Exploring new methodologies and perspectives on the road safety impacts of bus priority

Road accidents have and will remain a major concern as cities around the world continue to grow. The safety problem is likely to worsen as population growth is accompanied by increased travel. For many cities, these trends have led to a greater provision of public transport as private vehicles become a less viable mode of transport. With the rise in public transport travel, it comes as no surprise that road management agencies are turning to an increased application of priority measures to improve the travel experience for commuters. For buses, the provision of priority measures has typically been justified based on travel time savings and operational benefits. Although recent years have seen the advances in research valuing the wider ridership, mode shift and environmental benefits of bus priority schemes, including the network wide benefits, a major issue that has yet to be considered in bus priority planning is the road safety impacts of providing priority schemes. This thesis therefore aims to develop an in-depth understanding of the road safety implications of implementing bus priority through an investigation of accident records and conflicts in Metropolitan Melbourne. It is structured around six approaches that had been established to fill the knowledge in the area of bus priority safety effects. Each approach is the focus of a thesis chapter where the research context is discussed before the research methodology is presented. Results and key findings that emerged from subsequent analyses were used as a basis to understand the implications of implementing bus priority in the context of bus priority planning and research. The first approach concerns an exploration of the safety effects of bus priority at the aggregate level. Here, a before-after safety evaluation of both “space based” and “time based” bus priority was carried out to understand its effects at the network and bus route levels. A before-after accident type analysis was done to examine whether accident counts or nature of accidents had changed following the implementation of bus priority. Results of the safety evaluation based on the Empirical Bayes approach showed that the implementation of bus priority treatments led to a 14% reduction in accidents. “Space based” treatments (mainly bus lanes) yielded a stronger positive safety effect (18.2%) compared to “time based” ones (11.1%). In terms of fatal and serious injury accidents, a drop of 42 to 29 per annum was recorded. Given that different design types are available in before-after safety evaluation, the second approach focuses on understanding how the choice of comparison group type affected the bus priority safety estimate. Using the Empirical Bayes (EB) and Comparison Group (CG) approaches, it was found that the effect of using different comparison group types led to discrepancies in the final safety estimates. It is likely that these differences were due to the (necessary) omission of sites with zero accident history and effect of matching treatment sites with similar sites in the CG approach. A new approach that combined both EB and CG results showed promise as a more precise safety estimate was obtained. The third approach relates to an investigation of bus accidents at the route level. Using two mainstream modelling methodologies (MENB - Mixed Effects Negative Binomial and BPNN – Back Propagation Neural Network modelling), risks factors in bus accidents were explored with particular attention paid to the safety effect of bus priority. Results showed that bus priority led to lower occurrence for certain accidents types. The MENB and BPNN model results showed that bus priority had the effect of reducing route section level accident frequency by about 53.5%. The MENB model recorded better performance which pointed to benefits in adopting the MENB approach to account for time- and location-specific effects in accident count modelling. The fourth approach concerns the analysis of bus accidents in terms of vehicle, driver, roadway and environmental factors. This was done to identify the significant risk factors in a bus company database of accidents where bus drivers were deemed to be at-fault. Similar to the third approach, the aim was to understand the effect of bus priority on drivers’ at-fault probability in bus-involved accidents. Results from mixed logit modelling showed that bus length / age, driver’s gender / age / experience / accident record, road type, speed limit, traffic / daylight conditions, and the presence of bus priority affect the likelihood of bus drivers being at-fault in bus-involved accidents. For bus priority, the effect was found to be random as bus priority only reduced the at-fault likelihood for some 57.8% of drivers. The fifth approach centres on an investigation of the bus priority effect (bus lanes) at a corridor-level through micro-simulation. The focus was on conflicts at intersections and bus stops as the introduction of bus lanes was expected to have most impact on traffic movements at these locations. Results showed that the provision of bus lanes, regardless of whether they are created through space reallocation or creation, lead to a reduction in conflicts at intersections and bus stop locations. These pointed to lower rear-end and lane change accident risks for vehicles when bus lanes are in place. The sixth approach concerns an estimation of crash risk for vehicles that are behind a slowing or stationary bus at a bus stop in a mixed traffic configuration. This was done to quantify the safety benefit delivered by bus priority schemes that segregate buses from mainstream traffic. Using recorded travel behaviour and accident history of a representative road corridor, the average crash risk of vehicles that were in conflict with buses was found to be 0.0154% (with a standard error of 0.0063%). Based on the assumption of an average of thirty such conflicts occurring daily, it works out that there is an approximate 80% chance of one or more accidents taking place annually as a result of buses slowing down or being stationary at bus stops. Overall the thesis presents a range of advances in knowledge in the area of bus priority. Through the six approaches, new light has been shed on the safety effects of bus priority. The thesis concludes with a synthesis of the findings, in which its implications in the context of bus priority research and planning as well as opportunities for future research are presented.