In this paper, we present evidence of strong relationships between traffic flow conditions and the likelihood of traffic accidents (crashes), by type of crash. Traffic flow variables are measured using standard monitoring devices such as single inductive loop detectors. The key traffic flow elements that affect safety are found to be mean volume and median speed, and temporal variations in volume and speed, where variations need to be distinguished by freeway lane. We demonstrate how these relationships can form the basis for a tool that monitors the real-time safety level of traffic flow on an urban freeway. Such a safety performance monitoring tool can also be used in cost-benefit evaluations of projects aimed at mitigating congestion, by comparing the levels of safety of traffic flows patterns before and after project implementation. (A) "Reprinted with permission from Elsevier".
Abstract