Page 199 - Materials Chemistry, Second Edition

L1644_C05.fm Page 172 Monday, October 20, 2003 12:02 PM

Uncertainty of these parameters reflects directly on the outcome of any environmen-
tal impact method. Empirical inaccuracy (imprecise measurements), unrepresenta-
tive data (incomplete or outdated measurements) and lack of data (no measurements)
are common sources of parameter uncertainty. Weidema and Wesnæs (1996) describe
a comprehensive procedure for estimating combined inaccurate and unrepresentative
LCI data qualitatively and quantitatively. Although this procedure may substantially
improve the credibility of LCA outcomes, uncertainty analyses are generally com-
plicated by a lack of knowledge of uncertainty distributions and correlations among
parameters.

5.2.2 MODEL UNCERTAINTY

According to conclusions presented by various authors, the predicted values for
environmental impact and risk generally respond in a linear manner to the amount
of emitted pollutant. Moreover, in life-cycle impact assessment (LCIA) and IPA,
thresholds for environmental interventions are disregarded. Additionally, in LCIA
the derivation of characterization factors causes model uncertainty because these are
calculated with the aid of simplified environmental models without considering
spatial and temporal characteristics.

5.2.3 UNCERTAINTY DUE TO CHOICES
In many cases, performing choices is unavoidable in environmental impact analysis.
Considering the step of LCI from LCA, examples of choices leading to uncertainty
include the selection of the functional unit (or definition of the allocation procedure
for multioutput processes), multiwaste processes, and open loop recycling. More-
over, the socioeconomic evaluation step in LCA and IPA is an area in which choices
play a crucial role. Although experts from the social sciences have suggested many
different weighting schemes, only a few are operational and no general agreement
exists as to which one should be preferred.

5.2.4 SPATIAL AND TEMPORAL VARIABILITY
In most LCAs environmental interventions are summed up regardless of their spatial
context, thus introducing model uncertainty. Temporal variations, in turn, are present
in LCI and other impact assessment methods. In general, variations of environmental
interventions over a relatively short time period, such as differences in industrial
emissions on weekdays vs. weekends or even short disastrous emissions, are not
taken into account.

5.2.5 VARIABILITY AMONG SOURCES AND OBJECTS
In LCA or in other impact assessment methods, variability among sources and
objects may influence the outcome of a study. This means, for example, that some
variability in LCIs may result from differences in inputs and emissions of comparable
processes within a product system (due to the use of different technologies in
factories producing the same material). Furthermore, variability among objects exists

© 2004 CRC Press LLC