Application of a Crash-predictive Risk Assessment Model to Prioritise Road Safety Investment in Australia

Australia experiences many similar strategic road safety challenges as most European countries. These include the objective to strongly reduce fatalities and serious injuries, budgetary constraint requiring prioritisation of road investment, and the growing need for improved integration of road transport to drive efficiency. The current National Road Safety Strategy 2011-2020 aims for a 30% reduction in fatal and serious injuries, a step on a path towards the Safe System (Vision Zero). This aim is made more challenging by the geographically scattered nature of fatal and serious (severe) crashes on the road network, especially on routes with moderate traffic flows or in regional areas. This problem has led to reducing economic returns from conventional road safety initiatives based on treatment of crash-cluster locations or lengths. This paper shows how the Australian National Risk Assessment Model (ANRAM) addresses this challenge and assists state road agencies to meet the aims of the national strategy. ANRAM is used by the agencies to assess the risk of future severe crashes across their road networks. It is then used to prioritise those sections based on severe crash estimates, less susceptible to random variation (the scatter). The model facilitates creation of strategically-aligned infrastructure investment programs to reduce the severe crash risk and to estimate future crash savings. These are used together with program capital costs to carry out economic analysis and to support funding decisions. The paper outlines how ANRAM uses a crash-predictive approach to first estimate mean severe crashes per road section, by crash type. Then, it shows how ANRAM adjusts the estimate using risk algorithms which build on the International Road Assessment Program (iRAP) protocols. These algorithms use detailed road attribute data and research evidence about crash risk potential associated with road design/infrastructure, traffic speeds, and likelihood of vehicle conflicts. Finally, the model uses historical severe crash data in an Empirical Bayes validation technique to provide additional confidence in the estimates. ANRAM has enabled Australian road agencies to scope and prioritise proactive road safety investment options before severe crashes form a historical data clusters. The paper presents several examples of recent programs which reflect jurisdictional priorities, unique local conditions and available resources. The paper concludes with a discussion on how the approach used in ANRAM could have relevance to effective programming of road safety investment for European road agencies. It also suggests additional benefits beyond road safety, e.g. though providing inputs into road data inventories used in transport modelling. Language: en