You are on page 1of 9

Chapter 17A

An Illustration of the LCA Technique

Reinout Heijungs

Contents is not one unique interpretation and implementation of


the principles of LCA, and any attempt to illustrate
17A.1 Introduction ...................................................... 375
LCA by means of a concrete example should be
17A.2 Goal and Scope Definition................................ 376 understood as an illustration indeed. The purpose of
17A.3 Inventory Analysis – The Basic Case................. 376 this section is therefore to point out the main idea and
17A.4 Inventory Analysis – Some Complications ........ 378 methodological principles of LCA by means of a
hypothetical example with fictitious data.
17A.5 Impact Assessment – Definition
In doing so, we will concentrate on the computational
of Impact Categories, Selection
of Characterization Models, content of LCA. Of course, the entire concept of LCA
Classification and Characterization................ 381 embraces (or: may embrace) many types of activities and
17A.6 Impact Assessment – aspects, including involvement of stakeholders, collecting
Normalization and Weighting ......................... 382 of data, verification of data, manipulation of data, and
17A.7 Interpretation .................................................... 382
carrying out peer reviews. To what extent stakeholders
play a role and how this role should be embedded is
17A.8 Concluding Remarks ........................................ 383
something that not only is under discussion, but which
Reference............................................................................ 383 also varies from situation to situation. The same applies
to the process of peer review. Obtaining and verifying
17A.1 Introduction data are cumbersome activities which require a good
deal of process-technological and ecological and
toxicological knowledge. In contrast, the recipes that are
It has been mentioned in Chapter 12 that Life Cycle used to manipulate the data are of an almost mechanical
Assessment (or LCA) is one of the analytical instruments nature. There is not much disagreement (although at
within the toolbox for analyzing the interface of some places there is some). Strangely enough, almost all
economy and environment. This section provides a textbooks discuss those parts that are difficult and
worked example of the LCA technique. It is a fairly controversial, while the more or less rigorous basis is
elaborate demonstration of a quite simple hypothetical often left out of consideration. We feel, however, that a
product system. No real case study and data were presentation of these basic mechanisms of the method
chosen, the reason being that the number of data items for LCA is indispensable for a proper understanding of
would be too large to be sorted out in an educational the meaning of the results of an LCA.
setting. All data are therefore fictional and no claims All calculations that are made in this chapter are
should be made with respect to the results of the simple enough to do by hand or with a spreadsheet.
exercise. With respect to the methods that are presented, However, the use of dedicated software facilitates both
a similar though weaker statement must be made. There procedure and calculation of LCA. The calculations
presented below have been made with such a software
R. Heijungs ()
Institute of Environmental Sciences, Leiden University (CML),
tool, CMLCA, which is an abbreviation of Chain
The Netherlands Management by Life Cycle Assessment. The software,
e-mail: heijungs@cml.leidenuniv.nl as well as the file that contains the hypothetical data,

Jan J. Boersema and Lucas Reijnders (eds.), Principles of Environmental Sciences, 375
© Springer Science + Business Media B.V. 2009
376 R. Heijungs

can be downloaded, for educational purposes, free of


production of
charge, from the internet (www.leidenuniv.nl/cml/ssp). fuel [P7]

fuel

17A.2 Goal and Scope Definition production of


electricity [P3]

The ISO standards for LCA are quite extensive with electricity
respect to the goal and scope definition, and many of the
aspects that are mentioned, in fact duplicate many aspects production of production of
of the inventory analysis and impact assessment. This is copper [P6] glass [P5]
because carrying out the analysis is often an iterative glass
process of trial and error: initial modeling assumptions copper

are refined during the course of the analysis and it is only


production of
through a gradual process of improvement that the final bulb [P2]
results are reached. In this illustration, such a trial and
error approach will not be followed. This then reduces bulb
the content of the goal and scope definition phase to a
few aspects only. These steps relate to defining the exact use of bulb
[P1]
question for which the LCA is carried out.
In this example we choose to analyze and compare disposed bulb
two alternative methods of lighting a room: with the light

traditional incandescent bulb and with the modern incineration of


energy saving fluorescent bulb. Of course, a compari- bulb [P4]
son of one incandescent bulb and fluorescent bulb is not
very useful. We choose to compare the product alterna- Fig. 17A.1 Flow chart of processes for one of the two example
tives on the basis of the function ‘having 10 hours of products: The incandescent bulb. Note that the word ‘bulb’
should be understood here as ‘incandescent bulb’ throughout
light of an intensity of 1000 lumen’. This has conse-
quences for the power of the two product alternatives:
an incandescent bulb of 60 W and a fluorescent bulb of
15 W. Both have different characteristics with respect to Every box represents a unit process, and every unit
life span and material composition, but these aspects process needs to be represented with data. The data are
are addressed in the inventory analysis. The two alter- tabulated in the process sheets of Table 17A.1.
natives differ in more respects, like frequency of the For the fluorescent bulb the flow chart looks slightly
light emitted, warming-up time and so on. We decide in different. For simplicity, however, it has been assumed
the goal definition of this example case to ignore these that the process ‘use of incandescent bulbs’ is replaced
differences. In other cases, however, they may be felt to by ‘use of fluorescent bulbs’, and that similar changes of
be insurmountable, so that the two products are not incandescent into fluorescent are needed at other places,
regarded as functionally equivalent, and a juxtaposition but that the structure of the flow chart is identical to that
and comparison of the two is not a sensible idea. of Fig. 17A.1. The data for the unit processes that are
not yet listed in Table 17A.1 are shown in Table 17A.2.
The basic calculation that takes place in the inventory
analysis is to scale each unit process linearly with an
17A.3 Inventory Analysis –The Basic Case appropriate scaling factor so as to satisfy a commodity
balance for every product, material, service, or waste.
The main activity of the inventory analysis is collec- This is referred to as the inventory problem. Let us con-
tion and manipulation of process data. We will first do sider the commodity balances for the case of the incan-
this for the incandescent bulb. The flow chart of unit descent bulb. The external demand is ‘10 hr incandescent
processes is shown in Fig. 17A.1. bulb light’. This means that we must find scaling factors
17 An illustration of the LCA technique 377

for each unit process such that the scaled sum of outflows
Table 17A.1 Data for the unit processes in the flow chart that minus the scaled inflows of incandescent bulb light
constitutes the life cycle of the incandescent bulb
exactly yields a net outflow of 10 h incandescent bulb
Process 1: Use of incandescent bulbs
Economic inflows: 1 incandescent bulb
light. It is easily seen that the scaling factor for the unit
10,000 MJ electricity process ‘use of incandescent bulbs’ is determined by this
Economic outflows: 1 disposed incandescent bulb rule at a value of 10/5,000 = 0.002. All other scaling fac-
5,000 h incandescent bulb light tors are still undetermined. However, scaling of the use
Process 2: Production of incandescent bulbs process brings about that the inflow of incandescent bulbs
Economic inflows: 1,000 MJ electricity is set to 0.002. Now, we can construct the commodity bal-
10 kg glass ance for incandescent bulbs: the scaled sum of outflows
5 kg copper minus the scaled inflows of incandescent bulbs should
Economic outflows: 1,000 incandescent bulbs
exactly balance to 0. The scaled inflow is, as was shown
Process 3: Production of electricity
above, 0.002, and there is only one unit process (‘produc-
Economic inflows: 500 kg fuel
Economic outflows: 1,000,000 MJ electricity
tion of incandescent bulbs’) that produces this item, with
Environmental emissions: 1,000 kg CO2 to air an unscaled magnitude of 1,000. The scaling factor for
100 kg SO2 to air this unit process is therefore determined to be 0.002/1,000
Process 4: Incineration of disposed incandescent bulbs = 0.000002. In this way, we can go through the entire flow
Economic inflows: 100 disposed incandescent bulbs chart of unit processes. Table 17A.3 presents the results
Environmental emissions: 100 kg CO2 to air of this calculation for both product alternatives.
2 kg copper to soil Clearly, knowledge of scaling factors can not be the
Process 5: Production of glass aim of the inventory analysis. The knowledge one
Economic inflows: 100 MJ electricity desires is about emissions of pollutants and use of nat-
Economic outflows: 1,000 kg glass
ural resources per functional unit of every product
Environmental resources: 1,000 kg sand
alternative considered. Scaling factors are only an
Process 6: Production of copper
Economic inflows: 10,000 MJ electricity
intermediate step in obtaining these elementary flows.
Economic outflows: 100 kg copper A scaling factor affects all flows of a unit process.
Environmental resources: 1,000 kg copper ore Above, we have applied them only to the economic
Process 7: Production of fuel flows, like electricity and glass, not to the environmen-
Economic outflows: 1,000 kg fuel tal flows, like CO2 and sand. Therefore, we must apply
Environmental resources: 1,200 kg crude oil the scaling factors to the environmental coefficients of
Environmental emissions: 200 kg CO2 to air each unit process as well, and calculate the emissions
5 kg SO2 to air and resource uses of every unit process in relation to
the reference flow of 10 h incandescent bulb light.

Table 17A.2 Additional data for the unit processes in the flow Table 17A.3 Scaling factors for the two product alternatives on
chart that constitutes the life cycle of the fluorescent bulb the basis of 10 h of light
Process 8: Use of fluorescent bulbs
Incandescent Fluorescent
Economic inflows: 5,000 MJ electricity
Unit process bulb bulb
1 fluorescent bulb
Economic outflows: 1 disposed fluorescent bulb [P1] Use of incandescent bulb 0.002 0
[P2] Production of incandescent 2 E–6 0
25,000 h fluorescent bulb light
bulb
Process 9: Production of fluorescent bulbs [P3] Production of electricity 2 E–5 2 E–6
Economic inflows: 3,000 MJ electricity [P4] Incineration of disposed 2 E–5 0
20 kg glass incandescent bulbs
150 kg copper [P5] Production of glass 2 E–8 8 E–9
Economic outflows: 1,000 fluorescent bulbs [P6] Production of copper 1 E–7 6 E–7
[P7] Production of fuel 1 E–5 1 E–6
Process 10: Incineration of disposed fluorescent bulbs [P8] Use of fluorescent bulbs 0 0.0004
Economic inflows: 100 disposed fluorescent bulbs [P9] Production of fluorescent bulbs 0 4 E–7
Environmental emissions: 200 kg CO2 to air [P10] Incineration of disposed 0 4 E–6
4 kg copper to soil fluorescent bulbs
378 R. Heijungs

Table 17A.4 Inventory table, containing the emissions of pol- cow breeding yields milk, meat, manure, skins, calves,
lutants (outputs) and the extractions of resources (inputs) for the sperm, and some other co-products, cracking of crude
two product systems investigated
oil yields a whole range of fuels, and electrolysis of
Emission/resource Incandescent bulb Fluorescent bulb Unit
kitchen salt yields chlorine, caustic soda and hydrogen
Outputs
gas. When an LCA study is carried out for a PVC pack-
CO2 to air 0.024 0.003 kg
age, it means that production of chlorine is involved.
SO2 to air 0.002 0.0002 kg
Copper to soil 4 E–5 1.6 E–5 kg This unit process is, however, intricately connected to
Inputs the production of caustic soda and hydrogen gas. If we
Sand 2 E–5 8 E–6 kg compute a scaling factor for chlorine production, we
Copper ore 0.0001 0.0006 kg automatically include a certain amount of caustic soda
Crude oil 0.012 0.0012 kg
and hydrogen gas in our product system.
Thinking in terms of equations that correspond to
economic flows and unknowns that correspond to unit
Finally, it is usual to aggregate the magnitude of each processes, the occurrence of multi-functional pro-
environmental flow over all unit processes into the cesses can be understood to create more equations than
inventory table, that lists the total emissions and unknowns (Heijungs and Suh 2002). Such a set of
resource uses that are associated to the reference flow. equations is in general not solvable without additional
Table 17A.4 provides the inventory table of the incan- assumptions. The extra equations are often not com-
descent bulb and the fluorescent bulb. patible with the other ones. In terms of equations and
In this hypothetical example, fluorescent bulbs unknowns, there are several solutions to this problem:
are better on all aspects, except for the extraction of
• Add unknowns until there are as much as there are
copper ore.
equations
In practice, the use of matrix algebra facilitates the
• Delete equations until there are as much as there are
computations, especially when there are loops in the
unknowns
flow chart. Such loops correspond to recursive relations
• Try to find an approximate solution
in the system, and they are quite normal. Production of
electricity requires coal, and mining of coal requires Remarkably, current thinking in LCA concentrates on
electricity, and there are numerous examples of this the first solution. Procedures that effectively add
kind. See Heijungs and Suh (2002) for a presentation unknowns to the set of equations are known as alloca-
of the algebraic structure of the inventory analysis. tion procedures. The terminology may be somewhat
confusing, as there is no relation with the economic
mechanism with the same name, which deals with the
choice of options under scarcity of budget or other
17A.4 Inventory Analysis – Some
resources. Adding unknowns to the equations corre-
Complications sponds to adding unit processes to the system.
There are three principally different ways to add unit
So far the straightforward approach. There is often one processes to the system. The first way is to divide the
important complication. For the incandescent bulb, multi-functional process into two or more mono-
there are seven unit processes, so we need to compute functional processes, simply by examining the process
seven scaling factors. Furthermore, there are seven in more detail and disentangling the process into a
intersectoral commodities, so there are seven commod- number of constituent processes. For instance, it may
ity balances. Each balance is an equation, so we have happen that a plant produces two products with two
seven equations in seven unknowns. Under normal con- different machines. With a global view, the plant can be
ditions, such a set of equations can be solved. However, described as one multi-functional process, but closer
it may happen that the number of equations is not equal inspection reveals that there are in fact two mono-
to the number of unknowns. An important class of unit functional processes. It will be clear that this method is
processes that give rise to this complication is those of limited value. It will not solve problems related to
that provide two or more functions at the same time. truly coupled production, like the case of chlorine and
There are numerous examples of such processes: caustic soda.
17 An illustration of the LCA technique 379

The second way is known as the substitution method. points. This stresses the importance of a good procedural
The idea is that co-production of caustic soda and embedding, to prevent LCA working as a ‘hired gun’.
hydrogen gas makes that these come available to the Let us return to the example. We change the unit
market, and that less of these chemicals need to be process data of Table 17A.1 by making the electricity
produced by the traditional methods. Thus, co-producing production process (process 3) a multi-functional pro-
a certain amount of caustic soda in chlorine production cess. We assume that process heat now is collected to
effectively substitutes (or avoids) other ways of producing be delivered to users; see Table 17A.5.
this amount. It seems reasonable to subtract the fuel and Under normal conditions, the inventory can not be
material needs and the emissions of the substituted solved. We will carry out the allocation in two differ-
process from our PVC packaging system. In other words, ent ways, to demonstrate how results depend on such
we add a specification of a unit process ‘production of choices.
caustic soda’ to our flow chart of PVC packaging, and For the substitution method, we need to introduce data
we will be prepared to find negative scaling factors for for an ‘avoided’ or ‘substituted’ process. This is a process
this substituted (or avoided) unit process. In practice, it that produces heat, without co-producing electricity. We
may be difficult to find unit process specification for will assume the process data to be as in Table 17A.6.
substituted processes. After all, these are activities that Notice that we now have a process producing a flow
have not taken place, and providing data for things that ‘co-heat’, and a process producing a flow ‘heat’. The
are not there is of course cumbersome, both technically distinction in name has been made to emphasize the
and philosophically speaking. fact that there are two ways of production, hence there
The third way may be called the partitioning method. are two ‘brands’ so to say. It is like tomatoes cultivated
The idea is that we can assign a ‘fair’ share of fuel and with traditional and with organic culture. Although the
material needs and emissions to the different functional tomatoes themselves may be indistinguishable, we
outputs of a multi-functional process. For instance, 50% must give them a distinct feature, and hence a different
to chlorine, 40% to caustic soda, and 10% to hydrogen name. A crucial step in the substitution method is to
gas. Or, more sophisticated, 50% of the electricity needs explicitly announce that the heat that is produced in
to be assigned to chlorine, but the entire 100% of the process 11 substitutes the heat co-produced in process
leaked chlorine. If we analytically decouple a multi- 3'. Hence, we establish the (possibly debatable) equiv-
functional process into a number of mono-functional alency between the flows heat and co-heat. Now, there
processes, and if we are allowed to choose independent are eight processes and eight flows, and the system of
scaling factors for these originally coupled processes, we equations can be solved. We find new scaling factors
will have introduced more scaling factors, some of which and a new inventory table (see Table 17A.8 below).
(those for caustic soda and hydrogen gas production) will
be found to be zero. The large problem with the partitioning
Table 17A.5 Additional data for the unit processes in the flow
method is, of course, that the numerical values of the chart that constitutes the life cycle of the fluorescent bulb
so-called allocation factors is difficult to decide upon. Process 3': production of electricity and co-heat
Main movements are towards reflecting physico-chemical Economic inflows: 500 kg fuel
causalities (based, for instance, on thermodynamics) and Economic outflows: 1,000,000 MJ electricity
towards reflecting socio-economic causalities (based, for 1,000,000 MJ co-heat
instance, on market prices). Environmental emissions: 1,000 kg CO2 to air
100 kg SO2 to air
We will not address the other possibilities for finding
an answer to the inventory problem. However, it is clear
that, whether we choose for allocation or for one of the
Table 17A.6 Data for the unit process of heat production,
other solutions, whether we choose within allocation for which is assumed to be an ‘avoided’ process when the multi-
substitution or for partitioning, and how we choose within functional process 3' is allocated with the substitution method
these approaches for one particular mode of obtaining the Process 11: Production of heat
required data, the inventory analysis of an LCA is not Economic inflows: 0.2 kg fuel
based on purely mechanical manipulations with purely Economic outflows: 1,000 MJ heat
Environmental emissions: 0.3 kg CO2 to air
empirical data. Value judgments enter the analysis, most
0.04 kg SO2 to air
clearly at the allocation step, but also at several other
380 R. Heijungs

For the partitioning method, we proceed quite dif- Table 17A.9 Inventory table for the same modified system as
ferently. Process 3' is separated into two independently in Table 17A.8, but now allocated with the partitioning method
variable processes: one (process 3a) which produces Emission/Resource Incandescent bulb Fluorescent bulb Unit
electricity without heat, and one (process 3b) which Outputs
CO2 to air 0.02 0.0026 kg
produces heat without electricity. For some reason (for
SO2 to air 0.0016 0.00016 kg
instance, based on energetic value), we decide to allo- Copper to soil 4 E–5 1.6 E–5 kg
cate 80% of the burdens to electricity and 20% to heat. Inputs
Such a partitioning key may be based on economic Sand 2 E–5 8 E–6 kg
values, on energy content, or on a different physical or Copper ore 0.0001 0.0006 kg
socio-economic parameter. Anyhow, given these allo- Crude oil 0.0096 0.00096 kg
cation factors 0.8 and 0.2, this leads to the allocated
process specifications as in Table 17A.7.
Again, we have eight flows and eight processes, and we have seen, to the choice of the allocation method
the system of equations can be solved. and to the choice of the ‘avoided’ process or allocation
We may observe that the results are different. The factor, but they apply at many more places, like setting
results with the substitution method are as in Table of system boundaries, degree of region- and time-
17A.8, whereas Table 17A.9 shows the results with the specificity of the processes, assumptions with respect
partitioning method. to consumer behaviour, and so on. It is impossible to
We will not discuss the advantages and disadvantages illustrate all these complications here, but we will bring
of the several approaches to allocation. There exists an forward one topic, related to consumer behaviour.
extensive literature on this. For here, it is important to The functional unit was set at 10 h of light, with
recognize the fact that the inventory analysis is not a two different kinds of bulb. It can be argued that
mere mechanical processing of data, but that difficult consumers will be less precise in switching off lights
and debatable choices must be made. These apply, as when they leave a room if they use economical, or
‘green’ for that sake, bulbs, like incandescent bulbs.
Table 17A.7 Results of the application of the partitioning One could carry out a survey, or just speculate, on the
method with allocation factors of 0.8 and 0.2 to the data for the degree to which this behavioural difference exists.
multi-functional process 3'. This then could lead to an adjustment in the basis for
Process 3'a: Production of electricity comparison, for instance comparing 10 h of light with
Economic inflows: 0.8 × 500 kg fuel an incandescent bulb against 12 h of light with a
Economic outflows: 0.8 × 1,000,000 MJ electricity
fluorescent bulb. One could, on the other hand, also
Environmental emissions: 0.8 × 1,000 kg CO2 to air
0.8 × 100 kg SO2 to air
argue that such a behavioural difference should not be
Process 3'b: Production of co-heat
taken for granted, that consumers should be informed
Economic inflows: 0.2 × 500 kg fuel on the adverse effects of lighting fluorescent bulbs
Economic outflows: 0.2 × 1,000,000 MJ co-heat when no one is in the room, and that a proper
Environmental emissions: 0.2 × 1,000 kg CO2 to air comparison should be based on how we expect that
0.2 × 100 kg SO2 to air people use their bulbs in future.
Many similar questions can be posed. There is not
Table 17A.8 Inventory table for the system, modified with a one single answer to such questions, and it should be
co-producing process, and allocated with the substitution clear that a refined reflection on the exact goal and
method
scope of the LCA may be needed. This may well
Emission/resource Incandescent bulb Fluorescent bulb Unit
involve the consultation of experts and stakeholders
Outputs
from different disciplines and pressure groups. It should
CO2 to air 0.017 0.0023 kg
SO2 to air 0.0012 0.00012 kg make clear that LCA is not the objective process with
Copper to soil 4 E–5 1.6 E–5 kg reproducible results that was strived for some time ago.
Inputs It is generally recognized by now that LCA is a tool for
Sand 2 E–5 8 E–6 kg decision-support, and that this, amongst others, implies
Copper ore 0.0001 0.0006 kg that a proper in-context-placing of the LCA process
Crude oil 0.007 0.0007 kg and its results is needed.
17 An illustration of the LCA technique 381

17A.5 Impact Assessment – converting the LCI results to so-called indicator results.
Definition of Impact Categories, For ecotoxicity for the incandescent bulb, for instance,
we have contributions by SO2 and by copper: (20 ×
Selection of Characterization Models,
0.002) + (10,000 × 4E – 5) = 0.04 + 0.4 = 0.44 m2 eco-
Classification and Characterization system (see Table 17A.12).
We see that characterization results in a table that is
The inventory tables that were discussed above illus- shorter (in a real-world LCA study, several hundreds
trate what life cycle inventory analysis is about. But of inventory results are condensed into a dozen of
they do not show the complexities that are involved in indicator results) and that is easier to understand.
a typical case in real life. Then, the number of unit However, we still cannot formulate an unambiguous
processes may grow to 500, with a more or less equal preference, as fluorescent bulbs are much better on
number of economic flows. Although this makes the ecotoxicity and global warming, but somewhat worse
analysis more difficult and time consuming, the use of
dedicated LCA software will have no problems with
systems of this size. The number of environmental Table 17A.10 Hypothetical characterization factors for three
flows is a different subject, however. It can rise up to impact categories, derived from three hypothetical characterization
models
1,000. From a computational point of view, this is not
Category 1: Resource scarcity
especially problematic, but the interpretation of the Environmental resources: 0.1 RDU/kg sand
results becomes unfeasible. How is one to make sense 10,000 RDU/kg copper ore
of a table with 1,000 different items, ranging from 250 RDU/kg crude oil
dichlorobenzene to sulfur dioxide and from bauxite to Category 2: Ecotoxicity
whales? It is most likely that one product alternative is Environmental emissions: 20 m2 ecosystem/kg SO2 to air
better for one emission while the other is better for the 10,000 m2 ecosystem/kg copper to soil
other one. The impact assessment condenses for each Category 3: Global warming
product alternative the long inventory table into a small Environmental resources: 0.5 kg CO2-equivalents/kg SO2 to air
number of impact category indicators, or even into a Environmental emissions: 1 kg CO2-equivalents/kg CO2 to air
single weighted index.
The first steps of impact assessment deal with the
Table 17A.11 Classification table, containing the inventory
definition of impact categories and the selection of
items assigned to the selected impact categories
characterization models. In this example, we will define
Impact category Incandescent bulb Fluorescent bulb Unit
three hypothetical impact categories: resource scarcity,
Resource scarcity
ecotoxicity and global warming, and we will associate Sand 2 E–5 8 E–6 Kg
these with three hypothetical characterization models. Copper ore 0.0001 0.0006 kg
These express the impacts of a number of interventions Crude oil 0.012 0.0012 kg
in terms of resource depletion units (RDU), square Ecotoxicity
meters of ecosystem affected, and kg of CO2-equivalents SO2 to air 0.002 0.0002 kg
with respect to global warming. Characterization Copper to soil 4 E–5 1.6 E–5 kg
Global warming
factors are derived from these models, and their
CO2 to air 0.024 0.003 kg
hypothetical values are listed in Table 17A.10. SO2 to air 0.002 0.0002 kg
Classification is a step that sometimes is misunder-
stood: it is the assignment of inventory results to the
selected impact categories. Notice that no calculations Table 17A.12 Characterization table, containing the converted
are performed, and that the characterization factors and aggregated inventory items assigned to the selected impact
categories
provided by the characterization models are only used
Incandescent Fluorescent
in determining if a certain item is assigned to a certain Impact category bulb bulb Unit
category or not. Table 17A.11 gives the table that result
Resource 4. 6.3 RDU
after classification of Table 17A.4. scarcity
The next step is characterization. This involves the Ecotoxicity 0.44 0.16 m2 ecosystem
use of the characterization factors for numerically Global warming 0.024 0.003 kg CO2-equivalent
382 R. Heijungs

on resource scarcity. A final judgement depends, Table 17A.15 Hypothetical weighting data, used in weighting
amongst others, on the societal importance of resource of normalized indicator results
scarcity in relation to the other impact categories. And Impact category Value Unit
this also depends on the meaning of the numbers. Resource scarcity 5 –
Ecotoxicity 35 –
Global warming 60 –

17A.6 Impact Assessment –


Table 17A.16 Weighting table, containing the converted and
Normalization and Weighting aggregated normalized indicator results
Impact category Incandescent bulb Fluorescent bulb Unit
Normalization aims to add still more meaning to the Weighted index 3.4 E–9 1.5 E–9 year
indicator results. This can be done by expressing them
relative to a reference value. A usual choice for this refer-
ence value is the annual global size of the impact catego- Application of these weights to the normalized
ries. Given an inventory of annual world-wide emissions results yields Table 17A.16.
and resource extractions, the tables with characterization Now, the first ‘unambiguous’ judgement appears:
factors can be applied to compute the indicator results the fluorescent bulb beats the incandescent bulb by a
for the world over 1 year. This provides a reference value factor of more than 2.
for each impact category. See Table 17A.13.
The normalization data can now be used, with Table
17A.14 as result.
Observe the peculiar unit ‘year’: the normalized 17A.7 Interpretation
indicator results all have the same unit, which expresses
the period of time that the functional unit of product What has been presented above is of course a caricature.
can be held responsible for the world problems. Of The final result (3.4 against 1.5) suggests that fluorescent
course, this time period is very small for 10 h of light. bulbs are better, but we know that this ignores
Also observe that the decision problem has not been uncertainties, data gaps, assumptions, value judgements,
solved yet: the fluorescent bulb is still better in two and many other sources of error. A systematic analysis
respects and worse in one respect. But the removal of of the robustness of the results is an essential ingredient
arbitrary choices of unit facilitates making the impact of a careful decision-making procedure. In LCA, the
categories commensurable. Suppose that somehow, interpretation phase is meant to provide an explicit place
e.g., by shadow prices or questionnaires, it has been for such analyses. Unfortunately, operational methods
found that the inter-category weighting factors are as and clear rules are not yet standardized. In this example,
in Table 17A.15. we will show a small number of examples.
One example has already been shown. This was in
Table 17A.13 Hypothetical normalization data, based on Section 17A.4, where two different allocation methods
world for 1 year, used as a reference value in normalization were discussed and their results were compared. In general,
Impact category Value Unit one should make different choices for at least:
Resource scarcity 1 E11 RDU/year • Product performance characteristics
Ecotoxicity 5 E9 m2 ecosystem/year
• System boundaries
Global warming 1 E10 Kg CO2-Equivalent/year
• Process data
• Allocation methods and factors
Table 17A.14 Characterization table, containing the con- • Choices of impact categories
verted and aggregated inventory items assigned to the selected
impact categories
• Characterization factors
• Normalization data
Impact category Incandescent bulb Fluorescent bulb Unit
• Weighting principles and data
Resource scarcity 4.0 E–11 6.3 E–11 year
Ecotoxicity 8.8 E–11 3.3 E–11 year All types of results will be obtained. Monte Carlo techniques
Global warming 2.4 E–12 3.0 E–13 year may be helpful here, but they cannot be used for all types
17 An illustration of the LCA technique 383

of uncertainties. In the end, one will develop an idea of the for counterintuitive results. They also help one to focus
bandwidth of the results, and consequently of the robustness on key issues for process and product improvement.
of a judgement. For instance, when all reasonable choices Thus one can step from using LCA for static assessment
lead to 3.4 with a standard deviation of 1 for the incandescent to using LCA for adaptive analysis.
bulb and 1.5 with a standard deviation of 1 for the fluorescent
bulb, one may still say that the latter beats the former. But
when the standard deviations are much larger, and the
confidence intervals largely overlap, decision-making will 17A.8 Concluding Remarks
become problematic.
Apart from the parametric or probabilistic variation In this chapter, the working details of LCA have been
of data and the variation of assumptions and choices, described. We sometimes touched upon issues relating
there are interpretation-related issues that reside more to the possibilities and limitations of LCA. However,
at the procedural level. Are the product alternatives we did not explore that area. Chapter 12 discusses such
really comparable, also from the perspective of con- issues at length.
sumers? Have the right stakeholders been involved in Although we focused on the computational aspects
the correct way? These questions may have a large of LCA, we still were able to only discuss the most
impact on the usefulness and credibility of the LCA important aspects in a not too rigid way. A full discus-
study, but they are not discussed in this more model- sion, based on matrix algebra and including references
ing-oriented example. to other material, is in Heijungs and Suh (2002).
Another useful approach to interpretation is the
contribution analysis, in which results are decomposed
into contributing processes, flows or categories. One
may, for instance, find out that 90% of the weighted Reference
index for the incandescent bulb is due to ecotoxicity,
and that this in turn is largely determined by the emission Heijungs, R., & Suh, S. (2002). The computational structure of
of copper. Such results are useful for providing a check life cycle assessment. Dordrecht: Kluwer.

You might also like