Data‐Driven Endogenous Capacity Optimization for Specialized Warehouse Location in Industrial Distribution Networks

This paper develops a mixed‐integer linear programming (MILP) model for joint warehouse location and endogenous capacity planning in two‐tier industrial distribution networks. Unlike traditional capacitated facility location formulations that select capacity from a finite set of predefined levels, the proposed model determines each warehouse's size as a continuous decision variable, eliminating the capacity mismatch that commonly arises in discrete‐sizing approaches. The formulation integrates facility construction costs, land acquisition costs, segmented operational cost tiers capturing economies of scale, and product‐handling restrictions associated with specialized warehouse operations. Binary product terms in the objective are fully linearized via McCormick auxiliary variables, yielding a pure MILP solved efficiently by CPLEX branch‐and‐bound. Computational experiments on a 37‐city European logistics network validate the model under two configurations. Unlike prior single‐instance evaluations, the framework is tested through 50‐instance Monte Carlo experiments with demands perturbed by up to and costs by up to , achieving a mean cost reduction of over the fixed‐capacity benchmark (95% CI: ; paired ‐test ). © 2026 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.