Trains Per Hour in the Downtown Tunnel

63 Replies to “Trains Per Hour in the Downtown Tunnel”

1. The existing stations feel dingy and dated. If major work is proposed in DSTT1, we may as well go all the way and add platform screen doors, along with new walls completely separating the air volume of the tunnel from that of the station. The rest of the public spaces could be then be completely reimagined, I would suggest looking to the Elizabeth Line as inspiration rather than the architecture of newer underground Link stations, which feel somewhat goofy and unrefined from an architectural standpoint..

   1. I used to find the DSTT stations were too modernist and geometric. But over the years I’ve noticed more details, and Westlake has a lot of art deco elements and how they’re placed. Look at the clock, the pendant lamps, horizontally across the mezzanine as a whole, the Westlake station sign, etc. I tried to link to images but the ones I found aren’t directly linkable. Here’s a search page of them. (Only some of them are our Westlake Station.)

      The murals have commercialistic elements, but one is like Alice in Wonderland with Andy Warhol items, and the other is a mishmash of pieces from the mindset that got you Picasso and the Cubists. So it’s the kind of art they were experimenting with in the 1920s art deco era, and it’s confined within the murals’ frames.

      Does anyone remember a third mural that’s no longer there? It had three women’s silhouettes from the back, with hats appearing as large black circles. The images were made up entirely of their clothing — I looked for bits of their hands or something organic but there was none, as if they were people shapes assembled from clothes and accessories. It looks like it was removed to make another information board. Does anyone remember when or why it was removed?

      Pioneer Square is the second-most interesting station artistically. again with elements of art deco. Symphony is the most modern, but it is the financial district so that sleek effect goes with it. Intl Dist has those origami-like paintings that some people may like but not particularly.

      The worst station was Convention Place, looking like something that would complement a freeway interchange. Fortunately that one was destroyed. It used to be my home station for eleven years, and I used to find it depressing, both the station design and especially walking across the freeway to Capitol Hill. I often took a bus to Westlake to avoid walking across the freeway or using that station. Fortuntely, the more recent dog-walking park gives something to look forward to as you walk across the freeway, and makes the ugliness span two blocks rather than four, so it cuts it down considerably.

      The DSTT stations are about the most visually appealing ones in the whole system. The only one I’d say is comparable is the “found items” wall on the U-District station platform. And second the escalators down from the Capitol Hill northern entrance, with lamppost lamps as if it were outside. Most of the rest of the stations are ho-hum station design with one or two ridiculous art pieces. With the DSTT stations, the whole station is art and has an integrated theme throughout, and each theme is very different.

      1. Your CPS style comment is sound, but it had great practical benefits. It had ramp access with Olive Way, the reversible lanes, and the DSTT. It had an electric trolleybus substation and bus layover. For the dual mode buses, it had electric trolleybus overhead.

         We will miss it forever. It probably could have become a pathway to Ballard Link via the Terry Avenue right of way. Dow led the county to sell it to the Washington State Convention Center. The sale ended bus service in the DSTT prematurely; but for the sale, as Link headway is still long, there could be several routes to/from the south on the DSTT (e.g., 550, 101, 150, 124, 131, 132, C and H lines). The sale led to the transit capacity crisis of 2019 and a decline in ridership.

         The convention could have expanded on a different block. The CPS block could have sold after Link headway was short. The nearby blocks have high rises.

         CPS had always been a TOD target; the question was when and at what cost to transit and riders.

   2. I love the DSTT stations, except for Symphony, which is meh. They don’t feel unrefined at all, and each have a unique charm and character. I’d take them over a modern train station designed to be as sterile as possible.

   3. There are two different approaches to station appearance. One is that all the stations should have uniform architecture as if it’s an aspect of branding. The opposite is that every station should have unique architecture so that a rider can glance out their window in a subway and know what station they’re in.

      DSTT was the latter. ST uses the former now.

      I much prefer the latter. The station name signs are often not quickly visible out a train window. Since I’ve commuted on trains much of my adult life, I’ve often looked up from something and been unsure where I am. Unique architecture is instantly informative about where I am, while it can take awhile to find a name on the station walls (especially if standing, waiting riders block the sign). It also helps when someone can’t quickly recognize our alphabet for whatever reason, like foreign visitors or children or visually impaired people. Learning to get off at the light blue station with brown bricks is quite easy if not almost automatic after awhile.

      There is a way to blend the two with different colors. If one has a green triangle theme and another has pink salmon images, they can look alike but also be quickly differentiated even if the main feature is square glass panels.

      I also wanted to give a cautious nod to stations as advertising revenue opportunities. I’ve seen stations that look like they’re an Apple store as a way to generate revenue for the transit system. I think it’s better to not provide situations where station appearance can change every few months or years in this way.

      So do others have a preference? I lean towards permanent differentiation.

      1. I don’t have the same experience with trains in multiple countries, but I do agree that uniqueness (and as you say, the level of that uniqueness can vary) is a better approach than the current ST strategy.

      2. I like the differentiation; it gives character to a neighborhood the station is in, and as you say, is an easy identifier, such as Asian art for Chinatown International District station. Maybe keep the theme and swap out the artworks every few years.

   4. The DSTT stations are the nicest stations in the entire system by far, they are unique and very attractive with stone, art, color. They are timeless.

      The Sound Transit designed stations are all bland soulless ugly gray and metal boxes with some splashes of outdated Pantone color of the year from 2011.

      1. Yeah, and those bland ST Link glass squares as the main feature aren’t likely to age well either. They have all the charm of a white, square tile bathroom.

      2. PS. Even the earliest subways laid tiles that have brick-shaped dimensions and were offset. Hence the term “subway tile”.

         ST didn’t get it.

2. There are three different capacity considerations:

   1. The number of trains plus the capacity of each train.

   2. The platform themselves.

   3. The vertical paths (escalators, elevators, stairs).

   Each one needs to be analyzed. And while each one has a finite capacity, delay begins to kick in once it gets to about 70 percent.

   And anywhere that has rail-rail transfers will get bigger rider surges. Those surges can temporarily flood the system and add delays. That’s another consideration.

   In discussing trains per hour, it’s obviously focused on #1. I just wanted to mention the other two to note that they exist.

3. The DSTT capacity limitation can be limited by where trains reverse direction, particularly if trains are long and drivers must walk from one end of a train to another.

   This challenge is what ST is currently facing at Lynnwood once 2 Line trains begin operation in addition to the 1 Line.

   The best way to maximize DSTT train capacity is to have branches at each end of a line. This approach can even enable three branches rather than just two. Yes there may be a short wait for riders at the merge points if a train can’t keep an exacting schedule, but if there’s some recovery time within the schedules the delay can be quickly dissipated. Automation can really help this.

   Sadly, our advisory groups and elected officials are painfully unaware of this fundamental aspect of rail system design. Staff don’t seem to ever offer a primer about rail transit operations generally before corridor planning begins. It’s just one of many problems that emerge when the feedback approach is “what do you want to build?” based on whims, backroom lobbying and personal preferences as if it’s a second-grade art exercise.

   It’s a basic flaw in the Everett Link planning as two branches would be so much better for operations rather than this current snaky routing with half the trains being short-turned and having to slot into service between the trains beginning in Downtown Everett.

   1. Agreed. Too bad Paine Field route isnt a branch. Also to the south a branch off to Renton or Kent/Auburn. Might as well split these longer lines on the outskirts and have frequent headways closer in, all while providing Link service to more areas outside the I-5 corridor.

      1. Yeah it’s just regrettable that the Snohomish County leaders were so obsessed about creating the snaky alignment partly because they were never educated about its effect on train operations. Had they planned two branches, they could have also greatly improved travel times from Everett southward as well as saved building a few miles of track (and hundreds of millions of dollars). Add to that how ST could schedule trains to time a cross platform transfer where the branches meet, actually making the transit trip faster between the two branches.

      2. Then how would people from Everett get to Paine Field on Link? Part of the motivation was that people from the north (Marysville, Skagit County) would park at Everett Station and take Link to their Paine Field jobs, to reduce congestion on the Casino freeway and surrounding streets between Everett and Paine Field.

         I never thought that was realistic, but that was one of Snohomish’s stated reasons for the Paine Field detour.

      3. “Then how would people from Everett get to Paine Field on Link? Part of the motivation was that people from the north (Marysville, Skagit County) would park at Everett Station and take Link to their Paine Field jobs, to reduce congestion on the Casino freeway and surrounding streets between Everett and Paine Field.”

         Of course, the big issue is that there is just one station west of 99 in the plan near Seaway. And that site is surrounded on at least two sides by stride or freeways.

         And one of the two planned stations on 99 (the closer but still 3/4 of a mile from Paine Field) is technically deferred without funding.

         Meanwhile a center platform at Mariner could accommodate a scheduled, short timed transfer between a train from Everett to a train towards Paine Field and it would be just as fast as one making the out of the way jog over to Seaway. The Seaway station would be lost, but the station could then be “moved” to become an end station at Paine Field instead or even it could run further to Seaway. I’ll mention that the Seaway station has the lowest number of forecasted boardings on Everett Link Extension.

         Keep in mind too that Everett Station is supposed to have a 1000 space garage (possibly using an existing parking lot) but the Board is looking to defer future parking garages anyway.

         I’m pretty skeptical that Link will be an enticing option for most people working in the SW Industrial Center mainly because there’s just not that many jobs within walking distance and the area has massive free parking lots. .

      4. I imagine companies will have to have shuttles from the Link station because it’s too far to walk, but will they?

      5. Mike, that makes no sense, is that something Snohomish leaders have actually said? The primary chokepoint in Everett is the Snohomish river, both the I5 and US2 bridges have terrible congestion; once a driver gets across the river they might as well stay on the freeway. Also, the Everett garage isn’t that big and is already full with riders trying to get to Seattle. Paine field diversion connects workers who live within the ST taxing district to jobs with the RTD. I’ve never heard a politician talk about Link intentionally serving riders outside the taxing district.

         I expect there to ultimately be a branch somewhere in Snohomish. Given ST staff forecast ridership to be high enough to merit both lines running until Mariner, that’s the logical place to anticipate a junction.

      6. “that makes no sense, is that something Snohomish leaders have actually said?”

         Yes. I attended a lot of ST board meetings in 2014-2016 so I probably heard it there.

         “once a driver gets across the river they might as well stay on the freeway.”

         That was my thought too. It would probably take longer to transfer to Link and continue driving, and then when you get the Paine Field station you’d have a last-mile problem. So I didn’t think it was very realistic.

         But that’s part of the mindset of these exurban officials. Snohomish wants Sounder partly for people from Whidbey Island transferring at Mukilteo, and Link to entice people from the north to switch to Link at Everett Station to Paine Field. Both of those kinds of riders are from outside the ST district and don’t pay ST taxes, so why spend megabucks on infrastructure for them? And Snohomish wants Link at Paine Field in order to say it has high-capacity transit there because international companies expect it. And Tacoma wants Link to the airport to attract employers to Tacoma.

         All that ignores the real predominant cross-section of travel needs by the bulk of ST subarea taxpayers, so that’s why I’ve been screaming that the regional+local transit network needs to meet their trip needs. That’s the only way to get car use down to 50%.

      7. Great ideas for connecting more areas.

   2. I’m not sure if there was ever a good way to build Everett Link. It just doesn’t have the density along the way. It is like building light rail in Yakima. Maybe if you can leverage some existing tracks it work. But otherwise it just isn’t big enough to warrant the expense.

      With Everett you almost immediately run into contradictions. You can reduce costs dramatically by running on the surface. But then it is especially slow for long trips. The same is true for stop spacing. If Snohomish County is really dense and has lots of destinations between Lynnwood and Everett, you add a bunch of stations. But that makes long distance trips (into Seattle) much worse.

      At first glance a branch could solve the problem. One branch could follow the freeway while the other covers Snohomish County. Unfortunately, neither branch would be a good value. The ridership would be too small and the cost too high. The line that follows the freeway would not be significantly faster than an express bus to Everett. Yet the express buses carry relatively few riders. Likewise, Snohomish County is just too small to get a lot of ridership *between* stations in Snohomish County. Nor is there much between Lynnwood and Northgate. It is basically a bunch of park and rides, one after another. It is just too small and spread out to work very well either way.

      1. So you are just ignoring the Swift network? Ash Way and Mariner should be strong bus-rail transfer nodes. Honestly I’m baffled by any objection these first 3 stations, unless your goal is the redirect money outside of the county. Lynnwood is an excellent anchor for CT to terminate myriad of routes, but it’s still very suburban/commuter oriented. AW & Mariner anchor the middle of Swift routes; Link to Mariner plus Swift Blue, Orange, and Green give CT the high frequency grid to support all-day ridership.

         I think the path forward with Link in Snohomish county is much more straightforward than sorting out what needs to happen in Seattle. Start by extending up to Mariner. This can be 1 big-ish project or easily broken into 2 small projects (L>Ash Way, Ash Way > Mariner), each of which are comparable to an Angel Lake type extension. This is a low cost extension and interlocks with the Swift network to create a strong, coherent grid network in SW Snohomish.

         Next, extend from Mariner to OMF-N. The specifics can be figure out by a future generation, but the overall region is going to need an OMF-N to support Link growth (otherwise ST will need to plan for a 4th OMF somewhere else in King or Pierce). Include a transfer to Blue Swift somewhere, leveraging the Green & Blue Swift lines to connect all of Everett to Link.

         Third, extend towards downtown Everett, a major regional destination with the infrastructure (street grid, permissive zoning, existing uses & large lot sizes, etc.) to support significant growth over the next few decades. Do we get to Everett via Paine Field, via Evergreen Way, via I5, or maybe a combination with branching? Who knows! It literally doesn’t matter today.

   3. Staff don’t offer a primer to the Board because the information would empower the Board to ask pointed questions.

      The phrase, “Everybody hates a know-it-all.” is plangent here: the Board is composed of successful politicians who have learned (at least) one thing from business: “The Customer is always right.” “The Customer” for a politician is their constituents writ large. So they do not WANT to become Subject Matter Experts — of anything, really — because they might slip up in some meeting and get technical on a topic they know and understand.

      Better to leave to thinking to the Techbros, right?

   4. Lynnwood is a potential catastrophe. They should not have omitted the track-picking cross-overs just inbound from the platforms. Outbound side trackwork should be limited to pocket tracks. As it is now, the only available platform-picker scissors is south of the deferred station at 220th. That is about a minute from the platforms, which means that trains can only be turned by replacement of the operator. The time to walk a four-car Link train is too much of the six minutes max that a train can occupy one of the platforms when lines are at eight minute headways.

      I guess that’s why they put a scissors on the outbound side and included walkways on both tail tracks. What will finally become SOP will be the replacement operator boarding the last car in the station, riding into the tail and then taking over the train. Will the original operator ride back into the station or we they have to be forced to walk? We shall see.

   5. The turn around issue for the drivers is a very easy fix. Just have another operator enter the back of the train at Lynnwood, the original operator then exits when the train re-enters the southbound platform.

4. This is all very theoretical.
   Because three lines come from different branches, even if you perfectly time the trip in scheduling tool, 2 Line and whatever future line won’t be precisely on time. Automation won’t be able to mitigate delay at closing door. Suppose 2 Line heading to Lynnwood was delay at Judkin Park because door closing delay by 30 seconds, will this 2 Line train still be able to make it to its scheduled slot?
   So the real question isn’t how many trains can tunnel carry hourly, but how close two trains can be spaced in DSTT and how many seconds of delay can be tolerated so that 2 Line can merge to DSTT safely before next 1 Line.

   That’s also why a lot of high frequency metro lines don’t have branch or don’t have many branches.

   1. Suppose 2 Line heading to Lynnwood was delay at Judkin Park because door closing delay by 30 seconds, will this 2 Line train still be able to make it to its scheduled slot?

      At worst it means another train is delayed. But with proper signaling equipment, the delay can be spread out over the other train’s activities so it is hardly noticeable. Of course this assume they are significantly late. Right now the most the trains run is every 7 minutes 30 seconds. That means they would run every 2 minutes 30 seconds in the tunnel. If the tunnel can handle trains every 90 seconds then that gives them a full minute of leeway.

      a lot of high frequency metro lines don’t have branch or don’t have many branches.

      Yeah, but a lot of them do. What is interesting is that it is rarely a priority. The Boston Green Line comes to mind. Or the Muni Metro. In both cases they could use another line. The could also including new stations (unlike our second tunnel) but instead both agencies have focused on expansion in places that don’t help the crowding issue at all. For that matter, the New York City Subway could definitely use a lot of work to reduce the complexity and allow trains to run more independently. But for the most part, they have other priorities.

      That is what makes the second tunnel thing so weird. They are trying to build something *before* we know if this will actually be a problem. Meanwhile, in cities where it really is a known problem, they just don’t treat it as a priority. It is just weird that this somehow somehow became a “must have”. It is like buying snow tires when you live in Miami when you never leave the city. Yeah, sure, someone who know might move to West Virginia and you might want to drive up there to visit them in the winter. But plenty of people get by with all-season tires and chances are you’ll never make that drive.

      1. Link is on track to beat MUNI Metro and MBTA green by ridership and both of them run slower than Link. They are sub-system of a bigger urban rail systems while Link is the system for Seattle. It has to be more efficient than them.

         As for NYC subway, a lot of core sections shared by three lines actually have 4-tracks although some of the junctions are not fully grade-separated, so that’s not exactly the same as DSTT.

      2. Sounds like a good reason to spend money on building toward Everett and delaying the new tunnel indefinitely.
         I’ve experienced Boston, NYC, Mexico City, Shanghai, other subway/rail systems. I’m not a transp expert, but your MBTA Green Line reference feels spot on.

   2. The NYC subway has what Alon Levy called “reverse branching”, like the N and W that run together in Queens, but then split in Mahattan to two different areas.

   3. “They are trying to build something *before* we know if this will actually be a problem. ”

      You have to start constructing a second underground line ten years before the first line reaches capacity, or you’ll have a gap of overcrowding until the second one is finished. And since it takes five more years to decide to build it and vote on it, and five years of planning, you have to start the process fifteen or twenty years ahead. Otherwise you get into a situation like the DC Metro was in, where the central core was at capacity but they couldn’t build another line quickly enough.

      1. And since it takes five more years to decide to build it and vote on it, and five years of planning, you have to start the process fifteen or twenty years ahead. Otherwise you get into a situation like the DC Metro was in, where the central core was at capacity but they couldn’t build another line quickly enough.

         My point is that isn’t a big deal. Would anyone in DC prefer that they focus on potential capacity issues first and then, twenty years later, finally serve a particular suburb? Of course not. The crowding is a problem but not the worst thing.

         Look at Toronto. It has talked about a “relief line” for decades. They are finally getting around to building it. Yet if you were ask someone about transit in Toronto the first thing they would complain about it is the slow speed of the streetcars. In other words, they would much rather have additional (fast) metro coverage rather than a relief to the crowding. Or they would just like them to speed up the damn trams. Or run better buses. To be clear, I’m not saying that the Ontario Line isn’t great. But most of the value comes not from relieving congestion but from adding additional service.

      2. Subway capacity isn’t a hard line. There are several ways to manage demand to deal with overcrowding too. It’s all a matter of signals and switches and real-time dispatching.

         For example, if a train gets overly delayed at Judkins Park, the dispatcher doesn’t have to slot a delayed 2 Line train in sequence. It can send two 1 Line trains in a row; and by the 2 Line train being on the heels of a 1 Line train, it will be less crowded to boot. Or the dispatcher can run a train “express” for several stations to encourage people to get off a crowded train (”This train will now become express to U District. If you are only going to Westlake, Capitol Hill or UW please get off at the next station and wait for the next train.”) and get back on schedule. I’ve witnessed both scenarios on trains.

         This is why seasoned operations management are valuable, and why agencies hire people from other systems. An experienced dispatcher will know how to quickly adjust things in realtime, kind of like a pro football player who can adapt a play as needed.

      3. DC Metro has attempted several strategies to discourage discretionary at the peak times.

         At one point, the fares were higher during peak periods.

         And once they’ve had paid parking only at peak hours. They would make drivers pay between 3 and 8 pm to leave the garage. If they got out earlier or waited until later they could exit for free. It was a parking garage version of express lanes.

         The point being is that there are ways to use pricing to ease overcrowded trains by encouraging a few riders to travel at other times — and pricing is one tool.

      4. *Would anyone in DC prefer that they focus on potential capacity issues first and then, twenty years later, finally serve a particular suburb?*

         As someone who lived in the DC area for almost a decade (until 2016) and used Metro to commute the entire time – yes, this was what everyone wanted. No one wanted more suburban expansions, at least beyond access to Dulles. The main issues affecting rider experience were crowding, escalator reliability, and system disruptions that resulted from multiple lines sharing single points of failure like Rosslyn. And that was with three trunk line tunnels through the core of Washington. Of course Metro ridership was also 8x Link.

      5. As someone who lived in the DC area for almost a decade (until 2016) and used Metro to commute the entire time – yes, this was what everyone wanted. No one wanted more suburban expansions, at least beyond access to Dulles.

         Yes, but my point is that was *after* it had already served many suburbs. By the time it opened it was already a fairly suburban system. Yet they expanded long before they bothered to deal with crowding in the core. The plans for dealing with the crowding are also sensible. They involve adding coverage in DC proper (Georgetown mainly) and not just adding a line right next to the existing one. Yet even that won’t happen for a while, as it is deemed too expensive. They are focused on making the trains faster and more reliable along with additional bus service. The point being, I don’t think anyone (other than maybe folks in Georgetown) wishes they had built it in a different order. It is crowded, but as you mentioned, people wanted a trip to Dulles. The folks in the suburbs wanted to get to DC. I really don’t think DC Metro planners have any major regrets (again, other than not covering Georgetown). Partly that is because it is so extensive. Crowding is basically a nice problem to have.

         In contrast it is pretty to see how ST could regret a lot of their decisions. It was hard to serve First Hill but why not serve 23rd & Madison instead? That would have made for a very good connection to RapidRide G. For that matter there are just a lack of stations in general (e. g. only two in the UW). Meanwhile, Lynnwood Link doesn’t have that many riders. Maybe they should have served Aurora instead. Then there are things like light rail in the first place (even though it is essentially a metro). Or adding just enough grade-separation to keep costs really high but not enough to automate the trains or even prevent common accidents. This is all without ST3 (which is full of potential regrets).

         The point is, even if the trains are a bit crowded downtown it is highly unlikely it would reach be one of the top complaints about the system. It is understandable that is the case in DC, but that is because the system is just so much better.

   4. This is something I’ve come to appreciate about TriMet’s 3 track stations.

      Back when they were still able to operate MAX fairly frequently (blue line at 7 minute headways, and yellow, red, and green at 15 minute headways), the third track at Gateway would be a place to pause one of the red or green line trains for a somewhat longer station stop while the next blue line train got far enough ahead of it.

      So, ideally, IDS would be 3 tracks and at least 2 (or perhaps 3 or 4) platforms. In the minimalist scenario, from west to east, you’d have:

      ========== Southbound Track =============//==
      [][][][][][][]Platform[][][][][][][][][][][]//
      ==========Center Track================<
      [][][][][][][]Platform[][][][][][][][][][][]\\
      ==========Northbound Track=============\\==

      When a train arrives from Line 2, it goes on the center track. That way, if there is a line 1 train still at the platform or without enough space ahead, the Line 2 train would wait at the platform until there was sufficient space ahead of it.

      Southbound, it doesn’t matter, as the spacing between trains is already established. The junction break happens after they’ve left the platform.

      Unfortunately, they don’t have the space for that unless there’s a complete rebuilding of CID station.

      What it looks like they DO have enough space for is a station on the ramp entering the tunnel from the old freeway right of way. It would not be a level station, but most ground level stations have some slope to them anyway. It can’t be too steep, but it doesn’t have to be 100% level. As an example of a non-level station, take a look at Park Avenue MAX Station. It looks level if looking at it straight on from the street view, but if you look at the parking garage and how the bottom floor disappears into the hill, you can see that both the road and the station are sloped at about a 3% grade.

      So, if you absolutely had to, you could put a “North Stadium” station or some similar on the ramp, so that 2 Line trains could wait at a platform before joining the main line at CID.

      However, considering the length of the approach and junction inside the tunnel just south of CID, and the ramp length, as a practical matter there’s probably no real issue with having trains wait on the ramp.

      1. That was supposed to have preformstted text displayed with tags forcing the monotype font but apparently that doesn’t work.

      2. Yeah Atlanta does that too. Usually if a system was designed to run Y-shape service, they built the junction station (like CID to Link) with a third platform.

   5. Manhattan is full of three-line subway stations that branch at each end. It’s the busiest in the US. Look at Lines 1/2/3. There’s also 4/5/6 and A/B/C.

      DC Metro finally figured out to vary Silver Line end stations in Maryland for the Blue/ Orange/ Silver triple line operations.

      1. You gotta be specific here which section you are referring to. A lot of Manhattan’s subway sections have four tracks that serve three lines. Two express train lines share the express tracks and one local train has its own tracks. So it is not that bad.

      2. Manhattan has 4 tracks.

         It’s A/C/E, B/D/M/F

      3. I usually used the 8th Avenue services (A/C/E) because I assumed three lines would be the most frequent, and avoided the number lines because they were the only line in their tunnel, but I later read that the numbered lines tend to be more frequent than the lettered lines, so I wonder if that makes up for it.

      4. Mike,

         A/C/E also has 4 tracks up till Central Park west. Then it mixed with IND Sixth Avenue Line B/D. That’s still 2 lines per track.

      5. Not sure about frequency difference but I suppose that’s possible because IRT lines (numbered lines) feature smaller train cars (shorter both in width and length) compared to lettered lines.
         The most noticeable difference is that letter lines train (like A/C/E) have 4 door on each side while numbered lines only have three doors. So that’s possible they make them run more frequency to make up for the capacity limitation.

   6. Two things. First, platform doors bring the occurrence of door holding way down. Close the platform doors five seconds before the train doors and you have “won”. Platform doors don’t have to “bounce back” until the final three feet where they do in order to avoid crushing folks or breaking bones.

      Second, CBTC can “straighten out” conflicts at merging junctions. It simply chooses one of the two trains conflicting to be subordinated and slows it down so that it doesn’t run into a red light.

      1. Tom,

         Yes CBTC will improve the situation, but nobody know by how much it can do and if that’s enough to run three service patterns each with 8-minute headway on the same track. The 2-minute frequency standard for CBTC will be compromised by some unique Link features because this is not really a metro system like how slow 2 Line train can run merging into Central Link.

         A similar case can be found in BART in bay area where Oakland Wye rather than train control signal is the bottleneck of the entire thing. This is the kind of bottleneck signal upgrade and full automation won’t even fix.
         Oakland_Wye

         CBTC can only do so much. Otherwise many systems won’t even need double track the entire mainline. I think a lot of people recognize the construction and design issue of a second tunnel but downplay the challenge of operating capacity. They hope that more advanced signal control and automation level will solve everything. Well it won’t and maybe things are just as complicated as building a second tunnel but right now we just do better job quantifying the challenge on the construction side.

         Elizabeth Line in London is designed to run 24 trains per hour. It features very modern version of CBTC with GoA2 level automation and platform screen door Link will never be able to catch up. Its operation will still fail during extreme scenario and needs some kind of human interference because its full capacity gives very little tolerance for delay.
         I was there two years ago days before Christmas. During the evening peak, they pushed headway between trains to less than 30 seconds but all the trains were held before station multiple times and all the local stops were skipped. This is a system with both things you mentioned and perhaps above and beyond.

5. Over the course of the first 1 minute 45 seconds of this video, 2 São Paulo metro westbound Line 3 trains depart Tatuape.

   In the event anyone wants visual representation of what this level of frequency looks like.
   https://youtu.be/3b9mUQdHeEs

6. One situation where the antiquated signaling system in the DSTT was very apparent was back when the trains were sharing the tunnel with buses. Buses operated normally and stopped behind one another and opened their doors. But, the train had to follow a much strictly policy, where the front of the train could not enter a station until the back of the bus in front of it had completely exited it, and a bus behind the train could not enter a station until the back of the train in front of it had completely exited it.

   This not only created delay by making gaps between trains and buses much larger than between buses and other buses. It also meant that, whenever the train caught up to a bus in front of it (very common, as the train loads passengers much more quickly than a bus does), passengers on the train were forbidden from transferring to the bus the train had caught up to, and were, instead, required to wait as long as 15-30 minutes for another bus (and the downtown tunnel is an awful place for a long wait, with almost nowhere to sit and constant, blaring announcements). And, with the trains used at the time just 1-2 cars long, none of it was even physically necessary, as the platforms were plenty long enough for a train and 1-2 buses to stop alongside each other with the doors open. The only thing preventing it was the signal policy.

7. “Back when they were still able to operate MAX fairly frequently (blue line at 7 minute headways, and yellow, red, and green at 15 minute headways),”

   MAX frequency went down? Why?

   1. Not enough passengers, operators or finances.

      I’m not entirely certain, but I think work from home took away a bunch of the payroll tax.

8. “This train will now become express to U District.”

   Has Link ever done this?

   1. You can’t just do it mid way as some riders may count on getting off at the skipped stations. The only way to do so would be to ask those riders to get off the train to wait for the next one.

      1. @ Martin:

         Yes it was an inboard announcement. It helped that there was a train tailing this one.

         It was 4:30 pm in Chicago. The train was at crush load so the doors took a long time at each station before they could be closed.

         Then the announcement was made. About 40% of the riders got off at one last extended dwell. The we zoomed past several platforms and I got to O’Hare much faster than I was expecting.

      2. To add another fun example of this: I’ve seen trains go express mid-way through a trip on the MBTA Green Line many times. The operator makes an announcement, and at least once I recall them explicitly saying it was for “headway adjustment.” It usually seemed like it was for that reason—to correct bunching. This would happen even without crush loads and long dwells. The lack of anything but wayside signaling, the maintenance problems, and the significant amount of branches that merge into ~1 tunnel all seem like possible explanations.

         In any case, the passenger experience wasn’t always so great if you missed the announcement or couldn’t make out which stops they were skipping. The PA system on those old Kinki Sharyo “Type 7” LRVs wasn’t particularly clear as I recall.

   2. My anecdotal observation is that, even out-of-service Link trains still stop at every station, albeit, without opening the doors for passengers. This seems to suggest that every train stopping at every station is a deeply ingrained assumption in Link’s signaling system, which would have updated to allow a train that is running behind schedule to skip stations to make up time.

      1. I believe that Chicago train that I was on may have momentarily stopped at each station — but the driver could keep the doors from opening. The time savings was in reducing the long dwells at each station because of the crush loads. What is normally a quick 20-30 seconds when doors could be opened and shut was probably over 2 minutes at each platform. That adds up, even if it’s just a few stations.

         It was 35 years ago so my memory is a bit fuzzy.

   3. I still do think we should’ve have built our system with parallel tracks in consideration.

      Express trains (that skip lower ridership stops that are too close to each other) are an excellent addition to any transit system, and speed up people’s trips especially if it is timed right for people to jump between trains.

9. Minor mathematical error. A train every 8 minutes is 7.5 trains/hour (15 trains every 2 hours), such that three lines in the tunnel each with 8 minute frequencies would mean 22.5 trains/hour. This would mean a 2m 40s train frequency in the tunnel, and be 2.5 trains over their 20/hour limit.

   1. Clockface scheduling is important when a transit route isn’t frequent, like every 20 or 30 minutes. But it’s much less important when trains can run every 10 or 12 minutes. It’s mathematically possible to schedule a train line to run 6.67 times an hour (or 20 trains a hour with three lines) for example. It doesn’t have to be an integer or an even number.

      1. For sure. Also, 6.67 trains/hour is just a train every 9 minutes, so it would be quite easy to schedule. That’s actually good point, even limited to ST’s current tunnel capacity ceiling (20 trains/hour), you can interline all three while taking only a 1 minute frequency hit.

   2. I rounded 7.5 to 8.

10. For sure. Also, 6.67 trains/hour is just a train every 9 minutes, so it would be quite easy to schedule. That’s actually good point, even limited to ST’s current tunnel capacity ceiling (20 trains/hour), you can interline all three while taking only a 1 minute frequency hit.

Comments are closed.