Distribution network redesign under flexible conditions for short-term location planning

E-commerce has experienced a remarkable growth in recent years. This trend is gradually driving retailers to shift from typical warehouse rental contracts with rigid conditions and long-term commitments to flexible alternatives that provide storage space on demand. This strategy gives retailers the ability to respond dynamically to variations in demand, thereby improving customer service. In this paper, we address this trend by considering a two-echelon, multi-commodity distribution network operated by a retailer and whose configuration can be changed over a multi-period planning horizon. Location decisions concern the intermediate facilities and are framed by flexible and scalable conditions. Accordingly, the retailer needs to decide which company-owned facilities should be retained, and where, when and for how long new facilities should be leased from a set of potential sites with given capacities. In this setting, a facility can be leased at the same location for several non-consecutive periods. In addition to location decisions, also procurement, inventory, and distribution decisions need to be made to satisfy customer demand at minimum total cost. We develop a mixed-integer linear programming formulation for this problem and propose various families of additional inequalities that prove to be very useful to find high-quality solutions. To measure the impact of the new business model, we also consider alternative approaches with limited flexibility and scalability that are frequently encountered in practice. Our computational study gives useful insights on the trade-offs achieved by each of the different approaches with respect to total cost, network redesign decisions, and various logistics functions.