Expert Elicitation and Data Noise Learning for Material Flow Analysis using Bayesian Inference

TL;DR

This paper introduces Bayesian methods with expert elicitation to better quantify and update uncertainties in material flow analysis, demonstrated through a case study on U.S. steel flows.

Contribution

It develops an expert elicitation procedure for MFA priors and a method to learn data noise simultaneously, enhancing uncertainty quantification in MFA.

Findings

Reduced parametric uncertainty with data integration

Modest reduction in data noise uncertainty

Multi-year data improves inference robustness

Abstract

Bayesian inference allows the transparent communication of uncertainty in material flow analyses (MFAs), and a systematic update of uncertainty as new data become available. However, the method is undermined by the difficultly of defining proper priors for the MFA parameters and quantifying the noise in the collected data. We start to address these issues by first deriving and implementing an expert elicitation procedure suitable for generating MFA parameter priors. Second, we propose to learn the data noise concurrent with the parametric uncertainty. These methods are demonstrated using a case study on the 2012 U.S. steel flow. Eight experts are interviewed to elicit distributions on steel flow uncertainty from raw materials to intermediate goods. The experts' distributions are combined and weighted according to the expertise demonstrated in response to seeding questions. These aggregated distributions form our model parameters' prior. A sensible, weakly-informative prior is also adopted for learning the data noise. Bayesian inference is then performed to update the parametric and data noise uncertainty given MFA data collected from the United States Geological Survey (USGS) and the World Steel Association (WSA). The results show a reduction in MFA parametric uncertainty when incorporating the collected data. Only a modest reduction in data noise uncertainty was observed; however, greater reductions were achieved when using data from multiple years in the inference. These methods generate transparent MFA and data noise uncertainties learned from data rather than pre-assumed data noise levels, providing a more robust basis for decision-making that affects the system.