CLOSE OVERLAY

Safety Performance Functions for Rural Road Segments and Rural Intersections in Michigan

Project Details
STATUS

Completed

START DATE

10/01/15

END DATE

03/31/18

FOCUS AREAS

Safety

RESEARCH CENTERS InTrans, CTRE
SPONSORS

Michigan Department of Transportation

PARTNERS

Subcontractor to Michigan State University

Researchers
Principal Investigator
Peter Savolainen

About the research

This study involved the development of safety performance functions (SPFs) for rural road segments and intersections in the state of Michigan. The facility types included two-lane and four-lane state trunklines (divided and undivided), rural county roadways (paved and gravel), signalized intersections, and minor-road stop controlled intersections. Data were compiled from several sources for thousands of rural road segments and intersections statewide. These data included traffic crashes, traffic volumes, roadway classification, geometry, cross-sectional features, and other site characteristics for the period of 2011-2015. These data were assembled into separate files based on the facility type, jurisdiction, and federal aid status. The Highway Safety Manual (HSM) base SPFs were then calibrated using the Michigan-specific data, which demonstrated significant variability in terms of the goodness-of-fit of the HSM models across various site types, due in part to the very high proportion of deer crashes on Michigan’s rural highways. Consequently, Michigan-specific SPFs were estimated, including simple statewide models that considered only annual average daily traffic (AADT), as well as regionalized models that accounted for regional differences in drivers, weather, topography, and other characteristics. More detailed models were also developed, which considered additional factors such as shoulder width, driveway density, horizontal curvature, median presence, road surface type, and intersection skew. Crash modification factors (CMFs) were estimated, which are used to adjust the SPF crash estimates to account for differences related to the site characteristics. Methods for prediction of crash frequency by collision type and injury severity were also established. Depending on the facility type, this was performed either by using separate SPFs, severity distribution functions (SDFs), or crash distributions. Ultimately, the results of this study provide a number of tools that allow for proactive safety planning activities, including network screening and identification of high-risk sites. These tools have been calibrated such that they can be applied either at the statewide level or within any of MDOT’s seven geographic regions to accommodate unique differences across the state. The report also documents procedures for maintaining and calibrating these SPFs over time to account for temporal changes that occur across the network.

 

 

TOP