Title

Assessing Crash Occurrence On Urban Freeways Using Static And Dynamic Factors

Keywords

Crash occurrence; Negative binomial models; Roadway geometrics; Traffic volumes

Abstract

Both roadway characteristics and traffic volume affect the expected number of crashes on a freeway section. While roadway features are mostly static, traffic is mostly dynamic. With the availability of loop detectors along urban freeways, real-time traffic flow could be measured and used to model the crash risk of different freeway sections under varying traffic volume scenarios. This could help evaluating the safety of freeway sections in real-time. Moreover, different sections might experience higher risks in certain times of the day based on the specific traffic volume enabling traffic management centers and emergency response to be prepared. Modeling crash frequency has been an important and effective way to identify potentially high risk locations on all types of roadways. Researchers have applied frequency modeling techniques such as the Negative Binomial models to overcome the problems associated with the multiple regression techniques. Most of these models use aggregate measures of traffic volume (e.g. AADT), but limited studies have identified the precise and representative traffic volume of a location. The main objective of this work is to develop a model for the frequency of crash occurrence on a 36-mile stretch of I-4 in Orlando, combining traffic, geometric and roadway features to identify the significant factors that contribute to traffic crashes on freeways, and to evaluate the different forms of representing the traffic volume measure at the location of crashes. Different forms of traffic volume were attempted including data obtained from 138 dual loop detectors installed on I-4. The results of the model provide strong evidence for the significance of the traffic volumes immediately prior to the crash.

Publication Date

12-1-2005

Publication Title

Advances in Transportation Studies

Issue

5

Number of Pages

39-51

Document Type

Article

Personal Identifier

scopus

Socpus ID

63849307989 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/63849307989

This document is currently not available here.

Share

COinS