Using Probe Vehicle data to Understand Bottlenecks and Congestion.

摘要

A bottleneck becomes active whenever demand exceeds capacity. There are many different freeway features that can cause bottlenecks, e.g., grades, lane drops, merge and weaving sections (Banks, 1990). Conventionally an active bottleneck is modeled as if occurring at a discrete point or short distance along the freeway, e.g., as Bertini and Leal (2005) succinctly put it, an active freeway bottleneck is, 'a point on the network upstream of which one finds a queue and downstream of which one finds freely flowing traffic.' Point bottleneck models are widely used and are well established in traffic flow theory. Contrary to such conventional wisdom, in this report we present evidence that modeling the bottleneck mechanism as if it occurs at a single point along the road is too simplistic and we show that the mechanism appears to occur over an extended distance for some bottlenecks. A few previous papers have made a similar argument for various reasons (e.g., Coifman et al., 2003; Chung and Cassidy, 2004; Ogut and Banks, 2005). The present work highlights additional features that are obscured by the single point assumption. We adopt the term 'apparent-point-bottleneck', APB, to specify the location where one would place the point bottleneck model while underscoring our belief that the actual bottleneck mechanism occurs over an extended distance.