LOS and Travel Projections: The Wrong Tools for Planning Our Streets

Gary Toth is director of transportation initiatives with the Project for Public Spaces. This post first appeared on PPS’s Placemaking Blog.

Would you use a rototiller to get rid of weeds in a flowerbed? Of course not. You might solve your immediate goal of uprooting the weeds — but oh, my, the collateral damage that you would do.

Yet when we try to eliminate congestion from our urban areas by using decades-old traffic engineering measures and models, we are essentially using a rototiller in a flowerbed. And it’s time to acknowledge that the collateral damage has been too great.

Image: Andy Singer
Image: Andy Singer

First, an explanation of what I call the “deadly duo”: travel projection models and Levels of Service (LOS) performance metrics.Travel projection models are computer programs that use assumptions about future growth in population, employment, and recreation to estimate how many new cars will be on roads 20 or 30 years into the future.

Models range from quite simplistic to incredibly complex and expensive. Simple models deal primarily with coarse movements of vehicles between cities, while complex models deal with the intricacies of what happens on the fine grid of urban areas. To be truly accurate, growth projection modeling can be expensive. Therefore, absent compelling reason to do otherwise, most growth projections tend to be done using less expensive techniques, which usually lead to overestimates.

Levels of Service (LOS) is a performance metric which flourished during the interstate- and freeway-building era that went from the 1950s to the 1990s. Using a scale of A to F, LOS attempts to create an objective formula to answer a subjective question: How much congestion are we willing to tolerate? As in grade school, “F” is a failing grade and “A” is perfect.

Engineers decided that LOS “C” was a good balance between overinvestment in perfection and underinvestment leading to congestion. In urban areas, a concession was made to accept LOS D, representing slightly more restricted but still free-flowing traffic. LOS is commonly (actually, almost always) calculated using travel projections for 20 to 30 years into the future.

Using basic traffic models and LOS C/D to plan and design the interstate system was a no-brainer in the 1950s, ’60s and ’70s. When deciding how many lanes to build on a freeway connecting major cities, a sensitivity of plus or minus 10,000 trips a day could be tolerated, and the incremental difference in cost to plow through undeveloped land was relatively insignificant.

Good approach, wrong setting

I’m not going to look back and quibble with the general philosophy of how the interstates and the associated high-speed freeways were planned and designed. On many levels, the approach made sense.

But it became increasingly less persuasive when applied to the rest of our road network. Unlike interstates and freeways, most roads exist not just to move traffic through the area, but also to serve the homes, businesses, and people along them. Yet in search of high LOS rankings, transportation professionals have widened streets, added lanes, removed on-street parking, limited crosswalks, and deployed other inappropriate strategies. In ridding our communities of the weeds of congestion, we have also pulled out the very plants that made our “gardens” worthwhile in the first place.

It’s worth remembering, too, that not all congestion is bad. John Norquist, former Mayor of Milwaukee and current CEO and President of the Congress for New Urbanism, suggests that congestion is like cholesterol: there is a good kind and a bad kind.

What makes the prevailing situation even more troubling is that there are no comprehensive requirements dictating the use of either LOS or travel modeling in transportation planning and project design. The “Green Book” from the Association of American State Highway and Transportation Officials (AASHTO) (more formally known as “A Policy on Geometric Design of Highways and Streets”) clearly states that these are guidelines to be applied with judgment — not mandates. So does the Federal Highway Administration’s “Highway Capacity Manual.”

The idea that we must rid our roads of  any and all traffic congestion is, in fact, a self-imposed requirement. As Eric Jaffe wrote in an article for Atlantic Cities in December, 2011:

Although cities aren’t required to abide LOS measures by law, over the years the measure hardened into convention. By the time cities recognized the need for balanced transportation systems, LOS was entrenched in the street engineering canon.

Worse yet, many designers size a road or intersection to be free-flowing for the worst hour of the day. Sized to accommodate cars during the highest peak hour, such streets will be “overdesigned” for the other 23 hours of the day and will always function poorly for the surrounding community.

If that isn’t troubling enough, LOS is often calculated using traffic predicted 20 years into the future, even in urban settings. Until the forecasted growth materializes, the roadway will be overdesigned, even during the peak hour. Overdesigned roadways encourage motorists to drive at higher speeds, making them difficult to cross and unpleasant to walk along. This degrades public spaces between the edges of the road and the adjacent buildings, encourages people to drive short distances, and generally unravels a community’s social fabric.

Let me repeat: Contrary to what you may hear, there is no national requirement or mandate to apply LOS standards and targets 20 years into the future for urban streets. This thinking is a remnant from 1960s era  policy for the interstate system, and has erroneously been passed down from generation to generation.

Image: Andy Singer

So what are the right approaches?

Asking the simple question, “Do you want congestion reduced at a particular location?” is a question out of context. It’s like asking you whether you want to never be stung by a bee again. Of course, the answer will be yes. But what if I told you that to in order to never suffer a sting again, every plant within a several mile radius would have to be destroyed — and that you could never leave the area of destruction?

You would have a completely different answer, I’m sure.

The question that needs to be asked in urban settings is not whether you ever want to sit in congestion again. Who does? The question is whether you want to eliminate congestion on your Main Street 24 hours a day, 7 days a week, 365 days a year — knowing that the consequence would be a community with decimated economic and social value, increased reliance on car use, increased crashes, and, ultimately, more congestion.

Recognizing the need for balance, a number of entities are beginning to promote approaches sensitive to the context.

I was the New Jersey Department of Transportation’ s project manager for  the “Smart Transportation Guide” (STG), adopted jointly by the state DOTs in Pennsylvania and New Jersey.   The STG directs DOT designers to consider the tradeoffs between vehicular LOS and “local service.” It goes on to say that if the street in question is not critical to regional movement, that LOS E or F could be acceptable — and that designers may actually need to design to slow down cars.

The Institute of Transportation Engineers, an “international association of transportation professionals responsible for meeting mobility and safety needs” also promoted this concept in its landmark “Context Sensitive Solutions Guidelines for Urban Thoroughfares.” Florida DOT has adopted multimodal LOS standards, and cities like Charlotte, N.C., have elevated pedestrian and bicycle LOS to the level of that for automobiles. We have a long way to go, but the door is opening.

Creating balanced standards for roadway design will benefit transportation as well. In the Netherlands, the “Livable Streets” policy led to a remarkable improvement in safety on their roadways. They started in the 1970s with a crash rate 15 percent higher than in the U.S., and now have a crash rate 60 percent lower.

Design with the community in mind

It’s time for communities and transportation professionals alike to accept that we have been using the wrong tools for the wrong job. LOS and travel modeling may be effective when sizing and locating high-speed freeways, but are totally inappropriate in every other setting. If travel modeling with high rates of growth is used to make street decisions, your community may be doomed to a series of roadway widenings or intersection expansions. If vehicular LOS C or D performance measures are adopted as non-negotiable targets, major road construction will be heading your way.

Village, suburban and city streets need to be designed with the community in mind using the PPS principle of Streets as Places to  create a vision for a great community and then plan your streets to support that vision.

Lets not be fooled by the appearance of science behind Levels of Service and Traffic Modeling. As I pointed out in an interview with Wayne Senville that was published in the November 2010 “Planning Commissioner’s Journal,” LOS standards are easy to understand — and that’s exactly what makes them so dangerous.

14 thoughts on LOS and Travel Projections: The Wrong Tools for Planning Our Streets

  1. One interesting part is that to get optimal LOS, A, the best road is an unused road. Im sure Alaska is packed with roads with a LOS of A.

    One thing this article doesnt go into is that C is still free-flowing. That is, A generally means one is able to exceed the speed limit with no problems.

  2. Your basic understanding of the planning process misses a critical step, and leads you to wildly incorrect conclusions.  Please inform yourself on mode assignment before publishing misleading diatribes.

    Forecasting growth in travel demand is critical for any effort to invest in transit.  Any argument that we throw out transportation planning methods is reckless.  The people who might listen to you are the ones who need these methods to support transit.  People who want roads will push their agenda perfectly well without any substantive planning of any type!

  3. @40daebbed12b53745f7f9f21456e6154:disqus …it seems your argument is with a straw man.  I see nothing here that suggests throwing out anything, and you also do not name the ‘critical step’ you claim is missing in the process described here.
    The article explicitly addresses forecasting, and advocates for a broader context within which forecasting should occur, one which includes more than just automobile congestion and one which would put transit and pedestrian and bicycle planning in the mix.

  4. This stuff has to be taught in undergrad and graduate civil engineering programs. Once engineers are in the workforce, everything they’ve been taught up to that point comes down to “LOS A=Good, LOS F=Bad, maximize auto capacity at all costs”. A bunch of planners in the room telling engineers that their models are irrelevant is likely to go over like a roasted pig at a vegetarian convention.

  5. Coming from engineering school, they taught LOS as is THE criteria in determining designing for capacity.  It’s easy to understand but at the end, there are flaws with the approach, just like any other ones anyone can think of.  Of course, LOS being so auto-centric does not help the cause of other “alternative” transportation modes.

  6. Matthew Jones, please try reading my comment again.  Perhaps try reading up on some basic transportation planning as well.

    He is omitting the mode choice model step in the process.  You know, the part that shows that when roads get congested, more people take transit?!

    A little dose of understanding would help a lot here.  There is nothing wrong with demand forecasting or level of service analysis; it’s all a matter of how they are applied and the political choices that are made. 

    Railing against basic transportation planning methods is not productive.

  7. The big flaw with planning based on LOS is the implicit assumption that the level of auto traffic is immutable. Plenty of studies have shown that if there is no parking, people will drive less. If there is good public transit, people will drive less. If biking is safe and fast, people will drive less. If driving is made to cost more, people will drive less. This all begs the question why we’re not actively taking steps to decrease driving, preferably to the point where few people use or own private automobiles, especially in urban settings. In the end, the costs associated with a transportation system based on the personal automobile are huge. Despite these costs, the autocentric system very often fails on every metric compared to the alternatives. A system entirely based on alternatives like rail or bicycle can easily offer faster door-to-door travel times, and with far less use of valuable public space. This in turn will free up the now much narrower roads for their primary functions-delivery of goods/services, and emergency vehicles.

    I’ll agree with the basic premise of LOS planning, namely that we should never tolerate the longer than optimal travel times associated with congestion. However, we can only reach this lofty goal if we move most users to modes with inherently higher capacity, plus uniform performance characteristics to allow reliable scheduling. This almost always means some type of rail vehicle running on grade-separated right-of-way.

  8. @40daebbed12b53745f7f9f21456e6154:disqus There is no evidence that local DOTs bother with mode choice modeling when they do future planning. If they did, then they would uniformly opt for rail instead of road for all future expansion if LOS and capacity increases are the primary goals. It’s far easier to reach these goals with rail without disrupting communities. When you think about it, roads should really only exist for emergency vehicles, delivery vehicles, bicycles, and buses. Everything a private auto does can be done better by something else nearly all of the time. Over urban distances bike, subway, bus, or light rail are superior. Over suburban distances commuter rail is better. And over intercity distances HSR is far superior to driving. Really, there is no good reason to accommodate travel by private auto to the extent we do now. The “don’t build it and they won’t come” philosophy should be applied instead. The only place private autos should be accommodated is places like Alaska which are too sparse to support any other mode (and which can support LOS A with a two-lane country road).

  9. I just don’t see any ‘railing against basic transportation planning methods’ going on in this post, nor in my comment about @40daebbed12b53745f7f9f21456e6154:disqus ‘s straw man — I’m sorry, but it’s not clear which is intended.  I think Toth has included several spots where mode choice is part and parcel (pun intended) of the entire process.  And I still don’t see how the discussion of LOS in this post excludes the idea of mode choice.  I’m no traffic engineer, so feel free to point me to sources on mode choice.  I’m primarily familiar with a bit of Downs regarding convergence, and I would welcome suggestions.  Please, don’t condescend when your argument is questioned.  I’m pushing for explanation and speaking plainly; that ought not offend.

    “There is nothing wrong with demand forecasting or level of service analysis; it’s all a matter of how they are applied and the political choices that are made” — this description of what you seek seems like the exact spot where Toth begins, and then he goes further to get at a few of the implications of designing for place rather than simply the automobile.

Leave a Reply

Your email address will not be published. Required fields are marked *


The Projections Fallacy

Cross-posted from Streets.MN. We spend billions every year in this country on our transportation network, large percentages of it based on traffic projections. This despite the fact that we have a long record of not being able to accurately project traffic. The answer isn’t better projections but a better transportation system, one that is robust […]

Court: Don’t Spend Billions on Outdated Travel Forecasts

Cross-posted from City Observatory.  Last week, the Washington Post reported that the U.S. District Court in Washington, D.C., has ordered new ridership projections for the proposed Purple Line light rail line, which will connect a series of Maryland suburbs. Like any multi-billion dollar project that serves a densely settled metropolitan area—and this one connects some […]

Revisiting the Peak Car Debate

Cross-posted from the Frontier Group. I’ve never liked the term “peak car.” First, it was always unclear exactly what was supposed to be peaking – total vehicle travel, per-capita travel, car ownership, or all of the above? Second, like peak oil before it, “peak car” applies a catchy name to a collection of concepts that […]

The Great Traffic Projection Swindle

This is the final piece in a three-part series about privately-financed roads. In the first two parts of this series, we looked at the Indiana Toll Road as an example of the growth in privately financed highways, and how financial firms can turn these assets into profits, even if the road itself is a big money loser. […]