Distribution network design problem considering warehouse types: framework, model, and algorithm

E-commerce growth drives enterprises to optimise distribution networks, balancing rising logistics costs against increasing customer service demands. We investigate a distribution network design problem involving multiple warehouse types. Unlike traditional models with predetermined locations for central or regional depots, our model treats warehouse types as decision variables, resulting in a more flexible yet complex network structure. We propose a decision support framework outlining key implementation steps, including mechanisms for human-algorithm interaction. Building on this framework, we develop a mathematical model that minimises total logistics costs while maximizing on-time delivery fulfillment rates. Potential extensions for specific application scenarios are discussed. Our solution employs a two-layer heuristic algorithm: the outer layer optimises location decisions while the inner layer constructs network links, accommodating operational-level constraints. The algorithm's extensibility for evolving decision-maker objectives is addressed. Numerical studies demonstrate that for simplified instances, our method performs comparably to commercial solvers. Practical implementation is validated through a case study with a Chinese household appliance manufacturer.