Life cycle assessment. Impact assessment uncertainties

Over the last decade, the development of impact assessment methods for the LCA (Life Cycle Assessment) tool has provided practitioners with increasingly exhaustive impact results that are consistent with the objectives of studies [G 5 605][G 5 615] . These methods have also introduced ever greater sophistication and uncertainty. Assessing the uncertainties of impact results has thus become a key issue for the credibility of the tool and for supporting decision-making processes based on its results.

In LCA, uncertainty is generally defined and classified according to its sources, i.e. models, data and the practitioner's choices . Various methods of analysis, all based on probability and interval theories, have been proposed. Yet the word "uncertainty" refers not only to the content of the information (i.e. its value) that can be correctly analyzed by these theories, but also, at the same time, to its truth, understood in the sense of its conformity with reality and/or the objectives of the study (what we call confidence). Uncertainty about an impact assessment model (e.g. for the greenhouse effect) concerns the value of its variables (what are the errors and variability affecting these values?), but also their nature (are these variables the most relevant to consider in the model, given its objectives and the phenomena to be modelled? Are they correctly considered?) Taking these two elements (value and confidence) into account requires more detailed classification and characterization.

In this article we will illustrate a practical and comprehensive methodological framework for defining, classifying and evaluating uncertainties in impact assessments. The principles and operational procedures will be presented with the help of numerous application examples.