Page 351 - Materials Chemistry, Second Edition

5.3 Techniques for Result Analysis 335

• Product development and improvement
• Strategic planning
• Public policy making
• Marketing.

All of these applications may steer decisions even if those exclusively depend on
results of an LCA only in rare cases. It is therefore advantageous to formulate results
as quantitatively as possible and to indicate uncertainties of data and techniques.
This problem was already observed by SETAC and several authors at the time when
the LCA standardisation was still in process. 14)
Following the first round of standardisations (ISO 14040 to ISO 14043:
1997–2000), the quality-oriented method development became, under the
slogan ‘uncertainty’, an established working field. 15) It was realised that quality
management goes beyond the obvious problem of data quality and also concerns
methodological objectives such as

• selection of system boundaries;
• allocation rules, impact categories and indicators, including weighting factors if
present;
• the underlying framework of values.

Data as such are best suited for a mathematical examination, whereas the influence
of assumptions is best evaluated by alternative scenarios in the form of sensitivity
analyses.

5.3.2
Mathematical Methods

In view of the specific difficulties in the procurement of highly suited inventory
data or of their adaptation to a specific case, classical error calculations after Gauss
(± standard deviation) are seldom applied.
Heijungs and co-workers 16) discern five numerical types of analysis suited for
data analysis in the interpretation phase (partly already in the inventory analysis!):

• Contribution analysis
• Perturbation analysis
• Uncertainty analysis

14) Fava et al. (1994), Chevalier and Le T´ eno (1996), Kennedy, Montgomery and Quay (1996),
Kennedy et al. (1997), Coulon et al. (1997), Le T´ eno (1999) and Hildenbrand (1999).
15) Braam et al. (2001), Huijbregts et al. (2001, 2003), Huijbregts, Heijungs and Hellweg (2004),
Heijungs and Kleijn (2001), (Marsmann, 1997), Ross et al. (2002), Ciroth (2001, 2006), Ciroth,
Fleischer and Steinbach (2004), Heijungs et al. (2005), Heijungs and Frischknecht (2005) and
Lloyd and Ries (2007).
16) Heijungs and Kleijn (2001) and Heijungs et al. (2005).