Page 222 - Materials Chemistry, Second Edition

L1644_C05.fm Page 195 Monday, October 20, 2003 12:02 PM

the atmospheric emissions of the same unit without AGTS were evaluated. Each
simulation has been made in one separate run per scenario. Because of the inherent
variability of the MC model, it is not possible to affirm that the set of values relative
to the input variables used in the run of the current situation are going to be the
same as those in the former situation. The reason for this is that every run occurs
in different ways due to the generation of random numbers. In order to verify the
importance of this variability on the final outcome, both simulations were also carried
out in one run, and the results obtained showed negligible variations.
Figure 5.10 presents the results of the substances considered in this study in the
current situation with advanced AGTS. In the x-axis, it is possible to observe the
amount of pollutant emission per energy produced. The y-axis shows the probability
of each value of the life-cycle emissions. As mentioned before, 10,000 iterations
were carried out with the software Crystal Ball. The mean atmospheric emissions
of the heavy metals per 1 TJ of electricity produced by the incinerator were 7.50 ¥
–1
–1
10 kg with a normal standard deviation of 5.50 ¥ 10 kg/TJ. Because the density
distribution of the results is best adjusted by log-normal density function, a geometric
–1
mean of 6.10 ¥ 10 kg for HMs/TJ and a geometric standard of 1.92 were calculated.
–1
On this basis, a 68% confidence interval from 3.18 ¥ 10 kg/TJ to 1.17 kg/TJ was
obtained.
Because the other pollutants can be adjusted well by a log-normal distribution,
all the atmospheric emissions were treated in the same way, with their mean in the
68% confidence interval. In this framework, the calculated values of m and s for
g
g
–2
–2
As, Cd and CO were, respectively, 3.37 ¥ 10 kg/TJ (2.30), 2.62 ¥ 10 kg/TJ (1.67)
2
and 2.43 ¥ 10 kg/TJ (1.37).
The CO emissions of the incineration process were not determined by the
2
measurements assumed to have a log-normal distribution. In this case, the total
amount of waste treated was multiplied by the percentage fraction of plastics present
on it because this material is the only one in the mixture component that originally
comes from fossil fuels. Taking into account that 1 kg of plastic burned produces
approximately 2.0 kg of CO , the total CO produced by the waste incineration was
2
2
divided by the gas volume emitted to the atmosphere through the stack in order to
determine the CO concentration released. Thus, for this pollutant a normal distri-
2
5
bution with a m of 3.10 ¥ 10 kg/TJ and a s of 1.13 was obtained.
g
g
2
The profile of the LCI results for SO — 2.12 ¥ 10 kg/TJ (1.29) — to the same
2
situation is similar to those for NO and CO due to the same order of magnitude in
x
the s for the incinerator’s emissions. In Figure 5.10, the LCI results obtained by
g
MC simulation for the particulate matter (particles) in the former situation (Scenario
2
1) are also illustrated (m 1.50 ¥ 10 kg/TJ, s 1.93). If the results of the former
g
g
situation and the current situation are compared, a clear change can be seen in the
probability distribution from a log-normal to a rather normal one after the installation
of the advanced AGTS.
The mean values with confidence intervals for the former situation and the
current situation related to all studied pollutants are presented in Figure 5.11. Heavy
metals were only considered as a summed parameter. For the PCDD/Fs, HMs, SO 2
and HCl a clear reduction can be observed with the installation of the advanced
AGTS, especially for the first one. On the other hand, for CO , CO, PM and NO x
2
© 2004 CRC Press LLC