How do E-Bikes measure up? Analyzing speed differences and network impacts of São paulo’s bikesharing system

Abstract Speed is a pivotal factor in cycling mobility, influencing travel times, convenience, accessibility, and mode choice. However, empirical models of cycling speed, particularly for electric bicycles, remain limited. This study aims to examine the impact of infrastructure and topographic determinants on cycling speeds and network performance of electric bicycles, compared to conventional bicycles. Using a comprehensive GPS-based dataset from São Paulo’s Tembici bikesharing system, we employ linear mixed-effects models to analyze detailed variations in cycling speeds at the tracking point level, and linear regression at the trip level to assess overall trip dynamics. Novel variables, such as car traffic data from Uber, are incorporated, along with interactions between road segment length and cycling infrastructure type. Results show that electric bicycles consistently achieve higher speeds than conventional bikes, especially on challenging terrains and longer trips, due to motor assistance. Car speed significantly reduces speed for both conventional and electric bicycles. Segregated cycling infrastructure significantly increases speeds, with the greatest benefits observed for e-bikes on steep gradients. At intersections and traffic lights, e-bikes show greater sensitivity to stops in the segment-based analysis, but they recover more quickly at the trip level. E-bikes also led to a 14.2% improvement in the global efficiency of the cycling network, a measure of how effectively the network supports movement between all pairs of nodes, and our analysis of betweenness centrality showed that e-bikes change the importance of specific routes, making longer roads more critical. These findings underscore the importance of targeted investments in cycling infrastructure, particularly segregated lanes, to optimize e-bike performance, and support sustainable urban transport systems.