Fusion of conflicting information for improving representativeness of data used in LCAs

A well-established framework for information fusion to the field of LCA where disparate data may need to be fused to perform the assessment under certain conditions can be introduced and will enhance data quality and LCA credibility.

Abstract

Employing representative data is necessary for producing a credible LCA informing decision making process. When the data is available from multiple sources, and in incompatible formats such as point estimates, intervals, approximations, and may even be conflicting in nature, it is important to synthesize it with minimal loss of information to enhance the credibility of LCA. This article introduces a framework for information fusion that can serve this purpose within the current operational procedure of LCA. The character of information gathered from multiple sources is inherently different than that exhibited by the information generated by a single random source. The framework of possibility theory can be used to merge such heterogeneous information as demonstrated by its application in the diverse fields such as engineering, finance, and social sciences. This article introduces this methodology for LCAs by first introducing the theory behind data modeling and data fusion with possibility theory. Then, this framework is applied to the disparate data from literature on the manufacturing energy requirements for semiconductor device fabrication, and also to a hypothetical example of linguistic inputs from experts in order to demonstrate the operationalization of the theory. A flowchart is provided to recap the framework and for easy navigation through the steps of merging procedure. The framework for fusion of information applied the numerical and linguistic heterogeneous data in the LCA context illustrates that this methodology can be implemented relatively easily to increase the data quality and credibility of LCA. This can be done without making any changes in the usual preferred way of conducting an LCA. Information fusion may be performed either after the sensitivity analysis identifies the most impactful categories that need further investigation, or it can be performed upfront to the select input categories of interest. The article introduces a well-established framework of information fusion to the field of LCA where disparate data may need to be fused to perform the assessment under certain conditions. This framework can be easily implemented, and will enhance data quality and LCA credibility. We also hope that data entry software such as ecoEditor make provision for the data entry mechanism necessary to enter fused data.