集中充电式换电网络中, 充换电站的布局对电动汽车补能服务能力影响显著。现有换电网络规划多侧重于设施选址和建设成本, 较少考虑电池配送带来的时空资源约束和对网络运营成本的影响。针对换电站和集中充电站构成的集中充电式换电网络的运行和规划问题, 该文首先提出一种考虑电池动态配送的换电网络联合规划方法, 并构建了以运营商和用户综合成本最小化为目标的充换电站联合布局规划模型; 其次, 通过建立配送时效与电池动态供给的量化关系, 构建了包含电池充电、换电和配送过程的充换电站运行状态协同仿真模型, 并将其嵌入联合布局规划模型, 实现选址结果与运营调度的动态关联; 最后, 以天津市滨海新区为例, 验证了该模型的有效性。研究结果表明：相比于传统规划方法, 该文所提方法在换电网络规划阶段, 通过对电池充电、换电和配送等运营环节的建模和仿真, 使换电网络综合成本降低了5.98%, 电池采购数量减少了13.60%。该文研究结果可为集中充电式换电网络的规划决策、电池资源优化配置和换电站运营管理提供参考。

Objective: The rapid adoption of electric vehicles has highlighted the urgent need for efficient and reliable recharging infrastructure. Battery swapping technology offers advantages over traditional plug-in charging, including faster turnaround times, grid-friendliness, and improved space efficiency, and has garnered increasing attention. A centralized-charging battery swapping network comprising swapping stations (BSS) for battery replacement and centralized-charging stations (CCS) for dedicated battery charging can optimize grid load distribution and reduce infrastructure costs. Nonetheless, existing planning approaches focus on facility siting and construction costs, largely overlooking spatiotemporal resource constraints caused by battery distribution and its impact on network operation costs. This oversight results in suboptimal resource allocation, increased operational costs, and conflicts between operator investments and user service quality. To address these challenges, this study proposes a joint planning method for centralized-charging battery swapping networks. This approach integrates dynamic battery logistics with infrastructure siting, aiming to minimize the total cost, which comprises infrastructure, operational expenses, and user time losses. Methods: This study develops a mixed-integer optimization model to formalize the co-location and capacity planning of BSSs and CCSs. The objective function minimizes the annual comprehensive cost, which includes construction and equipment costs for CCSs and BSSs, battery procurement and logistics costs, and user-related costs such as taxi operational losses due to travel, queuing, and swapping times. Constraints include proximity-based demand allocation using Voronoi partitioning, maximum queue length limits to ensure service quality, and a CCS-BSS linkage that ensures each BSS is served by its nearest CCS. A nested simulation framework couples planning with dynamic operations to capture the operational intricacies. Battery logistics are modeled as a multi-vehicle routing problem with hard time windows, which is solved using insertion heuristics after virtual-node transformations to accommodate dynamic delivery requests. Delivery costs include distance-based fuel and lease expenses. CCS charging follows the "shortest time charge first" scheduling, with charger counts derived from daily demand and charging rates. Battery inventories are updated through the operational simulation of BSS and CCS, which calculates minimum procurement thresholds based on non-negative stock levels. The model utilizes an elite-preserving genetic algorithm to optimize siting decisions, iteratively refining planning based on simulation feedback, including delivery costs and battery inventory requirements. Results: The framework was applied to a case study in Tianjin Binhai New Area, where the optimal configuration consisted of 7 BSSs and 2 CCSs. The cost breakdown revealed that battery procurement was the dominant expense, followed by infrastructure and user time losses, while logistics costs were minimal. Model validation against static logistics baselines demonstrated a 13.60% reduction in battery procurement and a 5.98% decrease in total cost, resulting from the integration of planning and dynamic logistics. Sensitivity analysis revealed that the battery configuration at swap stations, including battery reserve quantity and delivery request thresholds, had a significant impact on the operation of the entire battery swapping network. Additionally, the maximum queue length constraint balanced service level and station construction cost; a smaller queue length required more stations and higher costs, while the total battery purchase quantity varied with queue length to maintain service levels. Conclusions: This paper integrates the operation and planning of battery swapping and CCS into a unified model that dynamically links site selection with battery delivery costs and user time loss. This approach addresses the shortcomings of traditional planning, which often overlooks delivery costs and the quantity of batteries purchased. Experimental results show a reduction in battery purchases and compressed cost compared to plans that ignore dynamic battery delivery, demonstrating its effectiveness in resource optimization and cost control. Additionally, this study's detailed co-simulation model, which captures battery charging, swapping, and delivery processes, enables multidimensional coordination of charging scheduling, battery reserves, and delivery route planning. Moreover, sensitivity analysis confirms that considering dynamic delivery guides reasonable site selection and resource allocation, while further controlling the scale of battery purchases and reducing comprehensive costs.