Photo by Mike Bjork

ST’s new 1st quarter 2010 ridership report is out.  As ST was reporting monthly Central Link ridership through March, there’s no new information on that front.  For other modes, though, from 1Q 2009 ST Express ridership was basically flat, while Tacoma Link was down 5% and Sounder declined about 7%.  Overall, ridership was up 36% thanks entirely to Central Link.

However, Link’s on-time rate was a horrific 71% for the quarter, compared to an ST goal of 90%.  An on-time trip, as I’ve complained before*, is defined to be one that leaves its terminal station less than one minute late. ST spokesman Bruce Gray explains that the 90-day quarter had a total of 35 days of maintenance work (mainly for noise reduction and WSDOT work); thanks to the limitations of the measurement system, all trains that deviated from the printed schedule (or weren’t run at all) were counted as late. Gray adds that ST is “not happy with the 70% rate” and on-time results have “gotten a lot better over the last month or so” as a lot of that maintenance work has wrapped up.

What’s more, Gray added that even under these conditions, 90% of trains were spaced within three minutes of their published headways.  A maintenance delay that slowed everyone on the line, or truncated it at Sodo, didn’t necessarily mess up the spacing between trains.  Indeed, I’ve argued before that when headways are short this is actually a much better measure of reliability.

Maintenance doesn’t explain all the late trips.  The report also blames bus/train operations in the tunnel, which is unique in the world, and therefore a source of schedule risk.  Ben’s going to have a more in-depth report on reliability next week.

*In that post, I cited a 99.5% reliability report for Central Link’s 3rd quarter, although no number was available in the 4th quarter.  Gray wasn’t prepared to elaborate on the origin of the 3Q number, although he understands that what data was gathered in 2009 was not very reliable.  Indeed, public Link schedules didn’t even exist until the 4th quarter, so it’s not clear how these metrics were defined.

62 Replies to “ST 1Q Ridership Up, Link Reliability Poor”

  1. If there are a lot of them and they come frequently, then I could give a rip if they’re late, because I don’t even need to see the schedule.

    1. I think that thousands of people would agree with you, but we need some sort of metric to determine the success of the agency.

      1. For success of LINK, you could measure passenger load vs that of the bus route it replaces and time of travel same route. Add in cost of equipment but break out operations costs and capital equipment. Then for that passenger load counts show what the equivalent bus/cab level of service would have cost.

        Then you’ll know whether the agency made the right choice…

      2. Add one more thing, don’t just look at the route is replaces, look at the total ridership in the region it serves before it came online and after.

        Also, when calculating bus/cab service cost, add in road maintenance costs, and adjust for increased quality of service (proof of payment, spacious waiting area, etc.)

        Even with these changes, we will only know if LINK was a good idea in 10/20 years time.

    2. There may be frequent enough Links to meet some theoretical definition about not needing a timetable, but I still miss my bus if the 6:27 am Link going north from Columbia City is late.

      1. My bus usually pulls from the holding area into the International District/Chinatown Station right after my Link train clears the station. So the margin is obviously tiny.

      2. Yikes. If you can make that on a somewhat frequent basis, then I’d say ST is doing pretty well. I try to keep everything I drive between 1-5 minutes late to allow folks to make connections. One exception would be starting out in the tunnel: Those trips I try to run right on time to avoid getting overloaded although I do drag my feet when the load is light and there isn’t anybody in view behind me.

        Since trains have priority, starting any bus route in the tunnel is an exercise in blind faith. Sometimes I’d arrive at Bay D (550) and run into a red light since the train is pulling out. Other times I’ve got a yellow and can sail right through at a pace I have control over. The difference, for the 550 at least, can result in a clean well-timed run through the tunnel vs. running 5 minutes late and having people jam onto my bus making me even later. My old tripper, 550/29, had a trip running back to Bellevue at 4:04 that could have been a case study in bus/train scheduling along with “bus bunching”.

        Keep in mind that Metro trains very little on this subject other than “pull out on time” and “don’t run early”. The rest we figure out along the way.

    3. The frequency is one thing, but the duration of the trip is another. If the trains were all taking 20 minutes longer than they claimed on the timetable, but they had extra trains so that the headways were still at whatever it’s supposed to be, the delay wouldn’t show up in the headways / wait times, but the train taking 20 extra minutes would still inconvenience people. It’s one thing to be at the station within 2 minutes of when they’re supposed to be. It’s quite another to get from Seatac to Westlake in the 32 (or whatever) minutes they claim it takes.

      1. The scheduled time between Seatac and Westlake is 38 minutes, most of the trains can manage it with less of a delay than 2 minutes (depending on the operator and MLK traffic.)

        The only times that the trains take “20 extra minutes” are when there is an emergency on the Right Of Way, ie. accidents, coach breakdowns, police actions. These are inevitable, the only difference between the trains and busses are that the trains cannot re-route in most cases. When we have track maintenance, we have to take trains off the line and reduce headway to accomodate the fact that the trains have to take turns in a section of track. A great deal of planning goes into reducing that headway and timing it so that the trains don’t sit idle mid line (at a station.)

        We ONLY add trains when there is a high volume crowd from the stadium (Sounders games) and we rarely, if ever, have more than the 12 trains in service during peak hour 7.5 minute headways.

    4. Good point. I can’t say I have ever looked at a Link schedule. You’re dead meat if you don’t have a timetable for most bus routes, particularly during off-peak hours.

      Which brings up another subject: while Link headways remain fairly constant, and the Westlake -> Airport trip is almost always 35 minutes (and will always be 35 minutes or less) my bus route got longer and longer with each passing year…until the recession hit. I noticed that before each service change the routes I used would be 2-3 minutes late – consistently. But when the service change occurred, the planners would simply add another 2-4 minutes on to the schedule. Which shows that reliability can be “fixed” for buses – but the same trip takes longer and longer, illustrating yet again how Bus Rapid Transit can be such a contradiction in terms. It gets less rapid and more expensive with each passing year.

      1. The single largest delay for buses, outside of traffic, is people fumbling for their fares. It doesn’t seem like a lot of time on an individual basis, but when you watch 20-30 people load and pay with cash, you realize quickly the need to give people an incentive to use ORCA. (Washington DC’s system provides an incentive to use their SmarTrip card by giving a discount to passengers who load larger amounts onto their card.) Off bus payment systems, like those used for Swift and eventually RapidRide, will help speed boarding even more.

        More passengers = more people fumbling for fares = more delays. With true BRT, not BRT-lite like ST Express or even RapidRide, those delays are minimized so you won’t see as much schedule degradation. 3 doors and better bus layout will also help quite a bit.

      2. The average trip time on Link between SeaTac and Westlake is about 40 minutes. Don’t know where you got 35 minutes.

        The average trip time between SeaTac and Westlake on the 194 bus was about 30 minutes — or 10 minutes faster than Link.

      3. A citation is needed for your 40 minute figure. For now, I’m much more inclined to listen to ST’s 37 minute figure.

        Also, if you want to play the find-one-bus-route-that-is-a-little-faster-then-link-when-there-is-no-traffic-and-you-have-a-reliable-schedule-in-hand game, then factor in headways, reliability, comfort, and safety into the equation.

      4. 4 out of 6 trips I took in the last month between Westlake and Airport incurred delays in the DSTT, one had an operator change, and one had extended dwell time at Rainier Beach inbound. I figured my average to be 39 minutes, 2 minutes more than the scheduled time.

        Using John Niles’ numbers (which I suspect Norman to be the “observer”) we get an average of 39 min 58 sec = 40 minutes, excluding the 3 trips affected by construction/maintenance, we get an average of 39 minutes.

        That leads me to simplistically think that most of the delays are caused by traffic in the DSTT.

      5. And delays along MLK Jr Way. I have been on trips where Link trains stopped at five red lights, which obviously adds a few minutes to that trip. Changing operators usually adds about one minute.

        The one bus route, Chetan? The 194 between Westlake and SeaTac just happens to be the bus route which Link replaced. The 194 was almost always under 30 mintues on that trip. I would say the most common time between Westlake and SeaTac was 28 minutes on the 194. The few times it got delayed more than a few minutes is what raised the average trip time to about 30 minutes.

      6. Except you always conveniently forget that the 194 only served downtown and the airport. Link serves many more locations and thousands more people than the 194 did.

      7. My hunch, based on unscientific observation, is that Oran is correct and the DSTT is the most unreliable segment of Link today. I suspect that various artifacts of dual operations are much of the issue. Maybe as the system is extended and the DSTT becomes rail-only reliability will actually improve.

        The other minor delays I’ve experienced have been due to operator change and signal operations in the Rainier Valley.

      8. Zed: there was no bus that ran on the Link route along MLK with only 3 bus stops, about a mile apart, like Link. So, what route would you compare Link to? A local bus that stopped every few blocks? Is that a fair comparison, in your opinion?

        The main reason Link may be faster than a bus along MLK Way is the very small number of stops Link makes. You could have a bus route along MLK Way with the same number of stops as Link, and that bus should run just about the same speed as Link, especially if it were a SWIFT-style bus with off-board payment, non-tie-down wheelchair area, on-board bicycle racks, etc.

        The other point being that with buses you can have several different routes, instead of the one Link route. You can have the 194 express between downtown and the airport, which was significantly faster than Link; then a local bus down MLK Way, with stops closer to where people live; then an express down MLK Way, similar to Link; and some of those buses down MLK could continue to Tukwila and the airport, for the few people who ride Link from MLK Way to Tukwila or SeaTac. All those bus routes combined would have cost a fraction of what Central Link cost, and you would have a much faster trip between downtown and the airport.

      9. Seven of my last 20 trips on Link had delays along MLK Jr Way. Some of these trips had only one delay, and some had multiple delays. Usually for red lights, but sometimes for, “the train is being held due to traffic ahead.” You have heard this recording on Link trips, have you not, as the train sits in one spot for a minute, or several?

      10. Norman,

        The 42X also had limited stops along MLK on its way downtown. A few more than Link, though.

      11. I’ve heard the “train being held due to traffic ahead” message most often in the tunnel, especially when entering Westlake or IDS, very rarely on MLK.

        Add two more data points to the Westlake-SeaTac Link trip. 36.3 and 35.1 minutes, dragging my average down to 38 minutes. No tunnel delays, no MLK delays, plus a wheelchair user and people from the airport and going to the Sounder game filling every seat, baggage accounted for.

      12. My last two trips, leaving Weestlake at 3:00 and leaving SeaTac at 3:42, took 38:40 and 40:15, respectively. The first trip had no noticeable delays — it just took 38:40; the return trip had a long dwell time at Rainier Beach, and a long dwell time at Stadium station, and thus took 40:15. Neither stopped at any red lights along MLK Way, yet both were over 37 minutes.

      13. That was probably a driver slowness issue. That really bugs me on the inconsistency of the train operators. They’d fail miserably if they had to operate a train in Japan.

        At first the priority area seats were up for luggage, when more people boarded along the way they took the seats and moved luggage out into the aisle. I guess 7 seats were taken by luggage, excluding priority area. It came to the point where there were so many people standing, it really didn’t matter.

      14. Now for the bad news.

        It seems that Metro was running the post-game shuttles in the tunnel (I thought they cancelled those but the buses were just signed “S Kirkland P&R” or “Northgate”). It just takes forever for those things to load. Buses and trains were backed up in the tunnel. It took me 9.5 minutes to get from IDS to Westlake.

        Also another trip from Westlake (2:24) to SeaTac/Airport, packed with Sounders fans, took 51 minutes. The train sat at Columbia City for about 10 minutes. The operator never announced a thing, not even the “traffic ahead” message. That really drives me crazy when riders are left in the dark. Northbound trains were running normally. When I got to Airport I found that one of the doors wasn’t working (Car #119). Don’t know if that was the cause of the delay. The following train arrived shortly.

        The return trip from SeaTac to Westlake took 36.7 minutes. No delays but I felt the operator was dragging it on MLK. The trailing train followed it back (bunching). However, the next SB departure from Westlake was on time (4:18).

        I think I’m going to get into the habit of filing complaints with Sound Transit more often. I see it this way: the more citizen complaints they get, the more pressure is on them (and partners Metro, SDOT, etc.) to get things right. The purpose is not to make light rail look bad but to make it the best it can be. If you agree, I urge you to do the same.

      15. “I think I’m going to get into the habit of filing complaints with Sound Transit more often. I see it this way: the more citizen complaints they get, the more pressure is on them (and partners Metro, SDOT, etc.) to get things right. The purpose is not to make light rail look bad but to make it the best it can be. If you agree, I urge you to do the same.”

        I agree. There was no excuse for the lack of an announcement/apology when your train was sitting at Columbia City. That sort of thing is really, really bad for PR. You should complain when that happens.

      16. I have to agree with the DSTT being a contributing factor for the LINK delays. Now the solution isnt as easy. I think a two part approach is required here.

        First, Moving all the motorcoaches to the surface. Having the tunnel closed for 2 years(?) for construction showed us that it was possible, and practical.

        Second, Eliminate the RFA, except in the DSTT from Westlake to International District. Not only does it bring us in line with Portland’s fare policy, it will help with the fare evasion problem.

        Now, the real catch will come, how much slower will that make the buses in downtown seattle? ORCA isnt exactly fast, although its accurate at reading the cards. changing the fareset, any fare disputes will slow things down, but with most people using a pass it shoudlent add too much time.

        I think we should also go to PoP on all buses, and upgrade our fareboxes with TRiM units, or intergrated transfer printers (magntically encoded like Vancouver BC, etc.) for cash fares, and re-enabling inter agency transfers again. This way the transfer doesent lie when checked, and we can also install orca readers at the rear doors on routes for quicker ingress/egress. Also would help cut down on fare evasion.

      17. I will agree on the operator consistancy. The city of Seattle requires that we run a minimum 7 minute headway on MLK. If, during off-peak 10 minute headway, a train is running 4 minutes behind, it causes it’s follower to have to wait an extra minute. Watching my board, I can usually identify the operator by the space between them and another train.

        Bus “Pay as you enter” in the DSTT is the worst cause of delay after sporting events. Imagine 60+ people fumbling around for their bus fare.

      18. Bus “Pay as you enter” in the DSTT is the worst cause of delay after sporting events. Imagine 60+ people fumbling around for their bus fare.

        But the DSTT is in the RFA!

  2. Public Interest Transportation Forum has measured light rail reliability by how much the trip varies in scheduled trip length time.

    We’ve looked at two samples of rides, with the latest results from February and March posted under title, “New sample of 40 Central Link light rail runs indicates delays of 3 minutes or more in about a third of trips, up from one quarter of trips last autumn.”

    Posted at .

    1. One key variable missing here is the standard by which “on time performance” is measured.

      For instance, ST uses a 10 minute standard for their buses and a 7 minute standard to guage Sounder reliability. King County Metro measures “on time performance” using 5 minutes as the marker.

      So, if Link was held to the same standard as John Niles’ terribly unreliable buses, the 71% figure would probably shoot up to 99%.

      But if that was the case, this entire discussion would be moot, and John Niles wouldn’t have much to obsess about this weekend.

      1. King County Metro buses have 76% on-time performance in 2008 and expected to increase to 80% in ’09 and ’10. So far, ST buses are more reliable than Metro buses using the same metric. If Metro’s 2009 projection was reached, ST is still more reliable but remains under ST’s 85% target.

        I don’t know what the distribution of late trips on Link are so I’m not going to speculate.

      2. Less buses on the road (when budget crisis rubber hits the road next year) = more reliability overall, ironically.

    2. Interesting release. I requested reliability data from ST two weeks ago, and was told I’d have it by today. We’ll see.
      I was concerned about your sample size, but it looks like it’s not too far fetched now. I’ll post the number they send me. (trains over 3 minutes late beyond the 37 minute posted trip time).

  3. All the more reason for realtime arrival information. And I mean actual realtime info, not just a two-minute or “train now arriving” warning.

  4. Something interesting that I’ve noticed is that on a perfect stretch where Link doesn’t stop for lights along MLK Way or buses in the tunnel, the times between stations are less, and sometimes much less, than what ST publishes. One station to station time I always clock lower is Tukwila to Rainier Beach where it’s usually 6-7 minutes and not the published 10 minutes.

    I do understand that erring on the side of more time is much more beneficial because of foreseen and unforeseen consequences, I think every transit system in the world does this. So I’m not sure how “early” factors into this on-time equation?

    1. In this equation it’s kind of abstract. But to the rider, it makes the next train seem like the previous one that’s just late.

    2. I have found Tukwila/Rainier Beach to be quite consistently 9 minutes, as scheduled. There is no reason for there to ever be delays on this segment, unless there is maintenance being performed on the tracks, which ST did only late in the evening (after 10 pm, I believe), so that should have affected only a small percentage of trains.

      1. Actually 10 to 1 is about 10% of trains. Weekends also saw a lot of maintenance and accounted for a bunch of off-schedule operations. That’s probably another 20% or so potentially affected.

  5. Given the reality of combined bus/rail operations, printing a schedule only set them up to fail. There are some definite benefits to our unique tunnel operations, but departure time consistency is NOT one of them. There will always be immense variability as long as buses are in the tunnel, especially during times when the RFA is not in effect and it takes 3 minutes per station for everyone to unfold their dollar bills on the bus in front of LINK.

    A couple of suggestions for improved reliability:

    (1. Could they end the practice of stopping at O&M to switch drivers or give staff a ride? In my anecdotal experience this happens every 4th trip or so, and adds at least a minute to the run. Come on, SoDo station’s NOT that far away…!

    (2. I think we could look to operator variability as well. Some operators have much longer station dwell times than others, and some are also very conservative with their running speed even on elevated or tunneled ROWs. There is just no reason to go through the Beacon Hill Tunnel at 20 mph, or under 50mph anywhere between Rainier Beach and TIBS.

    Overall however I think LINK’s frequency makes it more functionally reliable than these figures suggest. I’m so grateful to have it!

    1. Interesting metrics would be:

      1. What number/percentage of riders take the same route, and use the same stops everyday.

      1a. What percentage of these people are taking a non-optimal route, because they don’t have the information/map/web access to inform them.

      2. What number/percentage of riders are taking a unique trip…like a job interview…where they had to plan, or research the ride which they may never, or rarely take again.

      2a. What means did they use to plan: map, computer, phone, friend.

  6. Has ST given any reason for the 11% drop in Express Bus on-time performance (94% YTD 2009 vs. 83% YTD 2010)? That’s a rather large decline!

    1. Well if you read the note right next to the table, you’ll see they changed the definition.

  7. I’ve noticed lately when I ride Link in the afternoon and get on board at Mt. Baker Station, there will often be two southbound trains passing through before my next northbound train. This may be bad for reliability, but I think it is clever usage of the space in the tunnel beyond Westlake to queue up more trains and provide more frequent southbound service during the afternoon rush.

    Shhh. Don’t tell the sports fans that those five southbound trains in fifteen minutes are blowing ST’s on-time performance.

  8. Those of you pining for a return of the 194 with its limited baggage capacity and endless pay-as-you-exit queues at Sea-Tac are welcome to patronize the private-enterprise that is the Gray Line Downtown Airporter:

    Certainly many of those that attack LINK would rather that all such transportation services be run for profit and charge $1.32 per mile.

    1. The Airporter is only $11? Considering it probably never takes 51 minutes because there’s a Sounders game, or gets delayed for an hour because a car turns into it illegally, I might well be willing to give up a pitcher of beer to not have to deal with the stress of taking LINK to the airport.

      And how are you getting $1.32/mile? Google Maps tells me the Airporter route is 14.5 miles long, which at $11 is $0.75 per mile. Which is the same rate you pay to go 3 miles on Metro at peak, for what it’s worth.

      1. About the Airporter… indeed, it is $11, but times vary depending on which of the eight hotels you depart from. If you get on at the Madison Renaissance, it can be a 55 minute journey to SeaTac, as that is the first of the eight stops downtown. IF you depart the Westin, it is 34 minutes, and from the Warwick, 29 minutes.

        FROM Westlake Mall, the old 194 was about the same speed as from the Westin. HOWEVER, the 194 came in at door 00, Arrival (Baggage Claim). So if you were flying United or Alaska, you got to hike the length of the terminal. With Gray Line, there are TWO drop points on the UPPER deck, AKA Departures/Ticketing. so you need only bring baggage from the front door nearest your airline.

        Link offers a great service, but WOULD have been even better if it had looped in under the airport with a station at each end, rather than the straight shot stop 1000 yards east. Now that it is built, I would like to opt for the Pedicab or Golf Cart transfer service… For the elderly cruise ship passenger with bags, a very long walk… especially when once you get to the terminal you find your airline at the OTHER end of the run.

        (yes, I know, they should walk after all, but point is you have almost a million transits with the 223 cruise ship sailings this summer, this ties back to being tourism friendly…)

      2. and of course all of the Airporter Runs, along with the 194 of old were sensitive to traffic issues, of which Link is far less susceptible.

  9. Last week Sounder’s arrival time at Kent Station drifted upward 3 minutes.

    I checked the schedule and it had changed.

    There appears to be more slow stretches on the ride back from Seattle, where the train slows down, or stops.

    Maybe BNSF is picking up business?

  10. I live two blocks away from Rainier and Henderson, an oasis of public transit options, and work down town. When I moved in, I had a blast working all of the routes and figuring out which ones worked best. When I first moved in I agonized a lot over train vs bus. It’s a mile walk to the Rainier Beach light rail stop, but the trip is faster to westlake than any bus. After timing it a for a week I realized it’s a wash, train is more walking and more comfortable. Bus is less walking, less comfortable, very reliable. So when the weather’s nice I train, when it rains I bus. Coming home I love it, because I just wait at westlake and which ever one shows up first I take. When I lived in Columbia City, my route was shorter, but waiting for either the 34 or 7 was beyond frustrating.

    Props to the drivers of the 106, I can show up at my bus stop at 6:44 every morning and never miss the 6:45 bus! Reliability of pick up time is like gold to me.

  11. “An on-time trip… is defined to be one that leaves its terminal station less than one minute late.”

    Could you imagine if that was Metro’s definition of ‘on-time’? Their on-time rate would be something like 5%.

  12. the idiots who run red lights on MLK and hit the trains certainly do not help! But the tunnel is a real problem with delays. Buses breaking down, wheel chair riders needing a ramp and then needing to be strapped in slowing the bus down it all contributes to delays for the light rail.

Comments are closed.