Designing a resilient and sustainable multi-feedstock bioethanol supply chain: Integration of mathematical modeling and machine learning

The escalating global demand for sustainable energy has intensified the focus on bioethanol as a renewable resource. This paper presents a novel optimization approach for designing and planning a resilient and sustainable supply chain for bioethanol production, utilizing second and third-generation biomass. The proposed approach adeptly determines crucial strategic and tactical decisions for the supply chain while effectively addressing epistemic uncertainties and disruption risks associated with the production process. These decisions, encompassing optimal locations, capacities, technologies, production, distribution, and transportation choices, are strategically navigated in three distinct phases. In the first phase, a combined data envelopment analysis and artificial neural network approach selects optimal sites for establishing microalgae production facilities. The second phase presents a mixed-integer linear programming model to optimize the bioethanol supply chain design by minimizing total costs and meeting sustainability constraints on greenhouse gas emissions and employment. The model seamlessly integrates multiple biomass feedstocks, incorporating wheat straw, corn stover, and microalgae. Addressing disruption risks, the third phase employs a novel robust stochastic-possibilistic programming to effectively tackle uncertainties in costs, prices, and yields. The implementation of the proposed optimization approach using real data from a case study demonstrates its utility as a decision support tool for resilient and sustainable bioethanol supply chain design. Computational results notably highlight the superiority of the proposed robust approach, showcasing over 11 % cost savings compared to its deterministic counterpart.