Survival Analysis of the Likelihood and Duration of Traffic Flow Breakdown at Freeway Merge Bottlenecks

Abstract

Studying traffic flow breakdown and its duration is critical for freeway operation efficiency. In this paper, a breakdown probability model and a breakdown duration model were established based on the survival analysis method, respectively. The product limit method was used for the breakdown probability model, where the breakdown likelihood was calculated as functions of both mainline flow and the combination of mainline and ramp flow. We also consider the impact of different ranges of ramp flow and truck proportion. The log-rank test results showed that the group with the lowest ramp flow (less than 132 veh/h) affected the breakdown probability significantly differently from the other three groups. However, there was no significant difference in the impact of various ranges of truck proportion. A Cox PH model was constructed to reveal the hazardous variables that influence breakdown duration. The increase in average upstream volume during the breakdown episode, the standard deviation of the downstream occupancy, and the difference of upstream speed will reduce the breakdown duration. While the difference in occupancy and speed between upstream and downstream will prolong the duration. The results of this paper have guiding significance for alleviating the congestion caused by traffic flow breakdown.