Wow. That's disingenuous at best. What would the equivalent be for trains? Any station with greater than 3 minute headways is a C, and six minute headways is an F? Good post, Dan.

by Dave Murphy on Dec 2, 2011 3:40 pm • link • report

This is interesting, but is this for "separated" auto-traffic only highways? Frankly, we don't want our urban residential arterials to be "free-flowing".

by oboe on Dec 2, 2011 4:04 pm • link • report

I'd seriously doubt that the grading is based exclusively on speed. There's likely a probability incorporated that is attempting to highlight the risk that traffic will get worse. I take as evidence the notation in the chart on "Light to Moderate", "Heavy", "Congested", and "Severe". As the number of cars increases, so does the likelihood that an accident will occur, as well as the impact that such an accident will have on congestion.

by Loren on Dec 2, 2011 4:12 pm • link • report

Hmm... needs more asphalt.

Don't you just love playing rigged games?

by Cavan on Dec 2, 2011 4:13 pm • link • report

Loren, LoS is based on speed. It's as simple as Dan describes. It's really, really simple. It has nothing to do with quality of life or human perceptions. It is only an outdated traffic metric to justify decisions that were already made to cover more land with asphalt.

by Cavan on Dec 2, 2011 4:15 pm • link • report

What would the equivalent be for trains? Any station with greater than 3 minute headways is a C, and six minute headways is an F?

The better way to grade transpo infrastructure is something closer to their ROI. An expensive highway or train that is little used is an "F" in my book because it's a big waste of money. An "A" grade for a highway should be 2000/cars per lane per hour, which is the max rate of flow. If the highway was owned by a private business, I guarantee that's how they would think of it. When Disney World has really long lines for their rides, they don't consider that failing, they call it success.

by Falls Church on Dec 2, 2011 4:22 pm • link • report

Of course, this is for highways. On local streets, I believe LoS is measured by wait at intersections.

The biggest fail is that LoS is for cars only. Want to get LoS from F to a C? Then limit crosswalk time heavily. If a pedestrian has to wait 2 minutes to cross, who cares, its not measured. But 35 seconds for a car? oooooooooooog

by JJJJJ on Dec 2, 2011 4:24 pm • link • report

And back of the envelope calculation tells us that any highway with an LOS above D is a money loser - the taxes generated by the traffic in the road aren't enough to pay for its construction and maintenance.

by egk on Dec 2, 2011 4:35 pm • link • report

egk- assuming that the people's time and trips are all for worthless activities, like commerce... :)

by Douglas Willinger on Dec 2, 2011 4:44 pm • link • report

@JJJJJ is right, LOS on streets in based on the idea of "control delay" i.e. time lost slow or stopped at a traffic control: stop, yield (for roundabouts) or signal.

There is a lot of work going on regarding multimodal level of service that include pedestrians, bicycles and transit. See NCHRP Report 616 http://onlinepubs.trb.org/onlinepubs/nchrp/nchrp_rpt_616.pdf

And the 2010 edition of the Highway Capacity Manual which has a newly added section on multi-modal urban streets methodology. This section is being expanded further in the current research work to update this chapter for the next edition as well.

What I am saying is that the research and resources are now available. Will the agencies adopt them as a matter of practice in evaluating transportation projects?

by Some Ideas on Dec 2, 2011 4:51 pm • link • report

Actually, there are different standards for different types of roads and for intersections. In particular, the intersection standards record seconds of delay during peak periods/hours. I believe "F" kicks in around 55-60 seconds. Most urbanized area (Tysons?) intersections are routinely "F" rated; the question is how bad an "F" it is. Some NOVA intersections reach 180+ seconds delay in the peak hour!

by Lilguy on Dec 2, 2011 4:52 pm • link • report

@Cavan/Loren:

Level-of-Service is actually based on much more than just speed. For example, the primary determining measure for freeway segments is vehicle density, not speed. Speed plays a factor, but it's by far not the only factor. And as others have mentioned, for intersections it's based on delay.

by Froggie on Dec 2, 2011 5:44 pm • link • report

In any event, the game is still rigged, since roads in the "A" and "B" range are clearly overbuilt. "C" would be breakeven and--if you were a toll road operator--"D" and "F" would be making money, since far more traffic would be using the road (and hence revenue being collected) than what it would take to maintain it.

by Steve S. on Dec 2, 2011 6:37 pm • link • report

@Froggie is pretty much right. Freeway LOS is based primarily on vehicle density, with traffic over 45 passenger cars per mile per lane rated as LOS F. LOS C, for comparison, is up to 26 pc/mi/ln. As you can imagine, density does correlate with speed, but even at LOS E, vehicles are still moving at 50mph or more.

LOS for signalized intersections are based on completely different criteria. The letter grade is assigned based on average signal delay experienced per vehicle. Basically, at signalized intersections, if the average delay per vehicle is greater than 80 seconds it is rated at LOS F. LOS C is anywhere from 20 to 35 seconds of delay per vehicle.

One important thing I would point out is that the same freeway segment (or signalized intersection) can be rated at LOS F at, say 4:30 p.m. on a weekday but be at LOS C at 8:30 p.m. and LOS A at 3 a.m. and LOS B at 2p.m. on a Sunday. Unfortunately, most traffic studies are undertaken by first determining the peak (worst) morning hour and evening hour and ONLY evaluating those two hours. Sometimes, a midday peak hour or a Saturday peak hour is included depending on the situation (Saturday for a heavy retail area might be worse than any weekday peaks). Basically, nearly every traffic study done is basing its conclusions on studying peak hours, even if everything operates at acceptable LOS the other 22 hours of the day.

And yeah, this is only for vehicles. There are some LOS standards for pedestrians, but they aren't considered well-defined or particularly useful in the traffic engineering world form what I've seen. I do hope to see seriously upgraded methods for this soon as they really need to be quantified, in my opinion.

by Thom on Dec 2, 2011 8:13 pm • link • report

I found it interesting that they could not figure out why traffic was improved on the Dulles Toll Road from 2008 to 2011. Its going to improve a whole lot more in the coming years.....

by mcs on Dec 2, 2011 10:13 pm • link • report

I found this article to be misinformed and simplistic. It would not be a reach to suggest there was an agenda by writing this, but it was not an obvious one.

by Anon on Dec 3, 2011 1:47 pm • link • report

Care to explain what the article left out about the level of service, Anon? Or are we going to settle for the "not an obvious agenda" agenda?

by DAK4Blizzard on Dec 3, 2011 4:04 pm • link • report

Dan's right: LOS D/E is the sweet spot for highways, with both speeds staying high and enough demand to achieve high "throughput", but there is considerable risk of failure. For less-than-high achievers such as me, the correlation with grades in school rings true. For a tough course, it wasn't worth the additional effort to get an A or a B, while just getting by was very satisfactory but courted a significant risk of failing. A dilemma for highway planners, looking out to 2040, is that aiming for (say) LOS C/D could easily result in LOS A or B (overbuilt) or LOS F at which travel times can double. Despite years of effort to improve travel forecasting, long-range forecasts are often 20+% off.

by mark on Dec 5, 2011 10:39 am • link • report

The worst thing about LOS metrics is that they focus on local impacts and ignore regional ones. So an infill development will push all nearby intersections to LOS F and be penalized or forced to downsize, if it gets approved at all. In contrast, a greenfield sprawl-burb has no LOS impacts because nobody else is out there, even though it generates long-distance commuting that fills up roads and streets for miles. We create regional traffic impacts by trying to prevent or mitigate local ones in urban areas.

by Alice on Dec 8, 2011 2:06 pm • link • report