Cul-de-sac Hell and the Radius of Demand

This is interesting:

Research by Lawrence Frank, Bombardier Chair in Sustainable Transportation at the University of British Columbia, looks at neighborhoods in King County, Washington: Residents in areas with the most interconnected streets travel 26% fewer vehicle miles than those in areas with many cul-de-sacs. Recent studies by Frank and others show that as a neighborhood’s overall walkability increases, so does the amount of walking and biking—while, per capita, air pollution and body mass index decrease.

I especially appreciate this graphic, because it’s a nice illustration of a crucial transit concept: the radius of demand:


(The image on the left happens to be Woodinville, a suburb of Seattle, while the one on the right is Seattle’s Ballard district.  But they could be any comparison of c. 1950 and c. 1880 street networks in the New World.)

Imagine that the red dot in the center of these images was a rail rapid transit station, as these can generally draw walk trips from a radius of about 1 km.  Most planning maps that you see will simply draw a circle of 1 km radius around the station, and call that the “catchment” or market area of the station.

Cul-de-sac circles

A circular radius of demand is often called an “air distance,” or as we used to say, “as the crow flies.”  Humans, however, have to go around barriers.  These include not just major walls such as a freeway (the wide orange line in the left image) but also simply the absence of a path in the desired direction.  So the real radius of demand is defined by the network of streets and paths.  Actual walking distances along the road network are indicated in blue.  Until recently, you didn’t see these diagrams often in planning studies, because walking paths were just too complicated to calculate.  Now, thanks to increasingly rich GIS databases of pedestrian networks, it’s getting easier.

The beauty of fine-grained grid street patterns (not to be confused with the beauty of transit network grids) lies in your ability to reach most of the area of the 1 km circle in a 1 km walk along the network.  We can’t expect any neighborhood to offer a 1 km walking distance to any point on a 1 km circle.  (Such a goal would fill the entire area of the circle with needed walking paths, leaving no room for a city.)  But the classic grid structure of streets is remarkably good at maximizing the walk radius while still leaving space to build things.  In the image above, the walkable radius in the suburban cul-de-sac street pattern on the left is less than 30% of the area of the circle, while in the fine-grained inner city street grid on the right, it’s 64% of the circle.

Transit, in short, has a direct interest in steps to increase the permeability of the street network, steps that make a street network more like the map on the right and less like the one on the left, because such changes enlarge its market.  Cities that value transit (or walking or cycling for that matter) should be looking for strategic interventions where small additional segments of ped-bike path can dramatically shorten ped-bike distances and thus expand the actual walkable radius of demand around a transit station or stop.

Here’s an example from Hayward, California, which I remember from my brief 1990 stint at AC Transit:


The triangle in the center is a large mobile home park.  It is about 750 m in extent from north to south, and bounded by the freeway on the west and a fenced drainage canal on the east.  The only street nearby on which you could run a reasonably frequent transit service is the east-west arterial on the bottom of the image (Industrial Parkway).  (The local bus system actually does something more complicated than this, but that’s beside the point.  Clearly, if you were trying to run relatively fast and efficient bus routes, you’d stay on major arterials like Industrial Parkway, and focus on improving the pedestrian access there, rather than driving on much slower and more circuitous paths into the surrounding neighborhoods.)

At the very bottom of the triangle is a bridge over the canal.  My recollection is that in 1990, that bridge wasn’t there.  Regardless of whether that’s true, the point is that if the bridge weren’t there, almost everyone in the mobile home park would be more than 400 m (1/4 mi) walk from Industrial Parkway even thought most of the park is within that air distance.    This kind of situation was built all the time.

The little bridge at the southern tip of the park obviously changes everything.  Now, most of the park has a reasonable direct walk path to a stop on Industrial Parkway.  That’s better mobility for the people who live there, for access to nearby commercial as well as to transit.   I point to this bridge because it’s the kind of small intervention that cities and local transit advocates should be looking for.  It doesn’t have to be a street, of course, just a pedestrian link.  Such links can do a great deal to re-connect highly disconnected suburban street patterns, and thus bring much larger areas into the walk radius of transit.

To see the effect of an abundance of these links, here’s a slice of Australia’s sprawling national capital city, Canberra.  Even in the most car-dominated and sprawl-loving era, Canberra has always maintaned an ethos of connection to nature, and has manifested this in abundant park strips — most of them also off-street bicycle paths — lacing the city.  These paths completely change the game.  Here’s a bit of the Canberra district of Florey, designed in the 1970s.  It’s at the same scale as the slice of Hayward above.


The east-west arterial across this image is a very important street for the transit system.  In fact, our Strategic Plan for Canberra’s public transit identified it as a bus rapid transit corridor.  Right now, it’s a big fast arterial, but notice that as it turns into a transit corridor, its stops will have very easy access into the surrounding suburban fabric, even though that fabric is made of classic car-oriented design.  The key is that while there are plenty of cul-de-sacs, they are almost always “pierced” by a pedestrian link that allows people to keep walking in a desired direction.   Zooming in, you can see these links into the parkway corridor from each of the cul-de-sacs at the top of the image:

Florey zoom

In short, this structure is a labyrinth for cars, as cul-de-sacs are meant to be, but for pedestrians and cyclists it’s more like a grid.  Every city should be looking for places where they can retrofit this sort of pedestrian connection.

32 Responses to Cul-de-sac Hell and the Radius of Demand

  1. Adam Parast May 12, 2010 at 5:04 am #

    This is exactly what I’ll be working on when I graduate in a few months. I few notes.
    – New pedestrian/bike connections does not equal new road. In fact not building a roads gives non-motorized transportation an advantage could be enough to get people to walk or bike to a close destination.
    – A technical way of doing this analysis is doing a crows fly to network distance comparison. This is called a route directness index and it is the foundation of what the analysis tool I’ll be working with (

  2. Xavier May 12, 2010 at 6:01 am #

    Ha ha good ol’ Canberra 😉

  3. Beige May 12, 2010 at 6:49 am #

    This leads to another important point: Where I live, that bridge would be repair priority number eleventy-billion if it had to be closed due to damage. Hey, you can always drive the long way around. If the bridge and the roads connecting to it were fine but a cross street was under repair, the intersection would be dug into a deep trench in early spring, and left totally blocked until the very end of the project in late fall. Other, “more important” intersections would get temporary gravel paths across except for the few days the actual work makes that impossible, but this route wouldn’t be important enough, you can always drive the long way around.
    Even when the shortcut exists, the department of cars and highways needs to begin thinking about the possibility that a multi-kilometer detour down a narrow two-lane with no sidewalks and no shoulders won’t meet every need.

  4. Tom West May 12, 2010 at 7:24 am #

    With an ideal grid system, the area within 1km walk along the street is about 63% of the crow-fly area (a 1km circle). (Tobe exact, it’s 2/pi).
    The air vs. ground distance ios important when it comes to route spacing. Again, going with a 1km walk distance an ideal grid, the “air distance” implies you could put rouets every 2km. However, the groudn distance shwos you have to place routes every 1km with each route’s stops offset by 0.5km from adjacent routes to cover everybody.
    The offset would be fine if all grid streets were the same, but they aren’t… you always get major grid streets, and the intersections of those attract stops, which leads to a sub-optimal system.

  5. Herbie May 12, 2010 at 8:24 am #

    In St. Louis, the city closed a fenced opening that people on one neighborhood had been using to access a local MetroLink station. Closing that fence meant residents were ~600-700 meters from the station instead of less than 400 meters.
    Many other MetroLink stations, particularly Blue Line stations, have severe access issues within the radius of demand.

  6. David in Ottawa May 12, 2010 at 8:54 am #

    Sounds rather like you’re advocating the “fused grid”, though perhaps without knowing it.
    Unlike the Transitway, this Ottawa-born idea might actually be a good one (though in classic Ottawa style, good ideas born here don’t get implemented here – just the bad ones). Decades earlier, Noulan Cauchon, founder of Ottawa’s first town planning commission, came up with a hexagon-based grid system, but that system didn’t get implemented here or anywhere else since it was proposed in the 1920s and by the time the post-war building boom got underway the cul-de-sac model was the preferred model. At any rate, hexagon-based grids ensure a higher degree of connectivity than rectilinear grids, meaning that even more of the area within a given radius is reachable*, and, moreover, given the nature of hexagons as compressed circles, it’s far easier to devise routes that reach everyone without leaving diamond-shaped areas that go unserved or creating areas of significant overlap, avoiding some of the issues that Tom West raises in his post.
    *As an aside, you haven’t given a name to the ratio of the area reached by the stated walking distance against the claimed catchment area as set by the air radius… it would be helpful to have a term for this concept.
    Lest anyone think I’m a cheerleader for Mr. Cauchon, I’m not. He was instrumental in getting shelved for good a previous City Beautifulesque plan for Ottawa (the Holt/Bennett Plan) that would have ensured the survival of rail for both inter and intra-city travel in the heart of downtown Ottawa. He was the beginning of an anti-rail ethos in Ottawa’s planning history that has done long term damage to Ottawa’s urban fabric and which continues to this day in its bizarre obsession with BRT. He was an enabler to the bus jams that Ottawa now suffers from daily.

  7. John May 12, 2010 at 9:56 am #

    I’d like to point you towards what I think is a beautiful street grid in Shaker Heights, OH (Cleveland) designed specifically to minimize walking distance to LRT stations. Look at how the streets are angled towards the rail stations in the middle of Van Aken Blvd.

  8. Tessa May 12, 2010 at 11:42 am #

    The only problem with those kinds of pedestrian connections in cul-de-sacs that you mention is that they’re often really hidden. I remember when I first moved into a new suburban neighbourhood in North Vancouver, I would peer down streets wondering whether there was a convenient pedestrian connection at the end of it, or wondering whether I would get stuck at the cul-de-sac and have to retrace my steps back to the less direct connecting road. It’s a real pain for anyone who’s not familiar with an area, and it’s likely to affect decisions as to whether to use transit if they don’t know for sure what the best route is.
    I’m personally not a fan of fused grid. It’s one of those things that should only be looked at in a way to repair a much more damaged street system, while new areas should really stick to the traditional grid, I think, or at least a similarly well-connected street pattern.

  9. Alon Levy May 12, 2010 at 12:03 pm #

    In your experience, does the cul-de-sac-with-pedestrian-links solution work? I’m asking because in many neighborhoods, I can’t imagine it working. Those pedestrian links are like alleys, so late at night, they get sketchy; in high- and even medium-crime neighborhoods, this makes them unsafe. For example, common lore in Upper Manhattan is that the pedestrian step-streets crossing Morningside and St. Nicholas Parks are unsafe late at night. Remove retail from an area and its safety depends on having a high per capita income.

  10. Jarrett at May 12, 2010 at 2:50 pm #

    Alon. There’s no general rule about that, but I’d say that even if the ped links are only used in daytime and rush hours, that’s enough reason to value them.
    Remember that if you’re on the ped link next to the cul-de-sac, you’re right outside the window of two homes, so someone is likely to hear you scream.
    The other issue is the parkway that the ped link leads to in Canberra’s case. Those are problematic in terms of a safe pedestrian environment, becuase the ped links from the cul-de-sacs each lead to a lonely path across the park strip.
    But the same principle would apply if the arterial were right next to the cul-de-sacs, connected for peds but not for bikes, and that, I think, would be as safe as any other arterial-without-facing-retail situation.

  11. Jarrett at May 12, 2010 at 3:59 pm #

    Tessa.  True enough, although if you can't pull up the aerial photo layer of Google Maps on your mobile phone, you will be able to do it soon! 

  12. anonymouse May 12, 2010 at 4:02 pm #

    Note by the way that it’s just as common to see this sort of suburban hell street pattern embedded in a larger grid of major streets spaces at precise, geographically aligned half-mile intervals, which is generally where the bus service then runs. It’s typical for these streets to actually have nothing on them: just a wall, with the houses behind having their backs to the main street.

  13. EngineerScotty May 12, 2010 at 4:43 pm #

    Many of the cul-de-sac hell neighborhoods here in Beaverton have pedestrian/cycle cut-throughs, however in many cases the walkways are rather isolated–often running between the backyards of properties (much like an alley) with no direct access to properties, and tall fences obscuring the views of the walkway from the properties (and vice versa).
    Most such things occur in low-crime neighborhoods; still, not a place many folks would like to be alone at night in.

  14. GeigerBC May 12, 2010 at 5:15 pm #

    @John – Your Ohio link was interesting. Do you happen know how they phased the traffic signal in combination with the light rail at that location?

  15. Daniel May 12, 2010 at 5:57 pm #

    Tessa, yes, agree. Perhaps those semi-hidden pedestrian links/shared bike paths are more easily discoverable if documented in local street directories/Google Maps/etc.
    In my neck of the woods, Google Maps can’t navigate me on foot through a park or along a pedestrian-only route, but at least the dominant paper street directory (the excellent “Melway”) does include such paths (including its online version).

  16. Brent Palmer May 13, 2010 at 3:14 am #

    One helpful thing would be signage. Post an ordinary street sign pointing along the pathway, and you convey the message that it’s a continuation of the existing street. Attach information further down the pole, including a map of the immediate area and directions to the nearest bus stop or train station.

  17. Peter Parker May 13, 2010 at 3:24 am #

    Strongly agree that links like this should be built, but a view that they (or the fused grid) are anything better than a very distant second-best option is mistaken IMHO. Fused grid also reminds me of Radburn design – a disaster in public housing estates.
    Just as public transport should have the same routes night and day for legibility, the same should apply to pedestrian access to it.
    Either an actual closed gate or a perception that an alleyway is unsafe at night reduces the permeability of the pedestrian network, which is equally important to that of transit routes IMHO.
    The local papers here often stories about bashings in alleyways and requests for them to be closed (including from police). Much like a bus lane that can readily be converted to cars, alleys do not have the perceived permanence of railways and through-streets. And the best way to preserve through-streets is to ensure houses need them for their vehicular access, as per the traditional grid.
    Anywhere where there is no vehicle access or passive surveillance from buildings is regarded as unsafe.
    Graffiti is strong evidence of this and contributes to feelings of insecurity. Almost every back fence backing onto a railway line or narrow alley is daubed in graffiti.
    But place a car-carrying road beside the railway, and preferably also the front fences of houses, and the graffiti is reduced by 95%, even in poorer neighbourhoods.
    Hence neighbourhood support for the closure of alleys is likely to be higher than for their creation and a few bashings or bottle throwings is likely to provoke calls for closure.
    There’s a yawning gap between what is ideal for transit planners and the general community. A better outcome might be to identify houses in strategic positions (eg at the end of dead ends), purchase them when for sale and redevelop, possibly by building single storey units partially facing the walkway. This might eliminate the walkway-beside-high wall that only attracts vandals and contributes to a feeling of vulnurability.

  18. John May 13, 2010 at 9:04 am #

    Where there are at-grade crossings on Van Aken Blvd and Shaker Blvd, the signals seem to be actuated by the trains. A train won’t cross an intersection until the end of a signal phase, so occasionally the train waits briefly for cars to cross. There also seems to be a good clearance phase to get any cars in the median waiting to cross opposing traffic off of the tracks before the train gets a signal to proceed.

  19. David Sucher May 13, 2010 at 10:58 am #

    I guess I am a bit lost.
    Of course grids are good.
    But why use Woodinville as an example for the very reason that the freeway acts as a huge barrier with only one way through it — an underpass.
    The limitation would be there even if the grid was ideal. Yes it’s good to find some hard research that an interconnected street grid produces fewer vehicle miles.
    But why use Woodinville with a major barrier as an example?
    What am I missing?

  20. Daniel May 13, 2010 at 11:52 am #

    For many existing subdivisions, a pedestrian link will be more politically palatable than connecting in with a full road. When a developer around here wanted to punch a road into an existing subdivision (to meet minimum connectivity requirements in the local code), they encountered fierce resistance from the suburban neighborhood association. You know, unsavory characters, cut-through traffic, etc.
    When this didn’t work, the developers came back to the table with an 8′ multiuse path option, and the neighbors still hated it but slightly less. They are now proposing a 5′ path, and some of the resistance is starting to slow down. I highly doubt the developer will win this one, but you can imagine a situation where a neighborhood association might give in and allow a pedestrian connection. Baby steps.

  21. Jarrett at May 13, 2010 at 12:31 pm #

    @David Sucher.  I'm sure Prof Frank chose Woodinville precisely because its situation is so extreme, as his aim was to prove a point about how the entirety of street network patterns — including obstacles such as freeways but also just disconnected networks — lead to higher car use with all that that entails.  Note that the walkshed in the Woodinville example is blocked not just by the freeway but also by the disconnection of the network on the west.  It's clearly chosen to be an extreme example.  I suppose he could have picked an intermediate example, maybe with disconnected streets but no freeway, and thus begun to define the shape of a curve between the two endpoints.  Canberra, with curvy streets that discourage straight-line walks but also lots of ped links between cul-de-sacs, might have been other intermediate data point on that curve.  But he still makes a clear point based just on looking at the extremes.  I suppose he's defining an axis or spectrum on which others can plot intermediate examples in more detail.

  22. Stephen Schijns May 14, 2010 at 10:22 am #

    I have long thought that more effective signage would help this situation, particularly for cyclists – when a dead end (for autos) street is signed as “cul de sac” or “dead end” or “no exit”, even though it has a ped / cycle link, I would like to see a tab on the road sign to the effect of “path available” or “link to ___ street” (or perhaps even a simple symbol). Has this been implemented anywhere?

  23. Jack Horner May 15, 2010 at 4:15 am #

    1. re unsafe alleys at the end of cul-de-sacs: possible compromise – have a full street grid, but close selected side streets to traffic at the arterial road end and make it a pockethankerchief park also functioning as the walkway to the arterial. From the point of view of motorists its a culdesac.
    2. Where did the fashion for what I call ‘three grades of spaghetti’ street patterns come from? Seems to date from the 1960s in Australia. Was there ever any debate about the implications? Or was it just a fad that over time weaseled its way to the status of received wisdom without any real thought? Since it is so obviously bad from a public transport/walking/ cycling point of view, why is it still so strong even among planning authorities that pay lip service to environmental sustainability? What is wrong with grids? Do developers believe that most people really want limited access culdesac developments? Is there any sold evidence of what people really do want?

  24. Jarrett at May 15, 2010 at 4:25 am #

    I don’t think the street patterns of Australia are much different from those in the US of the same era. I think that in trying to make suburbs appealing as something unlike the city, the curved streets worked subliminally to suggest relaxation and openness compared to the urban grid.

  25. MNSegler May 15, 2010 at 6:12 am #

    Clear signage is common in many European cities to indicate prohibition of cars or sometimes walking speed restrictions for local access cars while permitting bicycles and pedestrians. I’ve also seen much more attention to pedestrian and bicycle right of way during construction. There are just so many people walking/riding in many areas that people would complain loudly if through access was limited during construction.

  26. Paul C May 16, 2010 at 2:44 pm #

    From what I understand of the cul-de-sac system.
    The biggest reason for its design was to prevent vehicles from cutting through. Some thing that was common in the grid system. Of course now there are places in a grid where they have put up blocks to prevent a vehicle from getting through. But a pedestrian or bicyclist can still get through. So the road doesn’t look like a dead end.
    The other part I never liked about the cul-de-sac system. Is how in a lot of cases they built a lot of roads with curves. Which I feel causes a poor usage of land in terms of density. At some point one property is going to be smaller or bigger than another property because of the curved roads.

  27. Dan November 24, 2010 at 2:24 am #

    Ironically, there IS a pedestrian link at almost the exact location of the red dot in the top left (Woodinville) image. The original study must not have known about it.

  28. Greg Beaumont April 20, 2011 at 7:44 am #

    The real reason for cul-de-sacs is simple – money – more units per acre (less road) – soley developer driven.

  29. Al Dimond May 11, 2011 at 1:41 pm #

    How about this as a reason for choosing this extreme Woodinville location: there actually is a bus stop there, for a route serving the highway (I think it’s Sound Transit’s route 522, along WA-522, which is a freeway at that point).
    I am not all that familiar with the area, but I think it’s safe to say that the stop is there but it doesn’t serve very much.
    Another example in suburban Seattle is the recently-opened Mountlake Terrace freeway bus station. It’s in the median of I-5 near 236th St. SW, but it only has a pedestrian connection to the transit center on the east side of I-5. Getting there from the west, and maybe the north (it’s hard to tell whether there are walking trails through the cul-de-sacs from a map), really have to do extra walking to get there. The station is a really fast stop for the express buses that use it, but I’m not sure if anyone lives within a quarter-mile’s walk of the actual platform!

  30. Al Dimond May 13, 2011 at 9:33 am #

    My previous comment aside, It’s not much worse to get to the freeway median platform than to get to the P&R in general… the way the area is built is a real challenge for transit and walking. Also, adding the freeway median stop allowed consolidation of express bus routes, nearly doubling frequency of the ST 510 and 511 routes, so it’s a significant win for me even though I’ve never used that stop myself.

  31. George Carty May 14, 2011 at 12:51 pm #

    I think the curvy road layouts of many suburban areas were intended to improve sight lines for drivers in the interests of safety.

  32. Max Schneider March 4, 2012 at 4:22 pm #

    @ Stephen Schijns
    Yes there is a street sign (in Germany) that shows cul de sacs which are permeable by bicycle or walking: see