Congestion: Average Travel Time vs. Reliability

I had the opportunity to attend a “Travel Time Reliability” workshop at Metro a few weeks ago (I never said I wasn’t a wonk). The speaker was kind enough to forward the PowerPoint since I wanted to blog about a few of the slides.

My point here is that when most people think about congestion, they think in terms of capacity constraints. But the reality is that the pain has to do more with unreliability. Not that the two aren’t related – more capacity does make you less sensitive to SOME of the factors that create reliability problems, but it’s an expensive answer and there may be more cost effective approaches (with way fewer negative environmental effects than adding capacity).


Here’s the psychology, and the problem with the way we usually measure now.

More after the break.


Managing the system DOES improve performance. In this famous experience in Minnesota, ramp meters were turned off for a period (the ‘before’ segment in the slide). The result was longer after travel time AND less reliability.


I think this is the critical concept out of the discussion. The pain of congestion is not the difference between the free-flowing travel time (11 min in this slide) and the average travel time (+4 min). The pain is the extra 7 minutes of buffer you have to add to make sure you’re not late more than once in every 20 trips (95% case = 1 day in 20). I suspect we’d have to put $Billions in to work on the 4 minutes, but we can probably spend $Millions and work on the 7 minutes!

5 Responses to Congestion: Average Travel Time vs. Reliability