Mapping Aluminum Flows in the United States Using Advanced Bayesian Material Flow Analysis

This paper presents a Bayesian material flow analysis (MFA) of aluminum in the U.S. economy, reconciling a 2017 baseline and updating the system through 2022. We introduce ARCADE (Abiotic Resource Cycles, Allocations, and Dynamics in the Economy), a unified Bayesian MFA framework integrating system definition, data pedigree, and probabilistic reconciliation under physical constraints. ARCADE combines government statistics, trade records, and industry reports into a physically consistent network and reports uncertainty-aware metrics. Results show continued U.S. reliance on imports of unwrought and finished aluminum products. The posterior median end-of-life (EOL) collection rate was 66.8% [57.2%, 79.9%] in 2017 and 67.6% [62.3%, 73.9%] in 2022. Median domestic retention of collected EOL scrap was lower in 2022 (52.7% [49.6%, 55.6%]) than in 2017 (56.4% [51.7%, 60.5%]), although the credible interval for the change spanned zero. Across robustness cases varying source weighting, pedigree-to-uncertainty mappings, and prior strength, domestic retention remained below 60%. Spatial analysis identified mismatches between EOL scrap generation and domestic remelting capacity. These findings show that improved collection alone may not increase domestic recycling when processing constraints persist. We observe a continued import dependence, efficient circulation of prompt fabrication scrap, and limited domestic retention of collected EOL scrap.