Development of Safety Performance Functions for Network Screening of Roadway Departure Crashes in Virginia

About the report:

Roadway departure (RD) is recognized as one of the eight emphasis areas in the Virginia 2017-2021 Strategic Highway Safety Plan (SHSP), and the Virginia Department of Transportation (VDOT) currently uses counts of RD crashes to identify locations for RD safety improvement. However, identifying locations based on crash counts is subject to bias and inaccuracy, leading to ineffective or erroneous outcomes.  The safety performance functions (SPFs) VDOT has been using for statewide network screening might be used for RD safety improvement, but this could lead to undesirable outcomes in that the current SPFs focus on all crash types and RD safety issues are believed to be different from those of other crash types.  This study was designed to develop SPFs for statewide network screening for RD safety improvements to overcome this issue.  RD SPFs were developed for 16 site types ranging from rural 2-lane segments to urban freeway segments with 8 or more lanes.  It should be noted that crashes within 250 feet from an intersection are excluded according to VDOT’s RD definition.

Of the 96 RD SPFs (16 site types x 2 severity levels x 3 functional forms) initially investigated, 93 RD SPFs were successfully developed for RD network screening.  One site type did not result in statistically significant RD SPFs for fatal and injury crashes, likely because of its small sample size.  The study found that the RD SPFs vary in their functional forms of annual average daily traffic (AADT)across the site types.  The logarithmic functional form of AADT, regarded as a standard for an SPF, is deemed suitable in general for a typical range of AADT. However, that form could be severely deviated from the true relationship in the data.  The study also found that the functional forms of AADT vary by injury severity.  The forms are generally similar between the two severity levels, yet there are some cases where the difference is substantial.  Accordingly, the functional form of AADT in an SPF for RD crashes should be determined for each site type and by severity level (all RD crashes and fatal and injury RD crashes) separately whenever possible. 

Based on the findings, the study recommends that the RD SPFs developed in this study be incorporated in the current VDOT statewide network RD screening procedure.  Specifically, the final RD SPFs determined by goodness of fit and prediction capability measures (Tables11 and 12) are recommended for use.  For some site types, more than one final SPF are provided, allowing VDOT to select the most appropriate one for network screening considering the practicality of implementing the RD SPFs.  Development of separate SPFs by site type and by severity was desirable to avoid possible inaccurate outcomes of network screening.

Authors