|Abstract or Summary
- Large truck-involved crashes have a significant impact on both the economy and society. They are associated with high injury severities, high crash costs and contribute to congestion in urban areas. Past studies have investigated the contributing factors of large truck-involved crashes, however a study isolating the spatial and temporal effects is lacking. This thesis aims to bridge that gap as well as provide practical applications to improve safety from a large truck perspective through two new frameworks. This thesis contains two standalone documents, each detailing the spatial and temporal transferability framework, separately. These frameworks provide additional information that can be utilized in the development of planning tools to ultimately improve safety.
Random parameters logit models (i.e. mixed logit models) were utilized to help identify the contributing factors of large truck-involved crashes. One advantage of the mixed logit model is that it can account for the unobserved heterogeneity in the model which relaxes the independence of irrelevant alternatives (IIA) property. A series of log likelihood ratio tests were utilized to determine if transferability, spatial or temporal, was warranted.
The first document details the spatial transferability framework which is demonstrated through a case study on large truck-involved crashes in urban areas in Oregon and Texas. Strict regulations imposed on the trucking industry limits the variability of heavy-vehicle configurations and enhance the standards for truck drivers (as opposed to passenger vehicle drivers). Encouraging consistency between large trucks is one way to improve safety and has also lead to the investigation of commonalities between large truck-involved crashes in two spatially distributed regions. The results of the log-likelihood ratio tests indicate that spatial transferability is not warranted between Oregon and Texas. Key differences were non-driver or 'uncontrollable' characteristics (e.g. weather, light conditions and time of day) while driver related characteristics (e.g. gender, age and restraint use) had similar impacts. Since the major differences include non-driver characteristics, perhaps a regional model with similar 'uncontrollable' characteristics is warranted.
The second document illustrates the temporal transferability framework which is applied to large truck-involved crashes in urban areas in Texas. Traffic patterns, light conditions and driver behavior vary throughout the day and consequently can have a varied impact on large truck-involved crashes. The results of the log likelihood ratio tests indicate that temporal transferability is warranted and the database was divided into five time periods to be analyzed separately. Traffic flow, light conditions, surface conditions, month and percentage of trucks on the road were among the significant differences between the crash factors of each time period.
The two proposed transferability frameworks, spatial and temporal, provide new information that can be integrated into safety planning tools and more sharply guide
decision-makers. For example, the results of this thesis can help to pinpoint temporal or spatial-related countermeasures. In addition the results of this thesis can help in the allocation of limited resources (i.e. help prioritize projects), minimize economic loss and help decision makers improve safety from a large truck perspective (e.g. modify trucking regulations).
Finally, this thesis provides a foundation for future research. As indicated in Chapter 2, a future study to evaluate the feasibility of a regional large truck-involved crash model between neighboring regions and the development of a national crash data reporting standard are potential ideas for future research. Chapter 3 stressed the importance of time of day on large truck-involved crashes which can serve as the basis to study the safety and economic impacts of time of day shifts of truck freight movements to off-peak periods. In summary, this thesis involves original research that expands the literature and provides a new foundation to analyze large truck-involved crashes.