Page 201 - Materials Chemistry, Second Edition

10 Life Cycle Impact Assessment 187

potential, due to the sheer number of substances (i.e. elementary flows) that may be
assigned to this category and the variation in their environmental persistence and
potential toxicity. It is much more certain to consistently characterise an impact
category to which only a handful of elementary flows are assigned showing impact
potentials that range only three or four orders of magnitude from the least to the most
impacting elementary flow (e.g. eutrophication, acidification or global warming).
With the exception of photochemical ozone formation, there is no other impact
category that covers even 100 different elementary flows. In this respect, there is
hence a factor of >1000 between other impact categories and the toxicity categories
(human health and ecotoxicity). This means that due to the large variety of sub-
stances with a toxicity potential, there will always be a very large uncertainty
inherent in these categories, although developers will eventually be able to lower
some of the model and parameter uncertainties currently observed. Excluding them
from the assessment because of their uncertainty would therefore mean that toxicity
would never be considered in LCA, which clearly risks violating the goal of LCA to
avoid problem-shifting from one impact category to another. Besides, the uncer-
tainty of assigning a zero-impact to a potentially toxic elementary flow by
neglecting the toxicity impact categories is certainly higher than the inherent
uncertainty of the related characterisation factors.
The solution rather lies in the way we interpret such inherently uncertain impact
potentials, whereas a more certain impact indicator may allow for identifying the
exact contribution of each elementary flow to the total impact in this category,
toxicity indicators allow for identifying the (usually 5–20) largest contributing
elementary flows, which will constitute >95% of the total impact. A further dis-
tinction between these will not be possible due to their uncertainty. Assuming that
an average and complete LCI may contain several hundreds of potentially toxic
elementary flows, one can then disregard all the remaining (several hundred) flows
due to their low contribution to total toxicity. A further discussion and recom-
mendations can be found in Rosenbaum et al. (2008).
Overall uncertainty in LCA is comprised of many different types of uncertainty as
further discussed in Chap. 11. Variability (e.g. spatial or temporal/seasonal) may also
be an important contributor, which should by principle be considered separately, as its
contribution can be reduced to a large extent by accounting for it in the characterisation
as discussed above for spatial variability and regionalised LCI and LCIA. Uncertainty
in LCIA can only be reduced by improved data or model quality, essentially coming
from updated LCIA methods, which is a good reason for a practitioner to keep up with
the latest developments in LCIA, which may well lead to less uncertain results than the
method one has been using for ten years. Most existing LCIA methods do not present
information about the uncertainty of the characterisation factors.

10.2.3.13 What Are the Main Assumptions?

In current LCIA methods, some assumptions are considered as a basic requirement
in the context of LCA: