Analysis of Time Headway Distribution on Korean Multilane Highway Using Loop Event Data

Abstract

Vehicular time headway is of fundamental importance in traffic engineering. This research investigated theoretical headway models for a multilane highway in Korea using loop detector data. To analyze time headway by traffic flow states, the collected data were categorized into five flow rates and tested for randomness. Subsequent theoretical modeling was performed at the five flow rates. The Gamma distribution at low to intermediate flow rates, and the Pearson VI distribution at all flow rates were fitted by the Kolmogorov-Smirnov (K-S) statistic. Model parameters were obtained from conventional maximum likelihood estimation (MLE). Analyses were performed on residuals and parameters of the fitted models, as well as various headway statistics. Notably, parameters of the Pearson VI changed markedly during level of service (LOS) C, which presumably reflects the boundary between low and high traffic flows.