This study proposes a production hub location (PHL) problem that integrates the classical multiplant lot-sizing and hub location problems. The PHL problem is to determine the location of production facilities, lot-sizing, inventory, hub location, and the distribution of multiple commodities from plants to customers, with an objective to minimize the total production, setup, inventory, hub operating, and transportation costs. The PHL problem applies to manufacturing companies that either have built a hub-and-spoke distribution network or are accessible to such a network through collaborations with other third-party logistics companies. Because the PHL problem is [Formula: see text]-hard, we propose an exact method that integrates dynamic programming and Benders decomposition (DPBD) for solving the problem. The DPBD method is enhanced by exploring several problem properties, such as a multicut reformulation, the generation of Pareto-optimal cuts, a two-stage hub elimination and restoration procedure, and the inclusion of a novel heuristic procedure. We compare the PHL model with several related models theoretically and computationally with newly created benchmark instances. The computational results indicate that the proposed model can reduce the total costs and facilitate better network designs and system decisions, highlighting the value of an integrated approach. We also provide managerial insights into the benefits of integration and show the efficiency of the DPBD method through an extensive number of computational tests. History: Accepted by Andrea Lodi, Area Editor for Design and Analysis of Algorithms—Discrete. Supplemental Material: The online supplement is available at https://doi.org/10.1287/ijoc.2023.0339 .
Read full abstract