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

Area of Science:

• Environmental Science
• Systems Analysis
• Statistical Modeling

Background:

• Material Flow Analysis (MFA) quantifies material life cycles but faces challenges with limited and uncertain data.
• Existing MFA methods struggle with under-determined systems and infinite possible solutions.
• Bayesian statistics offers a framework to incorporate prior knowledge and quantify uncertainty in data.

Purpose of the Study:

• To develop a novel Bayesian methodology for Material Flow Analysis (MFA).
• To enhance computational scalability and reliability of Bayesian MFA by relaxing mass balance constraints.
• To demonstrate the effectiveness of the proposed approach in handling data gaps and disaggregated systems.

Main Methods:

• Developed a novel Bayesian MFA methodology relaxing mass balance constraints.
• Proposed a mass-based, child and parent process framework for disaggregated systems.
• Utilized posterior predictive checks for data inconsistency identification and parameter selection.

Main Results:

• The novel Bayesian MFA approach improves computational scalability and reliability of posterior samples.
• Relaxing mass balance constraints enhances performance compared to existing Bayesian MFA methods.
• Weakly informative priors significantly improve estimation accuracy and uncertainty quantification, even with data gaps.

Conclusions:

• The proposed Bayesian MFA framework is feasible and effective, even with significant data gaps and disaggregation.
• Bayesian methods, particularly with weakly informative priors, offer a robust solution for complex MFA.
• The methodology aids in identifying data inconsistencies and improving model reliability for environmental assessments.