Roadside clear zone width at expressway embankment section based on accident probability prediction

ObjectiveThis study investigates the roadside clear zone width to improve the roadside safety of cold-region expressways in icy and snowy conditions, especially the special condition of low friction coefficient on ice and snow pavement. It aimes to provide theoretical support and practical guidance for the design, operation, and management of clear zones along expressways in cold regions, enhancing the safety performance of roadside operations for cold-region expressways.MethodEight factors influencing roadside accident risks were considered, i.e., radius of circular curves, hard shoulder width, longitudinal slope, side slope ratio, and side slope height, road surface friction coefficient, vehicle speed, and vehicle type. 2 808 datasets were collected through PC-Crash low-friction simulations. The simulation data were fitted within the clear zone boundary, i.e., the lateral position where the vehicle decelerated to 40 km/h. A baseline model was established for calculating expressway roadside clear zone width in cold region in icy and snowy conditions. Additionally, Bayesian network model was developed to predict the roadside accident probabilities, and this model was then applied to accident loss calculations.ResultThe proposed Bayesian network model achieved a roadside accident prediction accuracy of 91.27%, effectively indicating the likelihood of roadside accidents. Using the simulated roadside clear zone width as the benchmark, the revised model yields a maximum relative error of 4.6% and an average relative error of 3.1%.ConclusionBy taking both accident loss and engineering cost into account, the baseline model was adjusted to form a revised model for calculating the expressway roadside clear zone width in cold regions. The area under curve reaches 0.921 9, and the prediction accuracy rate is about 91.27%, proving the effectiveness and accuracy of model in predicting probability of roadside accidents.