Exploring Tram Road Safety with a Focus on Tram Priority

The demand for public transport is increasing in many cities due to population growth and higher pressure on limited road space. With the rise in public transport travel demand, road management authorities are choosing to implement preferential treatments for on-road public transit vehicles, known as transit priority. Tram priority is one form of on-road transit priority implemented in different parts of the world and takes a wide variety of forms. Regardless of the particular configurations, the aim of tram priority is to decrease tram travel times and improve service reliability. Previous studies have explored the operational benefits of tram priority measures. However, the implementation of tram priority measures adjusts the nature of road spaces and can have road safety impacts. The literature suggests that only a handful of studies have examined the safety impacts of tram priority, and those were largely subjective with mixed and incomplete results. In addition, limited clarity is provided by previous studies on the risk factors including tram priority associated with tram-involved crash frequency and severity. <br> With these current knowledge gaps in mind, the aim of this research is to explore the road safety impacts of implementing tram priority measures on the Melbourne tram network. To attain the research aim, five research components (or tasks) are defined. Each component is the focus of a thesis chapter where detailed research context, methodologies and the key findings are presented. <br> The first task focused on evaluating the overall safety effects of tram lane and signal priority measures in Melbourne using the Empirical Bayes (EB) before-after safety evaluation approach. The results showed that tram lane priority acted to reduce crashes for all road users by 19.4% and signal priority measures acted to reduce crashes by 13.9%, with a combined safety benefit of 16.4%. This task also investigated the changes in specific crash types before and after priority implementation. The outcomes indicated that tram lane and signal priority measures were effective in reducing vehicle- and pedestrian-involved all injury crashes by 17% and 19% respectively. In addition, the results showed reductions in on-path collisions (64%), and collisions among vehicles moving in both the same (12%) and opposite directions (23%). <br> The second task focused on evaluating the safety impacts of the replacement of older design tram stops with platform stops, using a Comparison Group (CG) before-after safety evaluation approach. This task only focused on pedestrian-involved collisions, as previous research has identified that these comprise the majority of crashes at tram stops. The results showed that pedestrian-involved all injury crashes reduced by 43% after platform tram stop installation. This task also adjusted the conventional CG analysis outcomes to consider differences in passenger volume at the treatment and comparison sites. The adjusted results suggested greater reductions in pedestrian-involved crash rates (crash counts per 10,000 passengers) after platform stop installation: an 81% reduction in pedestrian-involved all injury crashes and 86% reduction in pedestrian-involved fatal and serious injury crashes per 10,000 passengers. <br> The third task focused on investigating tram safety at the route section level using the Random Effects Negative Binomial (RENB) model. The key aim of this task was to investigate the impact of all forms of tram priority measures on tram-involved crashes. In addition, this task explored other key traffic, transit and route factors influencing tram-involved crashes. The results of the RENB model indicated that tram route sections with tram lane priority experienced 14% fewer tram-involved crashes than on routes without tram lane priority. Tram routes with a larger proportion of signalised intersections with signal priority treatments and a higher proportion of platform stops had less crashes. Results also showed that tram service frequency, stop spacing, route section length, traffic volume, and tram travel speeds significantly affected tram safety. <br> The fourth task involved identifying factors influencing the probability of tram-involved serious injury crashes using a logistic regression modelling approach, with a particular attention given to explore the impact of tram priority treatments. The results showed that the probability of serious tram-involved crashes was 41% higher on tram routes with tram lane priority than on tram routes without lane priority. This research component also identified that low floor trams, older trams, lower traffic volume and higher average tram travel speeds were more likely to increase the incidence of serious tram-involved crashes. Also, serious tram crashes were less likely to occur during winter and autumn on the Melbourne tram network. <br> The final task investigated the detailed road user and road design factors influencing tram safety from a tram driver viewpoint by conducting tram driver focus groups. The key focus of this task was to identify how tram driver safety perception varies between routes with and without tram priority features. The outcomes from the focus groups showed general agreement with the analytical study outcomes obtained from the first four tasks. In particular, physical separation of tram right of way, the presence of platform tram stops and provision of the ‘hook turn’ signal treatment at intersections were perceived as the more safety effective tram priority measures by Melbourne tram drivers. In addition, tram drivers observed that other road users were unaware of the safety issues around trams, had a poor understanding of the road rules applicable to sharing the road with trams and often violated tram-related road rules. These were identified as key road user factors for tram-involved crashes by the tram drivers. <br> In summary, this thesis provides a number of original contributions to knowledge in the field of tram road safety with a focus on tram priority through adopting more robust methodological approaches than have been previously applied to this research area