Freeway Crash Prediction Models for Long-Range Urban Transportation Planning

The need for safety assessment tools for long-range transportation planning was not seriously recognized until the Transportation Equity Act for Twenty First Century (TEA-21) established a requirement related to safety considerations in the planning process of Metropolitan Planning Organizations (MPO). Currently, most MPOs do not assess the safety consequences of alternative transportation systems. The goal of this dissertation is to develop practical tools for assessing safety consequences of freeways in the context of long-range urban transportation plans. Data for freeway segments and crashes were obtained from North Carolina Department of Transportation and Tennessee Department of Transportation. Three different modeling approaches were utilized for analyzing crash and freeway data. These approaches are the analysis of variance, regression analysis, and classification tree analysis. Separate models were developed for each crash type by severity for non-interchange segments and interchange segments. Appropriate independent variables and model forms were selected based on the availability of future data for the variables and statistical measures. The prediction performance of the different developed models was assessed and compared. The research reveals characteristics and patterns of crashes on urban freeways and provides crash prediction models that can be utilized in the long-range transportation planning process.