The Sound Transit Service Implementation Plan (SIP) is a goldmine of great data. Lizz has already reported on some of the big-picture stuff, but check out this graphic depicting load factors on Link.

As one who experiences Link’s load factors daily, I’m fascinated by the patterns that I can recognize. We see the two rush hour peaks, and that the most crowding occurs to and from the Rainier Valley. There are also inconsistent peaks around lunchtime and 10pm, which I assume are related to day and night games.

It’s also interesting that the practical limit appears to actually be about 200 people per car. Frequent riders will recognize the case where it is actually impossible to add riders. Given the seating layout, I don’t see how a “crush load” of 252 is feasible.

43 Replies to “Load Factors on Link”

  1. I agree. The service implementation plan is always fascinating. Yet I still want more.

    Specifically, I would love to see load on a particular station. They list overall ridership per hour (the graph you referenced) and they list daily ridership for each station, but I have no idea how people are using individual stations. I can only surmise.

    For example, it is quite possible that the Angle Lake station is being used almost exclusively during rush hour, while Capitol Hill and the UW stations are used throughout the day. But what about SeaTac? Is it being used primarily by commuters, or for people who take the train to catch flights that fly all hours of the day?

    None of this matters very much right now, but it will likely matter in the future. Running a train every five minutes or so between downtown and the U-District makes sense if there is demand all day long (and my guess is there is). But what about Lynnwood? Should we send trains there every five minutes in the middle of the day, or should we stop half of those at Northgate? Knowing how individual stations (or segments of the line) are used would help in answering those questions.

    1. Anecdotally, nearly everyone I see getting on or off at SeaTac has luggage, which probably means passengers catching a flight. Now that Angle Lake Station exists, there’s a lot less reason to use SeaTac for pick-up and drop-off or A-line transfers than there was before, a conclusion that is supported by the SeaTac station ridership numbers dropping a bit since Angle Lake Station opened.

      1. And many times I’ve used SeaTac station where my small backpack belies that I’m flying, so the number of airport users might be even higher than it appears

    2. What commitments were made in terms of frequency to Snohomish County voters/taxpayers/potential Link users in the ST2 ballot measure?

      1. With just about every Snohomish County bus funneling into Lynnwood Station, I wouldn’t worry too much about frequency dropping.

        If anything, North King may end up paying to boost tunnel frequency at some point because the trains will already be too full when they reach Northgate.

      2. ST last showed both lines going to Lynnwood in 2024-5. However, the overcrowding in North Seattle could be so bad that riders won’t be able to board when it opens. At that point, there would be a huge outcry.

        Who gets the blame?

      3. My biggest issue with ST3 is that the ST2 riders will lose service quality once ST3 projects open (like station and vehicle overcrowding) — and ST ignored those impacts and failed to explicitly budget for those mitigations.

      4. I’m firmly in the camp that if Link’s worst problem is overcrowding, it’s been wildly successful

      5. ST3 overcrowding won’t be everybody getting on north of Lynnwood because most of them will already be on Link in ST2. I can believe a gradual 10% or 25% increase over ST2, but 50% or 100% would at least two decades if ever. And some of the ridership will be doing different patterns, like Everett to north Seattle, Seattle to Lynnwood, and King County to the Everett Industrial Center. The latter will add to top crowding only if they board south of U-District station, but Boeing Everett is not the huge draw from everywhere that downtown Seattle is, and the further south you go, the fewer people probably work there. In any case they’d show up as reverse commuters, or alongside commuters from the south to UW, so not in the most crowded part (U-District to downtown AM, reverse PM).

      6. That’s just not true, Mike.

        1. The new SLU.LQA subway will encourage more riders to go to Westlake to board Link headed north to Capitol Hill and places north, adding crowding between Westlake and UW. It’s been often shown that the segment from Westlake to Capitol Hill will be the most crowded Link segment, and ST3 makes it worse.

        2. 522 BRT will add riders to the North Seattle segments of Link that otherwise would be going by bus to Downtown Seattle, adding to the most crowded North Seattle segments. Of course, the horrible 145th station design is terrible for the 522 BRT except for it providing a layover space.

        3. The second Kirkland-Issaquah line on the Eastside will affect the needs at the transfer stations (South Bellevue and Wilburton), probably needing an additional platform to facilitate transfers. There is no plan to modify these stations — even though they could still be modified before East Link opens.

        4. The ST2 extensions to Federal Way and Redmond may have been promised in ST2, but the subsequent designs based on later ridership studies did not assume that those extensions would be in place.

        5. The SODO station design is not set up to accommodate transfers, now planned to happen as a result of the second West Seattle line and swapping of the Link line to the new tunnel.

        6. I can’t imagine that ST will continue a Tacoma-Seattle express route once Link goes to Tacoma. That line will also have commuters from South Federal Way to Downtown Seattle as well as those from Pierce County. That will add riders to Link.

        7. The infill stations at Graham, 130th and Boeing Access Road will add riders between those stations and Downtown Seattle.

  2. Another thing I would love to see is a breakdown of the trip pairs. Since our system requires it, this shouldn’t be too hard to gather. There are only two pairs which you can deduce right now, and those are on the ends. Over 2,000 people a day take the train from Capitol Hill to the UW. Over 400 ride the train from Angle Lake to SeaTac. But I am curious about other pairs. For example, how many people are taking the train from Capitol Hill to Rainier Valley. Or how many people ride the train from the UW to SeaTac. As Link extends farther, and the bus system adapts to it, knowing this will help with the planning.

    1. I’m 100% behind this. Metro Transit in St. Paul/Minneapolis uses on-board surveys to ask people where they boarded and where they are going, then extrapolates it out to all passengers by time of day.

      See page 25 on this presentation PDF.

  3. the most crowding occurs to and from the Rainier Valley.

    How do you come to this conclusion. It could be that the cars are most “full” in the middle of a line where people are passing through. The destinations are obviously DT, Capitol Hill and UW which account for about 32k or the daily 37k riders. What stands out to me is the highly symmetrical on/off numbers for the DT stations. I’d expect this create the biggest “crowding” issue as wholesale changes are a lot more difficult than adding a few riders.

    1. The basis of my conclusion is that highest loads are NB in the morning and SB in the evening.

      1. The reason loads are higher in the AM commute is because people have a lot more flexibility to do things like shop, go out to eat, visit with friends, etc. after work so the load is spread out in the PM.. Given the very limited on/offs it’s a stretch to say the crowding occurs to/from the RV. People are primarily going to/from downtown and have to go through the RV.. As many people are going to/from the Capitol Hill Station as from all the RV stations combined. A big surprise for me is that the overwhelming ridership is to Capitol in the AM. That is, it’s more of a jobs center than residential. I guess the college is the primary driver of this in spite of DT being the largest jobs center in the region and the primary transfer point to everywhere.

      2. Thinking about that a little more I guess you can’t make the assumption that just because the numbers are higher for NB offs and SB ons that people are going in a particular direction. It’s a no brainer for end of the line station Angle Lake and a good bet for Seatac and TIB. But from CHS you can’t know if more people are originating their trip there or not. UW obviously is the inverse of Angle Lake with respect to on/offs NB vs SB but I’d venture a guess more people commute to UW than from the U-Distiict.

      3. More people commute to UW, but the number commuting from the U-District is also large. However, a lot of them aren’t on Link yet: they take the 70 or 49 rather than going to UW Station for just one or two stops, especially if they’re in the northern U-District. I was puzzled when Metro made the 49 the most frequent route on Capitol Hill when the 43 would have been a more logical Link shadow. But the first day of U-Link I took it up the University farmers’ market at 50th, and I went to the Roosevelt library and then thought of something I wanted from the Capitol Hill library, and so I took the 49 back because it was practically door to door. And when the bus is 15-20 minutes to Capitol Hill and 30-35 minutes to downtown, and transferring to Link is only slightly less especially if you have a walk at the end, then the bus isn’t that bad. All this will change when U-District station opens and Link is a much faster way.

  4. Also crowding could vary from car to car. Anecdotally at the airport 90% or more of people board the front 2 cars.

    1. This. Was only the first car observed, were cars observed at random, was the reading an average of the two/three cars? Each will tell you something different, each measures something other than what you think, each has its problems.

      At the ID, I board the third car routinely. Works well heading north, heading south the third car is better but is still stuffed at rush.

      1. To be clear, I have no idea. But if I were to guess, I’d say they use the tap-on/tap-off data – plus some extrapolation for paper tickets and freeloaders – to know the load of the train with fairly high fidelity, and then divide that by the number of cars in the train.

      3. Thank you for the pic, I was wondering what that was. My guess then is that the measurement was an average or median.

    2. Generally, people tend to board at whatever car is closest to the entrance, which varies from station to station (*). A well-designed system should intentionally have different stations favor different cars so that the net effect is a well-balanced passenger load throughout the train.

      Many of the stations downtown, the last car in the northbound direction is in an awkward location behind the stairs, so tends to not get much use. By contrast, the UW Station has the escalators drop you off near the first and last cars, while the elevators position you for the middle car.

      (*) If the train isn’t immediately about to leave, it’s actually more optimal to walk the platform to position yourself to be closer to the exit when you get off, but most riders don’t bother to do this.

      1. I think that, astime goes on, more and more people will choose to board at doors near where they want to go at their exit station. It will likely be more noticeable when there are four-car trains. Others may avoid the most crowded doors and start boarding at doors likely to be less crowded.

      2. Once we get full length open gangway trains, you can walk to your exit door even if the train is about to leave. You just have to walk through the train

      3. BART has open gangways. Most people still pick their favorite door because they likely don’t want to push their way though a jam-packed train car as well as open the heavy doors between cars. The same thing would probably happen here if we had open gangways.

      4. BART’s inter-car doors are not open gangways. True open gangways have no doors, the gangway is about as wide as the car interior itself such that it functions as a long continuous space, and often people are permitted to stand in those areas.

      5. I mostly ride between UW and Capitol Hill, so when I’m going northbound I get on the last car to put me at the southern escalators which go all the way up to the bridge. Southbound I get on the back of the first car or the front of the second car to position me for the southern escalator. If I’m going out the north side I do the opposite, because if you’re at the front of the first car or the back of the last car you get out behind the escaltaors and have to backtrack and I don’t like that. I must be lucky with times because the trains are rarely full so there’s no reason to go to the rear car to get a seat.

  5. Does anybody know if LINK has any passenger counting equipment? When I was on light duty, my assignment was to get data from these mechanisms on buses.

    Mark Dublin

    1. I think they have automated passenger counters at the doors, but I’m not sure how accurate that is when people are standing near doors.

  6. This diagram raises lots of questions for me.

    – How do the load factors look between each station? Is this just the highest point and the locations vary? Some trains appear more crowded in the Rainier Valley while others appear to be more crowded north of Westlake.

    – What’s the difference between two and three car trains?

    – Are there seasonal variations in loads, especially in the summer when UW and othe schools have fewer students?

  7. I would agree with Martin that 252 is way too high for a possible capacity. People have to be able to get out of the car at their station! Others won’t want to get too far away from a door and the doors can be at crush loads today (Link riders aren’t motivated to step up into aisles in the current cars). I think 200 is probably the realistic capacity.

    1. You want to rebuild the underground stations at hundreds of millions of dollars? Maybe by the time we do that the UW will relent on extending the Triangle tunnel to the station.

    2. > … double tall …
      Almost certainly never. The height of existing tunnels downtown constrains the height of trains. I’d be surprised if there is adequate clearance to add a useful upper deck. Adding extra height top the existing tunnels is likely to be both prohibitively expensive and disruptive. Moreover, the cost of a tunnel goes up almost quadratic ally in its diameter: adding extra height that will never be necessary is almost always a huge waste of money.

      The good news, is that except perhaps between Downtown and The U, there is plenty of planned capacity for almost any believable medium term passenger growth. Moreover, double decker trains are considered a bit of an anti-pattern for high frequency, frequent stop services. First, they require much longer dwell times. Second, they don’t add as much extra capacity as one might think: certainly much less than doubling.

      If the capacity of the current system becomes a genuine problem, there are a large number of things we can try to fix the problem, in rough increasing order of cost. 1 & 2 are already on the horizon, although there may be short-term constraints on their full implementation due to a near-term shortage of rolling stock.

      1) End joint ops, and use the freed capacity to run more frequent trains.
      2) Run four car trains.
      3) Get rid of the unused cabs (this requires new rolling-stock, and some infrastructure changes to support maintenance and storage of harder to divide trains — we already barely use the operational flexibility afforded by the extra cabs in service)
      4) Ban manspreading, get rid of some or all seats, ban bikes, restrict the size of luggage. These are financially cheap, but perhaps politically difficult.
      5) Go to open gangway stock (like 3, but more so, may also require some localized improvements to the loading gauge..)
      6) Run more frequent trains. 2 1/2 minute headways are simple, but would require new signaling systems. Peak headways of 90 seconds are managed by many systems worldwide, but require appropriate signaling, rolling stock, and passenger behavior. On our system, they would probably require significant upgrades to the clearing capacity of the busiest stations, perhaps including Spanish style platforming.
      6) Even longer trains. Expensive: the signaling system needs to support the longer trains, and station and maintenance facilities need to be expanded to support the longer trains. Note that open gangway stock and selective door opening make this somewhat less burdensome than it sounds, but it’s still a big expense.
      7) Parallel tunnels. Already planned for DT. I worry a little about SLU and a lot about Westlake-Broadway. If the latter becomes a problem, a single bore peak express tunnel from Northgate to SLU and Westlake, joining one or the both of the DT tunnels might provide a cost effective solution. Alternatively, a parallel tunnel, with a couple local stops between CH and the current University station might work.

    3. Does anywhere in the world have double-tall trains? Once you’ve filled 4-car articulated trains (which are like 8-car traditional trains, don’t forget), and done the low-hanging fruit like eliminating interior driver’s compartments, who h by the way is cheaper than lengthening stations), the next logical thing is to build a parallel line. That will divert people who live closer to the second line or equidistant between them, and serve new people and geographic areas on the other side of the second line, AND make the entire network more usable and make people more comfortable downsizing their number if cars because more of the city is accessible by frequent/fast transit. That’s why cities often have two or three lines going North instead of just one, and why they don’t raise the first line to extraordinary capacity instead.

      It burns me up when Sounder or the Chicsgo So lengths platforms. That fits more people into existing trains but foregoes the opportunity to make trains more frequent or closer tomorrow neighborhoods, which would both serve people better and attract more people to transit.

      1. Sounder cars are bilevel as are many commuter train rolling stock like Caltrain’s and Chicago Metra’s gallery cars. Amtrak long distance cars are double deck. Internationally, you’ll find bilevel cars running in Paris, Zurich, and Tokyo to name a few. But these are all trains sized to run on mainline railways. Most subway tunnels aren’t built large enough to accommodate such trains.

      3. Sounder is a bit different because ST doesn’t own the lines so the incremental cost of an extra run is high, unlike link where the cost is primarily fixed.

        I’ll support 10 car sounder because that is forecasted to be needed once we have all day sounder

    4. I would add bypass tracks in the mix of options. That’s especially true for lines that stop at least a dozen times before reaching a major downtown.

  9. Pingback: Podcast #51: Load Factors

Comments are closed.