Monday, February 24, 2014

Tour-based models and Walking

Only recently have travel models advanced to tour based models that recognize that all trips are not made from the home, but that people 'run errands', and combine several trips into a single tour. This has implications not only for automobiles but for all modes.

Every trip begins and ends with a walk. It is not always a very long walk, but it always exists. Every transit trips begins with a walk to the station/stop. And ends with a walk from the station/stop to the final destination. For automobiles, the initial walk distances is typically very short, as most people park closely to where they live. But the walk after parking, across the parking lot, or out of the parking garage, may be quite long. In both cases, the portion of trip spent in-vehicle is only part of the journey.

It would be better if each portion of that journey was considered a trip on a tour. Transportation models ought to consider pedestrian trips, and pedestrian scale transportation networks will require further developments in this area. Considering only the in-vehicle time ignores the significance of the walking portion of these trips, and the significance of the walking environment.

Much of the research on roads is on the effect of pavement quality on travel speed. There should likewise be extensive research on the effects of the quality of the pedestrian environment on both the likeliness to walk and the distances walked. This research will have two strands: The first geometric, measuring network qualities such as connectedness and the directness of paths through the network; the second qualitative, measuring the quality of the travel path.

Secondly, for development considerations, it would be better to consider each portion of the trip (Walk, Vehicle, Walk) as a separate trip. Specifically for purposes of retail gravitation. If the vehicle access point (station/stop/garage) is a trip end, then centers where a greater number of trip ends concentrate should be more attractive. Where someone catches a train (or keeps their car) may vary by virtue of what else is there.

Friday, February 21, 2014

Attached Housing, Detached Garages?

In January, I visited my friend near Washington, DC, where he owns a very expensive house. It was not a small house, but it was an attached house. But it was a very well designed attached house. With a detached garage, which was attached to other garages.

It is easy to forget that the Ranch or Rambler house of American suburbia was (in its time) quite an innovation. Historically, most garages were not attached. For the narrow and deep lots of streetcar America, it was impossibility. So most garages were alley access to the rear, and many were little more than sheds.  Putting the garage and house under a single roof was quite an innovation.

In most of America, attached housing still labors under a stigma. It is still second-class housing, for people unable to afford better. But I wonder--would more people accept attached garages? Were a developer to detach garages from houses, and then attach the garages, would the result be a saleable product?

Perhaps. But it would have to be well designed, and designed in such a way to reduce the amount of land area devoted to roads. The size of an average two-car garage is 400 SF. Detaching them from houses and aggregating them doesn't change the actual number of SF of ground area used. The size of an average two-car garage is 400 SF. Detaching them from houses and aggregating them doesn't change the actual number of SF of ground area used.

There is a reason that apartments have parking lots rather than individual garages. Rather than having individual driveways accessing individual parking lots, there is a single driveway accessing multiple parking lots. But that also constitutes a loss of square feet to the owners--a driveway is still usable living space.

To make it work, the garage clusters would need to be peripheral, located on the edge of the development. Yet to maintain the garages as personal and private space they would need to by connected to the homes. This suggests a cul-de-sak format, with garages arranged around a bulb that has ready access to the roadway.

Not an attractive urban pattern. And thus not an efficient development pattern.

Shout out: BusNinja


Monday, February 17, 2014

More Marchetti's Constant


"Why we're reaching our limits as a one-hour city"

Specifically, I was struck by this paragraph:
The one-hour-wide city, in Sydney, is reaching its limits. A city that has got 20 people a hectare and 40 kilometres an hour will become dysfunctional after about 2.5 million people.
One of the topics I'm very interested in is who gets rail--how big (and how dense) does a metro area have to be before it gets rail? (Or equivalent fixed-guideway transit).

My prior 'rule of thumb' has been about 2 million people. I would be VERY interested to see where (and how) the author arrives at that number. The author, Peter Newman, is a professor of Sustainability in Australia, who popularized the term 'automobile dependence' and write a highly cited book about it.

Friday, February 14, 2014

Marchetti's Constant

Reading about Marchetti's constant today, which suggests that all humans strongly prefer to spend about an hour a day in travel. (Half hour in, half hour out). This supposedly holds across all cultures, in all contexts.
Walking about 5 km/hr, and returning back to the cave far the night, gives a territory radius of about 2.5 km and an area of about 20 km^2.
20 km^2 is about 5000 acres. Conveniently, Salt Lake City's blocks are about 10 acres, so I have a ready conversion metric, so that means 500 blocks, which is a gird 22 blocks on a side. #

This range supposedly applies to villages with agricultural fields and the area contained within the outer walls of ancient cities. This implies that it applies to modern cities as well, or rather the walkable portions of them. Which implies a size limit for Transit Oriented Developments as well. Given a half hour time budget for commuting, of which at least some is taken up by the transit trip itself (say half). This suggests that a 10 minute walk*** to a transit station may actually be more appropriate. Indeed, Calthorpe's original TOD concept* had a 10 minute walk to a heavy rail, from a max distance of about 2000'.

This suggests that the 'Density Gradiant' near transit stations should be an extremely steep one, with elevator apartments (3+ stories) adjacent, and single family homes a half mile away. Because people beyond that half mile aren't likely to walk to the station*.

TOD (Transit Oriented Development) is frequently done badly. Largely because we lack these metrics. If TOD is going to work, the majority of residential units have to be next to the station. Not behind the parking lot, not beyond a belt of adjacent commercial development. Right next to it. Let the Park&Riders walk through that access path to the station. Direct them down a single avenue, running the gauntlet of convenience retail every day. Done right, ground-floor retail could actually be made to work, and in  way that inconveniences the minimum number of apartment dwellers.

In a way, NAM** does us a bad turn. The idea was for transit oriented metropolis, with independent cities clustered around transit stations. Each with their own 'neighborhood center' core. of retail and office development. Nothing like that has ever been built. A 'Transit Metropolis', where transit is embedded in the urban fabric, would look very different.

Recall that most 'New Urbanism'**** is greenfield development, not infill. Even if we fix the street connectivity and raise the density, any given metro area has less than 50 transit stations. 50 possible TODs, each with about 288 acres within 2000' of the station. Some tiny fraction of the total urban area.

*Certainly, some will, but they will be the fast walkers: Young, able, athletic.  Biking to station probably occurs from outside the half mile circle. Presuming a bike to travel 3 times as fast as a walker, say the 'Bike Zone' extends out to 1.5 miles.
**The Next American Metropolis
***People coming from longer distances walk faster. I stroll at about 3.0 miles per hour, and race-walk at about 4.0 miles per hour.
**** AKA 'New Suburbanism'

#How big can a CBD (Central Business District) get before it is 'too big'? It can't scale in size to population, certainly.

The paper is happily available online here

Wednesday, February 12, 2014

Bus Stop Location

Two issues concerning bus stops: 1) Their location; 2) Bus bulbouts

Bus stops on the far side of an intersection are superior to buses on the near side of an intersection. There is a naive belief that a bus already stopped for a red light can load and unload passengers at the same time. For a limited number of childless, young, able bodied passengers, this is true. However, the majority of Americans are neither young nor able bodied, a statement doubly true for many bus riders. They are often old, infirm, disabled, or towing children. Further, during peak periods, the time required to board a number of passengers exceeds the length of the 'red' portion of the traffic signal (<30 seconds). Trying to synchronize the length of a load/unload cycle for a bus with the stop/go cycle is bound for failure.

Any stopped bus blocks right-turning cars, either trying to turn off the road, or trying to turn onto the road. For automobiles, the location of bus-stops is moot. But for buses, far-side stops mean that 50% of the time, a bus will be able to travel through on a green light, rather than having to stop at every light, as if it were red, in order to pick up passengers. Given that signalized intersections represent the majority of delay time for urban travel, this represents a significant savings of time and an increase in travel speed. 

A naive student once lauded the virtues of bus bulbouts as an attribute of an effective transit system. Nothing is further from the truth. Bus bulbouts are the are an attribute of an auto-dominated transportation system. A bus without a dedicated lane travels at about 10 mph (including stops). The more frequently a bus must stop, the slower it travels. Urban traffic travels at about 17 mph. This leads to a demand for 'bus pullouts', where a bus leaves the travel lane. Exiting and re-entering the travel lane takes time, making buses even slower. Solutions include reducing the number of stops and improving bus stops. The latter is worthwhile if there is more than one stop within a 400m distance. (Bus poles are cheap and often too frequently placed).The latter improves bus 'dwell' time at a given stop, typically by improving curb and gutter infrastructure to permit level boarding with minimal gap between curb and bus. (One real advantage of rail over bus is faster boarding, due to more efficient 'docking'). 

Monday, February 10, 2014

Transit System Planning

To effectively model Transit Demand, a simulation program capable of using both a gravity model and a network model is required. Effectively, a Travel Demand Model simulator such as CUBE. Use a familiar model, but make it possible for agencies to "Be the DoT, not the BPR". (Department of Transportation, not Bureau of Public Roads). Map each type of transit to a functional class of road. Treat Light Rail as a highway, regional buses as arterials, flex routes as local streets. Connect centroids only to places within walk distance. Weight propensity to walk inversely to distance. Include a coefficient for quality of the built environment. Treat the first two as 'limited access' facilities, where board is only possible at stops. Lacking Cube, however, it should be possible using ESRI Network Analyst.

Friday, February 7, 2014

On Free Transit

Talinn, Estonia, has free public transit. 

Two things of note:
a) "passenger demand of just 3 percent — and attributed most of that gain to other factors, such as service improvements and new priority lanes for buses"
b) "if any modal shift is happening, it’s that some people are walking less and riding transit more"
The former suggests that service quality and speed matter more than fares. The latter suggests that they are spacing their bus stops too closely (400m), so that the buses are acting a pedestrian circulator, rather than a pedestrian extender. The latter is the more important point.
For buses, there is increasing marginal cost (of delay) per passenger. Every stop makes the journey longer. The more people on the bus, the greater the cumulative delay. If boarding takes 15 seconds, and there are four passengers on the bus, the result is 60 seconds of person-delay. If there are 8 people, the resulting delay is twice as large.
For a pedestrian facing the choice whether to walk or wait for the bus, the per passenger delay caused by boarding is immaterial. The only time-cost is waiting for the bus. If a bus comes every 5 minutes, it is worth waiting about 3 minutes to catch the bus. A comfortable walking pace is 5 m/s, so for any distance over 180m**, it is worth wait to catch a bus. On which basis, bus stops would be placed every 180m. So every four blocks would add a minute of travel time***, per passenger. Ergo, as headway (buses/hour) rises, distance between stops should rise as well.
* 3 minutes x 60 seconds/minute x 5 meters per second
**About 1 SLC block.
***Four blocks-->four stops x 15s delay/stop
'Best Bus', and the 'Bad Bus'. 

Wednesday, February 5, 2014

Reflections on Metropolitan Form and Binary Cities

Accessibility is not purely a field effect, but also a network effect. Rather than the center broadcasting a field effect, the network broadcasts a field of effect, with the strength of that field proportional to the network time-distance to the center. 
Take Binary Metropoli such as Dallas-Ft. Worth or Minneapolis-St. Paul, where two existing monocentric cities have become conjoined. The patterns of sub-centers is distinct from emergent polycentric, where monocentric cities have developed sub-centers. In monocentric cities with radial transit routes, sub-centers emerge as  function of distance from the center. As sub-centers develop sufficiently that their level of activity density warrants it, orbital routes between peripheral sub-centers will develop. For binary metropoli, the pattern is different. The intersection of radial branches between the city centers should emerge as a sub-center.

How does a sub-center differ from an 'Edge City'? Sub-centers have to do with the efficient allocation of retail and services minimize the overlap in market areas, and minimize transportation travel time to the centers.

To use the 'hierarchy of retail' as an example:

  • Regional malls...down to neighborhood,
  • Community Center/Power Center -- big box anchored, 100,000 SF 
  • Neighborhood, 55-60k SF..
  • Grocery anchored
  •  Specialty/In-line (Strip malls)
  • C-Store
The distribution is inversely proportional to their frequency. 

Edge Cities, in contrast, contain more than retail. Garreau characterized them in terms of office development. High-rise office development was traditionally the purview of downtowns, and its emergence outside them was kind of scandalous. High rise office development is associated with higher order services--not personal services and retail, but what is often called "Producer Services", or "Professional Services", characterized by FIRE industries (Finance, Industry, Real Estate).

Tuesday, February 4, 2014

Walking Along the Transit line

Spent a bit thinking about the special case we started with*--walking ALONG the transit line. Basically comes down to two cases: 
1) Get off at Station X, walk to destination.
2) Get off at Station X+1, walk back to destination

For Case 1
--If destination is less than halfway between Station X and Station X+1, walk back.
--If destination is MORE than halfway between Station X & Station X+1, walk to station X+1
Case 2 is much more interesting, and has to do with the walk ratio between the two. 
Getting off at station X+1 means you have to walk to the destination, and then walk back to Station X+1.

Now, for the transit option (Case 2) to be useful, it must be quicker than the transit option (all else equal).
The time to reach a destination using transit is*:
t=d/T + 2x/W

where t = time, d= distance, T = transit travel speed, x = distance of destination from Station X+1, and W is walk speed.
It's 2x, because leaving Station X+1, you still have to walk back to x+1.

Case 1 is always just d/W.
So an area is more accessible to Case 2 than Case 1 when d/T + 2x/W<d/W
If transit is 2x as fast as walking, so that T=2W... (2W, 3W, 4W,6W) we get:

d/2W + 2x/W<d/W
d/3W + 2x/W<d/W
d/4W + 2x/W<d/W
d/6W + 2x/W<d/W
Now I'll assume that the distance between Station X and Station X+1 is 12, and W =1 (making the math simply)
12/2*1 + 2x/1<12/1 --- >   6+ 2x < 12 --- >   x <3
12/3*1 + 2x/1<12/1 --- >   4+ 2x < 12 --- >   x <4
12/4*1 + 2x/1<12/1 --- >   3+ 2x < 12 --- >   x <4.5
12/6*1 + 2x/1<12/1 --- >   2+ 2x < 12 --- >   x <5

Effectively, there is a little 'bubble' around Station X+1 that it is faster to use transit to access, and that bubble gets bigger in proportion to the ratio between the two. However, that bubble does not grow in proportion to the increase in speed. 

      3             2*W
      4             3*W
      5             6*W

Rare is the urban transit that is 6 times the speed of walking. My intuition tells me that most urban rail* is about 3 times the speed of walking, on city street. For any location less than half way, it is always faster to walk. For any location less than 2/3, it is almost certainly faster to walk.

To some extent, this explains the value of streetcars as economic development tools. As 'slow' transit (2.x walking speed), the 'bubble' at the end is very small, so it's very rational to get off and walk the full distance between stops, so that all locations along the corridor benefit from the pass-by pedestrian traffic. If economic development is the sole aim of streetcars, going slower (so long as they are faster than walking) doesn't actually hurt. But getting people to ride streetcars in the first place requires them to provide transportation benefit to a degree where waiting for the streetcar and riding the streetcar is faster than walking.
For longer distances (of light rail magnitude), let me double the distance. I'll assume that the distance between Station X and Station X+1 is 24**

24/2*1 + 2x/1<24/1 --- >   6+ 2x < 24 --- >   x < 9
24/3*1 + 2x/1<24/1 --- >   4+ 2x < 24 --- >   x < 10
24/4*1 + 2x/1<24/1 --- >   3+ 2x < 24 --- >   x < 10.5
24/6*1 + 2x/1<24/1 --- >   2+ 2x < 24 --- >   x < 11
Bigger distances, bigger transit accessible 'bubble' at Station X+1, (all else equal). Makes less sense to make the full walk. This suggests more frequent stops, more economic development, as more pedestrians walk along the transit corridor. The historic form of retail during the transit-centric streetcar age (thin, deep stores) confirms this. Access to street frontage is what matters. Tempting to make really long stop spacing, but there is a distance decay on how far people are willing to walk. Pretty sure their is a way to compare this to walk-trip distance decay (1/x^2) to determine optimum stop spacing for accessibility.

I am not sure how walking ALONG transit compares to TOD access in terms of overall accessibility provided. We habitually calculate 'area accessible to transit' using PI*r^2. When we actually consider how people travel in an urban environment (along street frontages), travel along*** a transit corridor is where the action is. More unique urban land is exposed to a pedestrian (extracted from their car) walking along a transit corridor, than to a pedestrian walking to a destination in proximity to transit station (as there is no return trip along the corridor). Note I do not say the pedestrian has more accessibility (as per our discussions of accessibility isochrones around stations), as each station imposes a time-cost that reduces the average travel speed on transit, and thus the area accessible from the transit network. 

*Headway doesn't enter into it, as all option include boarding and disembarking once.
**This could probably be turned into a graphic like the attached.
***This suggests that a retail arcade between two stations (even if only along 1 side of the tracks) would do rather well, if placed between two closely spaced stations.