Abstract

Battery swapping for refueling has been embraced by major electric vehicle (EV) manufacturers. To facilitate the operations of battery swapping, this paper studies the management of a battery swapping network that operates in a ”swap locally, charge-centrally” way: batteries are swapped at local swapping stations and transported to a central charging station for recharging. It remains unclear how to replenish batteries from central charging stations, fulfill uncertain demand, and trans ship batteries among swapping stations to guarantee a high service level. In this paper, we exploit the network as a repairable inventory management problem with lateral transshipment. Combining the ideas from process flexibility, we first develop a stochastic dynamic program to characterize coupling network operations. Then we introduce a novel robust satisficing model from a target-oriented perspective to address the curse of dimensionality and distributional ambiguity. Specifically, we first employ a utility-based Service Level Measure by constraining the severity of inventory violations caused by uncertain demand. We solve this problem by leveraging an event-wise ambiguity set and a pooling-group-based enhanced linear decision rule. Out-of-sample tests demonstrate that our model outperforms several benchmarks in terms of robust ness and target attainment. The major insights are twofold: (i) Incorporating lateral transshipment in the service network reduces substantial costs compared to a system without transshipment. Furthermore, a chaining network captures most of the bene f its of a fully flexible network, and (ii) increasing the cost budget can enhance the robustness, whereas increasing the penalty cost reduces the probability of inventory violation by building more battery inventory.

Key words

Battery swapping, Adjustable robust optimization, Robust satisficing, Distributionally robust optimization, Lateral transshipment