B. Persaud, T. Saleem, M.E. Granados, T. Rajeswaran
Crash modification factors (CMFs) are needed for estimating safety benefits in cost-benefit analysis and project prioritization in planning and designing road infrastructure treatments. In fulfilling this need, there are limits to what can be achieved with traditional crash-based analytical methods, and research returns in this area are rapidly diminishing despite a dearth of knowledge. In particular, CMFs that vary by application circumstance to facilitate transferability and evaluation of individual sites continue, by and large, to be elusive, as are CMFs for treatments that are applied in combination. Some of these knowledge gaps are likely to occur in the future as traffic flow and safety on roads become increasingly impacted by the various levels and mixtures of vehicle automation that now seem inevitable in the not so distant future. This paper is based on relatively recent research that focused on application of surrogate measures to address these knowledge gaps by better estimating surrogate measures from microsimulation and prediction models and established robust statistical relationships between them and crash frequency, from which CMFs that vary with application circumstance for single as well as combination treatments could be inferred. The main objective of the paper is to consolidate the recent research that established relationships between surrogate measures and crashes for various site types in demonstrating the potential of using this knowledge for estimating crash modification factors and functions that could not easily be estimated from traditional crash-based methods.
Keywords: crash modification factors; microsimulation; automated vehicles; surrogate measures; conflicts