The Train of the Future: Paris Metro Line 14 (RMTransit)
How house architecture is changing (Nick Lewis, an interior designer in Vancouver BC)
This is an open thread.
82 Replies to “Sunday Double Feature: Future Train & Architecture”
ST Link ridership data is out for August and September. It shows ELSL ridership continuing to be at, or exceeding, ST’s pre-opening “high” estimate for ridership. And the data aligns with ST statements that ELSL ridership is exceeding projections.
ST Link ridership data for LLE is too early to draw any conclusions from, but we are also hearing statements from some officials that LLE is exceeding pre-opening forecasts.
So what does this emerging trend of ST underestimating ridership mean for the future?
Well…..circling back to that way too long and way too meaningless discussion on interlining and frequency, it means that the previous option of reducing frequency to 4&3@10 wouldn’t even be a viable solution.
Previous analysis showed that 4&3@10 just barely satisfied estimated demand, with maybe just a small demand exceedance in the urban core, But that was with earlier ridership estimates that now appear to be low, and before WFH flamed out and the tech companies started demanding their workers return to the office 5 days a week.
So if the previous option of 4$3@10 becomes unworkable due to higher than projected demand, and the original solution of 4&4@8 is unworkable due to fleet size, what is the next best option?
Increase frequency beyond 8-mins!
Operating both lines at 3@7.5 would exactly match 1-Line capacity under the 4@10 scenario, while significantly improving 2-Line delivered capacity over the 3@10 scenario. Total LRV fleet needs would be slightly more than the 4&3@10 scenario, but less than for the full 4@8 scenario.
Slightly improving frequency to 7.5 mins instead of 8-mins while standardizing train sizes at 3-cars has several advantages beyond just higher capacity and better frequency:
1). Fleet size is less than for the 4@8 goal, but only slightly higher than for the 4&3@10 option.
3). Spares and gap trains would also be standardized at 3-cars, freeing up LRV’s and LRV storage space.
4). Consistent train lengths are good for the customer base.
5). Synchronous scheduling is also good for the user base.
6). Meeting capacity needs while delivering better than promised frequency exceeds expectations with the customer base.
7). Solution is workable today. Waiting for delivery of 10 more LRV’s in 2028 is most likely not required.
However, if ST also significantly under predicts demand for FWLE, then there still might be a capacity issue between IDS and SoDo, even for the 3@7.5 solution. Detailed analysis required.
This is an excellent example of how interlining, combined with fleet size restrictions, can actually lead to improved frequency.
It shows ELSL ridership continuing to be at, or exceeding, ST’s pre-opening “high” estimate for ridership.
ST estimated 6,000 riders on the starter line*. For the five full months since opening day it has now average 5,561.2 riders per day**. So it is basically performing close to what was expected. The only time that rider estimates were way more than expected was UW Link. This was understandable, as the huge increase in ridership was due to a major restructure by Metro (centered around Link) that was not obvious.
The closer you are to the opening date the more likely you are to get an accurate estimate. It is difficult to find original estimates for East Link (if someone has them, please share). I found some estimates for Link as a whole from 2008. These numbers include ST2 (but not ST3). They estimated 286,000 Link riders by 2030 once Link goes from Federal Way to Lynnwood (and across the lake). This seems absurd. In contrast more recent estimates for a full East Link (across the lake and all the way to Downtown Redmond) at around 50,000. This seems far more likely.
Sound Transit expects about 6,000 daily riders on the starter line, but the agency has projected about 50,000 daily boardings on the 2 Line by 2030, by which time it will be extended to Seattle and Lynnwood and to Downtown Redmond.
** Feel free to double check my math. It isn’t trivial to gather the information (you can’t filter by line). But I selected the following stations: Bellevue Downtown, BelRed, East Main, Overlake, Redmond Technology, South Bellevue, Spring District, Wilburton. I then copied each weekday average and then averaged those.
I agree with your other point. Given a limited number of train cars, you can do two things: Run smaller trains frequently, or larger trains infrequently. In general Sound Transit has leaned towards the latter. They could easily run trains down Rainier Valley every 8 minutes all day long, but they don’t. Hopefully they will at least continue to run trains every 8 minutes during peak, but it wouldn’t shock me if they pinch pennies and decide to run longer trains infrequently. This isn’t as good for riders, but it saves them money on operations. Time will tell.
If the trains are automated, the operating cost to run fewer longer trains vs. more shorter trains is very minor, so you may as well choose the “more shorter trains” option, since less wait time makes for happy passengers and boosts ridership. As an added bonus, if the “more shorter trains” is decided back before the line is built, this allows the stations to be smaller, saving an enormous amount of money in capital cost. This is essentially what Vancouver did with its SkyTrain system.
By contrast, the “fewer longer trains” option incurs a worse rider experience and higher capital costs, but, if the line is not automated, it saves the agency a lot on labor cost, since it’s fewer drivers that need to be hired and fewer driver-hours they have to pay for. So, essential the choice towards fewer longer trains is really just fallout from Sound Transit being unwilling to consider automated trains.
Thanks for the heads up on the ridership data being restored to the web page after being pulled down for what I am sure was a well-justified reason.
It is hard to predict FWLE ridership without knowing where the STX route restructure is headed.
But even if the restructure heads toward truncations at FWCCS, it may be advantageous to wait until after the World Cup to do those truncations.
We could also take a lesson from the SCM, and do some price-gouging during the FMWC. Create a long-distance or even regular STX fare of $4 to $6 that takes effect before the FMWC, but accepts the usual monthly passes as full payment. The higher fares could then go away with most of the redundant bus service, probably just for the peak direction, since FWLE totally loses the race with STX 594 off-peak.
But I am glad you agree that standardized train lengths is a desirable trait for any service pattern. The main reason to run shorter trains during peak is to be able to run shorter trains off-peak, and thereby lower maintenance costs, while having more LRVs available for maintenance during the day. Even in the Before Times, I expected train lengths would drop down to two or three cars after the Great Conjunction, depending on line and whether FWLE and LLE were open yet.
I don’t see how you can conclude this. There are two overarching reasons:
1. The bulk of the East Link riders have not begun riding 2 Line yet.
2. Daily ridership is not the same as the most crowded hour of ridership.
I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway. The seats Link DEIS assumed 8 trains an hour (7.5 minute frequency).
I generally feel that all the trains should be the same number of cars. That’s especially true since Link does not tell a waiting rider how many cars are in a train on its arrival signs and announcements (unlike BART and Muni Metro). So I agree that 3 car trains aren’t optimum and should be avoided.
I also think that a little breathing room between trains is good for operations. Late trains are not uncommon and breathing room in the schedule helps the following trains get back in schedule. Of course, the typical rider only cares about when the next train arrives — and if every train is running 20 minutes behind schedule, a rider won’t notice or care as long as the wait is similar.
And since Covid, the peak surges are reduced and the ratio of busiest hour to daily ridership has dropped. Of course, more employers (like Amazon) are going back to requiring workers to appear at the office so that surge will start to appear more pronounced again.
Finally, I will say that the schedules at higher frequencies do not have to align perfectly to the minute nor do they need to be symmetrical. It’s nice but not absolutely required. It can even be better to have a second train on the heels of a first train to quickly pick up any riders that can’t board that first train. Riding subways in places like Chicago and New York and Boston can offer witness to that and many other operations strategies done to deal with crowding issues and delayed trains. It’s a big reason why I wish ST would hire more managers and dept heads from high-volume established systems rather than from a related industry.
Train frequency an important consideration for sure. It’s something that must be monitored. It’s great that you care! I’m only saying that the average daily data trends for a much lower volume segment doesn’t translate into conclusions about the crowding issues on the whole line.
@Al S,
“ I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway. The seats Link DEIS assumed 8 trains an hour (7.5 minute frequency).”
You are correct, 6-min frequency is not on the table. ST previously temporarily ran 6-min headways due to some issues with LRV availability, but they have no plans to do so again. And for a variety of reasons.
ST ridership modeling is also currently based on 8-min frequency, not 7.5. So going to 7.5-min frequency yields a 6.7% increase in delivered capacity over the 8-min scenario. Which is small, but still significant.
But the crux of the proposal is not based on the 6.7% boost in ridership that comes from 7.5-min headways. It comes from the change to 2-Line.
The 1-Line would stay exactly the same. 4@10 and 3@7.5 are exactly the same. Both operational scenarios delivery 24 LRV’s per hour across any given screenline. So no change to the 1-Line.
The difference is on the 2-Line, and therefore also on the interlined section through the urban core.
ST’s current option is to run 3-car trains at 10-min headways on the 2-Line, with maybe some 4-car trains mixed in after 2018 when they get 10 more LRV’s.
3-car trains at 10 mins delivers 18 LRV’s per hour across any given screenline, whereas switching to 3-car trains at 7.5 min headways would deliver 24 LRV’s per hour. That is a 33% boost in delivered capacity on the 2-Line, and an effective 14% boost in delivered capacity in the interlined section.
The capacity crunch is on the interlined segment, so that 14% boost becomes very significant.
Essentially the proposal is to find an operational scenario that has LRV requirements between the 4@8 baseline (which can’t be implemented due to LRV availability), and the 4&3@10 scenario (which doesn’t solve the capacity crunch problem due to emerging higher than anticipated demand for Link).
The proposal does this by providing an effective capacity boost on the 2-Line (and therefore interlined) segments, while delivering exactly the same capacity on the 1-Line. Thus it ends up being a sort of intermediate option, with higher delivered capacity where it is required (urban core), but with lower total LRV requirements than the 4@8 baseline.
The proposal also helps with LRV requirements by having a standard 3-car train length, which is operationally simpler, but also helps reduce the LRV requirement by eliminating the need to hold 4-car spare and gap trains in reserve. Everything would go to 3-car trains, even spares and gaps.
I guess I feel like anything more frequent than 10 minutes doesn’t require specified arrival time each hour. Such accuracy is very much desirable for 15 minutes or less frequency. But once below 10 minutes I’m would be more concerned about avoiding crowding.
For example, maybe only trains that are more frequent only need to arrive in Downtown Seattle between 7:50 and 8:50 in the morning. So an extra train runs a round trip to make that happen. That’s different than running several extra trains over a three hour period.
All this conceptual scheduling is interesting. However, at the end of the day it first is about decent frequency which I see as 10 minutes. After that, it’s all about avoiding people getting left on the platform. And a seasoned rail dispatching staff should get a thorough understanding of how peak demand goes and figure out what of many options works best to keep that from happening. I feel like they shouldn’t be locked into just one option of specific arrivals times and train lengths.
@Al S,
“….I feel like anything more frequent than 10 minutes doesn’t require specified arrival time each hour. ……once below 10 minutes I’m would be more concerned about avoiding crowding.”
I agree 100%. Which is why I find this focus on ultra-low frequencies that some people on this blog have to be so misplaced. Because once you get below the current level of 8-min headways adding more frequency provides minimal benefits. And Link will soon be operating at 4-min headways in the urban core anyhow.
Na, the crux of the proposal is to provide more capacity to alleviate crowding, while still providing a solution that doesn’t bust the bank on LRV requirements. Because we simply don’t have the LRV’s (yet) to solve this problem with brute force.
The side benefits of slightly better frequency and synchronous scheduling are just happy fallouts of the 3@7.5 option. The main goal was to solve the crowding problem without requiring a high number of LRV’s.
once you get below the current level of 8-min headways adding more frequency provides minimal benefits.
That is absurd. There is a direct relationship between frequency and ridership. The relationship isn’t linear — it is a curve as expected. But it doesn’t drop to zero at 8 minutes or even 6 minutes (which is the general standard for metros). This means that there are some people who will take the train if it is running every 4 minutes, but not every 6 minutes. You lose even more if the train runs every 8 minutes. We are not talking about people complaining — these are people who hate waiting so much that they will find another way to get there. If you doubt me just check the research.
Consider what happens every day in every building in the city. People hate waiting so much they will sprint to an elevator and ask that everyone else be delayed a few seconds just for them. And people do it, too! They hold the elevator because we can all relate. This is an elevator, which typically comes every thirty seconds or so. If it isn’t there within a minute people start looking around and complaining, or at the very least wondering what is going in.
By the way, this is different than what Al is talking about. He is talking about a schedule. I agree with his point. Partly it is due to the nature of most of our stations. It takes a while to get to the station. A schedule only does you much good. What is more useful is to see when the next train is arriving as you enter the station so you can decide whether it is worth going fast or not. Because no one wants to get to the platform only to find the train pulling away. Unless you are in some place like Vancouver, but that’s only because they run their trains very frequently. We do not.
“4@10 and 3@7.5”
What does this format mean?
“4@10 and 3@7.5”
4-car trains at 10-minute headways (24 train-cars per hour)
3-car trains at 7.5-minute headways (24 train-cars per hour)
Passengers perceive the difference between 10-minute, 8, 6, 5, and 2-3 minutes. 2-3 minutes is an excellent rider experience like New York or London. 5-6 minutes may be the most you can expect in a medium-sized city. 8 minutes and you wish it was 6. 10 minutes you’re starting to feel impatient, especially if it’s part of a 2-3 seat ride. If the other two buses run every 15 minutes, that’s potentially 10+15+15 = 40 minutes you’re waiting. 15 minutes is the most common on American light rails, and that really feels like the agency/community isn’t taking transit seriously. It’s one thing if buses run every 15 minutes, but the rail is supposed to be the core of the city’s transit system, so it at least should run every 10 minutes or more. 15 minutes is only appropriate for late night or very minor secondary lines.
@Mike Orr,
Of course at some level people can perceive the difference between a 8-min headways and 6 min headways, but the difference in average wait times is tiny (4 mins vs 3 mins), and that isn’t my point anyhow.
What I said was that the higher the frequency the less impactful small changes in frequency are. You enter an area of diminishing returns. And that is a fact. Going from 4-min headways to 2-min headways just isn’t going to have as much impact s going from 40-mins headways to 20-mins.
And, while there is no doubt that frequency has some impact on ridership, at the higher frequencies the effect is small. And the frequency change is more likely to drive changes in cost faster than changes in ridership.
Doubt that? Look no further than CT. When CT increased frequency 20% on Swift Green during the recent restructures the increase in ridership was 3%.
That represents a fairly weak coupling between frequency and ridership, and that weak coupling is true at most transit agencies, at least if they are operating at higher frequencies.
The good news for Metro is that this weak coupling works both ways. Metro has a very good opportunity to cut their costs and at least delay their day of reckoning with their financial Cliff by cutting frequency on some of their higher frequency, lower ridership routes.
The impact to Metro ridership won’t be that large for small reductions in frequency, but the savings in direct operating costs will be much more substantial. And, since Metro is operating at a fare box recovery ratio of about 8%, the risk of entering a transit death spiral is very small.
@Mike Orr,
Oh, and I forgot to mention.
The best way to avoid compounding transfer penalties is to simply avoid the transfer in the first place.
I try my best to avoid transfer penalties by structuring my trip to be on single seat rides. Or by just walking. I just don’t see the point of waiting for a bus that is scheduled to arrive in 15 minutes, *if* it is even on time or *if* it will even arrive at all.
And I don’t think our local bus agencies will ever be able to offer the frequency or the reliability of Link. They would go bankrupt if they even ever tried. So the transfer penalties are sure to remain.
I don’t think our local bus agencies will ever be able to offer the frequency or the reliability of Link.
The RapidRide G runs every six minutes. The RapidRide E runs every 7.5 minutes (midday). The 7 did as well (until the driver shortage). So they’ve done it before.
I try my best to avoid transfer penalties by structuring my trip to be on single seat rides.
Except the vast majority of trips within the city involve a transfer. Link only goes to a handful of places. Various buses make a lot more stops, but the options are still quite limited. They are in every city. You can’t have a really good transit system with a lot of transfers. Consider Vancouver. In my opinion they have the best transit system on the West Coast and the best transit system for a city its size in North America. It is highly dependent on transfers. They have a very good grid that allows people to make simple, easy, quick transfers.
To be clear — I’m talking about making a perpendicular transfer. You can argue that same direction transfers shouldn’t happen, or be rare. But if you do that you are dismissing a huge part of Link. If that is your argument then buses like the 71, 72, 73 and 74 should continue downtown. So should all of the Community Transit express buses (bring back the 400 series and send the 800-series buses to the UW). Fortunately neither Metro nor CT takes that approach and truncates the buses, thus forcing a transfer. That is a trade-off that in turn enables better frequency — for trips involving a transfer or not.
What I said was that the higher the frequency the less impactful small changes in frequency are.
Bullshit. You wrote:
once you get below the current level of 8-min headways adding more frequency provides minimal benefits.
You didn’t write “diminishing returns” or “less and less benefit”. You wrote “minimal”.
Oh, and your statement implied that either we run every 8 minutes all day long (which is simply not true) or that ten minutes is OK in the middle of the day.
But you didn’t make a claim about diminishing returns. I did. I mentioned the curve (in ridership/frequency). Thus the strawman you are are currently slaying was already destroyed. Allow me to beat that thing to a pulp.
Another (more technical) way to put this is elasticity. The between ridership and frequency is not completely elastic. If you double frequency you won’t double ridership. But it isn’t completely zero, either. It is a curve, with higher elasticity the more frequent you are. This is rather obvious. Going from 40 minutes to 20 minutes is a lot better than going from 20 minutes to 10 minutes. No one is arguing that. That strawman is dead, dead, dead.
But that doesn’t mean that going from 10 minutes to 5 minutes (or anywhere less than 8) is meaningless. According to this study it averages .27 under ten minutes. In other words if you go from ten minutes to five minutes you should expect roughly 27% more riders. The key word here is “roughly”. As they make clear there are a lot of different variables. But that is the average. Again, it gets smaller and smaller the more frequent you go. But again — this is ridership. Just because ridership doesn’t change doesn’t mean it doesn’t matter. Fare increases are relatively inelastic, but no one wants to pay more than they have to.
In other urban contexts, rail transit is important for its ability to carry large number of riders per vehicle, and hence per driver, usually by combining cars into trainsets. … This capacity advantage can be relevant in high-volume situations, particularly when frequencies get down to the three-minute range.
(Emphasis mine.) Notice it says three minutes. In other words, once you reach three minute frequency (not eight) there really isn’t much value to riders in running more frequently. Very few people care. Of course some do — the way some won’t walk a quarter mile to a bus stop — but very few do. The problem, of course is that running every three minutes — although great for riders — is expensive.
So agencies make a compromise. What is typical: Six. So much so that it has been published. To copy a comment on Human Transit:
Vuchic’s 2005 Urban Transit text lists the following desirable headways:
– up to 6 minutes for “short urban trips”
– 6-12 minutes is “still satisfactory for urban trips up to 5-10 km”
– up to 20 to 30 minutes “for longer trips, such as regional commuting trips for which passengers use schedules”
Now consider Metro and the fact that they run the G every six minutes. SDOT might have screwed up the project, but the logic is sound. It is what just about any agency around the world would do. No one would question why you would run a bus every six minutes, given the nature of the trip and the (relatively) large investment in infrastructure. For the same reason, running the trains every six minutes — all day long — would actually be quite appropriate.
And yes, things will get better for folks in the north end as soon as East Link gets here. I am very much looking forward to it (as I’m sure many other people are). I just wish more people in the region could experience that sort of frequency (or something to it: like six minutes).
@Ross and Al
You guys are overthinking it. Lazarus doesn’t care about the 8/10 minute frequency because they are on the lynnwood segment and gets doubled the frequency
If you’re 30 seconds late for a bus running every hour, that 30 seconds has actually cost you an hour.
So, with many sparse areas only able to justify hourly service, high frequency can be extremely helpful.
It’s not that they justify hourly service, it’s that American transit agencies underestimate what they justify or don’t have the resources to completely serve their service area.
People are not worried about the average wait time. They care more about the wait time if you just missed a train. Is that a minor inconvenience, or something that makes them also just miss their bus that runs every 15-20 minutes? How much earlier do you need to leave to make sure you still get there on time even if you just missed a train? Or how late are you if you left “on time” and hoped for the best? 15 minutes versus under 5 minutes is a huge difference for most people. 15 minutes for example is the point where your medical appointment gets cancelled and rescheduled.
@Glenn in Portland,
“ If you’re 30 seconds late for a bus running every hour, that 30 seconds has actually cost you an hour.”
Yes, but if I need to plan ahead to take a bus like I plan ahead to take long distance train or an airplane flight, then I am likely to just drive.
And I’m sorry, but if I do need to transfer to an infrequent bus, I’m going to take a little personal responsibility and not try to time it within 30 seconds or less.
The Starter Line’s expectations were very low, so if it exceeds them slightly it’s still very low. I did my monthly Lake Hills trip Friday afternoon and there were four people in my car.
I still believe Lynnwood will have higher ridership than either the Eastside or Federal Way for demographic and geographic reasons. The Eastside has a lot of rich people who feel they’re too wealthy or suburban to take transit, and work-from-home tech workers and employer shuttles. Snohomish County has lower average income and a higher percent of people in service jobs often with unusual shifts. North Seattle is just 15-20 minutes from Lynnwood and has a wide range of destinations.
From downtown Bellevue 20 minutes gets you to downtown Seattle, which may not be a compelling reason to leave downtown Bellevue. From downtown Redmond 20 minutes still leaves you in the Eastside (downtown Bellevue, in particular).
South King County and Pierce County have demographics more like Snohomish County, but Link’s travel times are longer and less competitive with express buses. 20 minutes from KDM gets you to Rainier Beach. 20 minutes from Federal Way gets you to between TIB and Rainier Beach. 20 minutes from Tacoma Dome (with the extension) gets you to Federal Way.
Among the issues of frequency, capacity, and train length — train length is the least important. Higher frequency is a material advantage to passengers: it makes transit more convenient, makes more trips feasible, allows people to do more activities in a day, and generates ridership. Varying-length trains just have the inconvenience of having to rush to the last car if you’re standing beyond it, but they don’t do anything about the wait time or how many activities you can fit into a day.
@Mike Orr,
“ Higher frequency is a material advantage to passengers: ”
Yes, but I think people on this blog tend to approach the issue of frequency from the POV of the “before times” of Metro only operations. Back then frequencies tended to be very low, and going from 30 min frequencies (or worse!) to 15 min frequencies made a huge difference. Whereas going from say 10-mins to 8-mins in the current world is not nearly as significant.
Stated another way, the higher the frequency the less small changes in frequency will matter.
Additionally, the type of route the frequency is delivered in will also have an effect. Small changes in frequency on commuter type routes (where rider schedules are baked in and repetitive) will have less impact on the ridership base than small changes in frequency on routes in the urban core (where a higher percentage of trips are spontaneous).
Stated another way, the higher the frequency the less small changes in frequency will matter.
Of course, but as I wrote up above, that doesn’t mean it approaches zero at ten minutes. Far from it. It basically never reaches zero. Consider elevators:
User studies have indicated that for an office building, a waiting time of less than 20 seconds is excellent and 40 seconds is poor.
Think about that for a second. 20 seconds if fine, but waiting forty seconds is terrible. Obviously you can’t have trains running that often. But if anything you have it backwards. We aren’t used to a really frequent system or even a standard system. Six minutes is as close to a standard as you’ll find. Not for peak, but for midday service. Some systems run more often. Toronto runs their trains every 2-3 minutes peak and every 4-5 minutes off-peak. But Vancouver has 6 minute midday frequency for their core. Some of the branches run less often, but they are quite far away from the city.
This brings up an important point. The shorter your trip, the more frequency matters. If you are taking a train to Portland then hourly service is fine. If you a couple miles then it is the opposite. (Again, the studies back this up. The frequency-ridership curve approaches zero more quickly with commuter rail then it does urban transit.)
Thus the branching in Vancouver is basically fine. The farther out you are from the core of the city the more likely you are to be taking a long trip. This is the basic idea with Stadbahns. They branch outside the city, but inside the city there is very good frequency.
Link doesn’t quite do that. Places like Beacon Hill and Mount Baker — soon to be part of a branch — are certainly within the city. East Link is interesting because Judkins Park is the one station that isn’t in the suburbs. Overall you don’t lose that much on East Link with ten minute frequency. But you definitely lose a lot with the main line (between CID and Rainier Beach). The answer is to simply run the trains every six minutes (midday). Three minute service would be overkill (but still appreciated) for the shared section; six minutes service would also be overkill (but still appreciated) in a lot of the distant areas; but six minutes would be appropriate between CID and Rainier Beach.
This brings up another issue. Our system is very long and ST has been reluctant to have turnbacks. If our system just ran from Northgate to Columbia City then it is likely the trains would run more often. If we had turnbacks then it would be fine to run within the core twice as often, while going out to the suburbs less often. In my opinion the obvious turnbacks are 148th (to the north) and Rainier Beach (to the south). That would mean running every twelve minutes (midday) outside the city and every six minutes inside it (I am counting 148th as “in the city”). But ST simply isn’t interested. There is no focus on frequency. There is no discussion of trade-offs, cost/benefit — none of that.
“I think people on this blog tend to approach the issue of frequency from the POV of the “before times” of Metro only operations. Back then frequencies tended to be very low, and going from 30 min frequencies (or worse!) to 15 min frequencies made a huge difference. Whereas going from say 10-mins to 8-mins in the current world is not nearly as significant.”
People on this blog know what subway frequencies are in other countries and in the top three US transit cities (New York, DC, Chicago). They’ve experienced some of those networks and seen what role transit plays in their societies, and they want the same here.
10 minutes to 8 minutes, or 8 minutes to 6 minutes, is a minor difference but it adds up. Every minute longer is a psychological pressure, and it loses some would-be riders. When there’s a cost-effective solution (automated trains) that can run every 5 minutes even in smaller cities (Vancouver) and ST doesn’t choose it, that frustration boils over.
Non-foamers are even less amused by long waits than transit fans are. When you get in your car you don’t have to wait up to 5 or 10 or 15 minutes before you can start it, you just go. Mass transit intrinsically requires waiting, but agencies/communities should aim to make that as short as possible. Long waits are the #1 complaint about transit: people would rather ride 5 minutes longer than wait 5 minutes longer, because at least you’re moving.
lol you say it’s meaningless and then go ahead and agree on the synchronous frequency which is literally what I was talking about.
> Solution is workable today. Waiting for delivery of 10 more LRV’s in 2028 is most likely not required.
Anyways just glad you weren’t going to pretend those extra trains were arriving immediately.
If you combine the identical frequency requirement along with what Ross said
> I agree with your other point. Given a limited number of train cars, you can do two things: Run smaller trains frequently, or larger trains infrequently. In general Sound Transit has leaned towards the latter.
And the lengthened routes, you guys can easily figure out the conclusion.
And the lengthened routes, you guys can easily figure out the conclusion.
Are you saying ST will use any excuse to run less often? Maybe. But it is clear that they don’t have to. There are enough train cars to run frequently — it is just a matter of priorities.
@WL,
“Synchronous” in this context does not refer to two lines being synchronized. It refers to the scheduling of each line.
A synchronous schedule (also called a clock-face or cyclic schedule) is a schedule where the frequency is selected in a manner to insure that something happens at the same time every hour.
7.5-min headways on Link are synchronous because each train will arrive at the same time every hour. A rider need only know at what time a train will arrive at a given station and the train will arrive at the same time every hour, regardless of which hour it is.
8-min headways are not synchronous because the train will arrive at a different time in alternating hours. A rider needs to know the schedule to know when the train will arrive, because it arrives at a different time in alternating hours.
Any given line can have a synchronous schedule, even if it is a standalone line.
Daily ridership is not the same as the most crowded hour of ridership.
I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway.
Good points, but it doesn’t really change the point Lazarus was making. If we have a shortage of trains cars and too much crowding we have two choices:
1) Run long trains infrequently.
2) Run short trains frequently.
Now consider these real-world options for that idea:
1) Four-car trains every ten minutes.
2) Three-car trains every 7.5 minutes.
The first requires 24 trains an hour (6 trains an hour times 4 cars per train). The second also requires 24 trains (8 trains an hour times 3 cars per train). It is exactly the same! In terms of number of trains cars as well as total amount of throughput, it is exactly the same. The first option saves money, while the second option is better for riders. It is up to the agency to decide which they prioritize more.
Where things get interesting is when you have extra train cars. Assume for a second you don’t have 24 train cars (per hour) but 28 or 30. You don’t have enough trains to run four-car trains every 7.5 minutes. The thing to do is run the trains every 7.5 minutes, but add train cars every so often. So if lots of people leave work at 5:00 then you run a few four car trains then. Or you alternate. The really nice thing is that you can apply them to one line or the other.
This is nothing new, even within our system. Look at the report on Link crowding from 2019: https://www.soundtransit.org/sites/default/files/documents/2020-service-implementation-plan.pdf#page=85. Notice that capacity per train goes up and down. Also notice that ridership goes up and down. They tried (largely successfully) to match the two. They ran relatively frequently and sometimes they had lots of train cars, sometimes they didn’t. In contrast the capacity is steady on the weekends. That is because they were infrequent, and since they were running infrequently they might as well run longer trains.
There are good reasons to run longer trains, with the same LRV throughput per hour during peak. Cost savings is probably not one of them.
If you run 4-car trains all day that are nowhere close to full, then you are running up the mileage on the additional LRVs. That increases maintenance costs 33% for the off-peak hours, which has to be weighed against the savings from having a handful fewer peak operators. It also shortens the useful life of the LRVs, hastening the day ST has to order fleet replacement.
Longer trains can also significantly reduce the number of LRVs available for the maintenance crew to work on during the day.
OTOH, having more LRVs available to work on off-peak means more of them will get a mid-day cleaning, which does not come for free.
There are probably more nontrivial costs and benefits to weigh, but I don’ think either of us have the information available to do useful armchair budget analysis.
For what it’s worth, when MAX started they’d couple and uncouple cars at Ruby junction, turning 2 car trains into 1 car trains during mid-day and back to 2 car trains in the afternoon. By the mid-1990s they decided it was cheaper to just run two car trains all the time than even the very fast coupling and uncoupling allowed by fully automatic couplers.
I did wander from my original point, which was simply that we just don’t know how crowded the trains will be at the busiest times. We have good indications from existing station activity but we just ont know.
I do want to push back a bit looking at initial Lynnwood Link ridership. It’s way below what we were told. The pre-opening press releases said that there would be 47K to 55K riders on an average weekday by 2026. The September data for the 4 stations is about 8.5K boardings or 17K assuming even two-way activity. Adding 2 Line trains and 130th St Station won’t increase demand by 200%. So it’s only running about 1/3 of what the public was told last summer. How can the ridership be at 1/3 and service be at 1/2 (no 2 Line) yet there is overcrowding? The answer appears to be bad math calculations by staff and/or consultants.
That said, there are still crowding reports coming in. Certainly the Lynnwood trains will get crowding relief when 2 Line opens.
Half of September was pre-bus restructure, when Snohomish County commuters had 11 more 1-seat rides downtown, the Swift Blue Line still terminated at Aurora Village, and South Shoreline Station was served by zero bus routes instead of the current five.
October’s ridership numbers will be a better indication of the interim normal.
Half of September was pre-bus restructure,
Yes, and that is why we’ve held off on an analysis of it. We may end up writing something about Lynnwood Link before we get October numbers just because ST has been fairly slow with their reports. But if we do, expect a followup once we have October numbers (for this very reason).
Worth noting: While I expect many of the stations to have an increase in ridership in October, I would also expect Northgate ridership to drop even lower (for the same reason).
There is also the point that UW and other schools started their fall sessions in latter September. UWS and UDS boardings ought to come way up.
There is a major problem with more frequent trains: MLK. Even though trains can run more frequently, they are going to get stopped more often at traffic signals. MLK often gives as much as 45 seconds to get pedestrians across the street (and that’s not counting time for yellow lights or left turn phases that cross train tracks). The MLK signals go for months malfunctioning, adding even more delay.
Right now, the left turns from eastbound Alaska to northbound MLK is triggering all the time with no turning cars — adding another 30-45 seconds of delay for Link trains if they can’t get priority. It’s been that way for many weeks.
Why have trains coming 2 minutes more often on average if a rider will just lose those 2 minutes at a downstream traffic signal? The only advantage is being protected from weather.
Getting stuck at an MLK signal also creates reliability challenges. If the intent is to have evenly spaced 1/2 Line trains headed towards Lynnwood, a slightly lower frequency will make that much easier to run.
The ugly fact is that ST miscalculated the train cars needed by 2026. Three car trains more frequently won’t solve the problem. If anything trains will be slower on average if they are all 3 cars and running more frequently (rather than 4 cars and less frequently) because of the greatly increased odds that the 1 Line trains will experience traffic signal delays on MLK.
I think the MLK signalling issue can be fixed. The problem is, the timing is built off the requirement that pedestrians have to be able to cross the entire street in one go, but doing that without impacting car capacity on MLK requires long light cycles that suck for everybody.
At least at the intersections with stations (e.g. most people crossing the street only need to cross halfway anyway), the light should be designed with shorter cycles where pedestrians cross just half the street at a time. They can also update the ped signals to permit crossing during left turn phases where the cars on the half of MLK being crossed are stopped anyway.
I don’t remember the details, but I know Not Just Bikes has a video where they talked about traffic signals in Amsterdam. Even at big, complicated intersections, they manage to do it with much shorter cycles and less waiting than is the case here.
Let’s talk about the rubber seat coverings.
I’ve seen enough to know where the “non-destinational” riders tend to nap: in the seats closest to the end of the LRVs.
I’d love for ST to run the experiment of installing the rubber seat covers on these seats, and see what happens.
The second video is by Nick Lewis, whom we haven’t had before and is usually off-topic (interior design), but I’ve been following him for a year. Today’s and yesterday’s videos, and a few in previous weeks, are an attempt to branch out to a wider variety of voices and some peripheral issues, so that we’re not just showing the same three people all the time and always looking at subway networks and bashing American transit and zoning; we do enough of that already. I’m hoping the broader scope of voices and peripheral topics is of interest in itself, and we may find some unexpected things the commentariat latches onto and wants more of.
I really like his interior design content and honestly very relevant to urbanism because well designed spaces can make living in them more comfortable. What materials you pick, what lighting you choose, how you arrange your furniture, are you right sizing your furniture for the space, what kind of curtain or window treatments will work in this space, what I should paint my walls, etc etc etc. These are all very helpful in developing your own personal style but also to creating a very worthwhile house to live in.
I also like that he says IKEA is good for interior design as well if you know what to look for and shop smartly rather than brush it off as cheap throw away furniture like many perceive IKEA as. And that not all high end furniture brands like RH aren’t always worth it.
Re Gemini‘s walk-and-talk videos, we may feature them at some point; it’s just that there are higher-priority ones right now. I’ve been watching some of them. There’s the excoriating review of Lynnwood station-area land use, which got a ton of comments in a previous open thread. He’s also done a mixed review of the Beacon Hill station area, the 2 Line Starter Line and the Microsoft campus, and one on the Overlook Walk I haven’t watched yet.
I just wish he’d gather a few more facts rather than guessing on things, not state guesses as certainties as much, and use a stopwatch to time things rather than “I guess it was 10-15 minutes”. We could answer a lot of his guesses about why the station areas are the way they are, Beacon Hill’s history, what Lynnwood is constructing around the station, etc, because we’ve been debating them for years. Still, the video’s show what a typical new rider’s impression is like, and one from Eastern Europe (the Balkans, as he said in the climate video).
For a language anecdode, I said his accent sounded to me like Russian, but that other Slavic languages have similar characteristics so I wasn’t sure. That’s based on me studying Russian in the 80s and 90s, having Russian friends, and visiting Russia in the 90s. But I also visited London and and met a guy from Serbia in a youth hostel. His English accent had several of the same sounds, wordings, and mannerisms I’d thought were Russian — and that’s on the opposite end of the Slavosphere. The Slavic languages diverged later than Western European ones (Germanic, Romance, Celtic), so it’s not surprising that they have a smaller variation.
Downtown Redmond Link Extension progress. A recent Sound Transit video. A couple of good shots of the Marymoor Village and Downtown Redmond station areas.
Would be nice to see much more accurate real time arrival predictions especially as this technology is about 25 years old. There is still a big issue with ghost buses on OneBusAway and all the apps, that don’t show the actual bus and default to scheduled time which on many routes like the 8 is completely meaningless. Even on brand new routes like the G I see this. Also at the beginning of routes can be hard to know regardless with real time predictions since the bus make not have left yet or may have just left and therefore be arriving at your stop in 2 minutes which is little time to react. Its really depends whether the bus does leave its terminal on schedule. Anyhow I write this because schedule reliability seems to be quite bad, sure a lot is auto congestion along the route but there are also many other factors and it would be nice for current riders and prospective riders to have very accurate predictions when the bus will arrive.
Some of that is from the vehicle tracking system, and can’t be helped.
However, I would find the following information helpful, which PDXBus (the Portland equivalent to OneBusAway) shows:
• Is the displayed information just the scheduled information or is it actual data?
• Has the trip been officially cancelled? (PDXBus shows the scheduled time with a red X through the time and the word “Cancelled” under the time)
• How long has it been since the system received location data?
• How long has it been since the program refreshed its feed from the server?
There have been a number of times that I’ve misjudged the time I had available to get a Pugetopolis bus because the information about the last time the location was received wasn’t shown.
In last weeks discussion about branching and frequency, one if the bottlenecks was that reversing trains at Lynnwood is tight.
Then wherever the 2 Line reverses (assuming the 3 Line goes to Downtown Everett) a whole other reversal challenge is introduced. That’s because 2 Line trains would be reversed into tracks where 3 Line is running through.
So to me that suggests branching the line north of Lynnwood.
The idea of having Everett and Paine Field as separate branches appears to be a political non-starter.
But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied. Perhaps ST should explore this.
Some of the advantages:
1. The branched ends makes it easier to blend the lines.
2. The messy property acquisition for both the Mariner Station and the 128th St segments goes away.
3. The new tracks can follow the SR 525 tracks between Alderwood and 99 (Evergreen Way), and follow 99 to Airport Road — providing more infill stations.
4. Link would arrive in Downtown Everett a few minutes earlier.
5. The Everett connection to the industrial area remains intact.
6. An extension to South Everett Park and Ride becomes possible.
Of course, it would be more miles of track. So it may be significantly more expensive even if it takes less real estate acquisition..
It’s one of those Sunday musings that dawned on me. I’m not advocating for it — merely pondering if it would be more useful to Snohomish.
[Ed. Assumed Paine Field was meant above.]
“But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied”
That’s what ST plans to do with the 2 Line. It’s also a potential early phase in Everett Link construction, if Everett gets delayed further or even if not.
I guess I’m merely suggesting that two lines diverge (branches) at Alderwood rather than having one just turn around earlier than the other one.
one of the bottlenecks was that reversing trains at Lynnwood is tight.
How so? You mean there isn’t enough room to put the trains? Or are you saying they can’t reverse in time. If it is the latter it is simply because they are pushing the envelope when it comes to doing this (as a way to keep as many trains in operation as possible). This goes back to the train car shortage. One alternative would be to turn back half the trains at Northgate. This would mean running trains every 16 minutes to Lynnwood though, and that is pretty much a non-starter. But when the trains are running twice as often it seems quite reasonable. We really don’t need train running every 4 minutes north of Northgate. 8 minutes would be enough.
But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied.
I’m not quite sure what you are suggesting, but a branch in the north end is quite reasonable. The problem with branching in Lynnwood is that:
1) It will still be extremely expensive.
2) It won’t get that many riders.
In contrast a branch in Seattle could easily get a lot of riders. The obvious one is UW to Ballard. Other options include branching towards Lake City or Aurora (north of the UW). Unfortunately we didn’t plan for this, which greatly complicates things (and adds to the cost quite a bit). There is another consideration:
We might be better off with an independent, automated light-metro. For example consider two worlds:
1) Ballard to UW interlined train running every ten minutes (midday). Riders could continue to Capitol Hill and downtown. This means a train running every ten minutes to Roosevelt, Northgate and other north end destinations as well.
2) Trains running every three minutes from Ballard to the UW. Trains running every five minutes to Roosevelt, Northgate, etc.
It is quite likely that the second option would be much cheaper than the first. Yes, it means people have to transfer. But it also means that people headed to the UW from Ballard (etc.) would have much better frequency (along with people in Roosevelt, Northgate, etc.).
If you asked me ten years ago I would have supported the first idea. Now I am strongly in favor of the second. There really haven’t been major changes to transit since rubber tired buses replaced much of the world’s trams. But one of the biggest has been automation. In my opinion you better have a damn good reason *not* to make your system automated. We don’t.
It appears that you are still hoping for that Ballard branch, Ross, and it cannot be done. At least, not for less than however much the tunneling between Roosevelt and Ballard would cost plus at least three billion, probably more, for the Open Heart surgery required at 52nd and Brooklyn.
Nobody at ST is smoking that bowl of Mendocino’s Finest.
A line across Wallingford makes a lot of sense sometime in the future (if there is one), but not as a spur off The Spine.
It appears that you are still hoping for that Ballard branch
No. I clearly wrote it was not my preference.
It cannot be done. At least, not for less than [a bunch of money]
That is wrote as well. Basically you are just repeating my points but in a way that suggests that I disagree with them.
You are right, Ross. I did not read carefully enough. My apologies.
“How so? You mean there isn’t enough room to put the trains? Or are you saying they can’t reverse in time?”
It’s lots easier to reverse all the trains at the end station than it is to slot one that turns around at an intermediate stop. At an end station the trains simply get directed to one platform or the other. Or they may go further using scissor tracks that allow for both tail tracks to be use as alternating sidings.
It’s helps lots to have a siding to reverse a train when one line stops and the other continues . It’s really disruptive without a siding.
A full branching with two tracks on each branch is actually like having two sidings. It’s one more level of flexibility beyond just one siding to reverse a train.
Yeah if we could work backwards a branch should have been in North Seattle. One that far north is less productive. My thought was simply that this configuration would keep all the same general station locations shown in ST3.
To be clear, it’s simply leaving the 2 Line alone to end at Mariner. It’s just running 3 Line tracks down 99 and 525 to connect just north of Alderwood rather than run down 128th to the Mariner Station — a segment and station that will require taking many homes. And trains from both Everett and the Paine Field area would get to Seattle a few minutes faster since they wouldn’t have to jog to go to the Nariner Way station.
So you are proposing to branch at Alderwood Mall, running one line up 525 to 99, up 99 to the current proposal then and on to Everett? But you’d still build alongside the freeway to Mariner? That’s spending a lot of money in the boondocks to avoid adding a pocket track at Mariner.
It’s lots easier to reverse all the trains at the end station than it is to slot one that turns around at an intermediate stop.
Of course. That is one of the drawbacks. You can see the list of “Cons” for strategy 3 (Northgate peak turn-back service) on the report:
* Increases operational complexity and risk of system delays
* Requires additional operators to turn trains quickly
* Requires additional security officers to ensure passengers have disembarked short turning trains
* Eliminates use of Northgate pocket track for trains that need to come out of service
All legitimate risks and trade-offs. But it is also a good way to deal with the (temporary) train car shortage. I’m not saying it is the best option though. One nice thing is that by the time the trains get to Federal Way we’ll know how many trains we need for the East Side. My guess is three-car trains will be fine. We may be able to purchase enough trains by then as well. If not I could see them retain ST Express service during peak (at least until they get more trains). Metro will probably truncate their buses, but these buses don’t carry very many people. Some will prefer transferring to Link but most will likely just keep taking the bus (or Sounder). Link’s big selling point will be for trips to places like Highline College and SeaTac (from the south) this won’t effect crowding at all.
I think Mariner is the right junction – my hope is that ST delivers Lynnwood to Mariner as a “Phase 1” project with Mariner junction (specifically, a station with a junction immediately to the north). Future political leaders can choose build out to Paine Field, further along I5, both, or neither.
525 isn’t a compelling transit corridor, mostly because between 99 and 5 it is a closed access freeway: CT’s Journey 2050* does not put any bus route on that segment. Either way there will be a station at 164th and at 99 to intersect with Swift Orange & Blue respectively. 164 & 525 is intriguing because of the lack of an interchange, but then 99 & 525 is a messy interchange while 99 & Airport Road is not, so that’s a wash.
There are two east-west transit corridors between Lynnwood and Everett, 164th (Swift Orange) and 128th/Airport (Swift Green). Link should connect with both of those to create a strong grid. Branching at Alderwood & running along 525 and I5 would be overkill because the primary value of the station at 164th is the connection to Swift; two stations on 165th adds little. At Ash Way there is no good way to turn away from I5. Therefore the junction occurs at Mariner.
Also, because Ash Way and Marinerare Link-Swift transfers, which makes the high frequency of two lines very compelling. A junction at Alderwood would mean inferior Link-Swift transfer times.
Only reason to run along 525 instead of I5 is if WSOT offers up some ROW such that it is materially cheaper. Losing the transfer with Everett-East King buses (at Mariner or a modified Ash Way) would a bummer, but definitely worth it if 525 was $1B cheaper than I5.
*https://www.communitytransit.org/long-range-plan
my hope is that ST delivers Lynnwood to Mariner as a “Phase 1” project
I think you are forgetting that Link expansion is not cheap. If the trains were running on the ground or on old railroad tracks then any expansion would make sense. But every new mile of track costs a fortune and will result in very few new riders and very little time saved. ST is basically trying to leapfrog existing transit in Snohomish County and it is really bad idea.
It is worth noting that CT used to run a lot of buses to downtown. Replacing these buses with a train is quite reasonable. But now you have an ideal bus/rail interface, a gigantic park and ride lot and you are basically done. I would invest in projects like the Ash Way bidirectional ramps (only $65 million!) and more BAT/HOV/Bus lanes. But other than that, the main thing Snohomish County needs is more bus service. It is weird enough that CT has “BRT” in various areas (with ten minute midday service) while many core routes have half hour frequency. But building a very expensive subway expansion in areas that have so few riders is just a bad way to spend money. You end up with a land of haves and have-nots, and it isn’t based on potential ridership. Some people have great transit while the vast majority do not. And for many of those with “great transit” all it would mean is that they drive to a different park and ride.
Consider the buses that we know about in Snohomish County. The 510 carries less than a thousand riders. The 512 carried 1,200 riders last month and that will likely go down even more (since the 512 truncation didn’t occur until midway through the month). I have no idea what relevant CT buses carry (does anyone?) but my guess is not that many. The existing ridership is just way too low to think that an extension makes sense for the area. Again, if it was a typical surface light rail system (i. e. cheap to build) then I could see it. But this is a full-on metro — complete grade separation, huge stations — this just isn’t cheap. Given the very small number of riders that would benefit it just isn’t worth it.
I’ll just note that the backwards “C” path of Everett Link Extension will take a 3 Line train 33 minutes to get from Downtown Everett to Lynnwood City Center. That’s 6 minutes longer than it takes to go from Lynnwood City Center to Westlake.
STX 512 shows schedules between 29 and 32 minutes today. Of course that path today has an extra 2-4 minutes to get to the platform as well as a wait for a train (at 5 minutes for each line it’s only a 2-3 minute wait).
To be clear, I’m not saying it should be done. I’m merely pondering if it would make sense to have a branch separating 2 Line and 3 Line generally. This version is just a way to keep the ST3 stations.
Maybe it’s just a futile attempt to make the really poor productivity of Everett Link Extension better — and it may not even do that.
Just like RapidRide and Link have wildly different performance expectations, we should have a similar expectation for Swift and Link. A full Swift line getting 7,000 riders is very productive while a long Link segment with 7,000 riders is pretty weak.
Just like RapidRide and Link have wildly different performance expectations, we should have a similar expectation for Swift and Link. A full Swift line getting 7,000 riders is very productive while a long Link segment with 7,000 riders is pretty weak.
Exactly.
It is worth noting that it takes about fifteen minutes (according to Google) to drive from the Lynnwood Station to the Everett Station. The bus makes an additional stop on Broadway and then stops at South Everett and Ash Way. The stop on Broadway probably doesn’t cost much time. The stop at South Everett does require the bus to slow down from 55 MPH to a complete stop and then speed up again. But these two stops only cost a couple minutes, if that.
The Ash Way bus stop costs a lot more. Google says it takes about 4 minutes to go northbound at 2:00 pm (https://maps.app.goo.gl/2ky72UK6zTyAG4oSA). It probably takes longer when there is a lot of traffic. In contrast it would take less than a minute if they added ramps to the north. This means saving at least 3 minutes while still serving Ash Way. Thus it means end to end time of around 20 minutes while being far less susceptible to traffic delays. Even with a transfer that is much faster than Everett Link.
Of course another option is have the 512 just skip Ash Way. This would save at least four minutes. People could still get from Everett to Ash Way (via the 201/202, which does a better job serving the apartments along the way). They could still get an express from Ash Way to Lynnwood (via the 201/202 and 513). If timed, these could provide 7.5 minute frequency (at no extra cost). This is a trade-off, but if the goal really is faster service between downtown Everett and Seattle/Lynnwood then this is the best way to provide it (while we wait for them to add the ramp at Ash Way).
Furthermore, the bus could be extended up Broadway in Everett. Right now it does this weird zig-zag. I’m not sure it is worth even deviating to Everett Station (because again the 201/202 does that). The vast majority of the buses that serve the transit center also cross Broadway. Either way, the bus should go up Broadway to College Station and thus provide a lot more one-seat rides.
Of course the counter-argument is that it isn’t about trips like that. It is about trips like Everett Station to Mariner or Alderwood to Paine Field. Fair enough. I just don’t think there is any way you will get enough riders to justify the cost.
Everett Link is not about connecting Everett to Lynnwood. That should be obvious given the Paine detour. Yes, extending Link further north does very little to improve travel times for buses already on I5 – so what? It is not for those riders. Yes, CT should invest more in non-Swift route frequency, but again, Link is about enhancing the Swift network and better connecting the Green, Orange, and Blue (northern part) to the broader region.
A discussion comparing Link to buses does need to be careful to distinguish between boardings and ridership . If there are 7,000 boardings, that is going to be 14,000 in ridership unless a significant number of riders make return trips on another mode.
I also think it’s also worth mentioning that all the Everett Extension Stations got about 40 percent more riders in the 2040 ridership forecasts than West Seattle Link Extension stations did — even though the capital costs appear about the same.
Everett Link is not about connecting Everett to Lynnwood.
And in turn that means it isn’t about connecting Everett to Seattle. So it is basically for travel within Everett. Fair enough. The problem is that building a subway system for that many riders is a very poor use of money. It is like building a subway system in the Tri-Cities. What??? Why not have decent bus service first?
That is the fundamental problem with both Tacoma Dome Link and Everett Link. It doesn’t make sense for people going from Tacoma or Everett to Seattle. But it also doesn’t make sense for trips within the city (or within a few miles of the city). It is way too expensive for the number of riders it will get. This should be obvious when you look at the cities, but there is plenty of evidence for this as well.
For example look at ridership for the 512 from before the pandemic: https://www.soundtransit.org/sites/default/files/documents/2020-service-implementation-plan.pdf#page=98. This was back in the days when ridership was higher, and ST showed us which way people were taking the buses (and trains). The 512 was one of the best performing buses in ST system, and yet it only got about 2,500 riders a day. Of those, less than 150 were riding within Everett, or even between Lynnwood and Everett.
Or consider the Green Line. It is a fairly good proxy for Everett Link. It is quite fast, and has a limited number of stops (although still more than Everett Link will have). It shares three of the future stops. It carries less than 4,000 riders a day.
This goes back to what Al said. This is good for a bus, but really, really bad for a subway line.
“1 Line trains to Angle Lake are arriving about every 15-20 minutes beginning at 7:08 PM until further notice due to a switch issue at Lynnwood City Center.
In effect: Nov. 3, 7:22 p.m.
Routes: 1 Line”
This is now the second night in a row. Yesterday things were even more messed up. Not sure what the problem was. I think it was an issue at Northgate.
I took Link into Snohomish County for the first time ever yesterday for a legitimate errand and I was quite impressed. I had a morning meetup with a friend near downtown Edmonds. From Ballard, this journey would entail routes 40+512+130 and would take almost 75 minutes. After Link’s opening, it’s the 44+Link+909 and it took me 50 minutes. Some observations:
1) MLT station is beautiful, despite it being next to I-5. Looking toward the north, the SB train suddenly emerges from a cluster of Evergreens as if it were coming out of a forest. It’s pretty unique compared to other stations.
2) CT has route 909 that acts almost as a “shuttle” route. I say shuttle because the length of the route is very short and it designed to align with the Kingston ferry, making the schedule odd and infrequent. But once you’re on the bus, it gets the job done rather quickly.
3) CT is using First Transit for the 909 (and likely all other 9XX routes). So that means the route is operated by a Double Decker, which was an overkill for my mostly empty trip. There were 3 other people onboard in each direction.
I took Link home yesterday after the SeaSquawks game and was also very impressed.
Supposedly ST was running 10 min headways, but I don’t know if they deployed any gap trains. That said, Link handled the massive crowds well and got us to our destination in less time than it used to take me to get out of the parking lot.
It’s a huge improvement. Just wish the Squawks would improve a bit.
Just like any other route, the size of the bus for CT 909 depends on peak ridership.
In 909’s special case, it saves at least five minutes over the Lynnwood options for riders getting off the ferry and heading south on the 1 Line.
I fully expect eastbound ridership ends up being larger than westbound ridership.
Does anyone know when the comment period ends for the ST Express fare reduction? Or is it just a survey, to see if some organized group has reasons to oppose the reduction, before opening a formal comment period?
I see nothing about it on the next Rider Experience Committee agenda.
November 4th (today) according to the email announcement.
I helped a friend of mine move into an Overlake Village apartment the other day. It felt unreal to me how much car traffic there was over there. My friend was confused at why her building didn’t have a parking space for every resident. She has been parking on the street… but most of the streets do not have many parking spaces, so she just parks in a general traffic lane. There were many other people doing that. These are streets with only one lane in each direction, so with one lane nearly always blocked, by necessity drivers in each direction were taking turns.
So I thought that proximity to decent transit (2, B, 245) would be a selling point for that area, but a lot of the residents evidently aren’t interested.
It sounds like your friend is in that group of Overlake Village residents who aren’t interested in nearby public transit like the 2 Line, B Line, route 245, etc. Did you ask her, or do you know why, she is so dependent on her car considering where she lives?
It’s the trade off of where the transit was built. The east link line alignment on the freeway running route rather than closer to retail on bel-red or 8th street. This makes it run faster for commuter style transit trips but for general grocery shopping trips or restaurants a lot harder or impossible
“The east link line alignment on the freeway running route rather than closer to retail on bel-red or 8th street.”
It’s not on the freeway; it’s in the middle of a growth center. I was surprised Link wasn’t on 8th but its location makes sense, and I think it’s where Forward Thrust would have gone too.
At Overlake Village station it loses the plot and is located where hardly anybody in the village can easily walk to it. So that was a mistake. However, ST would point to the large apartment buildings under construction between Safeway and the station, so it will look somewhat better when the area is built out.
The route to Redmond Tech station is along the freeway because that’s where the P&R and the Microsoft campus are. One could imagine the campus in highrises adjacent to downtown Bellevue and a subway station but not a freeway exit like in New York or other countries, but Microsoft was part of the remote-office-park-on-an-exurban-freeway-exit movement in the 1980s and that’s its nature.
> The route to Redmond Tech station is along the freeway because that’s where the P&R and the Microsoft campus are
I’m talking about the section between Overlake Village and to Bellevue.
I bring it up because this use case
> So I thought that proximity to decent transit
explains why the light rail isn’t used for such use cases such as retail/grocery etc.. but it potentially could have with a different alignment. Which is relative for future alignment choices for other link extensions.
Yeah, I agree with Wesley. It is interesting because I was ready to write a comment defending the path along the freeway. It would have made points similar to the ones Mike made. But then I took another look at the area and I see what WL is talking about. From the Bel-Red Station Link curves towards the freeway, but I could easily see it continuing straight on Spring until it intersects Bel-Red Road. You could add a station there, at 140th & Bel-Red Road. From there you could continue above Bel-Red Road until 148th, or better yet, 152nd. Add another station there (at 152nd & Bel-Red). Then go up 152nd until you get close to the freeway. The next station would be close to where they put Overlake Village Station (it would have the same name). Then just follow the freeway to Redmond Tech.
This gives you a couple more stations that are similar to the Spring District or Bel-Red stations. You have some existing retail and apartments, but you also set the stage for future growth. Of course this would have been a lot more expensive, and would have resulted in a lot more opposition. Remember just about everything thought the train should go from South Bellevue Station to Downtown Bellevue via Bellevue Way (so you could add a station closer to downtown) but that was rejected.
Getting back to the original point, I don’t think we can expect an area that has been car-centric for generations to suddenly transform overnight just because their is a train nearby. Far more likely is a change is the type of development (fewer parking lots, more high rises) but that depends on both the attitude of the developer and what the city allows. But even if that leads to an area that is less car-dependent, it doesn’t mean behavior will change right away.
The original point isn’t about the Link alignment, or Overlake’s layout. The original point is about someone moving to transit orientated development, surrounded by good transit, but isn’t interested in using transit, and wants to continue to rely on their car. Cramer didn’t explain why his friend moved to TOD, but doesn’t want to take transit. Commenters can guess all they want, but until Cramer explains it, no one knows. Maybe his friend has an extreme bus phobia, like one commenter does.
> The original point isn’t about the Link alignment, or Overlake’s layout. The original point is about someone moving to transit orientated development, surrounded by good transit
It’s related because even with good transit, the train needs to go to good destinations. Their friend or others in general probably want to reach say crossroads mall or even to reach grocery stores like fred meyer/ safeway the train doesn’t really help since it doesn’t go there. It is not necessarily strictly bad, but it is a trade off. The route near the freeway prioritizes commuters.
> Commenters can guess all they want, but until Cramer explains it, no one knows.
Sure, but we might as well discuss what we can know about which is how good or bad the transit access is to amenities.
“The original point is about someone moving to transit orientated development, surrounded by good transit, but isn’t interested in using transit, and wants to continue to rely on their car.”
That happens everywhere. Some of my neighbors in southwest Capitol Hill drive to everything not in the immediate village, although maybe not to work. They drive for the same reason my dad and roommate did, because that’s what they’re used to, and they overestimate how difficult it would be to take transit somewhere. Some have motorcycles. Why do they live here? Some to be close to Amazon or the Microsoft shuttle, so they don’t drive to work.
One worked on a tugboat so he was gone nine months a year and wanted a small studio for his break months. We went to the Ballard Farmers Market, and the first few times he insisted on driving us, but finally I convinced him to take the bus and avoid circling for parking, and he did and said it was convenient and took the bus to the farmers market afterward, but he still drove everywhere else.
I’ve seen this in even in larger cities, except in New York and maybe DC. I’ve seen people in Chicago drive practically from one L station to another, and in San Francisco from one BART station to another. They didn’t do it all the time but they did it at least sometimes.
“I helped a friend of mine move into an Overlake Village apartment the other day. It felt unreal to me how much car traffic there was over there. My friend was confused at why her building didn’t have a parking space for every resident.”
That’s the suburban mindset. Both Overlake’s layout and wide streets and your friend’s expectations are part of it. This is the American Dream, supposedly. It’s the Los Angeles model of “urbanism”. We suggest making it really walkable and inviting for pedestrians like Ballard, Capitol Hill, San Francisco, or Chicago, and we get told that’s un-American, not what 90% of people want, and not what the customers who are keeping the shops in business want.
The problem with Overlake is that there is no mixed use or urban commercial core. All the new housing has close to nothing on ground level, and all of the retail nearby is designed in the car-centric style with buildings behind parking lots. You can walk to Safeway but you don’t feel like you belong. So I don’t think it’s strange to want a car when you live there, your friend made a rational choice given her circumstances. The area is designed for people to take the train to work and use a car for everything else.
There are always multiple excuses, no matter where one lives, as to why people say they drive rather than take transit.
That is a common situation, Sunny. So much so that a study on the efficacy of light rail spent extra time focusing on it:
Many cities have seen new developments on “New Urbanist” principles: apartments with mid-rise units and a mix of commercial and residential development aimed at satisfying most residents’ daily needs without having to drive. Many of these developments are also transit-oriented, to allow for travel outside the development, such as to downtown jobs.
… [However] …
New, isolated developments are rarely large enough to be self-contained or offer the amenities of true city centers. Residents who want to travel to specialty stores or jobs not readily accessible by the existing transit network—and in typical low-density U.S. cities, this is almost all of them—will need to own cars. Once they own cars, there’s no reason not to use them for all trips, especially if zoning policies guarantee copious parking.
My guess is the area has some aspect of this. Of course things can change, and an area can evolve to be less car-centric. Broadway on Capitol Hill used to have so many automobile dealerships it was known as “Auto-Row”. It is probably the least car-centric area in Seattle now. Things change (it just might take a while).
HAPPY ELECTION DAY!!!
Turn in your ballot at a ballot drop box by 8 pm.
If you are not registered to vote, there are a limited number of places you can still register, and then privately cast your ballot.
There are also a handful of places where you can vote privately.
Others will have to provide the link with that info.
VOTE OR GET IN LINE BY 8 PM. You are discouraged from dropping your vote in a mailbox, as it won’t be counted unless it is postmarked by today.
ST Link ridership data is out for August and September. It shows ELSL ridership continuing to be at, or exceeding, ST’s pre-opening “high” estimate for ridership. And the data aligns with ST statements that ELSL ridership is exceeding projections.
ST Link ridership data for LLE is too early to draw any conclusions from, but we are also hearing statements from some officials that LLE is exceeding pre-opening forecasts.
So what does this emerging trend of ST underestimating ridership mean for the future?
Well…..circling back to that way too long and way too meaningless discussion on interlining and frequency, it means that the previous option of reducing frequency to 4&3@10 wouldn’t even be a viable solution.
Previous analysis showed that 4&3@10 just barely satisfied estimated demand, with maybe just a small demand exceedance in the urban core, But that was with earlier ridership estimates that now appear to be low, and before WFH flamed out and the tech companies started demanding their workers return to the office 5 days a week.
So if the previous option of 4$3@10 becomes unworkable due to higher than projected demand, and the original solution of 4&4@8 is unworkable due to fleet size, what is the next best option?
Increase frequency beyond 8-mins!
Operating both lines at 3@7.5 would exactly match 1-Line capacity under the 4@10 scenario, while significantly improving 2-Line delivered capacity over the 3@10 scenario. Total LRV fleet needs would be slightly more than the 4&3@10 scenario, but less than for the full 4@8 scenario.
Slightly improving frequency to 7.5 mins instead of 8-mins while standardizing train sizes at 3-cars has several advantages beyond just higher capacity and better frequency:
1). Fleet size is less than for the 4@8 goal, but only slightly higher than for the 4&3@10 option.
2). Standardized train lengths simplifies operations.
3). Spares and gap trains would also be standardized at 3-cars, freeing up LRV’s and LRV storage space.
4). Consistent train lengths are good for the customer base.
5). Synchronous scheduling is also good for the user base.
6). Meeting capacity needs while delivering better than promised frequency exceeds expectations with the customer base.
7). Solution is workable today. Waiting for delivery of 10 more LRV’s in 2028 is most likely not required.
However, if ST also significantly under predicts demand for FWLE, then there still might be a capacity issue between IDS and SoDo, even for the 3@7.5 solution. Detailed analysis required.
This is an excellent example of how interlining, combined with fleet size restrictions, can actually lead to improved frequency.
It shows ELSL ridership continuing to be at, or exceeding, ST’s pre-opening “high” estimate for ridership.
ST estimated 6,000 riders on the starter line*. For the five full months since opening day it has now average 5,561.2 riders per day**. So it is basically performing close to what was expected. The only time that rider estimates were way more than expected was UW Link. This was understandable, as the huge increase in ridership was due to a major restructure by Metro (centered around Link) that was not obvious.
The closer you are to the opening date the more likely you are to get an accurate estimate. It is difficult to find original estimates for East Link (if someone has them, please share). I found some estimates for Link as a whole from 2008. These numbers include ST2 (but not ST3). They estimated 286,000 Link riders by 2030 once Link goes from Federal Way to Lynnwood (and across the lake). This seems absurd. In contrast more recent estimates for a full East Link (across the lake and all the way to Downtown Redmond) at around 50,000. This seems far more likely.
* https://www.theurbanist.org/2024/04/30/riders-swarm-east-link-light-rail-on-opening-weekend/. The full quote is:
Sound Transit expects about 6,000 daily riders on the starter line, but the agency has projected about 50,000 daily boardings on the 2 Line by 2030, by which time it will be extended to Seattle and Lynnwood and to Downtown Redmond.
** Feel free to double check my math. It isn’t trivial to gather the information (you can’t filter by line). But I selected the following stations: Bellevue Downtown, BelRed, East Main, Overlake, Redmond Technology, South Bellevue, Spring District, Wilburton. I then copied each weekday average and then averaged those.
I agree with your other point. Given a limited number of train cars, you can do two things: Run smaller trains frequently, or larger trains infrequently. In general Sound Transit has leaned towards the latter. They could easily run trains down Rainier Valley every 8 minutes all day long, but they don’t. Hopefully they will at least continue to run trains every 8 minutes during peak, but it wouldn’t shock me if they pinch pennies and decide to run longer trains infrequently. This isn’t as good for riders, but it saves them money on operations. Time will tell.
If the trains are automated, the operating cost to run fewer longer trains vs. more shorter trains is very minor, so you may as well choose the “more shorter trains” option, since less wait time makes for happy passengers and boosts ridership. As an added bonus, if the “more shorter trains” is decided back before the line is built, this allows the stations to be smaller, saving an enormous amount of money in capital cost. This is essentially what Vancouver did with its SkyTrain system.
By contrast, the “fewer longer trains” option incurs a worse rider experience and higher capital costs, but, if the line is not automated, it saves the agency a lot on labor cost, since it’s fewer drivers that need to be hired and fewer driver-hours they have to pay for. So, essential the choice towards fewer longer trains is really just fallout from Sound Transit being unwilling to consider automated trains.
Thanks for the heads up on the ridership data being restored to the web page after being pulled down for what I am sure was a well-justified reason.
It is hard to predict FWLE ridership without knowing where the STX route restructure is headed.
But even if the restructure heads toward truncations at FWCCS, it may be advantageous to wait until after the World Cup to do those truncations.
We could also take a lesson from the SCM, and do some price-gouging during the FMWC. Create a long-distance or even regular STX fare of $4 to $6 that takes effect before the FMWC, but accepts the usual monthly passes as full payment. The higher fares could then go away with most of the redundant bus service, probably just for the peak direction, since FWLE totally loses the race with STX 594 off-peak.
But I am glad you agree that standardized train lengths is a desirable trait for any service pattern. The main reason to run shorter trains during peak is to be able to run shorter trains off-peak, and thereby lower maintenance costs, while having more LRVs available for maintenance during the day. Even in the Before Times, I expected train lengths would drop down to two or three cars after the Great Conjunction, depending on line and whether FWLE and LLE were open yet.
I don’t see how you can conclude this. There are two overarching reasons:
1. The bulk of the East Link riders have not begun riding 2 Line yet.
2. Daily ridership is not the same as the most crowded hour of ridership.
I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway. The seats Link DEIS assumed 8 trains an hour (7.5 minute frequency).
I generally feel that all the trains should be the same number of cars. That’s especially true since Link does not tell a waiting rider how many cars are in a train on its arrival signs and announcements (unlike BART and Muni Metro). So I agree that 3 car trains aren’t optimum and should be avoided.
I also think that a little breathing room between trains is good for operations. Late trains are not uncommon and breathing room in the schedule helps the following trains get back in schedule. Of course, the typical rider only cares about when the next train arrives — and if every train is running 20 minutes behind schedule, a rider won’t notice or care as long as the wait is similar.
And since Covid, the peak surges are reduced and the ratio of busiest hour to daily ridership has dropped. Of course, more employers (like Amazon) are going back to requiring workers to appear at the office so that surge will start to appear more pronounced again.
Finally, I will say that the schedules at higher frequencies do not have to align perfectly to the minute nor do they need to be symmetrical. It’s nice but not absolutely required. It can even be better to have a second train on the heels of a first train to quickly pick up any riders that can’t board that first train. Riding subways in places like Chicago and New York and Boston can offer witness to that and many other operations strategies done to deal with crowding issues and delayed trains. It’s a big reason why I wish ST would hire more managers and dept heads from high-volume established systems rather than from a related industry.
Train frequency an important consideration for sure. It’s something that must be monitored. It’s great that you care! I’m only saying that the average daily data trends for a much lower volume segment doesn’t translate into conclusions about the crowding issues on the whole line.
@Al S,
“ I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway. The seats Link DEIS assumed 8 trains an hour (7.5 minute frequency).”
You are correct, 6-min frequency is not on the table. ST previously temporarily ran 6-min headways due to some issues with LRV availability, but they have no plans to do so again. And for a variety of reasons.
ST ridership modeling is also currently based on 8-min frequency, not 7.5. So going to 7.5-min frequency yields a 6.7% increase in delivered capacity over the 8-min scenario. Which is small, but still significant.
But the crux of the proposal is not based on the 6.7% boost in ridership that comes from 7.5-min headways. It comes from the change to 2-Line.
The 1-Line would stay exactly the same. 4@10 and 3@7.5 are exactly the same. Both operational scenarios delivery 24 LRV’s per hour across any given screenline. So no change to the 1-Line.
The difference is on the 2-Line, and therefore also on the interlined section through the urban core.
ST’s current option is to run 3-car trains at 10-min headways on the 2-Line, with maybe some 4-car trains mixed in after 2018 when they get 10 more LRV’s.
3-car trains at 10 mins delivers 18 LRV’s per hour across any given screenline, whereas switching to 3-car trains at 7.5 min headways would deliver 24 LRV’s per hour. That is a 33% boost in delivered capacity on the 2-Line, and an effective 14% boost in delivered capacity in the interlined section.
The capacity crunch is on the interlined segment, so that 14% boost becomes very significant.
Essentially the proposal is to find an operational scenario that has LRV requirements between the 4@8 baseline (which can’t be implemented due to LRV availability), and the 4&3@10 scenario (which doesn’t solve the capacity crunch problem due to emerging higher than anticipated demand for Link).
The proposal does this by providing an effective capacity boost on the 2-Line (and therefore interlined) segments, while delivering exactly the same capacity on the 1-Line. Thus it ends up being a sort of intermediate option, with higher delivered capacity where it is required (urban core), but with lower total LRV requirements than the 4@8 baseline.
The proposal also helps with LRV requirements by having a standard 3-car train length, which is operationally simpler, but also helps reduce the LRV requirement by eliminating the need to hold 4-car spare and gap trains in reserve. Everything would go to 3-car trains, even spares and gaps.
I guess I feel like anything more frequent than 10 minutes doesn’t require specified arrival time each hour. Such accuracy is very much desirable for 15 minutes or less frequency. But once below 10 minutes I’m would be more concerned about avoiding crowding.
For example, maybe only trains that are more frequent only need to arrive in Downtown Seattle between 7:50 and 8:50 in the morning. So an extra train runs a round trip to make that happen. That’s different than running several extra trains over a three hour period.
All this conceptual scheduling is interesting. However, at the end of the day it first is about decent frequency which I see as 10 minutes. After that, it’s all about avoiding people getting left on the platform. And a seasoned rail dispatching staff should get a thorough understanding of how peak demand goes and figure out what of many options works best to keep that from happening. I feel like they shouldn’t be locked into just one option of specific arrivals times and train lengths.
@Al S,
“….I feel like anything more frequent than 10 minutes doesn’t require specified arrival time each hour. ……once below 10 minutes I’m would be more concerned about avoiding crowding.”
I agree 100%. Which is why I find this focus on ultra-low frequencies that some people on this blog have to be so misplaced. Because once you get below the current level of 8-min headways adding more frequency provides minimal benefits. And Link will soon be operating at 4-min headways in the urban core anyhow.
Na, the crux of the proposal is to provide more capacity to alleviate crowding, while still providing a solution that doesn’t bust the bank on LRV requirements. Because we simply don’t have the LRV’s (yet) to solve this problem with brute force.
The side benefits of slightly better frequency and synchronous scheduling are just happy fallouts of the 3@7.5 option. The main goal was to solve the crowding problem without requiring a high number of LRV’s.
once you get below the current level of 8-min headways adding more frequency provides minimal benefits.
That is absurd. There is a direct relationship between frequency and ridership. The relationship isn’t linear — it is a curve as expected. But it doesn’t drop to zero at 8 minutes or even 6 minutes (which is the general standard for metros). This means that there are some people who will take the train if it is running every 4 minutes, but not every 6 minutes. You lose even more if the train runs every 8 minutes. We are not talking about people complaining — these are people who hate waiting so much that they will find another way to get there. If you doubt me just check the research.
Consider what happens every day in every building in the city. People hate waiting so much they will sprint to an elevator and ask that everyone else be delayed a few seconds just for them. And people do it, too! They hold the elevator because we can all relate. This is an elevator, which typically comes every thirty seconds or so. If it isn’t there within a minute people start looking around and complaining, or at the very least wondering what is going in.
By the way, this is different than what Al is talking about. He is talking about a schedule. I agree with his point. Partly it is due to the nature of most of our stations. It takes a while to get to the station. A schedule only does you much good. What is more useful is to see when the next train is arriving as you enter the station so you can decide whether it is worth going fast or not. Because no one wants to get to the platform only to find the train pulling away. Unless you are in some place like Vancouver, but that’s only because they run their trains very frequently. We do not.
“4@10 and 3@7.5”
What does this format mean?
“4@10 and 3@7.5”
4-car trains at 10-minute headways (24 train-cars per hour)
3-car trains at 7.5-minute headways (24 train-cars per hour)
Passengers perceive the difference between 10-minute, 8, 6, 5, and 2-3 minutes. 2-3 minutes is an excellent rider experience like New York or London. 5-6 minutes may be the most you can expect in a medium-sized city. 8 minutes and you wish it was 6. 10 minutes you’re starting to feel impatient, especially if it’s part of a 2-3 seat ride. If the other two buses run every 15 minutes, that’s potentially 10+15+15 = 40 minutes you’re waiting. 15 minutes is the most common on American light rails, and that really feels like the agency/community isn’t taking transit seriously. It’s one thing if buses run every 15 minutes, but the rail is supposed to be the core of the city’s transit system, so it at least should run every 10 minutes or more. 15 minutes is only appropriate for late night or very minor secondary lines.
@Mike Orr,
Of course at some level people can perceive the difference between a 8-min headways and 6 min headways, but the difference in average wait times is tiny (4 mins vs 3 mins), and that isn’t my point anyhow.
What I said was that the higher the frequency the less impactful small changes in frequency are. You enter an area of diminishing returns. And that is a fact. Going from 4-min headways to 2-min headways just isn’t going to have as much impact s going from 40-mins headways to 20-mins.
And, while there is no doubt that frequency has some impact on ridership, at the higher frequencies the effect is small. And the frequency change is more likely to drive changes in cost faster than changes in ridership.
Doubt that? Look no further than CT. When CT increased frequency 20% on Swift Green during the recent restructures the increase in ridership was 3%.
That represents a fairly weak coupling between frequency and ridership, and that weak coupling is true at most transit agencies, at least if they are operating at higher frequencies.
The good news for Metro is that this weak coupling works both ways. Metro has a very good opportunity to cut their costs and at least delay their day of reckoning with their financial Cliff by cutting frequency on some of their higher frequency, lower ridership routes.
The impact to Metro ridership won’t be that large for small reductions in frequency, but the savings in direct operating costs will be much more substantial. And, since Metro is operating at a fare box recovery ratio of about 8%, the risk of entering a transit death spiral is very small.
@Mike Orr,
Oh, and I forgot to mention.
The best way to avoid compounding transfer penalties is to simply avoid the transfer in the first place.
I try my best to avoid transfer penalties by structuring my trip to be on single seat rides. Or by just walking. I just don’t see the point of waiting for a bus that is scheduled to arrive in 15 minutes, *if* it is even on time or *if* it will even arrive at all.
And I don’t think our local bus agencies will ever be able to offer the frequency or the reliability of Link. They would go bankrupt if they even ever tried. So the transfer penalties are sure to remain.
I don’t think our local bus agencies will ever be able to offer the frequency or the reliability of Link.
The RapidRide G runs every six minutes. The RapidRide E runs every 7.5 minutes (midday). The 7 did as well (until the driver shortage). So they’ve done it before.
I try my best to avoid transfer penalties by structuring my trip to be on single seat rides.
Except the vast majority of trips within the city involve a transfer. Link only goes to a handful of places. Various buses make a lot more stops, but the options are still quite limited. They are in every city. You can’t have a really good transit system with a lot of transfers. Consider Vancouver. In my opinion they have the best transit system on the West Coast and the best transit system for a city its size in North America. It is highly dependent on transfers. They have a very good grid that allows people to make simple, easy, quick transfers.
To be clear — I’m talking about making a perpendicular transfer. You can argue that same direction transfers shouldn’t happen, or be rare. But if you do that you are dismissing a huge part of Link. If that is your argument then buses like the 71, 72, 73 and 74 should continue downtown. So should all of the Community Transit express buses (bring back the 400 series and send the 800-series buses to the UW). Fortunately neither Metro nor CT takes that approach and truncates the buses, thus forcing a transfer. That is a trade-off that in turn enables better frequency — for trips involving a transfer or not.
What I said was that the higher the frequency the less impactful small changes in frequency are.
Bullshit. You wrote:
once you get below the current level of 8-min headways adding more frequency provides minimal benefits.
You didn’t write “diminishing returns” or “less and less benefit”. You wrote “minimal”.
Oh, and your statement implied that either we run every 8 minutes all day long (which is simply not true) or that ten minutes is OK in the middle of the day.
But you didn’t make a claim about diminishing returns. I did. I mentioned the curve (in ridership/frequency). Thus the strawman you are are currently slaying was already destroyed. Allow me to beat that thing to a pulp.
Another (more technical) way to put this is elasticity. The between ridership and frequency is not completely elastic. If you double frequency you won’t double ridership. But it isn’t completely zero, either. It is a curve, with higher elasticity the more frequent you are. This is rather obvious. Going from 40 minutes to 20 minutes is a lot better than going from 20 minutes to 10 minutes. No one is arguing that. That strawman is dead, dead, dead.
But that doesn’t mean that going from 10 minutes to 5 minutes (or anywhere less than 8) is meaningless. According to this study it averages .27 under ten minutes. In other words if you go from ten minutes to five minutes you should expect roughly 27% more riders. The key word here is “roughly”. As they make clear there are a lot of different variables. But that is the average. Again, it gets smaller and smaller the more frequent you go. But again — this is ridership. Just because ridership doesn’t change doesn’t mean it doesn’t matter. Fare increases are relatively inelastic, but no one wants to pay more than they have to.
Consider the case for streetcars based on capacity (https://humantransit.org/2009/07/streetcars-an-inconvenient-truth.html):
In other urban contexts, rail transit is important for its ability to carry large number of riders per vehicle, and hence per driver, usually by combining cars into trainsets. … This capacity advantage can be relevant in high-volume situations, particularly when frequencies get down to the three-minute range.
(Emphasis mine.) Notice it says three minutes. In other words, once you reach three minute frequency (not eight) there really isn’t much value to riders in running more frequently. Very few people care. Of course some do — the way some won’t walk a quarter mile to a bus stop — but very few do. The problem, of course is that running every three minutes — although great for riders — is expensive.
So agencies make a compromise. What is typical: Six. So much so that it has been published. To copy a comment on Human Transit:
Vuchic’s 2005 Urban Transit text lists the following desirable headways:
– up to 6 minutes for “short urban trips”
– 6-12 minutes is “still satisfactory for urban trips up to 5-10 km”
– up to 20 to 30 minutes “for longer trips, such as regional commuting trips for which passengers use schedules”
Now consider Metro and the fact that they run the G every six minutes. SDOT might have screwed up the project, but the logic is sound. It is what just about any agency around the world would do. No one would question why you would run a bus every six minutes, given the nature of the trip and the (relatively) large investment in infrastructure. For the same reason, running the trains every six minutes — all day long — would actually be quite appropriate.
And yes, things will get better for folks in the north end as soon as East Link gets here. I am very much looking forward to it (as I’m sure many other people are). I just wish more people in the region could experience that sort of frequency (or something to it: like six minutes).
@Ross and Al
You guys are overthinking it. Lazarus doesn’t care about the 8/10 minute frequency because they are on the lynnwood segment and gets doubled the frequency
If you’re 30 seconds late for a bus running every hour, that 30 seconds has actually cost you an hour.
So, with many sparse areas only able to justify hourly service, high frequency can be extremely helpful.
It’s not that they justify hourly service, it’s that American transit agencies underestimate what they justify or don’t have the resources to completely serve their service area.
People are not worried about the average wait time. They care more about the wait time if you just missed a train. Is that a minor inconvenience, or something that makes them also just miss their bus that runs every 15-20 minutes? How much earlier do you need to leave to make sure you still get there on time even if you just missed a train? Or how late are you if you left “on time” and hoped for the best? 15 minutes versus under 5 minutes is a huge difference for most people. 15 minutes for example is the point where your medical appointment gets cancelled and rescheduled.
@Glenn in Portland,
“ If you’re 30 seconds late for a bus running every hour, that 30 seconds has actually cost you an hour.”
Yes, but if I need to plan ahead to take a bus like I plan ahead to take long distance train or an airplane flight, then I am likely to just drive.
And I’m sorry, but if I do need to transfer to an infrequent bus, I’m going to take a little personal responsibility and not try to time it within 30 seconds or less.
The Starter Line’s expectations were very low, so if it exceeds them slightly it’s still very low. I did my monthly Lake Hills trip Friday afternoon and there were four people in my car.
I still believe Lynnwood will have higher ridership than either the Eastside or Federal Way for demographic and geographic reasons. The Eastside has a lot of rich people who feel they’re too wealthy or suburban to take transit, and work-from-home tech workers and employer shuttles. Snohomish County has lower average income and a higher percent of people in service jobs often with unusual shifts. North Seattle is just 15-20 minutes from Lynnwood and has a wide range of destinations.
From downtown Bellevue 20 minutes gets you to downtown Seattle, which may not be a compelling reason to leave downtown Bellevue. From downtown Redmond 20 minutes still leaves you in the Eastside (downtown Bellevue, in particular).
South King County and Pierce County have demographics more like Snohomish County, but Link’s travel times are longer and less competitive with express buses. 20 minutes from KDM gets you to Rainier Beach. 20 minutes from Federal Way gets you to between TIB and Rainier Beach. 20 minutes from Tacoma Dome (with the extension) gets you to Federal Way.
Among the issues of frequency, capacity, and train length — train length is the least important. Higher frequency is a material advantage to passengers: it makes transit more convenient, makes more trips feasible, allows people to do more activities in a day, and generates ridership. Varying-length trains just have the inconvenience of having to rush to the last car if you’re standing beyond it, but they don’t do anything about the wait time or how many activities you can fit into a day.
@Mike Orr,
“ Higher frequency is a material advantage to passengers: ”
Yes, but I think people on this blog tend to approach the issue of frequency from the POV of the “before times” of Metro only operations. Back then frequencies tended to be very low, and going from 30 min frequencies (or worse!) to 15 min frequencies made a huge difference. Whereas going from say 10-mins to 8-mins in the current world is not nearly as significant.
Stated another way, the higher the frequency the less small changes in frequency will matter.
Additionally, the type of route the frequency is delivered in will also have an effect. Small changes in frequency on commuter type routes (where rider schedules are baked in and repetitive) will have less impact on the ridership base than small changes in frequency on routes in the urban core (where a higher percentage of trips are spontaneous).
Stated another way, the higher the frequency the less small changes in frequency will matter.
Of course, but as I wrote up above, that doesn’t mean it approaches zero at ten minutes. Far from it. It basically never reaches zero. Consider elevators:
User studies have indicated that for an office building, a waiting time of less than 20 seconds is excellent and 40 seconds is poor.
Think about that for a second. 20 seconds if fine, but waiting forty seconds is terrible. Obviously you can’t have trains running that often. But if anything you have it backwards. We aren’t used to a really frequent system or even a standard system. Six minutes is as close to a standard as you’ll find. Not for peak, but for midday service. Some systems run more often. Toronto runs their trains every 2-3 minutes peak and every 4-5 minutes off-peak. But Vancouver has 6 minute midday frequency for their core. Some of the branches run less often, but they are quite far away from the city.
This brings up an important point. The shorter your trip, the more frequency matters. If you are taking a train to Portland then hourly service is fine. If you a couple miles then it is the opposite. (Again, the studies back this up. The frequency-ridership curve approaches zero more quickly with commuter rail then it does urban transit.)
Thus the branching in Vancouver is basically fine. The farther out you are from the core of the city the more likely you are to be taking a long trip. This is the basic idea with Stadbahns. They branch outside the city, but inside the city there is very good frequency.
Link doesn’t quite do that. Places like Beacon Hill and Mount Baker — soon to be part of a branch — are certainly within the city. East Link is interesting because Judkins Park is the one station that isn’t in the suburbs. Overall you don’t lose that much on East Link with ten minute frequency. But you definitely lose a lot with the main line (between CID and Rainier Beach). The answer is to simply run the trains every six minutes (midday). Three minute service would be overkill (but still appreciated) for the shared section; six minutes service would also be overkill (but still appreciated) in a lot of the distant areas; but six minutes would be appropriate between CID and Rainier Beach.
This brings up another issue. Our system is very long and ST has been reluctant to have turnbacks. If our system just ran from Northgate to Columbia City then it is likely the trains would run more often. If we had turnbacks then it would be fine to run within the core twice as often, while going out to the suburbs less often. In my opinion the obvious turnbacks are 148th (to the north) and Rainier Beach (to the south). That would mean running every twelve minutes (midday) outside the city and every six minutes inside it (I am counting 148th as “in the city”). But ST simply isn’t interested. There is no focus on frequency. There is no discussion of trade-offs, cost/benefit — none of that.
“I think people on this blog tend to approach the issue of frequency from the POV of the “before times” of Metro only operations. Back then frequencies tended to be very low, and going from 30 min frequencies (or worse!) to 15 min frequencies made a huge difference. Whereas going from say 10-mins to 8-mins in the current world is not nearly as significant.”
People on this blog know what subway frequencies are in other countries and in the top three US transit cities (New York, DC, Chicago). They’ve experienced some of those networks and seen what role transit plays in their societies, and they want the same here.
10 minutes to 8 minutes, or 8 minutes to 6 minutes, is a minor difference but it adds up. Every minute longer is a psychological pressure, and it loses some would-be riders. When there’s a cost-effective solution (automated trains) that can run every 5 minutes even in smaller cities (Vancouver) and ST doesn’t choose it, that frustration boils over.
Non-foamers are even less amused by long waits than transit fans are. When you get in your car you don’t have to wait up to 5 or 10 or 15 minutes before you can start it, you just go. Mass transit intrinsically requires waiting, but agencies/communities should aim to make that as short as possible. Long waits are the #1 complaint about transit: people would rather ride 5 minutes longer than wait 5 minutes longer, because at least you’re moving.
lol you say it’s meaningless and then go ahead and agree on the synchronous frequency which is literally what I was talking about.
> Solution is workable today. Waiting for delivery of 10 more LRV’s in 2028 is most likely not required.
Anyways just glad you weren’t going to pretend those extra trains were arriving immediately.
If you combine the identical frequency requirement along with what Ross said
> I agree with your other point. Given a limited number of train cars, you can do two things: Run smaller trains frequently, or larger trains infrequently. In general Sound Transit has leaned towards the latter.
And the lengthened routes, you guys can easily figure out the conclusion.
And the lengthened routes, you guys can easily figure out the conclusion.
Are you saying ST will use any excuse to run less often? Maybe. But it is clear that they don’t have to. There are enough train cars to run frequently — it is just a matter of priorities.
@WL,
“Synchronous” in this context does not refer to two lines being synchronized. It refers to the scheduling of each line.
A synchronous schedule (also called a clock-face or cyclic schedule) is a schedule where the frequency is selected in a manner to insure that something happens at the same time every hour.
7.5-min headways on Link are synchronous because each train will arrive at the same time every hour. A rider need only know at what time a train will arrive at a given station and the train will arrive at the same time every hour, regardless of which hour it is.
8-min headways are not synchronous because the train will arrive at a different time in alternating hours. A rider needs to know the schedule to know when the train will arrive, because it arrives at a different time in alternating hours.
Any given line can have a synchronous schedule, even if it is a standalone line.
Daily ridership is not the same as the most crowded hour of ridership.
I didn’t think 10 trains an hour (6 minute frequency) was on the table anyway.
Good points, but it doesn’t really change the point Lazarus was making. If we have a shortage of trains cars and too much crowding we have two choices:
1) Run long trains infrequently.
2) Run short trains frequently.
Now consider these real-world options for that idea:
1) Four-car trains every ten minutes.
2) Three-car trains every 7.5 minutes.
The first requires 24 trains an hour (6 trains an hour times 4 cars per train). The second also requires 24 trains (8 trains an hour times 3 cars per train). It is exactly the same! In terms of number of trains cars as well as total amount of throughput, it is exactly the same. The first option saves money, while the second option is better for riders. It is up to the agency to decide which they prioritize more.
Where things get interesting is when you have extra train cars. Assume for a second you don’t have 24 train cars (per hour) but 28 or 30. You don’t have enough trains to run four-car trains every 7.5 minutes. The thing to do is run the trains every 7.5 minutes, but add train cars every so often. So if lots of people leave work at 5:00 then you run a few four car trains then. Or you alternate. The really nice thing is that you can apply them to one line or the other.
This is nothing new, even within our system. Look at the report on Link crowding from 2019: https://www.soundtransit.org/sites/default/files/documents/2020-service-implementation-plan.pdf#page=85. Notice that capacity per train goes up and down. Also notice that ridership goes up and down. They tried (largely successfully) to match the two. They ran relatively frequently and sometimes they had lots of train cars, sometimes they didn’t. In contrast the capacity is steady on the weekends. That is because they were infrequent, and since they were running infrequently they might as well run longer trains.
There are good reasons to run longer trains, with the same LRV throughput per hour during peak. Cost savings is probably not one of them.
If you run 4-car trains all day that are nowhere close to full, then you are running up the mileage on the additional LRVs. That increases maintenance costs 33% for the off-peak hours, which has to be weighed against the savings from having a handful fewer peak operators. It also shortens the useful life of the LRVs, hastening the day ST has to order fleet replacement.
Longer trains can also significantly reduce the number of LRVs available for the maintenance crew to work on during the day.
OTOH, having more LRVs available to work on off-peak means more of them will get a mid-day cleaning, which does not come for free.
There are probably more nontrivial costs and benefits to weigh, but I don’ think either of us have the information available to do useful armchair budget analysis.
For what it’s worth, when MAX started they’d couple and uncouple cars at Ruby junction, turning 2 car trains into 1 car trains during mid-day and back to 2 car trains in the afternoon. By the mid-1990s they decided it was cheaper to just run two car trains all the time than even the very fast coupling and uncoupling allowed by fully automatic couplers.
I did wander from my original point, which was simply that we just don’t know how crowded the trains will be at the busiest times. We have good indications from existing station activity but we just ont know.
I do want to push back a bit looking at initial Lynnwood Link ridership. It’s way below what we were told. The pre-opening press releases said that there would be 47K to 55K riders on an average weekday by 2026. The September data for the 4 stations is about 8.5K boardings or 17K assuming even two-way activity. Adding 2 Line trains and 130th St Station won’t increase demand by 200%. So it’s only running about 1/3 of what the public was told last summer. How can the ridership be at 1/3 and service be at 1/2 (no 2 Line) yet there is overcrowding? The answer appears to be bad math calculations by staff and/or consultants.
That said, there are still crowding reports coming in. Certainly the Lynnwood trains will get crowding relief when 2 Line opens.
Half of September was pre-bus restructure, when Snohomish County commuters had 11 more 1-seat rides downtown, the Swift Blue Line still terminated at Aurora Village, and South Shoreline Station was served by zero bus routes instead of the current five.
October’s ridership numbers will be a better indication of the interim normal.
Half of September was pre-bus restructure,
Yes, and that is why we’ve held off on an analysis of it. We may end up writing something about Lynnwood Link before we get October numbers just because ST has been fairly slow with their reports. But if we do, expect a followup once we have October numbers (for this very reason).
Worth noting: While I expect many of the stations to have an increase in ridership in October, I would also expect Northgate ridership to drop even lower (for the same reason).
There is also the point that UW and other schools started their fall sessions in latter September. UWS and UDS boardings ought to come way up.
There is a major problem with more frequent trains: MLK. Even though trains can run more frequently, they are going to get stopped more often at traffic signals. MLK often gives as much as 45 seconds to get pedestrians across the street (and that’s not counting time for yellow lights or left turn phases that cross train tracks). The MLK signals go for months malfunctioning, adding even more delay.
Right now, the left turns from eastbound Alaska to northbound MLK is triggering all the time with no turning cars — adding another 30-45 seconds of delay for Link trains if they can’t get priority. It’s been that way for many weeks.
Why have trains coming 2 minutes more often on average if a rider will just lose those 2 minutes at a downstream traffic signal? The only advantage is being protected from weather.
Getting stuck at an MLK signal also creates reliability challenges. If the intent is to have evenly spaced 1/2 Line trains headed towards Lynnwood, a slightly lower frequency will make that much easier to run.
The ugly fact is that ST miscalculated the train cars needed by 2026. Three car trains more frequently won’t solve the problem. If anything trains will be slower on average if they are all 3 cars and running more frequently (rather than 4 cars and less frequently) because of the greatly increased odds that the 1 Line trains will experience traffic signal delays on MLK.
I think the MLK signalling issue can be fixed. The problem is, the timing is built off the requirement that pedestrians have to be able to cross the entire street in one go, but doing that without impacting car capacity on MLK requires long light cycles that suck for everybody.
At least at the intersections with stations (e.g. most people crossing the street only need to cross halfway anyway), the light should be designed with shorter cycles where pedestrians cross just half the street at a time. They can also update the ped signals to permit crossing during left turn phases where the cars on the half of MLK being crossed are stopped anyway.
I don’t remember the details, but I know Not Just Bikes has a video where they talked about traffic signals in Amsterdam. Even at big, complicated intersections, they manage to do it with much shorter cycles and less waiting than is the case here.
Let’s talk about the rubber seat coverings.
I’ve seen enough to know where the “non-destinational” riders tend to nap: in the seats closest to the end of the LRVs.
I’d love for ST to run the experiment of installing the rubber seat covers on these seats, and see what happens.
The second video is by Nick Lewis, whom we haven’t had before and is usually off-topic (interior design), but I’ve been following him for a year. Today’s and yesterday’s videos, and a few in previous weeks, are an attempt to branch out to a wider variety of voices and some peripheral issues, so that we’re not just showing the same three people all the time and always looking at subway networks and bashing American transit and zoning; we do enough of that already. I’m hoping the broader scope of voices and peripheral topics is of interest in itself, and we may find some unexpected things the commentariat latches onto and wants more of.
I really like his interior design content and honestly very relevant to urbanism because well designed spaces can make living in them more comfortable. What materials you pick, what lighting you choose, how you arrange your furniture, are you right sizing your furniture for the space, what kind of curtain or window treatments will work in this space, what I should paint my walls, etc etc etc. These are all very helpful in developing your own personal style but also to creating a very worthwhile house to live in.
I also like that he says IKEA is good for interior design as well if you know what to look for and shop smartly rather than brush it off as cheap throw away furniture like many perceive IKEA as. And that not all high end furniture brands like RH aren’t always worth it.
Re Gemini‘s walk-and-talk videos, we may feature them at some point; it’s just that there are higher-priority ones right now. I’ve been watching some of them. There’s the excoriating review of Lynnwood station-area land use, which got a ton of comments in a previous open thread. He’s also done a mixed review of the Beacon Hill station area, the 2 Line Starter Line and the Microsoft campus, and one on the Overlook Walk I haven’t watched yet.
I just wish he’d gather a few more facts rather than guessing on things, not state guesses as certainties as much, and use a stopwatch to time things rather than “I guess it was 10-15 minutes”. We could answer a lot of his guesses about why the station areas are the way they are, Beacon Hill’s history, what Lynnwood is constructing around the station, etc, because we’ve been debating them for years. Still, the video’s show what a typical new rider’s impression is like, and one from Eastern Europe (the Balkans, as he said in the climate video).
For a language anecdode, I said his accent sounded to me like Russian, but that other Slavic languages have similar characteristics so I wasn’t sure. That’s based on me studying Russian in the 80s and 90s, having Russian friends, and visiting Russia in the 90s. But I also visited London and and met a guy from Serbia in a youth hostel. His English accent had several of the same sounds, wordings, and mannerisms I’d thought were Russian — and that’s on the opposite end of the Slavosphere. The Slavic languages diverged later than Western European ones (Germanic, Romance, Celtic), so it’s not surprising that they have a smaller variation.
Downtown Redmond Link Extension progress. A recent Sound Transit video. A couple of good shots of the Marymoor Village and Downtown Redmond station areas.
https://youtu.be/7BN9Vu7J4VI
Would be nice to see much more accurate real time arrival predictions especially as this technology is about 25 years old. There is still a big issue with ghost buses on OneBusAway and all the apps, that don’t show the actual bus and default to scheduled time which on many routes like the 8 is completely meaningless. Even on brand new routes like the G I see this. Also at the beginning of routes can be hard to know regardless with real time predictions since the bus make not have left yet or may have just left and therefore be arriving at your stop in 2 minutes which is little time to react. Its really depends whether the bus does leave its terminal on schedule. Anyhow I write this because schedule reliability seems to be quite bad, sure a lot is auto congestion along the route but there are also many other factors and it would be nice for current riders and prospective riders to have very accurate predictions when the bus will arrive.
Some of that is from the vehicle tracking system, and can’t be helped.
However, I would find the following information helpful, which PDXBus (the Portland equivalent to OneBusAway) shows:
• Is the displayed information just the scheduled information or is it actual data?
• Has the trip been officially cancelled? (PDXBus shows the scheduled time with a red X through the time and the word “Cancelled” under the time)
• How long has it been since the system received location data?
• How long has it been since the program refreshed its feed from the server?
There have been a number of times that I’ve misjudged the time I had available to get a Pugetopolis bus because the information about the last time the location was received wasn’t shown.
In last weeks discussion about branching and frequency, one if the bottlenecks was that reversing trains at Lynnwood is tight.
Then wherever the 2 Line reverses (assuming the 3 Line goes to Downtown Everett) a whole other reversal challenge is introduced. That’s because 2 Line trains would be reversed into tracks where 3 Line is running through.
So to me that suggests branching the line north of Lynnwood.
The idea of having Everett and Paine Field as separate branches appears to be a political non-starter.
But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied. Perhaps ST should explore this.
Some of the advantages:
1. The branched ends makes it easier to blend the lines.
2. The messy property acquisition for both the Mariner Station and the 128th St segments goes away.
3. The new tracks can follow the SR 525 tracks between Alderwood and 99 (Evergreen Way), and follow 99 to Airport Road — providing more infill stations.
4. Link would arrive in Downtown Everett a few minutes earlier.
5. The Everett connection to the industrial area remains intact.
6. An extension to South Everett Park and Ride becomes possible.
Of course, it would be more miles of track. So it may be significantly more expensive even if it takes less real estate acquisition..
It’s one of those Sunday musings that dawned on me. I’m not advocating for it — merely pondering if it would be more useful to Snohomish.
[Ed. Assumed Paine Field was meant above.]
“But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied”
That’s what ST plans to do with the 2 Line. It’s also a potential early phase in Everett Link construction, if Everett gets delayed further or even if not.
2 Line: Mariner-Redmond
3 Line: Everett-West Seattle
That’s true Mike!
I guess I’m merely suggesting that two lines diverge (branches) at Alderwood rather than having one just turn around earlier than the other one.
one of the bottlenecks was that reversing trains at Lynnwood is tight.
How so? You mean there isn’t enough room to put the trains? Or are you saying they can’t reverse in time. If it is the latter it is simply because they are pushing the envelope when it comes to doing this (as a way to keep as many trains in operation as possible). This goes back to the train car shortage. One alternative would be to turn back half the trains at Northgate. This would mean running trains every 16 minutes to Lynnwood though, and that is pretty much a non-starter. But when the trains are running twice as often it seems quite reasonable. We really don’t need train running every 4 minutes north of Northgate. 8 minutes would be enough.
But the idea of having one branch at Mariner and another do both Industrial Area/Paine Field and Everett has not been studied.
I’m not quite sure what you are suggesting, but a branch in the north end is quite reasonable. The problem with branching in Lynnwood is that:
1) It will still be extremely expensive.
2) It won’t get that many riders.
In contrast a branch in Seattle could easily get a lot of riders. The obvious one is UW to Ballard. Other options include branching towards Lake City or Aurora (north of the UW). Unfortunately we didn’t plan for this, which greatly complicates things (and adds to the cost quite a bit). There is another consideration:
We might be better off with an independent, automated light-metro. For example consider two worlds:
1) Ballard to UW interlined train running every ten minutes (midday). Riders could continue to Capitol Hill and downtown. This means a train running every ten minutes to Roosevelt, Northgate and other north end destinations as well.
2) Trains running every three minutes from Ballard to the UW. Trains running every five minutes to Roosevelt, Northgate, etc.
It is quite likely that the second option would be much cheaper than the first. Yes, it means people have to transfer. But it also means that people headed to the UW from Ballard (etc.) would have much better frequency (along with people in Roosevelt, Northgate, etc.).
If you asked me ten years ago I would have supported the first idea. Now I am strongly in favor of the second. There really haven’t been major changes to transit since rubber tired buses replaced much of the world’s trams. But one of the biggest has been automation. In my opinion you better have a damn good reason *not* to make your system automated. We don’t.
It appears that you are still hoping for that Ballard branch, Ross, and it cannot be done. At least, not for less than however much the tunneling between Roosevelt and Ballard would cost plus at least three billion, probably more, for the Open Heart surgery required at 52nd and Brooklyn.
Nobody at ST is smoking that bowl of Mendocino’s Finest.
A line across Wallingford makes a lot of sense sometime in the future (if there is one), but not as a spur off The Spine.
It appears that you are still hoping for that Ballard branch
No. I clearly wrote it was not my preference.
It cannot be done. At least, not for less than [a bunch of money]
That is wrote as well. Basically you are just repeating my points but in a way that suggests that I disagree with them.
You are right, Ross. I did not read carefully enough. My apologies.
“How so? You mean there isn’t enough room to put the trains? Or are you saying they can’t reverse in time?”
It’s lots easier to reverse all the trains at the end station than it is to slot one that turns around at an intermediate stop. At an end station the trains simply get directed to one platform or the other. Or they may go further using scissor tracks that allow for both tail tracks to be use as alternating sidings.
It’s helps lots to have a siding to reverse a train when one line stops and the other continues . It’s really disruptive without a siding.
A full branching with two tracks on each branch is actually like having two sidings. It’s one more level of flexibility beyond just one siding to reverse a train.
Yeah if we could work backwards a branch should have been in North Seattle. One that far north is less productive. My thought was simply that this configuration would keep all the same general station locations shown in ST3.
To be clear, it’s simply leaving the 2 Line alone to end at Mariner. It’s just running 3 Line tracks down 99 and 525 to connect just north of Alderwood rather than run down 128th to the Mariner Station — a segment and station that will require taking many homes. And trains from both Everett and the Paine Field area would get to Seattle a few minutes faster since they wouldn’t have to jog to go to the Nariner Way station.
So you are proposing to branch at Alderwood Mall, running one line up 525 to 99, up 99 to the current proposal then and on to Everett? But you’d still build alongside the freeway to Mariner? That’s spending a lot of money in the boondocks to avoid adding a pocket track at Mariner.
It’s lots easier to reverse all the trains at the end station than it is to slot one that turns around at an intermediate stop.
Of course. That is one of the drawbacks. You can see the list of “Cons” for strategy 3 (Northgate peak turn-back service) on the report:
* Increases operational complexity and risk of system delays
* Requires additional operators to turn trains quickly
* Requires additional security officers to ensure passengers have disembarked short turning trains
* Eliminates use of Northgate pocket track for trains that need to come out of service
All legitimate risks and trade-offs. But it is also a good way to deal with the (temporary) train car shortage. I’m not saying it is the best option though. One nice thing is that by the time the trains get to Federal Way we’ll know how many trains we need for the East Side. My guess is three-car trains will be fine. We may be able to purchase enough trains by then as well. If not I could see them retain ST Express service during peak (at least until they get more trains). Metro will probably truncate their buses, but these buses don’t carry very many people. Some will prefer transferring to Link but most will likely just keep taking the bus (or Sounder). Link’s big selling point will be for trips to places like Highline College and SeaTac (from the south) this won’t effect crowding at all.
I think Mariner is the right junction – my hope is that ST delivers Lynnwood to Mariner as a “Phase 1” project with Mariner junction (specifically, a station with a junction immediately to the north). Future political leaders can choose build out to Paine Field, further along I5, both, or neither.
525 isn’t a compelling transit corridor, mostly because between 99 and 5 it is a closed access freeway: CT’s Journey 2050* does not put any bus route on that segment. Either way there will be a station at 164th and at 99 to intersect with Swift Orange & Blue respectively. 164 & 525 is intriguing because of the lack of an interchange, but then 99 & 525 is a messy interchange while 99 & Airport Road is not, so that’s a wash.
There are two east-west transit corridors between Lynnwood and Everett, 164th (Swift Orange) and 128th/Airport (Swift Green). Link should connect with both of those to create a strong grid. Branching at Alderwood & running along 525 and I5 would be overkill because the primary value of the station at 164th is the connection to Swift; two stations on 165th adds little. At Ash Way there is no good way to turn away from I5. Therefore the junction occurs at Mariner.
Also, because Ash Way and Marinerare Link-Swift transfers, which makes the high frequency of two lines very compelling. A junction at Alderwood would mean inferior Link-Swift transfer times.
Only reason to run along 525 instead of I5 is if WSOT offers up some ROW such that it is materially cheaper. Losing the transfer with Everett-East King buses (at Mariner or a modified Ash Way) would a bummer, but definitely worth it if 525 was $1B cheaper than I5.
*https://www.communitytransit.org/long-range-plan
my hope is that ST delivers Lynnwood to Mariner as a “Phase 1” project
I think you are forgetting that Link expansion is not cheap. If the trains were running on the ground or on old railroad tracks then any expansion would make sense. But every new mile of track costs a fortune and will result in very few new riders and very little time saved. ST is basically trying to leapfrog existing transit in Snohomish County and it is really bad idea.
It is worth noting that CT used to run a lot of buses to downtown. Replacing these buses with a train is quite reasonable. But now you have an ideal bus/rail interface, a gigantic park and ride lot and you are basically done. I would invest in projects like the Ash Way bidirectional ramps (only $65 million!) and more BAT/HOV/Bus lanes. But other than that, the main thing Snohomish County needs is more bus service. It is weird enough that CT has “BRT” in various areas (with ten minute midday service) while many core routes have half hour frequency. But building a very expensive subway expansion in areas that have so few riders is just a bad way to spend money. You end up with a land of haves and have-nots, and it isn’t based on potential ridership. Some people have great transit while the vast majority do not. And for many of those with “great transit” all it would mean is that they drive to a different park and ride.
Consider the buses that we know about in Snohomish County. The 510 carries less than a thousand riders. The 512 carried 1,200 riders last month and that will likely go down even more (since the 512 truncation didn’t occur until midway through the month). I have no idea what relevant CT buses carry (does anyone?) but my guess is not that many. The existing ridership is just way too low to think that an extension makes sense for the area. Again, if it was a typical surface light rail system (i. e. cheap to build) then I could see it. But this is a full-on metro — complete grade separation, huge stations — this just isn’t cheap. Given the very small number of riders that would benefit it just isn’t worth it.
I’ll just note that the backwards “C” path of Everett Link Extension will take a 3 Line train 33 minutes to get from Downtown Everett to Lynnwood City Center. That’s 6 minutes longer than it takes to go from Lynnwood City Center to Westlake.
That’s according to ST:
https://www.soundtransit.org/system-expansion/everett-link-extension
STX 512 shows schedules between 29 and 32 minutes today. Of course that path today has an extra 2-4 minutes to get to the platform as well as a wait for a train (at 5 minutes for each line it’s only a 2-3 minute wait).
To be clear, I’m not saying it should be done. I’m merely pondering if it would make sense to have a branch separating 2 Line and 3 Line generally. This version is just a way to keep the ST3 stations.
Maybe it’s just a futile attempt to make the really poor productivity of Everett Link Extension better — and it may not even do that.
Just like RapidRide and Link have wildly different performance expectations, we should have a similar expectation for Swift and Link. A full Swift line getting 7,000 riders is very productive while a long Link segment with 7,000 riders is pretty weak.
Just like RapidRide and Link have wildly different performance expectations, we should have a similar expectation for Swift and Link. A full Swift line getting 7,000 riders is very productive while a long Link segment with 7,000 riders is pretty weak.
Exactly.
It is worth noting that it takes about fifteen minutes (according to Google) to drive from the Lynnwood Station to the Everett Station. The bus makes an additional stop on Broadway and then stops at South Everett and Ash Way. The stop on Broadway probably doesn’t cost much time. The stop at South Everett does require the bus to slow down from 55 MPH to a complete stop and then speed up again. But these two stops only cost a couple minutes, if that.
The Ash Way bus stop costs a lot more. Google says it takes about 4 minutes to go northbound at 2:00 pm (https://maps.app.goo.gl/2ky72UK6zTyAG4oSA). It probably takes longer when there is a lot of traffic. In contrast it would take less than a minute if they added ramps to the north. This means saving at least 3 minutes while still serving Ash Way. Thus it means end to end time of around 20 minutes while being far less susceptible to traffic delays. Even with a transfer that is much faster than Everett Link.
Of course another option is have the 512 just skip Ash Way. This would save at least four minutes. People could still get from Everett to Ash Way (via the 201/202, which does a better job serving the apartments along the way). They could still get an express from Ash Way to Lynnwood (via the 201/202 and 513). If timed, these could provide 7.5 minute frequency (at no extra cost). This is a trade-off, but if the goal really is faster service between downtown Everett and Seattle/Lynnwood then this is the best way to provide it (while we wait for them to add the ramp at Ash Way).
Furthermore, the bus could be extended up Broadway in Everett. Right now it does this weird zig-zag. I’m not sure it is worth even deviating to Everett Station (because again the 201/202 does that). The vast majority of the buses that serve the transit center also cross Broadway. Either way, the bus should go up Broadway to College Station and thus provide a lot more one-seat rides.
Of course the counter-argument is that it isn’t about trips like that. It is about trips like Everett Station to Mariner or Alderwood to Paine Field. Fair enough. I just don’t think there is any way you will get enough riders to justify the cost.
Everett Link is not about connecting Everett to Lynnwood. That should be obvious given the Paine detour. Yes, extending Link further north does very little to improve travel times for buses already on I5 – so what? It is not for those riders. Yes, CT should invest more in non-Swift route frequency, but again, Link is about enhancing the Swift network and better connecting the Green, Orange, and Blue (northern part) to the broader region.
A discussion comparing Link to buses does need to be careful to distinguish between boardings and ridership . If there are 7,000 boardings, that is going to be 14,000 in ridership unless a significant number of riders make return trips on another mode.
I also think it’s also worth mentioning that all the Everett Extension Stations got about 40 percent more riders in the 2040 ridership forecasts than West Seattle Link Extension stations did — even though the capital costs appear about the same.
Everett Link is not about connecting Everett to Lynnwood.
And in turn that means it isn’t about connecting Everett to Seattle. So it is basically for travel within Everett. Fair enough. The problem is that building a subway system for that many riders is a very poor use of money. It is like building a subway system in the Tri-Cities. What??? Why not have decent bus service first?
That is the fundamental problem with both Tacoma Dome Link and Everett Link. It doesn’t make sense for people going from Tacoma or Everett to Seattle. But it also doesn’t make sense for trips within the city (or within a few miles of the city). It is way too expensive for the number of riders it will get. This should be obvious when you look at the cities, but there is plenty of evidence for this as well.
For example look at ridership for the 512 from before the pandemic: https://www.soundtransit.org/sites/default/files/documents/2020-service-implementation-plan.pdf#page=98. This was back in the days when ridership was higher, and ST showed us which way people were taking the buses (and trains). The 512 was one of the best performing buses in ST system, and yet it only got about 2,500 riders a day. Of those, less than 150 were riding within Everett, or even between Lynnwood and Everett.
Or consider the Green Line. It is a fairly good proxy for Everett Link. It is quite fast, and has a limited number of stops (although still more than Everett Link will have). It shares three of the future stops. It carries less than 4,000 riders a day.
This goes back to what Al said. This is good for a bus, but really, really bad for a subway line.
“1 Line trains to Angle Lake are arriving about every 15-20 minutes beginning at 7:08 PM until further notice due to a switch issue at Lynnwood City Center.
In effect: Nov. 3, 7:22 p.m.
Routes: 1 Line”
This is now the second night in a row. Yesterday things were even more messed up. Not sure what the problem was. I think it was an issue at Northgate.
I took Link into Snohomish County for the first time ever yesterday for a legitimate errand and I was quite impressed. I had a morning meetup with a friend near downtown Edmonds. From Ballard, this journey would entail routes 40+512+130 and would take almost 75 minutes. After Link’s opening, it’s the 44+Link+909 and it took me 50 minutes. Some observations:
1) MLT station is beautiful, despite it being next to I-5. Looking toward the north, the SB train suddenly emerges from a cluster of Evergreens as if it were coming out of a forest. It’s pretty unique compared to other stations.
2) CT has route 909 that acts almost as a “shuttle” route. I say shuttle because the length of the route is very short and it designed to align with the Kingston ferry, making the schedule odd and infrequent. But once you’re on the bus, it gets the job done rather quickly.
3) CT is using First Transit for the 909 (and likely all other 9XX routes). So that means the route is operated by a Double Decker, which was an overkill for my mostly empty trip. There were 3 other people onboard in each direction.
I took Link home yesterday after the SeaSquawks game and was also very impressed.
Supposedly ST was running 10 min headways, but I don’t know if they deployed any gap trains. That said, Link handled the massive crowds well and got us to our destination in less time than it used to take me to get out of the parking lot.
It’s a huge improvement. Just wish the Squawks would improve a bit.
Just like any other route, the size of the bus for CT 909 depends on peak ridership.
In 909’s special case, it saves at least five minutes over the Lynnwood options for riders getting off the ferry and heading south on the 1 Line.
I fully expect eastbound ridership ends up being larger than westbound ridership.
Does anyone know when the comment period ends for the ST Express fare reduction? Or is it just a survey, to see if some organized group has reasons to oppose the reduction, before opening a formal comment period?
I see nothing about it on the next Rider Experience Committee agenda.
November 4th (today) according to the email announcement.
I helped a friend of mine move into an Overlake Village apartment the other day. It felt unreal to me how much car traffic there was over there. My friend was confused at why her building didn’t have a parking space for every resident. She has been parking on the street… but most of the streets do not have many parking spaces, so she just parks in a general traffic lane. There were many other people doing that. These are streets with only one lane in each direction, so with one lane nearly always blocked, by necessity drivers in each direction were taking turns.
So I thought that proximity to decent transit (2, B, 245) would be a selling point for that area, but a lot of the residents evidently aren’t interested.
It sounds like your friend is in that group of Overlake Village residents who aren’t interested in nearby public transit like the 2 Line, B Line, route 245, etc. Did you ask her, or do you know why, she is so dependent on her car considering where she lives?
It’s the trade off of where the transit was built. The east link line alignment on the freeway running route rather than closer to retail on bel-red or 8th street. This makes it run faster for commuter style transit trips but for general grocery shopping trips or restaurants a lot harder or impossible
“The east link line alignment on the freeway running route rather than closer to retail on bel-red or 8th street.”
It’s not on the freeway; it’s in the middle of a growth center. I was surprised Link wasn’t on 8th but its location makes sense, and I think it’s where Forward Thrust would have gone too.
At Overlake Village station it loses the plot and is located where hardly anybody in the village can easily walk to it. So that was a mistake. However, ST would point to the large apartment buildings under construction between Safeway and the station, so it will look somewhat better when the area is built out.
The route to Redmond Tech station is along the freeway because that’s where the P&R and the Microsoft campus are. One could imagine the campus in highrises adjacent to downtown Bellevue and a subway station but not a freeway exit like in New York or other countries, but Microsoft was part of the remote-office-park-on-an-exurban-freeway-exit movement in the 1980s and that’s its nature.
> The route to Redmond Tech station is along the freeway because that’s where the P&R and the Microsoft campus are
I’m talking about the section between Overlake Village and to Bellevue.
I bring it up because this use case
> So I thought that proximity to decent transit
explains why the light rail isn’t used for such use cases such as retail/grocery etc.. but it potentially could have with a different alignment. Which is relative for future alignment choices for other link extensions.
Yeah, I agree with Wesley. It is interesting because I was ready to write a comment defending the path along the freeway. It would have made points similar to the ones Mike made. But then I took another look at the area and I see what WL is talking about. From the Bel-Red Station Link curves towards the freeway, but I could easily see it continuing straight on Spring until it intersects Bel-Red Road. You could add a station there, at 140th & Bel-Red Road. From there you could continue above Bel-Red Road until 148th, or better yet, 152nd. Add another station there (at 152nd & Bel-Red). Then go up 152nd until you get close to the freeway. The next station would be close to where they put Overlake Village Station (it would have the same name). Then just follow the freeway to Redmond Tech.
This gives you a couple more stations that are similar to the Spring District or Bel-Red stations. You have some existing retail and apartments, but you also set the stage for future growth. Of course this would have been a lot more expensive, and would have resulted in a lot more opposition. Remember just about everything thought the train should go from South Bellevue Station to Downtown Bellevue via Bellevue Way (so you could add a station closer to downtown) but that was rejected.
Getting back to the original point, I don’t think we can expect an area that has been car-centric for generations to suddenly transform overnight just because their is a train nearby. Far more likely is a change is the type of development (fewer parking lots, more high rises) but that depends on both the attitude of the developer and what the city allows. But even if that leads to an area that is less car-dependent, it doesn’t mean behavior will change right away.
The original point isn’t about the Link alignment, or Overlake’s layout. The original point is about someone moving to transit orientated development, surrounded by good transit, but isn’t interested in using transit, and wants to continue to rely on their car. Cramer didn’t explain why his friend moved to TOD, but doesn’t want to take transit. Commenters can guess all they want, but until Cramer explains it, no one knows. Maybe his friend has an extreme bus phobia, like one commenter does.
> The original point isn’t about the Link alignment, or Overlake’s layout. The original point is about someone moving to transit orientated development, surrounded by good transit
It’s related because even with good transit, the train needs to go to good destinations. Their friend or others in general probably want to reach say crossroads mall or even to reach grocery stores like fred meyer/ safeway the train doesn’t really help since it doesn’t go there. It is not necessarily strictly bad, but it is a trade off. The route near the freeway prioritizes commuters.
> Commenters can guess all they want, but until Cramer explains it, no one knows.
Sure, but we might as well discuss what we can know about which is how good or bad the transit access is to amenities.
“The original point is about someone moving to transit orientated development, surrounded by good transit, but isn’t interested in using transit, and wants to continue to rely on their car.”
That happens everywhere. Some of my neighbors in southwest Capitol Hill drive to everything not in the immediate village, although maybe not to work. They drive for the same reason my dad and roommate did, because that’s what they’re used to, and they overestimate how difficult it would be to take transit somewhere. Some have motorcycles. Why do they live here? Some to be close to Amazon or the Microsoft shuttle, so they don’t drive to work.
One worked on a tugboat so he was gone nine months a year and wanted a small studio for his break months. We went to the Ballard Farmers Market, and the first few times he insisted on driving us, but finally I convinced him to take the bus and avoid circling for parking, and he did and said it was convenient and took the bus to the farmers market afterward, but he still drove everywhere else.
I’ve seen this in even in larger cities, except in New York and maybe DC. I’ve seen people in Chicago drive practically from one L station to another, and in San Francisco from one BART station to another. They didn’t do it all the time but they did it at least sometimes.
“I helped a friend of mine move into an Overlake Village apartment the other day. It felt unreal to me how much car traffic there was over there. My friend was confused at why her building didn’t have a parking space for every resident.”
That’s the suburban mindset. Both Overlake’s layout and wide streets and your friend’s expectations are part of it. This is the American Dream, supposedly. It’s the Los Angeles model of “urbanism”. We suggest making it really walkable and inviting for pedestrians like Ballard, Capitol Hill, San Francisco, or Chicago, and we get told that’s un-American, not what 90% of people want, and not what the customers who are keeping the shops in business want.
The problem with Overlake is that there is no mixed use or urban commercial core. All the new housing has close to nothing on ground level, and all of the retail nearby is designed in the car-centric style with buildings behind parking lots. You can walk to Safeway but you don’t feel like you belong. So I don’t think it’s strange to want a car when you live there, your friend made a rational choice given her circumstances. The area is designed for people to take the train to work and use a car for everything else.
There are always multiple excuses, no matter where one lives, as to why people say they drive rather than take transit.
That is a common situation, Sunny. So much so that a study on the efficacy of light rail spent extra time focusing on it:
Many cities have seen new developments on “New Urbanist” principles: apartments with mid-rise units and a mix of commercial and residential development aimed at satisfying most residents’ daily needs without having to drive. Many of these developments are also transit-oriented, to allow for travel outside the development, such as to downtown jobs.
… [However] …
New, isolated developments are rarely large enough to be self-contained or offer the amenities of true city centers. Residents who want to travel to specialty stores or jobs not readily accessible by the existing transit network—and in typical low-density U.S. cities, this is almost all of them—will need to own cars. Once they own cars, there’s no reason not to use them for all trips, especially if zoning policies guarantee copious parking.
My guess is the area has some aspect of this. Of course things can change, and an area can evolve to be less car-centric. Broadway on Capitol Hill used to have so many automobile dealerships it was known as “Auto-Row”. It is probably the least car-centric area in Seattle now. Things change (it just might take a while).
HAPPY ELECTION DAY!!!
Turn in your ballot at a ballot drop box by 8 pm.
If you are not registered to vote, there are a limited number of places you can still register, and then privately cast your ballot.
There are also a handful of places where you can vote privately.
Others will have to provide the link with that info.
VOTE OR GET IN LINE BY 8 PM. You are discouraged from dropping your vote in a mailbox, as it won’t be counted unless it is postmarked by today.