Crowded Link train at rush hour

Crowded southbound train at University Street Station during evening rush hour
Photo by SounderBruce, flickr

Sound Transit CEO Peter Rogoff gave us a tight range on the practical maximum number of light rail vehicles (LRVs) that can be in service at one time regularly given the current fleet, in one of his Board reports last Thursday: ST Ops was able to pull together a 17th 3-car train on Saturday, January 21, for the huge Womxn’s March, on “a moment’s notice”. So, we know 48 LRVs can be used regularly. It might even be 49 or 50. 51 is doable for special events. Neither baseball games occurring almost every other day in the summer, nor a year of the One-Center-City mashup while ST waits for its next order of LRVs to arrive in 2019 can be considered special events.

The tunnel buses are scheduled to be kicked upstairs with the September 2018 service change. As part of that service change, all SR 520 buses are proposed to cease coming downtown, and instead be diverted to UW Station. As Zach pointed out Thursday, absorbing this new ridership with the current LRV fleet may be a challenge.

A more modest version of this restructure was originally explored as part of the U-Link restructures that took effect last March. The concept has enjoyed support from this blog. Let’s take a look at how much more ridership Link will have to absorb with this restructure, and the various scenarios for how to handle it.

PM peak-of-peak hour downtown buses on SR 520
Route Eastbound Westbound
King County Metro 252 2
King County Metro 255 8 5
King County Metro 257 2
King County Metro 268 2
King County Metro 311 4
ST Express 545 9 6
Total 27 11

For purposes of this post, we are not going to talk about SR 520 exit ramp congestion, SR 520 construction, Montlake congestion, escalator breakdowns, elevator breakdowns, long transfer hikes, Husky football shuttles pushing regular buses aside, or Link Light Rail being brought to a halt for hours due to damage to the catenary. We’re just going to assume a reasonably smooth transfer between routes 252, 255, 257, 268, 311, and ST Express 545, and Link, at UW Station, and that Metro and Sound Transit will decide to go through with this restructure.

We know that even with 6-minute headway, some of the southbound trains are getting really tight. Increasing headway will only exacerbate that problem, unless there are more 3-car trains. The addition of up to 750 peak-hour passengers (on game nights) transferring at UW Station to head downtown would add ca. 75 passengers per current peak-hour train.

With 2-3 minutes expected to be removed from the scheduled time to travel through the downtown transit tunnel once the buses are kicked out forever, and another minute possibly being removed for the overly-long forced dwell time at SeaTac Airport Station, the very-best-case scenario we can expect for end-to-end travel time between UW Station and Angle Lake Station is 44 minutes.

One possibility contemplated in the OCC restructure proposal is to add turnback trips between UW Station and Stadium Station. The travel time for turnback trips between UW Station and Stadium Station would be 14 minutes in the best-case scenario. Turning the train back south of Stadium Station means either having the operator walk the length of the train, or having two operators on board, so it adds risks, costs, and perhaps new headway-control issues.

Minumum trains to do peak loop(s)
Headway in each loop Full loop Turn-back loop (Stadium to UW)
6 minutes 16 trains 6 trains
7.5 minutes 13 trains 5 trains
8 minutes 12 trains 4 trains

Doing the 6-minute headway / combined 3-minute headway plan would require most of the trains in both loops to be 2-car consists. Only 4-6 trains maximum could run as 3-car consists. That might yield one extra southbound LRV to Angle Lake during the peak-hour, or it might be just like now (24-25 LRVs per hour).

The 7.5-minute/3.75-minute scenario, if it were to run with 3-car trains on the full loop and 2-car trains on the turnback loop, would still require at least 49 LRVs in service, putting it in the gray area of feasibility. It would have only as much peak-hour southbound capacity to Angle Lake Station (24 LRVs) as is currently being used.

If that really tight 4-train loop could be done in the 8/4 scenario, the deployed LRV fleet would just barely be able to run as all 3-car trains with both loops running at 8-minute headway. This scenario would actually yield less southbound capacity to Angle lake Station during the peak hour (22.5 LRVs) than what is currently being used.

None of the turnback interspersion scenarios allow for PM peak-hour ridership growth heading south from downtown, beyond currently available space on the trains.

If ST opts for the approach of just running all 3-car trains all day, and sticking with the 6-minute headway after the buses get kicked out of the tunnel, those 48 LRVs needed to do the peak loop leave nothing for turnback trips, except maybe having a 2-car train ready to send north from the SODO base at the height of peak to carry an extra 400 passengers to UW Station, and then immediately deadhead back to base.

It won’t be apparent until this summer, when the impact of the additional 2500 daily southbound trips created by the opening of Angle Lake Station last September is fully felt, to see if even 30 LRVs per hour can handle the OCC crush. But unlike the various turnback-loop scenarios, the approach of just running the full loop at 6-minute headway looks likely to be in the right ballpark for southbound ridership demand.

Now, lets look at northbound ridership impact, since that is what is being used to justify the turnback loop. Let’s assume ridership of 60 passengers per bus headed out of downtown on those 27 peak-hour buses headed northeast. That’s roughly 1600 additional peak-hour passengers for Link to carry north to UW, or 160 per current train. If you’ve been in Westlake Station during evening rush hour, you’ve probably noticed that the southbound trains are consistently more crowded than the northbound trains. The new SR 520-transfer ridership would justify adding about 10 peak LRVs northbound, before accounting for existing spaciousness. Moving to all 3-car trains running the full loop would add about 6 peak LRVs in each direction. That should be plenty. To the extent there are crushloads, having them on a train where everyone will empty out 6 minutes later seems better than having them on a train where most of the riders might be on for at least 27 minutes.

The turnback scenarios yield anywhere from 40-46 LRVs per hour per direction between UW and Stadium Station. That seems like serious overkill in the context of new ridership from UW/SR 520 transfers.

Sticking to the simple approach of just running all available trains at 6-minute headway on the full loop, there might be some passengers who can’t fit during the spike at the summit of peak. But having a few passengers occasionally wait 6 minutes for the next train to head home because everyone is heading out at 5:30 pm is a first-world problem over which I’m not going to lose any sleep.

Good news: Sound Transit deployed floor arrows at Westlake Station this week.

They are much farther from the platform edge than the floor arrows used in most other train stations around the world, but hopefully they will help passengers be in the right place to let deboarders get out quickly, so that boarding can happen more quickly.

32 Replies to “Can Link Fleet Handle One Center City Ambitions?”

  1. Another point worth considering is whether the extra influx of transferring riders warrants better headways for a greater portion of the day. For instance, can the definition of “peak” hours be extended by a couple hours, or can the midday headway be improved from 10 minutes to 8 minutes?

    And then, of course, there’s the service reductions for minor holidays, when Link runs a Saturday schedule, which means 10-minute headways, not 6, during the height of rush hour. This MLK day, ST decided to do this with only 2-car trains, even though major holidays like Christmas have been getting 3-car trains. The result was a crushload, even without all the new, transferring riders.

    1. How many 2-car trains did you see on MLK Day? I was told that it will be 3-car trains on all weekends and holidays, until further notice.

  2. When in 2019 are the new trains coming?

    Also I’ve seen in comments before that ST can’t run 4-car trains until Northgate, and when pressed why, the reason is lack of trains (which are coming in 2019, a good while before Northgate).

    Once the new trains arrive in 2019 (pre-Northgate), are there any operational obstacles to running a 3/4 car weekday train rollout instead of the current 2/3, and all 4-car trains for huge events like the Womxn’s March?

    1. They may have to add a third between-car barrier at each station. But that’s chump change within their rounding errors.

  3. I’ve suggested going to the 7.5/3.75 operation along with some others. The idea has gotten pushback from those that think the 90 second additional weight to get to the Rainier Valley and points south is not desirable. Has this changed any minds?

  4. One of the potential difficulties of a mid-line turnback is driver breaks. The current 6 minutes gives drivers at least 12 minutes at each end or at least 10 minutes after the reverse process is achieved.

    The scheme of 3-4 minute trains as a second line would seemingly give drivers only 6-8 minutes at UW Station. Is that a problem? A “seat slide” may be required (drivers hop off one train and wait for the next one at UW Station) as one solution.

    1. I’d expect we can improve on both this and the turnaround lag time by having the new operator board the north end of the train when it’s southbound at Stadium, taking over in the turnaround pocket (so no one has to walk the entire train), and the old operator disembarking when the train pulls back in at Stadium northbound. The old operator then gets a reasonable break (perhaps 2 trains worth) and becomes the new operator of another train.

  5. I’m putting out a lot of effort writing comments for everybody else to tear to pieces. Best I can do. But this is about the fifth time in two week my screen has blocked solid and lost twenty minutes of work. Have e-mailed more than once without an answer.

    So finish up [On Topic]: 1. given the Convention Center economy of Bremerton and the concept and execution of the Deep Bore Tunnel, and all the wailing about approaching surface disaster, transit is in good fighting condition to keep the buses in there and let Seattle give their removal its best shot.

    2. Long years’ experience of vertical transit breakdowns should be legal and political clout enough to rebuild the shiny crap ourselves, and make the guilty beautifully pay for it. First week in office, approach has worked for US Chief Exec, at least re: helpless immigrants and humiliated armed forces and professional diplomats.

    Same real estate market that ‘s destroyed regional transportation has doubtless left construction scaffolding with more elevators than we’ll ever need. Made to run safely after fast assembly.
    We’ve likely got a lot fewer stations. Minimalist appearance is legit. architecture form. With disabled and elderly passengers, we could get sued for a lot more money than we’ve so far litigated. Let’s check fine print. Is the T-name anywhere in evidence?

    3. Joint ops and One (Convention) Center City have equal credibility. First reeks of laziness left to mold, second of bullying by anonymous money. Let’s do same for signal system as above for elevators, and after 17 years give our bus drivers and train operators the training they need as a racing team, rather than a sulking Demolition Derby. Hours saved by loss of losing operations should repay expense with two decades of interest.

    Have read that mental depression is “Learned Helplessness.” Solid rightward turn of our country’s politics indicates that unyielding single minded stubbornness to the point of vicious is a mighty anti-depressant. Opposition posture? Fear of calling oneself liberal for forty years says it all. As do results of this approach for, among other things, public transit.

    Any advice on screen lock-up, most stubbornly vicious Deepest Thanks.

    Mark Dublin.

  6. Open gangway cars if/where practical and single ended cars are parts of the solution, however at the end of the day you’re squeezing in x amount more capacity and still running up short. Reality is we need 4 – or more – cars, we need more trainsets, and this problem will not go away on its own.

  7. What seems like overkill on the urban core today might seem like undercapacity in the future. Increased bus/rail intercepts at the north and south ends of downtown will drive significantly increased demand in the urban core while only making minor increases to demand elsewhere.

    Additionally, it is my understanding that the two busiest stations on the line are ID and Cap Hill. What is really needed to manage demand is some knowledge of load factor and or ridership between stations – particularly between ID Station and stations to the south. If that demand can be met without all trains being 3-car, then the best system is really adding the turnback loop.

    In any case, this is a good problem to have. And the problem goes away in 2019 anyhow. At that point the turnback loop is surely the way to go.

    And the speed and reliability improvements attained by kicking the buses out of the DSTT will help the situation too. Reducing the cycle time is equivalent to adding physical capacity.

    1. Don’t forget that many Sounder riders use DSTT to get to/from IDS and King Street station and as # of bus trips have decreased , the bulk of these use Link – all will after 2018.

    2. As someone who commutes Columbia City/ID there are definitely a lot of Sounder folks catching whatever comes first in the tunnel from the ID to points further north in the morning and making the return in the evening. However, there is also a decent amount of folks getting on Link at the ID to head to the Rainier Valley. My guess is the churn there is pretty even when Sounder is running.

    3. Lazarus,

      You are right that today’s post only goes so far as guessing whether the SR 520 restructure, by itself, would blow Link’s capacity with the current fleet. As you can tell, I think it can.

      Add on a bunch of buses from the south and I-90 terminating at ID/CS, and the number of transferers could get overwhelming. I don’t have hard numbers on those beyond number of buses proposed to no longer traverse the Central Business District.

      The busiest stations, based on Link boardings, are Westlake, UW, SeaTac Airport, Capitol Hill, and International District/Chinatown, in that order.

      It is the overlap of UW-ID/CS and Westlake-south-end trips that is threatening to burst Link capacity.

      You’re probably right that we don’t need 3-car trains all day, but we will to meet southbound demand during the PM peak of peak once the buses get kicked out and the SR 520 buses get kicked out of downtown. The PM peak is also the time when having turnback trips would do the most good. Doing them only off-peak would be kinda goofy. But unless we run a bunch of peak 2-car trains on the Angle Lake loop, there are very few LRVs available — maybe not any — to do the turnback loop.

      1. @Brent,

        The important data per this discussion is not boardings but ridership. A given station might have a relatively large number of boardings, but if the number of passthrough riders is small then overall demand at that location is still small and could still be met with short LR trains. This is of course most true of the Airport Station.

        It has always been my understanding that total system ridership peaks at ID Station, although I honestly can’t tell you if that peak is greatest to the north or south of ID Station. And I’ve heard that system ridershipmis almost as high between Cap Hill and WL Stations.

        What is really needed per this discussion is the screen-line data on ridership between adjacent stations. If the system ridership between Stadium and Lander Stations requires 3-LRV trains then it will be hard to run a turnback line with our current fleet. If not, then there will be options for a turnback line prior to 2019.

      2. Right now, the point where trains are the fullest southbound is after Pioneer Square Station during PM peak. A lot of north-end riders disembark at International District. They are second-fullest south of University Street Station, and third-fullest south of Westlake. I’m not sure where northbound ridership is highest, but it is noticeably less than southbound ridership in the PM peak. Even at Westlake, where most south-end riders depart, it is still noticeably less than the southbound Link ridership that is just starting to get crowded.

        Either lower headway (impossible on MLK) or a series of 3-car trains headed south will be needed southbound. The only realistic hope for turnback trips in 2018 is if ST figures out that the peak-of-peak is at a different time northbound than southbound.

      3. A rider-per-segment at a peak hour could easily be at a different place than a rider-per-segment for all day riders. This is missing data that the public does not know – so we are all speculating. That data would also inform us about the viability of a turnback.

        With Link daily ridership almost doubling because of U-Link, it’s a pretty strong likelihood that the heaviest daily point is on either side of Westlake.

        A final issue is whether we may just need only a few extra relief trains arriving Diwntown between 7:15 and 8:15, and leaving Downtown between 4:45 and 5:45. A time-of-day crowding study would seem to help to frame the best solution.

      4. @Brent,

        That is my point exactly. To use a republican buzz word that they always mis-use, the goal here is to “right size” the system such that the resources are placed where they are needed and not wasted where they aren’t needed.

        To phrase it differently, when considering a turnback line there are two constraints on the system – capacity on the core interlined section and capacity on the single line section.

        If ridership demand south of Stadium Station can be met with 3-car trains operating at 7.5 min headways then the first option to consider is single car trains on the turnback line – and surprising as it might seem, this actually works.

        Why consider a single car turnback line at 7.5 min headways? Because it puts 32 cars per hour in the urban core while only using a total of 44 LRV’s. The full loop option at 3-car trains and 6 min headways actually puts 30 LRV’s per hour in the urban core while requiring 48 total LRV’s. This the single car turnback option at 7.5 min headways provides 7% more service in the urban core while using 8% less of the total fleet.

        What if we just matched the full loop fleet utilization at 48 LRV’s? To do this the turnback line would operate 4 2-car trains and one single car train. Such a system would have the same fleet utilization as the 3c@6min full loop system, but would provide 36 cars per hour in the core – a 20% increase in core capacity at the same utilization rates.

        Starting with a single car turnback line also allows for seamless increase in capacity in 2019 when the new LRV’s start to arrive. The full system could go back to 6 min headways or the turnback could go to full 2-car trains. Or both.

    4. Short-term everything converges at Westlake because almost all the bus routes transfer at 3rd & Pine, and that’s on top of the downtown shopping and Pike Place tourists and businesspeople in downtown offices, plus the fact that Westlake is a pass-through point between north Seattle and Sounder/Amtrak/SeaTac. In the future as North Link. East Link, and Ballard Link are built out, more people can transfer outside downtown. I do like how UW and Angle Lake and eventually Northgate are becoming car-storage points and transfer points for all events in central and east Seattle — events anywhere the line exists — and that will become more so in the future. But for now a lot of people still have to transfer at Westlake because that’s where the transfer is.

      . In the future as Link is built out

    Brent, please check the ST 2016 SIP, page 118 It showed a longer cycle time and 19 trains required for the Angle Lake round trip. the cycle time: 49+49+8+8 = 114; divided by 6 (headway) for 19 trains. With two-car trains, 38 used for Angle Lake. If joint operations end, the peak running time might decrease, but probably not enough to save a train. There could be discussion of a turnback train between UW and SODO with joint operations ending prematurely with the WSCC expansion.

    1. That SIP was written in 2015, before the real-life travel time was known. The time it really takes to travel from TIBS to Angle Lake Station in the current schedule is exaggerated by a couple minutes, but still only 48 minutes, not 49. The trains dutifully hold up at SeaTac Airport Station so as not to get ahead of the schedule.

      I believe there is still variance of several minutes among operators, so there may still be some trips taking well over 50 minutes. I was on one that took 54, with a two-minute wait to pull into UWS. So, yes, my math is optimistic.

      The difference in cumulative dwell time between a 2-car train and a 3-car train can be more than a couple minutes over the course of a trip.

      Minimizing trains is going to take additional training beyond the traditional “qualification”.

      1. Shortening dwell times is more than a matter of training operators. Riders also need to be trained, to get them to move towards the doors before the train stops, and to stop running and waving for the train to wait for them to board.

        Too many riders are too lackadaisical when it comes to boarding and alighting. We are a big city transit system now, and both drivers and riders need to get used to that. Rules that apply to 30-minute bus service don’t work for 6-minute train service.

  9. Too bad ST decided on 1500v overhead. TriMet has 26 or so spares in its fleet thanks to recession era cutbacks in operators.

    They run 960v or so on parts of the TriMet system so the 750v units might work on ST lines, but it would require a careful check.

    1. Glenn, this is very important for us to understand. Can you tell us how an electric rail system decides on the voltage of its trains? And what the advantages an disadvantages are of each voltage? Would really help.


      1. The voltage is really a nominal voltage.

        Supposedly, TriMet retuned the substations in a number of locations to put out a higher voltage because in the end it saves them money. They loose less in the resistance in the wire and thus more actual work for the power consumed. The systems on the cars are able to take higher voltages than the nominal 750v given for the nominal wire voltage. The overhead is so severely overbuilt that it can tolerate a bit more than 750v too. If you Google hard enough You will find some references to engineering reports about the section isolator between the 750v nominal voltage system on parts of MAX and the areas where it has been increased to 825v nominal.

        On a basic direct current circuit the power lost in transmission increases with the square of the current. With the electronic static converters, the amount of power draw stays fairly constant for the given demand but the current draw drops as the voltage increases. So, there are some benefits to allowing the voltage to push the upper limits of the selected nominal voltage.

        (This gets complicated because the electronic switching introduces a magnetic component and the losses from that part increase with voltage, but basically the system is simple direct current.)

        So, you set the voltage to be low enough to not fry the equipment, but high enough that you aren’t loosing a bunch of power getting it to the train through the wires. In the case of MAX this has become 825v nominal, except apparently (according to one LTK report available online) for a few substations where the main transformer from the power company doesn’t have the correct power taps to allow the increase.

        750v is nice as a selected nominal voltage because it is what all the recently built light rail lines are doing these days, but it increases the cost relative to ST’s 1,500v because the lower voltage needs more substations. TriMet has a substation around every two miles or so. This means not just buying the land and building the substation, but paying the utility connection fees and maintaining it.

        If you have a station every mile or so like TriMet does, it isn’t so hard to add a substation to station construction. Northgate to Everett with several stations would mean a bunch of substations with no additional infrastructure for them to be built into. You’d have to weigh the costs of all those extra substations with potential benefits of using a more common voltage.

        So, long term the choice of 1500 v for ST’s lines is understandable, but it does mean dealing with the limits of a system that isn’t as common in North America as 750 v.

      2. Mark, a 1500-volt system requires fewer substations along each line. The higher voltage pushes the juice farther than a 750-volt system does, so power stations can be farther apart. Fewer substations translates to lower costs. It’s really just about that simple. Other nations have 1500-volt systems; it’s not all that unusual.

      3. Mark, you’d be familiar with this example: the South Shore Line and Metro Electric in Chicago use a 1,500 V DC electric system. Many interurbans used this voltage for reasons already stated.

      4. Or higher.

        Spokane and Inland Empire Railroad used 6,600 volts AC for its longer distance lines. Some of that was reduced need for substations but some of that was some of their customers (including at least one paper mill) purchased both freight service and electric power from the railroad, and the overhead contact wire was also used as the transmission line.

  10. How firm is our information about the time-frame for each phase of upcoming changes? Especially, what part of our system is going to be blocked or slowed when? The more accurately and definitely we know that, the easier it’ll be to react.

    Also a good idea to make every part of the plan as flexible as possible, including, at the beginning, day by day, or hour by hour. Unexpected situations are hard enough to handle now- something it would be good to start addressing now.

    Our communications are terrible, and under stress could see to it the whole region is blocked. From what I’m hearing, this isn’t a problem of machinery and computers, but lack of training and coordination. Many if not most slow-ups in the DSTT now stem from this exact trouble.

    But above all, it’s already going on forty years past time to eliminate oldest and worst problem of all. However and whoever, temporary or permanent, a project like this has got to be under the control of one agency. With a leader (preferably only one) who’s familiar with every technological and operational thing about it, as well as being able to administer and lead by example.

    Good “marker” is whether the candidate speaks in the grammatical “active voice.” Not “it will be done.” Always: “I, or whomever I tell, will do it.” Also: “Every decision I sign will be mine.”

    Can think of two people I’ve met in the transit world I’d give this to. Also pretty well know that neither one of them will take it on a bet, like people used to say when PCC streetcars were new. Which history proves is best proof of leadership.

    But really crucial need is for everybody involved, agencies, employees, and the general public to get so frustrated they grab somebody by the lapels and demand they lead. And have somebody good already prepped and primed when pressure gauge hits the red. Meantime, word has to be put around to the politicians that work doesn’t start at all ’til, like with everything from Uber to walk-through LINK, there’s one set of hands at the controls.

    Mark Dublin

    BTW- anybody else getting a beat-up-file icon about “Unresponsive?” Pretty soon whatever’s causing it isn’t going to like my own response at all.

  11. Just decrease the ridiculous dwell times, and you can probably increase the number of pax per hour each LRV can carry a few percent. Instead of dwell times which seem to average 25-30 seconds, have dwell times of only 10-12 seconds like Toronto’s subway has.

Comments are closed.