Jarrett Walker had a terrific in-depth post last week about pulses (AKA ‘timed transfers’) and how 20 minute service could ever be worse than 30 minute service. I was mulling over the concept when I realized how familiar the situation sounds– turns out Sound Transit is doing exactly that for its Issaquah-Seattle 554 express route.
Metro designs a lot of Eastside routes on clock-face headways, where you have regularly timed departures after the hour. You can tell how this would be pretty ideal for a typical configuration of a local feeder route on 30 min service with a frequent trunk route on 15 min service. That way, as long as you make your local bus, you typically don’t have to wait more than a few minutes to make your connection.
Lengthening headways on your trunk route from 15 minutes to 20 minutes drastically reduces the quality of the feeder-trunk transfer. If connecting from a local route, it turns two good transfers an hour into one– and that’s not something that makes transit look very attractive. Eastgate, in particular, seems to be a hub that Bellevue is eying for future growth, so any reduction in service there is going to be disappointing.
To be clear, the 554 is a trunk route so I don’t actually think 30 minute service would be better than 20 minutes. Trunk service usually has to be maintained at the highest-level frequencies you can afford, and I fully understand that 20 minutes is the best we can get right now. That being said, however, if we do ever get out of dire straits, achieving synergy will be one of our better bets at making transit more viable.
One of the things that drives me nuts about eastbound 554’s schedule is that it is poorly timed with meeting westbound 550s on Mercer Island. During non-peak hours (when 555/556 don’t operate), a 550/554 trip would allow me to get from work back to Issaquah more quickly than riding the excruciatingly slow 271… except there’s always a sufficiently long layover at Mercer that the running time savings for the 550/554 option are negated.
Would like to see 555/556 chopped in two and realligned when funding is better. An all day route connecting redmond, overlake, bellevue, south bellevue, eastgate, and issaquah would be nice.
RapidRide Line B will cover the Redmond, Overlake, Bellevue section fairly well.
except its not “RAPID”…itll take 40 mins to go from Redmond to Bellevue
554 and 545 are critical routes for those living on the eastside, they just dont get the love they need on the weekends. Will be interesting to see how that changes once tolls are started. I predict some higher loads on 545 and probably 554 as well.
In the case of the 554, I’ve had that route on a short list of routes I’d like to see kicked downstairs to the tunnel, when space becomes available (and I won’t launch into my predictable rant about why the 101/102/150 still come downtown at all). Down-stairing the 554 might bolster its ridership enough to justify keeping the 15-minute headway, and save a little on the time to perform a run.
Transfers become clunky not just due to bad timing, but by spreading out the places where one might attempt to make the transfer, with deciding whether to wait on 2nd Avenue or in the tunnel being one of the most common conundrums.
If the 150 is going to keep going downtown for the rest of eternity, I think it makes less sense as a tunnel route than the 554 and all other remaining eastside routes do. Unlike with the 101/102/106, there is no exercise in waiting to see if Link or some other bus comes first. So why not swap the 150 and 554 between downstairs and upstairs?
Probably will happen when U link is up and running…but it will only be till the busses get evicted from the tunnel.
Does the tunnel really change ridership that much? It’s never affected my choice of route. Of course it’ll all be immaterial once u-link opens.
Prior to the 316 being placed in the tunnel it was a big factor in how I got home. Do I choose to stand outside in the rain/dark and how long do I wait before deciding the 316 already passed or is a no-show before walking back to the tunnel from 3rd & Spring? If I caught the surface 316, it was often a faster ride home since the transfers at Northgate TC often left me stranded for 10-20 minutes waiting for the 5/16/345/346 to get across I-5.
Now that the 316 is in the tunnel I go straight from the office to the tunnel (don’t even need to go outside). The 316 schedule is fixed and is never more than 5 minutes late (usually a train delay). I can make an easy decision whether I’ve missed it based on time or whether I see fellow riders on the platform and hop the 41 and hope for a short transfer at Northgate. I have also noticed many new riders on the 316 in the last 10 months that it has been in the tunnel, the ride time was significantly shortened by the 16 stop consolidation and the tunnel change. I actually can sleep in half an hour longer thanks to the shortened ride.
Consider if you’re going Eastbound to Mercer Island or the Ranier freeway stop. Ideally, you could take either the 550 or 554, whichever comes first. But one runs on the surface and the other is in the tunnel. I’ve been kept out of a 554 after an evening Qwest Field event when the bus was crush loaded and couldn’t take on about twenty people.
When I was showing my younger brother around Downtown Seattle for the first time (right before I got my motorized wheelchair and just before the opening of Central Link), I usually used the tunnel to do in-Downtown trips. Saved massive headaches over steep hills.
Yeah, I know buses are going to get the boot, but in the meantime, I try to pick routes that go into the tunnel when going into Seattle.
With the lost of the 564 to South Hill, when I go see my Dad on weekends, I usually take the 230 to the 550 to the Sounder to the Tacoma Link to the 3 to the 212 (and get picked up in Steilacoom) to get to Pierce County (I’m not stupid enough to take a bus down I-5 on a Friday evening). It’s much easier, being in a wheelchair to get off at IDS, take the elevator to street level, wheel over to the Weller Street Bridge, and take the elevator down to the Sounder platform.
Hang on, let me get this straight… You are going to South Hill and you take Sounder, to Tacoma Link, to 3, to 212…But that only goes to Lakewood Towne Center and you would need 410 to get to South Hill. Instead, you could get off Sounder at Puyallup and take the 402 up the hill to get to the Transit center…
Unless for some unfathomable reasin you want to spend that much time on transit.
Nathan: I’m coming from Microsoft and going to DuPont. Used to take the 230 to the 564 to the 410 to the 204 to the 212 (the old 585 would’ve really helped, but as it was axed before I started doing that route)
Now I take the 230 to the 550 to the Sounder to the Tacoma Link to the 3 to the 212.
My Dad picks me up at the Steilacoom ferry landing
It’s all about avoiding southbound Interstate 5 on a Friday evening
Oops (a comment editor would really help). The 230 was only when I was coming from my house.
* Took the 564 straight from Overlake TC
* Now take 566 from Overlake to 550 to Sounder
The logic of it all… The way this is laid out it ensures connection FROM the 30 minute headway bus to the more frequent bus but not in reverse. If the pulse is wide enough it can allow two way insurance of transfer. I’m just working with the numbers presented since not all transfers are done at a park&ride. On the street you have to hop off one bus and onto another, they’re not both in the same spot at the same time so having one bus come 3 minutes before the other only works in one direction.
Going from a bus with 30 minute headways to 15 minute headways
Transfer at 57 after = 3 minutes
Transfer at 27 after = 3 minutes
Reverse route – Going from a bus with 15 minute headways to a bus with 30 minute headways
Transfer at 00 = 27 minutes
Transfer at 15 after = 12 minutes
Transfer at 30 after = 27 minutes
Transfer at 45 after = 12 minutes
Average transfer time (3 + 3 + 27 + 12 + 27 + 12 = 84 / 6) equals about an 14 minute average transfer time.
Going from a bus with 30 minute headway to 20 minute headway
Transfer at 57 after = 3 minutes
Transfer at 27 after = 13 minutes
Reverse route – Going from a bus with 20 minute headway to a bus with 30 minute headway
Transfer at 00 = 27 minutes
Transfer at 20 after = 7 minutes
Transfer at 40 after = 17 minutes
Average transfer time (3 + 13 + 27 + 7 + 17 = 67 / 5) equals about an 13.4 second average transfer time.
If I run out of my house and catch the first bus at somewhat random intervals my transfer times will be the same either way. If we’re only talking about transfer stations and the pulse is wide enough it changes everything and the 15 minute headway scenario pulls ahead substantially as long as you always have time to make that transfer and buses are never late. This never happens.
Having said that I still prefer 15 minutes headway buses connecting to 30 minute headway buses because of consistency. Consistency plus convenience equals ridership and the more often a bus comes the better given enough money to support it.
If my math is wrong I apologize, I went over it a couple of times and found errors, they may not have been the last.
Yeah good point.
This is one of those situations where it depends on the context. The question becomes is the transfer heavily directional or not. If transfers are even in both directions it is ideal to have the feeder service arrive before and leave after the trunk route. If transfers are heavily directional though it usually is best to have timed transfers in the heavier direction (towards downtown in the morning, away in the afternoon).
So to build off what you calculated above, I would multiple the waiting time by the number of people making that transfer. Then sum those numbers up and you will have a more full picture of transfer delay by scenario.
And this is why “timed transfers” should be the pure domain of suburban commuter-oriented service and have nothing to do with urban connectivity
This was covered recently on another thread, but I think the distinction needs to be reinforced. In the city, core routes with truly high frequencies, bi-directionally, need to be achieved.
I frankly don’t think we should even have the same planners working on urban and suburban route structures, lest the (worthwhile) lessons of one be (destructively) applied to the other.
There’s a couple of things here that deserve mention.
First, it is very, very hard to measure transfer patterns effectively. I cannot tell you how many times I have seen transfer connections set up but no one knows if the connections are actally used.
Timed transfers are for more than “suburban commuter-oriented service”. You need to take away that “commuter-oriented” line, as close to 50 percent of transit trips (varies by system) are not work related.
From ORCA data, it should be possible to see if specific transfers are being made. I do hope that the planners will look at all that data they’re collecting. Anonimized, of course.
Grant,
If it were a true timed transfer or pulse, the 15-minute service would probably arrive at 12, 27, 42 and 57 and the 20-minute service would arrive at 17, 37 and 57 after the hour. Likewise, the 30-minute service arriving at 57 and 27 would depart at 00 and 30.
I used to take a bus in San Diego that ran in 15 minute intervals directly from UCSD to a big transit center (Old Town). At the transit center, there were two trolley lines that were co-linear, but terminated there, so you would have simultaneous departures East and South. The trolleys were also run on a 15 minute schedule. Now, the express bus continued to downtown, stopping near the stops of the other trolleys.
It was extremely frustrating as when the bus was on-time and encountered no traffic (about 80% of the time), the two trolleys would be leaving just as I entered the transit center. It was completely mistimed. If anybody wanted to go Downtown from Old Town, they could take the trolley. So everyday, there was a seemingly unnecessary 15-minute wait. I wonder if the bus had been planned in a pulse from it’s origin, which was about ten minutes from UCSD.
It brings up the point that transit planning depends on the priority levels that are chosen for the lines, and which points the planning starts from. If they shifted my bus times back 5-minutes, then they might have had to change the timing of another 10 lines that pass through the origin of my bus. It must be hard to completely revamp the time-tables from a centrally planned perspective.
Ever since the last Metro shakeup, my transfer time on MT 181 (Twin Lakes P&R) to ST 578 (Seattle) at Auburn Station has grown from 20 minutes. I realize that smooth transfers aren’t always possible given 30-minute headways on the feeder line _and_ the trunk line, but still… It’s uncomfortable. I just hope that connections were improved elsewhere in the system.
Oops. Make that “has grown from less than 10 minutes to over 20 minutes.”
how about some comments on the original Lee post? Does the Walker piece implicitly criticize the ST management and Board decision to cut service and provide odd headways? was not the promise of the regional express network one of high frequency trunk lines? could ST not provide 15-minute headway service all day on routes 522, 545, 550, 554, 566, 574, 577, and 594? how much slower would east Link be provided? the regional express program uses a relatively small portion of the ST budget. The STB shared the budget briefings in the fall.