The one-seat slowdown

In the article about the 19th Avenue Subway, I wrote briefly about the plans for the J to cover the portion of the M route in Ocean View (Randolph, Broad, and San Jose). Residents of Ocean View are now faced with a trade-off when trying to get downtown. They will experience a faster ride if they take the J to SFSU Station and transfer to long, fast M trains to continue inbound. However, some people are not happy about the loss of a reasonably quick one-seat ride, since their current route will be replaced with the J coming all the way around through Balboa Park. The J is going to spend a lot more time in mixed traffic than the M does currently, and so taking it downtown from Ocean View is likely to be a slow and unreliable ride. While we should think about what will keep the J reliable, we might also want to consider doing something else to mitigate the concerns of Ocean View residents.

Let’s take a look at the N. A large percentage of this route is at-grade, and getting from one end to the other is very slow. This sounds familiar! During commute hours, when subway space is limited and people from the forties need to get downtown in a reasonable amount of time, Muni runs an express route along this line. The NX-Judah Express aims to get people from further out to downtown quickly by skipping lots of stops in the middle and using faster one-way, timed-light roads where possible to keep the buses moving. Plenty of bus routes have matching express routes during commute hours. The 1-California and 38-Geary each have two express routes which use Bush and Pine to get downtown quickly. The 7X-Noriega Express uses Oak, Fell, Franklin, Gough, Golden Gate, and Turk. Express routes for the 8 and 14 use freeways. But most light rail lines do not need an express route since the subway generally obviates the need for them. To support a light rail express line, we need to be able to get people from far-flung areas onto fast roads and keep them there until we get close to downtown. With the J extension to SFSU, the line might become eligible for such treatment.

280 and 101 can be used to transport people express from the Glen Park areas to downtown more quickly than it would take to crawl all the way around and finally get into the subway. Stops between 19th Ave & Randolph St and San Jose Ave & Santa Rosa Ave would be serviced by the express bus, after which it would take 280 via Baden and either Circular Ave or Monterey Blvd. From there, it can get on 101 via the Alemany Maze interchange. The Central Freeway will allow the bus to exit on Mission to service Van Ness & Market. The bus can make normal J stops up Market from there. (Other designs might make different stops downtown or use a different route to get there, as the NX tries to serve the Financial District more specifically. I tried to make this serve existing downtown J/M stops as closely as possible.)


Now that the JX services stops up to Santa Rosa, we can turn around more J trains at Glen Park. This already happens during commute hours, but it’s rare, and with the JX to supplement service past Glen Park we will be able to get away with turning some trains around more quickly than we would be otherwise. This would help keep the route more reliable for residents further inbound on the route.

Of course, there are other concerns at play here. I don’t normally like introducing express buses, because they are an expensive proposition with regard to how much service they actually provide. Muni will be forced to increase its peak bus fleet to accommodate the extra commute-time load. The J station at Glen Park is somewhat confusing if trains are turning back there, since trains can’t cross over to the inbound track from the outbound side. I’m not sure how expensive installing this extra track would be. Furthermore, extremely heavy traffic on the freeways may mean the bus won’t save any time after all! However, if this idea gets more people on board with the 19th Avenue Subway by assuaging concerns over the reliability of the J and preservation of a somewhat quick one-seat ride to downtown from Ocean View, then I think it’s worth studying.

The Tenderloin gap, part III

The last two parts (12) of this series focused on a north-south bus north of Market, but we never laid out the remainder of its journey further south. I think there are two major strategies for such a route—bring it into a major connection point or continue with the north-south gridline. What would these look like?

The most easily accessible major connection point is 3rd & 20th, with connections to the T, 22, and 48. With such a design, this route would effectively be participating in a secondary downtown radial pattern along with these routes. The T brings people in from the north and south, the 22 from the west and far northwest, and the 48 from the southwest. This bus would then bring people in from between the 22 and the T. This is even more prominent once the T is sent through the Central Subway.


If we were to do this, land use at the nexus would need to support this anchoring of many frequent lines. Mission Bay is nearby, but the neighborhood is somewhat concerned about its expansion southward. The area is already zoned “urban mixed use” (UMU), which is a conversion from old industrial land use types to support more residential and commercial development. Think lofts built in old warehouses with ground-floor retail, and the like. While we like mixed use zoning, most of the parcels in this area have height limits of between 40 and 68 feet. This does not allow for very much density, and surely would not be able to support a second radial pattern. Perhaps the new Warriors stadium will provide the Planning Commission with a reason to raise the height limits in the area, but I wouldn’t hold my breath for that. Given that the T and 22 are such high-frequency and high-capacity lines and we plan to add another one, we’d probably want to raise height limits to at least 100 feet to take advantage of this. The AHBP might have been useful to provide taller buildings with more affordable housing in the area. While raising the height limits is a possibility, it is a remote one and we might be overserving this area quite a bit in the meantime.

If we were to continue the gridline to the south, we have a perfectly nice candidate already available—the 19-Polk! Given that the 19 is already bracketed by this bus and the 47/49, all of which are designed to be more frequent, we can scrap the 19 and use its southern leg for this bus instead, which leads to an increase in service for those riders. Since the 19 hits Market so close to where this bus would, it seems duplicative to try and fit an extra gridline in Potrero Hill, and we would probably have to introduce some confusing hopscotch to serve 6th St instead of 8th. This isn’t ideal because 6th St doesn’t cross over into Potrero like 7th and 8th do, but 8th St would be overserved if we ran both this bus and the 19 on it.


Muni Forward

SFMTA’s Muni Forward plan has a number of service changes planned which, if implemented, change the calculus for the two options outlined above. Under the proposed changes, there will be a lot of rerouting on the east side. The 9R-San Bruno Rapid has added a stop in front of San Francisco General Hospital, so the 33-Ashbury/18th will no longer turn down Potrero Ave and go down to the hospital. Instead, riders will be expected to transfer at 16th St and take the 9R one stop south. The 33 will instead continue eastward and take over this current leg of the route for the 22-Fillmore, which uses 17th and 18th Sts to eventually end up at 3rd & 20th. The 22 will then be rerouted to use the current route of the 55-16th St, and serve Mission Bay. The 19 is also rerouted to serve SFGH under this plan, with the 48-Quintara/24th taking over its route to Hunter’s Point. A new 58-24th St will take over the eastern part of the 48’s route and go to 3rd & 20th.

This would break up the 3rd & 20th radial hub, as the frequent 22 is replaced with the less frequent 33, and the fairly frequent 48 is replaced with the infrequent new 58. Add all this to the zoning changes needed to maximize the benefit from creating this radial hub, and it looks like this option will not work for us in the long term.

On the other hand, replacing the 19 south of Market is still a good idea: our frequent gridline has an anchor (SFGH) and the route is fairly straight. Thus, our bus can take over for the 19 no matter how many of the Muni Forward recommendations are implemented.


This new route also has the advantage of getting rid of that confusing bit it encounters as it crosses over Cesar Chavez, although the route is shorter, so some riders will have to take the post-Muni Forward 48 to 23rd & De Haro or Rhode Island to get on the new bus. In any case, this is still better than the other option should the Muni Forward recommendations come to pass. The pain of making this connection will be ameliorated by the increase in frequency.

Overall I like rejiggering the 19 instead of creating an entirely new line, since I think it works better with the land use. The other option looks nicer from a network perspective, but it is viable only if Muni does not follow through on some of its proposed changes, and if we get some help from the city to upzone the 3rd & 20th area. In effect, this ends up being a change of the 19 route, which shouldn’t be such a big deal—Muni Forward is already full of big route changes!

Completing the grid: The Tenderloin gap, part II

Last time, I talked about what I felt was a gap in Muni service. I’ll try and sketch out a line that would fill in this gap.

Since we are talking about a gap in service between Van Ness and Stockton, the best way to start would be to try and send a line right down the middle. As they are nine blocks apart, the middle lies between Leavenworth and Jones. We’re going to want to pick one of those streets to serve as the base for our line. A bus along this corridor will serve:

  • Civic Center Station
  • Boedekker Park on Eddy
  • Tenderloin Playground on Ellis
  • St. Francis Hospital on Bush
  • Grace Cathedral and Huntington Park on California, bordering which are many large hotels as well
  • The Pacific Ave neighborhood commercial district
  • Michelangelo Playground on Greenwich
  • The squiggly bit of Lombard for tourists, Yick Wo Elementary, and George Sterling Park if you’re willing to hike up to it
  • Fay Park and the San Francisco Art Institute on Chestnut
  • Russian Hill Park on Francisco
  • Conrad Square, Aquatic Park, and Fisherman’s Wharf

We should also allow easy transfers to east-west lines in the area, which would be the 1, 2, 3, 5, 10, 12, 30, 31, 38, 45, and 47. Of these, the 1, 5, 30, 38, and 47 are part of the frequent grid, so connections with these are especially important.

Designing the line

Our first instinct might be to use both streets in a one-way split, but one-way splits have a number of drawbacks. They make lines more confusing, as I talked about last time. They also reduce the area that is within walking distance of both directions of the line—a good explanation of why (and why we want to avoid this) can be found here, with diagrams. Our hope is to minimize use of one-way splits where possible.

Leavenworth and Jones are zoned similarly in Nob Hill and Russian Hill, so that’s not going to inform our decision. Since the 19 exists on the Van Ness side, and northbound buses south of Geary are further away than Stockton, let’s try Jones first.

North of California, Jones is a two-way street, so we’ll use it for both directions of this new bus line. Most buses in this area go up to North Point, so that’s our main destination, and then riders will easily be able to transfer to the 47 to head east and west. The 30, E and F are also very close to North Point & Jones. We’re also going to need to turn the bus around, and we can do so with a small one way loop like on the 19, extending from North Point to Beach.


If we look closely at this map, though, we’ve done something impossible—Columbus Ave has a median in some areas, and one such place is right where Jones would cross it. We can’t go through here. Our proposal might include breaking up this median and only allowing Muni vehicles to cross. This would require a lot of extra signaling and would complicate traffic patterns on Columbus that the median was presumably intended to simplify, so I’m going to consider this unsatisfactory. We can look at some changes in routing that avoid this problem section of Columbus.


Both strategies take us to North Point. Each one has its own complication: the first suggestion makes the line a bit wiggly, and the one-way loop terminating the second option is needlessly large. I’ll choose the first option for now.

Below California, Jones is a one-way street running south. We can make the bus use Leavenworth for the northbound leg, turn right on California, left on Jones, and continue up with the path as mentioned before. We’re trying to avoid one-way splits as much as possible, but here it seems justified. These streets are one-way, and since they are less than 500 feet apart, the one-way split doesn’t affect the service area too much.


For now we are running the bus down to McAllister, which is right before it hits Market. Let’s try out our route. We send an imaginary bus up from Market on Leavenworth, where it turns right on California, and left up Jones. We get to Green St and start going down a huge hill… in fact, this hill is so steep that our bus goes into free-fall!

The steepest grade any Muni vehicle negotiates is 22.8%, which is along the 24-Divisadero as it travels on Noe St between 26th St and Cesar Chavez. The 24 uses trolleybuses specifically because this hill was too steep for motorcoaches to climb at the time. Modern motorcoaches can do the job, as you might find when a Castro street fair closes off a whole bunch of streets and they can’t use the overhead wires anymore. The buses and trolleybuses might be able to transit steeper grades, but we’ll consider it an upper bound for now, since we don’t actually know if anything steeper will work. Jones between Green and Union is a 26.2% grade—even if we could get down this hill, it’s immediately followed by a 29% grade between Union and Filbert, and going back up later is definitely out of the question. (You can find official grade maps here.)

We can try and route around this hill as well, but we risk making the route too complicated. I’d rather move the base of the route from Jones to Leavenworth. The steepest part of the line north of California is a comfortable 22.2% grade between Union and Filbert, and the end of our route already moves over to Leavenworth anyway. This generally serves the same area while keeping the line as simple as possible. We’ll still try to use Jones south of California because Leavenworth is one-way northbound during that segment.


Let’s try the imaginary bus again. We send it up Leavenworth, where it makes it all the way to North Point, loops around and starts back towards Market. We come down to California, make a left and then a right around the odd guard rail… and promptly bottom out on the crest! Jones has a 24.7% grade between California and Pine, so putting a bus on it is not a safe bet. This means we can’t use Jones here, either! We have to explore some backup plans.

  1. Use Hyde to get to Bush or Post, take it to Jones, and use that to go towards Market.
  2. Use Mason to get to Pine or Sutter and take it to Jones.
  3. Use Hyde the whole way down to Market.
  4. Turn one lane of Leavenworth into a contraflow bus lane.


Option #1 recreates some of the leapfrog one-way splits we saw in the confusing 27 route, which we’re trying to avoid in creating this line, so that’s not a good idea. Option #2 entails a 3 block wide one-way split, which is one of the problems with the 8 and 30 that we are trying to work around by having a mid-Tenderloin line. Option #3 is workable, but risks nudging the service area a little too close to the Van Ness bus brigade. It does not require drastic infrastructural changes, so we can keep it on the table for now.

Option #4 is very interesting. Contraflow bus lanes are often used to turn one-way streets into two-way streets for transit purposes. Other traffic will only be able to go northbound on Leavenworth, but buses will have exclusive access to a southbound lane. I don’t believe they are currently used in San Francisco. This requires a little bit of infrastructure work such as painting the lane red and adding proper signage, as well as adding traffic lights at each intersection. An intermediate proposal might use a contraflow lane for a short time until the bus can get over to Jones, probably via Bush or Post. However, Leavenworth is a large street south of Geary. If we are going to use the contraflow lane north of it to avoid the California crest on Jones, we might as well take a lane out of the wide part to improve simplicity and speed.

Here are our possible alignments:


Through any of these options, we can connect Fisherman’s Wharf with Market at Civic Center Station.

Stop selection

Local buses in San Francisco stop every two to three blocks, so we can follow this model when picking stops. Ideally, we want to line up the stops with the east-west lines, so we can maximize its usefulness as a grid line.


If both the northbound and southbound trips use Leavenworth, then all of the stops north of Market are marked here. If we use Hyde for the southbound leg, we can use the same set of cross streets, but we’ll want an extra stop on Grove before continuing southeast on 8th St. The McAllister and Grove stops are very close to Civic Center Station. I also added an optional stop on Bush to serve the hospital more directly, since hospitals are destinations that you might want to add stops next to even if there are other stops nearby. For precedent on this in the current Muni network, check out the 9R-San Bruno Rapid, which stops in front of SF General Hospital on Potrero Ave even though it stops again one block away on 24th St.

Unfortunately, we are not done—this route only takes us to Market, and grid lines are more useful when you don’t have to think about where they start and stop. In the next installment, I’ll draw the rest of the line south of Market, and we can decide what a good anchor for the other side of this frequent line would be.