California Partners for Advanced Transportation Technology

A study was conducted to verify C. F. Daganzo's behavioral theory of multi-lane traffic flow (1, 2). This study was conducted by teams from San Diego State University and the University of California at Berkeley who worked independently on a series of case studies to test predictions derived from the theory. The results of the study suggest that some of the phenomena predicted by Daganzo do occur, but not at all locations, and that the underlying behavioral assumptions are oversimplified. Specifically, the types of flow- density (or flow-occupancy) relationship assumed by Daganzo were found to occur at some sites but not others; semi-congested states and fast waves between semi-congested and fully-congested states, as predicted by Daganzo were observed at one site; an increase in average time gaps indicating a "loss of motivation" assumed by Daganzo was observed at one site but not at others; speeds were found not to be equalized among lanes in congested flow, contrary to Daganzo.s assumption and most past literature; redistribution of flow among lanes was observed at flow breakdown despite the absence of speed equalization, contrary to Daganzo's behavioral assumptions; and distinct capacity and discharge flow states predicted by Daganzo were not observed downstream from queues. Observations not directly related to the test of Daganzo.s theory included details of lane-by-lane speed behavior in congested flow, a case in which the location of the point of maximum density upstream from a bottleneck may have influenced its capacity, and observations of general characteristics of incident recovery flow.