University of California Transportation Center

This paper presents the most complete picture yet of moving bottlenecks on freeways, including experimental observations and a theory. The experimental observations include the “fingerprint” of a moving bottleneck on a series of loop detectors, and a set of controlled experiments in which moving bottlenecks were artificially introduced in the traffic stream. The paper also contrasts this evidence with current theories and describes a new one that is consistent with the data.

High-resolution oblique plots of loop detector data from freeway I-880 in Oakland (California) are used to analyze the aforementioned fingerprint. They clearly display the presence of the bottleneck and its evolution in time and space, including the precise location in space-time where it appeared. The data also reveal a fleeting but real change in the drivers’ car-following attitude shortly after the bottleneck’s appearance.

The controlled experiments reveal that the flow downstream of the bottleneck increases with the speed of the bottleneck when the bottleneck holds back a queue—in contradiction with two previous theories (Gazis and Herman, 1992, and Newell, 1993).

The new theory includes these as special cases. It treats the moving bottleneck as a boundary condition that can be integrated with kinematic wave (KW) theory and also with variants of this theory that account for multiple vehicle types and changes in driver psychology. The empirical evidence suggests that the lengths of queues upstream of moving bottlenecks and the ensuing vehicle delays can now be predicted with good accuracy.