Multilevel facility location optimisation: a novel integer programming formulation and approaches to heuristic solutions
研究了多级设施选址问题，提出一种新的整数规划模型，并设计了多起点贪婪和禁忌搜索两种启发式算法，通过大规模随机算例验证了有效性。
This paper focuses on Multi-level Facility Location (MFL) problems, a critical component in supply chain design. Foundational tasks involve selecting plants, warehouses, distribution centres, and retail stores (markets) to maximise profits while considering related constraints. The problem has a single-assignment property. In that, a retail store is served via a single bundle of products transferred from a plant to a warehouse, then to a distribution centre, and finally to the store. Furthermore, each retail store has preferences for products from specific plants. Each selected facility incurs a one-time fixed cost, and there are upper bounds on the number of facilities of each type that can be selected. Transporting a single bundle of products from one upper-level facility to a lower-level facility incurs cost. Adapted from hub-location literature, the problem is formulated as a variation of the quadratic assignment problem. We propose two heuristics as solution approaches: (1) a multi-start greedy heuristic, (2) a multi-start tabu search. Extensive computational experiments with the heuristics are provided for randomly generated large-scale problems, and sensitivity analyses, supported by appropriate statistical methods, are used to validate the effectiveness of the heuristics’ results.