Refining a crash risk framework for urban bus safety assessment: Evidence from Sardinia (Italy)

Assessing the crash risk on bus routes would be pivotal to increasing the transit system safety. All past studies largely focused on the proposal of some crash risk indexes, and only one (recent) research has computed the bus crash risk in transit services by developing a bivariate (frequency and severity) risk model. Conversely, as far as the authors’ knowledge, no research attempted to quantify the probability of bus crash occurrence as a risk component and present results by geographic information systems. This study covers these gaps by refining an existing framework to better adhere to the original and well-accepted definition of risk. Specifically, it models the bus crash risk on (parts of) routes as a function of the probability of having a crash, the related severity, and exposure terms. Next, (parts of) routes are ranked according to the risk value and classified by a multilevel scale, to show those with the highest risk. Experiments highlight the viability of this framework using 3000+ raw bus crash data records. The results are synthesised by straightforward charts and maps. This framework helps support public transport companies in enhancing safety performance because it could be implemented in a transit safety managerial system for monitoring and certification purposes according to safety norms and EU directives.