Evaluation of Geometric and Traffic Characteristics Affecting the Safety of Six-Lane Divided Roadways

The analysis of traffic crashes that occurred in Florida in 2002 has shown that six-lane divided highways had higher crash rates than two-lane and four-lane highways. Variation of roadway geometrics and traffic characteristics on six-lane highways compared to four and two-lane highways are suspected to be the source of the difference. The goal of this study was to make a comparison of crashes on four and six-lanes in terms of descriptive analysis and to build a crash prediction model which can be used to predict annual number of crashes. Detailed analysis of geometric and traffic data collected from the archives and in the field was undertaken to establish the correlation between independent factors and crash frequency. The results of descriptive analysis showed distribution of crashes on four and six-lanes to be dependent. In most crash attributes, crash rates on six-lane sections were higher. A Zero-Inflated Negative Binomial (ZINB) crash prediction model was developed. The response variable in the model was crash frequency and independent variables were roadway geometrics and traffic characteristics. Crash frequency was found to have a very strong relationship with most of the independent variables indicated by their p-values. Section length, number of access points, sidewalk, percentage of trucks and traffic volumes had a positive coefficient with the crash frequency. Posted speed limit, shoulder width, median width, surface width and curvature had negative coefficients indicating that when they increase in magnitude, crash frequency decreases. Transformation of some variables like posted speed limit, horizontal degree of curve from numerical to categorical resulted into a well-fitted model.