Crash Surrogate Modelling for Traffic Operations on Freeways

Abstract

Crashes in freeways are recognized as being more dangerous and severe than those on urban streets. It is of great importance to analyze and improve traffic safety of freeways. Compared to the traditional count data regression models, crash surrogate indicators are widely recognized as a better alternative due to many advantages. Many crash surrogate indicators were proposed and widely applied to measure traffic conflicts in different situations. However, due to some drawbacks and limitations, these surrogate indicators are not suitable for freeway safety evaluation. Freeways often operate under a synchronized state, where a minor disturbance can lead to serious consequences due to uniform high speeds and small headways. Aiming to propose capable surrogate indicators for freeways, we fully examined the crash mechanism of the car-following scenario, and developed a tree structure to describe possible conflict scenarios. In the thesis, two new surrogate indicators of Braking Rate to Accommodate a Disturbance (BRAD) and Aggregated Crash Index (ACI) were proposed. These indicators can measure the minimum braking rate required by the following vehicle to avoid a crash, and where overall risk of the car-following scenario is investigated.