Premium
Influence of Inventory Data Sets on Life‐Cycle Assessment Results: A Case Study on PVC
Author(s) -
Peereboom Eric Copius,
Kleijn René,
Lemkowitz Saul,
Lundie Sven
Publication year - 1998
Publication title -
journal of industrial ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.377
H-Index - 102
eISSN - 1530-9290
pISSN - 1088-1980
DOI - 10.1162/jiec.1998.2.3.109
Subject(s) - representativeness heuristic , life cycle assessment , life cycle inventory , environmental science , categorization , reliability (semiconductor) , computer science , database , statistics , production (economics) , mathematics , artificial intelligence , economics , macroeconomics , power (physics) , physics , quantum mechanics
Summary This study compared six widely used European life‐cycle assessment (LCA) inventory data sets, identified those, data elements that introduce major differences, and quantitatively determined the influence of these data elements for a cradle‐to‐gate LCA o f polyvinyl chloride (PVC).Large differences in data (10‐ I 100%) were found. Data on substances with recognized high environmental impact and easily determined emissions and environmental impacts, like those related to energy, show the least differences. Process‐specific emissions show larger differences. Substantially more substances emitted t o air than t o water or soil are reported, and differences between the values are less. Furthermore, various inventory data sets donot always cover the same substances. Often, individual substances, such as specific (chlorinated) hydrocarbons and metals, are collectively categorized rather than individually reported. Specific data elements o f the inventory causing many differences were geographical, temporal, and technological representativeness; categorization o f substances; naming of substance categories; use of different category definitions: system boundaries; and allocation method. The influence of these differences on LCA results, determined through sensitivity analysis, was significant, typically 10‐ 100%. Results emphasize the importance of appropriate and explicitly described data sets and the necessity o f sensitivity analyses. Results also show the need for a regularly updated and openly available database with high quality data. The availability of such a database would improve the reliability of LCA and thereby stimulate its application.