Determinants of the Congestion Caused by a Traffic Accident in Urban Road Networks
Zhenjie Zheng,Zhengli Wang,Liyun Zhu,Hai Jiang
DOI: https://doi.org/10.1016/j.aap.2019.105327
IF: 5.9
2019-01-01
Accident Analysis & Prevention
Abstract:Non-recurrent congestion is frustrating to travelers as it often causes unexpected delay, which would result in missing important meetings or appointments. Major causes of non-recurrent congestion include adverse weather conditions, natural hazards, and traffic accidents. Although there has been a proliferation of studies that investigate how adverse weather conditions and natural hazards impact road congestion in urban road networks, studies that look into determinants of the congestion caused by a traffic accident are scarce. This research fills in this gap in the literature. When a traffic accident occurs on an urban link, the congestion would propagate to and affect adjacent links. We develop a modified version of the Dijkstra's algorithm to identify the set of links in the neighborhood of the accident. We first measure the level of congestion caused by the traffic accident as the reduction in traveling speed on those links. As the impact of congestion varies both in space and in time, we then estimate a generalized linear mixed-effects model with spatiotemporal panel data to identify its determinants. Finally, we conduct a case study using real data in Beijing. We find that: (1) the level of congestion is mostly associated with the types of the traffic accidents, the types of vehicles involved, and the occurrence time; (2) for the three types of traffic accidents, namely, scrape among vehicles, collisions with fixed objects, and rear-end collisions, the level of congestion associated with the first two types are comparable, while that associated with the third type is 8.43% more intense; (3) for the types of vehicles involved, the level of congestion involving buses/trucks is 6.03% more intense than those involving only cars; (4) for the occurrence time, the level of congestion associated with morning peaks and afternoon peaks are 5.87% and 6.57% more intense than that associated with off-peak hours, respectively.