Category: Urban Design

Sunnyside Yards Redevelopment

Sunnyside Yards, lying along the LIRR Main Line immediately adjacent to the site of my proposed Sunnyside Junction, span about half a square mile (1.3 km^2) of mostly vacant land, with some big box retail with ample parking at its eastern margin. The short distance to Manhattan has already made Western Queens increasingly desirable (538’s Nate Silver called Sunnyside the third best neighborhood to live in in New York); the new rail junction would make this vacant land into prime real estate, making it feasible to sell air rights above the yards in a similar manner to how much of East Midtown was developed with air rights over the Grand Central tracks.

I would like to discuss how this should be done. This can be thought of as not just a particular Sunnyside question, but also my general ideas for how to do good transit-oriented development, and even more general ideas for how to develop new sites for dense urbanity.

First, the development would be mixed-use. This is because there’s both commercial and residential demand near Manhattan. More speculatively, this could cause the Long Island City secondary CBD to expand eastward, from Hunters Point and Queens Plaza toward the proposed station. In any case, the station should be expected to have high-intensity retail and office buildings immediately adjacent.

On the other hand, the development should be integrated into the existing neighborhoods on both sides of the yards, in terms of both street layout and development intensity. This is not the place to test out new ideas of urbanism; the streets should look as similar as possible to those of Sunnyside and Long Island City. Here is one way to map out streets: note the block size is similar to that of the surrounding areas. The same should be true of street width.

The best way to combine the two goals – retaining existing neighborhood context and allowing high-intensity commercial development near the station – is for the city to have progressively higher-intensity zoning proceeding from the margins to the station itself. Away from the immediate station area, medium-rise buildings such as those of Upper Manhattan (excluding projects) should suffice, and the city should not try to ram high-rise buildings against neighborhood opposition. This would also be friendly to small developers, turning this into the anti-Atlantic Yards. Needless to say, there should be no parking minimums, since the area would be dense and well-served by mass transit.

The overall density of such development could be compared to the mid-rise neighborhoods of Upper Manhattan, such as Morningside Heights and Washington Heights. Morningside Heights has 40,000 people per km^2, and so does Washington Heights when one makes sure to exclude its ample parkland. Morningside Heights has a lot of open space and many jobs, but it’s also higher-rise than Washington Heights (excluding the projects, again). Either could be taken as a basis of comparison, by which standards the 1.3 km^2 over the yards should support about 50,000 people.

Sunnyside would effectively get a second core, around the station, in addition to the existing core along Queens Boulevard. Although the development could spill over, raise rents, and produce gentrification, by itself it would not change the existing neighborhood much, which is fine as Sunnyside is pleasant as it is. Even the Queens Boulevard semi-highway works remarkably well there: the 7 el does not produce too much noise, and instead breaks the boulevard in half, making it look narrower and producing a good street wall for each of the boulevard’s halves.

Bear in mind that out of everything I have proposed in this blog’s history, I would peg this as the least likely to happen: the development I’m advocating spurns big monolithic development. Instead, the city would just map out streets, enact mild zoning restrictions to prevent the community from rejecting the plan for fear of Manhattanization, and perhaps attract a few anchor tenants and companies to build immediately next to the station. In contrast, the present process of redevelopment in New York is laden with collusion, with big developers getting land for megaprojects for less than it’s worth. The city would give a developer not only the yard land but also neighborhood blocks around it, which would be turned into a modernist urban renewal hell instead of a higher-intensity version of the same neighborhood.

My sliver of hope is that the extra transit service coming from the new junction station, and the fact that at the margins of the land the new development would look hardly different from the existing blocks, would reduce neighborhood opposition. Often the dominance of big developers in cities comes from neighborhood opposition to change, creating an arduous process of obtaining variances and schmoozing with city officials that small business cannot afford. I would peg the chances of neighborhood approval at low to moderate, the chances of such a plan happening in case of neighborhood approval at low, and the chances of such a plan happening in the absence of neighborhood approval at zero. What say you, Sunnyside-area bloggers?

Planned Cities

The back and forth between Steve and me about his proposed pedestrian-oriented city led me to think more about planned cities, as his is. Although it’s normal among urbanists (for example, Jane Jacobs) to contrast organic cities with planned cities, there are really two aspects of planned cities, one economic and one design-based.

A city can be planned in terms of its urban design. A master plan came in and created its street layout, which is in some way regular. For the purposes of this discussion, it does not matter whether it was planned according to the principles of the Enlightenment, the Garden City movement, or modernism; however, this city will typically have grander major boulevards than unplanned cities, because they are meant to be the focus of civic and state power. A city can also be planned in terms of its economy and location: it was built out of scratch, typically to serve as a new national capital. Those two aspects of planned cities are more or less independent, and criticisms of one do not always carry over to the other.

We can present this using the following table of examples:

Economically Planned
 Economically Organic
Organic Design London, Tokyo, Boston, Rome
Planned, Enlightenment principles Washington, Toronto, St. Petersburg New York, Philadelphia, Paris, Chicago
Planned, Garden City Kuala Lumpur Tel Aviv (Old North)
Planned, modernist/postwar Brasilia Singapore, Mumbai (ex-slums), US Sunbelt, Tel Aviv (north of the Yarkon)

There are several things one should note about the above table. First, there are no economically planned cities with organic design, at least none I’m aware of: a city that exists only because of a national master plan will also have a master plan dictating its urban form. The closest things to it are a preexisting organic city that became larger when the national capital was moved there, for example Rome or Bonn, and a planned city that had some spontaneous development before being subsumed by a master plan, for example St. Petersburg. In fact the reason I found Steve’s proposed city a little weird is that it tries to approximate the features of an unplanned city (very narrow streets) in a setting that is completely planned (the city is plopped in a new rural subdivision).

Second, economically unplanned cities usually have downtown districts with no urban planning, regardless of how the rest of the city looks. This is true in New York (Lower Manhattan), Tel Aviv (Jaffa, pre-1920s Hebrew neighborhoods such as Neve Tzedek), and Singapore (the older parts of the CBD, Chinatown, Little India). When a city has no pockets of unplanned areas, for example Paris, it’s often because the entire city was razed and rebuilt according to a master plan; this is equally true of planned parts of cities, which contain few hints of the areas that were razed to make room for them.

Third, many, though not all, problems that Jacobs and other traditional urbanists associate with planned cities are problems of economic planning and not urban planning. A good example is Jacobs’ attacks on the Garden City movement, in particular Ebenezer Howard and Patrick Geddes, for hating traditional cities. Such an attack could be justified by looking at new suburban subdivisions as well as Kuala Lumpur, built from scratch to be the capital of Malaya.

In contrast, Jacobs’ criticism looks misplaced when one looks at Tel Aviv’s Old North, a dense and walkable neighborhood built entirely according to a master plan specified by Geddes. If you reread my criticisms of Tel Aviv, you’ll see that they are more about what more modern planners have done with it than the Old North’s basic urban form. The urban form is walkable, and the narrow streets are inherently less car-friendly than those of New York or Paris, but the traffic engineers have nonetheless optimized traffic for cars, and the national government rejected a subway in the 1960s for anti-urban nationalist reasons.

In other words, although the urban form of Steve’s Triangle City is likely to be similar to that of Rome, Tokyo, Kyoto, and other unplanned cities, its social and economic characteristics are not. They are more likely to be similar to those of Kuala Lumpur, Brasilia, Washington, and master-planned suburbs.

At the same time, some of the problems of modernism remain true in both planned and unplanned cities. Singapore’s projects and cul-de-sacs are as socially isolating as those of American suburbia. The arterials are not walkable, and the government’s response to huge pedestrian volumes at the intersection of Orchard and Paterson was to construct pedestrian underpasses. The same is true of the newer parts of North Tel Aviv, north of the Yarkon, though the high incomes there prevent the gang problems common in American projects.

Fourth and finally, it is rare but possible for a city built from scratch to become a true import-replacing city, depending on circumstance. Toronto did; although it was built from scratch as Upper Canada’s capital, safely away from the US border, it grew and became Ontario’s primate city, and eventually eclipsed Montreal as Canada’s largest city. The same is true for numerous cities built or designated as imperial capitals before industrialization, e.g. Beijing, Berlin. I cannot tell whether Kuala Lumpur has achieved the same feat, but its size suggests it has. Washington has not – its economy depends almost entirely on government. Master-planned suburbs practically never do.

Sprawl is Auto-Oriented

Steve Stofka has a post detailing his ideal new city, built on principles of high density through very narrow streets, and an interconnected, pedestrian-friendly grid. Its population is given as 30,000, and its area as about 2 square miles, or 5 square kilometers. Although it’s meant to illustrate urban design principles, in a similar matter to my description of my ideal urban arterial street, the proposed location in exurban South Jersey raises questions about why new subdivisions remain auto-oriented.

Let us zoom out and see the general location of this Triangle City. It is exurban; there’s no access except by narrow local roads. This means that residents would drive to all locations outside the town, forcing high car ownership and raising car use well beyond the ability of narrow, pedestrian-oriented streets to carry them. Such towns can be built along potential commuter rail lines, but most of the time the stations along these lines are already developed at vastly lower density than 6,000/km^2. This alone means that such new development would not look like the narrow-street neighborhoods of Tokyo, which have rapid transit to get people to the rest of the metropolis.

Even if a rail line were built, it would not do much to make such development less auto-oriented. A town of 30,000 is very far from self-sufficient; its residents will do much of their shopping outside, making car ownership even more attractive. Morningside Heights, a neighborhood of 33,000, is not self-sufficient either, but because it’s part of a contiguous urban fabric with rapid transit links to the rest of the city, it has low car ownership.

For a more reasonable take on why sprawl absolutely requires cars and long drives, here’s commenter JJJ writing on his blog about a proposed Fresno sprawlburb of 10,000 people:

A self-sufficient community? So much laughter. 2,800 homes may be able to support a supermarket, a CVS, a few nail salons and chinese take-out….but people want more. People want to shop at Target, Best Buy, Costco, etc, none of which will EVER locate in a community of that size (they require much larger customer bases). People want a Cinema and bowling alley. People want a large selection of restaurants.

So residents will have to drive for 30 minutes, through the only road, which is used by those heading to the lake for recreation, or passing through to the national parks. That means congestion.

And the future supermarket, library, pharmacy, elementary school etc? The people who will end up working there won’t move into an expensive rural suburb. They’re going to live in Fresno. And every single one of them will have to drive in that one road. The idea that the retirees will be so self-sufficient that they will be busing tables at a small chinese restaurant is hilarious.

In other words, in a society where most people are rich enough to own a car, building sprawling subdivisions makes car use inevitable. Transit-oriented development done right should start in the dense cities and expand outward slowly, making it feasible to do all of one’s errands on foot, by bike, or by public transit.

Meticulously planned subdivisions usually flop in terms of urban quality, but if one wants to build any, they should be surrounded by or adjacent to existing high-intensity urbanization. This is the reason for gentrification: such areas are almost never empty, and as a result the only way to create them is to displace existing neighborhoods, either organically (Williamsburg) or by fiat (Manhattanville).

Why Transit Should be in the Fast Lane

Local buses tend to use the slow lane, which in North America means the rightmost lane; this is how they access the curb to pick up passengers. New York’s painted bus lanes on First and Second Avenues are to the right, with the buses slower than the cars both in perception and in actual practice.

Occasionally, transit uses the fast lane, especially if it’s BRT or a streetcar; for some of the access challenges of boarding not from the curb, see an old Human Transit thread on the subject. The issue of whether there should be sidewalk- or median-adjacent transit lanes came up in comments on Cap’n Transit’s blog. So let me explain why higher-grade transit than local buses, which means rail or BRT, should run in the median, with boarding from raised curbs either on the sides or in the center.

1. Service identity. This is probably the overarching concern, especially on the question of whether to have raised curbs or instead stop traffic in the slow lanes and have people cross to the bus from the sidewalk. ITDP’s magnum opus standards for full-fat BRT virtually take median running for granted, and only consider alternatives when the right-of-way is constrained. This is also mentioned as the highest grade of BRT in a conference paper examining BRT on city streets.

2. Fewer conflicts. Using pedestrian-friendly two-phase stoplights, it is impossible to eliminate turn conflicts, though in Delhi they found that median running (right lane in India) had fewer turn conflicts. In addition, it’s possible to eliminate conflicts with cars entering or exiting the parking lanes, as well as stopped cars left near the curbside lanes.

3. Median lanes are politically easier to physically separate, since separating them does not deprive cars of curbside access. If cars can physically violate transit lanes, they will, either accidentally or intentionally (my mother’s car’s GPS guidance routinely sends her along the tram-only lanes). As APTA mentions in its own standards for BRT,

One major advantage of a median busway is that there is typically no demand for other vehicles to stop in the center of the street for purposes such as parking or as a breakdown lane. As a result, there is a lot less reason for vehicles to want to occupy the center of the road and less resistance to creating a physical barrier separation between the busway and the adjacent general traffic lanes.

Point #3 is what killed the proposal for the 34th Street Transitway, which would have run two-way on one side of the street with one direction running contraflow. The NIMBYs on East 34th Street complained specifically about curbside access, using such language as “Delivery and service trucks… no longer have direct access to buildings and stores along stretches of 34th Street.” Most issues they raised involved curbside access or else bus noise adjacent to the street, both of which would have been solved by median lanes.

To add to what Steve Stofka is writing about grids, if I had to design a street from scratch, it would look a lot like a two-way version of a Manhattan avenue, with bus lanes in the middle. It would be 30 meters building to building, and about 20 curb to curb; this is enough space for two parking/loading lanes (2.5 meters) buffering pedestrians from moving traffic, two car travel lanes (3 meters), and two median bus lanes (3-3.5 meters), with room left for physical separation (measured in centimeters). Raised curbs for stations should add 3-4 meters, at the expense of either parking or sidewalk space once every few hundred meters; one advantage of trams, or buses with doors on both sides, is that they can use less space-consuming island stations.

Quick Note: Safe Streets, Safe Cities

Everyone should go read Jan Gehl’s post on Streetsblog about good urban design, excerpted from his book Cities for People. I have nothing to add, except to underline one part that’s often underrated among urbanists: the role of parked cars as buffer between moving cars and pedestrians or cyclists. Compare this photo with this photo, and ask yourself where the cyclist is better protected.

I generally tend to be very supportive of Manhattan’s design. The streets may be wider than elsewhere, but that translated mostly to increased pedestrian space. Manhattan’s 18-meter side streets have 1-2 driving lanes and a parking lane on each side; so do the 12-meter side streets in Tel Aviv, the difference being that in Tel Aviv cars park with two wheels on the sidewalk. As long as there’s an adequate street wall and the buildings are not set back from the street, it isn’t a real problem. As Gehl notes, there are many ways to make cities livable short of the ideal of Venice, in which cars begin where the city ends.

Little Things That Matter: Stoplight Phasing

In Manhattan, most intersections have two stoplight phases: one permitting all north-south traffic, and one permitting all east-west traffic. Each phase lasts about 45 seconds, ensuring that pedestrians can cross even the widest avenues in one go with time to spare.

In Tel Aviv, the signalized intersections are almost never as in Manhattan. Even intersections of major streets with side streets will usually have three phases, and intersections of two major streets will usually have four, permitting conflict-free turns; turn conflicts with pedestrians exist on such intersections, but are uncommon.

From the traffic engineer’s perspective, Tel Aviv intersections are better – they’re supposed to be safer and smoother for the driver, with none of the snarl that happens when a car driving on Upper Broadway tries to turn left. They’re also hell for anyone not in a car, since waits are much longer, and to compensate for the larger number of phases each phase is shorter. This discourages enough pedestrians as to reduce the number of pedestrians for cars to hit, creating an illusion of even more safety.

If there’s enough car traffic, then streets with complex stoplight phasing are uncrossable 75% of the time. But if the street is median-divided, this is even worse, because the traffic engineers try to optimize car traffic, which means the pedestrian green on the two halves of the street is unsynchronized. At some intersections, one direction of a crossing is pessimized for the pedestrian: that is, after crossing one half of the street, the pedestrian will have to wait nearly a full cycle to cross the other half.

There are emerging calls for complete streets, which include such important features of walkability as wide sidewalks and frequent crosswalks. But the frequency of the crosswalks is only partial consolation if the stoplights are optimized for high car speed rather than high walking speed. For a healthy, young individual, the difference between not having to wait at intersections and having to wait a minute and a half every 300 meters is the difference between walking at 6 km/h and walking at 4 km/h. Traffic engineers do not tolerate phasing that slows down cars by 33%, and should not tolerate phasing that so slows down pedestrians.

The above does not apply as much to low-traffic areas such as Downtown Athens, Georgia, because there are sufficiently few cars that locals ignore pedestrian stoplights anyway. But in a large city with many cars such as Tel Aviv, it’s difficult to cross safely on red. As a result, streets that are no wider than a Manhattan avenue can take multiple minutes to cross, and one such street, Ibn Gabirol, divides the neighborhood it passes through in a way that 42nd and Broadway never do.

Ibn Gabirol and similar streets are not suburban arterials. They’re not like Athens’ Baxter Street. They have a street wall, very wide sidewalks, and mixed uses, just like Manhattan avenues. They also have enough foot and car traffic that they don’t feel desolate. They feel very walkable, as long as you stay on one side; it’s when you try to cross that their auto-oriented nature becomes apparent.

More on Density

Commenter Benjamin Hemric replies to my previous post on Manhattan density, arguing that,

1) It’s very interesting that people seem to have very, very different goals and objectives here — and it’s good that these are expressed and out in the open. Otherwise people wind up talking past one another.

The Jane Jacobs goal (which is where I feel I’m coming from) believes there should be HEALTHY (self-generating, market-based) cities (for a variety of reasons, including economic and social ones) — and more and more of them to meet a growing world population. (I think Glaeser also shares this part of the Jacobs viewpoint.)

The saving of the environment (to extend what I see as the Jacobs’ viewpoint) is a by-product. And it isn’t from piling more and more people into a relatively small part of a city but from having many successful high density districts — and by having many dense cities, which by their very nature are greener than suburban sprawl.

2) It seems to me that it’s important to remember that the outer boroughs (to use the NYC example) and nearby northern New Jersey are not pristine undeveloped land, but already pretty well built up albeit oftentimes at relatively low density areas — so making them denser also helps save the environment.

The implied question is as follows: the New York region wants to add more people; where should we plan on housing them?

My environmentalist answer is that they should be housed in dense areas – it doesn’t really matter which ones, as long as they’re walkable and transit-accessible. Even marginally dense areas are okay as long as it’s part of a concerted effort at TOD – say, residential and commercial upzoning in Eastern Queens along the LIRR in conjunction with offering rapid transit-like service levels on commuter rail.

The Jacobsian answer that Benjamin is giving is that they should be housed not in the densest neighborhoods, but in somewhat less dense neighborhoods, on the theory that they’d become greener due to the additional flux of residents. This is, in principle, a good idea. The residents would cause more environmental impact than in Manhattan and live in what are now less walkable neighborhoods, but would induce such development that the impact of existing residents would drop. But it’s not clear which effect dominates, and since both options are much better than any alternative, both should be legal and encouraged; there’s no need for landmarking to force people out of the Village and into Brooklyn.

But that’s all in principle. In practice, densifying outer-urban neighborhoods is a political nightmare. Christof Spieler once wrote about how in Austin, development is governed by a coalition of NIMBYs and suburban developers and boosters. The result: it’s hard to increase density in existing urban neighborhoods, and easy to develop greenfield exurbs. In New York, a similar thing is happening, in reverse – it’s easy to develop in Manhattan’s non-landmarked areas, and hard in the outer-urban neighborhoods and the suburbs. (New York’s exurbs are also growing very quickly, but are too remote and lightly populated to matter.)

Adding density to parts of Queens and Brooklyn where it would introduce a tipping point in favor of walkable urbanism may well be harder than repealing landmark restrictions in the Village. Community boards are always drawn from the wealthier and more connected segments of society, and in those areas they invariably own a car. A new development in Flushing was saddled with extra parking, and NIMBYs all over outer-urban New York oppose dedicated bus lanes due to loss of car lanes.

The conclusion is that the alternatives to density increases in Manhattan are more parking garages all over the Outer Boroughs, and greenfield suburban development in the few parts of the suburbs in which there’s space (typically nowhere near rail). I’m all for walkable densification along outer ends of subway lines, or if commuter rail modernizes then also near train stations. Wake me up when that happens. Until then, the best approach is supporting political reforms to make both Manhattan densification and outer borough densification easier.

Suburban TOD

Hicksville is located 43 kilometers east of Penn Station on the LIRR Main Line. It’s a major job center of eastern Nassau County, with 25,000 jobs and a rather large shopping center adjacent to the train station. The station itself gets about 8,000 weekday boardings, more than any other suburban LIRR station except Ronkonkoma.

However, the station has no TOD. The shopping mall is transit-adjacent, but the route to it from the station passes through parking lots at both ends. The station is surrounded by thousands of parking spaces; a recent reconstruction of a parking garage cost $364 $36.4 million for 1,400 spaces, which at $26,000 per space is more than the per-rider cost of such expensive transit lines as Second Avenue Subway. As a result, the total number of people getting off at Hicksville in the AM peak is 700, for a rail share of 3%.

Such failure is quite common in the US. Leaving aside stations explicitly configured as park-and-rides, such as Metropark or Ronkonkoma, off-CBD stations have to have at least some retail and office space usable by reverse commuters, on the pure financial grounds that reverse-peak service is nearly free to provide. Otherwise, light rail and subway trains run empty in the reverse-peak, and commuter trains park downtown, leading to outsized costs for CBD railyard expansions.

For a comparison of how good TOD looks like, see this industry presentation about Tokyo’s Tsukuba Express. As is normal in Tokyo, the line is very expensive: $140 million per km for a line that’s just 26% in tunnel, though the tunnel percentage is much higher within the central city. But per rider this is not too bad, because as the images in the presentation demonstrate, intense TOD followed construction. Stations are surrounded by high-density office and residential buildings and not parking lots.

A theme I am going to revisit is that high construction costs should not be an excuse to scale down service levels. It may be expensive to develop on the parking lots adjacent to the station, but the ridership is always worth it.

If there’s to be a transit revival, it’s imperative to increase mode share at major suburban centers. The transit mode share for people working in Manhattan is 75%, while the auto share is only 14% – and the auto mode share is dominated by the suburbs that use the GWB, rather than Long Island. There’s some room to expand Manhattan employment, but not enough to make a dent in the region’s car use. It’s critical to instead make it easier to use transit and harder to drive to work in such secondary downtowns as Flushing and Jamaica, and in such major suburban centers as Mineola and, yes, Hicksville.