Investigation of traffic safety and stability through simulation and theoretical modelling

• Main Author: Dong, Tianyu
• Other Authors: Zhu Feng
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Civil engineering::Transportation
• Online Access: https://hdl.handle.net/10356/165135
• Institution: Nanyang Technological University

Traffic safety attracts increasing attention from both academia and industry, because road accident is one of the significant factors leading to death or injuries for human beings. In order to reduce traffic accidents caused by human-related errors, numerous countries and governments devoted efforts to developing more effective traffic policies and regulations. It is necessary to conduct multi-angle traffic safety analysis in various scenarios. In-depth and comprehensive analyses of traffic safety have the potential to assess the safety level, and figure out the essential factors leading to accidents, thereby providing useful guidance to mitigate the occurrence of accidents. In recent decades, the development of communication and automation technologies has injected new vitality into intelligent transportation systems (ITS). However, the impact of these new technologies on traffic safety is still unclear. With consideration of the scarcity of accidents data for emerging technology vehicles, the first study of this thesis employs simulation to investigate the safety impact of two kinds of emerging technology vehicle technologies, adaptive cruise control (ACC) of automated vehicles (AVs), and cooperative cruise control (CACC) of connected automated vehicles (CAVs), in both homogeneous and heterogeneous traffic. Both theoretical stability analysis and simulations are conducted in homogeneous traffic, and the results indicate that CACC outperforms ACC from a safety perspective. In heterogeneous traffic simulation, not only the market penetration rate (MPR) but also the platooning intensity is considered. Results show that the collision risk increases with the increment of MPR of AV, and increases first and then decreases with the increment of MPR of CAV. The collision risk decreases with the increase of platooning intensity for both AV and CAV. The first study adopts micro-simulations to investigate traffic safety. However, simulation results are dependent on the car-following (CF) model selection and simulation settings. In order to eliminate the dependency, sensitivity analysis is usually required to obtain reliable conclusions. Nevertheless, it is time-consuming and resource-demanding to perform the sensitivity analysis, particularly when a large number of parameters are involved. To further address the limitations of simulations, the second and third studies of this thesis aim to theoretically study traffic safety based on a general CF model. Through the theoretical studies of traffic safety, we mathematically formulate traffic safety, which enables researchers to better understand the mechanism of traffic safety. Specifically, the second study sets out to investigate the relationship between traffic stability and safety. Traffic stability and safety are believed to be closely related. Though they are always investigated separately, a clear relationship between traffic stability and traffic safety is missing in the literature. Therefore, the traffic safety criterion is derived from a generic CF model, and then the derived criterion is compared with the traffic criterion. Two definitions of traffic safety are investigated, i.e., no crash and low potential collision risk quantified by surrogate safety measures (SSMs). Both the local stability criterion and string stability criterion are compared with the safety criterion. The results show that when we consider an infinite-size vehicle fleet, there is no crash when and only when the traffic fleet is string stable. While string stability is a sufficient but unnecessary condition for the non-crash criterion of a fleet with a finite number of vehicles. The findings of this research have the potential to not only improve the theoretical traffic safety analysis, but also provide guidance for automated vehicle design and safety level assessment of human driving behaviours. The results of the second study provide insights into under what conditions the traffic is considered safe or unsafe. In the third work, we move more in-depth into the propagation of collision risk. Specifically, the third work extends the stability analysis method which focuses on the propagation of disturbance over time and space to the propagation of collision risk. Comparative analysis is conducted between traffic stability and collision risk stability. It shows that the risk local stability condition is consistent with the traffic local stability condition. However, the boundary conditions of risk string stability are different from traffic string stability for non-linear risk indicators. In the rational parameter configuration range, stable string traffic may still be risk string unstable. Further, numerical simulations and microscopic traffic simulations with various CF models are performed to verify the theoretical derivation and the applicability of the proposed risk stability analysis.