Estimating the safety benefit of separated cycling infrastructure adjusted for behavioral adaptation among drivers; an application of agent-based modelling

Separated cycling infrastructure is a key strategy employed by urban and transport planners to reduce car vs cyclist crashes.We constructed an agent-based model (ABM) to explore the potential effects of introducing progressively greater levels of saturation (e.g., more kms) of separated cycling infrastructure into a transport network in which drivers also demonstrated behavioral adaptation in response to increased exposure to cyclists as suggested by the safety in numbers (SiN) theory.The findings highlight that if behavioral adaptation among drivers is assumed to be a strong mechanism underpinning cyclist safety, the introduction of low levels of separated cycling infrastructure across a network (e.g., few kms) may provide little or no reduction in car vs cyclist crashes. This is due to the countervailing effects that separated infrastructure may have on drivers’ exposure to cyclists; a fundamental contributor to the concept of behavioral adaptation.This study demonstrates the utility of ABMs to explicitly define and model candidate behavioral mechanisms associated with cyclist and vehicle interaction when estimating the interaction of infrastructure and behavioral mechanisms proposed to underlie cyclist safety. Practically, it suggests that greater saturation of separated cycling infrastructure across transport networks may be required to reduce overall car vs cyclist crashes in circumstances where behavioral adaptation is also a strong mechanism contributing to cyclist safety.