Spatiotemporal variability and prediction of e-bike battery levels in bike-sharing systems

TL;DR

This paper analyzes the spatiotemporal patterns of e-bike battery levels in Barcelona's bike-sharing system and introduces a Markov-chain model to predict bike availability and battery status, aiding operational management.

Contribution

It provides novel insights into e-bike battery dynamics and develops a predictive model to enhance fleet management in bike-sharing systems.

Findings

Bikes near city center have shorter rest periods and lower battery levels.

The Markov-chain model accurately predicts bike availability and battery levels.

Spatiotemporal analysis informs better operational strategies.

Abstract

Bike Sharing Systems (BSSs) play a crucial role in promoting sustainable urban mobility by facilitating short-range trips and connecting with other transport modes. Traditionally, most BSS fleets have consisted of mechanical bikes (m-bikes), but electric bikes (e-bikes) are being progressively introduced due to their ability to cover longer distances and appeal to a wider range of users. However, the charging requirements of e-bikes often hinder their deployment and optimal functioning. This study examines the spatiotemporal variations in battery levels of Barcelona's BSS, revealing that bikes stationed near the city centre tend to have shorter rest periods and lower average battery levels. Additionally, to improve the management of e-bike fleets, a Markov-chain approach is developed to predict both bike availability and battery levels. This research offers a unique perspective on the dynamics of e-bike battery levels and provides a practical tool to overcome the main operational challenges in their implementation.