Development of a new conflict-based safety metric for freeway exit ramps

Abstract

In this article, the safety implication of freeway exit ramps is examined using a newly developed conflict-based safety metric. A multi-step Monte Carlo simulation procedure is proposed to estimate the crash probability of a conflict (i.e. the conflict metric), considering the stochastic nature of human reaction (i.e. reaction time) and vehicle capability (i.e. maximum braking rate). Motion prediction methods are introduced to determine possible crash occurrence during the Monte Carlo procedure, for the two main conflict types (i.e. rear-end conflict and lane-change conflict) frequently occurring within freeway exit ramps. Notably, the conflict metric is able to differentiate two types of crash occurrences of lane-change conflicts: sideswipe and rear-end crashes. Safety evaluations on 16 freeway exit ramps based on microscopic traffic simulation is conducted; based on the proposed safety metric, a traditional time to collision–based conflict metric, as well as actual crash data, was proposed. The proposed metric shows the power in identifying relative safety among multiple freeway exit ramps and potential in directly predicting actual crashes. Thus, it could be considered as a promising tool for safety evaluation on freeway exit ramps, providing practitioners and professionals valuable information in traffic design, control, and management of freeway exit ramps.