Page 216 - Materials Chemistry, Second Edition

L1644_C05.fm Page 189 Monday, October 20, 2003 12:02 PM

5.8.1 APPLICATION OF THE FRAMEWORK TO LIFE-CYCLE INVENTORY
OF ELECTRICITY PRODUCED BY A WASTE INCINERATOR

This section presents an application of the framework for uncertainty assessment
corresponding to the case of LCIs mentioned earlier. For this purpose, the industrial
process of electricity produced by the MSWI, discussed in Chapter 1 to Chapter 4,
was once more selected. The following goals were proposed for the sake of a more
didactical and practical example:

1. Assigning probability distributions to the parameters considered in the
study
2. Assessing the uncertainties and variations in the calculation of the LCI
table
3. Determining the most relevant parameters in such LCI by sensitivity
analysis

5.8.1.1 Assigning Probability Distributions to Considered
Parameters
The predominant pollutants identified and quantified during the implementation of
the LCI for the MSWI study were selected by a combined quantitative and qualitative
approach. The quantitative selection consisted of a dominance analysis performed
on the basis of the results in the impact assessment carried out using the eco-indicator
95 method (see Chapter 3). Figure 5.9 presents the contribution of the considered
pollutants to the total environmental potential impact measured by the eco-indicator
95. As a selection criterion, only the emissions with a contribution to the total
environmental impact higher than 1% will be selected for the uncertainty assessment.
The results of the quantitative selection established that the atmospheric emission
of cadmium (Cd), carbon dioxide (CO ), chloridric acid (HCl), nickel (Ni), sulfur
2
dioxide (SO ), other heavy metals (HMs) and particulate matter (PM) would be
2
taken into account. Moreover, because of their carcinogenity and consideration as
primary air pollutants in the ExternE project (EC, 1995, 2000), arsenic (As), carbon
monoxide (CO) and PCDD/Fs were also to be considered (Figure 5.9).
A proper determination of the probability distribution is possible if data are
extensively available, as in the case of measured emissions, electricity production,
working hours and flow gas volume. Here the probability distributions were calcu-
lated from experimental data provided by the LCA study (STQ, 1998) and by the
MSWI director by means of a report (Nadal, 1999) or personally. Based on a relevant
number of measurements and their inherent variations, the normal or log-normal
distribution was selected as the best-fitting probability density function for the
respective types of data. The quality of the fitting was assessed by the Kolmog-
orov–Smirnov test for parameters with less than 30 measurements and by the Chi 2
test for parameters with more than 30 measurements.
The software Crystal Ball allowed carrying out this fitting of probability distri-
butions. The variation of the emissions in the study was enormous due to the constant
variation in the waste’s incinerated composition. The concentrations of the

© 2004 CRC Press LLC