Mettre en contexte les resultats d'une Analyse de Cycle de Vie: Developpement de facteurs de normalisation canadiens et problematique de la definition des frontieres.

摘要

In Life Cycle Assessment (LCA), the normalization step follows the impact assessment step and can be used as a support for interpretation. It calculates the magnitude of a potential impact in respect of the total of a given reference. Even though normalization is optional in LCA according to ISO standards (International Organization for Standardization (ISO), 2000a), it has the advantage of expressing LCA results on a common basis that allows the determination of the relative importance of the different effects to the selected reference. These results are expressed in a common unit, and thus are in a suitable form for the (possible) following step: weighting. Among the two existing approaches, namely internal and external normalization, external normalization has the advantage to place the LCA results in a broader context, for example a geographical context (country, continent or world) (Norris, 2001). The use of external values of reference allows a consistency check of the Life Cycle Inventory (LCI) of the studied products. The values of reference used in external normalization correspond to the total impacts of the selected geographical area and are called normalization factors (NFs).;When the reference corresponds to a global scale, there is no difference between production and consumption. At a continental or national scale, the difference increases as imports and exports gain importance (Wilting et Ros, 2009). Traditionally, the NFs are calculated by considering the production activities. For comparative reasons, and in order to be able to validate the previous hypothesis, calculation of the Canadian NFs was carried out by using the same approach. On the other hand, it is recognized, in the literature, that a consumption-based approach is more suitable to assess the responsibility of countries relatively to their environmental impacts (Peters et Hertwich, 2008; Wilting et Ros, 2009). Therefore, the second hypothesis was: the difference between production and consumption activities of a country can be highlighted by a comparison of the NFs calculated according to these two points of view.;The main objective of this project is to develop reference values adapted to the Canadian context. It can be divided into two distinct sub-objectives, corresponding to the two hypotheses that were previously defined, (1) calculate Canadian NFs considering production activities; and (2) calculate Canadian NFs considering consumption activities and compare the resulting NFs from the two approaches.;Parameter uncertainty evaluated by the semi-quantitative approach based on the Pedigree matrix (Weidema et Wesnaes, 1996) showed that uncertainty of the NFs depends on impact categories. Temporal variability represents the error made by choosing the year 2005 as reference. Temporal variability does not contribute significantly to the total uncertainty but can be an issue if the emissions follow a tendency2 in time or change abruptly due to legislation modifications. Uncertainty due to choices was evaluated with a scenario analysis. Model uncertainty was highlighted by reiterating the calculation of the NFs for three different impact assessment methodologies: IMPACT 2002+ (Jolliet et al., 2003), TRACI (Bare, 2002) and LUCAS (Toffoletto et al., 2007). The observed differences suggest that it is justified to use a geographically adapted methodology.;It is generally recommended to use NFs adapted to the geographical context in which the LCA study is carried out (Udo de Haes et al., 2002). NFs were developed for various countries and continents (e.g. Netherlands (Breedveld et al., 1999; Huijbregts et al., 2003a; Sleeswijk et al., 2008), Denmark (Stranddorf et al., 2005), Europe (Jolliet et al., 2003) or the United States (Bare et al., 2006)) but up to now, never in a Canadian context. This leads to the first hypothesis, i.e. that it is necessary to calculate NFs adapted to the Canadian context because economical and industrial activities differ from one country (or continent) to another, which leads to significant differences in terms of total impacts. It is also important to calculate uncertainty related to the NFs has it allows LCA practitioners to refine and/or moderate their interpretation of the normalized results. Although often discussed in a qualitative way, the NFs' uncertainty was never quantified.;The NFs based on Canadian consumption were calculated from the individual consumption relative to the five following fields: food, transport, housing, consumption goods and public services. The use of the augmented process-based hybrid methodology allowed the use of both physical and monetary data in the realization of the assessment. For 3 damage categories, out of 4, the consumption-based NFs are approximately twice as high as the consumption-based NFs. This highlights the difference between production and consumption activities in Canada. (Abstract shortened by UMI.);2Increasing or decreasing tendency