We consider a routing problem where orders are transported just-in-time from several suppliers to an original equipment manufacturer (OEM). This implies that shipments cannot be picked up before their release date when they are ready at the supplier and should be delivered as close as possible to their due date to the OEM. Every shipment may have a distinct due date but all shipments loaded onto the same truck arrive at the same time. The performance of the transportation network is optimized by finding an allocation of shipments to trucks and routes for each truck that minimizes the total earliness-tardiness cost. These penalties are caused by deviations between the truck arrival times at the OEM and the due dates of the loaded shipments. To solve the problem, we introduce a metaheuristic approach based on large neighborhood search, which we combine with an efficient local search scheme that allows the evaluation of neighborhood solutions in worst-case logarithmic time despite the nonlinear objective function. Our algorithm can find high-quality solutions to large instances with 200 shipments in less than 12 minutes of CPU time. From a practical perspective, our computational tests indicate that a too small truck fleet or very limited time differences between release and due date can dramatically affect the punctuality of the deliveries.

我们考虑一个路由问题，其中订单从多个供应商及时运送到原始设备制造商 (OEM)。这意味着当货物在供应商处准备就绪时，不能在其发布日期之前取货，并且应尽可能在接近到期日期之前将其交付给 OEM。每批货物可能有不同的到期日，但装在同一辆卡车上的所有货物都会同时到达。运输网络的性能是通过找到卡车的装运分配和每辆卡车的路线来优化的，从而最大限度地减少总的提前-迟到成本。这些处罚是由于卡车到达原始设备制造商的时间与装载货物的到期日之间存在偏差造成的。为了解决这个问题，我们引入了一种基于大邻域搜索的元启发式方法，我们将其与一种有效的局部搜索方案相结合，该方案允许在最坏情况对数时间内评估邻域解决方案，尽管目标函数是非线性的。我们的算法可以在不到 12 分钟的 CPU 时间内为 200 个出货量的大型实例找到高质量的解决方案。从实际角度来看，我们的计算测试表明，卡车车队太小或发布日期和到期日期之间的时间差非常有限会显着影响交货的准时性。