Title: Life cycle impact assessment based on decision analysis
Author: Jyri Seppälä, Finnish Environment Institute
Status: Systems Analysis Laboratory Research Reports A86, June 2003
Keywords: life cycle impact assessment, LCA, decision analysis, valuation, environmental impact assessment, forest industry, metals industry, models, theories.
Life cycle assessment (LCA) is a popular tool for evaluating environmental impacts of products and services. However, the methodological choices and framework to assess environmental impacts in LCA are still under discussion. Despite intensive development worldwide, few attempts have been made hitherto to systematically present the theoretical bases of life cycle impact assessment (LCIA). In this study, the decision analytic foundations for LCIA are illustrated. It is shown that the typical aggregation equation used in LCIA for calculating indicator results can be derived from multiattribute value theory (MAVT) if certain simple assumptions hold. The decision analysis framework presented in this work offers additional values for all the phases of LCIA. A value tree, a tool used for structuring multicriteria decision making problems, can be helpful for selection of impact categories and classification. MAVT can clarify the debate concerning marginal and average approaches in the determination of characterisation factors. On the basis of MAVT, normalisation is needed before weighting. The methods and experiences of preference elicitation derived from the field of decision analysis can be utilised in the determination of subjective characterisation factors and impact category weights. Furthermore, experiences and techniques for the sensitivity analysis of multi-criteria decision models can be utilised in LCIA. In addition, MAVT assists in the calculation of impact category indicator results and total impact indicator results according to the appropriate aggregation equations. The decision analysis framework presented in this work is flexible and suitable for different impact assessment approaches developed in LCIA and it can help the methodological development with which the non-linearity aspects of impact assessment are taken into account. It is shown in the work that site-dependent characterisation methods can easily be fitted into the framework. In a case study of the Finnish forest industry a Finland-specific impact assessment model utilising the results of other tools, such as air quality and transport models and even expert judgements, was developed. In addition, the so-called ratio estimation method for the elicitation of impact category weights was applied and developed so that interval-valued ratio judgements could be used in the uncertainty analysis of the model. In the case of the Finnish metals industry, decision analysis impact assessment was applied to produce a solution in which global, regional and local environmental problems were assessed in the same framework. In both case studies, experts working with the environmental issues valuated impact category weights with the help of decision analysis techniques. In the work it was shown that MAVT provides a foundation for a logical and rational approach to impact assessment in LCA. In the future, there is a need to demonstrate quantitatively the differences between LCIA conducted according to MAVT and according to current practices. Furthermore, there is a need for research to study the strengths and weaknesses of the different decision analysis methods for LCIA purposes.