Distribution network deployment for omnichannel retailing

International audience; This article studies the distribution problem of brick-and-mortar retailers aiming to integrate the online channel into their operations. The article presents an integrated modeling approach addressing the online channel-driven distribution network deployment (e-DND) problem under uncertainty. The e-DND involves decisions on operating fulfillment platforms and on assigning a fulfillment mission to those platforms, while anticipating the revenues and costs induced by order fulfillment, replenishment, delivery, and inventory holding. To model this problem while taking into account the uncertain nature of multi-item online orders, store sales, and capacities, a two-stage stochastic program with mixed-integer recourse is developed. Two alternative deployment strategies, characterized by allocation of orders, inventory positioning, delivery schema, and inbound flow pattern decisions, are investigated using this model. The first deployment strategy investigates the ship-from stores practice where the on-hand inventory is used for all sales channels. The second deployment strategy additionally considers the advanced positioning of inventory at a fulfillment center in the urban area where the online orders are requested. To solve the two-stage stochastic model with integer recourse, an exact solution approach combining scenario sampling and the integer L-shaped method is proposed. Numerical results, inspired by the case of a European retailer, are provided to evaluate the performance of the deployment strategies and the efficiency of the proposed solution approach.