Page 201 - Materials Chemistry, Second Edition

L1644_C05.fm Page 174 Monday, October 20, 2003 12:02 PM

5.3.2 FORMER UNCERTAINTY ASSESSMENT IN IMPACT PATHWAY
ANALYSIS

The IPA is a quite complex approach and hence risks lack of reliability in the final
results. In the same way as with other environmental analysis methods, uncertainty
is the key problem that makes it difficult to convince decision-makers based on the
outcomes of a study.
One of the most interesting experiences is that reported by Rabl and Spadaro
(1999), in which they evaluated the uncertainty and variability of damages and costs
of air pollution by means of analytical statistical methods. In this case, the authors
observed that the equation for the total damage is largely multiplicative, even though
it involves a sum over receptors at different sites. This conclusion comes from the
principle of conservation of matter, which implies that overprediction of the disper-
sion model at one site is compensated for by underprediction at another; the net
error of the total damage arises mostly from uncertainties in the rate at which the
pollutant disappears from the environment.
In the same reference, the authors discuss the typical error distributions related
to the factors in the equation for the total damage, in particular those related to two
key parameters: the deposition velocity of atmospheric dispersion models and the
value of statistical life; according to Rabl and Spadaro (1999), these are close to
log-normal. They conclude that a log-normal distribution for the total damage
appears very plausible whenever the dose–response or exposure–response function
is positive everywhere. As an illustration they show results for several types of air
pollution damage: health damage due to particles and carcinogens, damage to build-
ings due to SO , and crop losses due to O , in which the geometric standard deviation
3
2
is in the range of 3 to 5. Results and conclusions such as those presented by Rabl
and Spadaro illustrate the necessity of dealing with uncertainty assessment in IPA
in spite of its high level of complexity.

5.4 INTRODUCTION TO MONTE CARLO SIMULATION

Based on the previously mentioned experiences regarding uncertainty analysis in
LCA and ERA studies, especially to IPA, it seems that the use of a stochastic model
helps to characterize the uncertainties better, rather than a pure analytical mathe-
matical approach. This can be justified because the relevant parameters follow a
different frequency distribution. In this case, one of the most widespread stochastic
model uncertainty analyses is the Monte Carlo (MC) simulation. In a wide approach
to perform an MC simulation, the parameters under evaluation must be specified as
uncertainty distributions. The method makes all the parameters vary at random
because the variation is restricted by the given uncertainty distribution for each
parameter. The randomly selected values from all the parameter uncertainty distri-
butions are inserted in the output equation. Repeated calculations produce a distri-
bution of the predicted output values reflecting the combined parameter uncertainties.
According to LaGrega et al. (1994), MC simulation can be considered the most
effective quantification method for uncertainties and variability among the environ-
mental system analysis tools available.

© 2004 CRC Press LLC