Relationship Between Highway Pavement Condition, Crash Frequency, and Crash Type

Abstract

Research has been conducted to relate highway pavement conditions with vehicle accidents. Utilizing the Tennessee Pavement Management System (PMS) and Accident History Database (AHD), this study developed 20 negative binomial regression models to examine the relationships between pavement condition parameters, crash frequency, and crash types. The modeling results indicated that either Present Serviceability Index (PSI) or International Roughness Index (IRI) was a significant pavement condition parameter for predicting the crash frequency of highway segment; whereas, the Rut Depth (RD) was not statistically significant in the crash prediction models. Due to the collinearity between PSI and IRI, it was found that when the two parameters were applied together into the crash regression models, the statistical regression results could not be well explained. Comparing to the IRI and RD models, the models’ goodness-of-fit results indicated that PSI models had the best performance consistently in predicting the frequency for each type of crashes. Additionally, regression results indicated that either right shoulder width or left clearance to median up to 8 feet wide would significantly reduce the crash frequency.