Abstract
Statistics indicate that accidents at signalized intersections have become a significant safety problem throughout the United States. Right-angle collisions are ranked highest in terms of severity at signalized intersections, while their frequency is second only to rear-end collisions. One key component that is lacking is a tool to help identify excessive risk of right-angle collisions. This paper proposes a practical method to evaluate the frequency of right-angle collisions on the basis of the frequency of short postencroachment times (PET). PET and crash data collected from signalized intersections in Indiana were used to calibrate right-angle crash prediction models. Regression results reveal that PET frequency is a key determinant of right-angle crashes and is capable of discriminating varying safety levels within a location as well as across locations. Several evaluation examples are presented to illustrate how the method can be used and how the estimation results can be interpreted.
