Investigating Vehicle Roadway Usage Patterns on the Shanghai Urban Expressway System and Their Impacts on Traffic Safety
Rongjie Yu,Yiyun Wang,Mohammed Quddus,Jian Li,Xuesong Wang,Ye Tian
DOI: https://doi.org/10.1080/15568318.2020.1722869
2021-01-01
International Journal of Sustainable Transportation
Abstract:The urban expressway system serves as a key role in the roadway transportation system. It provides an efficient and comfortable approach for long-distance travel within the city. However, the safety status of the urban expressways is becoming a critical issue as the high-frequent traffic crashes have severely influenced the traffic operations. Among the safety influencing factors, including traffic operational parameters (such as traffic speed and volume), geometric features and traffic participants' characteristics (such as vehicle roadway usage patterns), the traffic operational parameters and geometric features have been widely investigated. However, the impacts of traffic participants' characteristics on traffic safety have never been examined. This unprecedented study aims to link vehicles' roadway usage patterns with traffic safety through crash frequency analyses. First, the roadway usage patterns were identified using Latent Class Cluster Analysis (LCCA) based on their traveling rates. Then, the hourly-based crash frequency analysis data were formulated with traffic operational parameters, geometric features and crash data. Finally, crash frequency analysis models were developed to unveil the relationships between the crash occurrence and their influencing factors. The modeling results showed that the Random Effects Hurdle Negative Binomial Model (REHNBM) provided better goodness-of-fit. And it concluded that higher proportions of vehicles with low-level roadway usage pattern would substantially enhance the possibility of crash occurrence; while the proportions of vehicles with the medium-high-level roadway usage pattern had negative impacts on crash occurrence probability. Finally, safety improvement recommendations and strategies based on the modeling results were put forward.