Testing Daganzo's Behavioral Theory for Multi-lane Freeway Traffic

This report describes the detailed, albeit still preliminary study of traffic on stretches of two different freeways.  Both were plagued by merge bottlenecks.  The first of these sites is the Gardiner Expressway, a 3.3 km long freeway stretch in Toronto, Canada.  The site was selected because of its suitable geometry (i.e. its merge bottleneck) and its well-tuned loop detectors located upstream and downstream of the bottleneck.  The site thus provided for an exceptionally good “laboratory” for testing Daganzo’s behavior theory of drivers (Daganzo, 1999).  It turns out that the observations from this stretch qualitatively match the theory in a number of important ways, as will be described in this report. The second site is a 1.8 km stretch of westbound Interstate 24 just upstream of the Caldecott Tunnel in Berkeley, California. This site provided a means for verifying Daganzo’s theory for “California  conditions.” It is especially suitable for this study thanks to its very disruptive bottleneck and to its numerous vantage points (i.e., adjacent hillsides) from which to videotape traffic. Four cameras were strategically deployed along this freeway stretch. The detailed traffic data (manually) extracted from these videos were, like the Toronto data, found to be qualitatively consistent with much of Daganzo’s theory.