Modeling heavy vehicle crash and injury severity

Every year, nearly 1.2 million persons are killed and 50 million are injured in road crashes around the world. Road crashes are anticipated to be among three top leading causes of deaths in the world by 2020. In Australia, almost 1,400 people are killed and 32,000 people are severely injured in road crashes annually. Of the different types of vehicles involved in crashes, heavy vehicles are a major traffic safety concern, due to their higher likelihood of involvement with fatal and severe injury accidents. In Australia, heavy vehicles contribute significantly to the nation's economy because they are the major means for transporting goods in the country. In addition, it is predicted that heavy vehicle traffic will increase by 50 per cent by 2030. Therefore, the increase in the number of heavy vehicles will add to safety concerns because the probability of vehicle crashes increases by five per cent when the heavy vehicle percentage is higher than 30 per cent of total traffic volume. On the other hand, although heavy vehicles comprise only a small percentage, roughly 3% of the total registered vehicles, this type of vehicle is involved in 18% of total road fatalities. Therefore, the reduction in the number of crashes involving heavy vehicles has been proposed as one of the key performance indicators in the National Road Safety Strategy 2011-2020 for Australia. To reduce the trauma of heavy vehicle crashes, more research is needed to provide a better understanding of the factors influencing the frequency and severity of these crashes. The aim of this research is to identify the factors influencing heavy vehicle crashes and injury severity in Victoria, Australia. Therefore, in this research project, three studies were carried out to provide evidence-based recommendations to enhance the safety of heavy vehicles and save lives on Australian roads. In the first study, a crash severity model is developed to determine the variables influencing single-vehicle crashes involving heavy vehicles at intersections and mid-blocks. In the second study, a crash injury severity model is developed to determine the neighbourhood socioeconomic variables that influence injury severity in heavy vehicle collisions. Finally, in the third study, a crash injury severity model is developed to determine the causes contributing to injury severity in heavy vehicle angle collisions. In the first study, the objective was to identify the factors differentiating between single heavy vehicle collisions at intersections and mid-blocks using a binary logit model. The results show that single-vehicle crashes involving heavy vehicles at intersections are more likely to occur on main roads and highways, whereas crashes at mid-blocks are more likely to occur on higher speed roads, divided two-way roads, roads with special facilities or features (e.g. bridges), and roads with higher percentages of heavy vehicle traffic. Intersection crashes are also more likely to involve vehicles that are turning left or right, resulting in angle crashes, whereas mid-block crashes are more likely to involve vehicle overturning. The primary objective of the second study was to identify the neighbourhood socioeconomic characteristics affecting injury severity in heavy vehicle collisions. Specifically, the study explores the influences of the socio-demographic characteristics of the neighbourhoods where road users live and where the crashes occur. This study uses a multinomial logit model. In addition to neighbourhood socioeconomic variables, such as education, English language proficiency, occupation, income, and birthplace, other variables affecting road user injury severity, including environmental, temporal, road user, road, and vehicle characteristics, are considered as control variables. The results show that road users residing in neighbourhoods with more people born in Australia have higher injury severity, while road users living in neighbourhoods with more people with a university education and working in the sales profession have lower injury severity. Furthermore, crashes occurring in neighbourhoods with more people working as professionals are more severe. The findings present mixed results for the variables including technical education, clerical jobs and people born overseas for the neighbourhoods where the road users live, and variables such as people born in Australia, sales jobs and English language use for neighbourhoods where the crashes occur. The socio-demographic characteristics of the neighbourhoods where the road user resides and where the crash occurs contribute significantly to the road user injury severity in collisions involving heavy vehicles. It is important to emphasise that these neighbourhood socio-demographic characteristics should be used as a supplement to the information provided by the standard collision hotspot analysis. Finally, the main objective of the third study was to identify the factors contributing to injury severity in angle crashes involving heavy vehicles, in order to provide insights into improving traffic safety. The secondary objective was to compare the binary logic, skewed logistic (Scobit) and random parameters logit models in terms of their accuracy in identifying the factors contributing to injury severity in heavy-vehicle angle crashes. The findings indicate that the skewed logit model performs slightly better than the standard binary logit and mixed logit models in terms of the goodness of fit. The factors influencing injury severity in angle crashes involving heavy vehicles include occupants' gender, age, restraint use, and whether occupants are ejected or not, vehicle age, type, movement, fire status, point-of-impact and damage, time-of-day, road classification, posted speed limit and number of occupants involved in the accident. The output of this research will provide evidence-based recommendations to improve the safety of all road users, including heavy vehicle drivers on Australian roads. This study will also contribute to advancing knowledge in the field and will provide road safety professionals with more information and knowledge on the advantages of using statistical models, especially the Scobit model, in traffic safety studies