They all started out as suburbs. Grab a map going back far enough and you’ll see your favorite dense, urban, walkable mixed-use district as a speculative plat on the edge of the then-developed area.

In Houston, this includes neighborhoods like the Medical Center, the Fourth Ward, the Montrose, and the Heights. Midtown Manhattan was once a neighborhood of single family homes. The high-rises and rapidly gentrifying neighborhoods of Chicago’s North Side were once that city’s outer periphery. Almost every block in Los Angeles was, at one point or another, within a stone’s throw from an operating farm.

This wasn’t a bad way to build cities. Problem is, at some point we decided that the natural tendency of desirable neighborhoods to increase in density and intensity was a bad thing, and we adopted restrictive zoning regulations and deed restrictions to ensure that things didn’t change. This killed off the ability for future urban districts to form; almost all of our favorite urban places were platted before World War II.

The restrictions which insulated postwar neighborhoods from the gradual process of building and rebuilding also accelerated sprawl, since an absence of increased density in established neighborhoods resulted in more demand for new neighborhoods at the periphery. This in turn has as its very predictable side effects long commutes, auto-dependent lifestyles, and gigantic queues of cars at the close of the K-12 schoolday.

Clearly, the only alternative left is to toss the yard and the car and embrace a bright new future of townhomes and light rail feeder buses. But wait just one minute… what if we built suburbs like we did before the war?

What if new developments retained the big yards and abundant parking that so many Americans have grown to love, while older developments gradually filled in with denser housing, walkable retail, and more intensive amenities? What if we built single-family residential neighborhoods today that could accept skyscrapers 100 years from now? Is such a thing even possible? Would anyone be interested if it was?

Enter the New Suburbanism

Horticulture instead of Architecture – New Urbanism communities are master-planned like any other plan from a single architect; a perfect whole, to be filled in piece by piece. Once built to plan, New Urbanist communities do little more than age gracefully. New Suburbanism recognizes that livable cities are grown over time, not designed on a sheet of paper or a CAD screen. Rome wasn’t built in a day, nor were any of our other favorite urban places.

History, not Philosophy – New Urbanism is descended from some of the loftiest planners and thinkers of the 19th and early 20th century, dreams of ideal cities on a more human scale before modernism took hold. New Suburbanism is based off the history of how cities actually grew; piecewise, through subdivisions and “additions,” with the original housing stock and other structures slowly replaced by denser, more urban uses.

Suggestive, not Prescriptive – New Urbanism creates a unifying vision and pre-ordained “sense of place” for a community, then adopts strict rules and guidelines to ensure that all subsequent development meshes with and contributes to this vision. New Suburbanism intentionally leaves room for the differing and conflicting views and contributions of builders, designers, planners and developers now and in the future.

Houston operates one of the most complex local bus networks I’ve ever been exposed to. This isn’t intrinsically negative nor positive; it just is. Houston’s local bus complexity is comparable to Washington, DC – more comprehensible than New York, but more complex then most of its peers, Chicago and Los Angeles included.

So what’s good? What’s bad?

Good: Semi-express routes
Semi-express routes are going out of style around the country as cities extend rail and shift formerly radial routes into feeders. But in a city as sprawling and expansive as Houston, they’re essential. These routes include the 56-Airline, 44-Acres Homes Limited, the 20-Long Point Limited, the 131-Memorial, the 163-Fondren, the 88-Hobby/San Jacinto, and the 137-Northshore. All run local outside 610, turning express inside.

Good: Intuitive Crosstown routes
Houston’s crosstowns run the gamut from high-frequency urban routes with heavy ridership to outer suburban lines with a few riders on an hourly bus. One common feature among them is intuitiveness. Western routes like the 67, 19, 46, and 32 mostly stick to a single crosstown arterial. Other routes may alternate arterials to serve the most densely-packed neighborhoods, like the 42-Holman or the 23-Crosstimbers. The 26/27 orbital is fairly easily comprehended by most, while even the more complex routes serve an obvious function. The 73-Bellfort makes a seemingly endless series of jogs and turns, but in the process manages to link the Galleria, Greenway Plaza, the Medical Center, and Hobby Airport into a single continuous route. That’s impressive.

Bad: Loops at the end of long routes
Loops by themselves aren’t a bad thing. Where a bus route has no clearly-defined endpoint, like a transit center or a shopping mall, loops can increase the service coverage of a route while inconveniencing a minimum of riders. The problem occurs when loops exist at the end of long routes. As route length increases, the required layover time also increases, negating the convenience of the loop and ridership-building capability of the loop.

A decent contrast can be found on the West side. The 72-Westview is a relatively short route that links the Memorial City area and a few neighborhoods to Northwest Transit Center. It’s timed so that the loop can be traversed continuously, with all layover time absorbed in a dedicated bus bay at NWTC. A few miles north are the twin loops of the 50-Heights’s Rosslyn and Hollister branches. The 50 is a lengthy route stretching across town, with one-way travel times that can approach two hours. Layovers for this route can approach the 20-minute mark, which means that if you live on the “wrong” side of the loop, it’s almost always faster to walk to the other side than to ride through the layover point. Directions provided by Metro’s own trip planner bear this out.

Bad: Excessive Tails
Tails made sense in the 20′s and 30′s when many traveled by transit. They allowed operators to split service in the lowest-traveled (and often lowest-density) areas of routes to serve more riders. The problem with tails is that many new transit riders have little experience with urban bus networks, and in an era of GPS and online trip-planners, people are less locationally-aware than ever.

Tails are the biggest operational source of negative first-time rider experiences. Teenagers looking to explore, adults with a car in the shop or a desire to “try out the bus” dutifully plan trips, get on their route, and are then horrified as the bus turns down a different street than it was supposed to. Often, they won’t come back. No other source of first-time rider frustration is so easily preventable; other frustrations of this magnitude are cultural (i.e., “a homeless guy hit on me”) and will probably always be present on subsidized urban mass transit.

The best litmus test for tails is simple: do they make sense as separate routes? When Metro split the 81-Sharpstown from the 82-Westheimer, it ended years of suffering for new riders who got taken for a ride. And they’re both good routes in their own right; in Metro’s most recent round of service cuts, these routes are some of the few that saw increased frequency, primarily to relieve overcrowding. Other long-tailed routes – like the 25-Richmond – would similarly work with two separate numerals.

On the flipside are tails like the 33 and 14′s scenic tours of the Hiram Clarke area, the 36 and 50′s alternating service to the Northwest, and the 5′s indecision about where it’s destined for Griggs or Bellfort. These needlessly confuse riders, and could easily be replaced with equivalent service of much greater simplicity.

- – -

Up next: Tangled routes, Straight lines and Circulators

TRANSIT

Extend existing LRT to Lynnwood and Tacoma
Uncontroversial. SoundTransit’s long-term planning has always anticipated this. However…

Admit you just built a heavy-rail system
Since the dawn of electric railways, Americans have associated heavy “urban rail” systems with full grade-separation and third-rail power. This view was reinforced in the 60′s and 70′s when systems like BART, MARTA, and the DC Metro were built as effectively commuter subways. But this is a uniquely American view. In Japan, most heavy rail systems built since the 60′s have been constructed with 1500VDC overhead power, and many have been built to the same loading gauge as the national rail network. This facilitates promiscuous inter-operation and trackage-sharing among multiple private and public entities.

For instance, the Kintetsu Railway has a seamless connection between its electrified “commuter” lines and the Karasuma Line of the Kyoto Municipal Subway. This facilitates a single-seat ride from the temples of Nara to the temples of Mount Hiei; same train, different sects. Haven’t heard of Kintetsu before? Kintetsu is the parent company of Kinki Sharyo, which makes the vehicles used on Seattle’s LINK “light rail.”

In truth, the terms “light” and “heavy” rail refer not to weight or size but to capacity. And in this respect, Seattle’s system is clearly heavy rail. The design length for Seattle LINK, at four cars, is longer than any other US “light rail” system. Most US light rail systems operate trains of around 160′-200′. Los Angeles’s Blue Line extends that to around 260′. But Seattle’s LINK is designed around an ultimate train length of 400 feet, with four articulated vehicles mimicing an 8-car subway or commuter train.

Except for the Rainier Valley segment, most of LINK is built like a heavy-rail system. The entire Sea-Tac – Henderson segment is one continuous viaduct, with no level crossings. The entire Northgate extension is forecasted to be underground. Any future expansions to this system should preserve this preference for grade-separation. Some high-speed surface running might be appropriate between Burien and Midway, but major intersections (SR 516, 272nd) should be elevated, and the whole of Federal Way should be elevated, from north of 312th to south of 348th.

On the northside, cut and fill in the I-5 right-of-way should be the norm, although it might be more cost-effective to tunnel under the cemeteries and use the old interurban right-of-way north of 125th. The alignment should also take into account the willingness of Lynnwood/Edmonds/Mountlake/Shoreline to upzone some of the copious amounts of large-lot single-family that line both I-5 and the interurban ROW. Perhaps threaten to withhold stations, constructing a continuous express line from 145th to Alderwood if no rezones are forthcoming.

Finally, install platform gates. Distance-based fares and proof of payment have never quite gotten along, but Seattle’s system will – once these extensions are built – be the most complex fare system ever implemented on a POP basis. Adding faregates will also “decriminalize” a split-second decision to stay on the train, since one can just add exitfare. This is simple for most stations, but poses a bit of a problem for the MLK corridor, the only part of the system that actually resembles “light rail.” The ultimate solution might involve enclosed stations with platform doors, or simply fencing the platforms and installing cameras to catch people attempting to sneak in on the tracks; this is what the Japanese railroads do. A uniquely American take on this system might involve a video feed that could be watched live, via 3G, by a Seattle Police officer hidden away on a side street, ready to swoop in on a moment’s notice to ticket fare evaders.

Swap the Freight Trains
Right now there’s two railroad lines between Tacoma and Seattle. There’s the BNSF, which carries most of the freight traffic, Sounder, and Amtrak. And there’s the UP, which carries a few freight trains a day. Only one of these corridors has room for expansion, and it’s not the one Sounder serves. So what to do?

The BNSF line has the advantage of running through the center of all the small towns along the route, where any urban redevelopment would take place. The UP line skirts those towns. That makes the BNSF ideal for continued passenger service, which means the freights should move to the UP. It’s not too far-fetched; UP operates over BNSF via trackage rights north of Tukwila and south of Tacoma. Only between the two cities are there two lines, a historical quirk that has mostly to do with a Milwaukee Road interchange point that was abandoned 30 years ago.

This entails a few things. First, buy the UP line outright, exclusive of the Fife yards, and allot a certain number of trains/day “as-of-right” based on the capacity of the existing single track line. Second, double track the entire UP between Tacoma and Tukwila, connecting it at either end to the existing multi-track BNSF line. This includes a new bridge across the Puyallup River. Third, connect the UP with the BNSF at Auburn. There is a railroad “wye” at Auburn literally underneath SR-18 that controls all traffic over Stampede Pass. This will need a connection to the UP, which can also be used to access BNSF’s substantial Auburn yard facility. Fourth, trade space on the BNSF for space on the new, double-tracked UP. All that additional capacity can be used to siphon freights off the BNSF main line, clearing the way for more passenger rail. With freight trains removed, Sounder could operate at a 10-minute headway during rush hours – or a 20-minute headway with substantially-increased Amtrak Cascades service.

Co-opt the Dinner Train
Create a continuous Tacoma-Bellevue-Everett commuter line to rival the Tacoma-Seattle-Everett one, using the same tracks as the Spirit of Washington Dinner Train. It’s not particularly difficult. You just need to be willing to drop the cash. What kind of cash? Well, you’d need a grade separation at Monster Road to bring Bellevue-bound trains over the thick morass of Amtrak, Sounder, BNSF and UP freights. You’d need a station within walking distance of Downtown Bellevue, which means shifting the existing line west to a new location on fill or an elevated structure. Note that this is A LOT easier as long as the land between the tracks and 405 is occupied by auto dealerships. Once enterprising developers snatch up those parcels and turn them into mid-rise apartments, it’ll become nearly-impossible. And you’d probably need to add a track between Snohomish and Everett Station, considering that the Stevens Pass route is currently *the* corridor for interstate intermodal freight destined to/from the Ports of Seattle and Tacoma. You don’t want to mess that up with commuter trains.

Use 520, not I-90
Lake Washington has two bridges. One is in dire need of replacement; one isn’t. It’s intuitive that any fixed-guideway transit connection across the lake should be built as part of a new 520 bridge, rather than attempting to retrofit the I-90 bridge (which is operating perfectly well without light rail, thank you very much). But there’s other reasons, too; it’s doubtful that Issaquah or North Bend will ever grow to a size warranting LRT, so preserving the I-90 HOV is a good long-term investment towards increased express buses to those areas. And while an I-90 crossing creates a circuitous route from Seattle to Redmond, a 520 crossing could have a high-speed junction with separate branches to Redmond and Bellevue, creating a direct route to both cities. (Make it a high-speed “wye” and you could have direct Bellevue-Redmond service, too.)

Monorail should also be seriously considered as the technology of choice for this run. The former Seattle Monorail Project got a bad rap for a variety of things relating to engineering (Bombardier’s exit from the bidding process, cost overruns) and political (reduced MVET revenues) challenges. But this doesn’t reflect negatively on the technology. The abrupt variations in grade and swift crosswinds of the floating bridges are easily dispatched by rubber-tired monorails, which cannot be shaken off their central beam. LRT is a little more finicky. And if Downtown Seattle merchants and condo-owners won’t sign off on elevated stations over 2nd/5th avenues, the system can simply be terminated at Westlake. After ten years of forced transfers to the Downtown tunnel, the political will can probably be found to extend the monorail through downtown and, who knows, maybe even out to West Seattle.

HIGHWAYS

Replace some interchanges
There are entirely too many cloverleafs in the Puget Sound region. From south to north: replace SR512/I-5 with a half stack (loop ramps can remain for traffic to/from South Tacoma Way). Replace SR-18/167 with a full stack. Add flyover ramps from SR-18 WB to I-5 SB and SR-18 EB to I-5 NB. Note that this might require a braided ramp configuration with the westbound onramp from Weyerhauser Way.

405 @ 167 and 405 @ I-5/Southcenter both require full stacks. 518 @ 509 could use a flyover, but I’m not sure if it should connect SB 509 to EB 518 or WB 518 to SB 509. I don’t think anyone can predict this until the 509 extension to I-5 is complete, since that will significantly impact traffic in both of these directions.

520 @ 405 should be converted to full stack by raising the 405 mainlines one to one-and-a-half levels above their current grade and sticking all the stack ramps underneath. This is easier than it sounds, since 405 is in a valley there, with vertical sags on either side of the existing overpass over the 520 mainlines. 405 @ I-5/Alderwood is probably fine the way it is.

Get rid of the traffic lights
Seattle’s suburbs have far too many expressway-grade or near-expressway-grade state highways that have their capacity reduced by 1/3rd to 1/2 by traffic lights that serve everything from local streets to Costco entrances.

From north to south: Upgrade US 9 in a piecemeal fashion to siphon traffic off I-5. The entire roadway should be four lanes divided, with through-traffic overpasses at the existing signalized (or roundaboutized) intersections of SR-528/60th, Lundeen Park Way, SR-204, Chapel Hill Road, 20th, 32nd, Marsh/Airport, SR-96/Broadway, Cathcart, 180th, and SR-524/212th. Many of these would probably work as two-lane overpasses, reducing cost. SR-92 should end at SR-9 in a “trumpet” interchange, and SR-9 at US-2 should be a partial cloverleaf. SR-522 @ 212th/Paradise Lake would work as a SPUI, but some sort of custom job with several roundabouts might work better. SR-522 @ SR-9 is hemmed in on all sides by railroad tracks and existing development. An unorthodox solution would be to remove the ramp from eastbound SR-522 entirely, replacing it with a crossover 1/2 mile to the south where SR-9 and SR-522 are parallel.

To the south, SR-516 needs to provide a nonstop connection from 167 to I-5. The simplest way to do this is to disconnect the western segment of SR-516 from the eastern segment, treating them as two overlapping interchanges. 516 @ Meeker/Reith is an obvious SPUI, while 516 @ 167 might consist of a couple flyovers on top of the existing diamond. A more ambitious proposal would be to bypass old Kent entirely by extending SR-516 diagonally southeast to meet with the 277th expressway just east of E. Valley Highway. This would take out a substantial chunk of productive farmland, however.

In Pierce County, the 167 extension needs to be built, as does the Cross Base Highway (SR 704). Provision should be made for construction of a feeder system along SR 704 between Spanaway Loop Road and Mountain Highway, as well as at Canyon Road and SR 161. In a pattern Texans are familiar with, the overpasses need not be built now; simply construct the feeders, then add the grade-separations when traffic volumes grow to warrant them (they always do).

Wait Five Years
Adding flyovers and overpasses to reduce or eliminate traffic lights on major non-freeway state highways would have a drastic and profound effect on traffic patterns and volumes. With enough improvements to SR-9, for instance, traffic congestion on I-5 through Everett might vanish. Alternately, SR-9 improvements – especially when combined with commuter rail from Snohomish to Bellevue – might help push the Seattle sprawl into Skagit County, ultimately exacerbating I-5 congestion to where substantial capacity improvements (or a Sounder extension to Mount Vernon) are warranted.

The same thing is true in the south end. There’s no telling how SR 704 might influence commute patterns five or ten years after it opens, especially given Pierce County’s historically unpredictable approach to zoning policy. SR 516 improvements coupled with the SR 509 extension would combine to radically alter traffic patterns on I-5 in that area. Congestion relief provided by opening up the 405/I-5 and 405/167 interchanges might also increase the traffic load on I-5 south of Tukwila, or it might reduce it by making 167 a more attractive route for Seattle commuters.

There’s simply no way to tell for certain, and despite what high-priced consultants might try to tell you, there’s no computer model that can account for and accurately predict all of the variables in such a complex system. But most of Seattle is not lacking in capacity (520 excepted) – it simply makes inefficient use of it, by signalizing expressways and by pushing 2010′s traffic flow through interchanges that date to the 50′s and 60′s. Upgrading intersections should be the first priority.

People create their own pathways wherever there’s time or effort to be saved. In a city built according to a rigid plan, these pathways disappear as soon as some other use for the land is found. But in a city that evolves without a plan, it’s just as often that these pathways become enthroned as streets and thoroughfares. It’s why Broadway doesn’t mesh with the rest of the Manhattan grid.

Problem is, in the modern age there’s simply no way to escape rigid street plans. Organic evolution of pathways worked just fine when cities had open sewers and cesspools, when water was pumped from wells and modern conveniences like electricity and internet were unknown. But with the advent of fixed utilities there is a need for streets to be planned in advance of construction. It therefore behooves us to understand the principles which govern organic street evolution.

What are these?

1.) Polycentric Radial Topology. The shortest distance between two points is a straight line. In a rural area, this will take the form of a number of radial patterns leading to town centers – as in East Texas. In a dense urban area, radial streets leading to different centers of activity will cross, creating a complex pattern of streets of varying widths and angles – as in Madrid.

2.) Respect for Terrain. Secondary only to the desire to take the shortest route is the desire to take the easiest route. No man will walk across a mountain day after day if he can simply walk a couple extra miles around it.

For most of human history, people had no choice in the matter. The tools for large-scale earthmoving simply didn’t exist. Figuring out ways to literally move mountains enabled us to build freeways and high-speed rail lines. But the techniques which are needed to build an Interstate highway or a Shinkansen extension don’t necessarily need to be applied to *all* streets. It’s completely reasonable to balance the needs of modern traffic movement without pressing the delete button on entire topographic features.

It’s near-impossible to pull an example of this from the Houston area, so we’ll head to the Northwest – Sumner, Washington, to be exact. Sumner sits in deep river valley, scoured out of the surrounding terrain by centuries of volcanic activity. Any road heading east must immediately overcome a 600′ elevation gain. A couple of these roads are shown here: one hillside, two 45mph arterials.

Pretty Good

Not So Great

At the top is Sumner-Tapps Highway. You can deduce the limits of cutting and filling by where the deciduous trees are; evergreens predate the road, deciduous have grown atop the engineered slope in the intervening 30 or 40 years. At the bottom is Lake Tapps Parkway. It’s a newer road, so there’s still a lot of bare earth. The difference in the extent of cutting and filling done is quite striking.

In Houston, where the land is flat, developers lay out gently curving roads for purely aesthetic reasons. So it’s a little ironic that in the Northwest, where winding roads are suggested by the topography, new arterials are simply cut from the hillside. But this is a fairly new phenomenon, and towns like Sumner have an ample supply of pleasantly winding roads. In laying out an organic street network, it behooves one to create more of the latter and less of the former.

3.) Path Sharing. Path sharing is what separates a truly organic street form (like Madrid’s) from a linear, polycentric radial street pattern like Washington DC’s. What is path sharing?

Roads and other ways don’t rarely appear all at once. Rather, street networks evolve over time; they’re added onto, piece by piece. As different paths approach a destination, they’ll often merge. Perhaps the existing path is better-improved, or has more services. Or perhaps there’s a difficult topographic feature (like a river) which several routes may share.

Georgia, like Texas, is a state not covered by the Public Land Survey System, so rural highway networks follow the same organic highway pattern. In Dublin, highways converged in all directions to share a ferry across the Oconee River. There’s no ferry anymore (a four-lane highway bridge was constructed in 1952), but the radial highway networks on either side of the river still mesh together to share a single path through town.

Even in towns without rivers or mountains, highways will often share a single route through town. Remember the east Texas highways from the other post? A closer look at one of those towns reveals multiple instances of path sharing.

This is Rusk, Texas. To the northeast, Bagley Road merges with Texas 110, which itself then merges with US 84. To the south, FM 23 branches off FM 343 at a “t” intersection, then both roads split again. This entire division from one farm road into four happens in a span of less than one mile.

Putting it all together

A master-planned polycentric radial street network which respects terrain, and which shares paths where appropriate, can yield most of the same efficiencies for both pedestrian and auto travel as those that evolved gradually. The key is to design it with growth in mind. Most of the jams that occur in such networks happen because the overall layout didn’t support growth. For instance, if you design a network with path sharing, you have to plat different street rights-of-way. Putting a 60′ street into another 60′ street requires the merged path to be 80′ or more. You can get around this with couplets, bypasses, and various other tricks – but the title of this post is “Organic Streets Defined,” and any further discussion would fall squarely into the category of “Organic Streets Applied.”

You might wonder, if polycentric radial networks are so great, why don’t we see more of them? We’ll get to that in a minute. But first, here’s Madrid.

Do these streets make any sense to you? They didn’t to 19th-century Spanish Urban Planners. In Barcelona, they surrounded the old city with a giant, monolithic grid plan: the Eixample. Later additions to Madrid’s outer districts were similarly orthogonal. Thankfully, the old city managed to grow to considerable size before modern urban planning took hold. It’s still there to be studied.

Interesting thing about Medieval Madrid: Almost every street is radial to some park or square. How this happened is a chicken and egg question. Some long-established centers of activity, like the Palacio Real (a site that dates to the Moorish era), no doubt spawned their own radial streets. Other centers, like the Plaza Mayor, appear to have been cut out of the existing city fabric after much of the local street network was already in place.

Here’s the radial networks for six selected squares within old Madrid:

You can continue doing this for other squares or centers of activity. (there are many, many more in the Centro district alone.) Or you can try it in other medieval street grids – in Italy, Germany, the Middle East. The area around the Jokhang in Lhasa (the Tibetan religious city) followed a loose polycentric radial pattern before the Chinese government razed much of the original architecture in favor of concrete block apartments, squaring off the street network in the process.

Why don’t we see more polycentric radial street networks in the U.S.? Simple answer:

The Public Land Survey System PLSS covered the USA with a grid from the time of the first (white) settlers. The PLSS promoted easy subdivision of land within the grid system. Every square mile of the West and Midwest constitutes a 640-acre “section,” and one could cut up these sections to create a piece of property with a description something along the lines of: “The northwest 1/4 of the southwest 1/4 of the southeast 1/4 of Section 25, Township 21 North, Range 3 East, Willamette Meridian and Baseline.”

This enabled subdivision without surveying. Since the “section lines” all had defined lengths and bearings on file at the county records office, all the local distances could be inferred. This ease of surveying wasn’t just confined to land, though; you could lay out highway networks the same way. Need a county road? Take “the eastern 30′ of the NW 1/4 of the SW 1/4 of the SE 1/4 of Section 25, T21N R3E”, add in “the western 30′ of the NE 1/4 of the SW 1/4 of the SE 1/4 of Section 25, T21N R3E,” and you had yourself a nice 60′ county road ROW that “split the difference” between two different parcels of land.

But as you can see from the map above, not every state was surveyed under the PLSS. Texas is one of them. Different parts of Texas have a variety of survey systems dating back to Spanish rule. Some resemble the PLSS; the Panhandle is laid out on a rigid north-south-east-west grid, while the oilfields of the Permian Basin are surveyed on a cant of about 20 degrees off north. The eastern portion of Texas doesn’t have any single survey system, however. That’s why Harris County has such an odd shape.

With no overarching land survey system, highway makers – everyone from Spanish missionaries to TXDOT engineers – have always been free to follow the shortest route between two points.

The first thing to note about the East Texas highway network is that almost *nothing* is truly north-south or east-west. The second thing to note is the number of cities and towns with five six, or many more radial highways.

But this doesn’t even tell the half of it. Because it’s a large-scale map, smaller county roads aren’t shown. As one zooms in, more radial roads become apparent. The pattern is almost fractal.

Take the small town of Alto, for instance. On the large-scale map it appears to lie at the junction of two highways, one roughly north-south and one roughly east-west. But a closer look finds that Alto has no less than eight different radial highways.

Some of these are very, very old – like the Old San Antonio Road, arguably the first highway in Texas. Others are considerably more recent. But all continue the evolution of the polycentric radial highway network. And it’s not just a Texas thing – you can find polycentric radial highway networks all over Georgia and the Carolinas, too. But the “polycentric radial” concept doesn’t *entirely* explain how these networks came about. To do that, we need to introduce a couple more concepts… concepts which will have to wait until the next post.

But what about cars? To answer this, put yourself in the shoes of a transportation planner from the 50s. You’ve got a rapidly suburbanizing metropolitan area, and you’re tasked with designing a freeway network – from scratch. Why freeways? Because they make this example simpler. With freeways you can speak in terms of “number of lanes” as a rough shorthand for capacity.

Here’s Anyregion, USA. Remember 3/4/5 triangles from high school geometry?

Travel demand to/from Downtown follows a rough inverse proportionality to distance. Travel demand between the satellites is lighter, but still significant. What’s this demand look like?

Your task as master transportation planner is to figure out where to put the freeways and how many lanes they’ll need.

The first concept that comes to your mind is a grid. You figure all of the traffic between Uptown and Westfield will want to bypass Downtown, so you need at least two lanes all the way around the loop. You also figure the travel demand between Downtown and Overpark will split up evenly via Westfield or Uptown; some people live on the south side of Overpark, some people live on the east side, some people just need to swing through Uptown to pick up some Chinese take-out on the way home. What’s this look like?

You drop this at a regional council meeting. It’s the 50′s, so the wood-paneled boardroom is thick with cigarette smoke. You’ve just finished your presentation when the Mayor of Overpark coughs and raises his hand. Why do Uptown and Westfield get a “straight shot” to Downtown, he asks, while Overpark residents have to take the long way around? He’s got a point. Chastened, you head back to the drawing board to drum up a radial freeway system. After a few days, you come up with this:

Since Overpark now has a “straight shot” to downtown, you might as well save a couple bucks on the loop route by routing it directly between Uptown and Westfield. It’s not like suburb-to-suburb trips will have to go *too* far out of their way. The next month, you head back to the regional council meeting. You’re getting ready to present the new plan when you notice the Mayor of Overpark is back. And he’s got the entire city council in tow. Worse, they’ve enlisted the help of influential senator Hugo “Hot Air” Humphrey. Humphrey is a bigwig in transportation appropriations, a “go to” guy for highway builders. Facing defeat, you draw up a system that gives Overpark the same radial treatment as Downtown.

Along the way, some graduate EITs in your office have been crunching the numbers. One day you’re pondering the upcoming ASCE golf tournament when one of them eagerly knocks on your office door. Seems he’s found some interesting data points. Seems Overpark’s preferred freeway network is also the least costly!

The graphic shows it all. As the highway network morphs from grid to radial to polycentric radial, total lane miles go down, and the amount of concrete used *also* goes down. Could it be that the Mayor of Overpark was right all along? Crazy. Scratching your head, you remark that the right-of-way acquisition costs for the Polycentric system are probably a bit higher than the grid. But right-of-way isn’t all that expensive compared to improvements, at least not in your rapidly growing southern/western city.

What else might come influence the cost of the system? You think… interchanges! What sort of interchanges would be required for all these freeways to interface with each other?

As it turns out, the grid highway network – by forcing traffic onto a few “primary” routes – creates busy highway intersections that can only be solved with a four-level stack interchanges. Stacks are expensive. But the radial system spreads traffic across smaller freeways, many with only two lanes in each direction. These freeways have low enough traffic flows that engineers can build the cheapest interchange of all – the lowly cloverleaf. Other interchanges require that you add a couple flyovers to the basic cloverleaf design – a “half stack” – but the only place requiring a full stack is Downtown.

There’s a lesson here – but what is it?

For a region with many complex traffic flows, a polycentric radial street and highway network will tend to be more cost-effective than a grid.

Polycentric Radial streets, when applied at the pedestrian level, minimize walking distances to major activity points. And Polycentric Radial highway networks are more cost-effective than grids, at least in relatively undeveloped areas. Where else might we find such networks? And is there a simpler, shorter name we might use to describe them? Something less cumbersome than “Polycentric Radial”?

More posts are coming.

Dallas and Fort Worth’s freeways form sort of an odd mutant setup (there’s no picture here because frankly it defies easy explanation). As you get close to the center of Cowtown or the Big D, the freeway system looks radial. When you get out in the middle, it’s more of a grid. Not coincidentally, congestion is much worse in the middle – it takes a lot longer for the road system around DFW Airport to clear out than it does for radial routes like the Central Expressway. We’ll look at why this is in another post sometime later. But for now I want to talk about Polycentric Radial street patterns.

We only really have one polycentric radial urban street network in America: L’Enfant’s plan for the District of Columbia. The reasons for this are largely aesthetic. DC’s streets radiate to the seats of government power, and the wide avenues combine with these symbols to create vistas anchored by the Capitol, or the White House, or the Washington Monument.

Of course, DC traffic is horrendous. But this isn’t L’Enfant’s fault so much as it is the work of 75 years of municipal engineers turning every street into a one-way and putting a traffic light on every corner. DC would run smoothly (if slowly) if the government simply restriped every street as a two-way, replaced the lights with two-way stops and mini-roundabouts, and converted the circles to yield-on-entry.

This will never happen.

But in the meantime, Washington’s polycentric radial network is interesting in that it minimizes travel distances to important destinations. And for pedestrians, travel distance is everything. Cars and trains are capable of traveling at vastly different speeds depending on context, so we strive create contexts conducive to high speeds (freeways and high-speed rail lines). But pedestrians walk at about the same speed on dirt as on pavement. Grade separation means nothing. For a person on foot in an urban setting, travel time and travel distance are identical.

Looking back at Washington, L’Enfant’s diagonls aren’t confined to the White House and Capitol axes. The entire city is scattered with a number of circles and squares with multiple radiating avenues. This puts most Washingtonians within easy stroll of an outdoor hanging spot. Attempting to trace out all these different centers and their associated radials quickly gets unwieldy; instead, I’ve simply placed a “dot” at the location of each park with at least eight seperate radial streets.

“As-built” Washington differs from this plan somewhat. In the northern part of the city, fewer diagonal streets were actually constructed then shown. In the eastern part, more were – so there are *even more* squares with 8+ radials than you see here. A couple of the diagonals exist as railroad rights-of-way rather than public streets. And at least one of the squares shown here is now underneath a freeway interchange.

But the general concept remains the same; DC provides more people more access to more parks and gathering spaces than would otherwise be possible in a simple grid structure. This has wide-ranging applications.

Name: “WoW Roundabout”
Location: Washington Corridor, Houston
Cross Streets: Washington Avenue, Westcott Street, Arnot Street
Coordinates:
Modern Roundabout: Yes
Aesthetic Treatment:
Traffic Control: Yield-on-entry, advance signage

Name: unknown
Location: Westchase, Houston
Cross Streets: Seagler Road, Meadowglen Lane
Coordinates:
Modern Roundabout: Yes
Aesthetic Treatment: Old-growth oak tree preserve in center island; extensive landscaping throughout
Traffic Control: Yield-on-entry, advance signage

Name: unknown
Location: Museum District, Houston
Cross Streets: Main Street, Montrose Boulevard, Hermann Drive,
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Illuminated Water Fountain (Mecom Fountain)
Traffic Control: Northeast-southwest through traffic on Main Street has the right-of-way. All other entries have stop signs. In-circle yield signs for traffic turning left from Main Street. Advance warning signs on Montrose southbound, both directions on Main Street: CIRCULAR INTERSECTION text on yellow diamond.

Name: unknown
Location: Museum District, Houston
Cross Streets: Montrose Boulevard, Golf Course Drive, Hermann Park parking lots
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Bronze Sam Houston on horseback, on a pedestal. Brick walkways, grass, shrubs.
Traffic Control: Mixed. Entry from north (Montrose) has a yield sign. Entry from west (parking lot) has no traffic control; yield is implied through design features. Entry from east (Golf Course) has priority, with an in-circle yield for the conflicting movement (left and U-turns to Montrose northbound).

Name: Pearland “Roundabout”
Location: Pearland, Texas (south Houston suburbs)
Cross Streets: Pearland Parkway, McHard Road
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Large central island used as a detention pond. Monument signs with Shadow Creek Ranch logo facing north and west entries.
Traffic Control: North-South traffic on Pearland Parkway has priority, with in-circle yields for conflicting movements. Entering traffic from west (McHard Road) also has yield. Overhead, freeway-style guide signs (with diagrammatic arrows) on all entries. 25mph curve advisory plaque for southbound Pearland Parkway at north entrance. Southbound traffic makes a 120-degree turn in-circle to stay on Pearland Parkway. Northbound traffic follows a smooth, ~1800′ radius arc through the circle. Approach speeds are 45mph from south, 50mph from north.

Name: unknown
Location: River Oaks
Cross Streets: Larchmont Road, Ella Lee Lane
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Central island is a small park. Grass, trees, shrubs, park benches.
Traffic Control: White keep right / median signs at all four entries. No other signage.

Name: unknown
Location: River Oaks
Cross Streets: Briarwood Court, Ella Lee Lane
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Lawn with trees.
Traffic Control: White keep right / median signs at all four entries. No other signage. Briarwood Court is a cul-de-sac street which serves no more than a few lots on either side of Ella Lee.

Name: various
Location: Bridgeland, Houston (uninc. Harris County)
Cross Streets: various
Coordinates:
Modern Roundabout: No
Aesthetic Treatment: Grass, Asphalt
Traffic Control: ???

Transit planners use the Transit Capacity and Quality of Service Manual to measure existing services and analyze proposed changes. In the TCQSM, 1/4 mile, or 1320 feet, is considered the maximum distance that a non-transit-dependent user will walk to a local bus route.

Establishment New Urbanism takes heavily from DPZ’s transect concept, with a large emphasis on providing local-service retail within a “five minute walk” of most residences. Traffic engineers design crosswalks and other pedestrian facilities based on an AASHTO standard walking speed of 4 feet per second. 5 minutes @ 4 ft/s is 1200 feet. Faster pedestrians can cover a greater distance; 4.5ft/s yields a 1350′ radius, while I’ve clocked my own “brisk walk” speed at a bit over 6 ft/s, for a radius of 1800′.

A Pattern Language, the landmark 1977 environmental design tome that covers everything from furniture arrangement to city design, suggests that the local commercial districts which define the edges of neighborhood boundaries work best with a walk radius of about 1500 feet.

You see the pattern here. 1320. 1500. 1200-1350.

The effective walking distance to (i) transit, (ii) local service retail, and (iii) the commercial streets that define the edges of neighborhoods is so similar, why not combine the three into a single planning concept? The Everything Node. We already hit two of the three with “transit oriented development.” But TOD leaves out the role that major traffic streets play in defining the edges of neighborhoods, and besides, I’m not really a fan of the term.

What, exactly, defines TOD? Is it access to transit? Then a cluster of 1950s ranchers adjacent to a light rail stop fits the definition. What do you think of when you visualize “TOD”? Mixed use? A variety of shops within walking distance? Ample common areas? None of these have anything to do with transit. We ought to be promoting the “pedestrian-oriented development,” and perhaps we would be were it not for the unfortunate acronym that results. Think “POD” and you probably visualize some combination of dancing hipsters, overly-futuristic airport peoplemovers and christian rap-metal bands.

The Everything Node goes one step further than TOD, by mating transit nodes and local service retail with arterial-arterial intersections, supporting a wider variety of retail and with it, a more interesting neighborhood.

Simple, hierarchical definition of an Everything Node
1.) Street intersection or intersections
2.) Local-service retail centered on said intersection(s)
3.) Variety of other uses available within walking distance of said intersection(s)
4.) If contextually appropriate, transit service at or near said intersection(s).

Why the hedge on number 4? Simple: transit is a requirement in urban areas, but it’s a lot less essential in the hinterlands. And make no mistake; small towns are the very definition of an “Everything Node,” with retail, office, residential, industrial, and every other imaginable use within a few minutes’ walk. Many a Texas town routes major US highways around its courthouse, centering everything on their region’s most primary through-routes. And these small towns often contain efficiencies not seen in more urban developments. For instance, in Rockdale, Texas, the local mexican restaurant sits right across the street from the feed shop. One trip to Rockdale’s main street, and both you and your cows will be sleeping soundly.

In built-out Calthorpe projects (Issaquah Highlands and San Elijo Hills), the couplet (another term for a one-way pair) is split from a standard suburban arterial, with two intersecting arterials forming a “box” intersection. This is the simplest couplet-couplet intersection. But couplets are infinitely scalable; push several together and you achieve a set of meshed couplets, or a grid of one-ways. Meshed couplets have been a standard treatment for city street grids in CBDs since the mid-20th century, and if you live in a major metropolitan area, chances are you drive one on a regular basis.

To date, every meshed couplet in the United States was created by modifying a previously-established gridiron. As an urban design feature in new greenfield or large-scale infill development, meshed couplets emerge when two arterial-arterial “box” intersections (of the kind Calthorpe has built) occur in close enough proximity to require coordinated signal timing between them. This may occur in braided street networks, or in other street network topologies which will be discussed in future posts.

Meshed Couplets can support extensive networks of parallel and angle parking, enhancing sidewalk safety and reducing the need for surface parking lots. The smaller blocks and narrower street right-of-ways promoted by splitting opposite traffic streams likewise promote urbanism. There is a tradeoff between traffic speeds and walkability; larger blocks result in higher speeds, but reduce walkability, while the smallest blocks yield signal progression speeds that may not be acceptable to many jurisdictions.

Timing Meshed Couplets
There are two ways of going about timing a network of one-ways. The first (and more common) is to adopt an asymmetrical, “custom” timing based on existing traffic flows that greatly reduces the delay on major through-streets at the expense of other, less-used ones. Houston, Texas is a prime example; the peak-hour signal timing allows for a constant 30mph progression along many of the primary north-south streets, with the side-effect that many east-west travelers must contend with a sea of reds.

The second (and much less common) means of timing a grid network is four-way symmetrical signal timing. Four-way symmetrical signal timing balances bandwidth across all streets and all directions of travel, giving every direction a relatively similar chunk of the total cycle time. Downtown Portland, Oregon is one of the most extensive users of this signal timing pattern, which – when combined with Portland’s small 200′ block size – creates the ~15mph “bike friendly” signal progression, which we will study in a moment. But first, an illustration:

(click on the image for a larger version)

In such a network, the timed progression speed in all directions is the distance around a block divided by the signal cycle length, or:

In Portland, north-south streets are 80 feet, east-west streets are 60 feet, and interior blocks are 200′x200′. Thus the centerline distance around a block is (280+260+280+260), or 1080′. During off-peak hours, Portland uses a short 48-second cycle length, which yields a progression of 15.3mph. During peak hours, the signal cycle bumps up to 60 seconds (to reduce the percentage of time lost to yellow), which pushes the progression speed down to 12.7mph.

So, you ask, How do I design a grid for a 30mph progression? Well, first off, you’re crazy. Signal progression speed and speed limits are NOT the same thing. In any timed signal network, drivers will attempt to “keep up” with the timing, and if delayed (by a slow moving vehicle, for instance) will increase their speed beyond the signal progression speed to make up for lost time. If you build a walkable grid with 30mph arterials, you want traffic to *actually go 30mph*, which means you want to time your signals for 23 to 25mph. Let’s shoot for 23 as the starting point.

Plugging in 23mph and a 60-second signal cycle, we get a total block circumference of about 2000 feet. That’s about 500′ square. Now, 500′ square is too big for a walkable grid, so we’ve got three options here. First, reduce the cycle length (unlikely; your jurisdiction will balk at low cycle times). Second, reduce the traffic speed (the simplest option). Or Third, cut up the superblock. A 500′ block can be sliced and diced into a couple of 250′ blocks with two-way, local-service interior streets (which yields a grid on the scale of Fort Worth or Portland), or it can be cut up with a couple of narrow, one-lane one-way streets (which yields a grid on the scale of Philadelphia, Lower Manhattan, or Boston’s North End).

But like any signal-based network, there are significant downsides. Four-way symmetrical signal progression forces left-turning traffic to wait an entire signal cycle, just like a “perfectly timed” suburban supergrid. In Portland, if you come up on a fresh green and make a left turn, you’ll find that the light ahead of you turns red just before you arrive. At that point, you end up waiting an entire cycle length. This is an inconvenience if traffic is light, but if left-turning volumes are heavy, it has the potential to break down the entire network. You can “massage” the timing slightly to reduce this, but beyond a certain point you’ve effectively switched to an asymmetrical/custom grid timing, at which point you wonder why you’re still reading about Portland-style timing at all. So let’s talk about circulation.

Alternating circulation in a one-way grid allows for drivers from any direction to easily reach parking or driveway entrances fronting any street. This is essential for two reasons. First, you want to minimize the number of driveway entrances for reasons of pedestrian safety and comfort. A network with easy circulation promotes this. Second, you want to minimize the number of queued cars at lights. A network of one-ways makes this easy, by allowing for left-turn on red. Alternating right-turn on red and left-turn on red intersections create a “checkerboard” of blocks which can be circumnavigated without waiting for the light to change.

Each of the red blocks in the above photo can be circled in this way. In Downtown Houston, this clears left- and right-turning traffic out of the intersection before the next platoon of through traffic arrives. It also allows for extremely large parking garages of eight stories or more to depend on a single driveway entrance. The same circulation efficiency which allows Houston to work can be applied to developments of lesser intensity, as a means to minimize the number of lanes on all streets. Fewer lanes and narrower street rights-of-way translate into a more intimate and pedestrian-friendly environment, an environment which is good for the mind, body, soul, and adjacent retail tenants.