The Epidemiology Study Of Intersection Crashes And The Investigation Into The Effect Of An Intersection Collision Warning System

Since the first traffic accident resulted in the death of a human being, vehicles have threatened the safety of people’s lives and property despite their undoubted benefits. Intersections are one of the most dangerous locations on the road network, because they are a convergence point of vehicles traveling on conflicting paths with other vehicles and pedestrians. Intersections are also a contributing factor to traffic congestion and have a high prevalence of severe side-impact crashes. Therefore, Intersection crashes are responsible for a substantial portion of traffic crashes and traffic crash casualties. In order to develop more effective safety countermeasures to improve traffic safety at intersections, a better understanding of intersection crashes is necessary.Firstly, a descriptive epidemiological study of intersection crashes was carried out to establish the frequency of casualty intersection crashes by category. The dataset examined was provided by VicRoads, the state road authority in Victoria, Australia, for the period January2000to December2009. In addition, the circular distribution method was used to investigate temporal patterns in casualty intersection crashes.Logistic regression is an analysis method used in analytic epidemiology that has been proven to be a reliable means of revealing the relationship between the response and explanatory variables within the road accident domain. In this study, a logistic regression was conducted in order to reveal the risk factors affecting the severity of intersection crashes in Victoria, Australia. The potential risk factors identified included driver characteristics, vehicle features, environment and road factors and crash characteristics.While the logistic regression method was used as a statistical interpretation tool of intersection crashes, an innovative step was included that employed fault tree analysis (FTA) as a systematic interpretation tool, focusing on the identification of the basic events that can lead to an intersection crash. By constructing a fault tree, this method can describe the logical process from basic failure events (contributing factors) to the top outcome event (intersection crash) to present how the top event happens. During this study a pilot fault tree analysis was produced to categorize the contributing factors to intersection crashes. The data for the analysis was based on interviews conducted with crash-involved drivers and detailed inspections of the vehicle and crash site. The data used were sourced from the Australian National Crash In-depth Study (ANCIS), which provided detailed information for each intersection crash collected. Fault tree analysis was used in order to provide a new insight into the reasons behind the occurrence of intersection crashes rather than to simply examine their outcomes.There are some methods utilized to study the contributing factors affecting the severity of a traffic accident or leading to the occurrence of a traffic accident, such as logistic regression and fault tree analysis referred in the thesis. In order to assist this kind of research, a database dedicated for collecting the accident contributing factors is a potential requirement.In order to reduce the incidence of intersection crashes, many safety countermeasures have been proposed. One of the countermeasures under development is an intersection collision warning system (ICWS). The prototype intersection collision warning system uses dedicated short range communication (DSRC) technology, which enables high speed and low latency wireless communication between two vehicles or between a vehicle and the roadside infrastructure, to alert a driver of a potential collision. A driving simulator experiment was conducted, in order to better understand the effects of an ICWS on driver crash avoidance behavior at intersections. The driving simulator study was opted for over a real word driving experiment because of the expense, complexity and limited controllability of the latter.The studies performed in this thesis revealed a variety of interesting results. Circular distribution analysis revealed that the time of central tendency for intersection crashes resulting in casualty was15:19. The most frequent time period in which crashes occurred was between9:57and20:40, reflecting the period when most traffic movements take place in a24-hour period.The logistic regression study into the severity of intersection crashes showed that seven risk factors of the twelve selected for analysis were found to be statistically significant, including driver age and gender, speed zone, traffic control type, time of day, crash type and seat belt usage. This study found that male drivers as well as older drivers (aged65and above) had higher odds of being involved in fatal intersection crashes compared with their reference group. The intersection crashes occurring between midnight and early morning (0:00-5:59), in100km/h speed zones, or without traffic control had a higher odds of a fatal outcome than their counterpart categories. Furthermore, collisions involving pedestrians or a non seatbelt-wearing driver were more likely to be fatal crashes.The application of fault tree models to the Australian real-world crash dataset identified the most common factors contributing to intersection crashes. The most likely combination of contributing factors was that the driver misjudged the speed or gap and took no evasive action. This combination had a probability of occurrence of0.15.37%of the intersection crashes examined were found to be the result of intentional behaviors.A web-based contributing-factor database system was built using ASP.NET and ACCESS technologies. The database included user login, information input, information query and information presentation interfaces. There were two types of user using the database system:information input user and the information query user.The driving simulator experiment results showed that an ICWS, could reduce the number of intersection crashes by approximately40%to50%and shorten drivers’ braking reaction time in response to emergency events at intersections. It was found that drivers’ braking reaction time in response to an auditory warning was significantly shorter than braking reaction time in response to the visual warning in the cross traffic and the violating vehicle from right side (110R) scenario. As well, speeds were reduced in the cross traffic and the violating vehicle from left/right side (110L/R) and the violating vehicle right turn against the normal vehicle (121) scenarios and deceleration rates increased in the110L and121scenarios under the influence of ICWS.With the rapid development of the automotive industry and the dramatic increase in the size of the vehicle fleet, it is essential to better understand the nature and causes of traffic crashes. The descriptive epidemiological study of intersection crash frequency and the logistic regression study as the analytic epidemiological study on the severity of intersection crashes provided in-depth information of intersection crashes. Identification of risk factors and the discussion of the relative odds ratio between levels on the impact of the intersection crash severity would be beneficial for road safety stakeholders involved in the development of initiatives to reduce the severity of intersection crashes. As well, the contributing factors categorized by FTA could be helpful to visualize deduction of intersection crash causations. The construction of the contributing factor database would assist future studies into intersection crashes. The driving simulator experiment provided overall support for ICWSs based on the DSRC technology concept. In general, the results from this thesis provide a significant contribution to understand of intersection crashes and have broader implications for the development of effective safety countermeasures at intersections.