Modeling of Time Headway Distribution on Suburban Arterial: Case Study from South Korea

Time headway is of fundamental importance in traffic engineering. For any of the traffic simulations to address traffic problems to be effective, accurate vehicle generation is essential. Previous research into this subject has focused on stochastic modeling for uninterrupted facilities. Yet little researches for an interrupted facility have been performed. This paper proposed theoretical headway models for a suburban arterial in Korea. Using the laser sensor-based traffic detector, a large amount of accurate headway data was obtained. To analyze by traffic flow states, the headway data were divided into five flow states. Subsequent runs test rejected the randomness for the 5-9 v/m flow. Hence, theoretical modeling for the remaining flow (10-14 v/m, 15-19 v/m, 20-24 v/m, and 25-29 v/m) is performed. The goodness of fit tests by Kolmogorov Smirnov (KS) statistic revealed that the Johnson SB model for 10-14 v/m flow, the Johnson SU model for the other three flows fitted best. The Log-Logistic model was second-best for 20-24 and 25-29 v/m flow. The Log-normal model was also accepted for 25-29 v/m flow. Some analyses not only on the statistics of the headway, but the fitted model exhibited somewhat different characteristics compared to those of the headway from uninterrupted facilities. This paper can provide better understanding of the headway from interrupted facilities and starting point for developing a well-descriptive simulation model for signalized arterials.