Lack of geometric design consistency has been considered a major contributing factor to collision occurrence, and can therefore be used to indicate areas with higher potential collision risk. Providing designers with a tool to evaluate the design consistency of new highways during the design stage would help eliminate these areas before the construction of the highway, and thus provide a proactive approach to safety improvement. The most common criteria for design consistency evaluation have been based on the operating speed selected by drivers according to their own perception of the road. This paper presents the results of a comprehensive study conducted to develop operating speed and differential speed prediction models for two-lane highways and relate the evaluation measures to actual collision experience. Data on actual road alignments, traffic volumes, and collisions were collected for segments on seven different two-lane highways belonging to different highway classes. Speed data were collected on a sample of horizontal curves representative of all curves on these highways. The speed data were first used to model drivers' choice of operating speed and speed differential as they negotiate the horizontal curves. Regression analysis based on the negative binomial distribution was then carried out to relate the collision frequency to the design-speed margin consistency and operating speed consistency. It was found that operating speed consistency provided superior models, and is recommended as a consistency evaluation criterion.
Samenvatting