In this study, we propose a two-echelon multi-objective dual-sale channel supply chain network (DCSCN) model. The goal is to determine (i) the set of installed DCs, (ii) the set of customers the DC should work with, how much inventory each DC should order and (iv) the distribution routes for physical retailers or online e-tailers (all starting and ending at the same DC). Our model overcomes the drawback by simultaneously tackling location and routing decisions. In addition to the typical costs associated with facility location and the inventory-related costs, we explicitly consider the pivotal routing costs between the DCs and their assigned customers. Therefore, a multiple objectives location-routing model involves two conflicting objectives is initially proposed so as to permit a comprehensive trade-off evaluation. To solve this multiple objectives programming problem, this study integrates genetic algorithms, clustering analysis, Non-dominated Sorting Genetic Algorithm II (NSGA-II). NSGA-II searches for the Pareto set. Several experiments are simulated to demonstrate the possibility and efficacy of the proposed approach.
Lecture Notes in Artificial Intelligence 8481, pp.479–489