Friday, July 27, 2012

Conservative traffic studies are radical

I've been writing on the topic of traffic and models for a bit now, and the reason it's on my mind these days is the New Brighton Landing project which is being designed by New Balance. The Planned Development Area materials include an Exhibit E - Transportation Study, which has flagged my attention. New Brighton Landing will be built in three mostly vacant parcels on the edge of Allston and Brighton near Everett Street, on the south side of the Mass Pike. This spot is about half a mile from Allston's Union Square, which also happens to lie on the path to the Mass Pike ramps in Allston.

Union Square is a fairly active neighborhood, with the Jackson-Mann school on one side, the fire station on another, and a diverse set of restaurants, bars and other small shops scattered around the other edges. The key bus routes 57 and 66 connect here with the less frequent 64, making it a transit node. Union Square marks the intersection of Cambridge Street with the end of Brighton Avenue, and the beginning of North Beacon Street.

In an earlier iteration of the plans, there were plans afoot to widen Cambridge Street here from two to four travel lanes, and they were based upon projected volumes generated by the new development. For now, it seems like community input has caused them to shelve those plans, though nothing is certain.

What led them to consider such measures is the projected level-of-service for the intersection, after they applied certain traffic growth factors. The study is founded on the assumption that traffic levels grow 0.5% per year, every year. This assumption is baked into several places, e.g. page 17:

Based on discussion with the BTD, the study team developed Year 2012 traffic volumes for the New Brighton Landing’s transportation analysis by adopting the Year 2007 traffic data from the Lowe’s work and increasing the volumes by an annual growth factor. 
[...] The annual factor of 0.5% was applied to the Year 2007 counts to produce existing condition volumes for weekday a.m. peak hour (8-9 a.m.), weekday p.m. peak hour (5-6 p.m.) and Saturday Midday Peak hour (12:45-1:45 p.m.).

The trouble with this assumption is that regional and national trends have shown a peak in vehicle miles traveled that occurred sometime between 2005-2007, and a reduction ever since. As it happens, MassDOT offers an interactive map interface which can be used to query some of their actual traffic data. So I looked at locations around Union Square and Allston to see what historical data was available. As it happened, four appropriate readings were available, each having entries for 2002 and 2010, in terms of Annual Average Daily Traffic (AADT).

Location20022010Percent Change
Cambridge Street near Linden Street40,00037,797-5.5%
Cambridge Street near Union Square25,40023,971-5.6%
Brighton Avenue near Park Vale Avenue30,20028,547-5.4%
Harvard Avenue near Farrington Avenue17,20016,213-5.7%

It turns out that in the vicinity of Union Square, not only has traffic not increased by 0.5% a year, it has actually decreased by almost 0.7% per year. The following chart sums of the difference between the actual differences in traffic vs what the glib projections would have guessed.

Actual traffic levels decreased from 2002 to 2010.
The difference is nearly 10% of the total for all four and, crucially, straddles the zero line. That the BTD was going to consider roadway expansion in this area, on the basis of projections that were diametrically wrong, is extremely disturbing. The reasons behind the drop in vehicles are not clear. It could be the near depression, and perhaps counts will skyrocket again once prosperity returns. The problem with this hypothesis is that the national trend of downwards VMT was already being noted and written about prior to the financial crash.
While total driving in both rural and urban areas grew between January 1991 and September 2008, rural and urban VMT have been declining since 2004 and 2007, respectively. Amongst these collective driving declines, the nation shifted more of its VMT share to larger capacity, urban roadways.
And, notably, the data collector on the nearby Mass Pike indicates an upward slope over the same years that Union Square has a downward trend. So it seems more likely that there is a serious change in the way we are using these local roads.

The study claims that the New Brighton Landing development will bring additional traffic. They estimate 12,000 vehicle entries and exits a day. Most of that will be for the office space, some for the restaurants and retail, and also for the hotel being planned. The other major draw is the new sports complex, but presumably that will have its peaks at different times. The numbers are based on the Institute of Transportation Engineers' Traffic Generation handbook, which provides pat numbers for everything but the sports complex. While it is standard procedure to use such numbers, it seems wrong to base such estimates on studies generated outside of Boston, and outside of Massachusetts, in a completely different environment. Even worse, without any data, the study assumes that 100% of sports complex related trips will be made by automobile. Page 43:
To be conservative (highest impact), however, all ice rink and track and field facility trips have been assumed to arrive/depart by vehicle.
The chronic problem of traffic overestimation I debunked above is also referred to as a "conservative" approach. But the only "conservative" aspect is that it would lead to the destruction of the environment in the neighborhood. Such overestimation is actually quite radical if it leads to destructive road-widening.

A truly conservative approach would seek to mitigate traffic and preserve the quality of life in the neighborhood. That is why Transportation Demand Management, as deployed in Kendall Square, is likely the best way forward. Luckily, it seems that New Balance is on-board with this approach. They already sponsor Hubway, and are providing the funds for the new Commuter Rail station. They are also pouring a lot of resources into making the development area more friendly to walking, bicycling, and safe driving by fixing the decaying infrastructure and creating new, smaller blocks with street level retail.

Ultimately, I'm really happy about the project, and I'm glad they have decided to invest $500 million into the neighborhood. I just did not want to let the fallacies embedded in the transportation study slip by unnoticed, as they are pervasive problems that will likely need to be corrected in other, future plans as well.


  1. Traffic has declined for a number of reasons, starting with higher fuel prices, unemployment, fewer retail trips and other local spending reductions. Mass residents are sending money and jobs out of state to buy goods on the web instead of fighting traffic and traffic lights to spend at a local store and pay state sales tax. We need only continue to make travel harder to feed that trend. Buying will be as over 100 years ago with the Sears catalog because delivery is easier for the majority unable to afford horses or cars.

    What does the future bring? New natural gas discoveries should produce a glut. That and its cleaner burning will spur conversion from oil, which in turn takes pressure off of fuel prices. Increased vehicle efficiency will make drivers less sensitive to prices. With so many gas guzzlers on the road, Americans don't seem all that sensitive to price. They are more sensitive to lost and lowered income, cutting back and delaying on buying products and services. This recession is really bad. In previous recessions VMT never went down, rarely even staying flat.

    Traffic = economic activity. Let's hope there is a recovery.

    1. I hope for a faster recovery too, but I feel this one is different with regard to VMT. It isn't just the recession: overall VMT peaked in 2005-2006 -- well before the crash. And while vehicle counts dropped on some local streets, the counts on the nearby portion of I-90 have steadily increased, even throughout the recession. MBTA ridership numbers have reached new record highs as well. This is consistent with a mode shift towards walking for local trips.

      I'm not sure we can turn back the clock and bring retail back to its state several decades ago, nor should we try. If delivery is the future, so be it. We will adapt. There are plenty of businesses with services that cannot be delivered.

    2. I think the recession started before the official date. I think house prices/volume was slowing before the official recession. What had gone up for sure was gas prices.

      VMT is distorted by being mostly Interstate miles, and urban ones at that, so we get less information about how local spending and servicing is going. I feel a major impediment to local spending is how much more time consuming to travel roads have been made, along with producing more pollution from all the stop and go now. People don't want to leave home as much, watch movies at home instead of theaters, and have human interactions on the computer instead of face to face.