To address the rising issue of cyclist safety at signalized intersections, establishing therelationship between motor-vehicle and cyclist flows, geometric design and built environmentcharacteristics is fundamental. To date, there is very little empirical evidence of the impact ofthese factors on cyclist injury occurrence and on the most effective countermeasures in NorthAmerica. In this regard, this research (i) proposes a new approach to measure risk exposure and(ii) develops cyclist injury frequency models. Three separate definitions of risk exposure areused: completely aggregated flows, motor-vehicle flows aggregated by movement type andpotential conflicts between motor vehicles and cyclists. As an application environment, a sampleof signalized intersections on the island of Montreal was used, as well as data comprising ofdisaggregate motor-vehicle and cyclist flows. Several negative binomial models were fitted tothe data. Among the results, this study shows that cyclist collisions are sensitive to changes incyclist flows: a 10% increase in bicycle flow is associated with a 5.3% increase in the frequencyof cyclist injuries. Surprisingly, motor-vehicle volumes at the intersection level do not have asignificant effect on cyclist collision frequency. However, when motor-vehicle flows areconsidered based on movement type, it becomes apparent that right turn movements have a greateffect on accident occurrence. Similar results, indentifying right turns as having the greatesteffect on cyclist injuries, are achieved when potential conflicts are considered. It was also foundthat the number of bus stops in the proximity of intersections increases cyclist injury occurrence.Some geometric design factors such as presence of median and three-leg intersections weretested; however, their effect was found to be statistically non-significant.
展开▼