Authors - Sowmini Devi Veeramachaneni Abstract - Modern supply chain systems must balance economic efficiency with environmental sustainability. Traditional optimization approaches, such as linear programming (LP), provide optimal solutions but often struggle with scalability in large-scale networks. This paper proposes a clustering-based framework to reduce the computational complexity of supply chain optimization while preserving solution quality. The method groups suppliers and demand points using feature-aware clustering based on cost and emission profiles, and solves a reduced transportation problem using LP. Experimental results on a real-world dataset demonstrate that the proposed approach achieves near-optimal performance, with less than 7% deviation in profit and less than 2% deviation in emissions, while reducing computation time by nearly an order of magnitude. An ablation study further highlights the trade-off between computational efficiency and solution fidelity controlled by the number of clusters. The proposed framework provides a practical and scalable solution for large-scale, sustainability-aware supply chain optimization.
Open Zoom