Bayesian material flow analysis for systems with multiple levels of disaggregation and high dimensional data

• DOI: 10.1111/jiec.13550
• Publication Date: 2024-09-30T14:09:50Z
• Authors: Junyang Wang; Kolyan Ray; Pablo Brito‐Parada; Yves Plancherel; Tom Bide; Joseph Mankelow; John Morley; Julia A. Stegemann; Rupert Myers
• Keywords: Hyperparameter; Uncertainty Quantification

Abstract Material flow analysis (MFA) is used to quantify and understand the life cycles of materials from production end use, which enables environmental, social, economic impacts interventions. MFA challenging as available data are often limited uncertain, leading an under‐determined system with infinite number possible stocks flows values. Bayesian statistics effective way address these challenges by principally incorporating domain knowledge, quantifying uncertainty in data, providing probabilities associated model solutions. This paper presents a novel methodology under framework. By relaxing mass balance constraints, we improve computational scalability reliability posterior samples compared existing methods. We propose mass‐based, child parent process framework systems disaggregated processes flows. show predictive checks can be identify inconsistencies aid noise hyperparameter selection. The proposed approach demonstrated case studies, including global aluminum cycle significant disaggregation, weakly informative priors gaps investigate feasibility MFA. illustrate that just prior greatly performance methods, for both estimation accuracy quantification.