Implementing survival analysis to capture stochastic characteristics of saturation flow rate considering the impacts of adverse road-weather conditions

Abstract

Saturation flow rate (SFR) variations were analyzed using the video data collected at a signalized intersection in Winnipeg, Canada, to investigate the implications of adverse road-weather (RW) conditions for SFR distributions and characteristics. Survival analysis was implemented to develop stochastic SFR distribution functions considering censored data and a statistical analysis method was developed for determining the optimal critical vehicle (CV) for measurement of saturation headway. The analysis findings suggest that adverse RW conditions decrease SFR significantly and moves CV to the front of the queue while having little impact on SFR considering heavy vehicles. Furthermore, the findings imply that the conventional SFR estimation method overestimates the probability of saturation at a given flow rate. The proposed analysis method reveals stochastic characteristics of SFR and provides a method to estimate SFR distributions under different RW conditions, which is essential for improving the operation of signalized intersections, particularly in cold regions.