Empirical Evaluation of the Impacts of High-Occupancy Vehicle (HOV) Lane Collisions on Different Types of Lane Configuration in California

Abstract

This paper focused on the impacts of HOV lane collision on HOV facilities with different lane configuration in California. Based on the 5-minute loop detector data and HOV lane collision records from 17 study routes in the year 2008, the real-world collision-induced impacts were estimated by the proposed methodology, in which three major steps are included: 1) synchronization of HOV lane collision information from two databases, the California Freeway Performance Measurement System (PeMS), and the Traffic Collision Surveillance and Analysis System (TASAS); 2) identification of collision’s impact based on the traffic data; and 3) calculation of collision-induced delays. The Wilcoxon rank sum test and The Kolmogorov-Smirnov Test were used to verify the equality of the data sample distributions from HOV facilities with different lane configuration. A log-linear regression model was used to investigate influential factors on HOV lane collision impacts, such as lane configuration, collision duration, injury or not, peak period or not, collision type and weather condition. The K-S test result proves that the distribution of samples drawn from continuous-access HOV facilities is equal to that of limited-access ones, but for the adjacent general purpose (GP) lane the impact is different. The results from the regression analysis imply that the impacts of HOV lane collision on both HOV and the adjacent GP lanes may be different in different HOV lane configurations.