Comparing different methods for connecting bike lanes to generate a complete bike network and identify potential complete streets in Atlanta

Abstract

This study compares two different strategies for connecting bike networks – traditional design-based and algorithm-supported – to investigate how their results differ along metrics such as proportion of bike lanes along simulated routes and the resulting cycling stress. The objective is to find optimal strategies for connecting isolated existing cycling infrastructure to form complete networks that improve both active mobility and public transit ridership. By aligning the bike network with transit and activity locations, this research develops an algorithmic framework for generating a skeleton of multimodal networks best suited to become "complete streets." The network generated through an algorithm is compared with a proposed traditionally designed network to determine their relative network performance. The findings suggest that a judicious combination of traditionally designed, and algorithm-supported networks offer better cycling infrastructure than either strategy alone. In addition, algorithms can also be developed to indicate the potential for street segments to be complete streets.