|
Citation
|
La Torre F, Domenichini L, Branzi V, Meocci M, Paliotto A, Tanzi N. Accid. Anal. Prev. 2022; 178: e106852. |
|
Abstract
|
INTRODUCTION: In roadway safety management processes, Accident Prediction Models (APMs) represent the best available tools to analyse potential safety issues, to identify safety improvements and to estimate the potential effect of these improvements in terms of crash reduction. The Highway Safety Manual (HSM) provides consistent predictive methods for estimating the predicted average crash frequency, but an appropriate calibration is necessary to use them in contexts different from the ones where they were developed.
METHOD: The present study provides a contribution in this field of research providing a European APM based on the one proposed by HSM and introducing a new methodology to transfer the HSM to different European rural freeways. Specifically, a new set of jurisdiction-specific (JS) base safety performance functions (SPFs) have been developed as a function of annual average daily traffic volume and roadway segment length, considering JS base conditions specific for each different national network, different from the HSM base conditions. These new SPFs were then used for the calibration of the full models, and the results compared with those obtained applying the HSM predictive model. This allows to evaluate the potential benefits of calibrating jurisdiction-specific base SPFs, with different base conditions.
RESULTS: The results showed that the local SPFs development approach allowed to obtain a better fit than the HSM predictive model for the analysed European countries.
CONCLUSIONS: The findings suggest that the development of jurisdiction-specific base SPFs will offer more suitable APMs for countries that differ from the USA, thus a more reliable prediction could be obtained by applying this procedure. PRACTICAL APPLICATIONS: The use of JS SPFs allowed developing European APMs based on those proposed by HSM but applicable to different conditions. This represents a useful starting point for further analysis and improvements in accident prediction modelling.
Language: en |