Hazard based models for freeway traffic incident duration.

Area of Science:

• Transportation Engineering
• Traffic Management Systems
• Statistical Modeling

Background:

• Non-recurrent congestion from traffic incidents poses a major challenge for road network managers.
• Effective mitigation strategies require understanding factors influencing incident duration.
• Previous research has not fully explored the differential impact of various factors on diverse incident types.

Purpose of the Study:

• To investigate the influence of various factors on traffic incident duration on major roads.
• To develop and validate a comprehensive incident duration model using real-world data.
• To identify key determinants affecting the duration of different types of traffic incidents.

Main Methods:

• Utilized twelve months of traffic incident data from the Australian freeway network.
• Developed parametric accelerated failure time (AFT) survival models, including log-logistic, lognormal, and Weibull distributions.
• Employed models with fixed and random parameters, and a Weibull model with gamma heterogeneity to analyze incident duration.

Main Results:

• Weibull AFT models with random parameters effectively modeled duration for crashes and hazards.
• A Weibull model with gamma heterogeneity was optimal for stationary vehicle incidents.
• Incident characteristics (severity, type, towing needs), location, time, and traffic conditions significantly impacted duration; infrastructure and weather did not.

Conclusions:

• Incident duration is uniquely determined by incident type and specific influencing factors.
• Findings provide crucial insights for traffic incident management agencies to reduce incident clearance times.
• Reduced incident duration can lead to decreased congestion, fewer secondary incidents, and lower economic/human losses.