Safety Assessment of Freeway Active Traffic Management by Exploring the Relationship between Safety and Congestion

This study shows how Oregon crash incident data and PSU Portal traffic data can be combined to determine what factors lead to increased crash risk. Data related to Oregon highway 217 was used to conduct the analysis. The first analysis used Portal traffic data to determine Level of Service (LOS) and then determine the relationship between that and crash rate (Fatal and Injury) which was derived from ODOT’s crash database. Similar to studies conducted in other states, there was a clear relationship between LOS and crash rate, with worse LOS associated with increased crash rate. The second part of the study used Portal data again, but this time the mean and variation of the variables speed, occupancy and volume were calculated 5-10 and 10-15 minutes before a crash incident on Oregon 217 on both the upstream and downstream directions. The crash incidents this time were derived from the Traffic Management Operations Center (TMOC) incident data which gave incident times to the nearest minute as opposed to the nearest hour in the LOS study. Given the number of correlated predictors in the data, logistic regression modeling may have led to regression estimates with large variances. Instead, logistic lasso regression was used to select a subset of significant predictor variables to predict the probability of a crash occurring given the traffic conditions at the time. Increasing upstream speed variation and occupancy, and downstream occupancy variation, volume and volume variation were associated with increased crash risk. Slower or decreasing downstream speed was associated with an increased crash risk.


  • English

Media Info

  • Media Type: Digital/other
  • Edition: Final Report
  • Features: Figures; References; Tables;
  • Pagination: 32p

Subject/Index Terms

Filing Info

  • Accession Number: 01686733
  • Record Type: Publication
  • Report/Paper Numbers: FHWA-OR-RD-19-05, SPR 793
  • Created Date: Nov 26 2018 4:54PM