Safety Performance Functions for Intersections on Highways Maintained by the Virginia Department of Transportation

In recent years, significant effort and money have been invested through research and implemented safety projects to enhance highway safety in Virginia. However, there is still substantial room for improvement in both crash frequency and severity. As there are limits in the available funds for safety improvements, it is crucial that allocated resources for safety improvement be spent at highway locations that will result in the maximum safety benefits. In addition, intersection crashes play a significant role in the safety conditions in Virginia. For example, crashes at intersections in Virginia for the period 2003 through 2007 account for 43.8% of all crashes and 26% of fatal crashes. Therefore, identifying intersections for safety improvements that will give the highest potential for crash reduction when appropriate safety countermeasures are implemented will have a significant impact on the overall safety performance of roads in Virginia. The Federal Highway Administration (FHWA) has developed a procedure for identifying highway locations that have the highest potential for crash reduction (ITT Corporation, 2008). A critical component of this method is the use of safety performance functions (SPFs) to determine the potential for crash reductions at a location. An SPF is a mathematical relationship (model) between frequency of crashes by severity and the most significant causal factors on a specific highway. Although the SafetyAnalyst Users Manual presents several SPFs for intersections, these were developed using data from Minnesota. FHWA also suggested that if feasible, each state should develop its own SPFs based on crash and traffic volume data from the state, as the SPFs that are based on Minnesota data may not adequately represent the crash characteristics in all states. SPFs for intersections in Virginia were developed using the annual average daily traffic as the most significant causal factor, emulating the SPFs currently suggested by SafetyAnalyst. The SPFs were developed for both total crashes and combined fatal plus injury crashes through generalized linear modeling using a negative binomial distribution. Models were also developed for urban and rural intersections separately, and in order to account for the different topographies in Virginia, SPFs were also developed for three regions: Northern, Western, and Eastern. This report covers Phases I and II of the study, which includes urban and rural intersections maintained by VDOT. Statistical comparisons of the models based on Minnesota data with those based on the Virginia data showed that the specific models developed for Virginia fit the Virginia crash data better. The report recommends that VDOTs Traffic Engineering Division use the SPFs developed for Virginia and the specific regional SPFs suggested in this report to prioritize the locations in need of safety improvement.