An Exploratory Hazard-based Analysis of Traffic Accident Impact on Freeway Traffic Flow

Abstract

The traffic congestion can be categorized into two groups regards to its causes

Recurring congestion and Non-Recurring congestion. Of these two types, Non-Recurring congestion occurs more frequently that it occupies 60% of traffic congestion. The main causation of the Non-Recurring congestion is traffic accident. For making efficient plan for preventing congestion caused by the accidents on highway, this research aimed to analyzing impacts of traffic accident on traffic flow. For achieving this goal, cox proportional hazard model has been adopted. Four years of traffic accident records and Vehicle Detectors (VDs) data on the Gyeongbu line, the Seohaean line, the Yeougdong line in Republic of Korea (Korea) are collected for analysis. By using the hazard model, various factors of traffic accident that affects the traffic flow have been identified. If the traffic flow before the accident is stable, the hazard of getting worse level of service(LOS) is lower. Moreover, hazard is increased proportionally with the degree of vehicle damages and volume at upstream. Also, location of lane closure and involvement of heavy vehicle, alignment also have an effect on occurring congestion. Meanwhile, factor influence is different according to traffic flow characteristics before accident, and management goal. The cases of Model 1 (decrease in LOS from LOS A and B to LOS C), Model 3 (from LOS C and D to LOS E) are influenced by accident severity more than Model 2(from LOS A and B to LOS E). But, the cases of Model 2 are mostly affected by capacity decrease and demand. It means severely accumulated queue is needed for incurring congestion in Model 2. It is because Model 2 is the case representing conversion of the state (from free flow to severe congested flow). From this results, the research suggested what type of operations is needed to prevent congestion and what extent of operations is needed.