A combined framework for modelling non-secondary and secondary crashes

The objective of this study, is to develop a combined framework that allows both secondary and non-secondary crashes to be identified and modelled. This study further uses survival analysis to explore the relationship between primary and secondary incidents; something which has not been trialled before. This study focuses on Sydney's major motorways; the M4 Western Motorway and the M5 South Western Motorway, for the period 2011 to 2013. Crash geometric data provided by the Roads and Maritime Services from the Crashlink database and GlPSJCAM survey respectively. To determine the crash boundary, speed measurements for three and a half hours after the recorded time of the primary crash was used. The negative binomial was used to investigate crash risk factors specific frequency of non-secondary crashes. Separate models were generated for injury only crashes, and crashes that occurred during the AM and PM peak periods. For secondary crashes, the legit and hazard based approach was used to identify the explanatory variables that contribute to the occurrence of a secondary accident and how this changes with time from the primary incident. With an upper crash duration boundary of three and a half hours, the likelihood of a secondary incident was seen to be greatest during the first 60 minutes of the primary crash occurring. This likelihood was found to be heavily influenced by a combination of weather conditions and driver expectations. Data continues to be one of the main obstacles in creating a complete crash framework. To overcome the modelling limitations of this study in the future, a standardized and reliable data collection and recording method for all transportation agencies is essential.


  • English

Media Info

  • Pagination: 18p

Subject/Index Terms

Filing Info

  • Accession Number: 01676242
  • Record Type: Publication
  • Source Agency: ARRB
  • Files: ITRD, ATRI
  • Created Date: Jul 26 2018 10:45AM