Route selection guidelines and prioritization tools for efficient electrification of bus fleets

Abstract

The transition to electric buses (eBuses) is vital for reducing greenhouse gas emissions in public transit. However, challenges such as limited battery capacity and energy consumption necessitate careful route selection and fleet planning. This study analyzes transit route characteristics and employs machine learning models to predict fleet expansion needs. The analysis identifies that shorter routes with low speeds, stop density, and ridership are more suitable for initial eBus deployment as these routes are less likely to require fleet expansion. Moreover, eight binary machine learning classification models were developed to predict fleet expansion requirements. XGBoost model achieved the best performance with an accuracy of 0.84 and an AUC of 0.8457. Finally, SHAP analysis revealed route length, average temperature, and bus speed as key factors influencing fleet expansion. These findings offer practical guidance for transit agencies, enabling strategic route prioritization and effective eBus deployment with no fleet expansion till the technology matures.