28.12.2012 Views

LCA Food 2012 in Saint Malo, France! - Manifestations et colloques ...

LCA Food 2012 in Saint Malo, France! - Manifestations et colloques ...

LCA Food 2012 in Saint Malo, France! - Manifestations et colloques ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

GROUP 6, SESSION B: METHODS, TOOLS, DATABASES 8 th Int. Conference on <strong>LCA</strong> <strong>in</strong> the<br />

Agri-<strong>Food</strong> Sector, 1-4 Oct <strong>2012</strong><br />

182. Analysis and propagation of uncerta<strong>in</strong>ty <strong>in</strong> agricultural <strong>LCA</strong><br />

Xiaobo Chen 1,2,* , Michael Corson 1,2<br />

1 INRA, UMR1069 Sol Agro <strong>et</strong> hydrosystème Spatialisation, F-35000 Rennes, <strong>France</strong>, 2 Agrocampus Ouest,<br />

F-35000 Rennes, <strong>France</strong>, Correspond<strong>in</strong>g author. E-mail: xiaobo.chen@rennes.<strong>in</strong>ra.fr<br />

The confidence <strong>in</strong> <strong>LCA</strong> results depends primarily on the quality of source data and their pert<strong>in</strong>ence for the<br />

system studied. However, the need for large amounts of data leads to much uncerta<strong>in</strong>ty <strong>in</strong> impact estimates<br />

due to the data themselves: measurement, use <strong>in</strong> calculations, and f<strong>in</strong>al transformation <strong>in</strong>to impact estimates.<br />

The ma<strong>in</strong> sources of uncerta<strong>in</strong>ty <strong>in</strong> the data cha<strong>in</strong> <strong>in</strong>clude not only statistical uncerta<strong>in</strong>ty (mean and standard<br />

deviation) of the data, but also m<strong>et</strong>hodological choices <strong>in</strong> <strong>LCA</strong>, such as hypotheses made to represent the<br />

system of <strong>in</strong>terest, data representativeness, impact assessment m<strong>et</strong>hods, and allocation of impacts b<strong>et</strong>ween<br />

co-products. Therefore, consideration of uncerta<strong>in</strong>ty <strong>in</strong> <strong>LCA</strong> would provide more scientific <strong>in</strong>formation for<br />

decision mak<strong>in</strong>g. This topic is the focus of doctoral research recently begun at INRA that aims to (1) identify<br />

sources of uncerta<strong>in</strong>ty <strong>in</strong> agricultural-production systems, (2) analyse their propagation, and (3) estimate<br />

their relative contributions to the overall uncerta<strong>in</strong>ty <strong>in</strong> calculated impacts.<br />

Although some texts describe uncerta<strong>in</strong>ty generally as a lack of knowledge, its def<strong>in</strong>ition may change depend<strong>in</strong>g<br />

on the <strong>LCA</strong> steps <strong>in</strong> which it occurs (Huijbregts, 1998). Therefore, uncerta<strong>in</strong>ty is often classified<br />

accord<strong>in</strong>g to its nature and source (e.g., “natural” (i.e., variability) vs. “epistemic”; Van Asselt and Rotmans,<br />

2002). Epistemic uncerta<strong>in</strong>ty has been subdivided <strong>in</strong>to three sources: scenario, model and param<strong>et</strong>er (Fig. 1).<br />

This first step allows uncerta<strong>in</strong>ties from each source to be evaluated by correspond<strong>in</strong>g approaches.<br />

Monte-Carlo analysis is used most frequently to evaluate uncerta<strong>in</strong>ty <strong>in</strong> <strong>LCA</strong> (Bass<strong>et</strong>-Mens, 2009). With it<br />

one can estimate the <strong>in</strong>fluence of uncerta<strong>in</strong>ties <strong>in</strong> <strong>in</strong>put variables (us<strong>in</strong>g their probability distributions) on<br />

predicted potential impacts. However, this approach suffers some m<strong>et</strong>hodological bias due to correlations<br />

b<strong>et</strong>ween variables and poorly-known response rules. For example, the selection of appropriate distributions<br />

is usually based on literature, expert judgment, or empirical studies <strong>in</strong> other systems, which may <strong>in</strong>crease the<br />

complexity of model and param<strong>et</strong>er uncerta<strong>in</strong>ty. Moreover, Monte-Carlo simulation is commonly used for<br />

assess<strong>in</strong>g the <strong>in</strong>fluence of uncerta<strong>in</strong>ty <strong>in</strong> emission factors <strong>in</strong> LCIA, but it may not be appropriate for uncerta<strong>in</strong>ty<br />

<strong>in</strong> other steps, such as the def<strong>in</strong>ition of scope or functional unit or the <strong>in</strong>terpr<strong>et</strong>ation of results that consist<br />

of both subjective and objective uncerta<strong>in</strong>ty (Fig. 2). Although it is not necessary or possible to consider<br />

all uncerta<strong>in</strong>ties <strong>in</strong> <strong>LCA</strong>, subjective uncerta<strong>in</strong>ty should not be overlooked. More complex approaches (e.g.,<br />

fuzzy logic) exist, but their use rema<strong>in</strong>s marg<strong>in</strong>al. Currently, the m<strong>et</strong>hods used to describe uncerta<strong>in</strong>ty propagation<br />

<strong>in</strong> <strong>LCA</strong> have not tried to differentiate the various types of uncerta<strong>in</strong>ty but rather to aggregate them.<br />

Thus, more research is necessary to overcome barriers to analys<strong>in</strong>g uncerta<strong>in</strong>ty <strong>in</strong> <strong>LCA</strong>.<br />

Thesis work will beg<strong>in</strong> by identify<strong>in</strong>g and classify<strong>in</strong>g uncerta<strong>in</strong>ty <strong>in</strong> each <strong>LCA</strong> step, especially uncerta<strong>in</strong>ties<br />

frequently encountered when assess<strong>in</strong>g agricultural systems. With case studies, the research will identify the<br />

most important uncerta<strong>in</strong>ties, develop m<strong>et</strong>hods to categorise them, and work to estimate the contribution of<br />

each source of uncerta<strong>in</strong>ty to the overall uncerta<strong>in</strong>ty <strong>in</strong> output. By consider<strong>in</strong>g uncerta<strong>in</strong>ty <strong>in</strong> agricultural<br />

<strong>LCA</strong>, more compl<strong>et</strong>e <strong>in</strong>formation about environmental impacts can be given to decision-makers, who should<br />

consider uncerta<strong>in</strong>ty as an important part of decision analysis.<br />

References<br />

Bass<strong>et</strong>-Mens, C., Kelliher, F.M., Ledgard, S., Cox, N., 2009. Uncerta<strong>in</strong>ty of global warm<strong>in</strong>g potential for<br />

milk production on a New Zealand farm and implications for decision mak<strong>in</strong>g. Int. J.<strong>LCA</strong> 14, 630-638.<br />

IPCS, 2008. Uncerta<strong>in</strong>ty and Data Quality <strong>in</strong> Exposure Assessment. Geneva, World Health Organization,<br />

International Programme on Chemical Saf<strong>et</strong>y (Harmonization Project Document No. 6).<br />

Huijbregts, M.A.J., 1998. Application of uncerta<strong>in</strong>ty and variability <strong>in</strong> <strong>LCA</strong>. Part I: a general framework for<br />

the analysis of uncerta<strong>in</strong>ty and variability <strong>in</strong> life cycle assessment. Int. J. <strong>LCA</strong> 3, 273-280.<br />

Van Asselt, M.B.A., Rotmans, J., 2002. Uncerta<strong>in</strong>ty <strong>in</strong> <strong>in</strong>tegrated assessment modell<strong>in</strong>g. Int. J. <strong>LCA</strong> 54, 75-<br />

105.<br />

933

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!