From Traffic Analysis to Real-Time Control: A Hazard-Based Model to Estimate Post-Collision Recovery Periods

When a collision occurs, information on its time and characteristics is needed for traffic control purposes; however, the post-collision impact on traffic and the corresponding expected recovery and clearance time is still an important and under-studied subject of research. Though the collision-induced delays may be estimated given the prevalence of mobile apps like Google Maps, the durations of post-collision effects remain largely unexplored. Existing studies mainly focus on modeling the collision induced queue formation and dissipation. However, limited research focuses specifically on the modeling of collision duration per se. Therefore, the objective of this study is to utilize hazard-based modeling to estimate collision clearance time by factoring in various endogenous and exogenous factors such as collision severity, lane width, number of lanes, and surrounding traffic conditions. 2017 and 2018 collision and traffic detector data from the California freeway performance measurement system (PeMS) were mined for developing a semi-parametric hazard function of collision durations. The resulting time dependent hazard—the likelihood of clearing a collision site—can help traffic controllers and drivers make better decisions while coping with the uncertainty associated with incident events.


  • English

Media Info

  • Media Type: Web
  • Pagination: pp 103-111
  • Monograph Title: International Conference on Transportation and Development 2019: Smarter and Safer Mobility and Cities

Subject/Index Terms

Filing Info

  • Accession Number: 01729972
  • Record Type: Publication
  • ISBN: 9780784482575
  • Files: TRIS, ASCE
  • Created Date: Aug 28 2019 3:01PM