Compromises on North Broad

Jon Geeting recently picked up on an implicit idea in my Width, and the Perception of Width article: giving Broad Street the same design standard as e.g. Paris' Ave. Kléber. He has also run further with this idea than I ever conceived, including the idea of putting a BRT lane down Broad. While these are all good ideas in their own right, however, at a certain point Broad Street--especially North Broad--will demand, at a certain level, compromise.

The reason for this is a legacy of highway projects that never got built, particularly in the northern suburbs. While this is, as Alon Levy pointed out, good for mass transit (and a major reason why reactivations of the Newtown Branch and Bethlehem Branch (at least as far as Quakertown) are good ideas), it does impose a certain stark reality on Broad Street.

The 309 (blue) and 611 (red) corridors both feed into North Broad.

Both of the major highways serving the northern suburbs--PA-611 and PA-309--drain onto North Broad. Because of this, much of the traffic into Center City from central Bucks and Montgomery Counties comes down North Broad: while the Northeast Corridor May function as an escape valve, it only serves a small portion of the northwest corner of this area; commuters from Ambler, Doylestown, Flowertown, Glenside, Hatboro, Jenkintown, Warminster, or Willow Grove have little choice other than SEPTA or North Broad. And of course, commuters from (especially) Churchville and Richboro have none. This makes North Broad's demand far higher than South Broad's, one of the reasons why the current traffic pattern involves converting a parking lane into a through lane at peak.

This raises the difficulty that, while an MWB would be easy to implement on South Broad, it is somewhat more difficult on North Broad. While it certainly offers significant traffic calming and safety improvements, it doesn't really address flow interruption congestion (congestion caused by traffic lights), which the peak fifth lane functions as a band-aid for; implementing it as a reversible lane would  (a) make it quite difficult to fit everything else onto the street, in addition to being (b) quite ugly. And of course, the real solution to the flow problem--eliminating traffic lights--would be catastrophic for all other users. All of this implies that North Broad requires, at minimum, four through lanes with good light timing minimizing stops, which, of course, dashes any hope of BRT on the corridor. (Not that the 4 or 16, with the subway underneath, really need it.) That said, BRT would be useful on several other corridors, such as Erie, Allegheny, Lehigh, and Washington Aves. BRT could be run down the less busy South Broad corridor, but it is, if anything, even less useful there (cf. Reading Between the Lines).

So the key is attaining a tradeoff--a balance of priorities. To do this we need some sort of hierarchical modal evaluation criteria. This could be easily abused and turned into a checklist by the same kinds of numbnuts who eviscerated any pretense of urbanism in Syracuse, but having such a checklist exist, and be public knowledge, would be an invaluable tool in project evaluation and project criticism. But what would this entail? To see my answer, partitioned from this post due to increasing lengthiness (not that that's stopped me before) and as subject matter deserving its own post--stay tuned.

Width, and the Perception of Width

This has also run on Strong Towns and This Old City; these are the permalinks for both.

Looking at these two links--Hector Street, Conshohocken, and 6th St., Bella Vista--which is wider?
Looking at these two links--Broad Street, Philadelphia, and Avenue Kléber, Paris--which is wider?
Looking at these two pictures--

--which is wider?

The answer is neither. In each example, both streets are the same width, but the way the width is expressed is different. The middle example is, of course, familiar to Stroad to Boulevard readers; let us concentrate on the bottom example.

The top photo is of a complete street in the Netherlands. It has two travel lanes of 10 feet, a parking lane with tree bumpouts of eight feet, and ample bike and sidewalk facilities. Its Streetmix profile is thus

where the planter+tree on one side represents the bumpout, and the parked car on the other the parking lane. Visualize the tree above the car, and the car below the tree, to get a better idea of how it looks in reality.

By contrast, the US complete street's Streetmix profile is

Notice how scrunched up the sidewalk is, and how vulnerable the cyclists.

Both examples have nine-foot sidewalks and eight-foot parking lanes. In the Netherlands, the travel lanes are only ten feet wide, and the cycle path looks to be eight feet; in the US, the travel lanes are 12 ft. each, and the cycle path is six feet. Both add up to 70.

But also notice how the placement of each use in the street communicates information about it. In the US example, cars are clearly given priority, not just by being given wider travel lanes, but also because the bike lanes are placed in such a manner as to expand the travel lanes' clear zone. There are no bumpouts, and so the trees are scrunched along the curb, and pedestrians forced onto the sidewalk. Of the seventy feet, a whopping 52 is part of the car realm, just shy of 75%. This means that for every unit of pedestrian space, there are four of auto space, a ratio of 1:4.

By contrast, in the Dutch example, cars are clearly subsumed. The travel lanes are narrowed, and the parking lane is interspersed with bumpouts; the parking lanes now become the edge space. The bike lanes have been pulled away from potential conflicts, and widened slightly. The sidewalks remain the same width, but there are no trees to present barriers. In this example, a mere 36 feet of seventy is part of the auto realm, if you count the bumpouts as also part of it, giving a ratio very nearly 1:1.

Just as the optimal building height/street width ratio is 1:1, so too is the pedestrian realm/auto realm on a street. This ratio seems to be paramount in the psychological assessment of street width. Hence we can do the analysis we just did for a typical Dutch arterial vs. a typical US Main Street, and still get questions like

Why streets in the US are so wide?

There is space for two row of parked car on both sides and about three times the width of a vehicle in the middle. But technicaly there is no need to park on the road because most house have one or two garages and a lot of space in front of them. I think that`s a waste of space and structure.

and answers like

Because until a few years ago, people looked at safety data from rural roads, and found that roads with 12 ft lanes and wide shoulders had fewer crashes. So, they applied this to residential streets, and guess what? It hasn't worked. Instead of being safer, people drive down these streets as if they were on rural highways, at speed much too fast for neighborhoods.

In the future, you'll see roadways designed more closely for the functions they play in the overall road network, with narrower city and neighborhood streets.

The psychological assessment of width the commentator exhibited is what I am calling perceived width, or alternately the perception of width. Unlike absolute width, which is determined by the lot lines, perceived width is determined by the modulation of the street--how much is given over to cars and pedestrians (and whether bikes are treated as cars or pedestrians)--as well as setback depth (if any). Optimal modulation has the two major users evenly split the street, and setbacks, if they exist at all, be purely ceremonial (no more than 15 ft. or so), or be shielded by an edge along the lot line, created by trees or bushes. We'll assume no setback, since that kind of space is usually (a) green and (b) in the public way of neither. That is, it is neither of the auto realm nor of the pedestrian realm.

Cars need a spatial minimum. The absolute minimum is often thought of as ten feet a travel lane, and eight feet a parking lane; parking lane accesses can, however, be nine feet. Give-way streets need slightly wider travel lanes, so that the car giving way can pull into a space without having to go through a full parallel-parking maneuver. Let's say 14-18 feet for them.

Let us now turn to our first example. Both streets are fifty feet, with ten-foot sidewalks. This implies that both can support give-way traffic; neither does. The Conshy example, however, hosts a bus and thus cannot be pedestrian-prioritized space--that is, a home zone. The design given to it is therefore nearly optimal. This is what it looks like on Streetmix:

The bumpout is irregular, with four parking spaces between each one (on each side of the street). In this, it's like a slightly less verdant example of the Dutch complete street. This design offers a wider walking area than the Philadelphia example:

Both designs, while good, are, however, inferior to

which uses the bumpout/parking strategy to increase available space without losing trees, but in addition narrows the travel lane and adds a cycle path. In this design, 26 feet are part of the auto realm, and 24 the pedestrian--as close to a 1:1 ratio as a 50' street is likely to get.

The second example contrasts Philadelphia's Broad Street--a classic example of a grand avenue converted into a traffic sewer arterial, with Paris' Av. Kléber. The former looks like

this on Streetmix

and the latter, this.

I find myself going into Stroad to Boulevard territory here: notice that there is 30' of pedestrian realm on either side of the street, versus 16' on Broad Street. The bike and parking functions are both elements of a single shared service street, which the sidewalk spills onto; transit islands can also be installed on the green medians separating the two realms from one another. The bottom line isn't just that there is more people space--and less car space--on the Av. Kléber; it's also that one perceives of the Av. Kléber as being more humanized, and narrow, than one does of Broad, which feels more autocentric, and wide.

The perception of width is the psychological element of street design. Psychologically more humane streets tend to attract more pedestrian activity, and more small-scale, innovative, and experimental economic uses catering to pedestrians. It is, at engineering scale, the same thing a parklet signals: the idea that people care about the street as a street, and not just a place for automobile movement and storage.

Multiway Boulevards, Transit Avenues

Stroad to Boulevard is a blog dedicated to a single policy recommendation: the replacement of American-style arterial avenues (aka stroads aka wide boulevards) with multi-way boulevards, with examples mainly drawn from the French style. This is justified primarily by economic gardening: stroads fail, according to Strong Towns' Charles Marohn, to create places of value (i.e. offering a public environment nurturing of true value-creating enterprise over life cycles). In addition, the benefits of urban greenery are obvious, and (as Stroad to Boulevard's author points out) benefit congestion by decoupling through traffic from causes of friction, after Jarrett Walker.

Multi-way boulevards appear endemic to places of value in Europe, and rudimentary versions even occur in places of high value in the United States--see, for example, Commonwealth Avenue, Park Avenue, and K Street--but they have yet to be accepted here. Part of this is due to different focuses in transportation planning, engineering, and design: in Europe, a more multi-modal approach became more accepted and, yes, welcomed after the Dutch bike revolution, whereas in the United States the moving-cars-around orthodoxy continued to prevail. In essence, this means American transportation engineering is now a generation behind Europe's.

What I want to explore now are two things: firstly, the variety of configurations available for such boulevards, and secondly, ways to implement transit in the scheme.

What is the Minimum Width?

Stroad to Boulevard carries, essentially, a single cross-section, seen above, for a typical boulevard--100 feet. Of this, two access ways (curbside woonerven) flank a four-lane central artery. Each access way is 28 feet, and the central artery 44 feet, which implies that each driving lane is 11 feet. It is possible to tweak some of the design features--for instance, dropping a foot off each of the central passing lanes (not meant for trucks anyway) and adding it to the access lanes. As woonerven, the access area fulfills primary pedestrian and cyclist needs and secondary auto ones (primarily loading and parking)--to function, in other words, as a street--while the central artery is able to function as a road.

We'll call this full profile--bidirectional, with twin woonerven. It is intuitive that this would be optimal for 100-foot arteries, such as Broad and Market streets.

Ignore the school bus. There would be no bus lanes at this standard.

The narrowest a full-profile multi-way boulevard could be would have two driving lanes in the center, flanked by woonerven on either side; this is created by removing the 100-footer's inner passing lanes. Taking 28 foot as the standard woonerf width, this implies that the narrowest full-profile multi-way boulevard must be 76 feet--that is, (28*2)+(10*2). Since few (if any) American streets are exactly 76 feet wide, this implies that 80 feet is the smallest practicable limit, which we would fill out by adding a foot to both driving lanes and each access lane--(29*2)+(11*2).

If you split this in two, you would get a half-profile boulevard. The narrowest this could be would be 38 feet (or a more comfortable 40 feet)--assuming no pedestrian realm at all to one side. Add a sidewalk and you've got 50 feet.

The problem with this, however, is that most nearly every 50-ft street is a street, and thus not compatible with the MWB treatment. Even Paris' Boulevard Haussmann is a street. In fact, I do not believe there are any half-profile boulevards in existence--they gobble too much space with their access woonerven, provide inadequate parking relative to their full-street counterparts, and most 50-ft streets offer ample natural pedestrian realms--almost a 1:1 sidewalk/street ratio as long as the carriageway is held to reasonable widths (that is a ten-foot driving lane plus two eight-foot parking lanes, for 26 feet total).

So we can say the narrowest useful multi-way boulevard is 80 feet. This actually works well for most American cities, which tend to have wider streets that can better support this treatment.

Adding Transit to the Scheme
Most European multi-way boulevards have buses running in the outer through lanes, and some have  light rail in the median--see this example from Rotterdam. Such boulevards can have truly tremendous widths--Rotterdam's Schiekade looks to be, for example, some 160 feet wide--but we are focused on developing light rail along boulevards of more manageable width.

Girard Avenue is a good example: The 15 trolley runs in the median of what is here a 120-foot street. This represents an easy insertion into the 100-foot multi-way boulevard standard--see above. Likewise, a transit median on a 100-foot multiway boulevard, like would be a relatively simple insertion into the 80-foot standard--see below. Washington Ave. would be a good possibility for this treatment--see below.

The problem comes when you need to put a transit median into an 80-foot way, like Baltimore Avenue. Without the trolley median, it would be equivalent to 60 feet, too narrow for a full boulevard. Could a sacrifice need be made? Just the woonerven--Stroad to Boulevard shows us the way. While that post is about 66-foot streets, a crop of three feet on both sides brings us down to 60 feet, enough to insert the 20-foot transit median in between. This is, again, perfectly encapsulated in Rotterdam, on the Middellandstraat:

 Even where the street is too narrow to support full access woonerven, it's still possible to support a transit median. But what about 60-foot streets, like Germantown Avenue? This final class does require a sacrifice--Do you give mass transit a dedicated right-of-way and cut parking? Or do you mix traffic in two travel lanes and provide it? Both solutions require a tradeoff: with the former you have better throughput on all modes, but most constituencies demand parking and reject any proposal reducing it, Shoup be damned.

So in conclusion, we can say that, firstly, the nature of a multi-way boulevard makes 76 ft. its absolute minimum, as that width has two driving lanes coupled with mirrored woonerven; this can be retranslated into an 80-foot standard for most American municipalities; secondly, that the addition of transit lanes blows the width out a bit, and that a 100-foot transit boulevard is analogous to an 80-foot multi-way boulevard, 120 to 100, and so forth; thirdly, that transit provision on sub-100-foot streets requires a transit avenue, built like a Dutch complete street; and fourthly, that at 60 feet a tradeoff must be made between throughput and parking.

While this post is by no means perfect, consider it a framework for applying transportation engineering and urban design to the streetscape; consider it a framework for implementing European design standards in American situations.