Unraveling urban bike-sharing dynamics: Spatiotemporal imbalances in bike rentals and returns in Washington D.C.

Bike-sharing systems offer eco-friendly and flexible mobility solutions for short-distance travel but present new challenges such as device oversupply, random parking, and unmet demand caused by spatiotemporal imbalances between rentals and returns. While extensive research focuses on usage patterns of bike rentals alone, limited attention has been given to these imbalances and their relationships with various factors. This study investigated spatiotemporal rental-return imbalances using 2022 trip data from 356 bike stations in Washington D.C. Hourly station-level imbalances were calculated as the difference between bike rentals and returns and categorized as general or substantial (exceeding 10 bikes per hour). General imbalances displayed spatial concentrations in Columbia Heights (a mixed-use neighborhood) and Downtown (a central hub), but their temporal patterns were not notable. Substantial imbalances showed strong temporal trends: rentals exceeded returns primarily from 5 to 8 pm and fell short from 7 to 10 am, reflecting users' travel behaviors riding to these bike stations in the morning and departing from them in the evening. A Bayesian Additive Regression Trees (BART) model further identified distinct drivers: general imbalances were more found in census tracts with higher population densities, larger white percentages, diverse land uses, and poorly connected bike networks, while trip attributes including station-level rentals and returns, trip duration, and distance exhibited crucial nonlinear threshold effects on substantial imbalances. These insights guide targeted design, planning, and operational strategies for bike-sharing systems to address imbalances and enhance efficiency.