Page 352 - Materials Chemistry, Second Edition

336 5 Life Cycle Interpretation, Reporting and Critical Review

• Comparative analysis
• Discernibility analysis.
Contribution analysis generally serves the quantitative determination of parts to
the total result. For LCAs this may imply determining the contribution of one or
some life cycle sections to an impact category (indicator result). This approach is
often called the sectoral analysis and can be adequately plotted as a coloured bar
diagram (see Figure 4.4). Dominance analysis can be regarded as specific form of
contribution analysis in such that it determines the dominant contributions to the
result and therefore is suited for the identification of significant parameters. For
sectoral analysis, however, it has to be considered that, in addition to the selected
sectors, upstream processes (partial inventories related to raw material production)
are often determined as well. A dominance of the ‘production’ sector must therefore
by no means indicate that most emissions occur at one single production site!
Heijungs and Kleijn (2001, loc. cit.) point out that contribution analysis is suitable
for a variety of objectives, both on the level of inventory and of impact assessment
(with or without normalisation also followed by weighting, if required). It is possible
to determine the impact of emissions or of individual unit processes, and so on,
on intermediate or final results. Questions of utmost importance can be clarified
during this process, for example, where improvements of product systems can be
accomplished with highest efficiency. These objectives are particularly important for
non-comparative LCAs that predominantly serve product optimisation. Graphical
representations should be handled with prudence: pie charts, especially, are not
suitable for negative entries resulting from credit entries; for bar diagrams, the
latter can be represented by a zero baseline.
Perturbation analysis is related to sensitivity analysis, but mathematically more
exactly defined. The impact of marginal 17) (very small) alterations of LCA input
parameters is quantitatively examined (marginal analysis). It can thus be deter-
mined on how sensitively the calculated results react to uncertainties of input
parameters. Hence it can be deduced which input parameters require great pre-
cision and which are suitable for estimated values, because their influences on
the final result are small. The advantages of perturbation analysis are that existing
uncertainties need not be known and that further data material is not necessary; it
can be conducted with every LCA. The disadvantages can be the time demand, espe-
cially for complex systems and correspondingly high computation times (Heijungs
and Kleijn, 2001, loc. cit.).
The results of perturbation analysis are extremely valuable for system improve-
ments as those inputs can be identified whose reduction would imply large
reductions of environmental loads.
Uncertainty analysis is a systematic analysis of the propagation of input
uncertainties to output uncertainties, at which preferably Monte Carlo simulations
are applied. 18) If values of input parameters obey a certain probability distribution
(e.g. Gauss) a large number of arbitrarily chosen simulations (hence the name

17) Heijungs (1994).
18) Heijungs and Kleijn (2001), Morgan and Henrion (1990) and Huijbregts (1998).