefit from infill on commuter rail, and do not need Metro service.”
That’s a little dismissive. Infill commuter stations could serve New York Ave, but they run considerably south of Rhode Island Ave within the District. The Rhode Island Avenue corridor is straddled by inherently transit-oriented neighborhoods, originally built around the old streetcar line, from the central core on out to College Park. As long as we’re tunneling under H St., which already had a development renaissance in anticipation of the streetcar, I don’t know why we’d ignore Rhode Island Ave.
A Rhode Island Ave. line could create direct connections from Farragut to Shaw (with infill at Logan Circle/14th St.), Shaw to Rhode Island Ave/Brentwood (infill in Bloomingdale/Eckington), and go on to connect southern Brookland, Langdon, Woodridge, Mt. Rainier, Hyattsville, Riverdale Park, and College Park (with a couple connections to the Maryland Purple Line out that way). It’s certainly worth more than a Georgetown spur, in any case.
The Camden line is only 500m south of Rhode Island Ave. Seems like a waste to build a whole new subway line just to avoid that 5 minute walk.
Untangled 2018/01/02 – 03:33
You are doing it again: believing you have presented conclusive evidence of your claims. You haven’t. I am not even convinced that HK short-hop metro line (that terminates in the city!) outperforms RER-A. (And even if it does, it doesn’t prove what you seem to think.) And your attempt to discuss seating was a mess. What we really need to know is the capacity (total and seated incl. strapontin) of the non-duplex, 4-door MS61 car versus the new duplex, 3-door MI-09 car (it’s probably on those French websites that refuse to load). The thing is you continue to compare mostly-standing metro with mostly-seated RER (and BART etc). And then you compound it with unsubstantiated claims about duplex trains. As we know from past posts this is largely politically-driven spin.
All engineering designs look good on paper but many are found to be wanting when tested in the real world of conditions beyond expectation, or at their design limits. And like I said, that may yet be true for the HK Metro which I don’t believe has approached that 75,000/h design criterion. The fact is that RER-A has achieved 30tph and 60,000 pax/h (both directions), from 2011 when they were introduced. The continued relentless increase in ridership of RER-A has caused an increase in occasional line disruptions (which propagate up and down the line in the 3-branched 5-line RER-A).
The branching is a problem, but SNCF is doing nothing to fix it. It could do so, bundling it into the RER E extension: the RER E would take over the Poissy branch, giving the RER A full control of the Cergy and Saint-Germain-en-Laye branches. Even this would only give the RER E 12 peak tph on the west, so the RER E could even take over the Cergy branch and leave the RER A with just the RATP-owned segments. Instead, the RER E is only taking over the line to Mantes-la-Jolie, at 6 tph, with a really awkward (and as far as I can tell globally unique) 6+16 tph track sharing arrangement between La Defense and Rosa Parks, leaving the reverse-branch at Cergy as is.
Right. But decisions on transit issues are never just engineering etc., mostly (local) politics and money. And timing. Perfection only resides in theoretical treatises.
Also they already have a tonne of branching planned for the eastern expansion of RER-E (E1, E3, E5, E7). On the 2025 plan I see, it is RER-F that serves Mantes-la-Jolie. I suppose, too, that with the Grand Express plan it is expected a lot of pressure will be relieved on all the radial lines so these are considered short-term problems that will go away fairly soon?
“with the Grand Express plan it is expected a lot of pressure will be relieved on all the radial lines”
The Grand Express plan is not magic. Radials will still be inherently higher-ridership than circumferentials.
They have a lot of branching to the east, but all of these trains are planned to terminate at La Defense, instead of continuing west, even though the RER A and Transilien J and L could really use the relief. My crayon is still to have the RER E take over Cergy and Poissy (and points west, inc. Mantes-la-Jolie) and then connect the Transilien L lines serving Saint-Cloud to the RER A.
And as for relief lines, the worst crowding on the RER A is between Les Halles and Auber. GPX isn’t relieving Auber (and neither is the RER E extension), does nothing for passengers from points east to La Defense, and only offers limited relief to RER B-to-La-Defense transfers.
Really? And do you mean “continuing east”? Surely one or more of those E1 thu E7 would continue on thru La Defense to Auber etc. And surely with E (and F) being an alternate for the eastern banlieusards (and with F + M14-extension southerners too) to get to Auber it would bring some measurable relief to those currently using A.
But anyway, and in response to Eric, the concept of GPX and Grand Metropole Paris is not just to improve transport into Paris but to decentralize more things to extramuros so fewer people will need to commute into central Paris. Of course the radial lines will always have more traffic than the circumferential lines but they should still bring relief. This is a 20+ year plan, and will be fascinating to see unfold.
I mean continuing west. Those RER E trains from the Gare de l’Est network are planned to go through Magenta to La Defense and turn there, instead of continuing onward to Mantes-la-Jolie. Don’t get confused by the different appellations – the RER E and F are the same line between Rosa Parks and La Defense.
And so far the announced real estate development in the suburbs, e.g. in Aulnay, is more residential than commercial.
Right I can see that. They obviously share the tunnel (and maybe a lot of the track west of those tunnels?). But I misunderstood your original comment. You were talking of pax from eastern Paris while I was talking about those from western/SW Paris. I was also confused by “Rosa Parks” which is still labelled “Evangile” on my 2025 map (amusing to see that socialist Bertrand Delanoe broke with naming convention and got a bunch of those new T3 stations in the NW corner named after his fave socialist/feminist heroines, Rosa Parks, Ella Fitzgerald, Delphine Seyrig*).
However, this plan shows RER-F turning south on the eastern side of St Lazare-Auber and charting a (new?) route to Gare Montparnasse (via Musee Orsay) then out to the SW suburbs and in a giant loop eventually back to where it started at Mantes-la-Jolie. Is that still on the cards? Has it begun construction? Surely this line would divert pax from going to/changing lines at Chatelet, and thus relieve RER-A? In fact changing at the dreaded Chatelet (for RER-A) would tend to encourage pax to stay on F all the way to La Defense even if it is a longer ride.
…………………….
*I just refreshed my memory about Seyrig and discovered this (exactly 40 years ahead of Hollywood!):
The Saint-Lazare-Montparnasse connection is vaporware. I don’t think it’s even a good idea; if there’s money for a new RER line under the Seine, it should be a four-track line from the Invalides-Gare d’Orsay tunnel to Saint-Lazare, breaking the RER C in two and connecting each half to half of the RER E, to create northwest-southeast and southwest-northeast commuter lines.
OK. My first response is ‘tant pis’.
But isn’t your preference essentially identical to the Musée d’Orsay to Saint-Lazare route of the RER-F (on the 2025 fantasy map)? And linking it to Montparnasse is a no-brainer and kind of completes the conversion of Transilien lines to RER lines? In my non-expert 5-minute deeply considered judgment, it looks quite good!
If this applies to the part you have quoted then I would agree, the part you have quoted in the brackets is my own opinion on seating. I won’t pretend otherwise. The part on the loss of seats in longitudinal isn’t though.
My main point is that single level trains should be able to handle this situation better, when the RER is employing technology usually reserved for metros like CBTC, ATO or SACEM and is carry lots of passengers, then it should be to some extent compared to them. (But don’t mistake the comparison with metros for me saying that RER A is 100% metro-like, even if they switch back to single decker, I would still expect them to retain some, but nevertheless fewer, regional rail features. (like more seating than metros))
You know this could go on for ages so I’ll just say, at the end of the day, all trains are trains and are in many ways directly comparable to each other no matter what other people say. As long as a train rocks up and runs to schedule, most people won’t give a toss.
Also found this so I’ll just leave this here with some quotes:
Click to access 2009_Coxon_Burns_deBono_Napper.pdf
“widening doors is not as effective as having multiple doors”
“it takes longer to disembark a train from a position between two widely spaced doors than to maneuver between a three-door carriage in comparable capacity situations”
“A typical doubledecker carriage will take approximately 40% more passengers than a comparative length single-decker carriage. However, their dwell times are 0.3 seconds per passenger slower than single-decker rolling stock”
You’ve again ignored the arguably most important factors: seating & comfort on suburban commuter lines. Interesting the article you cite hasn’t ignored that at all! Naturally it didn’t surprise me to see that almost all their data on such comparisons come from analysing Paris transit:
Do you notice how those responses to “not having a seat” or “standing in a crowded train” are rather more than the 0.3 sec dwell penalty! Then there is this:
Plus, on duplex trains:
Note that while there is no doubt about higher capacity for duplex trains (as you cite, it elsewhere says +40%, not a trivial figure), how much this negatively impacts dwell times and pax accidents remains “some research indicates”. As a research scientist with 120 papers to my name, I can translate that for you: “we don’t know!” (If we actually knew we wouldn’t use that kind of equivocation, believe me. In high-quality peer-reviewed journals, one knows if you deploy such certainty you risk reviewers’ hackles rising and having another point taken off your review.)
Note that their Fig 2 shows the wide-3-door wide-vestibule arrangement (including 2+2 seating!) of the Paris RER (and for that matter their Metro)! They list the other strategies (less seating, more standing, more doors) as adopted by HK-MTR, which of course is short-distance inner-city Metro service as I have been saying from the very beginning of this exchange.
At any rate, I thank you for the paper. I think it does provide useful data and analysis. It also essentially does not in any definitive manner offer a clear-cut benefit of one over the other, except perhaps for high-capacity short-run Metro. Part of the reason I get my back up when you (and to a certain extent, Alon) start laying down your criticisms, “justified” with dubious inappropriate world examples, is that you seem to treat the people who design and run Paris transit as inexperienced amateurs who refuse to see common sense! But of course they are just about the most experienced designers of city mass transit in the world, and duplex trains on RER-A is their 4th iteration on this line alone, since 1977 (and they were scouring the world, including Tokyo, in 1970 to design the RER, trains and platforms). More than that, France has that peculiar thing (in the western world, almost non-existent in the Anglosphere; notably different in Asia) of the people in positions of both managerial and political control have actual engineering experience. (They sometimes have both.*) Or they aren’t so dumb as not to listen to expert advice.
IMO it is a big factor in why so much of the public infrastructure in France is so impeccable, and so much in the Anglosphere is so underdeveloped or awful. I followed the sorry story of the Sydney NW Metro and how the politicians (right-wing of course) completely–at the last minute–over-rode all the train experts, including a British one they brought out, and imposed their single-deck design. Many of those experts objected in public and some resigned. The rightwing politicians refused to make public the “advice” they claimed they based their decisions on.
On the RER design the French have balanced the capacity, including the crucial seated capacity, versus egress factors and ultimately dwell-time, and IMO come up with a good design. The alternative may be the more econocratic design (for entities in a commuter algorithm) that functions with (arguably) marginally better performance but which is hated by actual transit users.
Finally, if you actually lived in Paris and experienced in a daily routine these forms, you might change your mind. Incidentally the Sydney NW-Metro may get away with its design for decades because of patronage levels way below anything like Paris or HK etc. but it has wilfully discarded expert advice and burnt its bridges (well its tunnels that were ordered by the politicians to reduce the tunnel size specifically to exclude current Sydney duplex trains; an echo of Moses 50 years earlier building his parkway bridges too low for public buses!). We’ll see how the pax think; it is a 66 km route from Eppping to Rouse Hill in exurban NW Sydney, and that forces them to change trains to travel the last 10km into the CBD.) But of course NIMTOO: these politicians won’t be around when/if the shit hits the fan (indeed the NSW Premier resigned after a single year in office! and reverted to his investment banking career where he reputedly earns almost $3m p.a.)
……………………….
*As an example are Louis Sato and Philippe Essig, the authors of that 2000 report/review of the design process of the RER beginning in 1965. Both were graduates of the Ecole Polytechnique and the Ecole Nationale des Ponts et Chausées. Both had worked a decade overseas in actual infrastructure building projects. Essig became Managing Director of RATP (Paris Metro transit) from 1981 to 1985, then president of SNCF from 1985 to 1988 and then a minister in the government of Prime Minister M. Rocard. He became president of TransManche Link that built the Channel Tunnel.
Compare it to the former investment bankers, politicians, real-estate speculators (like Donald Trump!), lawyers or assorted political-lifers that determine our infrastructure in the Anglosphere, while scorning and discarding the true experts. Chosen and built by partisan amateurs, usually on short-term party-political partisan grounds, there is no need to look any further for explanations as to why 1. it is so inadequate and 2. so expensive.
The reason we’re in this thread in the first place is that this extra dwell time is reducing actual line capacity from 30 tph to 24. The MI 09 has 948 seats over 224 meters of length, and the MI 79 running in double traction has 624 over 208: 4.23/m vs. 3. But then the difference in capacity, caused by dwell times, means the bilevel only has 13% more seated capacity than the single-level train.
And that’s without going into the cost difference – the MI 09 costs €4-5 million per car, which is horrific. Bespoke trains are expensive. The total budget for the MI 09 order was €3.5 billion, for just 164 5-car trains; at a more normal cost (about half), there would be enough budget for about 9 km of new tunnel. Put another way, if RATP and SNCF had spent the same amount of money differently, there would have already been a tunnel from Saint-Lazare to La Defense for the RER E. In a more constrained environment, quadrupling Les Halles-Gare du Nord was estimated at €700 million a few years ago.
Well, that is an exactly 50% increase in seating capacity in a single train! And, again, your last sentence (above) totally ignores the “Patrons‟ response to „not having a seat‟ was the equivalent of an additional 5 to 14 minutes of travel time, this penalty increasing with the length of the journey. „Standing in a crowded train‟ was the equivalent to an increase of 27 minutes of „disutility‟.”
The difficulty is that the latter is tricky to quantitate and then even easier (for econocrats) to ignore. But patrons (users) don’t ignore it. Certainly it is the basis for my personal, if not professional, judgement. If I was travelling from Paris to the Saclay campus every day, I sure as hell don’t want to be standing. Do you really want to be forced to stand for such journeys (you complained about facing the wrong way; these trains give you a 50% higher likelihood of getting the seat you want!).
As to the costs, again it is the econocrat in you that emerges to dominate decision making, and the argument doesn’t convince me. Like your comment about PCs and the fact that Apple had occasional totally new OS that was not backwardly compatible with everything that went before: with this attitude there would be zero progress. Of course costs matter but you have to be very careful and discriminating about which costs really matter. First the cost is amortized over the 30 year life of the trains, and second, they may have imagined Alstom would pick up a lot of international orders if the design was successful for crowded big city transit (notwithstanding the loading gauge legacy issue). No accident that there are only a handful of train makers in this market, and only one of them (Bombardier) is in an Anglophone country (though not really, no accident iit is based in Montreal, francophone Canada:-).
Again, while technically some of your points may be correct, in the real world there are an awful lot of complications in making those decisions. For example, the doubling of the Les Halles-Gare du Nord tunnels, yet it is not RER-B & D that are suffering the most congestion. I like the sound of your alternatives to RER-F but actually isn’t that only tinkering with the current (fantasy plan) version. Would it make so much difference, for all the undoubted pain of the necessary changes? In such a complex thing as a transit network, surely it is “the perfect is the enemy of the good”?
Incidentally your point about “a tunnel from Saint-Lazare to La Defense for the RER E” is kind of rewriting history as originally it was intended to use (Transilien) tracks from St Lazare north thru Clichy & Courbevoie then down to La Defense. So, despite spending all that money on other things they have changed their mind to an even more expensive option. Just confirms my point about decision making. Cost (in today’s Euros) is only one factor.
Well, at least we’ve kind of moved from the stage where Michael was outright denying that there was absolutely no dwell time penalty with double-decker trains, that Spanish platforms were the shit, and that wider doors made up for fewer doors. Unless Michael wants to deny all these things again.
Despite Michael comparing a 0.3-second per passenger double decker dwell penalty (which per hour would easily add up enough to squeeze in more trains) with the 5-14 minute no seat penalty in a very awkward way, I think that there is still a strong case for single-level trains to return to RER A in the future, especially when there is only roughly a 10% seating penalty per hour.
Funny enough though, because it will be much faster, Sydney Metro, once it’s fully built, will probably save the 5-14 minutes that you refer to vs double decker, so no increase in disutility (you’re argument) even if they’re standing. And Sydney Metro will only have a 15% loss in seats per hour, not a bad sacrifice overall IMO.
*your
Untangled 2018/01/03 – 02:10
You’ve just gone and ignored everything. Again.
You do realize that those data on 0.3 sec dwell penalty is not from these trains but from older designs (pre-2002). In fact it was certainly a part of the analysis RER used to redesign the carriages, from 4 doors to 3 wider doors–to reduce this effect.
And your calculation of only 10% difference in seated pax per hour is typically inaccurate. Using Alon’s figures of 948 seats per duplex (MI-09) and 624 seats per simplex trains (MI-84) on RER-A gives 23,700 seats per hour for the duplex at 25 tph while the other gives 18,720 seats per hour at 30 tph. The non-duplex train is providing 26.6% fewer seats per hour, 2.7 times your figure. [But I believe this is for the 112m 5-car set and they are actually run as 10-car sets so double all this: 5.4 times your figure.]
But wait, there’s more! Both those trains have transverse 2+2 seating while your preferred longitudinal seating would reduce the seating by my estimate another 33%; but ok I’ll give you the benefit of the doubt and say just 25%. That gives the simplex version only ≈15,000/h at 30tph. This is 8,724 seats less per hour; or 58% more in the duplex format. [Or 17,448 extra seats per hour with the double-length trains sets.]
But wait, there’s still more!: Every train will have 50% less seating, and with trains only running at 30tph at peak for a few hours of the day, this negative factor would be spread across to non-peak hours where the difference would revert to 50% fewer seats per hour without the duplex trains. Even worse if longitudinal seating.
I think I understand why the engineers and transit experts at RATP and in the French government chose the duplex trains.
Yet again, you ignore that RATP considered all these factors in the design. And that you and a bunch of Anglosphere econocrats know best. Maybe that is why the only systems in the Anglosphere to remotely compare to Paris,ie. NYC and London, do not look good by comparison (even on paper; as a commuter, even worse IMO), and it gets worse the further you get from their CBDs.
And you’re dreamin’ if you think they’ll be running trains to Rouse Hill at even 12 tph compared to 24 tph of RER.
But ok, we (you) are going round in circles, with you consistently ignoring the factors that matter in a long suburban service versus a short city metro service.
Yes I do realise that the MI-09 will likely have less of a dwell penalty compared to the study but you were for some reason comparing that to the “disutility” time for standing even though there is no connection between the two, that’s what I wanted to point out. Although I still don’t that the MI-09 has fully eliminated the penalty.
As for seating, the old seating wasn’t very dense despite it being transverse, you could easily pack in a few more seats per car without a dramatic loss of comfort or standing space, even the MI-09 appears to be denser. MI-84 seems to have a density of 2.8 seats per metre if 624 seats is true, generally a single deck suburban carriage at 2+2 will have around 3.5 seats per metre. And both 948 and 624 figures are for 10 cars.
As for the French government, I’m pretty sure that the decision to order the MI-09 was made in haste and rushed so I doubt that what you say about them considering absolutely everything to be true, it was a carry over from alighted older train. Making decisions in haste doesn’t sound like holier than thou government you seem to be promoting.
And yes I do think they will be running, over 12 tph in new line and I’m not dreaming, you may want to look at the EIS. You should come.
Untangled 2018/01/03 – 08:57
You can change the topic all you like (seating density? the RER was specifically designed for more seating in more comfort), but I notice you haven’t refuted my calculations (I may have got the train length wrong but that was quite irrelevant to the “percentage seated” comparison which stands). Under any of the scenarios the number seated is much greater on the duplex versus the simplex trains, and even greater than a single-level train with longitudinal seating. Unless you are sticking to your 10%? So I take this as you ceding the argument.
I think your comments on the French government reveal your inherent bias. I certainly don’t claim they are perfect but your complaint (unsourced, seems opinion) that they make rushed decisions perhaps might commend some other governments to do the same. RER-A was planned from the mid-to-late 60s and opened in 1977–and it has become the most intensively used commuter line outside Asia. London CrossRail was actually planned even earlier–early post-war–but in its final actual incarnation (being explicitly modelled on RER-A) will open, maybe, late this year, a mere 42 years after RER-A &-B. We could go thru project after project in the Anglosphere that should have been built but hasn’t been, before moving onto tramways and then HSR …
As it happens I have just caught up with an article in the NYT last week (by Brian Rosenthal, 28 Dec; referenced in today’s CityLab; cites Alon’s articles on NYT construction costs) which has a significant section on comparing Paris (specifically M14 extension). One strong theme throughout the article is the lack of control of NY projects by the city, and total reliance on consultants. “In Paris, which has famously powerful unions, the review found the lower costs were the result of efficient staffing, fierce vendor competition and scant use of consultants.” While “Soft costs for (NYC) East Side Access are expected to exceed $2 billion. The project plan called for the hiring of 500 consultants from a dozen different companies, according to a 2009 federal oversight report.” “On East Side Access, it is sharing the contract with STV Inc., which recently hired the former M.T.A. chairman Thomas F. Prendergast. The contract was initially for $140 million, but it has grown to $481 million.”
I have banged this drum to Alon many times so let me do it again. NYT may be one extreme but throughout the Anglosphere there has been a deracination of the role of civil servants in big projects, so that they barely know the details of a project and must hire a consultant to answer the simplest of questions. Worse, there is a constant revolving door from those nominally in charge of the public interest and the companies involved: “A Times analysis of the 25 M.T.A. agency presidents who have left over the past two decades found that at least 18 of them became consultants or went to work for authority contractors, including many who have worked on expansion projects.” End result:
But while the (NYC) Second Avenue Subway cost $2.5 billion a mile, the (Paris) Line 14 extension is on track to cost $450 million a mile.
I think we (ie. you Untangled) could learn a lot from the French. But I have an inkling you may be one of those consultants sucking on the public teat in Oz?
Re Sydney NW Metro, an EIS will assess the worst-case-scenario re environment, ie. the highest tph the line could technically support. We’ll have to wait to see what happens in the real world when it starts operation. In the context of the discussion above here is something re this Sydney project:
Consulting firm Turner & Townsend were awarded the tender for cost planning services on 22 July 2011.[41][42][43] According to Turner & Townsend, the company’s role will be to “manage the project budget and demonstrate to the taxpayer that they are receiving maximum value for money”.[42] This is exactly what I am talking about. This kind of oversight should be by the civil service in the public interest. I am pretty sure the French don’t outsource such responsibility.
Talk about changing the subject, now you’re on construction costs (which I never raised in this thread anyway), but whatever.
If you want to talk about Sydney construction cost, let’s look at the Metro Northwest you refer to. It’s 23 km of new tracks at AU$8.3 billion but with some calculations and you end up at US$280 million per km. I know the 23km is only around 2/3 underground but the cost is not bad for a new upstart metro system that Sydney has never had before. But don’t forget that the 8.3b also includes the conversion of an existing 13km regional rail line with CBTC, level boarding, platform screen doors, etc.
And that existing line (but still newish) to be converted is fully underground and was initially built at AU$2.1 billion or with some calculations, you end up at US$126 million per km. Hardly excessive construction costs. If you add $2.1b and $8.3b together for the entire 36 km initial line and convert it, you end up with US$226 million per km. But knowing you, you would probably say that it’s very excessive, almost New York expensive and way more expensive than the US$280 million cost for the M14 extension.
If you seriously want New York construction costs though, you should at Melbourne’s 9km of tunnel at US$958 million per km, now that’s crazy.
Well, I notice you still don’t put up any argument against my numbers, so I just dismissed that and turned to a related subject.
Ha! Untangled scores an own-goal. The only reason for that is that they (politicians, against all expert advice and on a fictional report they refuse to make public) abandoned Sydney’s classic duplex trains that use that section. This means all the stations have to be rebuilt to accommodate the new single-level trains.
I think it is a indictment of Sydney transport planning that they had to resort to expensive tunneling at 50km out in the boondocks! That is the only new section of track from Epping out to Rouse Hill. For Epping to Chatswood they have closed the existing line for a year while they modify the stations–the users have to bus it. But when complete the new trains will terminate at Chatswood and passengers change to the standard Sydney train for the final 10km to the CBD. The whole journey is beyond 60km.
As to costs you are not comparing like-with-like. 50km in the boondocks versus in the hyper-dense and heritage Paris, or even inner-city Melbourne.
The worst thing I’ve heard about Sydney tunneling recently isn’t even the cost, but the complete opacity of the process; the public only gets to learn e.g. what the cost is from deliberate leaks.
Well apart from the 7-month itself, I don’t see anything wrong with it since there will be benefits, like I dunno a train every 4 minutes instead of 15 during the peak. And that Chatwood thing is a cross-platform change and it’s temporary until the next stage. That line has been heavily underserved since it was built, Metro will change that.
It’s not a complete like for like (the 226m vs 280m, or whatever) but it doesn’t matter as much since the M14 extension is mostly in the suburbs anyway. And while Cudgegong Rd can be call boondocks, you’re forgetting there are other significant hubs within the new 23kms like Castle Hill (which has a large, expanding shopping center and an apartment construction boom to contend with right next door) and Norwest/Bella Vista (one of the larger employment hubs).
True that a tunnel for this line isn’t ideal but the area is very hilly so a tunnel was probably the least worst option.
Melbourne’s new tunnel is still a joke in cost though.
Untangled 2018/01/04 – 08:02
You’re thinking of the proposed M14 extension deep into the southern suburbs. The NYT story was about the current short northern extension from St Lazare to Clichy. Anyway remember those Parisian suburbs are still much more densely packed than the kind of sprawl 40km out in NW Sydney.
In terms of costs and transparency in Australia (though it really is an Anglosphere, even a neo-liberal thing) I really strongly believe some control needs to be brought back within the civil service. That will mean rebuilding departments from the ground up, because for 25 years now they have been destroyed by politicians. Politicising heads of departments (on contract instead of career civil servants) is one of the big factors in producing this kind of thing. In NYC it has gone crazy and is part of the deep corruption (not necessarily illegal but that’s the problem, it should be illegal) via political donations and patronage and (consultancy) jobs for the boys etc. Do we really expect yet another bunch of exactly the same expert pool, ie. consultants like Turner & Townsend, to not exploit the system? Ludicrous. And you can be damned sure the French (and probably Germans and Swiss etc, and certainly the advanced Asians) don’t outsource such critical functions to industry creeps. As that NYT article explained about the M14 job, “RATP handled all the contracting. … Officials awarded dozens of contracts, and most garnered at least a half dozen bids, driving down costs.”
The Melbourne tunnel might be outrageous (I haven’t followed it; are you sure the costs don’t include a lot more than just the tunnel; it is unravelling the spaghetti of shared tracks/tunnels around Melbourne that should have been done half a century ago; btw, it seems the Sydney Metro cost does not include the trains?) but it couldn’t compare to the Melbourne East-West Link fiasco in which those consultants ripped off the public for a sweet $1bn (some say $1.5bn; no one can even say precisely!) for not building one mm of the tunnel. Now NYT they have the most expensive mile of tunnel in the world but I think EWL beats them because it is infinity!
Of course that brings me back to Sydney NW Metro which these conservatives (under Premier, now investment banker, Baird) are trying to emulate—of all bloody disastrous examples–the Brits who privatised their trains and are trying to do the same to the LU. There was one reason and one reason only why they decided, against all expert advice, to exclude the standard duplex trains: they want to privatise the network and this is going to be the first case. It will be run independently of the rest of the network (and can never be fully integrated) and its management will be outsourced, blah blah. And some of the politician creeps involved will end up profiting from it, in the same way that Max Moore-Winton, John Howard’s head of PM’s department while the Sydney Airport was privatised, then became the inaugural CEO then the Chairman of the private entity running this monopoly (and the biggest one in Australia). If it sounds like something you expect in Italy, it was. It should have been illegal and I remain unconvinced it wasn’t, and at the least it was pure corruption, if done in plain view. And you want to trust what these sharks cook up in reports they refuse to make public, and that every other expert condemns?
That opacity in cost is quite true and I don’t like it either. The reason they give is that the true cost won’t be known until the contracts are signed but they should provide estimates. (Although estimate are known to be widely off so they’re probably playing it safe, eg Light Rail estimated $1.6 billion but actual contract was $2.1 billion and that’s not the worst (although you can thank the French Systra for that blowout designing it so that only Alstom was competitive, should have fired them and asked the Germans to do it like in the Gold Coast)).
This is somewhat similar in spirit, if different in solution, to Alex Barclay’s 2013 proposal to WMATA’s capacity problem:
https://ggwash.org/view/33210/can-metro-add-capacity-without-a-downtown-loop
That plan involved 9.7km of new tunnel. It gave over the western half of the Red Line over to the Green Line, which was rerouted through a 4th and 6th streets subway; moved the Silver and Orange to an M Street subway, gave the old Blue Line route to the eastern half of the Red Line, and stub-lined the Yellow at Pentagon.
They should simply discontinue service at Arlington Cemetery and replace it with a bus route (perhaps even for free, if it prevents people from complaining about loss of service at Arlington Cemetery). The route could be Rosslyn or Lincoln Memorial (or both) to Pentagon via Arlington Cemetery. On a typical day throughout the year, the number of people actually using Arlington Cemetery is really low, low enough that a bus can handle it. It’s only during special events that Arlington Cemetery is slammed and “needs” a train to handle the crowds… perhaps on those days only, Blue-Line-as-we-know-it could be restored.
If Arlington Cemetery didn’t exist, what to do would become so obvious: Send the Blue Line over the Yellow Line bridge and terminate it at Greenbelt (essentially make Yellow Rush Plus the new Blue Line). That would incidentally also solve the “Green Crush” problems at Shaw/U St/Columbia Heights/Petworth.
Alas, now that it’s actually built, the veterans’ groups and Congress will never allow WMATA to close Arlington Cemetery permanently. Even though 100% of the region’s taxpayers don’t live there and 99.999999% of the region’s employers aren’t located there (the sole exception being of course the Cemetery’s workforce).
The problem is that this wrecks Pentagon-Rosslyn service.
Using the existing Union Station metro station for either the Yellow Line or an H Street subway won’t be easy: it’s very shallow, and is only a few dozen feet away from the main station’s western stub tracks. Making things worse, Amtrak plans to turn the area immediately abutting the existing Metro station into a concourse (and–eventually–into a second set of through tracks) in the next few years. As for the approach from the north, the Red Line weaves under the H Street Bridge (a.k.a. The Hopscotch Bridge) but *over* a tunnel under H Street that Amtrak owns and plans to continue to use. The Red Line can’t approach a deeper metro station from the north because it has to be at-grade with the Amtrak tracks before it reaches K Street, otherwise it would have to go under K Street as well–which it can’t. The east-west H Street subway, on the other hand, would need to be deeper than the existing metro station to clear Union Station and its future expansions.
This isn’t to say that the Yellow Line can’t use the existing station, but I think it would have to take a pretty southerly approach, much like the one that the existing Red Line uses today. In order to approach Union Station, the current Red Line turns south immediately after Gallery Place so that it can swing back towards the north at Judiciary Square. This is part of the reason why Gallery Place features its unfortunate T-shaped station instead of an evenly-squared cruciform like the one at Metro Center… and why Gallery Place is overcrowded right now.
TL;DR: I think it would be easier to connect the proposed H Street subway to the Yellow Line instead of the Red Line. That would allow a deeper Union Station metro station under H street, which is closer to future development anyway (e.g. Burnham Place and NoMa). I also tend to think the North-South component of the Yellow Line might be better under 9th or 10th Street. That would not only be closer to the center of the CBD, but it would also allow us to leverage the fact that the FBI building will need to be replaced eventually (and will therefore be a good place for a cut-and-cover station). A 9th/10th alignment would also be closer to the City Center development, and would tie in well with WMATA’s plans to connect Gallery Place and Metro Center with a pedestrian tunnel.
Greater Greater Washington did a rewiring of Metrorail (albeit more expensive than yours) a few years back. Make the jump: https://ggwash.org/view/33210/can-metro-add-capacity-without-a-downtown-loop
They note the mismatch in demand between the Shady Grove (very busy) and Branch Ave (a fair bit less busy) spokes, and propose to remedy that with a short-turn at Independence Ave/4th St SW station. I personally disagree with that detail; the level of mixed-use and residential development that has occurred around Waterfront and Navy Yard stations since that article was posted is nothing short of incredible — and we should be planning for maximum utilization of the new Green Line further south than Independence/4th. I propose a pocket track immediately outbound of Southern Avenue, where numerous bus routes from the south and the northeast converge, contributing to the higher entries/exits there. This plan gets rid of reverse branching without converting the Rosslyn/Pentagon segment to a mere shuttle. It also makes for a convenient way to connect the new Red Line’s H/9th-10th station to both Metro Center and Gallery Place, and straightens the curve east of Gallery Place currently on the Red Line by moving the Red Line north to H St as opposed to the current tunnel under G St, which will carry the Green Line once the rewired system begins operation in the new layout (if it happens, of course).
Alex Barclay (a guest writer at the time), in harmony with previous GGWash posts, emphasized the importance of cross-platform transfers where lines currently merge and diverge — minimizing transfer penalty and making the system that much easier to use (on a daily basis; this different layout from most stations will need to be clearly signed to minimize tourists’ and out-of-towners’ confusion, albeit a low priority among most transit systems). To enable cross-platform transfers, the current uppermost track at both Rosslyn and Pentagon stations will need to be moved down 1-2 levels (depending on what the best trajectory is for the new Orange/Blue tracks under the Potomac to Georgetown, under M St NW, and back to 12th St NW for Metro Center — it’s going to be deep, as street level in Rosslyn sits on a high bluff overlooking the river — not to mention Rosslyn’s current street-to-mezzanine escalators are the fifth-longest span in the Metrorail system, at 194 feet long (diagonal rise) — and quite steep, too).
This is currently happening. I pointed out this blog post when I responded to WMATA’s survey. Encourage y’all to take this survey as well: https://www.wmata.com/initiatives/plans/BOS-Study.cfm
Click to access 4A-Blue-Orange-Silver-Capacity-Reliability-Study.pdf
FYI, latest update on what’s happening at WMATA/Metrorail