To meet the Paris Agreement’s goal of limiting global warming to below 2°C while achieving net-negative carbon emissions [1], this study proposes an optimization framework for the strategic planning and siting of biorefineries across the United States. Biorefineries offer a promising pathway by integrating carbon capture and storage (CCS) with biomass-to-biofuel conversion [2].
Strategic decision-making is challenging due to the geographical dispersion of biomass resources, the technological variety of BECCS pathways, and climate uncertainty. To address these challenges, we developed a Mixed-Integer Nonlinear Programming (MINLP) model that optimizes biorefinery site selection to minimize net carbon emissions and total lifecycle costs. The model integrates datasets including county-level biomass availability from the U.S. Billion Ton Report, performance and emissions profiles of BECCS technologies (gasification, fermentation, pyrolysis), and cost components such as transportation, capital, and operations.
To reduce computational complexity in large-scale bioenergy infrastructure planning, we applied K-means clustering to group U.S. counties based on total biomass availability and geographic proximity. A total of 19 model scenarios were executed, comparing clustering configurations. The optimization model, formulated in Julia using JuMP, was solved using high-performance computing (HPC) resources.
Preliminary results from the 15-cluster configuration indicate that over 2.3×1012 kg of CO₂ can be captured, with a total feedstock cost of approximately $180 billion, transportation cost of $26.88 billion, and revenue of $645.21 billion. These outcomes are based on nine different biorefinery types selected by the model to be built across the 15 cluster locations. Among these, gasification with hydrogen production yielded the highest CO₂ capture at over 3.9×1011 kg, while polyethylene production generated the highest revenue at $140.53 billion.
The 100-cluster scenario was evaluated for a single scenario focusing only on gasification with hydrogen production. Results show a CO₂ capture of 9.78×1011 kg, feedstock cost of $48.36 billion, transportation cost of $13.26 billion, and revenue of $105.19 billion.
The model will be extended to run all 19 scenarios under the 100-cluster configuration to enable a comprehensive comparison with the 15-cluster. These findings demonstrate how predictive analytics and optimization can support resilient, low-carbon bioenergy infrastructure design by balancing spatial resolution with computational efficiency.
Keywords: Biorefinery Deployment, Net-Negative Emissions, Mixed-Integer Non-Linear Programming,
References
[1] Amahnui, G. A., Vanegas, M., Verchot, L., & Castro-Nunez, A. (2025). Achieving the paris agreement goals by transitioning to low-emissions food systems: A comprehensive review of countries’ actions. Environmental Science & Policy, 163, 103968.
[2] O’Neill, E. G., Geissler, C. H., & Maravelias, C. T. (2024). Large-scale spatially explicit analysis of carbon capture at cellulosic biorefineries. Nature Energy, 9(7), 828-838.