Jarrett Walker: New Geary BRT Option Could Provide Faster Service

When it comes to providing the fastest, most reliable bus rapid transit service on Geary Boulevard, cutting out bus passing lanes and “consolidating” local and express services might sound like a downgrade. But according to transit consultant Jarrett Walker, such a configuration could actually provide a superior level of service. It’s all about using the right metrics to judge travel times.

Would Geary BRT be downgraded by "consolidating" services into one line? No, says transit consultant Jarrett Walker. Image: SFCTA

Streetsblog asked Walker, author of the book and blog Human Transit, to weigh in on the debate between Geary BRT’s Alternative 3 (with bus passing lanes) and Alternative 3-Consolidated (without). In response, he pointed out that “if you combine both services into a single pattern, everyone gets more frequency.”

“Total trip time includes waiting time as well as in-vehicle time,” Walker wrote in an email. Providing one BRT service more frequently, rather than devoting some buses to local service, “should reduce average waits by enough to make up for a minute difference in travel time.”

“Rapid stopping patterns need to be much much faster than the local to justify having both,” he continued. “I have a little trouble believing that the differences in travel time are as small as [the SF County Transportation Authority estimates], but it would not surprise me if a single pattern ends up getting almost everyone where they are going sooner.” According the SFCTA, the “consolidated” BRT service would run its main stretch in seven minutes, while an express service would run it in six minutes, 10 seconds.

Jarrett Walker. Photo: ##http://pdcentre.ca/2011/07/jarrett-walker-presentation-now-available/##Planning & Design Centre##

Walker’s point on frequency seems very convincing, and since I’ve railed on the “consolidated” option as nothing more than a bid to preserve car parking for misguided merchants, an apology is in order. The benefit of shorter waiting times outweighing stop delays wasn’t made clear to me, and I was skeptical of a proposal that appeared late in the process — particularly since it would allow planners to dodge political backlash by saving parking spaces while reducing space for transit.

As Walker pointed out in his book, such misconceptions about quality transit service are common. “In most debates about proposed rapid transit lines,” Walker wrote, “the speed of the proposed service gets more political attention than how frequently it runs, even though frequency, which determines waiting time, often matters more than vehicle speed in determining the total time a transit trip will require.”

  • Mario Tanev

    Honestly, I am not a fan of the L (14L, 38L) and so on services for precisely this reason. They either stop running late at night, or their frequency is diluted by the locals. I think there may be a place for a line with BART-like stop spacing however. It is most useful to those making long-distance travel, whereas the normal BRT would be useful to those who want flexibility.

    On Geary past Van Ness this could be:

    Van Ness, Fillmore, Presidio, Park Presidio, Sunset.

    On Mission that would be useless because it would replicate the BART stop spacing. But since Geary doesn’t have anything like BART, such a set up might actually provide savings.

  • Anonymous

    Thank you for posting this! It’s exactly the point I was trying to make in the earlier debate.

  • Easy

    I worry about folks who are don’t want to lose “their stop” fighting & bloating the line back into slow, local service.

    Having a passing lane doesn’t preclude running the service as a single line, and gives the flexibility for buses to pass other buses when they are full or otherwise bunched up.

  • Michael Pieja

    Mr. Walker’s point is interesting, but doesn’t it assume that people are leaving their starting point to begin their journeys at essentially random times (e.g. without consulting realtime arrival info or a schedule)? If folks use real-time arrival data to time when to leave to go to the stop, the effect of service frequency on waiting time seems likely to be significantly less, at least as long as the frequency remains on the same order of magnitude.

  • david vartanoff

    What is it people don’t get about the utility of differing services in a single corridor? Some of us may be going out past 20th Ave others a few blocks. We are ALL customers. and in passing, why NOT provide a transfer stop at Divisadero??? If you are using Geary to access a crosstown, you DO want a quick ride. I should note that I have several times in the past pointed out that Muni could have improved the 38/38L in base day service hours simply by increasing the L’s in the mix. This came up when instead they tried the lame idea of cutting out all service at several stops in the Tenderloin, provoking legitimate anger from those who use them. The result was that Muni did nothing to improve service in that area and it is still abysmal.

  • Anonymous

    Ironically, I just made a comment at Alon Levy’s blog on the same topic about an hour before I saw this post, but using Denver as my example. http://pedestrianobservations.wordpress.com/2013/07/16/are-express-trains-worth-it/#comment-10163

    Limited-stop routes suffer reduced frequency at all stops as stated by Jarrett, but also uneven frequency at major stops, leading to guaranteed bus bunching. As many customers will take whatever bus comes first if they aren’t going very far, this can easily result in uneven vehicle loads, as most cram onto the first bus coming while the second bus is relatively empty.

    Express routes, on the other hand (38AX/BX), usually save a significant amount of time to be worth operating. By using a parallel, less congested street and running closed-door, buses are able to take advantage of traffic signals timed for vehicles and run at a high average speed.

  • Anonymous

    We should assume that people head to the bus stops at random times. For passenger convenience, rapid transit should be “show up and go” – you shouldn’t have to consult a timetable or NextBus before you head to your stop. I think both the 38 and 38L are already at ~8 min headways for most of the day, so we’re already in “show-up-and-go” territory.

  • Michael Smith

    This idea completely ignores the need for local service on Geary. Some people aren’t looking for increased frequency. Instead, they want the additional stops of a local. The local isn’t going to go away, especially since we don’t want the need for more stops to cause the BRT system to be watered down. Plus I can guarantee that the difference in travel time is going to be greater than 50 seconds. And yes, Jarrett is not taking into account the effect of the NextBus system.

    You were right in the first place. The lack of passing lanes is a horrible idea that is only being considered in order to preserve a few parking spaces. If San Francisco cannot build a reasonable BRT system then we shouldn’t waste money on yet another poorly designed project that we would be stuck with for decades.

  • Indeed, at their height the A, B, and C lines went to 10th, the Great Highway, and 33rd (respectively).

  • Anonymous

    This is a ridiculous assumption in the day of nextbus. If I had any inclination that I would be boarding MUNI in the next 30 minutes, my phone would be out checking nextbus, and I typically left my house such that my wait time at the stop would be on the order of 30 seconds. Unless frequency gets to the 3-4 minute range, *reliably* – nextbus has changed the game.

  • Andy Chow

    In the Richmond, 38L averages about 6 blocks per stop, and 38 averages 3 blocks a stop. Consolidating the service to about 4 blocks per stop only adds a about 3 to 4 additional stops for the 38L and about the same number of stop reduction for the 38 in that segment. Some folks may have to walk 300 (less than 1/2 the length of the BART platform) more feet to access the consolidated stop from the local stop.

    The distance of 4 blocks is about a 1/4 mile. This would be about the same distance as the Van Ness BRT.

    There should be enough ridership for an all-day express that stops at
    Union Square, Van Ness, Fillmore, Divisadero, Masonic, and then all BRT
    stops to Ocean Beach. This may be a better service option over the current limited set up, but that would mean a side BRT may be a better alternative in the eastern part of the corridor so that bus can bypass one another.

  • Andy Chow

    Not everyone is that tech savvy to check Nextbus all the time. Nonetheless Nextbus isn’t a substitute for not providing consistent and reliable service.

  • Ben Ross

    This is a great example of why people should stop looking at “BRT” as a package and instead look at what’s inside the package. Depending on circumstances, a dedicated bus lane can be most useful carrying all the buses in a corridor, carrying only limited-stop buses, or carrying express buses that don’t stop at all.

    If you start off by asking “how can we build the highest quality BRT” you are driven to choice #2 when one of the others might be better.

  • Anonymous

    Nextbus doesnt negate the importance of “show up and go” frequency.

    Would you rather walk to the bus stop whenever you want to go somewhere, or sit at home waiting for the right time to leave after checking Nextbus?

  • Anonymous

    Good point – let’s get rid of all this ridiculous parking and give Geary BRT a clear ROW and passing lanes so service is consistent and reliable

  • Anonymous

    I agree. Frankly we should be able to have 2 minute headways on the 38.

    I could check nextbus all I wanted but if there wasn’t going to be a bus for 20 minutes and I needed to be somewhere in 15, I’d get a cab.

  • Mario Tanev

    Jarrett Walker often talks about freedom (frequency is freedom). You’re right that NextBus alleviates the frustration of having to wait at the bus stop. But it doesn’t solve the freedom problem. You’re not free to go out any time you want and expect the bus to be there. The car gives you such freedom, and it’s important for transit to match that freedom.

  • Anonymous

    “Unless frequency gets to the 3-4 minute range, *reliably* – nextbus has changed the game.”

    Yes, and the whole point of Geary BRT is to provide reliable frequencies in the 3-4 minute range!

    You’re misinterpreting my point. For the purpose of planning Geary BRT, we should be assume that people head to the bus stop at random times. Obviously you would check NextBus before leaving if the line you were going to had 30 minute headways.

  • mikesonn

    My example: I ride the 30/45 lines every day. I leave the house assuming bus will be there because headways. This is often not the case. I go buy coffee (or whatever) and check nextbus to plan on actual next bus arriving. 3-4 min headways mean nothing if they bus doesn’t actually come that often.

  • Sprague

    Great point, Mario. A great transit system has frequent and reliable service. A great transit system provides greater freedom than an automobile because it allows its riders to continue their trip (ie. by hopping on the next bus) without having to return to wherever they left their car. Thank you Aaron and streetsblog for your advocacy of BRT on Geary that has the potential for delivering such frequent and reliable service.

  • Andy Chow

    If keeping those parking can lessen the opposition to the project, then it is a positive step. The passing lanes are just one of those nice to haves that would only provide some improvements in operational flexibility but not a lot, especially considering that bus service pattern can be changed to avoid having buses pass each other in certain areas and maintain travel time.

    This is no different than the “blended” scenario for HSR+Caltrain. It does add some operational constraints but gets most of the benefits with less impact to the communities.

  • mikesonn

    Great idea, let’s half-ass ALL THE THINGS!

  • Anonymous

    Sure, but that’s in congested mixed traffic lanes. The reliability issue should not apply to BRT in separated lanes. If we’re still seeing bunching on the 38 after this project is built, something has gone seriously wrong.

  • mikesonn

    Yeah, keeping parking is what has gone seriously wrong.

  • Andy Chow

    How is it half-ass when the proposal on the table is still a median busway?

  • mikesonn

    Yes Andy. You got me. Down with all autos. Give me all the road space! Because transit, cycling and walking already take up SOOOOO much of the available ROW in San Francisco. Best to keep what little left there is for cars so we can’t take away any more to make BRT actually be BRT. *smh*

  • Andy Chow

    This is a democracy and that there are many in SF rely on autos whether you or I think it is appropriate or not. This is a decent compromise to get the project going while delivering most of the transit benefits. I agree with Jerrett that we should review and question the operating patterns (local vs. limited).

    If the point is to fight every last inch, then keep fighting and keep stalling this project.

  • mikesonn

    Whatever Andy. You’ve really jumped the shark lately and you are fighting straw men of your own making.

    I want BRT that WORKS, not something that will only marginally improve the current horrible status quo on Geary. This doesn’t make me ideolog, it makes me someone who wants Muni to stop sucking so damn much.

  • Mario Tanev

    In theory it’s great to have just the right number of tiers of service people need (I would recommend a 38S which stops at every corner and runs once an hour to satisfy those complaining about stop consolidation). In practice you run into issues. Imagine you want to provide 5 minute headways overall, with 10 minute headways for the local and the limited each.

    1. If you need a limited, you know have a longer wait or you have to take the local half the time, negating the advantage of the limited.
    2. To provide the same headway, the limited needs more buses/operators since it’s slower, which means you sink more than 50% of your resources into it, which you could have spent in better service on your limiteds.
    3. Because many riders will just board whatever comes first, that creates bunching and uneven headways.

    Let’s not kid ourselves, the proposal here (and on Van Ness) is not for BRT, but for a European style, let’s not stop at every damn block, but still stop frequently, type service, with the extra amenities of BRT (dedicated lanes). After all, walking from Divisadero to Fillmore takes 10 minutes, which is about the average walk to a subway station in Europe. And BRT in its conception (Curitiba) is supposed to be a cheap substitute for the Metro, which means you shouldn’t easily be walking from one station to the other.

  • Andy Chow

    I think SF is too small to gain much from two tiers of BRT service. The local buses already make too many stops. Once you cut those stops then these routes would be too similar to the limiteds. If the ridership is strong it would be more useful to provide all day express service like the 8X that skips a lot of stops.

    SF is not LA, where the rapid/limited buses stop every 1/2 to a mile and where the routes are 15-20 miles long (compared to 6-8 miles of Muni’s).

    In Curitiba, the BRT stops are about 1/4 to 1/3 a mile and most stops don’t have passing lanes.

  • Anonymous

    Add a single wheelchair passenger into the mix and all bets are off. The design of Muni busses and platforms is all wrong. Other cities have grade-level boarding for wheelchairs and baby prams, and they don’t experience the same infuriating delays that handicapped passengers are forced to impose on the system (because of bad design, not their fault).

  • Anonymous

    We are talking about SF’s useless buslnes, not a useful and fast subway system. Most people need to make connections, which means unpredictable arrival times, because of unprotected transit lanes with shared intersections.

  • Anonymous

    This is the SF way of doing things in the public sector. What else would you expect?

  • Anonymous

    Then stop wasting money on all these pitiful incremental projects. Would be better to do one undergrounding of Van Ness for 10 blocks than to half-ass both Van Ness, Polk and Geary.

  • Anonymous

    “losing a stop” is more than about simply needing to walk and extra 200 meters. It directly impacts property values, customer foot traffic, and even area demographics.

  • Malena

    Why not to use some services like “http://www.goldparking.co.uk/ they are providing?

  • mikesonn

    Or not.

  • Easy

    Side-running BRT is almost never better, due to right-turning traffic (& parking traffic, if any) causing delays.

  • Easy

    NextBus doesn’t help on the second or third leg of a connection – other than letting you know whether you have to run to make the connection or can just walk. So high frequency is still preferable on any route you are changing to.

  • Droooo

    With the advent of NextBus the frequency takes a back seat, to me. I would rather have a faster trip knowing when the vehicle is coming than to trust that more frequency will deliver me faster, overall.


Geary BRT Plan Watered Down to Appease Parking-Obsessed Merchants

Update: This plan may not be “watered down” after all. See our follow-up report here. Planners are touting a new proposed configuration for Geary Bus Rapid Transit that would forgo bus passing lanes in order to preserve car parking to appease merchants. Separated, center-median bus lanes would be retained, and project backers hope the changes […]

Options for Geary BRT Come Into Focus

Just after San Francisco approved a preferred design for its first Bus Rapid Transit route on Van Ness Avenue, the SF County Transportation Authority showcased the latest conceptual proposals for a companion BRT project on Geary Boulevard. Geary BRT, which has been fraught with delays over the years, is expected to bring relief to riders on Muni’s […]
The County Transportation Authority warming up for a long afternoon and evening of comments before the final approval of the EIR for Geary BRT. Photo: Streetsblog

Geary Bus Rapid Transit Study Approved by County Transportation Authority

Yesterday evening at San Francisco City Hall, the County Transportation Authority Board unanimously approved the Geary Bus Rapid Transit project’s design and Final Environmental Impact Report (EIR). The approval brings the $300 million project, which has been a decade in the making, one step closer to fruition. For any readers just getting up to speed on […]

Is the Geary Bus Rapid Transit Project in Jeopardy?

Photo: plug1 If the Geary Bus Rapid Transit (BRT) project doesn’t get some love from advocates and the general public, the project could be in trouble, according to several people closely following the process. "I look to the left, I look to the right, all I see is opposition and criticism," says Joel Ramos, a […]