Citation

Santos MI, Oliveira PTMS, Larocca APC. Transp. Dev. Econ. 2022; 8(1): e10.

Copyright

(Copyright © 2022, Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s40890-021-00148-x

PMID

unavailable

Abstract

Researchers are constantly seeking for a quick, inexpensive, and easy-to-use way to mitigate the severity of road accidents. This study proposes the use of a simulator driver model, i.e., a dynamic vehicle model presented in a fractional factorial experiment for the identification of active factors in road accidents. The investigation was based on an accident database of a real two-lane highway that composed the scenario simulation, and tests were conducted in a 10-km path with 39 tangent-curve configurations. Seven factors, namely curve radius, path profile, path condition, virtual driver skill, speed, period of the day, and carload were considered and three replicates were performed per treatment. Speed and friction coefficient were randomly defined for each treatment within the range established for each level. The first experimental design was a quarter fraction, and 42 accidents were observed in 96 events. Speed exerted the highest influence on their occurrence, followed by curve radius, period of the day, and some second-order interactions. The second part of the experimental design was based on the results of the first. A half fraction factorial design with five factors and 14 replicates per treatment were performed, and 96 accidents were observed in 224 events. Speed exerted the highest influence on their occurrence, followed by period of the day, curve radius, virtual driver skill, and second-order interactions. The study has contributed to the understanding of the way drivers drive vehicles on a road with a small S-curve and to assessment procedures for analyses of risk factors and improvements in road safety.

Language: en