Calibration of HSM Predictive Methods on State and Local Rural Highways

Performance-based safety goals and objectives are more attainable with the use of the Highway Safety Manual (HSM). However, the safety performance functions (SPFs) in the HSM are not accurate because they are not calibrated to local conditions. Additionally, each SPF and crash modification factor (CMF) assumes a set of base site conditions which might not be realistic or representative of local roadways. The calibration procedures provided in HSM Part C Appendix A should therefore be modified to accommodate local data availability as well as roadway, traffic, and crash characteristics. Furthermore, a set of base conditions applicable to local highways should be determined. This study presents the application of the HSM to rural state highways in South Dakota and provides important guidance and empirical results regarding how to calibrate HSM models. Calibration guidelines have been developed to facilitate future calibration activities. The calibration was based on crash data from three roadway segment types and five highway intersection types during a five-year period (2008-2012). The calibration process includes establishing new base conditions, developing SPFs, converting CMFs to base conditions, and deriving calibration factors. Considering the sample size, base conditions, and prediction accuracy, four state-specific SPF models have been developed for two segment types (two-way two-lane and multilane divided) and two intersection types (two-way two-lane three-leg STOP and two-way two-lane four-leg STOP), respectively. The HSM SPFs are retained for other highway facility types. The values of calibration factors for crashes are derived by injury severity scale (i.e., killed, type A, B, or C injury, and property-damage-only crashes). Results show that the number of crashes observed in South Dakota can be drastically different from those predicted by the HSM. The HSM models in this study underestimate the total crash count on roadway segments for all rural state highway segment types in South Dakota. On the other hand, the models overestimate the total crash count at intersections for all the intersection types in South Dakota except multilane three-leg STOP sites. The HSM models overestimate fatal and injury crashes in all highway facility types. Property-damage-only crashes are sometimes overestimated by 200% and sometimes underestimated by 72%. The state-specific SPF models developed from South Dakota’s safety data perform better for facility types that are not adequately predicted by the HSM method. The recommendation is to use a hybrid method with both HSM and state-specific models for certain segment and intersection types to achieve high crash prediction accuracy.

  • Record URL:
  • Supplemental Notes:
    • This document was sponsored by the U.S. Department of Transportation, University Transportation Centers Program. Cover date: August 2020.
  • Corporate Authors:

    University of Wisconsin, Madison

    Department of Civil and Environmental Engineering
    1415 Engineering Drive
    Madison, WI  United States  53706

    South Dakota Department of Transportation

    700 East Broadway Avenue
    Pierre, SD  United States  57501

    Mountain-Plains Consortium

    North Dakota State University
    Fargo, ND  United States  58108

    Federal Highway Administration

    1200 New Jersey Avenue, SE
    Washington, DC  United States  20590

    Office of the Assistant Secretary for Research and Technology

    University Transportation Centers Program
    Department of Transportation
    Washington, DC  United States  20590
  • Authors:
    • Qin, Xiao
    • Chen, Zhi
    • Shaon, Mohammad Razuar
  • Publication Date: 2020-6-30

Language

  • English

Media Info

  • Media Type: Digital/other
  • Edition: Final Report
  • Features: Appendices; Figures; Maps; References; Tables;
  • Pagination: 90p

Subject/Index Terms

Filing Info

  • Accession Number: 01751955
  • Record Type: Publication
  • Report/Paper Numbers: MPC-20-416, SD2013-04-F
  • Contract Numbers: MPC-438; 311179
  • Files: UTC, TRIS, ATRI, USDOT, STATEDOT
  • Created Date: Sep 8 2020 10:43AM