An iterative decomposition algorithm for flexible two-echelon network design

This paper proposes a mixed integer programming model for a flexible two-echelon capacitated, multi-commodity, and multi-period network design problem. The model integrates several decisions of a supply chain and simultaneously plans production, inventory, location, and distribution. We consider a set of plants supplying intermediate facilities through which products are shipped directly to final customers. The model also includes real-world features of flexible delivery due dates and flexible location of the intermediate facilities. An iterative decomposition algorithm is proposed to solve this rich integrated problem. Three versions of the model are solved iteratively: two relaxation-based models and one restricted model. Solving the restricted model acts as a neighborhood search around the current best solution. Solutions obtained by the relaxation-based models guide the neighborhood search by identifying the binary variables to fix and at the same time promote diversity in the search. The results obtained by our computational experiments highlight the efficiency of the proposed method. Furthermore, managerial insights are presented on the utilization of flexibility in the obtained network design solutions, and how it relates to economies of scale.