Capacity drop at active bottlenecks: An empirical study based on trajectory data

Abstract

Capacity drop, a traffic phenomenon indicating that the discharge flow from a queue is lower than the pre-queue flow, is commonly observed at freeway bottlenecks. In the literature, the majority of empirical studies on capacity drop rely on aggregated traffic flow data. To fully understand the mechanism behind capacity drop, it is essential to analyze trajectory data, which captures the microscopic behavior of individual vehicles. However, the availability of high-quality trajectory data covering both sufficient spatial and temporal scope is limited. Consequently, existing theories and mechanisms to explain capacity drop from the perspective of vehicle behavior are predominantly analytical, lacking direct evidence to validate their impacts on contributing to capacity drop. This paper fills this gap by conducting a comprehensive empirical analysis of capacity drop using high-resolution trajectory data extracted from videos recorded by unmanned aerial vehicles. The empirical analysis examines the relative effects of various capacity drop mechanisms and reveals the following findings: (i) The late responses of hesitant vehicles during the acceleration process significantly contribute to capacity drop; (ii) the impact of response delay on queue discharge rate is more pronounced at lower congestion speeds; (iii) response delays primarily result from deceleration during car-following, followed by lane changes, with their combined effect having a more pronounced triggering impact. These findings are subsequently validated using data collected from another site. The findings presented in this paper are valuable for developing more accurate microscopic traffic simulation models and designing more effective traffic management and control strategies.