Modeling the Multi-objective Supplier Selection and Order Allocation with a Risk-Averse Approach and Considering the Impact of Disruption

Abstract

Managing disruption risks in supply chains, particularly in complex and uncertain environments, is a critical concern in industrial engineering and supply chain management. In this study, a mixed-integer nonlinear programming (MINLP) model based on stochastic programming was developed to simultaneously optimize supplier selection and order allocation in a centralized multi-product supply chain. The proposed model accounts for both local disruption risks (e.g., equipment failures) and regional risks (e.g., natural disasters or regional conflicts) and simulates the behavior of a risk-averse decision-maker using Value-at-Risk (VaR) and Conditional Value-at-Risk (CVaR) criteria. To solve the model, two multi-objective metaheuristic algorithms—Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Multi-Objective Particle Swarm Optimization (MOPSO)—were employed, with their parameters tuned using the Response Surface Methodology (RSM). The performance of the algorithms was evaluated across 30 test problems using common multi-objective assessment metrics, including NPS, MID, DM, Spacing, computational time, and objective function values. Computational results indicated that NSGA-II outperformed MOPSO in most key metrics, particularly in terms of Pareto front convergence and diversity, achieving the highest overall performance score and ranking first among decision-makers. Meanwhile, MOPSO demonstrated relatively better performance only in terms of computational time and achieving acceptable objective function values.