How do macroscopic traffic flow parameters affect time spent in conflict on freeways? A comprehensive analysis using...

Time spent in conflict (TSC) represents the total duration experienced by all vehicles involved in conflicts, serving as an important parameter in freeway safety assessment. Accurately analyzing the relationship between macroscopic traffic flow parameters and TSC is key to achieving effective safety assessment. This study aims to model TSC, accounting for data censoring and heterogeneity, and examine how macroscopic traffic flow parameters influence TSC.

Hazard-based duration models, including the Kaplan-Meier (K-M) model and the random effect Accelerated Failure Time (REAFT) model, were proposed to fit TSC. The K-M model analyzes the temporal and spatial heterogeneity of TSC, while the REAFT model is employed to fit the TSC distribution and to determine how macroscopic traffic flow parameters and their interactions impact TSC.

The K-M model results indicate that TSC exhibits temporal heterogeneity on months. The random effect Weibull AFT model is identified as the best fit for TSC, showing that the probability of traffic conflicts ending increases with longer TSC. Furthermore, this model quantifies the impact of macroscopic traffic flow parameters on TSC. Among these, density has the greatest influence on TSC, with an effect magnitude of 24.73%, while flow has a relatively minor impact, with an effect magnitude of only -0.6%.

This study accounts for the censoring and heterogeneity of TSC and reveals how macroscopic traffic flow parameters influence TSC, providing a theoretical foundation for freeway safety assessments. Furthermore, the findings support the development of advanced freeway safety assessment systems, enabling accurate crash risk predictions based on macroscopic traffic flow parameters.