New methods for modeling and integrating bicycle activity and injury risk in an urban road network

by Jillian Strauss

Institution: McGill University
Year: 2016
Keywords: Civil Engineering & Applied Mechanics
Posted: 02/05/2017
Record ID: 2124092
Full text PDF: http://digitool.library.mcgill.ca/thesisfile138869.pdf


Despite the fact that the bicycle modal split is low, an increase in urban cycling activity has started to appear in many urban areas in Canada and the United States. However, as bicycle flows increase so do concerns for cyclist safety which have been pushed to the forefront. In this context, the need for new planning tools and data collection methods to investigate bicycle activity and safety have emerged in the field of transportation engineering. Studies focusing on investigating the link between cyclist safety, geometric design and built environment characteristics as well as traffic conditions are rare in the current literature. Even rarer are studies focused on investigating safety across different facilities, for the different road users who share these facilities and the factors affecting their injury occurrence. Tools to estimate bicycle volumes at the entire network level are missing. All previous safety studies have considered a sample of sites and have not been able to compute and map risk in the entire network of intersections and road segments. Dangerous decelerations as a surrogate safety measure for cyclists has yet to be investigated for the entire network. To date little is known about cyclist speeds and travel times along segments and delays through intersections at the disaggregate level for the entire network. In order to address these limitations, the general objective of this thesis is to propose new methods to model and estimate bicycle activity and injury risk at different spatial levels combining different sources of data. The proposed methods are then used to identify risk factors as well as to map risk indicators based on accidents and hard braking along corridors and at intersections in the entire network. More specifically, the objectives of this thesis are, to: 1) develop a Bayesian modeling framework to simultaneously model injury and activity outcomes for cyclists and study the role of geometric design and built environment characteristics on both outcomes, 2) carry out a comparative analysis between the injuries, levels of flow and risk for the three modes at signalized and non-signalized intersections and investigate the impact of vehicle traffic on the safety of non-motorized modes, 3) improve current methods for determining bicycle exposure measures by combining manual counts, automatic counts and GPS trip data to estimate and map bicycle flows, injuries and risk throughout the entire network of intersections and road segments, 4) develop a methodology to obtain deceleration rate for cyclists at intersections and segments using GPS data, explore the relationship between observed injuries and deceleration rate, and 5) develop a methodology to estimate cyclist speeds, travel times and delays in a road network using GPS data and identify the factors affecting cyclist speeds along segments. Among the contributions, this thesis proposed a modeling framework to simultaneously study cyclist injury occurrence and bicycle activity to overcome the issues of endogeneity. A similar… Advisors/Committee Members: Luis Miranda-Moreno (Supervisor1), Patrick Morency (Supervisor2).