Page 200 - Materials Chemistry, Second Edition

L1644_C05.fm Page 173 Monday, October 20, 2003 12:02 PM

in weighting environmental problems during impact assessment due to variability
in human preferences. For instance, when a method such as the willingness-to-pay
(WTP) method is used to determine the external environmental cost due to a specific
damage, differences related to individual preferences cause inherent variation in the
final result.

5.3 WAYS TO DEAL WITH DIFFERENT TYPES OF UNCERTAINTY

Huijbregts et al. (2000) have offered solutions on how to deal with the issues of
uncertainty previously discussed. The tools available to address different types of
uncertainty and variability in LCAs include probabilistic simulation, correlation and
regression analysis, additional measurements, scenario modeling, standardization,
expert judgment or peer review, and nonlinear modeling. Scenario modeling
(Pesonen et al., 2000) should be especially useful in cases in which uncertainty
about choices and temporal variability is present.
When a model suffers from large uncertainties, the results of a parameter uncer-
tainty analysis may be misleading. In most cases, the consequence of decreasing
model uncertainty will be the implementation of more parameters in the calculation,
thereby increasing the importance of operationalizing parameter uncertainty in the
model. In the following two sections, we present an overview of previous efforts to
assess uncertainties in LCA.

5.3.1 EXPERIENCES TO ASSESS UNCERTAINTY IN LIFE-CYCLE
ASSESSMENT
So far the influence of data quality on final results of LCA studies has rarely been
analyzed. In spite of the lack of published case studies, several approaches to carry
out this kind of evaluation have been proposed during recent years. Nevertheless,
from a general point of view, the existing methods can be classified in qualitative
and quantitative assessments.
Qualitative assessment means describing the data used by characterizing its
quality. Weidema and Wesnaes (1996) and Weidema (1998) proposed using data
quality indicators depending on categories like reliability, completeness, temporal
correlation, etc. In turn, Finnveden and Lindfors (1998) suggested ranges for
various inventory parameters as rules of thumb. Quantitative assessment means to
quantify all inherent uncertainties and variations in an LCA. In order to perform
this task, many different analytical procedures have been applied. For uncertainty
analysis of LCI, Hanssen and Asbjornsen (1996) used statistical analysis, Ros
(1998) proved the fuzzy logic, and Maurice et al. (2000) as well as Meier (1997)
decided in favor of the stochastic methods. Regarding uncertainty assessment
within the impact assessment stage of LCA, Meier (1997), Hofstetter (1998) and
Huijbregts and Seppälä (2000) have reported results achieved using similar tech-
niques. Even when it is strongly effective, quantitative assessment is continuously
confronted with the problem that it is hardly possible to analyze all types of
uncertainties.

© 2004 CRC Press LLC