An optimization Approach for Supply Chain Network Design under Uncertainty

Abstract

This paper presents an optimization framework for supply chain network design (SCND) under uncertainty that integrates stochastic, robust, and distributionally robust methods with modern data-driven estimation to improve cost-effectiveness, resilience, and sustainability. The proposed approach models facility location, capacity, inventory and transportation decisions in a two-stage mixed-integer program, accommodates multiple uncertainty representations (scenarios, ambiguity sets, parameter distributions), and applies computational decomposition and sampling-based solution techniques. Numerical experiments with case studies demonstrate that hybrid stochastic–robust and distributionally robust formulations provide superior out-of-sample performance compared to purely deterministic or single-method formulations, particularly under limited data and disruption-prone environments. The findings highlight trade-offs among cost, robustness, and service levels and point to future research directions: integration with real-time data, multi-criteria sustainability objectives, and scalable solution algorithms.