Abstract The Electric Vehicle Routing Problem with Time Windows and Stochastic Waiting Times at Recharging Stations is an extension of the Electric Vehicle Routing Problem with Time Windows where the electric vehicles (EVs) may wait in a queue before the recharging service starts due to limited number of chargers available at stations. Since the customers and the depot are associated with time windows, long waiting times at the stations in addition to the recharging times may cause disruptions in logistics operations. To solve this problem, we present a two-stage simulation-based heuristic using Adaptive Large Neighborhood Search (ALNS). In the first stage, the routes are determined using expected waiting time values at the stations. While the vehicles are following their tours, upon arrival at the stations, their queueing times are revealed. If the actual waiting time at a station exceeds its expected value, the time windows of the subsequent customers on the route may be violated. In this case, the second stage corrects the infeasible solution by penalizing the time-window violations and late returns to the depot. The proposed ALNS applies several destroy and repair operators adapted for this specific problem. In addition, we propose a new adaptive mechanism to tune the constant waiting times used in finding the first-stage solution. To investigate the performance of the proposed approach and the influence of the stochastic waiting times on routing decisions and costs, we perform an experimental study using both small and large instances from the literature. The results show that the proposed simulation-based solution approach provides good solutions both in terms of quality and of computational time. It is shown that the uncertainty in waiting times may have significant impact on route plans.

中文翻译：

---

具有时间窗和充电站随机等待时间的电动汽车路线问题的基于仿真的启发式算法

摘要 带时间窗和充电站随机等待时间的电动汽车路径问题是带时间窗的电动汽车路径问题的扩展，其中电动汽车（EV）可能在充电服务开始前排队等候，因为数量有限。车站有充电器。由于客户和仓库与时间窗口相关联，除了充电时间外，车站的长时间等待可能会导致物流运营中断。为了解决这个问题，我们提出了一种使用自适应大邻域搜索 (ALNS) 的基于两阶段模拟的启发式方法。在第一阶段，路线是使用车站的预期等待时间值确定的。当车辆跟随他们的旅行时，到达车站后，会显示他们的排队时间。如果站点的实际等待时间超过其预期值，则可能会违反该路线上后续客户的时间窗口。在这种情况下，第二阶段通过惩罚时间窗口违规和延迟返回仓库来纠正不可行的解决方案。提议的 ALNS 应用了几个适合这个特定问题的破坏和修复操作符。此外，我们提出了一种新的自适应机制来调整用于寻找第一阶段解决方案的恒定等待时间。为了研究所提出方法的性能以及随机等待时间对路由决策和成本的影响，我们使用文献中的小型和大型实例进行了实验研究。结果表明，所提出的基于仿真的解决方案在质量和计算时间方面都提供了良好的解决方案。结果表明，等待时间的不确定性可能会对路线计划产生重大影响。