The green repositioning problem in the bike sharing system

Abstract

Bike Sharing Systems (BSSs) and environmental concerns have been receiving increasing popularity in transportation operations. In the BSSs, the distribution of bike demand often mismatches with the bike supply due to the asymmetric bike flows. Bikes need to be redistributed among stations to mitigate this mismatch using a dedicated fleet of vehicles, and the problem of determining the vehicle route and loading instructions is referred to as the bike repositioning problem (BRP). However, the fossil-fueled vehicles used in the BRP produce pollutants and greenhouse gases that do harm to the environment and damage the environmental creditability of BSSs. This environmental impact of bike repositioning has not received enough attention. In addition, the presence of broken bikes in the BSS not only causes user dissatisfaction but also wastes resources. The collection of broken bikes for either maintenance or recycling can save resources, making the BSS more environmentally friendly, but has rarely been noticed. Therefore, this study aims to capture the environmental issues in the BRP and develop green BRPs. This study looks into three different BRPs. The first one is a static BRP with two types of bikes (i.e., usable and broken bikes) using traditional internal combustion engine vehicles (ICEVs) powered by fossil fuels. The objective is to minimize the CO2 emissions of vehicles while achieving a perfect balance between the demand and supply of bikes at each station. A mixed integer linear program model is presented to formulate the problem and a commercial solver was used to solve it. The factors that affect the CO2 emissions are discussed. After that, this problem was extended through relaxing the perfect balance constraint and setting a time budget. The objective is to minimize the weighted sum of the CO2 emissions, the deviations from target inventory levels, and the penalty for uncollected broken bikes. An enhanced artificial bee colony algorithm was adopted to generate the vehicle route, together with two methods for computing the loading/unloading quantities at each stop. The effectiveness and efficiency of the proposed method are discussed based on the computational experiments conducted on the real-world BSS instances. Finally, a dynamic BRP using battery electric vehicles (BEVs) was examined. The objective is to minimize the weighted sum of the penalty cost for unmet demand and the charging cost of vehicles. A rolling horizon framework was adopted to incorporate the revealed inventory levels at bike stations and the vehicle load of bikes. The artificial bee colony algorithm with an embedded dynamic programming method for computing the loading instructions was proposed to generate solutions to the subproblem in each horizon. Computational experiments were conducted on a real-word BSS and problem properties were analyzed from three aspects: the horizon length, different penalties for failed rentals and returns, and the charging related settings. This thesis provides practical insights to the operators into reducing the CO2 emissions of the ICEVs and the employment of BEVs in the daily repositioning work of BSSs, making the BSS a greener transport mode to the public.