A quick response to an incident scene is crucial to understand the incident characteristics and implement an appropriate incident management including an advanced warning to on-coming drivers. In order to do this, incident response time on freeways for peak hours was modeled by a generalized form of Poisson regression (Negative Binomial regression model). This study utilizes a ten-year crash-data of the state of Ohio from 1990 to 1999. Incident response times on freeways were found by using crash log entries of crash-report time and crash-scene-arrival time in the crash data files. The additional data that affect the capacity of freeways such as number of lanes and traffic demand at where an incident occurred will help develop better models. The developed model can be used to provide an accurate incident response time on freeways for those who report incidents and a great resource for a real-time incident management system in ITS (Intelligent Transportation Systems) such as an on-coming incident information by variable message signs to drivers on freeways.
Abstract