Data-driven robust optimization for a sustainable steel supply chain network design: Toward the circular economy

Abstract

Steel is the indispensable cornerstone of modern life and its remarkable strength and adaptability render it a fundamental pillar of worldwide advancement and economic growth. However, the steel sector faces significant challenges, including the need to incorporate sustainable practices and circular economy principles, especially enhancing occupational safety, optimizing water and energy resource utilization, and effectively utilizing by-products within the steel supply chains. This paper addresses these gaps by presenting a bi-objective data-driven optimization model formulated to design a sustainable forward-reverse steel supply chain network. Sustainable development is studied by examining total cost, occupational safety, and environmental impacts of water and energy consumption and transportation system, concurrently. Incorporating circular economy principles, the collection of steel scrap under take-back policies and utilizing steel slag, a by-product of steel production, is considered. To address uncertainties, a data-driven robust optimization method is employed which harnesses historical data to construct a dynamic uncertainty set using support vector clustering. The model is applied to a real case study in the Iranian steel sector. Results demonstrate the effectiveness and robustness of the proposed method, showing that, on average, a 5 % increase in costs correlates with a 7 % improvement in environmental impacts.