As urban areas strive to reduce greenhouse gas emissions, bike-sharing systems emerge as a promising solution to decrease private vehicle reliance. This study presents a robust methodology to quantify the emission reduction potential of bike-sharing systems in Vancouver, focusing on three core steps: trip purpose classification, mode substitution simulation, and emission factor estimation. Trip purposes were classified using k-means clustering, distinguishing between commuting and leisure trips. Mode substitution was then simulated using a graph neural network trained on trip diary data, allowing for accurate predictions of traditional modes replaced by bike-sharing, such as cars, transit, and walking. Finally, emission reductions were quantified using MOVES, a comprehensive emission inventory model, accounting for road types, fleet composition, and temporal variations.
The mode substitution analysis revealed that bike-sharing systems predominantly replace car trips, with 54% of trips previously made by car, while 14% were from transit and 32% were walking trips. The results indicate significant environmental benefits, with each bike saving approximately 119 kg of CO2, 1.26 kg of CO, 0.08 kg of NOx, and 1.24 g of PM2.5 annually. These findings highlight the effectiveness of bike-sharing systems in reducing urban transportation emissions and emphasize their role as a cost-effective decarbonization strategy.
