A bi-level stochastic optimization model for reliable supply chain in competitive environments: Hybridizing exact method and genetic algorithm

Abstract

• Proposing two MINLP models for competitive supply chain using two-stage stochastic programming. • Utilizing a bi-level model to formulate the Stackelberg competition, where DCs are leaders and retailers are followers. • Discussing the effect of disruption risk on the performance of DCs. • Developing a hybrid metaheuristic method based on GA and exact solver to solve the problem. • Studying a real-life case from Alborz pharmaceutical company to show the potential applicability of the model. Managing supply chain operations in a reliable manner is a significant concern for decision-makers in competitive industries. In this article, two mathematical models considering competition and integrity in a three-echelon supply chain under uncertainty are proposed. The competition is formulated as a Stackelberg game such that the distribution centers have more power than the retailers. In the first model, decisions are made about the location and number of distribution centers (DCs), allocation of retailers, and the selling price of products. In the second model, based on the real world, the probability of risk and failure for the distribution centers are considered. Backup facilities should be established for unreliable facilities to meet the demands of retailers during disruption. To capture uncertainty, a two-stage stochastic approach is applied to model the problems. The first stage of the model belongs to the strategic planning and is not affected by randomness, while the second stage deals with tactical decisions depending on the realization of the first stage's random vector. In order to solve the problem, a hybrid genetic algorithm has been applied to large-scale problems. Numerical experiments have been conducted to assess the effectiveness of the proposed algorithm. Next, a sensitivity analysis is performed to recognize the most important parameters and evaluate the accuracy of our approach. Finally, to demonstrate the applicability of the model, the proposed model was implemented on the data of Alborz Pharmaceutical Company.