A novel clustering-based algorithm for distribution network design in pharmaceutical retail logistics

Purpose To establish an efficient distribution network, pharmaceutical retail companies have to address two major challenges. First, warehouse location decisions need to be integrated with daily transportation routes. Second, the multi-echelon structure of the pharmaceutical network and the large number of retail pharmacies further increase the difficulty of network design. This paper aims to address a two-echelon distribution location problem (2E-DLP) and provide pharmaceutical operations managers with an efficient algorithm for distribution network planning. Design/methodology/approach A two-stage clustering-based algorithm (TSCBA) is proposed to solve the 2E-DLP. It first uses a weighted K-means clustering algorithm to generate potential locations for central procurement warehouses and regional distribution centers. Then, a greedy algorithm is applied to estimate the transportation cost between each regional distribution center and its retail pharmacies, which is used to compute the total cost of different location results. The TSCBA is highly efficient in evaluating location results, especially when handling numerous retail pharmacies. Findings The performance of the proposed TSCBA is evaluated by different scale 2E-DLPs, which are generated based on the operational data from a leading pharmaceutical retail company in China. The computation results indicate that TSCBA is capable of striking a good balance between solution quality and computation cost. Originality/value This paper proposes a novel TSCBA for solving the 2E-DLP in pharmaceutical retail logistics. The strong performance of TSCBA can be attributed to the effective hybridization of two fast algorithms for the warehouse location problem and the vehicle routing problem. The TSCBA has significant potential to optimize two-echelon and large-scale distribution networks for pharmaceutical retail companies.