Redesigning multi-echelon integrated distribution networks using the Lagrangian relaxation heuristics

With constant changes in customer and supplier bases in the midst of rapid globalisation and rising e-commerce activities, supply chain professionals need to reassess the existing distribution network more frequently than ever before. Such reassessment may involve the complex array of the phase-out, relocation, consolidation, centralisation, and risk pooling of warehousing facilities across the entire supply chain. Since the supply chain includes an intertwined network of suppliers, customers and third-party intermediaries, the problem of redesigning the distribution network (DNRP) poses many challenges. To solve such a challenging problem, we develop and propose a nonlinear optimisation model. This model takes into account the multi-echelon links and dynamic interdependence among supply chain partners. The proposed model was structured within the dynamic knapsack framework where a known quantity of resources is available and demands for those resources arrive randomly over time. In this framework, we develop a new heuristic algorithm based on the Lagrangian relaxation, sub-gradient, and branch and bound methods for solving the DNRP. The practicality and computational efficiency of the proposed model and algorithm were verified through a series of model experiments under what-if problem scenarios. [Received: 9 May 2019; Accepted: 8 May 2020]