Optimizing the Strategic Decisions for One-Way Station-Based Carsharing Systems: A Mean-CVaR Approach

Abstract

The study focuses on the strategic decisions including on the location and capacity of stations and the fleet size for designing the one-way station-based carsharing systems. Under demand uncertainty, we introduce a two-stage risk-averse stochastic model to maximize the mean return and minimize the risk, where the conditional value-at-risk (CVaR) is specified as the risk measure. To solve the problem efficiently, a branch-and-cut algorithm and a scenario decomposition algorithm are developed. We conduct computational experiments based on historical use data and generate efficient frontiers so that the system operator can make a trade-off between return and risk. We then utilize an evaluation method to analyze the necessity of introducing risk. Finally, the efficiency of the proposed algorithms is elaborated through comparative experiments. Both branch-and-cut algorithm and scenario decomposition algorithm can tackle the small- and medium-scale problems well. For large-scale problems that cannot be solved by using an optimization solver or the branch-and-cut algorithm, scenario decomposition method can provide favorable solutions within a reasonable time.