Flexibility and consistency in inventory-routing

In many contexts, logistics is used to enable competitive advantages and cost savings. For some companies, logistics itself is its core competency (i.e. logistics providers). In this context, vendor-managed inventory (VMI) systems are one of the most up-to-date strategies allowing companies to reach a superior performance. Under a VMI strategy, the replenishment and distribution making process is centralized at the supplier’s level, leading to an overall reduction of logistics costs. In order to operate a VMI system, an Inventory-Routing Problem (IRP) has to be solved, simultaneously making inventory management and routing decisions over several periods. Our purpose is to introduce two new concepts, called flexibility and consistency, within the context of the IRP. Flexibility will be added through the possibility of sharing inventory among locations, making the concept of transshipment available within inventoryrouting. It is also useful to react quickly to changes in the demand in a dynamic and stochastic environment. Transshipment problems are typically characterized by movements of goods among entities of the same level, such as customers. They allow the system to share stockout risks and to increase the flexibility of the decision maker by increasing the number of sources from which goods can be transferred. We then introduce the Inventory-Routing Problem with Transshipment (IRPT), a problem in which the decision maker has the option to plan transshipment movements so as to minimize the total system cost. This problem arises, for instance, when solving stochastic Inventory-Routing Problems (SIRP) in a rolling horizon framework where one uses demand forecasts for the next time periods as approximations of the unknown demand. We present a formulation that allows transshipments, either from the supplier to customers or between customers. We also propose an adaptive large neighborhood search heuristic to solve the problem. This heuristic manipulates vehicle routes while the remaining problem of determining delivery quantities and transshipment moves is solved through a network flow algorithm. Our approach can solve four different variants of the problem: the IRP and the IRPT, under maximum level and order-up-to level policies. We perform an extensive assessment of the performance of our heuristic. Consistency will help offer higher quality of service, benefiting both supplier and customers with more regular services. We also propose the inclusion of consistency features within the IRP framework. They can be used