A two-stage approach to enhancing biofuel supply chains through predictive and optimization analytics

The escalating pressures of population growth, surging global energy needs, water shortages, reliance on fossil fuels, and urban air pollution underscore the critical demand for sustainable energy alternatives. Biofuels present a viable solution, yet their successful adoption hinges on an efficient supply chain. This study introduces a comprehensive two-stage optimization framework to advance the design and operation of biofuel supply chains. In the initial stage, a novel hybrid methodology integrates data envelopment analysis with artificial neural networks to identify optimal sites for agricultural waste collection facilities. This approach combines the performance assessment strengths of data envelopment analysis with the predictive capabilities of neural networks, enabling a data-informed site selection process. The second stage employs a mixed-integer linear programming model to optimize a closed-loop biofuel supply chain under uncertain conditions, targeting both cost reduction and minimized carbon emissions. A probabilistic scenario-based approach is utilized to address uncertainties, enhancing the model's real-world applicability. Additionally, the Lagrangian relaxation technique is implemented to achieve precise solutions while preserving computational efficiency. For large-scale scenarios, the study leverages the non-dominated sorting genetic algorithm and multi-objective simulated annealing to generate near-optimal solutions. A practical case study validates the proposed framework and provides decision-makers with clear and actionable strategies to optimize site planning, reduce operational costs, and enhance environmental sustainability in biofuel supply chain management.