Development of a methodology to compare and select an appropriate crash surrogate for freeways safety analysis: A case study on truck lane restriction strategies

Crash occurrence has been widely used as the ultimate measure of safety, for diagnosing traffic safety problems, and evaluating the effectiveness of safety improvements on all kinds of transportation facilities. However, these applications are usually fraught with problems due to rare and random occurrence of traffic crashes and the extremely long period required before a history of crashes is well built for solid statistical analyses. Hence, more and more efforts have been made to identify an appropriate crash surrogate which could be used as a substitute for or to supplement crashes for safety analyses. The most popular crash surrogate is traffic conflicts. A notable merit of traffic conflicts over actual crashes is that adequate data for statistical analyses can be collected over several weeks. However, a solid correlation between crashes and traffic conflicts has so far not been built and most previous studies on this subject were confined to the area of urban intersection safety. These reasons motivated the author to conduct this study. The purpose therefore of this study is (1) to propose a modified definition of traffic conflicts that will facilitate the use of the technique on interstate highways, (2) calibrate the parameters that indicate the occurrence of traffic conflicts, and (3) identify an appropriate crash surrogate for safety analysis on interstate highways. In order to illustrate the use of the identified crash surrogate, a case study was conducted to determine the safety and operational effects of truck lane restriction strategies on interstate highways.;This study redefined six severity measures of traffic conflicts for safety analysis on freeways and investigated the relationship between crashes and traffic conflicts collected from simulation. The correlation between crashes and traffic conflicts were built up through comparison of traffic conflicts and crash data collected from three sites on interstate freeways I-64, I-66, and I-81 within the jurisdiction of Virginia. Based on the results of correlation analysis between crashes and traffic conflicts with different levels of severity measures, the thresholds for all six severity measures were calibrated, and appropriate ranges of optimal thresholds identified. Before the calibration, the simulation package -- PARAMICS was calibrated for the first time with the target function of distribution of headways, a microscopic measure instead of the traditional macroscopic measures, e.g. travel time or LOS since the conflict events investigated in this study involve detailed interactions between individual vehicles. Crash prediction models were also built using frequencies of different types of conflicts as independent variables collected from one site and validated using data from the other two sites. As a result of the calibration and model development, the Time-To-Collision (TTC) with a threshold of 1.5 seconds was chosen as the appropriate crash surrogate for safety analysis on freeways.;In this study, a Surrogate Safety Assessment Methodology (SSAM) was developed for safety analysis using crash surrogate as the measure of effectiveness (MOE). The proposed SSAM and conflicts measured by TTC with a threshold of 1.5 seconds were employed to evaluate the effectiveness of truck-lane restriction strategies under different geometric (number of lanes, grade, posted speed limits, and ramp density) and traffic (demand volume, and truck percentage) conditions. In order to evaluate the effectiveness of truck lane restriction strategies comprehensively, different types of conflicts were combined into one measure after weights were given to them according to pre-crash maneuvers and vehicle types involved in a traffic conflict event. These severity weights were computed on the basis of investigation of the severity and cost of different types of crashes involving the corresponding pre-crash maneuver and vehicle types on interstate freeways in Virginia. These results were later used to develop guidelines for implementing optimal truck-lane restriction strategies on interstate highways in terms of reducing traffic conflicts and enhancing LOS in restricted lanes.;Results obtained from this study provide a theoretical basis for applications of crash surrogates as a substitute for or to supplement crashes for safety analyses on freeways. For the first time, the parameters (thresholds) of different types of conflict definitions are confined to reasonable ranges for safety studies on freeways. The calibration of traffic simulation package demonstrates a new microscopic target function - the distribution of headways which improves the credibility of traffic simulation in investigations of detailed interactions between individual vehicles. Severity analysis of traffic conflicts reveals the relationship between the severity of crashes and pre-crash maneuvers and vehicle types involved in crashes. The case study evaluated the impacts of truck lane restrictions on safety and operation and the guidelines produced thereafter facilitated the selection of appropriate truck lane restriction strategy under different geometric and traffic conditions.