We can manage what we can measure.

Peter Drucker
In Chapter 8, we learned about the four steps that are common to ah sustainabihity
assessment methods and discussed the first step, that of goal definition. In this
chapter, we will focus on the second step, which involves obtaining data for doing
theassessment, and then using it in .the calculations. We will consider various
sources of data for building process and input-output networks, and address prac-
tical issues such as allocating the resource use and emissions between multiple
products. Our approach in this chapter will rely on relatively simple mathematics,
with a more general, mafrix-based framework provided in Chapter 10. We wil also
cover approaches for assessing the effects of uncertainties in the inventory data.

Calculations with life cycle network modeis like those described in Section 8.3.2
require data about tese1ected activities and their fiows. Tjiese data are referred
fo as the life cycle inventory (LCI). übtaining data for a large life cycle is a
-formidable task owing to the diversity of processes and the need for quantifying
inputs and outputs of many types of raw materials, fuels, products, byproducts,
and wastes. Consider the ethanol life cycle shown in Figure 8.4. The inventory
requires data from diverse areas such as engineering, agriculture, and transporta-
tion. Using expertise in each area it is certainly possible to determine inputs and
outputs for each process. This could rely on fundamental knowledge and inod-
els about the underlying processes, or on practical knowledge about the fiows
in functioning systems. Traditionahly, efforts toward process modeling or indus-
trial data measurement have focused mainly on economically valuable fiows such
as raw materials and products. However, sustainability assessment requires data
about wastes and emissions, which may not be readily available froni traditional
modeis and databases.
Over the last several years, LCIs for a large variety of systems have been
compiled and have become available as free and commercial databases. These
inventories are usually based on data from relevant activities. To address corporate
154 9 Inventory Anaysis

a
Flow t•.

Benzene 'Ç
Carbon dioxide fossil
Carbon mnxide:
fossil
Cobalt :. :.:.Aii/uníf
Dmitrogen monox]de Airlunspeci
Electncity natural gas
at power plarit
Formaldehyde
Lead
Mercury
Methane fossil
Nitroen 9?r$.S
Particuates
and < 10 :

Sufiiroxid
.7.
r.

VOCs
- compo s)-
I.iT:I
Ddínmytraiig
pipeline, un
•Natural Ç pips
t ii1 Z' .........
Transport combrnatio -
truc avrge fiie1
muc "'' • l ( .%
Transpo4n,
-
.. : ¿*• e

proprietary concerns, data are collected from multiple processes of the same
activity, and only their average values are included in the inventory databases.
Typical inventory data for a process are shown in Table 9.1. Note the compre-
hensive nature of this data since it contains even relatively small fiows, such as
those of lead, mercury, and dinitrogen monoxide, and relevant activities such as
transportation. Such modules for various processes are used to develop process
network modeis, as we will learn in Section 9.2 and Chapter 10.
For input-output networks, data for each sector and fiows between sectors are
compiled by government agencies and are available in public databases. As we
learned in Section 8.3.2, these data are even more aggregated than those of process
 9.2 Cakfflations 155

The coefficients represent life cycle fiows per dollar of final demand in temis of producer prices.

modeis, but they comprehensively cover an entire region. Sorne typical data from
input-output life cycle networks are shown in Table 9.2. The coefflcient is the
total direct and indirect orlife cycle floj.perdollar of final demand, which is the
nonetary value of the consumption from the selected sector. tie final demand is
the consumption by consurners, not other economic sectors. Mathematical details
about the underlying calculations are presented in Chapter 10.
Sorne popular sources of process and input-output inventoiy data are summa-
rized in Table 9.3. Metliods and tools that rely on such data will be our focus in
subsequent chapters.

Pp.
With LCI data for multiple processes, it is possible to calculate various quantities
for the entire life cycle network formed by connecting the process modules. The
data in each process module are usually normalized to a unit output of the prirnary
product. For example, the inventory in Table 9.1 is normalized to produce 1 kWh
of the main product: electricity from natural gas. For relatively simple networks,
it is easy to determine the scaling factors for each module in order to connect
them and thus obtain results for the selected functional unit, as illustrated in the
following examples. For more complicated networks, it may be necessary to rely
on the rnathernatical framework described in Section 10.1 and software.
Example 9.1 A module Pl for fuel production has an input of 50 L of crude oil to
produce 20 L of fuel. It emits 2 kg SO2 and 10 kg CO2. A module P2 for electricity
generation produces 10 kWh of electricity from 2 L fuel. It emits 0.1 kg SO2 and
156 9 tnventory Analysis

--

r).
Source Scale Featuscs Refuu]j

NREL..

GRÉET

Ecoinvcnt
4.

YRI.
r
FAüstat

1 kg CO2. For a network constructed by connecting these modules, determine the

life cycle consumption of crude oil and the emissions of CO2 and SO2 in generating
1000 kWh electricity.

Solution
The two modules are shown in Figure 9.1. To gencrate 1000 kWh of elcctricity, P2
needs to be scaled by a factor of lO = 100. This scalcd-up module will require
200 L of fuel. To supply this amount of fuel, P 1 needs to be scaled by a factor of
= 10. The resulting life cycle network is shown in Figure 9.2. As can be seen
in this figure, the total consumption of crude oil is 500 L, the emission of SO2 is
30 kg, and that of CO2 is 200 kg.

Crude oil, 50 L Pl Fuel, 2 L P2 Electricity, 10 kWh

Fuel, 20 L
Fuel Electricity
p rod uct ion generation

2 kg SO2, 0.1 kg

CO2, 10 kg CO2, 1 kg

Figure 9.1 Modules for fuel production and electricity generation.

 9.2 Calculations 157

Crude oil, 500 L Pl Fuel, 200 L Electricity, 1000 kWh

.
Fuel Electricity )
production generation

SO2, 20 kg SO2, 10 kg

100 kg CO2, 100 kg

Figure 9.2 Life cycle network for producing 1000 kWh of electricity. The modules in Figure 9.1 were
scaled linearly to obtain this network. Module Pl 15 scaled by a factor of 10 and P2 by a factbr of 100.

Example 9.2 Determine the total amount of benzene emitted in the extraction
and conversion of natural gas into 500 kWh of electricity.
Solution
This problem involves two processes: the extraction of natural gas, and the con-
version of natural, gas into electricity. The data in Table 9.1 are for the conversion
process. They show that 10-8 kg of benzene is emitted per kWh of electricity. For
the extraction of natural gas, we need similar data. Sources such as the NREL
inventory database [1] contain such data. From this source, we can find the mod-
ule of "natural gas, at extraction site." It shows that benzene emission is 5.31E-06
kg per cubic meter of natural gas extracted.
From the conversion module in Table 9.1, generating 500 kWh of electricity
emits
kWh > 10_8 kg benzene — 5 x 10 kg benzene
500
kWh —
The volume of natural gas required to produce 500 kWh electricity is

500 kWh x 0.298 kWh = 149 m3 natural gas

For the extraction module, the benzene emission for 149 m3 of natural gas is
kg benzene
5.31 x 10—6 x 149 ni3 = 7.91 x 10 kg benzene
m3
Therefore, the total rnass of benzene emitted in the extraction and conversion steps
for producing 500 kWh electricity is
5 x 10 + 7.91 x i0 = 7.96 x i0 kg benzene
Example 9.3 Using the input-output (10) data in Table 9.2, determine the life
cycle emissions and resource use in the generation of 500 kWh of electricity in
the USA. The producer price of electricity is 5 cents/kWh.
Solution
The producer price of 500 kWh electricity is 500 x 0.05 = $25. This is the final
demand of electricity. Multiplying the coefficients for the electricity sector in
1 58 9 Inventory Analysis

Table 9.2 by this price gives life cycle fiows of 156:75 kg CO2 to air, 0.156 m2y
cropland, 4.05 x 10 kg benzene to air, 2.55 x 10 kg glyphosate to water, and
0.16 m3 groundwater.

Example 9.4 A gas-electric hybrid car uses 100 L gasoline and 200 kWh elec-
tricity. The producer price of gasoline is $0.39 per L, and of electricity is $0.05 per
kWh. The electricity is produced from natural gas. Determine the life cycle CO2
emission for this system based on (a) an input-output model and (b) constructing
a hybrid model that uses the process data in Table 9.1.
Solution
Using the input-output model for the USA in Table 9.2,

10 model life cycle CO2 emissions (0.503 x 0.39 x 100) + (6.27 x 0.05 x 200)
= 82.3 kg

In this case, the electricity sector represents the average for ah modes of generation
in the USA.
Knowing that electricity is generated from natural gas, we can use detailed
process data for this process and combine it with the JO data for gasoline. Thus,

Hybrid model life cycle CO2 emissions = (0.503 x 0.39 x 100) + (0.585 x 200)
= 136.6 kg

Note that in the hybrid model, CO2 emissions from activities upstream of the power
generation process are ignored.

Allocation. Consider a process that makes two products and emits 10Ó kg of
CO2. Each product is usd further in other processes. A practical challenge in
assessing the CO2 emission from each product's life cycle is partitioning the 100
kg emission between the two products. Such processes with multiple coproducts
are quite common. For example, as shown in Figure 8.4, corn farming produces
com and stover, the corn being used for food and the stover for maintaining soil
quality or for making cellulosic ethanol. Corn fermentation produces ethanol and
distiliers' dried grains, the former being used as a chemical intermediate and the
latter as animal feed. Similarly, power generation from coal produces electricity,
heat, ash, carbon dioxide, and other byproducts, and the fractionation of crude
oil produces a large number of products ranging from methane and ethane to
gasoline, aviation fuel, and tas. The foliowing illustration introduces the approach
and issues in the allocation of LCI data among coproducts..
Consider a process that converts resources A and B into two products, P and
Q, as shown in Figure 9.3. Since the products are likely to be used to make other
items or will be soid to different buyers, the raw materials and emissions need
 9.2 Calculations 159

A P
B

(a) System for allocation by partitioning or (b) AHocation by displacement

no allocation

Figure 9.3 Allocation methods.

to be allocated to each product. Three approaches for addressing this allocation

problem are as foliows.

i
,' - ç.
wr
D if

e
t[:

• Partitioning. This approach splits or partitions the quantity of raw materials

and emissions on the basis of a selected feature of the products, such as mass,
energy, or monetary value. Thus, in Figure 9.3 a, a quantity, say, x is partitioned
between the two products as foliows:
mp
Xp (9.1)
mp + mQ
mQ
IQ = (9.2)
mp + mQ

Here, the allocated quantities Ip and x are determined in proportion to their

rnasses, mp and mQ. Other allocation criteria could be monetary value, fuel
value, or exergy. This is the easiest and most common-allocation approach.
160 9 Inventory Analysis

However, it is also highly subjective, since seledting the partitioning criterion

is lot easy owing to the high degree of subjectiveness. Such subjectiveness
appears in many common problems, like the one described in Box 9.1. Fur-
therrnore, in problerns in which alternatives are being compared, the results
may be contradictory for different partitioning criteria, as illustrated near the
end of this section.
Displacement. Another approach for allocation is based on displacement or
system expansion. This is illustrated in Box 9.1. In this case, the byproduct
is considered to substitute the same flow as that produced by a conventional
process. The process being analyzed takes credit for displacing the conven-
tional process. Forexamp1e, converting sugarcane into ethanol also produces
bagasse, which is comrnonly converted to electricity. In the displacement
approach, this electricity from bagasse is considered to displace the electricity
obtained from conventional sources in the region. Then the emissions from
these conventional sources for producing the displaced electricity are sub-
tracted from the sugarcane ethanol analysis. Thís approach is ilustrated in
Figure 9.3b, where the emissions from the conventional process used to pro-
duce the byproduct Q, denoted by the shaded box, are subtracted from those for
the process being analyzed. A challenge that may be encountered in using this
approach is that the process being displaced may itself have sorne coproducts
that also need to be displaced and so on, resulting in an infinite displacement
problem.
• No allocation. If a process is not fufly understood, then ah the products from
it may not be known. In this case, partitioning or displacernent rnethods can-
not be applied. This situation is often encountered in ecological systems. For
example, solar energy drives a forest ecosystem, but the products and ser-
vicesprovided by this system may not ah beknown. Another reason behind
this approach is that partitioning and displacernent implicitly assume that the
coproducts can be produced independently, but this may not be possible for
many coproducts. Examples include corn and corn stover, the yellow and white
portions of an egg, and goods and services from ecosysterns. This approach
rnay be ilustrated with the help of Figure 9.3a. In this case, the allocated
quantities Xp and .r may be determined as
Xp-X (9.3)
XQ = 1 (9.4)
Cornparing these equations with Equations 9.1 and 9.2, we can see that in the
no-ahlocation approach, the ahlocation coefficient is unity for both products.
A challenge in this approach is that when such coproducts are combined, the
allocated quantities cannot be added since that will result in double counting.
 9.2 Cakulations 161

co2, 120kg CO2, 2 kg

•A P, $20, 40 kg
Q, 1 kg
B Q, $200, 200 kg

(a) Life cycle of product P and byproduct Q. (b) Lite cycle of process for producing Q.

co2, 120 kg

P, $20, 40 kg
) Q.......
200kg —200
Co2, —400 kg
(c) AHocation by displacement.
Figure 9.4 lllustration of allocation methods.

This can be understood by considering a process that combines P and Q. With

Xp and XQ calculated by Equations 9.3 and 9.4, respectively, we have xp+XQ =

2x, which is not possible since only the amount x was used to produce P and
Q. This approach is mainly used in the technique of emergy analysis, which is
described and ilustrated in Chapter 14 along with special rules that have been
devised to avoid double counting.

Example 9.5 For the life cycle network shown in Figure 9.4, the main product is
P and the byproduct is Q. The total CO2 emitted in the supply chain is 120 tons.
The masses and costs of P and Q are as shown in Figure 9.4a. Another process
for producing only byproduct Q is shown in Figure 9.4b. Allocate the CO2 ernitted
from life cycle 1 between the two products using partitioning, displacement, or
no-allocation.
Solution
The total CO2 emitted for life cycle 1 is x = 2O kg.
Partitioning. Using monetary values and Equations 9.1 and 9.2,

20
Xp$ x120
20+200
= 10.9 kg
200
ZQ$ = x120
20+200
= 109.1 kg

Similarly, using mass allocation, Xp,mass = 20 kg, and XQ,mass = 100 kg.
162 9 Inventory Analysis

Displacement. Figure 9.4c shows the modified network with byproduct Q con-
sidered to replace the conventional process that is currenfly producing Q. As
shown in this figure, replacing 200 kg of Q means saving the emission of 400 kg
of CO2. So the net CO2 ernission for producing P S XP,djsp = 120 — 400 = — 280 kg
CO2.
No allocation. Without allocation, we use Equations 9.3 and 9.4 to get
120 kg CO2 and XQ,none = 120 kg CO2.

Example 9.6 Two alternative processes are to be compared in terms of their

water use per unit of the desired product, P. Process 1 uses 10,000 L of water
to produce 200 kg of P and 50 kg of Q, which are worth $1000 ard $100, respec-
tively. Process 2 uses 12,000 L of water to produce 100 kg of P and 100 kg of R,
which are worth $4950 and $50, respectively. Using mass allocation and mone-
tary allocation, identify in each case the process that is more water-efficient per
kilogram of P.
Solution
The allocation coefficients for the two processes using mass and monetary
allocation are summarized in Table 9.4.
By mass allocation, th water required per kilogram of P is 0.8 x 10,000 = 8000
L/kg of P for Process 1, and 0.5 x 12,000 = 6000 L/kg for Process 2. Thus, by
mass allocation, Process 2 is to be preferred for P.
By monetary allocation, the water required per kilogram of P is 0.91 x 10,000
9100 L/kg for Process 1, and 0.99 x 12,000 = 11,880 L/kg for Process 2. By
monetary allocation, Process 1 is preferable for minimizing the water use in
making P.

Example 9.6 provides conflicting results for the two allocation methods, making
it difficult to choose between the two processes. Given the subjectivity involved
in choosing the allocation method, the recommended approach by practitioners
of life cycle assessment is to prefer the displacement approach whenever possi-
ble. If th.is is not possible, allocation based on monetary values is recommended.
üften, multiple allocation methods are applied to a problem that involves choosing
between alternatives. If the decision does not change with the allocation method,
then the decision is robust to allocation.
 9.3 Uncertainty 163

Sustainability assessrnent rnethods combine information from diverse sources,

which are bound to be uncertain. Three categories of uncertain information in
these methods are:

1 Parametric uncertainty. This is uncertainty in the measured or observed data. It

represents the inherent variability over time of the measured variables, sensor
bise, and difficulties in measurernent.
2 Scenario uncertainty. This represents variability due to the choices made by the
modeler. It includes the effect of the allocation method or value judgment used
for combining multiple outputs.
3 Model uncertainty. This is uncertainty in inathematical relationships. It
includes errors in the development of physical, input-output, or hybrid modeis,
owing to the simplification of complicated relationships, and the extrapolation
of modeis of similar systems.

Life cycle inventory data, like ah measured data, are uncertain owing to errors in
measurement, aggregation of data from multiple sources and systems, and system
variability. Therefore, it is natural to account for the effect of uncertainties in
the results to determine whether the difference in results is within the margin of
error or is statistically significant. In this section, we will learn about sorne basic
methods that are used in sustainabi]ity assessment.
AH rnethods for uncertainty analysis require information about the uncertainty
in the data and the modeis used in the analysis. This information is used to
determine the uncertainty in the calculated quantities. The nature of the analy-
sis approach chan ges with the availability of uncertainty information, which can
be of three types.

• Qualitative inforination. Often, only qualitative i.nformation is available about

uncertainty. In this case, rnethods such as pedigree analysis may be used.
• Quantitative information, but without distributions. In this case, statistical dis-
tributions are not available, but the range of variation rnay be assumed. In
such situations, the sensitivity analysis approach is used.
• Quantitative information with distributions. In this case, detailed statistical
distributions are available. The approach of Monte Carlo analysis may be used.

We will learn about the first two approaches.

Pedigree method. This approach is useful in the absence of quantitative infor-
mation about uncertainty. It relies on selecting a numerical score representing
the quality of the fiows and processes in the LCI. Pedigree matrices for assign-
ing numerical scores to LCI fiows are shown in Table 9.5; smaller values indicate
less uncertainty. The matrix in Table 9.5 is used to evaluate various fiows in an
164 9 !nventory Analysis

inventory in terms of their reiiabiiity and representativeness. Assigning the scores

requires insight into the nature of the inventory data, but sorne degree of subjec-
tiveness may be unav'oidable. If no information is availabie about each category,
the default value of 5 is assigned, indicating the highest uncertainty.
Example 9.7 For the fuel production module in Figure 9.1, the crude oil data
form an estimate from an industry document that is five years oid. The data are
from the same geographic region and same process type as that of the study, and
represents average data over a year for the primary process being used in the
region. Determine values of the pedigree matrix for this flow.
Solution
Based on Table 9.5, we see that for the category of flow reliability, the value is 4
since the data are documented. For temporal correlation, the value is 2 since the
data are less than six years oid. Continuing in this manner, the pedigree vector for
this flow is [4,2,1,4,1].
The pedigree method results in a matrix of uncertainty indicators for each flow
and process in the inventory. These data may be combined as a weighted sum
of fiows of the same type in each module and their uncertainti.es, foliowed by
normalization by the total flow:

IIkhikik
u — .5
>Ik hik
where i represents the impact category, j is the uncertainty type, and k is tiie type
of inventory module.
Examp!e 9.8 The LCI for truck transportation consists of diesel production and
transportation. Information about emissions of particuiate matter of sizes 2.5 and
10 microns and their pedigree matrices are given in Tabie 9.6. Determine indicators
for the uncertainty in the life cycle ernissions for each category.
Solution
Uncertainty in flow reliability for PM2.5 and PM10 are calculated by using
Equation 9.5 as foliows:
(2 x i0 x 5) + (6.27 x i0 x 1)
UPM2.5,1 =
2 x i0 5 + 6.27 x io-
= 4.9
(2.79 x iO x 1) + (6.3 1 x i0 x 3)
UPM1o,1 =
2.79 x io- + 6.31 x iO
= 1.0
 1iMT. (
A ator2 :-
-.

eof dat
•oUiftnce, onprmpre
tftt ,i
5'yéars
ron1 diffuent
ofiiutipi r unknown arca
3? 4

1Í1atd. ra of
5 fud ""
;
;
!J 1&Y
catóe$are
er1
a Verification may take place in several ways, e.g., by on-site checking, by recalculation, through mass balances or coss-check.s with other sources.
b Temporal difference refers to the difference betwcen date of data generation and the date of representativeness as defined by the goal of the projcct.
C
A related area of study is dcñned by the user and should be documented in the geographical metadata.
d Technology cat:gories are proccss design, operating conditions, material quality, and proccss scale.
C
The relevant market should be documented in the study. The default relevant market is measured in production units.
Adequate time period can be evaluated as a time period long enough tu even out normal fluctuations. The default time period 15 1 year, cxcept for emerging
technologics (2-6 rnonths) or agricultural projccts >3 ycars.
 ata colledtion
.j r. rncthod

-J
m.- •.
- o s
Lt
•- ±:, ...Q v'i'k
168 9 Inventory Analysis

Here, the second subscript, 1, stands for the first uncertainty type of flow reliabil-
ity. The calculated indicators for each uncertainty type and emission are shown in
I:he last two rows of Table 9.6. The results convey that the data for PM1O are of
higher quality since the uncertainty is lower in this case.
Sensitivity analysis. When information about the quantitative range of varia-
tion is available, this approach determines the sensitivity of the results to errors in
contributingvariables. This helps in identifying the variables that have the largest
effect on the variabiity of the results. More effort may then be directed toward
determining the uncertainty in these variables.
Example 9.9 The environmental impacts ZA and zB of two product options, Prod-
uct A and Product B, are calculated by combining their emissions x and y as
ZA = 2z + Sy and zB = 4x + y. Knowing that x = 10 and y = 50, which
product has a smaller impact? How robust are your results to uncertainty in the
data?
Solution
Based on the specified values of x and y, we ca1culate ZA = 270 and zB = 90.
Thus, Product B has a smaller impact.
To perform a sensitivity analysis, since we do not have information about uncer-
tainties in x and y, we will assume the uncertainty in each variable to be ±10
percent. This implies that x may take values between 9 and 11, while y is between
45 and 55. The results of the sensitivity analysis are shown in Table 9.7. These
results show the range of ZA to be 243-297, while that of ZB is 81-99. Thus, even
after accounting for the variabiity, Product B has a smaller impact. The results in
Table 9.7 also indicate the sensitivity of ZA and ZB to the input variables, z and y.
From the ñrst two rows we can see that for a ±10 percent change in 1, ZA vanes
by ±0.8 percent while ZB vanes by ±4.7 percent. For alO percent change in y, ZA
varies by ±9.3 percent and ZB vanes by ±5.9 percent. Thus, for these values of
x and y, ZA 5 more sensitive to changes in y, and ZB is almost equally sensitive
to changes in both z and y, but slightly more to the latter variable. These results
imply that if ZA 5 the variable of interest, we should pay more attention to getting
more accurate measurements of y. If zB is the variable of interest, the accuracy of
both variables is equally important.
 9.5 Review Questions 169

Example 9.10 For thesystem in Figure 9.4a, suppose that the quantity of CO2
emitted is related to the inputs of A and B as foliows: Xc02 = lOTA + 2OXB. For fue
values depicted in the figure, XA = 2 and 1B = 5. (a) Determine the sensitivity of
the calculated CO2 emissions to uncertainties in the two input variables. (b) If the
error in 1A is known to be ±20 percent and in xB to be ±10 percent, determine
the error in XCO2.
Solution
(a) Let us consider the effect of ±10 percent variations in XA and 1B Ofl XCO2. The
ránge of 'A will be [1.8, 2.2] and of 1B will be [4.5, 5.5]. The resulting range of
xCO2 will be [108, 132]. The value of CO2 is more sensitive to variations fl XJ3 tha.fl
in
(b) The range of 1A is [1.6, 2.4], and of B S [4.5, 5.5]. Then, CQ2 vanes between
[106, 134].

In this chapter we learned about the second step in sustainability assessment:

inventory analysis. This step involves obtaining data for ah the activities included
in the life cycle model. Databases containing such data are available for individual
processes and for economic sectors. For activities that produce multiple outputs,
emissions and resource use need to be allocated between these outputs. Allocation
is a subjective exercise, and we covered methods based on partitioning, displace-
ment, and no allocation. Finaily, since data are bound to contain errors, we learned
about the pedigree method and sensitivity analysis.

Key Ideas and Concepts

• Process network • Input-output network
• Hybrid network • Functional unit
• Aliocation by partitioning • Allocation by displacement
• Life cycle inventory • Uncertainty analysis

9.5 Review Questions

1. Explain the meaning of a scaling factor.

2. Define final demand.
3. Which are fue three methods that can be used to allocate emissions from a
process between multiple products?
4. If different ahlocation methods give conflicting results, what is fue
recommended approach?
170 9 Inventory Analysis

5. Which method should be used for uncertainty análysis if information about

uncertainty is qualitative?

Problems

9.1 Find the foliowing data from the appropriate database in Table 9.3. State
your assumptions, if any.
1. The fossil energy use and particulate matter emission to air from
hauling a ton of goods from Los Angeles to Las Vegas.
2. The carbon dioxide emitted from the electricity used in the production
of phenol formaldehyde resin.
3. Determine the top five emissions by mass from the cement sector
(NAICS 3273) in the USA.
4. Which countries are the top ten emitters of nitrous oxide (N2 0) from
fertilizer use in agriculture?
9.2 Process 1 produces P, Q, and R. The masses of P, Q, and R are 100 kg, 20
kg, and 5 kg, respectively, and their monetary values are $1000, $100, and
$20. This process emits 500 kg of CO2. Process 2 produces P and S with
masses of 100 kg and 5 kg and monetary values of $1000 and $0. This
process emits 1000 kg CO2. Process 3 produces 1 kg of S with CO2
emission of 0.8 kg. Choose the process for producing P that has the lewest
CO2 emission by using
1. mass-based allocation
2. monetary allocation
3. displacement and mass-based allocation
4. displacement and monetary allocation
9.3 A life cycle assessment of disposable coid cups conducted for Starbucks
Coifee Company and NatureWorks LLC [9] compared cups and uds made
out of polypropylene (PP), polyethylene terephthalate (PET), and
polylactide acid (PLA). The report provides the following information
about flow representativeness of the data for PET.
• Data are from 2002 to 2007 for the report prepared in 2009.
• Geographical coverage is for the USA.
• Technological data are based on detailed modeis of manufacturing.
Using this information, suggest imcertainty indicators for the four
categories of flow representativeness.
9.4 Sustainability assessment of emerging technologies can be veiy useful for
guiding their development in a way that prevents unintended harm. Since
such technologies are often not yet implemented, inventory data are
 9.5 Review Questions 171

estimated by theoTetical modeis. Using the pedigree matrix, determine the

quality indicators for flow reliability and flow representativeness.
9.5 An iritegrated gasification combined cycle (IGCC) plant can convert coal
and water to electricity and hydrogen. The carbon dioxide from such a
process may be captured by a double-stage solvent such as Selexol. The
main inputs to such a system are 5426 tons of coal and 21,787 m3 of water
to produce 560 tons of hydrogen, 30.3 MW of electricity, 11,253 tons of
pure carbon dioxide, and 1293 tons of carbon dioxide in flue gas. The cost
of the hydrogen is $2110 per ton, of the electricity is $1800 per MW of the
pure CO2 is $37 per ton, and of the flue gas is zero [lo]. Allocate the water
use among the products of this process by the foliowing approaches and
discuss the pros and cons of each approach for this allocation problem:
1. rnass-based allocation
2. energy-based allocation
3. monetary value-based allocation
What additional information wili be required to do allocation by
displacement in this problem?
9.6 A well produces 6000 standard cubic feet (scf) of gas for every barrel (bbl)
of hydrocarbon liquid.
1. Assuming a heating value of 1000 BTIJ/scf for gas and six million
BTU/bbl for hydrocarbon liquid, allocate the leakage (emission) from
this weli between the two products.
2. Considering a gas density of 25 g/scf and an oil density of 100 kg/bbl,
allocate the ernission in terms of mass.
3. Using prices of $3.50 per 1000 scf for gas and $90 per bbl for
hydrocarbon liquid, perform a monetary allocation.
9.7 The information about sorne LCI modules is as foliows.
• 1 kWh of electricity requires 0.64 kg of water, and the process ernits
1.12kg of CO2, 1.1 x i0 of SO2, and 1.97 x iO kg ofN0.
• 1 kg of polypropylene requires 31.9 kg of H20, 0.11 kWh of electricity,
and the process ernits 1.82 k.g of CO2 and 0.0031 kg each of SO2 and
N0.
• 1 kg of copper uses 4.38 kg of H20 and 0.49 kWh of electricity, and the
process emits 0.099 kg of CO2 and 0.41 kg of SO2
• 1 kg of steel consumes 0.22 kWh of electricity and 10.47 kg of H20,
and the process emits 0.074 kg of CO2.
• 1 kg of plastic extension requires 0.51 kWh of electricity, 4.7 x iO kg
of polypropylene, and 19.72 kg of H20.
• 1 paper towel dispenser requires a plastic extension that weighs 2.79
kg, 3.19 kg of polypropylene, 43.76 kWh of electricity, 0.17 kg of
copper, and 0.17 kg of steel.
172 9 Inveritory Analysis

Construct a process-model-based life cycle 'network for producing a paper

towel dispenser. Determine the amount of water used and the CO2 emitted
in the life cycle ofone paper towel dispenser. (Provided by Tanner
Anderson, Eric Bissonnette, Stephanie Gase, and Ruonan Zhao).
9.8 Answer the foliowing questions using data from Table 9.2.
1. A company spends $ 10,000 per year on courier services. Determine the
resulting life cycle emission of carbon dioxide and benzene. What type
of network model did you use forthis calculation?
2. A household's annual expenses include $1000 for gasoline, $800 for
electricity, and $300 for vehicle repair. Determine the cropland needed
to support these expenses. Explain which steps of the life cycle may
require the cropland most.
9.9 A life cycle assessment of an office building finds the climate change
impact of the building materials to be 4818 tons CO2 equivalent, of the
construction to be 822 tons CO2 equivalent, of the heating to be 11,219
tons CO2 equivalent, and of the electrical services to be 25,230 tons CO2
equivalent. The pedigree vector values for data representing these
activities are (2, 1, 2, 2, 2, 2), ( 2, 1, 2, 2, 2, 3), (2, 2, 1, 1, 1, 1), and (2, 2, 1,
1, 1, 1), respectively. Determine the pedigree vector values for the total
climate change impact of these activities.
9.10 In a product life cycle, the emission of CO2 and the use of water depend on
three variables, x•, ¡ = 1, 2, 3, according to the foliowing equations:
yco2 = 10011 + 1112 -713 ; YH2o = -12 + 5013. The nominal values
of z are 10, 50, and 5.
1. Using sensitivity analysis, determine the variable whose accuracy is
most important.
2. Ifthe range ófvariation of each variable is II [45,551,12 [24,361,
13 [4,7], determine the range of values for CO2 emissions and water
use. Also determine the most important variables.

References

[1] National Rencwable Energy Laboratory. US life cycle inventory database. www.
lcaconimons . gov/nrel/search, accessed November 3, 2014.
[2] Y. Yang, W. W. lngwersen, T. R. Hawkins, M. Srocka, and D. E. Meyer. USEEIO: a new
and transparent United States environmentally-extended input-output model.
Journal of Cicaner Production, 158:308-3 18, '01.
[3] Argonne National Laboratory. The greenhouse gases, regulated emissions, and
energy use in transportation (GREET) model. http: / /greet . es . ani . gov,
accessed January 31, 2012.
 References 173

[4] Swiss Centre for Life Cycle Inventories.'Ecoinvent life cycle inventory database.
www. ecoinvent . ch, accessed January 18, 2013.
[5] Envirorimental Protection Agency. National emissions inventory. www. epa. gov/
ttn/chief/trends, accessed August 22, 2015.
[6] Environmental Protection Agency. Toxics release inventory. www. epa. gov/
triexplorer, accessed August 22, 2015. -
[7] Food and Agriculture Organization. FAOSTAT. http: //faostat3 . fao. org,
accessed July 8, 2017.
[8] A. Edelen and W. W. Ingwersen. The creation, management, and use of data quality
information for life cycle assessment. The International Journal of Life Cycle
Assessment, 23(4):759-772, 2018.
[91 PE Americas. Comparative life cycle assessment Ingeobiopo1ymer, PET, and PP
drinking cups. www.natureworksllc . com/-/rnedia/The_Ingeo_
Journey/EcoProfile_LCA/LCA/ PEA_Cup_Lid_LCA_FullReport_
ReviewStatement_12 12 09_pdf . pdf, 2009, accessed November 23, 2018.
[lo] B. Kursun, S. Ramkumar, B. R. Bakshi, and.L.-S. Fan. Coal gasification by
conventional versus calcium looping process: a life cycle energy, global warming,
land use and water assessment. Industrial El- Engineering Chemistry Research,
53(49):18910-18919, 2014.