Tiffany Von Arnim (Flickr)
Tiffany Von Arnim (Flickr)

When Sound Transit decided to split the spine for the ST3 package – sending Everett trains to West Seattle and Tacoma trains to Ballard – it did so for a number of reasons. The unprecedented length of the spine corridor meant it was always infeasible to run trains end to end; and capacity concerns in the current downtown tunnel made adding a third line undesirable, leading Sound Transit to propose a fantastic second subway through downtown.

Now that ST3 has passed, it’s time to start making a good thing better. In particular, three operational disappointments lie ahead, all of which are fixable:

First, the still-quite-lengthy lines are planned to offer uniform frequency, meaning trains will cruise through Fife just as often as they do through SeaTac or South Lake Union. Even ardent defenders of the ‘light rail spine’ would have to admit both the asymmetry of demand between the city and outlying areas, and the need for additional urban frequency.

Second, the length of the lines and the nature of the radial commute mean that during peak hours passenger turnover will be relatively limited, leading to crushloads as early at Northgate or the UDistrict for the Red and Blue lines, leaving thousands of downstream passengers at UW or Capitol Hill with a consistently poor experience.

Third, this poor experience will occur while the brand new subway squanders its potential capacity, as the Ballard-Tacoma line cannot exceed 5 minute headways due to at-grade running on MLK Boulevard. With the same 400′ platform length constraints as the other tunnel, this means our brand new subway would be just half as utilized as the current tunnel.

What can be done? Expanding on an idea by Page 2 writer Devon Jenkins, I’d offer two possible solutions: turnback trains and point-to-point service. With regard to turnback trains, Sound Transit is already mulling the possibility of UW to Stadium trains as supplemental service to mitigate the likely 2018 closure of the tunnel to buses, so additional turnback options could potentially be considered under ST3, such as Ballard to SeaTac.

But it’s point-to-point service that offers the most dynamic (and operationally complex) option. Cities such as Washington DC, (WMATA), San Francisco (BART) and Denver (RTD), for all their faults, tend to offer greater frequency on core corridors by consolidating in the core and branching on the periphery. The limitations of this approach are clear: without grade separation between lines, the needed crossovers create fragile chokepoints that limit the realistic frequency of each line. Most of San Francisco and Denver’s individual lines run every 15-20 minutes because of the limitations of the Transbay Tube or Denver’s Central Corridor.

RTD System Map
RTD System Map

But in thinking about such a multi-line concept in Seattle, these limitations don’t necessarily apply as severely, especially with a second tunnel. A multi-line concept could go a long way toward right-sizing ST3 while retaining frequent service for everyone, matching supply and passenger demand while still providing the full suburban service we voted for.

Consider the hypothetical below.  In addition to the planned Red, Green, Blue, and Purple lines, this concept would add 3 more: Orange (Ballard-Redmond), Silver (Ballard-West Seattle), and Black (Lynnwood-SeaTac).

Map by the Author
Map by the Author


Consider the benefits:

  • Right-sizes frequency: central corridors would be served as often as every 4-6 minutes, while outlying stations would see service every 12 minutes.
  • Respects the frequency limits on MLK.
  • Fixes the crushloads: with turn back trains from both Lynnwood and Northgate, passengers weary of crushloads could know to wait for a Black or Blue line train.
  • Evens out demand in the tunnels: If each line ran every 12 minutes, each tunnel downtown would carry 3 lines at up to 4 minute headways.
  • Zero new track-miles: The only additional capital costs of this proposal are switches or possibly flyovers immediately south of International District Station, so that trains from either tunnel could head towards Redmond, Tacoma, or West Seattle.
  • Ease of maintenance: Any train could be maintained at any Operations and Maintenance Facility.
  • One-seat rides to SeaTac from both Ballard and Lynnwood.
  • One-seat rides from the Eastside to South Lake Union, Queen Anne, and Ballard, where bus service can never hope to be competitive. (and cough, Expedia). 
  • Creates the traditional West Seattle to Ballard line many had in mind since the monorail’s failure.

Weaknesses include the possibility of uneven headways depending on chosen turnback stations, massively increased operational complexity, continued wild cards such as MLK and the Ballard Bridge, and planned development near the Int’l District tunnel portals. But we should be thinking creatively about how to fix the known problems that lie ahead, and this admittedly radical concept is intended to get the conversation started.

175 Replies to “Frequency Where It Matters: Right-Sizing ST3”

  1. It’s super clever to route the Ballard “Green” line to SeaTac at the new ID superstation. This gives every terminus a one seat ride to every other terminus – which of course is amazing. we’ll have to keep this in mind when designing the new ID station to allow for this design.

    1. ID is already one of the busiest stations on the line with Amtrak and Sounder connections. We’ll need to invest in significant additonal pedestrian access to keep things moving.

      Widen the Weller St Brodge? New underpass tunnels for transit riders? We’ll have to get creative to avoid pedestrian gridlock.

      Try crossing the Weller Street bridge before a game or during rush hour to get a glimpse of the future.

      1. Would love for there to be some kind of pedestrian underpass between International District/Chinatown Station, the Sounder platforms and King Street Station.

        You’d have to get Sound Transit to give up some of their parking garage to make it work though.

      2. Soil conditions could be a problem for an IDS-King Street Station-Pioneer Square underpass. Hundred year old maps show a shoreline there. Tunnel project had to pump grout (a kind of waterproof cement), into the ground for at least several months before northbound boring could start.

        My idea would be a bridge between the entire International District and Pioneer Square- except result would look more like a glass skyscraper laid on its side, at a height where it can clear traffic on Fourth, and any other uses underneath.

        Main concourse would really be an indoor street. In airport fashion, there could be moving sidewalks in the floor.

        Pretty much the style of very old bridges in Venice- isn’t that where he King Street Station clock tower was copied from?- the result would really be a glassed-in shopping street, with stores and eating places.

        Structure would really make everything it touches into a single very active transit hub of a neighborhood, whose functions and activities would more than repay its cost.

        Mark Dublin

  2. Is there any chance of an 8th line tacking people form Redmond and Bellevue to the Airport and back

    1. This suggestion is similar to my thought below regarding a connection between Judkins Park and Mt. Baker. I believe this would be very improbable, given the current alignments of stations and tracks.

      1. That would be fantastic Jon, but yeah this post is intended to propose service concepts compatible with both current funding and the current System Plan. A line like you describe would need another public vote.

      2. Wouldn’t it require more trains and thus possibly another maintenance base? How will that fit into the budget?

    2. No doubt the region will build it, Ian. Wonder if viaduct structure inbound from Tukwila International can divide at the bottom curve for an eastbound line. Will probably be elevated the whole way. Like tunnel boring, elevated tracks will be a lot faster and more economical when it comes time for- call it “The 560 Line.”


  3. I guess in this day and age of saying anything to get elected prevails (‘lock her up”, “build a wall”) that ballot measures should be no different.
    I don’t doubt the demand from the extremes of the spine do not warrant urban core frequencies, but just weeks after the election, Everett and Tacoma are being asked to consider getting only half the service they were touted prior to voting. (ST3 Appendix 3 – Benefits). Going from 6 minutes to 12 minutes in the peak is half, right?
    Of course, there is precedence for cutting system capacity in half, as ST1 and ST2 jettisoned the documented 90 second headway capacity of the existing tunnel for 3-4 minutes max, resulting in the need to build another tunnel.

    1. Perhaps if ridership supports 6 minute headways all the way to Tacoma and Everett, the black & blue lines could be adjusted so that the black line travels from Tacoma to Northgate and the blue line travels from Redmond to Everett? I think this would still accomplish Zach’s goal of maximizing service in the core and fully utilizing both subway tunnels.

      1. You could alternate Tacoma to Everett and Tacoma to Bellevue solving my Bellevue to the Airport with no Transfers Problem and the Judkins Park to Beackon Hill Problem. You would need some sort of goofy Switch at International District Station or some diversion just north of Stadium Station

      1. Page 17 of 20 under Performance of Modes:
        “Link light rail is a 19-mile electric light rail line with 15 stations
        operating predominantly on exclusive right-of-way between SeaTac
        Airport and the University of Washington. Angle Lake Station
        will extend the line farther south in SeaTac later in 2016, and ST2
        investments will build about 33 miles of light rail service in the
        region. Trains run about every six minutes during peak hours and
        every 10 to 15 minutes off-peak and at night.
        With ST3, the light rail system will more than double again to
        116 miles with over 80 stations. Currently two-car and three car
        trains serve customers based on capacity needs, but station
        platforms will accommodate up to four-car trains for future service
        expansion as demand grows.”
        But hey, I get it. How many voters actually read deeper than voters pamphlet?

      2. It doesn’t matter how many people read it because that is nothing even close to a claim, pledge or promise of six minute frequencies.

      3. Exactly. What an odd reading of something that isn’t even there. It describes headways under what actually exists now (“trains run…”), then under ST3 description, i.e. what is proposed, it describes the proposed system itself (“light rail system will…”). There is no mention of headways under ST3, mandated or otherwise. You’d have a better argument if the system turned out to be 114.7 miles long instead of 116, since for some reason that’s the only concrete number given. Even the station count is a nebulous “over 80.”

        I would have voted, happily, against anything that gave the voters control over operational issues like train headways–particularly when it would involve, for the foreseeable future, empty trains rattling through the outskirts every 6 minutes.

        Zach’s proposal is a brilliant one and one that I hope gets a good look by ST.

    2. You could add 2 additional lines: Northgate (or Lynnwodd) to Everett and Boeing to Tacoma. This assumes that most of the trips to and from Everett and Tacoma wont need to go all the way to Downtown Seattle. It would provide additional frequency around the other metro hubs in the region. Could you also add a line for Redmond to Issaquah?

      1. If you add those other lines, then you might as well just boost frequency on the long haul routes.

        (redmond to Issaquah is a good idea, see comments below)

    3. This move would require ST Board approval, which would entail buy-in from Pierce & Snohomish reps (oh hey, advantage of the ST Board also being local politicians…)

      I think implementation of this plan would require overcrowding to actually occur, and for demand at the fringes of the lines to be anemic. Otherwise it is a self-fulfilling prophesy, as demand will be poor at the stations with low frequency.

      As Zach says, hopefully ST can plan ahead and build the stations to allow for this kind of operational flexibility. But I think politically we need to start with full frequency along just 3 lines and see how the ridership actually pans out for a year or two.

      1. Keep in mind these are off peak frequencies. And 12 minutes is only 2 minutes more than the 10 minute off peak frequency of Link now.

      2. We should build the junction to have operational flexibility for the future. As for whether we should reduce frequency below 10 minutes on any line, I have reservations about that. Rapid transit is most effective and has highest ridership when it’s less than a 10-minute wait.

      3. Rapid transit is most effective and has highest ridership when it’s less than a 10-minute wait.

        Right, and rapid transit is most effective and has highest ridership when it serves an urban area. On the fringes it doesn’t.

    4. Tough luck on indignation over 3 to 4 minute headways, Mic. Maybe Fox News can help you by blaming Hilary Clinton and Anthony Weiner for the double-cross. And tell her she deserves to always wait at least 5 minutes for the 60.

      Strange nobody is blaming Anthony for anything right now. Likely because his wife probably has brothers shopping for suitable Pakistani cutlery as we speak. But careful about airing petty resentment in general right now. Otherwise you’ll be accused of being a college student wimpified by one too many short-headway Safe Zones.

      What’s the sneering at defeated liberal buzz-term now? “Suck it up, Buttercup?” Love that one. She beat Sea Biscuit every race, but Franklin Roosevelt’s people buried her story so it wouldn’t hurt Hollywood posterity’s feelings. Brings back my own memory of having a Vancouver Route 4 trolleydriver apologize to me because after ten, service went back from four minutes to eight.

      But eat your liver over this one, Mic. By complaining about double-cross over headways shorter than safe following distance, you’re actually admitting ST and KC Metro’s performance has finally created the kind of passengers which top quality desperately needs:

      People who over-ride Spell Check so their Customer Service schedule complaints make “Stupid” have two o’s. OK, I know we can’t make Southcenter Curve squeal like a trumpet, but just dub in Gershwin for “Next Stop Boeing Access!”


  4. One could also argue that there are social equity benefits of retaining the one-seat ride between the Rainier Valley and UW/North Seattle/Lynnwood.

    I know this would be an unapproved (and probably very expensive) capital project, but while we are dreaming…the black line really ought to go straight up Rainier Ave from Mt. Baker to Judkins Park.

    1. You mean replace one of our most popular and congested buses (the 7) with a fast train while also greatly improving connectivity? Don’t be silly. First we have to build a giant bridge over the Duwamish, run a line to Fife and Ash Way …

      1. On a quiet night, Ross, I could beat the 106 from the Valley to Fifth and Jackson even though he got to use I-5. I’ve still got working prints for wiring the 7 into the DSTT.

        For a chintzy $12 million ramp and some bus lanes and signal pre-empts and a pantograph like those Swedish trucks, I could’ve beat LINK from Othello to IDS every race.

        Wonder if legalized transit racing could out-Imaginize the lottery?


    2. Why would you build a line just to bypass Beacon Hill, Stadium, and SoDo? Given Judkins Park station is in the middle of I90, there is no way this black line would allow a transfer at Judkins without building absurd flyover ramps and/or rebuilding I90.

      Ross, the 7 is crowded because people aren’t willing to transfer at Mt Baker. We need to fix the station access environment there, not build a new line.

      1. That could be a feasible route for a Metro 8 line. Mt Baker to Judkins Park, then over to MLK to to work up Leshi, Madrona, Madison Valley, Stevens Park, etc.

      2. The 7 is also crowded because it goes through a contiguously densely populated area. You know, the type of place that leads to high transit demand, even if the transit is slower than molasses (as the 7 often is). Oh, and Mike, the way to turn “a limited-stop service” into a transit line with urban stop spacing is to add more stops. I’m not saying that I would replace the 7 with a train, but doing so makes way more sense than most of what we approved in ST3.

        As far as Mount Baker station is concerned, I agree the station needs to be fixed. But the other reason people stay on the 7 is to reach other parts of the line (e. g. Jackson).

      3. The 7 also serves a tremendous amount of intra-RV trips (Genesee to Hillman City, 12th/Jackson to Mt Baker) that will not be replaced by Link. A lot of folks take rather short trips on the 7 as there are destinations all along the route. I would venture there are very few folks riding the route end to end.

      4. @ cuyahoga — I agree. That is my point. What I’m saying is that the 7 could be replaced by a brand new light rail line. You could have stops every half mile. It would carry a lot of people. It would not be what I would build next, but it would be better than most of what was in ST3, precisely because it has the type of ridership you describe. Not end to end, but all day short rides.

        Sorry for the thread drift, since again, it wouldn’t be what I would build next. A Metro 8 subway makes way more sense, as it would have similar stop spacing along with very good connections and exactly those sorts of rides.

      5. OK yeah if you are building that as a part of a Metro 8 type line, that makes much more sense.

        But a “Metro 7” line is just silly – the stop diet you’d need to have rail would remove most of the reasons people like riding the 7 north of Mt Baker.

    3. I’ve toyed with the idea of using a streetcar to connect the two stations (Mt. Baker and Judkins Park) using MLK and the I-90 lid. That same streetcar alignment could then continue north up 20th Avenue to Yesler, then tie into the existing FHSC tracks at 14th/Yesler. That would allow for all sorts of operational connection combinations. Add to that a streetcar branch ending at Harborview or Virginia-Mason or somehow near the new second tunnel station at Madison and the operational linkage options could be quite varied.

      1. You get what you pay for. It’s a fraction of the price because it’s a fraction of the ride quality. Why the rush to cheap out? Let’s build some *nice* transit instead of cutting corners all the time and pretending buses are good enough.

      2. @Mars — No. The primary advantage to rail is that a train can carry more people. But *our* streetcars don’t carry more people than *our* buses. So before you even get into the trade-offs, you’ve eliminated the main one for a streetcar. The main advantages to a bus, however, remain. It is cheaper to build (although still not cheap). It is much cheaper to move the route (e. g. If it was a bus, the First Hill streetcar could be easily altered to avoid the button hook if they simply skipped that stop). Buses can avoid temporary as well as long term obstacles (a car or construction). They can go up hills. You don’t need a special storage location (they can drive to the main bus barn at the end of the day). Surface rail, meanwhile, is a hazard to bikes.

        Both require a new set of vehicles (dual sided doors) as well as off board payment stations and platforms that allow level boarding.

      3. RossB – all of the pros and cons you cite come from the perspective of a transit system *planner*. I’m trying to remind you and others that things look very different from the perspective of a transit system *user*.

    4. Link can’t replace the 7 because of the in-between stops. Link is a limited-stop service. And both Metro and the city are planning to split the 7 into a Jackson-Rainier route north of Mt Baker and a 23rd-Rainier route south of Mt Baker by 2025.

      1. Based on density and ridership, Rainier Avenue should be a faster and more heavily used corridor than MLK. Pretty sure it had either streetcar or interurban when Safeco Field would’ve had lady athletes standing up in the saddle on a porpoise.

        Right now, with available lane and signal adjustments, whichever one goes over, or under, the LINK tracks under Fifth and Jackson the Route 7 will be exactly the kind of local supplement every LINK route will soon need.

        But based on the great old printed maps, reason Rainier had so little cross traffic south of Jackson was barbed wire fences. No record of the PCC at Rainier and Edmunds.


      2. A RV streetcar conversion could replace the Mt Baker to downtown portion of the 7 if it were linked into FHSC tracks, but given the current climate on streetcars and transit investment in general I give that almost zero chance of being built.

        We’d be more likely to see private money try to run a streetcar out to the central district on Jackson to jump start development on Vulcan properties out there… and I think the mood in the city is sour enough on streetcars and gentrification that that will never happen either.

  5. The “stem-section + branches” concept and point-to-point service are not radical new concepts, which is great because there are already many “lessons learned” that can be factored into the planning from the beginning. Here’s a few:

    1. When multiple lines merge into one (i.e. northbound at SoDo and Chinatown), any delays along one of those lines can spread throughout the network. This often results in trains stacked within the tunnel approaching the station. Mitigation: Multiple platforms for that approach would allow trains to continue loading/off-loading passengers rather than waiting in the tunnel, which would decrease delays by preventing trains from stopping multiple times. Ideally, each line in the same direction would get its own platform (an island platform for every set of two lines), with the tracks merging into a single track after the station. Examples: Munich Ostbahnhof for the S-Bahn, Münchner Freiheit for U3/U6, Max-Weber-Platz for U4/U5,

    2. The branches of each line should have similar demand characteristics, so that capacity can be optimized for both ends of the line in terms of train length. Otherwise, some trains would be crush-load approaching one end and under-capacity approaching the other end, leading to wasted capacity. This wouldn’t be an issue if ST had enough vehicles to run 4-car trains everywhere, all-day, but it’s an optimization strategy that would come in handy when there are not enough vehicles to provide the needed capacity everywhere.

    3. Further optimize point-to-point services based on origin-destination patterns and not just pure ridership. Use data to determine when and where point-to-point services are needed the most. Ideally, it would be implemented along corridors that have high transfer rates. Example: If lots of people transfer between the northern part of Line 1 and the southern part of Line 5 during peak periods, then it would make sense to run point-to-point services from the northern part of Line 1 to the southern part of Line 5 during peak periods. Also comes in handy for crowd control at stations…Examples: Munich U1/U2/+U7, U3/U2/+U8.

    1. Great comments, and yes these are common arrangements throughout the world. I meant ‘radical’ only in the context of current service plans for ST3. The good news is where you need the most land for junctions, you have the least constrained space (Stadium/Sodo), so we could possibly avoid level junctions.

      1. I use examples from Munich because I live there, and it happens to have great system-wide examples of concepts mentioned in the article.

  6. Is your proposed frequency table peak or off-peak? If the table is peak and every other trip on each line is peak-only, that means if it’s not rush hour, you’re talking about 24-minute headways to Tacoma and Everett. This is not only less than the 15-minute daytime Monday-Saturday headways Everett gets today with the 512, it also doesn’t mesh well with connecting buses that run every 30 minutes.

    That said, if the frequency proposal in the chart is the “all-day” level of service, with a few additional peak-only trains added on top of it, I think it could really work well.

    4-minute all-day headways between the UW and downtown also has a nice benefit of making a route truncation of SR-520 buses easier to swallow, since it would effectively eliminating the “wait time” component of the transfer to downtown.

      1. For off-peak, this sounds great. I see a problem with peak, but there is a good chance it will be solved before we reach full ST3 build-out. Tacoma-Seattle riders (most current ST riders in Pierce County) were given a trade-off with ST3: longer rides, but shorter headways. The trade-off was weaker for the peak: headways on the train (6 minutes) will actually be longer than the bus (5 minutes), but the increase in ridd time is less and the train is more reliable. However, changing from 6 minute peak headways to 10 with no compensatory improvement in speed will be seen as a bait-and-switch.

        The easy answer is Sounder. We’ll have to wait until we know what the negotiations with BNSF give us, but it will likely make the peak Tacoma-Seattle trip more desirable on Sounder than Link, which makes the Link headway reduction much more palatable. Some capital improvements on Sounder would be nice as well. We don’t have to fully implement Troy’s plan, but smoothing out some curves, grade separating the intersections, and adding a third line would reduce travel time and improve reliability on Sounder. A reliable 40-minute ride from the Dome to King Street would be nice.

      2. I didn’t mention it, but I’m also assuming the retention of Tacoma-Seattle express buses, if not by Sound Transit then by reinstatement via Pierce Transit with a premium fare. I think there will still be decent demand for nonstop service, which with proper priority management could achieve Tacoma Dome to Sodo in 35 minutes, or from Tacoma Dome to Westlake (via Seneca) in 45 minutes. Combine 10-12 minute all-day Link with 15-minute peak Sounder and 15-minute peak express bus, and I think you have a winning combo. Stagger the times properly at Tacoma Dome and you have departures every 4-5 minutes during peak.

      3. When BNSF is done with their current construction, we’ll have triple track from Tukwila to Auburn except for a pinch point of double track in Downtown Kent. Combine that with the eventual introduction of skip-stop or express trains and voila you have your 40-minute ride. Amtrak is scheduled for 43 minutes from Tacoma to Seattle today.

      4. In response to your second comment first: That fits with my guess, although I’m assuming you got it from the actual documents rather than a gut estimate like I did.

        To your first comment: personally, I use the Tacoma-Seattle express buses much more than any of Pierce Transit’s current service, but from an equity standpoint, improving local service will be more important than maintaining the expresses once Link reaches Tacoma. Proper management of the freeway would make the express bus more appealing (I timed it once at 29 minutes from Spokane St to the Dome in free-flowing traffic, so your estimate of 35 minutes from the Dome to Sodo sounds accurate), but (as has been pointed out many times on this blog) proper management of the freeway is unlikely to occur.

        If the poltical winds shift enough that we can get reasonable bus priority on the freeway, it will probably be part of the same urbanist tide that makes removal of I-5 between I-90 and SR 520 likely, so keeping the current Spokane St exit and truncating the bus at Stadium Station would probably be better than getting off at Senaca (an exit which would no longer exist). If we do keep the freeway through downtown, then all those left exits will make it difficult to get decent bus priority anyway, so we might as well stick to getting off at Spokane St.

      5. I’m no ocean physicist (I do know a couple), but overcoming the barriers to really removing I-5 through downtown would seem to require an impossibly high “urbanist tide”. Getting over the wall of reconfiguring downtown exit patterns in a space-neutral way should require a much lower (but still quite high) “tide”. Properly managing HOV-lane speeds… that’s a higher tide than we have now, but significantly lower than either of these. Right?

      6. Metro is planning to replace the Federal Way expresses, so there’s precedent for Pierce Transit to replace some of the Tacoma expresses if it chooses. (As if it had the money to.)

      7. I’m with Al. There is a huge difference between changing the 2+ HOV to 3+ HOV and removing I-5 from the middle of the city.

        Anyway, either way there will be demand for express buses, especially in the middle of the day. Meanwhile, ridership of the trains — even if they are frequent and there aren’t express buses — is likely to be very low. So low that it will be hard to justify running trains there very often. Every 12 minutes seems pretty generous, actually. Even at that ST will be bleeding a lot of money for the southern end of the spine.

      8. RossB, I highly doubt the South end will be bleeding money out of ST like you seem to assume. In my opinion, Sound Transit will likely give the axe to some of the routes when light rail to Federal Way opens to reduce redundancy or truncate some express buses at Federal Way and may extend out some bus route network possibly, like to Orting.
        As for the off-peak frequency, 12 minutes doesn’t sound generous, it sounds about right in terms of frequency.
        I feel like you assume no one really uses transit in the south end outside of peak. When I see a good amount of people actually riding the A midday. The 578 usually has a decent line of people waiting for it when I get off at Federal Way transit center mid morning to catch the A to Highline College.

      9. @Zachary — Oh, I know the numbers. It is all in the Sound Transit Service Implementation Plan. You are right, the number of people who ride buses from the south end in the middle of the day is decent. For a bus. But I really see very few people riding the train, since it will be slower to Seattle. They will take Sounder and they will take the express bus (if offered) but why would you take a train ride to Seattle in the middle of the day, knowing it is slower (and likely less frequent) than a bus? The only way they can generate decent ridership is if they kill the express buses.

        But I’m talking about a long distance hauler, not something like the A (a frequent stop bus highway similar to Swift). That bus is a different beast (although still not one of more popular buses). It is one thing to make a trip a few miles up the road on SR 99, it is another thing to take an hour long trip to a different city. Those types of trips tend to be planned, and thus frequency doesn’t matter as much. Anyway, the numbers for all the Federal Way and Tacoma express buses aren’t very high. It’s less than 10,000 for sure. I think it is less than 5,000 (but I haven’t counted it up). It is hard to justify running a train for that long with so few riders. All that being said, ST will likely run it every 12 minutes, as you said. ST will be losing money (a lot) but that is just the nature of systems like this.

  7. Another benefit – the many overlapping lines means you don’t need to use the major stations to do transfers, but can instead transfer at rarely used stations. An Everett to Seatac rider could transfer at the lightly used Sodo or Mountlake Terrace stations, rather than needing to fight (and adding to) the huge crowds at Westlake or International District.

    ST could even encourage this, labeling the lesser used stations as “less crowded transfers” on the service map.

    1. +1 – Also, these transfers would be on the same platform. The westlake and ID transfers would involve quite a walk.

      1. In fact, by deemphasizing the transfer you could probably reduce the footprint of both Westlake and Int’l District for the new tunnel, simplifying construction and reducing capital costs. That could possibly pay for the necessary IDS weaves.

    2. And when on stations with side platforms, that’s not an issue because you would only be transferring on the same side anyway, and it would be a reliable 6 minute transfer.

  8. “Zero new track-miles: The only additional capital costs of this proposal are switches or possibly flyovers immediately south of International District Station, so that trains from either tunnel could head towards Redmond, Tacoma, or West Seattle”

    Flyovers would be very expensive, but operationally far superior to level junctions. The cross-over delays are significant, and the opportunity to build a flyover is best taken when constructing the line in the first place.

    The Green Line in Boston comes to mind as an example of how a level junction causes major operational problems. It operates somewhat similarly to this concept – the eastern terminus at Lechmere is only served by 1-2 routes while others turn back earlier. It also has a level junction where the “E” branch departs from the mainline that severely constrains throughput and causes a lot of delays.

    Platform access for more than 2 tracks at turn back stations would also be vital, otherwise trains turning back will occupy a mainline track and likely delay thru-trains behind. A setup with 2 island platforms and 3 tracks (center track having access to both platforms) would avoid this.

    1. two islands and three tracks sounds ideal, but can ST incorporate this idea, and extra cost, into what they are already committed to doing?

      The additional lines and turn back track locations sounds wonderful. Much better than running single lines north and south and hoping for the best.

      1. Northgate will have a pocket track but I don’t believe it will have platform access. Kind of how Rainier Beach is set up today (though rarely used). Doubt the designs can be changed at this point – unfortunately ST is baking in operational inefficiency with the station and track layouts.

        In a 3-track 2-island setup, the operator can switch cabs easily by walking along the platform. A security sweep can happen simultaneously.

        Until we get trains with open gangways, a pocket-track turnaround seems like it would require another operator to help turn the trains. The pocket turnaround would also take more time to execute: security sweep on NB mainline track while 2nd operator boards, 1st operator moves train into pocket track, 2nd operator takes control and moves train back to SB platform.

      2. The operator can pull the train into the pocket track, secure it, then dismount and walk to the other end on the ground. No need for a second operator. ST would be doing this if they institute the UW-Stadium turnback line in 2018.

        I agree a 3 track-2 island setup is preferable for turnback operations because it eliminates the need to move the train, but its not required and doesn’t have a significant operational impact.

      3. Agree with Jason – benefit of having cabs at both ends! Something the STB commertariat seems to hate but might finally be useful.

  9. 1. Build the Georgetown bypass, to get rid of the silly MLK line 6 minute headway limitation that will affect the whole system.
    2. Dead end the MLK line at Rainier Beach, maybe interline it to Northgate at 10 minute intervals to start, reducing to 6 if demand allows.
    3. ???
    4. Profit?

    1. Get rid of the silly 6 minute Rainier Valley limit.Period.

      If TriMet is able to do 30+ trains per hour on a line that crosses a busy freeway exit ramp, there should be no reason why this limit should exist in the Rainier Valley.

      1. The problem is that you’re comparing 5+ miles of parallel street running, with many crossings and light synchronizing, with a single, perpendicular crossing. SDOT’s position is that you can’t go less than 6 minutes without a cascading effect on the whole corridor that would affect both cars and light rail.

      2. Busy Parallel roads should be even easier to deal with. The lights stay green longer for ML King than for the perpendicular roads.

        You could build one hell of a good signal synchronization system for what it would take to build the Georgetown loop.

    2. RapidRider,

      Damaging service in the rainier valley would require a vote.

      More useful to more people would be the proposal to connect mt baker && judkins park stations, but would be expensive and still require a vote.

      More useful to everyone, (and no public vote required) would be a speed limit increase in RV for Link.
      Is the speed limit under the purview of the city, or does the state have to get involved if Link and cars have different speed limits?

      1. I don’t know how you’re damaging service, by removing RV from Link’s trunkline.

        Extending 10 minute headways throughout the day would be increasing service. Again, evaluate demand and keep certain 6 minute headways at peak intact if needed. You’d be cutting off a one-seat ride from RV to Tukwila, Sea-Tac and 200th for those very small number of trips, in exchange, you’d greatly increase reliability and headways for the rest of the network.

        Another option is to build out the future Boeing Field Station to serve as the southern terminus of the RV line (assuming GT bypass serves Boeing Field), if maintaining RV->southward service is a priority. A quick, inline transfer would effectively as good as current service.

        As far as changing the speed limit or headways: I believe speed limit matching adjacent roads is an FTA requirement; 6 minute headways is based on SDOT’s position that any less would negatively impact MLK Way signaling.

    3. You don’t need to build a bypass. Just send the train somewhere else. East side would be the first choice, of course. The bypass is really a silly idea (although probably not the silliest thing ST has proposed). It is unfortunate that trains can’t run more often in Rainier Valley. But skipping Rainier Valley would mean well, skipping it. Ridership in Rainier Valley exceeds ridership south of it, and will likely continue to do so.

    4. Build underpasses on MLK! There are 4 main crossings that will constrain the whole system: Alaska, Orcas, Graham, Othello, Cloverdale. Close off all the other crossings! Start asap with Alaska and Othello.

      1. Yes, if they ever do anything to that route, that is what they will do. If Rainier Valley ever gets big enough to justify greater than six minute frequency, then this would make a lot of sense.

    5. The Georgetown bypass line was deleted from ST’s long-term plan in 2014. The supposed beneficiaries of it, South King and Pierce, don’t seem to care at all for it, nor for the faster travel time it would create.

      Seattle Subway still has it in a a SeaTac – Georgetown – Denny Way – 23rd – Mt Baker line.

  10. RapidRider, are you talking about a line straight south from Downtown Seattle, going past Boeing Field? If so, I’ve been with you from the beginning.

    I’m also glad to see some discussion of track design more complex than simply one set of rails in each direction. I’ve often wondered about express track in some places. Good we’re finally talking like a railroad.

    One of best postings ever, Zach.


  11. Potsdam has this type of arrangement on its tram lines.Only several remote line endings have a single route to the end. The rest have at least two routes to the end of the line.

  12. This is a great example of how we should be exploring different operational configurations. It’s important to assess a variety of configurations to maximize productivity and convenience. It’s too bad that you have to put it out there; ST should already be proposing to test these things!

    The value in doing this now is also to help define things like how we add turnback sidings, multi-platform transfers and estimated numbers of train cars.

  13. I think that the Eastside needs several more variations to test. Things like:

    – switching the purple and gold lines so that there is direct Kirkland- Seattle service and direct Issaquah -Redmond service (there is very little likelihood that a Kirkland-Issaquah direct through trip will be made by a rider)

    – offering a direct Issaquah – Seattle line if possible (a three line strategy to match the Downtown Seattle three-line timing sequence)

    Ultimately, the best service scenario will be based on loads, so we should just be making sure that all of the through connections are possible through switching, tail tracks and platforms.

    1. – switching the purple and gold lines so that there is direct Kirkland- Seattle service and direct Issaquah -Redmond service (there is very little likelihood that a Kirkland-Issaquah direct through trip will be made by a rider)

      I suggested something similar back in March, with Downtown-Kirkland, Downtown-Redmond, and Redmond-Issaquah lines. ST said that article contributed to the decision to interline Eastside trains at East Main instead of Wilburton, but in their opinion the overall concept unacceptably underserved the Seattle-Redmond market.

      1. Yes! I remember a reference to that.

        I thought ST’s response to you was notably and insultingly dismissive. ST staff is also being naive if they think that Eastsiders not in the Redmond corridor won’t be demanding direct Seattle service through their future board members.

      2. Why didn’t they demand it before then? Issaquah was adamant on having a Link line but never pushed for one-seat ride to Seattle. Some Issaquahites and Kirklandites might want it, but would they be enough to represent an entire city’s demand?

    2. Yes please! I was thinking the exact same thing. Giving Redmond riders 2 different lines downtown seems like a lower priority than giving Kirkland riders a 1 seat line.

    3. As an Issaquah resident, I would love alternating service on the Eastside: Trains coming from Seattle alternate between Kirkland and Redmond. I don’t think Issaquah to Seattle is on the table, that will be a 2-seat ride no matter the operational design, but Issquah to Redmond is far superior to Issaquah to Kirkland. The value of the lost frequency between Redmond & Seattle should be offset by the benefits of Kirkland to Seattle one seat rides?

      There can still be a frequency different – Zach’s article had double the frequency in Redmond than in Kirkland & Issaquah, which I think is appropriate. (If Redmond-Seattle has crowding issues, just catch the Redmond-Issaquah train and transfer at Wilburton – the train will be basically empty when coming from Kirkland)

    4. – offering a direct Issaquah – Seattle line if possible (a three line strategy to match the Downtown Seattle three-line timing sequence)

      That was long ago rejected by ST. They were pretty clear that they didn’t want the Issaquah line to weaken the headways on the Bellevue to Seattle line. A train could reverse directions at Wilburton, but that gets pretty silly. A rider from Issaquah would ride through East Main and downtown Bellevue twice before heading to Seattle.

      switching the purple and gold lines so that there is direct Kirkland- Seattle service and direct Issaquah -Redmond service (there is very little likelihood that a Kirkland-Issaquah direct through trip will be made by a rider)

      Very few people will ride this all the way to Issaquah (that is pretty much a given). But South Kirkland to Factoria or Eastgate is a reasonable trip. My guess it will make up a few hundred a day, or a substantial portion of the ridership.

      Meanwhile, South Kirkland to downtown Seattle via Link will never be that fast. If you are in South Kirkland, you can either take an express bus to the UW or downtown. Either way, even with an arduous and annoying transfer, that will be faster to downtown (and substantially faster to the UW, which is a major destination). The stops you skip, on the other hand (South Bellevue, Mercer Island, Judkins Park) are all minor destinations, and not worth worrying about.

      Plus, with all due respect to South Kirkland, it too, is fairly minor. There is only one stop on the line, and it is in a very low population density area. That could change, of course, but I don’t see it. Even if it did, the other six stations (east of Wilburton) are much more important than that single one.

      1. Zach’s layout has three lines in Downtown Seattle, which means that the two lines to the Eastside will have a time gap between two of the trains. They would have to be at 8 and 4 minutes, for example, to match the intervals in Downtown Seattle.

        That extra four-minute gap could serve a direct Issaquah-Seattle operation, even if it is only at peak period. That would thus not take anything away from Bellevue-Seattle service. All that would be needed is a pair of tracks to link the two lines so that trains don’t have to be reversed.

        Another option might be to move the Issaquah split back to Wilburton. The East Main merge concept has already been met with some neighborhood opposition already — so it’s entirely possible that ST will have to walk away from the East Main transfer proposal as the Issaquah line gets designed. I don’t like to see ST going back to Wilburton, but I’m not naïve enough to believe that it won’t happen.

      2. South Kirkland is a destination insofar as it’s a bus transfer away from the tech jobs in Kirkland. Nothing like Redmond or UW, but still a job destination.

        As for getting to downtown, you think one seat ride from S Kirkland will be slower than two seat ride transferring at UW? If I’m already on a bus when I get to S Kirkland, sure I’ll stay on the bus all the way to UW. But if I’m driving/walking/biking to the station, I think I’d rather catch the train all the way into Seattle rather than bus+rail.

        But really, I’m advocating for this type of alignment because Issaquah to Redmond is an actual line – lots of people live in Issaquah work at Microsoft, and even with the board curve it will be faster than bus or driving from Issaquah to Microsoft (yes, traffic on both 405 and the side streets at that bad). That said, a same-direction transfers in Bellevue for Issaquah-Redmond trip pairs isn’t all that bad, so it’s not super important.

      3. Eh, I take it back. Kirkland-Seattle and Issaquah-Redmond trips are same direction & same platform transfers, so the service improvements of 1-seat rides are pretty minimal.

        Even if noone rides it Kirkland-Issaquah end to end, it’s the most operationally straightforward way to ensure high frequency on Seattle-Redmond and lower frequency on Kirkland/Issaquah branches.

      4. I suspect that there will be overcrowding between Downtown Seattle and UW. Kirkland riders may prefer crossing on I-90 if there is direct service for that reason.

      5. >> Zach’s layout has three lines in Downtown Seattle …

        Right. It has six altogether, (three on each line). Each line has four minute headways. As mentioned below, it would be something like this:

        3:00 — West Track — Red
        3:02 — East Track — Green
        3:04 — West Track — Blue
        3:06 — East Track — Grey
        3:08 — West Track — Black
        3:10 — East Track — Orange

        Now add purple in there. I don’t see it working, but I haven’t tried to figure it all out. Maybe I’m missing something. Please, tell me how you do it, because I don’t see it while keeping three minute headways (max) in downtown.

      6. You’re right, Ross. It would be hard to shoe-horn in a seventh line from Issaquah through Downtown Seattle at Zach’s frequencies. To keep only six lines there would likely mean that the second line into West Seattle would be truncated before Downtown Seattle — or maybe that second line has a lower frequency so that Issaquah and West Seattle share that sixth slot and some of the other trains would terminate before Downtown Seattle — or maybe the seventh Issaquah line goes to West Seattle rather than Downtown Seattle because West Seattle needs more service to relieve overcrowding and that would give West Seattle three lines ending there.

        Zach’s proposal needs study, as he admits. Also, we’ll have ST2 opened before Issaquah and Kirkland goes to detailed design, and the way that we add service into the system will be much clearer after 2023-4 once real-world demands are established.

        And it’s hard to anticipate how something happens in station area land uses that may completely shift our operational thinking about Issaquah, like how SLU development has shifted our thinking about serving a Ballard alignment.

      7. Terminate at International District?

        By that pioint there are so many lines the transfer penalty shouldn’t be too bad.

        By the time that line gets operational it might be possible for people to consider alternatives that aren’t politically possible now. For example, make the CCC up to light rail standards and run some of the Link trains there. Issaquah to Seattle will probably be shorter trains you could run on the surface anyway.

  14. This network has a couple issues. The largest is the fact that riders headed to Bellevue from Downtown have to wait 12 minutes between trains. Additionally, it shows trains interlining in groups of both 2 and 3. Headways cannot be properly coordinated with both, leading to many areas with uneven frequencies. Finally, this is more a comment about the map than anything else, the network downtown is difficult to follow and understand. The proposal is interesting (and I like the map) but it is important to consider how branching frequency works.

    1. I didn’t do a formal modeling analysis, but I do recognize in the final paragraph the likelihood of uneven headways depending on the location of turnbacks and the ability of trains to adhere to their schedules.

    2. Bellevue would have six minute service all day. Here is an example (as copied from my other comment):

      3:00 — Red
      3:02 — Green
      3:04 — Blue
      3:06 — Grey
      3:08 — Black
      3:10 — Orange

      It could get better, assuming we shrunk headways on all the lines, but that would be mean less than six minute headways on Rainier Valley, which is not currently acceptable.

  15. There have been operational horror stories because of the lack of tail tracks:

    – SF Muni in the 1980s couldn’t turn trains around quickly at Embarcadero, creating notorious delays. It cost almost a billion dollars to fix that problem with a turn-around and some through-routing to Mission Bay.

    – SF Muni rolled out a schedule turning back the T line at Castro station – with a new train control system at the same time! The result was the Muni Metro Meltdown that was eased by giving up on the Castro Station turnaround.

    As ST moves forward with extensions, they should get serious about possible turn-back problems like these. Tail tracks are needed to make this plan work well!

  16. The more lines the better even if there running on the same track. Yeah also love the concept of three tracks, maybe even four. This is something I noticed in New York where the third track was used for express trains that don’t stop at every station.

  17. Very clever. I like it a lot. Not only does it have the advantages you stated, but you also have 2 minute frequency on the shared stations (on alternate tracks). If you want to get from Westlake to I. D., for example, you can expect a train every two minutes — you just have to pick the right track (east or west). Here is a sample schedule, for a train going by Westlake:

    3:00 — West Track — Red
    3:02 — East Track — Green
    3:04 — West Track — Blue
    3:06 — East Track — Grey
    3:08 — West Track — Black
    3:10 — East Track — Orange

    This is not only good for people who are headed from one end of downtown to the other, but folks who decide to transfer. For example, let’s say you are in Ballard and want to go to Bellevue, but you just miss the direct train (the Orange). Rather than wait 12 minutes, you take the next train (the Green). Then you transfer at Westlake (or I. D.) and you have two minutes to transfer to the Blue line. This is just about right, since you are going to a different track.

    My only concern is about capacity. The only really congested area will be between the UW and downtown, especially in the evening (downtown commuters mix with folks headed to Capitol Hill and the UW for the evening). Four minute frequency isn’t bad, but it isn’t the three minute headways that ST can comfortably handle (according to Of course everything could, theoretically, be shrunk by a third. Instead of running trains every twelve minutes, run them all every 9 minutes. So that would mean:

    3:00:00 — West Track — Red
    3:01:30 — East Track — Green
    3:03:00 — West Track — Blue
    3:04:30 — East Track — Grey
    3:06:00 — West Track — Black
    3:07:30 — East Track — Orange

    Sounds great, but unfortunately, that is unacceptable on Rainier Valley. As I’ve said many times, the problem with the Rainier Valley alignment is not that it is slow, but that it has very limited headways (six minutes). You could simply truncate one of the lines (e. g. the black) at SoDo during peak, but that means 9 minute frequency for Rainier Valley and SeaTac at peak, which is quite a bit worse than today. It is difficult to max out the section that really needs to be maxed out (UW to downtown) without making compromises elsewhere.

    It is possible, of course, that we really don’t need three minute headways anywhere in our system. Four minutes frequency with four car trains might handle the load just fine.

    I have opinions on the turn back issue, but I’ll leave that for another comment.

    1. Good comments. I didn’t want to worsen anyone’s existing service (such as in the RV), and I do think that 4 minute headways with 4-car trains will suffice while leaving a little more wiggle room for operational unpredictability due to branching.

      1. I agree. It is a solid plan from an operational and service standpoint. I think it is fine from a capacity standpoint. Here is a proposal if that ever becomes a problem:

        8:00 — Blue
        8:00 — Green
        8:03 — Red
        8:03 — Orange
        8:06 — Black
        8:06 — Orange
        8:09 — Grey
        8:09 — Blue

        You still have 12 minute cycles. So, that gives you consistent 3 minute headways for Northgate, Ballard and the East Side. That solves the capacity problem. You retain 6 minute headways for Rainier Valley and West Seattle, so that is better or the same for everyone.

  18. Eventually a RV bypass would be needed (maybe perhaps in ST 4), hopefully this bypass would serve the communities of Georgetown and perhaps South Park (though would have to cross the Duwamish River twice on this).

    Having the RV line dead end at Rainier Beach is a dead issue and would be opposed. RV needs to have access to South King County too. Maybe in the far future, the RV line can be extended to Renton with some kind of southern transfer point (aka. Boeing Access Road??)

    Question. Would you run all the route branches daily, or would it be like BART, where some route branches don’t operate at nights and Sundays?

    The pocket tracks are mainly to store extra trains or to store a disabled train (to get out of the way, so service can continue to operate.

    1. You wouldn’t need to cross the Duwamish at all. From the current mainline, you’d peel off south to either 4th or Airport Way (most direct). In Georgetown, you’d turn on Corson, then turn on East Marginal Way. With some fancy footwork, you could both serve the future Boeing Access Road Station and utilize the existing crossing over the Duwamish.

      It’d a tight squeeze, because you’d need to fit switches for the GT bypass into the mainline, while keeping in mind the future 400 LF BAR station, plus the fact that the I-5, BNSF and Duwamish crossings are all special cantilever sections. GT bypass would HAVE to serve the BAR station to be able to decouple RV line from the mainline. This may only be possible if BAR is on East Marginal Way, rather than on Boeing Access Road.

  19. Great idea. But what I worry about is somebody getting on the wrong train. How is somebody going to easily tell which line to get on? Can you please submit an appendix to this… cunning plan how?

    I don’t just want a square LED like on Trimet. I want a pledge of an auditory warning and more signage so that folks at International District Station intend to go to West Seattle but end up headed towards Tacoma.

    There you go. Good thinking.

    1. Between Route #s, Line Colors, and the big sign on the front of the train that says “West Seattle,” I think people will figure it out.

    2. This sort of thing happens every day all over the world. What’s proposed here is a bit like what you see today on BART, or parts of the NYC subway. Even here in greater Seattle we have lots of zones serving several different routes: in the bus tunnel, at popular on-street stops (Alaska Junction or 34th/Fremont), and in transit centers.

      It requires some attention to differentiating the routes and directions. What BART does is about the least you can do. But people will make mistakes even given the best chances. I certainly have.

      1. Thanks Al ;-). It’s just for the longest time I got confused when I was racing down to International District & Westlake Stations which way was northbound and which way was southbound. I still get lost sometimes trying to find the right exit at Westlake to the South Lake Union Streetcar.

  20. Ending three lines at Ballard could be a big operational bottleneck. If there aren’t sufficient tail tracks or extra platforms, this could get operationally disastrous very quickly.

    Would ST3 allow for funding one of those lines to have a Leary Way surface alignment to Fremont, and another line to have Market Street surface alignment to the Botanical Gardens/Chittenden Locks as a broader operational solution to the bottleneck? That would prevent this end-of-line bottleneck as well as improve rail station access to more areas of NW Seattle.

  21. At first glance: GENIUS!!
    At second glance: Oh, lord, the operational complexity. And the mediocre headways in the suburbs, and for one seat rides.
    At third glance: This brings up really deep and philosophical issues about transfer based vs. point to point systems, regional equity, simplicity vs. complexity, etc.

    Are one seat rides needed or useful when the transfers are between 4 minute lines? Would you wait 12 minutes for the right train, or just hop on the next one? The reduced frequency of Everett to Seattle seems fine – but, the reduced frequency of Lynnwood to Everett seems like a problem (that could doom development and therefore the line).

    I’m not sure how I feel on the whole. I definitely would not support this option without flyover junctions – it’s too likely to lead to cascading delays and unpredictability when there are issues in other parts of the system. A train you know is coming is more useful than a 1 seat ride.

    However, I think the increased utilization of the new tunnel and the reduction of transfer chokepoints (via one-seat rides and the addition of transfer opportunities) could make sense, if a flyover is built and the system can operate reliably.

    It would be useful to compare this system to one with simple turnbacks and transfers for a representative set of destination pairs.

    1. There are really good arguments for both frequent transfers and multi-line systems. I hadn’t seen the latter argued for, so I did it myself to get it talked about. The comment thread so far is fantastic and constructive both in its praise and criticism.

  22. Question, In I.D. station or Westlake station can you add a center Island and two sets of parallel tracks to prevent backups from forming inside the tunnel as well as a way to transfer from north bound to south bound tracks at the top and bottom of the island and add stairs down to the island?

    1. Did you not see the conceptual diagrams? At IDS the new tunnel will be well to the east of the present platforms and about thirty feet deeper. At Westlake the same situation will pertain with the addtional complication that the tracks will be crossing at a 90 degree angle. Not at grade, of course.

  23. There is another potential operating concern: overcrowding. Having different lengths of lines helps to enable riders at stations closer to Downtown get into a train. Turning some trains at SeaTac provides more capacity for airport arrivers and for RV residents. Turning some trains at Northgate provides capacity for North Seattle and UW riders.

    ThAts another advantage to Zach’s proposal here.

      1. As a Capitol Hill rider, this potential pain point is what got me thinking in the first place.

    1. The counter is that in the outbound direction, someone with a destination of Federal Way will be crammed into a train going southbound carrying both Tacoma bound passengers and passengers with destinations of SeaTac Airport. They may have to take an earlier train and transfer at the end of the line.

  24. How would the interlining work in Downtown Bellevue? Assuming the Blue and Orange lines would be running with 12 minute headways each, with a train arriving every 6 minutes (as in RossB’s example schedule), the headways in downtown Bellevue would have to be slightly irregular. There would be periods of time where a train would arrive every three minutes, followed by 6 minute gaps. For example:
    Blue Line 4:00
    Purple Line 4:03
    Orange Line 4:06
    Blue Line 4:12
    Purple Line 4:15
    Orange Line 4:18
    Blue Line 4:24

    Also, does anyone know whether three minute headways are even possible on that stretch of the line? Another option would be to increase the Purple Line’s frequency to every 6 minutes as well, thus creating even 3 minute headways in Downtown Bellevue. Of course, it seems unlikely that Issaquah-Kirkland is likely to create enough demand to justify six minute headways on that line.

    1. Good point. In general I don’t see anything wrong with an imbalanced schedule. This will have some sort of imbalance, unless you run just as many trains to Kirkland as you do Seattle. I agree that running trains to Issaquah as often as you run trains to Seattle seems silly and expensive. One option would be a turnback at Eastgate. It shouldn’t be too expensive to run there, and there should be half way decent ridership. But I would simply accept a 3, 3, 6 combination through Bellevue, while having a 6 minute Seattle to Bellevue and Issaquah to Kirkland pattern.

      I would assume that Bellevue can accept 3 minute headways.

  25. I find it strange that folks are working so hard to turn the trains the other way, ignoring the most common train transfer in our system. No, it isn’t Issaquah to downtown Seattle via East Main. Taking the bus to Mercer Island is faster. Nor is it South Kirkland to downtown Seattle (taking the bus to the UW is faster). Bellevue to the airport should be fairly popular, but it comes in second.

    The most popular change of direction will be Northgate to Ballard. Not the whole way, of course (that would, again, be faster on a bus) but UW or Capitol Hill to South Lake Union or Lower Queen Anne. These are all urban stops, relatively close together, which means they have high demand all day long. The close proximity also means that the transfer penalty is actually a substantial part of the overall time. It is the difference between slogging along on a slow bus (or calling Uber) versus actually using this multi-billion dollar system we are about to build. If we are going to make a curve in the system, it would be there.

    1. Given ST is planning on Westlake being a perpendicular intersection between the two lines, with the lines at different levels, I think a Ballard to UW line would be impossible. Given UW to Westlake is going to be the busiest section of the system, I’d rather actually build a Ballard to UW line before trying to build new tunnels downtown to enable Ballard to UW via downtown.

      You are most likely right that it will be the biggest transfer in the system, but I don’t see a way to fix that operationally – the transfer is build into the system design.

      And, you know, we can always remove some street parking to give the 44 some bus lanes. Certainly cheaper than any rail solution

      1. (Ballard to UW via downtown as a single line is impossible, to clarify. Ballard to UW as a new line is very much possible)

      2. They are two different issues, really. I’m thinking of UW/Capitol Hill to South Lake Union/Uptown. Those trips will occur via these two lines even if they build Ballard to UW (which is justified for different reasons).

        I agree that they will do nothing to enable a one seat connection between UW/Capitol Hill to South Lake Union/Uptown. But that is why it is silly to talk about one seat connections between Issaquah and downtown Seattle (a less common and less important trip).

    2. FWIW, I’m assuming Issaquah to Mercer island (or S. Bellevue) buses will disappear when Issaquah link opens. Maybe Metro buses from outside ST boundaries might go all the way to S Bellevue, but that’s it.

      For Bellevue to SeaTac, the system will also have 405 BRT to TIBS if people think downtown is too crowded to transfer.

    3. One solution to that is a branch tunnel from Capitol Hill Station to SLU. It would get you part of the Metro8 Subway.

      Barring that, I suppose you could always throw in one of the reversing loops TriMet is so fond of, so southbound in one tunnel can become northbound in the other.

  26. I’m having a mental block right now. Are the two downtown transit tunnels to share the same platform in each direction at ID station? I know they’ll be seperate at Westlake.

    1. I don’t think we know, yet. Water table, soil, building locations and operational switching issues at IDS will be complicated. I could even see that the current platform may have to relocate a bit southward.

      Zach’s scenario generally minimizes transfers. Only Link trips between east and south would need to transfer at IDS. If Zach’s scenario added an Eastside line that went to West Seattle or SeaTac, even that IDS transfer wouldn’t need focused attention.

      1. The reason I ask is that I was thinking what will be the frequencies from downtown to Bellevue. If there are two lines running from downtown to Bellevue at 10 minutes each (peak), than peak frequencies from Bellevue to downtown will be 5 minutes, but from downtown to Bellevue will only be every 10 minutes, because of the two separate platforms.

      2. The issue is not transfers. The issue is that if there are seperate platforms, when going from ID to Bellevue, one would have to choose which platform to go to first, halving the effective frequency.

      3. I see your point. Still, waiting for up to 10 minutes rather than up to 5 between trains isn’t that inconvenient. It may even take up to 3-5 more minutes more to get to a further platform at IDS boarding the next eastbound train anyway. It’s all part of a feasible station design assessment.

        If the second tunnel set of tracks had a center platform at IDS, even transfers between east and south would be greatly facilitated in Zach’s scheme.

      4. I’m also assuming ubiquitous electronic signage. When you’re at 5th/Jackson and headed to Bellevue, for example, you’d see that you just missed a Blue Line train on Track 1 and should instead catch an Orange Line train on Track 3. Or you’ll use the Transit App, etc. If you just blindly went to your customary platform, then yes you may have to wait longer.

      5. @Zach:

        That’s not a very good solution. Every trip between any station pair should use the same platform every time. Toronto has 2 different platforms if you’re traveling from St. George to Spadina or vice versa. But their trains always run every 5 minutes or less on both lines. Plus the platforms are really far apart at Spadina.

      6. At IDS and Westlake signs might be a practical solution, though in both cases the conceptual diagrams have the platforms at least two hundred yards apart. In Midtown, though, a rider is faced with walking several blocks between “Financial District” and either “University” or Pioneer Square.

    2. Depends a bit on what they wind up doing for the route. They have a basic concept but not a specific routing for the tunnel yet.

    3. No; a single platform serving all three lines would be severely stressed. And there is simply not enough room north of Jackson to have the necessary flying junction.

  27. This idea has one serious caveat. That is that for outbound trips from downtown, a person would have to decide which set of platforms to use. That isn’t necessarily fatal, but it shouldn’t be ignored either. In fact, for most people using the south end lines headways wouldn’t be four to six minutes to a destination outside the CBD, but rather eight to twelve minutes at the particular platform the rider chooses.

    1. If you only have 10 tph each coming from Bellevue, Rainier Valley, and West Seattle, then with two tunnels you have to make a choice between either “splitting” lines between tunnels like this or not maximizing your tunnel capacity.

      I can see two schemes that could max out the tunnels without splitting destinations downtown. The first, keeping only 10 tph (every 6 minutes) on each of the southern branches, would be 10 tph Bellevue-Northgate, 10 tph West Seattle-Northgate, 10 tph Rainier Valley-Ballard, and 10 tph Sodo-Ballard. Basically ST’s original “split spine” plan with some extra downtown trains to fill the Ballard tunnel.

      The second sends 20 tph Bellevue-Northgate, 10 tph Rainier Valley-Ballard, and 10 tph West Seattle-Ballard. West Seattle, Stadium, and Sodo would lack direct access to UW and Northgate, but it would max out both tunnels, and increase service to Ballard and Bellevue – both getting 20 tph to downtown Seattle.

  28. I have to just say, 12 minute headway from tacoma and everett would be pretty terrible for riders. Remember that many tacoma folks will be riding to federal way or the airport and don’t want to wait 12 minutes for a relatively short trip. I want to say that regardless of ridership, Tacomans have a right to frequent service.

    I just think that there’s a little seattle bias at play here. Yes, this plan might make service in Seattle excellent, but it would also make service in everett/tacoma sort of worthless.

    1. The 594 and 574 have twenty minute platform headways today. Twelve minutes isn’t ten, but it isn’t twenty either.

    2. Rule of thumb is under 15 minutes for good service, so 12 minutes seems acceptable. It can always be increased if demand exceeds forecasts

  29. Will driverless trains render much of this moot? One of the biggest reasons to run turnback trains is because of the high cost of running nearly empty trains end to end, but if the trains are driverless that will significantly reduce the marginal cost of operation.

    Evening out the demand in the two tunnel is still super helpful, especially at peak.

    (I’m assuming technology will get to the point that the RV can be driverless without needing to rebuild the line)

    1. ST earlier said it would look into a wider range of technologies for the Ballard-West Seattle line including possibly driverless, but it never committed to anything. Now Ballard will be connnected to Rainier Valley so that option is out. The Everett-West Seattle line will be fully grade-separated I think but the Everett-Redmond line has a couple surface segments in the Eastside (east of 120th). We could pursue those Rainier Valley underpasses I’ve been advocating but that goes beyond ST3’s budget.

      1. If the economy keeps booming, they can add those underpasses to the ST2 budget. And they should.

      2. That’s looking doubtful the way federal policies are headed. Voodoo economics is back. There probably won’t be an immediate drop and there may be a minor initial boost, but longer term the trajectory will probably be lower than it would have been without Mr T and Mr P and Mr B and the little T’s. It’s unclear which direction the final policies will go, or how closely they’ll match the players’ past proposals, but assuming they are the same as the pre-proposals over the past month, expect the tax-cut stimulus to be squandered, the deficit to rise, interest rates to rise, international trade to fall, culture-war issues to suck the oxygen from the room, and neglect of our real economic problems (an increasing number of people who can’t make ends meet, can’t pay increasing housing prices, and can’t spend to create other people’s jobs and income, not to mention environmental degradation and its effects on people’s health).

        The wisest thing I’ve read in the past year is that the problems working-class coal-country whites face and inner-city minorities face are the same, but people have come to believe they’re completely different problems and that one is more important the other. (Some think one is more important, others the other.) The next major challenge will be to get people to recognize they’re the same, and set up a no-bullshit way to fix them. Hint: Finland, Sweden, Canada, and New Zealand have shown the way.

        Then there’s concern the hothead might jump into another dumb war. Obviously, that could have a mixed impact on the economy, since with wars the constraints on federal spending go out the window, and the companies are lined up ready to produce.

        Seattle as always may see a different impact than the rest of the country, or at a different time. The tech revolution will continue, assuming companies still find cloud services worthwhile, and everybody wants to facebook and iphone and VR until the cows come home, and have online bargains delivered by drone in an hour. So the tech economy will still be up, whether or not it’s as high as now. But Boeing is the company whose planes single-handedly move America’s trade balance, and its orders are slacking, and a lower-trade environment may make American planes less desirable (China and Canada have some planes if you’re looking for an alternative), and that would affect Seattle’s economy.

        Metro’s long-range plan is based on the assumption that the economy will continue to grow as it has done. That would allow it to reach its 2025 and 2040 goals without additional revenue, if I understand right. But if the economy slackens, that would cut into Metro’s plans. We need most of the RapidRide lines and the Frequent service around it. Some of it seems unnecessary, such as 15-minute service on Aloha Street, and the 30-minute Covington-Snoqualmie express. So maybe 10-15% could be cut without hardship, and it would still be twice as good as today.But I’ve started worrying how much of a slump there might be in the next decade and how much it might affect Metro’s 2025 plans. Prop 1 is propping up extra service through 2020, but if there’s a recession that extra service will go away and Prop 1 will be holding up 2014’s level of service. 30-minute evening 5, 10, and 41, 30-minute daytime 65 and 75, that sort of thing.

      3. “expect the tax-cut stimulus to be squandered”

        Meaning, there’s already billions of dollars that’s just sitting in companies’ and rich peoples’ bank accounts and banks’ balance sheets, and they’re not inclined to spend it on productive investments or jobs or lending out. Instead they’re just sitting on it or using it to puff up stock prices and buy back stock and and do mergers and bid up real estate. Any tax-cut money they get would probably just go to the same things.

      4. I’m skeptical that the existing Link line can’t be retrofitted to be driverless in the future. Technically, a driverless train that travels along a fixed, known route, with limited at-grade crossings in fixed, known locations, is much easier than full-out driverless cars and buses, which are likely to be all over the road by the time Link service opens in Ballard.

        For instance, one of the current major limitations of driverless vehicles (the requirement for an ultra-detailed mapping of every possible road the vehicle might travel on) is not a big deal at all when the vehicle in question is a train that is constrained to traveling along its tracks.

        Obviously, ST isn’t going to make a formal commitment at this time, since it’s too early to say for sure when driverless technology will be proven enough for ST to want to use it (public transportation, in general, tends to be really conservative about these types of things – nobody wants to do it until somebody else has done it first). But if 50 years from now, the Link trains down MLK still require human drivers, I would be very surprised.

      5. +1

        Really, there are only three tools available to a Link train driver when presented with a car on the tracks: bell, horn and brakes.

        Today’s technology is quite capable of activating all three of these.

        The only question is: how good is driverless car technology at determining if another driver is going to do something stupid, and if so, which of those three (if not all) get activated?

  30. From the “Human Transit” guy.

    These kinds of ideas are theoretically beautiful and tend to be operationally hard.

    The desire to create single-seat riders makes your whole network completely interdependent, which means not much flexibility to (a) respond to surges in demand or (b) isolate impacts of disruptions.

    To take (a): In your plan you have limited ability to insert more trains on the peak in any demand-based way — for your maximum headway of 1.5-2 min or whatever it is, because the window you want to insert the peak train into isn’t there continuously along the whole path, due to all the branching. Instead, your only option for the peak is to turn up frequencies all over the network, which could be a poor fit to demand and very expensive.

    Note also that the combined frequencies are somewhat fictitious downtown because there’s no way to put all the trains to every destination at the same station. In your version, for example, you have half the W Seattle trains in one tunnel and half in the other. That’s unavoidable when you’re planning this kind of complexity.

    My advice is always to quit trying to provide single-seat rides but do try to provide fast connections (same-platform or shortest possible walk). Draw the minimum number of lines that do the job all-day, and then consider peak-only patterns that you might overlay, including options for extending shortlines further only on the peak.

    1. Separating them may not work very well though.

      Ballard to Westlake will be extremely busy. However, if it were a stand alone line it would be limited to one train every six minutes due to traffic issues south of the Mt Baker station.

      So, it will have to be combined with some other line.

      Then there’s the issue of the current plan putting lots of capacity where it isn’t needed, such as Lynnwood to Everett.

      Sorting out what lines get mixed with what other lines promises to be an interesting puzzle.

      1. ST plans to turn back the second Everett line (East Link) at 128th Street in south Everett. If off-peak ridership is really as bad as the doubters predict, that could be retracted further to Lynnwood.

        The system needs turnbacks for Ballard-Intl Dist and Ballard-SeaTac (or so). That would allow 6-minute all-day frequency in the core and 12 minutes in Tacoma if desired.

      2. I’m not convinced that ridership to Everett will be “as bad as the doubters predict” but I am quite convinced that South Lake Union and Lower Queen Anne will overwhelm trains running every six minutes.

        Turning at International District is OK, but but it means lost connectivity.

        If the lines are to be completely separated as Jarrett suggests, The turnback that is needed is Mt Baker. That at least allows the 7 and other routes a bit better connection into downtown. It might also help appease the Rainier Valley pitchforks that won’t want to loose their one seat ride to the UW, as the increased frequency will make transfers a bit better.

  31. If I have this right aren’t they building a turn back track at IDS? And if so maybe they intend during peak times to run Ballard to IDS every 3 minutes, but only let every other one go through to Rainier Valley.

    1. There is a turnback at Stadium for UW-Stadium runs. That would presumably be inherited for Ballard-Stadium runs. Whether it would facilitate Everett-Intl Dist I don’t know; I think West Seattle will have a separate elevated track in SODO.

    2. I believe that they can guide a train through the yards south of SODO as a loop, and not even have to have the driver leave the cab. If not, the yard access could also be used as a tail track.

      One key reason for exploring scenarios like Zach’s here is to see if there are siding and platform designs that should come into play. It’s much harder and more expensive to go back and fix something than to do it right the first time.

      1. People: I’m looking at the Lynnwood Extension EIS final design and I don’t see a third track/ tail track just north of Northgate. I know it’s been proposed but I don’t think it’s on the plans — at least not on the sheets from Northgate Mall to 130th Street.

        Is this just missing in the EIS document, or is there no turn-around siding/tail track north of Northgate? Without a tail track, trains in both directions will have to be held so that a train can switch directions on a cross track. With a train every 3-4 minutes in each direction, it’s not logistically practical without a siding. It’s the only way to hold a train while letting other trains pass so that the reverse direction train can then be dropped into service at the right interval.

      2. Al, you won’t find it there; it’s being built as part of Northgate. Northgate builds the tail past the mall and across 1st Ave where Lynnwood will connect to it.

  32. Cool ideas here, but I think it would be better to hook Issaquah up with downtown Redmond rather than South Kirkland. Provide complete coverage of the eastside on a single route. Tie South Kirkland up with Lynnwood or another north terminus.

  33. The south Link spine is very costly and pretty dumb but it has multiple purposes. It is not only a very slow radial line connecting the Tacoma Dome parking with downtown Seattle, but is also connects Tacoma with SeaTac via Federal Way. Link has passing tracks at Angle Lake and TIBS. So, could another line be added to the Zach map: Tacoma Dome to TIBS. It might use one-car trains and be evenly spaced with the Tacoma-Seattle trains.

Comments are closed.