Aunty’s Wheel

Don’t get me wrong, I love good ridership news as much as the next transit nerd, but at some point the wheel has to slow down. July’s Link weekday ridership was 16.9% higher than July’s of 2013. 16.9%. And July 2013 was no weak month, it was 10.7% percent higher higher than July 2012.

I’m going to say December. Looking at my charts, December of last year was when Link’s ridership growth really started spiking. Up until that point growth was pretty consistently averaging between 10 and 11 percent. Since then it’s been 5 points higher. What’s your prediction?

July’s Central Link Weekday/Saturday/Sunday average boardings were 37,354 / 32,873 / 27,135, growth of 16.9%, 0.1%, and 13.6% respectively over June 2013. Similar to June Saturday was low, however the weekend average still increased as a whole. Sounder’s weekday boardings were up 8.9% with ridership increasing on both lines. Tacoma Link’s weekday ridership decreased 3.3%. Weekday ST Express ridership was up 5.7%, with 22 out of 25 routes showing growth. System wide weekday boardings were up 9.2%, and all boardings were up 8.8%. The complete July Ridership Summary is here.

My charts below the fold.

102 Replies to “July 2014 Sound Transit Ridership Report – Where Does it Stop?”

  1. The glass is too small.

    Did anybody notice the ridership predictions from page 13 of Wednesday’s Transit Integration Report?

    Expected U Link ridership: About 35,000 weekday boardings by 2018 (65,000 on entire line)

    That means the existing portion of the line is expected to contribute 30,000 daily boardings, or less. Um, wait. We’re already there. On weekends. Weekdays have blown way past that number.

    Although this is all-day data, not peak-of-peak data, I sense a capacity problem during the peak-of-peak hour looming, especially when the 2014 Service Improvement Plan (which didn’t contemplate this year’s huge Link ridership growth) has Link close to capacity in 2016 and over capacity by 2019, even with the increase from 16 to 20 Light Rail Vehicles per direction per hour during peak.

    Does ST need to rethink the service plan for U-Link, and find a way to push more LRVs per peak hour through the transit tunnel?

    1. are you factoring in 3/4 car trains? or are your calculations based on the existing 2-car trains that are operated today

      1. I’m following ST’s Service Improvement Plan, which is repeated in the Transit Integration Report: 2-car trains every 6 minutes.

    2. “Over capacity”–a good problem to have!!! :) Tough choices will have to be made, but better integration can make this all a moot point in the end. Bring on the over capacity!

    3. Actually that isn’t the problem – the modeling from the FTA that’s used to project ridership is at fault. And it is done intentionally to make transit projects appear less worthwhile than say road widening programs. But when the truth comes out, the FTA, transit agency heads & politicos get to make big speeches & wave palm palms.

    4. It’s time to go back and look at the original projections for U-Link/Central Link (back when U-Link was going to be built first. You’re going to meet those projections. (I don’t remember exactly what they were.)

      Projecting from the current trends, U-Link + Central Link will probably have 85K+ weekday passengers soon after opening.

      U-Link will need to run 3-car trains all day from the moment it opens. At least.

      Sean is correct: FTA policies routinely sandbag transit lines by requiring them to underestimate ridership (badly), in order to prevent them from getting funded. (Meanwhile, estimates for number of cars on proposed new roads are always grossly overestimated, sometimes by factors of 4 or 5).

  2. but at some point the wheel has to slow down. …. Up until that point growth was pretty consistently averaging between 10 and 11 percent. Since then it’s been 5 points higher. What’s your prediction?

    The growth rate will depend a lot on how well the next round of bus route restructures are planned to feed into Link. There should be decent growth along the line in at least a few places, but to really have substantial growth the ridership needs to be drawn from a large area, and that ridership needs to be directed to Link not just because there happens to be light rail there, but because the combined bus + light rail mixture is superior to the bus service that was there before.

  3. During the PM peak, two-car trains are frequently full, as in crush loads. OK, not white-glove-packed-on crush loads, but loads that get annoying even for regular riders, and a real struggle for someone going to the airport with luggage. There’s no room for growth next year on those overcrowded trips, and such trips will continue to grow in number..

    I think it’s time for Sound Transit to take down that demising wall at the end of the Pine St. stub tunnel and allow Central Link to operate with three-car trains when necessary. The ULink fleet is here already, so the cars are available. We don’t need to wait until the day ULink opens to begin operating longer trains.

    1. The trains can run closer together, right? We’re not at the 4-minute headway limit imposed by ST, are we?

      I agree that running 3 or 4 car trains would be more efficient, but shorter headways would also improve the experience, possibly increasing ridership even further.

      1. more trains = more operators

        Yeah … I don’t understand why they can’t activate 1-2 cars worth of track past the Pine St Interlocking to allow for 3/4 car trains now at rush hour and for events in the city

    2. Roger,

      My impression is that what is holding ST back from doing as you say and temporarily configuring for 3 car trains in the stub tunnel is that they don’t want to complicate the SCADA cut-over for U-Link.

      That is their biggest concern with U-Link schedule right now and it is unclear what stub tunnel reconfiguring would mean per the current SCADA system and the eventual activation, testing and cut-over to the complete SCADA system. So I think they are being cautious and are only willing to touch the system the minimum number of times.

      And it would still take time to reconfigure the stub tunnel. In the end we would probably only get 9 months to a year of use out of it before U-Link opens and we get to max out at 4 car trains anyhow. In the mean time it would seem that going from 7.5 min to 6 min headways would probably handle the demand — if the DSTT can support 6 min headways under joint ops.

      1. The SCADA stuff is over my head, so I can’t comment on that. For this layman, I can’t grasp why activating another 100 feet of track is such a big deal. It wouldn’t even need to be powered.

        As for moving to 6-minutes headways, v. current 7.5, that should certainly be possible (it’s what ST kept promising up until a year or two before Opening Day) but what does it do to current DSTT bus ops?

        ST announced it would operate shorter headways on Seahawks opening day; how did that go? How short were they able to press the headways? Were there any problems created by the tighter headways? Inquiring minds want to know.

      2. Lazarus nailed it. The wall comes down when the signal system is installed, cut over, and ready for integrated testing.

        U-Link opening will force ST to contend with the need for 3-car trains early on.

        And where does it stop? It doesn’t. 35K today; 300K +/- within a decade.

      3. Yep, the Canada Line is at 140k/day at 4-minute headways just 5 years in, and that’s with a poor SkyTrain transfer design.

      4. And I recently read the ST monthly progress report on U Link and North Link. Their SCADA project timeline appears to be both called out as critcal and fixing it is overdue.

        FYI – they appear to have completely removed the float on U-Link and appear to be firmly planning for a 1Q2016 opening.

      5. Zach, railcan,

        The Canada line has lots of local stations and serves an urban area from start to finish at YVR. It’s extremely unlikely that Link as currently authorized will ever reach a daily ridership of 140,000. Three hundred thousand is ridiculous without in-city lines with closely spaced stations.

        Ballard-UW and some other cross-town just north and east of downtown might do the job. But the suburbs will never get Link there. They just don’t like it.

        Commanism! Commanism, I say!

      6. The Canada Line has several things in its favor of course, automation, full grade-separation, 4-5 minute headways, aggressive bus restructures to feed it, etc. But the biggest difference I think is the high levels of density they achieve in their suburbs, which is generally a product of Vancouver’s NIMBY inability to densify outside the core. I lived in a tiny cluster of density near UBC in West Point Grey, and from my 11th-floor perch it was all millionaire Craftsmans, forest preserves, and golf courses. The Cambie stretch of the Canada Line is only now starting to densify to respectable levels.

        But I think Link will crack 100k with its current authorization, but the only way to get much past that is to build those ‘in-city lines with closely spaced stations’. That’s what most of us are pushing for these days anyway, with comparatively lesser-patronized suburban extensions as the political price to get the urban rail we need.

      7. Here’s a question: Will the turn-around at the UW Station be a problem until 2023? Will there be enough tail track to take a train out of service and turn it around, or will the trains be reversed like they are at the Airport Station today? I’ve often witnessed slight train delays getting to the Airport Station because there are trains waiting on each platform and we have to wait until one pulls out. The turn-around time (and train delays associated with it) will affect what minimum headways can be once both U-Link and Northgate Link are operational. All in all, longer trains are going to be the best strategy for combating overcrowding on Central Link as opposed to more frequent trains.

      8. “high levels of density they achieve in their suburbs, which is generally a product of Vancouver’s NIMBY inability to densify outside the core.”

        What does this mean? You mean that NIMBYs have no power in the suburbs? But they don’t seem to have power in much of Vancouver either.

        “I lived in a tiny cluster of density near UBC in West Point Grey, and from my 11th-floor perch it was all millionaire Craftsmans, forest preserves, and golf courses”

        That sounds like Seattle. So again, what’s the difference? San Francisco has strong NIMBYs in almost the entire city and all the suburbs. Are you saying Vancouver is the opposite?

        I think it’s more that Canada and Europe give their governments more ability to think regionally and get things done, whereas ours are more balkanized.

    3. OK, not white-glove-packed-on crush loads, but loads that get annoying even for regular riders.

      Sorry, no. Not even that.

      I’ve seen the peak. Having to stand for a few stops, think for two seconds about positioning yourself at an angle that won’t block the aisle, and be generally aware of your fellow passengers — though all anathema to Seattle’s Me-ist culture — does not equate to the kind of fullness you imply.

      The only time those trains come close to being truly “full” is immediately after sportsball event.

      Central Link riders will be fine for the foreseeable future. I don’t know about 2-car U-Link riders.

      1. As someone who rides Link every day, I can say that d.p’s experience does not convey the fragility of the current state of things. Any sort of delay (widening headways) in the tunnel or special event aggravates flow inside the train car, and occasionally actually means people can’t board.

        And forget about bringing a bike on in the peak anywhere below Westlake, even if you’re willing to hold it.

      2. @d.p.

        Take a look at page 17 of ST’s 2014 Service implementation plan. The average max passenger load on southbound Link trips was spiking to 250 passengers during the PM peak. The report doesn’t say exactly when the data was from, but it seems to be at least a year old, so I’d expect current peak max loads going southbound to be approaching 300 passengers.

      3. And yet somehow there are never more than six human beings in the raised sections, there’s never anyone standing in the articulated part, there’s always one door with oodles of space by it, and so on…

        I guess it’s sort of like the claims that RapidRide is “crush-loaded” in the peak. But only really when the bunching gets terrible. And somehow I can always manage to find my way on with the slightest flex of my elbows and vocal chords.

        With 150 seats and supposed standing room for 250 on a 2-car train, PhillipG’s numbers suggest a 50%-60% average capacity, which supports your admission that this is an operational problem, and not an especially acute one.

        Also, since when does the primary purpose of mass-transport capacity utilization include the ability to drag 6-foot-by-4-foot slabs of metal and exposed grease everywhere you go at the height of rush hour? Cities the world over ban bikes from their highest-load services at peak, because scalability and minimized boarding times and internal circulation matter when you’re trying to move lots of people around efficiently.

        Maybe it’s time to de-Portlandia our expectations of how we share public resources in a our growing city.

      4. Your ability to divine what “always” or “never” happens is astonishing when arguing with people who use the system on a daily basis.

        There will be a time to ban bikes from Link during the peak, but not when we could easily double capacity by bringing forward some spending on a project that is massively under-running. I’m not sure what you think I’m “admitting” in stating it’s an operational problem. The point is running longer trains would solve a bunch of issues.

      5. Why? For further evidence that Seattleites are idiots about spatial allocation? I already know that.

        I just love that the Seattle solution to identified problems always involves either massive overcompensation (often requiring gobs of nonexistent money), or throwing up hands in resignation and doing nothing. It never seems to involve slight common-sense modifications or remotely encouraging the human capacity for adaptive behavior.

      6. I think slight common-sense modifications are exactly what people are recommending. That is, adding another train car (or two). Not that I want to knock down the wall. I’m no engineer, but it seems to me like we can wait a few more months. But I can understand the desire, and if it only costs a little, we might even make up for it with increased ridership (more pleasant ride as well as good publicity).

        Look, we are all aware of the countless mistakes that Link made when building this system. But guess what? It is surprisingly popular. Not as popular as it should be, given the money we spent (and are spending) but still, it is surprisingly popular (at least to me). The numbers keep growing even though we haven’t built the most important line yet (UW-Capitol Hill-Downtown). It’s possible that like “ethnic” food back in the 70s, Seattle residents are dying for good rapid transit, and will take what they can get, warts and all. Perhaps a better example is soccer (one of the more popular sportsball* events). The Sounders set all sorts of attendance records despite the lack of a championship and despite the fact that their league is equivalent to a second or third division league in Europe. But we are gaga over soccer (our numbers eclipse other cites). Maybe light rail is like that, which is why Sound Transit better prepare for it, when the first big line opens (in early 2016). Adding more trains will help, but increasing headways is even more important. I would say it is important on its own right (to improve the quality of the overall transit system) but if we have to use the “trains are too crowded” excuse to increase frequency, then I’m all for it.

        * Sorry to tease you about “sportsball”. I really like the term, and will use it amongst friends, especially the guys who don’t like sports.

      7. @d.p.

        A couple points:

        1) As I said, the average max load figure I quoted appears to be 1-2 years old- assuming that peak ridership has grown at the same rate as overall ridership, the current average max load should be about 15-25% higher- now around 288-313 riders. If we have another year of greater than 10%, we’d expect an average max load of around 350 peak Southbound riders even before U-Link opens.

        2) The 250 max load is an average. Assuming the variance is greater than zero, that means that some days, there were more than 250 riders on the train at some point.

        3) I’m skeptical that you can fit 200 people in a Link car under any but the most idealized conditions- in the real world, with luggage, bags, bikes, strollers, wheelchairs, etc, the max capacity will be somewhat less. Can some of this be improved by adaptive behavior? Sure, but even if everyone learns to pack themselves optimally, you don’t allow bikes at, peak, etc, when ridership is growing 10-15% YoY, the capacity increase that adaptive behavior buys you will be swamped by another year of ridership growth.

      8. I’m with Ross.

        Sound Transit should absolutely be planning to run 3-car trains (at minimum) from the moment U-Link opens, and if they really still aren’t, that represents a failure of foresight on par with the junking of the Montlake vent, the insane UW Station access architecture, and the refusal to future-proof for future urban expansions and connections anywhere.

        But none of this justifies the “Central Link has surpassed ¼ of Canada Line ridership, my god we’re running out of room!” Chicken-Littling in this thread. The trains are not actually turning away passengers. U-Link is 16 months away. We’re fine until then.

        Want a cheap common-sense modification? Make the upper sections of the cars 2×1 seating, and you’ll about sextuple the number of people who can comfortably choose that space for standing room. Even better, make both sides inward-facing bench seats. Those “inner” seats are so uncomfortably wedged-in anyway that virtually no one uses them during the 15 hours every day when the trains still run with excess seats.

      9. @d.p.

        Sound Transit should absolutely be planning to run 3-car trains (at minimum) from the moment U-Link opens,

        Unfortunately the 2014 SIP only forsees running 2 car trains at peak (see Table 5 on page 106). ST seems aware that this will be a problem, since the same table shows demand exceeding capacity starting 2018.

        Make the upper sections of the cars 2×1 seating

        It would be fantastic if ST converted all the 2+2 seating in existing vehicles to 2+1, and ordered all future vehicles only in 2+1 and bench configurations. 2+2 is sort-of OK when there are seats for everyone, but gets in the way once it’s standing-room only.

      10. I wonder if luggage racks could help with capacity. Yes, they’d take away a few seats, but as long as the train serves the airport, it will always be carrying passengers with luggage. Without luggage racks, suitcases inevitably block seats or standing room that could be used to squeeze in a few more passengers. (The theory that people can just sit for 40 minutes with a large suitcase on their lap is great on paper, but just doesn’t work in practice).

        I rode the Silver Line bus in Boston once, and the luggage racks worked quite well. While the bus was packed, it would have been a lot worse if everybody had to hold on to their suitcases, rather than set them down on the rack.

    4. Is there enough space to break a four car train into two two car trains for only the distance required to get each section reversed? You would have to station an operator down there for the duration of each peak, in order to operate the second half of each train, but that’s cheaper than having two car trains run the length of the line at increased frequency.

      TriMet used to break and join MAX trains in its stations quite a lot until they decided a bit of extra electricity is cheaper than increasing and decreasing train length. The fully automatic Scharfenberg coupler is pretty quick at coupling and uncoupling, and they would typically pull a single car train into the station at Ruby Junction, couple on to the back of it with a second car, and the train would then be on its way as a two car train. I was on a few where they did this, and the bump that you feel when the coupling happens really isn’t noticeable at all, and the delay was barely more than a standard station stop.

      1. Glenn,

        As usual you have an excellent idea. It seems like with eight-minute headways this could be managed. Six minutes might be a little tight but even that should work.

      2. Glenn – completely awesome. I never thought of this. We’re going to follow up with Sound Transit on this.

      3. Come to think of it, TriMet may have had to have one staff at Ruby Junction when they did this: one at the station platform to signal how close they were to coupling, and the operator to run the car exiting Ruby Junction shops into the rear of the train at the station.

        One possible problem is that the Link trains have a cover over their couplers, while the Bombardier MAX cars have everything open all the time.

      4. Gold star for Glenn!

        I’m not sure, but I don’t think there is enough room in the stub tunnel to pull this off, although ST has had substantial experience coupling 1-car consists into 2-car trains, and decoupling them in the stub tunnel.

        Consider this as an emergency plan for a day when Link’s capacity is expected to be overwhelmed: De-couple a 4-car train around at the Westlake northbound platform, reverse each segment, and re-couple at the other platform.

        That, and start working on BN&SF to get permission to run the Sounder fleet all day at minimum headway, if we know what that overwhelmed-capacity date would be.

        Pull that off, and the extra 50,000 voters who ride ST only once before the ST3 vote will be duly impressed. Or maybe some will become regular riders.

      5. Brent — There is very likely space to do this in the part of the stub that isn’t the turn around but, even if they can’t do it in the stub they could do it on the platform. The problem (if there is one) will definitely be about decoupling and re-coupling the trains (which I know nothing about) or staffing (which shouldn’t be a real problem.) They might also have operational problems with running the longer trains in the tunnel while buses are still there.

        If they can do 3xs during afternoon peak and 4xs on days likes Seahawks day 1&2 you are absolutely right — occasional riders will be more apt to vote yes. Right now rare transit riders tend to only ride on the worst possible transit days — its quite the cycle. Generally, I tend to think a yes on ST3 is all about getting a package with shiny things for each subarea to voters (Seattle and the Eastside in particular because they can pull the rest of the region) and that it will win easily. That said, more yes votes would be nice.

      6. As long as you don’t couple on a curve, so that the couplers are misaligned, coupling and uncoupling the Scharfenberg coupler is pretty easy and quick. It is a true complete automatic coupler, and was designed for this type of thing.

        The Japanese Shinkansen are able to do this at their stations in about 30 seconds:
        http://www.youtube.com/watch?v=RbWet68jZgk
        Too bad YouTube wasn’t around in 1996 or so. I think that is the last year I remember TriMet doing this process at Ruby Junction. After that, MAX popularity was high enough it was no longer worthwhile creating single car trains for mid-day service, as the low demand period was really only several hours long. The S70 cars were ordered with a cab at one end only and passenger seating filling the other end, as if there is ever a need for single car trains they can do that with any of the dozens of first and second generation Siemens cars.

        I don’t imagine Link has the neat automatically retracting covers over the couplers that those trains have though. Also, I would expect Link to have some sort of required safety test before a coupled train would be placed in service.

        I would imagine that there is a double crossover between the northbound and southbound lines?

        | |
        |\ /|
        | x |
        |/ \|
        | |

        If so, here is what I picture happening:

        1. Four car train enters Westlake. Passengers get off.

        2. Four crew board, and original driver heads for break. One crew boards each car. The coupler covers on the front and rear are opened.

        3. Train pulls forward so that it occupies the next signal block, releasing the station block for the buses (my hope is that they used signal blocks that are fairly short).

        4. The uncoupling process seen in the video happens. You don’t need a spotter for the uncoupling.

        5. The first two cars run into the northbound stub end track.

        6. The left facing crossover is then activated, so the next two cars have a diverging signal, and the signal will then be clear for them to move forward into the stub end track that is on the southbound track.

        7. Remember that we boarded four operators: one in each car. This means that the operator does not need to get out on the platform and then walk to the other end of the train. Instead, there is an operator at each end of each two car train. So, the northbound to southbound crossover is then activated, and the lead two cars run from the northbound track to the southbound track, and stop just after the southern switch. They should be able to position this with some precision, since there is an operator at the rear of the train to tell the actual train operator at the other end when to stop.

        This also frees up the northbound stub end for the next two car train.

        8. The last two cars pull forward until they couple to the first two cars. The operator that was in the rear of the first two cars should be able to get out and provide the required distance clearance hand signal to the approaching train. The space around the diverging track for the crossover should provide enough side clearance so this move can be done safely.

        9. Whatever tests need to be done to make sure that the resulting train is safe to operate at the start of the run can now happen.

        10. The train pulls forward into Westlake station. The actual train operator takes over from there, and while the doors are open and the passengers are boarding, the front end and rear end maintenance operator close the coupler covers at the front and rear.

        11. Sadly, there is no center platform, so the maintenance operators now have to go back up to the mezzanine and cross over to the northbound platform.

        The reason why I make a distinction between the operators for this move and the regular train operators is that these may have to be shop staff that do this, as official train operators may not be allowed to work with the couplers and their covers. Shop staff would be able to move cars as part of maintenance and other out of service moves, but not move them in revenue service – or at least as best as I can tell that is how the TriMet operating rules work.

      7. @Glenn,

        I’m normally very pro-union, but if there is any chance that it would not agree to a temporary codicil to the work rules to allow four shop staff at a time to make this train-breaking-reassembling operation at Westlake for a few months they are stupid beyond belief. The right wingers would trumpet this all over the region as a club against ST3.

        How does a flyer saying “Selfish Union Work Rules Make Your Daily Commute Harder” sound?

      8. I don’t think that they would have a union problem either way, since the moves involved here don’t actually carry any passengers. As non-revenue moves the mechanical staff could probably do the moves.

        However, from the sounds of it, they have been making and breaking trains from one to two cars in the stub tunnel already?

        If that is the case then they already have a procedure in place.

  4. I like how we’ve exactly matched weekend boardings on the opening month. There’s something comforting about a chart with the two endpoints touching,

    1. It’s actually 99 boardings short of the opening month weekend average, but I think that is well within the margin of counting error.

    2. It’s fascinating to see how busy the line is on the weekend. Most routes take a huge cut in ridership on weekends and that just isn’t happening on frequent, fast, and comfortable service…Something to that?

  5. All this sure seems to point to 3-car (or even 4-car?) runs once U-Link is open. It can’t come soon enough, though.

    1. Does ST even have enough cars to routinely run 3-car trains when U-Link opens? (ST will need to do so… but does that mean they need to order some more cars?)

  6. It’s getting faster too… The 12 month rolling average is up 14% from a year ago. Last year’s 12 month average increased 11% from the previous year.

    This is weekday boardings, so it shouldn’t be thrown off by the Seahawks. I can imagine needing to decrease headways AND cut down the demising wall so they can run bigger trains if it continues like this. By the time they open U-link, in March or April 2016, they could average 40k during that month. crazy.

    1. Ok, I guess it wouldn’t be necessary to run more frequently if they doubled the size of the trains. That’s obvious.

      1. I sense reluctance on ST’s part to reduce frequency from the current plan, for fear Metro will stuff more buses into the tunnel, and make it even slower than it is today.

        The solution may be a minimum performance standards agreement, in which, ST agrees to run longer trains at current frequency, and Metro agrees to move more buses upstairs if a certain threshold for run-time through the tunnel can’t be met consistently during the peak-of-peak hour.

      2. It wouldn’t be necessary, but more frequent service helps everyone. That is a huge part of the whole bus-train interaction thing that we’ve been talking about lately. If the train can come often enough, then a typical user can combine a bus and a train ride and not pay a transfer penalty. That is way it works in big cities (Toronto has two minute headways). That is the way any urban transit system is supposed to work. A significant number of your trips will be transfers (whether train to train or train to bus) and the transfer time penalty will be minimal. A system that does otherwise is more like commuter rail: sending people into the city several times a day.

      3. That works well for inbound trips. Unfortunately, for outbound trips, a frequent train isn’t enough. As long as the bus you have to connect to is still running only every half-hour, the connection will still be difficult. Essentially, you would have to plan for the worst-case train delay to determine how much time you need to allow, every single day.

      4. @asdf — I’m not going to defend half hour schedules. That is terrible, and everyone knows it. But at the same time, just improving the train frequency makes a huge difference.

        I think the key is both frequency as well as reliability. I’m not sure what the “worse case train delay” is. The whole idea of grade separated (or mostly grade separated) light rail is that there are no delays. The trains run consistently (unlike buses). There might be an unusual situation that delays the train, but those should be extremely rare (and can happen with any transportation system — including a one seat ride).

        For our system, the weak link is the bus, both because it is infrequent, and because it is unreliable. If the bus happens to be reliable then the transfer (either direction) is easy. For example, let’s say it is 2021 and I want to take a bus from downtown to Lake City. The old 73 has been replaced by a bus that starts in the U-District (similar to the 373). From downtown, it takes me ten minutes to get to the U-District. It runs every three minutes, and I figure it takes me two minutes to walk to the other bus stop. So, basically, the train ride takes 15 minutes (at most). So, if the 73 leaves at 3:30 PM, I need to be at the downtown train station at 3:15. If I get there just as a train is leaving, I know it will be tight, but work. On the other hand, if the train leaves every ten minutes, I either allow for that extra time (meaning the train trip now takes 25 minutes) or I need to check the train schedule. If the train schedule is favorable, then it is not a problem. But quite often, it isn’t, which means I either have to hustle to catch the bus, or wait around a long time. To make matters worse, either way I have to catch “that train”. This means that unlike the three minute train, I have to time my walk carefully. I can’t afford to get there right when a train is leaving.

        Keep in mind, that if one system is infrequent, it is much harder to justify changing the other system. For example, If the train runs every ten minutes from downtown to the U-District, it will be really hard, politically, to adjust the 73 so that it doesn’t go downtown. Downtown riders to the north end (areas north of the U-District) will hate the transfer, because it will take so much time. This is all for a simple transfer (two seat ride). It gets much worse if you try to justify three seat rides. For example, in 2025, I think it would make a lot of sense to ride the 41 from Lake City to 125th, take the train, then take a bus from the U-District to Fremont. That would actually be a fairly fast way to get to Fremont from Lake City (faster than driving during rush hour). But that only works if the transfers are small. If nothing else, we can minimize the time spent waiting for the train, which then means you are essentially looking at matching one bus with another. Having that fast, frequent, reliable “spine” means that it will be much easier to change the routes on the buses, making them more frequent as well as more reliable.

      5. The obvious solution to this problem is to simply show the connecting train times in the bus timetable, so people know when they must be in the train station in order to make the connection. Metro already does this with some bus routes. For example:

        22: http://metro.kingcounty.gov/schedules/022/s0.html
        65: http://metro.kingcounty.gov/schedules/065/s0.html
        346: http://metro.kingcounty.gov/schedules/346/s0.html
        931: http://metro.kingcounty.gov/schedules/931/s0.html

        An example from Perth, Australia: http://members.iinet.net.au/~jbbrown/Eval-Tools-Paper/www.transperth.wa.gov.au/timetablePDFs/Western%2038%2020110911.pdf

        Of course, this is a bit more tricky if the bus does not terminate at the Link station, which means that connections will be taking place in multiple directions. (Also, not everyone connecting to Link is going to Downtown.) In that case, just publishing Link schedules and possibly special timetables for the most important connections would help riders understand what time they need to be at the Link station in order to make the connection.

      6. Trips are always dependent on their weakest link, which usually means a half-hourly bus in the train+bus direction. However, feeders will be more reliable than the existing routes. If you look at the 73, its bottlenecks are the University Bridge, Eastlake, Denny Way, and the DSTT. All those go away. Likewise, the 30 was unreliable when the current 32 was part of it, because of the Fremont congestion and for a while it was rerouted on Aurora. All that went away when it was split. Likewiswe, I have high hopes for the 36/49 route because even though Broadway backs up, it doesn’t slow down nearly as much as 3rd Avenue or Jackson Street, and if it does stay on 12th until Denny that’ll be even more reliable. In north Seattle, pretty much any feeder will be reliable except those on 45th and Northgate Way.

      7. When people get on an elevator, they do not worry if that elevator will arrive in time to get their connection.

        Once you start running Link trains often enough, this same effect should take place as there is no significant delay. In fact, once the tunnel becomes train only the speed of the trains should be able to increase, so that ultimately faster speeds + high frequency = less time required than a one seat bus ride with the bus in the tunnel.

        Obviously various things would have to happen first for this to work well.

      8. The transfer experience has grown a lot more tolerable in recent years, at least for those with smartphones. Simply being able to whip out your phone and know immediately whether you made your connection or missed it makes a huge difference. That and Being able to quickly summon a ride on Lyft, Uber, or Car2Go on those occasions when the connection actually is missed.

  7. We can expect that ST Express will be the big winner in the coming months. With Metro suburban service about to take a big hit, folks will be flocking to the only alternative. Link will probably also get a bump at the Rainier Valley stations but it won’t be as great as the tsnunami that’s about to hit ST Express.

    1. So there’s a subtlety in there — does the reduction in tunnel-based suburban bus service also improve tunnel reliability for everyone else, including Link riders?

      1. I don’t believe that many of the Metro tunnel buses are running on routes which are going to be terminated. But it’s a good point you make. Any which are currently in the tunnel will certainly reduce trips in the tunnel.

      2. Fair point. Nothing in the September route revisions affects any existing tunnel service, and only the 106 revision makes any impact (rerouting to the surface, downtown via Yesler Way) in the February proposal.

        To that end, Metro did a weekend’s worth of high-volume tunnel operations testing a few weeks (months?) ago. Have we heard anything about the results of that test cycle?

    2. There are only a few places and times where Metro and ST Express overlap. Most ST routes do not have nearby Metro routes..

      1. For starters, the 242, 243, 250, 260, and 265 are all on the chopping. Time will tell how many of these riders switch to the 542, 545, or 555 instead. Of course, if these buses are already at capacity and ST has no money to add additional trips, these people could simply end up back in their cars.

      2. @asdf,

        ST had better find the money to add additional trips if they want to have any chance of getting a positive vote out of the East Side when ST3 comes up.

        The politics of this whole transit thing is getting constantly more tangled. What happens with Metro impacts Sound Transit and vice versa. People don’t much care which agency’s name is plastered on the side of the vehicle; they just care that it’s “public transit”. If it works, it shines a positive light on all public transit. If it is a hassle to use or fails to deliver its passengers on time to their destinations, it casts a shadow.

      1. That’s a bit of a stretch. Here in Vancouver we have beautiful hybrid buses with overhead racks (small, to be sure; not for suitcases) and reading lights. They were gotten for us by Aunt Patty from ARRA funds. Thank you Aunt Patty.

        People love them! Yes, there is a “rail bias”, but it’s not absolute. Give folks a fast, frequent ride on comfortable, quiet buses and they’ll ride. No, buses don’t trigger the same level of Transit Oriented Development that rail does. Point granted. But for the suburbs they can be a great way to cut the peak off the “peak”. The truth is that most people living in the “‘burbs” expect the flexibility of private car commuting. But if we can provide them with a pleasant commuting experience, some — perhaps not “many”, but enough to make a difference — will want to use it.

        Most western urban areas have multiple employment nodes, and they can’t all be connected by rail from which everywhere that the folks who travel to them originate. It just doesn’t make economic sense. So nice buses with priority must be a significant part of the solution.

        John’s statement is right.

      2. People will take buses up to a point. After the commute gets longer than a certain amount of time, the rail bias becomes stronger.

  8. In my opinion, the popularity of the system greatly out weighs its quality. We haven’t even built the most important part — the part of it most experts and your average resident knows will be popular — downtown to the UW. Integrating bus, bike and train service should further increase ridership, now that the county is looking at the issue in more detail. Given the density of the city (not that many people live close to a station) that could lead to a substantial increase in light rail ridership. Link should, in my opinion, be optimistic and assume that plenty of people will ride light rail. This means they need to address several issues:

    1) What will it take to increase headways from the UW to downtown? I’ve heard that if pipes are built, we can get this down to three minutes. Well, let’s do it. This will greatly improve bus to train interaction.

    2) Deal with the fact that Rainier Valley and Bellevue have very limited headways (six and eight minutes) while the demand for SoDo to UW service will exceed that. We shouldn’t drag down the entire system just because one part of it has a limitation. Nor should be spend billions duplicating service. It seems to me like a turnback line at SoDo makes the most most sense. There is no excuse for trains running less than every three minutes from SoDo to the UW. There will be demand, and Link might as well start talking about it.

    1. I’m not sure of the speed at which trains will be able to turn-around at UW or at Northgate (once each extension opens). Still, I think that this has a good chance of being the major delay point in the Link train operation once the extensions open. The north turn-around has a high likelihood of being the major Link operational problem in the next phase of our rail expansion, especially when both lines are operating.

      1. I’m surprised at that, given that everything to the north (so far as I know) is grade separated. In other words, is there a reason we can’t run trains from Lynnwood every three minutes? I really don’t care if they are half full until the U-District, as long as they can keep going to SoDo every three minutes.

      2. The turn-around has nothing to do with being grade separated. It will require a driver to make sure everyone gets off, then take a train out of service, then go to the opposite end of the train, start the train back in service (and wait for the control center to register the train) and likely board a bunch of new riders. There may also be some sort of “break” time for the driver or “recovery” time buffer to get a train back on schedule.

      3. There’s no turnback at UW or U-District, only Northgate. Originally the East Link trains were going to terminate at Northgate off-peak, although now they’re going to Lynnwood. But isn’t the turnback infrastructure still in the plan?

      4. I agree with Mike. My point about the grade separation is that as long as the line can handle the headway (three minutes) then it really doesn’t matter where the turnback is. It can turnback at Northgate or Lynnwood. At worse, we “waste” the time by running trains every three minutes from Lynnwood (as I suggested above). That is a small price to pay for three minute headways in the core of our system (UW to downtown).

        Also, I assume the turnback system has extra tracks, allowing for extra trains to be idle. This means that a train doesn’t need to immediately turn around, but can simply be pulled out of service. Meanwhile, a train that was sitting there (idle) starts up and heads the other direction. With just one extra train, that allows (theoretically) six minutes to turn around.

        As you can tell, I’m not expert on these issues. But I think these are the things that we should be looking at, and spending money on. If we add a few stations north of Lynnwood, it will be nice for the folks up there. But in terms of the overall system, we could get a lot more benefit for a lot less money if we invest in things like proper turnbacks, ventilation shafts, etc.

      5. Mike Orr: There clearly has to be a turnback at UWS; it will be the end of the line for five years. There is a crossover at the south end of the station to enable this. It will work operationally like Airport station.

        Al S.: You can speed up the turnback process by adding people. A second operator to enter the cab at the other end of the train to prepare it for the return trip. Station personnel to do the security sweep. The second operator could run the train south so the first operator could take his break until a later train arrives.

        And there are two platforms, so to get 6 minute headways, you have twelve minutes to turn the train around (ignoring any problems with late trains).

      6. The additional infrastucture to enable turnbacks at Northgate once Lynnwood Link goes into service is a pocket track (like those at Stadium and Rainier Beach) as shown in this large PDF. Before Lynnwood Link goies into service, the pocket track along with the tail tracks would allow storing three trains for the start of service or for increased frequency at the start of the PM peak.

        Another interesting thing in that alignment droawing is that there will be a “maintenance track” and universal crosover north of Maple Leaf portal and south of Northgate station. This will presumably be for MOW operations in the tunnel and on the elevated sections as indicated by the “High Rail Access” in the retained fill section at 1st NE.

      7. It never hurts to have crossovers in various places for use as unexpected turnbacks. Sometimes, maintenance or repair require a section of track be taken out of service for a time.

        Even on line sections that are completely grade separated, sometimes drunks in BMWs come crashing down out of the sky and land on the tracks anyway:
        http://blog.oregonlive.com/commuting/2012/11/trimet_releases_cost_estimate.html
        causing a section of track to be taken out of service for a time. In such cases, the ability to turn trains at non-normal turning points can be helpful.

    2. If we have extra capacity on the south end of the tunnel (the tunnel holds more than can fit on the tracks in Rainier and East Link combined) does it make sense to add something like West Seattle or Georgetown onto the south end of the line?

      1. For West Seattle and Georgetown, I propose an improved SoDo station along with lots of road improvements (extra HOV lanes, HOV ramps, etc.). I really think SoDo makes sense as the interchange for just about everything that comes from the south (and that includes buses from Tacoma). It would be a lot cheaper to invest in a turnback station there, and run trains frequently, instead of running trains to West Seattle. Train service to West Seattle is, unfortunately, ridiculously expensive for the ridership. So much so that Link proposes ten minute headways maximum. In other words, good BRT could easily provide double the headway, with good reliability.

        But that only makes sense if the transfer cost (in time) is minimal. Three minute headways between SoDo and Northgate is minimal. Eight minutes (which is what it is now) is not.

      2. No. It might make sense to run turnbacks, but it never “makes sense” to extend rail to an area which is built out with insufficient density to support rail and which has no likelihood of developing it.

        Georgetown is never going to densify; there are airplanes landing and taking off from Boeing Field a half mile away. West Seattle is never going to densify; the NIMBY’s are rampant.

        Now maybe in thirty years, if a tsunami of climate refugees comes to South King County, it might make sense to build an “express cut-off” from Boeing Access Road alongside Airport Way to the VMF with a station in Georgetown. But not before.

    3. Decrease headways”. The headway between two vehicles is the number of minutes elapsed between vehicles stopping. You want to increase frequency!

  9. With numbers like this ST really needs to move to 6 min headways. Doing so will add 25% more capacity with the existing 2 car trains. That should be more than adequate to cover existing demand and a reasonable amount of growth in 2015.

    To facilitate this Metro needs to start reducing the number of buses using the DSTT. This should be a fairly easy process. We know some northend tunnel routes will be vacate the tunnel when U-Link opens and instead be restructured as LR feeders. Metro should identify these routes now and surface them until they get truncated after U-Link opens.

    Surfacing a DSTT route for a few short months when you know it will be leaving the tunnel in 2016 is no hardship to anyone, and it is a good way to slowly progress to the future while still accommodating current needs.

    1. 7.5-minute frequency may not be enough, but 6-minute frequency is still overkill, and would only serve to slow down everything. ST is fully capable of dividing 60 by 9 for the peak-of-peak schedule, but has to give Metro 180 days’ notice if they are going to add a train trip, IIRC.

      The 106 is scheduled to be restructured out of the tunnel, though that doesn’t make up for the extra trips needed on the 550.

      If routes are to get surfaced early, they ought to be the ones that are planned to be on the surface forever. My vote is to surface the 101, 102, and 150, which ought to be entering downtown at the Seneca St. exit from I-5. Then Bay B can be split up between Mercer Island and not-Mercer-Island, or just go to having one bay, whichever method minimizes time needed to get the southbound bus platoons through.

      Northbound in the tunnel is actually worse, thanks mostly to the Bay A pile-up. Moving some routes to Bay B might do the trick, but if not, I’d surface the 255. That’s not ideal, given that it faces restructure, but all the other routes won’t be more than local connectors by the time Lynnwood Station opens.

      1. Dude. It’s pretty simple math. Link ridership is growing at 16.9%. Adding a ninth train to the existing 8 would add 12.5% to capacity. So doing what you suggest would actually result in Link falling further behind. And even if the growth in ridership falls back to around 12%, doing what you suggest would result in no improvement and would not solve the problem.

        Also, with nine trains per hour their arrival times would “walk”, which is a very poor design indeed.

        Na, Link is going to 6 min headways. It is not worth having ST play around with half steps when it will need to take the rest of the step in a few months anyway.

        Time to move toward the future. The future is 6 min headways on Central Link and a LR only DSTT. MESSING AROUND WITH SHORT TERM “solutions” that doesn’t move the region in that direction is a waste of taxpayer money.

      2. @Lazarus,

        Are you aware that Link peak headway on the south fork is planned to revert back to 7.5 minutes after East Link opens?

      3. @Brent

        It’s impossible to predict ridership 9 years in the future, but it will be interesting to see if 7.5 minute peak headways will be enough to meet demand south of IDS in 9 years. On one hand, it would be astonishing if the Central Link segment continues to have 10-15% annual ridership growth over the next decade, but on the other hand, U-Link and Northgate Link will probably help make the Rainier Valley a more desirable area to live and work, so I don’t think it’s completely crazy to consider the possibility that peak four car trains running ever 7.5 minutes will be packed south of IDS in 2023. If it does get to that point, with the 6 minute limit on headways through Rainer Valley, we won’t have much more room to expand service though.

      4. It seems evident from the numbers that ST should shift to 6-minute headway in peak immediately. Most of the buses are going to have to be kicked out of the tunnel eventually anyway… it’s time to start doing so.

        ST is also going to have to start looking for money to buy additional cars to lengthen trains.

Comments are closed.