downtown networks: the oblique approach

How important is it that the transit services converging on downtown all go right to the core of downtown's demand?  In a city with a strong downtown, a reactive, demand-driven view of planning will add more and more transit into downtown as demand requires, because the loads are always highest there.  But sometimes, it can make sense to avoid the center of downtown, or to slide past downtown along one edge.

Zach Shaner's new proposal for the core area of Seattle is a nice example.   (Click to enlarge and sharpen, or see JPG here:  JPG )


You don't have to know Seattle to follow this.  The core of downtown is the little shaded box on the left, where LINK light rail will be in a subway and other frequent buses will run on the surface.  Most of the rest of the map area is also dense, so this is an area where you would expect a high-frequency grid to be possible, easily serving trips from anywhere to anywhere. 

But the map area is also ferociously hilly, so fitting the classical grid to this terrain has always been a struggle.  Still, Zach suggests that once the LINK light rail extension northward opens, ending at the University of Washington just off the north edge of the map, something remarkably gridlike might come into being. 

Several of the design principles that I often advocate are at work here.  First, Zach has focused on running on the fewest possible streets so as to provide the maximum possible frequency.  Second, he's mapped both the existing and proposed networks so that you can see this frequency difference.

But he's also provided a nice example of the oblique approach to downtown, which is a way of balancing the distributive quality of a grid with the need to serve downtown's concentration of demand. 

Look at the magenta line coming from the southeast.  It's Line 7, now the single busiest bus line in Seattle.  Off the map to the south, it extends about 5.5 km (3.4 mi) onward through dense, old, and historically low-income areas of the Rainier Valley.  It touches the rail line at Mount Baker station, just inside the map, but it flows onward into the city, currently going right through downtown.

Zach proposes instead that Line 7 approach downtown but then bypass it just to the east, via Boren Avenue.  He would do this to create a new crosstown (north-south) opportunity to reach dense north parts of downtown (including the South Lake Union redevelopment area and Seattle Center, which is right where Line 7 exits the map on the west edge) without going right through downtown's heart.

What would the dense downtown ridership of Line 7 do?  Well, some of it would transfer to LINK where the two lines touch at the south edge of the map.  Others, depending on exactly where in downtown they were going, might transfer to any of several very frequent east-west services crossing the line.  But many who now seem to be going downtown would reveal that they're really going elsewhere.  This is the key.

Line 7 ridership shows massive boarding and alighting downtown, and downtown is certainly an overwhelming destination.  But you always have to ask if the concentration of downtown boardings is evidence of what people want or what the current network structure requires of them.  People going to the dense and fast-growing areas on the north edge of downtown will find Zach's revision a dramatic improvement.  Many others may be going still further north, to anywhere in the northern half of the city, in fact; they currently have to go through downtown to make their connections, because that's where their bus goes, but they could make the same connections to northward services from the revised Line 7, often with quite a savings of travel time.

Obviously, I don't have the data to validate exactly how many people are in each of these categories.  I'm not even advocating Zach's design, but I do think it illustrates an important design concept, one that you will never think of if you're focused entirely on where your current ridership patterns seem to be leading.

Many major cities are facing unmanageable volumes of buses squeezing through a tight downtown.  Sydney, where I lived for five years, has a remarkably similar predicament to Seattle's.  One solution, in Sydney as in Seattle, is to let go of the idea that radial services aimed generally for downtown need to go right through the very center of it.  If there is a sufficient diversity and richness of connection opportuntiies for reaching various parts of downtown, you can often create a better design by sliding past downtown obliquely, as Zach proposes that Line 7 should do. 

I don't advocate or oppose Zach's design, but I do think it's a nice illustration of how to fit the "everywhere to everywhere" network design principles (such as the high frequency grid) to a very difficult terrain.  It also raised some interestingly contrasting comments from different Seattle transit experts who've seen it; more on that in another post.

UPDATE:  Zach explains his proposal to the Seattle transit community here.

8 Responses to downtown networks: the oblique approach

  1. Mike Lindblom, Seattle May 13, 2011 at 3:22 pm #

    That’s an interesting concept and Boren Avenue (right past three hospitals, a high school, condo towers, public housing and cathedral) would greatly benefit from gridding. A transit drop from Boren into South Lake Union would fill a huge gap in service now. To make it work, four challenges: 1) Riders arriving from the 7 (or other crosstown routes such as the 48) don’t necessarily want a two-stop trip – so called “transfer penalty” even if headways are frequent. 2) dwell time is terrible on local routes whenever folks are not physically agile, or pay cash; 3) spillover cars clog Boren as they queue for the downhill crawl to I-5 freeway entrances and 4) parking must be reduced or eliminated on some streets, notably Madison, for buses to flow freely on the east-west part of the grid. But these are not insurmountable.

  2. mikef0234 May 14, 2011 at 2:11 pm #

    If a street has many routes operating on it, shifting some of the routes to a parallel street may make it so that two streets have frequent service. You might also end up with two streets that both have too-infrequent or too-unreliable service, especially if the routes sharing a street aren’t coordinated and especially off-peak.
    In Vancouver, Granville and Burrard downtown are a bit like the former, though Burrard is a bit infrequent off-peak and it seems as though the trolleys on Granville are coordinated to run in a convoy. Hastings and the Powell/Cordova couplet are a bit of the latter, and could maybe be combined to improve frequency on Hastings.

  3. Morgan Wick May 14, 2011 at 3:43 pm #

    See Zach’s original proposal for needed context, including the fact that part of the purpose of this redesign was to provide the Boren Ave service to a part of the city that had actually been requesting it since the First Hill Streetcar (the brown line east of Downtown) was determined not to run on Boren.
    Also, note that the yellow Line 8 is already a “periphery of downtown” route, but only on its north edge, and it serves different markets than the 7.
    Note also that even people who agree on the need for more crosstown routes can quibble with this plan. A historical crosstown variation of the 7 is Line 9, the north-south cyan line on the current map, which once-upon-a-time continued to the University District along the purple Line 49, and was an all-day frequent local instead of the express it is today. Restoring this route as the main north-south corridor in the Valley, following the gray line Zach uses for the 12 north of Yesler Way, would better reflect crosstown commute patterns that have been recognized in the past. Then the southern part of Line 12, which serves Beacon Hill (currently Line 36, which turns down Jackson to downtown) can head up Boren.
    Note that the current Line 60, coming from roughly the same direction, runs a couple blocks west of Boren between Madison and Yesler, before heading down Madison to provide redundant service on Broadway, more closely approximating the Boren corridor south of Madison. I’ve proposed keeping the 36 on its current downtown routing and stretching the 60 along the Boren corridor.

  4. Jarrett at May 14, 2011 at 5:37 pm #

    Morgan.  Thanks for the context, which is certainly necessary for forming a detailed view on this proposal.
    Once again, I do not endorse or oppose the proposal, but am merely using it as a good illustration of the oblique-approach concept, which I tried to present in a way that's clear to non-Seattle readers.

  5. Multimodal Man May 14, 2011 at 9:18 pm #

    The key is the grid has to be frequent. People often want some direct, non-radial connection but because it makes infrequent connections it turns out to be less useful (Jarrett has a good illustration for this). I would argue that 7.5 minute frequency for Link is not enough frequency to assume away frequent core routes from downtown Seattle. Route 2 moving to SLU and then Boren seems to assume by the look of the map that the Monorail provides a useful replacement for the 2.

  6. Zoltán May 15, 2011 at 5:13 pm #

    I’m interested in the possibilities for some cities with a largely radial network (sometimes dictated by the street layout) to combine that with a grid downtown.
    Something such would work along the lines of each route taking the closest and most obvious downtown street, and leaving by the closest and most obvious radial route after that, and forming a grid downtown with each route heading north-south or east to west as directly as possible. Baltimore is somewhat like this already, and I’ve been thinking about possibilities for a network redesign that would make it more like this.
    The benefit of doing that is that if a city can’t support the frequencies required to make the entirety of a grid pattern work, there are probably enough buses downtown to make high frequency possible, so downtown, at least, it can ensure simple two-seat rides to every different part of downtown you might want to reach.
    On the other hand, it reduces the number of transfers easily available downtown, in that you can’t easily transfer between routes running parallel through downtown, designed to never cross, which potentially makes anywhere-to-anywhere travel more difficult. This Seattle proposal has enough right angles in it to avoid that, but that’s largely a matter of getting around geographical obstacles.

  7. Rob May 15, 2011 at 10:56 pm #

    I haven’t had time to take in all the details, but I really like the thrust of this. Some picky stuff — Boren is an awful route because it stacks up with freeway-bound traffic. Transit there has not been reliable in the past. An alternative might be to use 8th, assuming the city would allow some contraflow treatments on the north side of the freeway. I can imagine a transit-emphasis corridor there that would serve First Hill, South Lake Union and maybe Queen Anne, with a direct connection at Convention Place station allowing more people to access all of those places.
    One thing that’s particularly great about this sort of thinking is that most of the future growth in downtown Seattle will be in the peripheral areas around it, on First Hill and South Lake Union. But these days very little transit serves either, and Link light rail will not contribute to serving these areas at all – in fact, when buses are kicked out of the transit tunnel, convention place station will close and there will be less effective service to the Denny Triangle area where the most immediate growth is likely. Getting more use and connections out of the existing bus network will be critical, and once Link opens there will be no reason for every route to follow the trains downtown.
    It’s hard to know for sure which new connections will be the best ones. Most transit planning is incremental, meaning that transit responds to demands on the system. I think in this case modeling might be useful to show which new connections would have the best ridership response, and which existing routes would lose riders if a transfer was needed to reach downtown.

  8. Jack Horner May 16, 2011 at 5:43 am #

    Similar issues arise in Melbourne, which has a Central Business District of 1600×800 metres (1 mile times ½ mile – the ‘Hoddle grid’, easily recognisable as it is offset to the north south grid of the inner suburbs). CBD-like activity now extends into nearby areas.
    The City Council has proposals to extend tram services on the CBD fringes (the darker lines on p.33 at Some of these are oblique in the sense discussed.
    Melbourne has good public transport service and high public transport mode share to the CBD. PT mode share drops sharply to CBD fringe areas outside the Hoddle grid, presumably because of the combination of worse service and easier parking.
    Remedying that will require more attention to through city and near city public transport. This needs a change in the transit planner’s mindset from ‘one seat to the city’ to ‘any time, any place in the inner area, using transfers’.