LCA Methodology Modular Approach

A New Approach for a Modular Valuation of LCAs

Andreas Ciroth 1., Giinter Fleischer 2, Karin Gerner z and Heiko Kunst 2
i GreenDeltaTC Tools & Consulting, Raumerstr. 7, D-10437 Berlin, Germany
2 TU Berlin, Institute of Environmental Engineering, CR2, Str. des 17. Juni 135, D-10623 Berlin, Germany

* Corresponding author (ciroth@greendeltatc.com)

comparison of alternatives is most interesting, a comparison

DOh http:l/dx.doi.orq/10.1065/Ica2003.08.132 yielding a ranking of the alternatives compared. In LCAs, this
ranking is often difficult to obtain. If an impact assessment on
Abstract
midpoint level (Udo de Haes et al. 1999) has been performed,
Goal Scope and Background. Qualitative valuation methods frequently some indicator results favour one alternative, and
carefully try to avoid an aggregation across impact categories. some the other. For example, product A has less climate change
However, such an aggregation often helps in obtaining a clear potential, while product B has a lower ozone depletion poten-
result for the valuation (which product scores better?). This ar- tial: H o w does the ranking look like, then?
ticle presents a new valuation method that uses an iterative ap-
proach. The application is demonstrated by the help of a case Basically, the aim of a valuation may be seen in that it pro-
study for electric motors in trains. vides and compares the values of the objects considered.
Methods / Main Features. The approach combines two exist- Commonly, a method pursuing this aim needs subjective el-
ing, unique valuation methods described earlier in literature, ements to be introduced in its procedure for assigning these
which both are of a rather non-aggregating nature, in line with values (Finnveden 1997). These subjective elements give rea-
ISO requirements, and were designed to be performed by LCA son to suspect whether the scientific validity of the result is
experts. The method is implemented in a computer software. violated, and this 'suspicion' may be addressed as a com-
Besides constants used within the method, the software needs mon problem of valuation methods within LCA.
as input solely indicator values from the Impact Assessment.
Results and Discussion. The iterative nature of these methods From the different valuation methods available (Ecosite
itself, and especially the combination of these methods, helps in 2002, Bengtsson 2000, Hildenbrand 1999), it can be ob-
achieving a valuation result for the LCA with not more subjective served that methods differ in where, at which place, and to
and aggregating elements than necessary. Subjective elements are what extent they make use of subjective elements, and also
clearly separated from others. The algorithm seems highly sensi- in the result they give. Some approaches provide not only
tive to changes in impact categories regarded as important ones. rankings (as ordinal results), but quantitative results. On
The implementation in software greatly eases the application of
the one hand, these methods further assist in decision mak-
the method by transferring routine work from LCA experts to a
machine. It ensures a reproducible result and prevents erroneous ing by transforming complex results into easy to understand
steps in a rather complicated valuation procedure. It further helps statements. But on the other hand, these results will be more
in hiding the complexity of the method from the user. assailable than rankings, because of the additionally used
Conclusion. The approach of combining valuation methods in subjective elements. For an advanced discussion of advan-
LCAs seems a fruitful one, and shows benefits when implemented tages and disadvantages see (Udo de Haes et al. 1999, Finn-
in computer software, in terms of usability, and in terms of a veden 2000).
more reproducible application. Care has to be taken to make
sure users know what they do when performing an automated The international standards, especially ISO 14040 and ISO
valuation procedure. 14042, name two different types of valuation methods:
Outlook. We see three ways for extending the approach, namely: grouping and weighting. Both grouping (sorting and rank-
(i) become part of a toolbox of different valuation procedures; ing of impact categories according to their environmental
(ii) explicitly cope with uncertainty, and (iii) include different importance) and weighting (using numerical factors for con-
values for normalisation, in different regions worldwide. The verting indicator results of different impact categories lead-
software will be made available also in a stand alone version. ing to a single score) (ISO 2000) are based on value choices
including subjective elements (Udo de Haes et al. 1999).
Keywords: Grouping; modular approach; ranking; semi quan- According to ISO, grouping and weighting are optional ele-
titative valuation; subjectivity interface; tool; toolbox ments of the life cycle impact assessment (LCIA), and their
use is restricted to internal LCAs. The most important re-
1 Introduction: Aims and Problems of a Valuation in Life quirements on grouping and weighting mentioned in ISO
Cycle Assessments 14040 and ISO 14042 are transparency of procedures and
(intermediate) results, and consistency of the procedures with
LCA as an ecodesign tool is often used to support decisions
goal and scope of the study (ISO 2000).
on products and services. To be able to reduce the environ-
mental impacts related to a product (for example by switch- The procedure of performing grouping or weighting is not
ing to another product, or by changing the product itself), a specified in the international standards. A standardisation

273
Modular Approach LCA Methodology

would make sense since recent works showed a consider- UBP F

able influence of the valuation procedures on the overall OF--- c (Brand et al. 1998) (1)
F, F,
outcome of an LCA, due to the valuation procedure itself,
with: s eco-factor
but also due to the subjective elements introduced in a spe-
UBP: Environmental burden score
cific valuation procedure (Hildenbrand 1999, Ciroth 2002). F: current flow per year
In this article, after an overview of valuation methods, a Fk: critical flow per year
new method that uses an iterative approach is presented and C: constant (10Wa for better handling of results)
discussed from a methodological point in chapter 2. In chap-
ter 3, the method is demonstrated in a case study, yielding Eco-points representing the quantified environmental burden of an
emission can be calculated via:
new discussion points from the practical application.
EP = OF.m~ (2)
1.1 Overview of valuation methods
with: EP: eco-point
Since 1990, several valuation methods for LCA have been
mi: emission of substance i (mass) related to the func-
developed and used for a large number of case studies tional unit
(Hildenbrand 1999). For a comprehensive description and
valuation of different methods see (Bengtsson 2000). As the final step, the eco-points are aggregated over all sub-
stances. Other distance-to-target approaches are discussed
The methods may be structured into different sets (Table 1).
in detail in (Seppiilfi and Hiim/il/iinen 2001).
The main difference is the type of result the methods pro-
vide: a single score (quantitative method) or an ordinal rank-
ing (semi-quantitative or qualitative method). 1.1.2 Eco-indicator 99

The methods mentioned as examples in Table 1 will be dis- The Eco-indicator 99 has been developed by an interna-
cussed very briefly in the following. tional group of LCA and environmental experts mainly from
the Netherlands and Switzerland as a 'damage oriented'
impact assessment method (PRE 2000), based on the Eco-
Table 1: Overview of ranking methods Indicator 95 impact assessment methodology (Goedkoop
Ranldng ~ t ~ : 1995). The aim was to develop a method to express the
Distance-to-target single score Ecopoints of BUWAL total environmental burden of a product in a single score
methods (Brand et al. 1998) (Goedkoop et al. 1998).
Endpoint-oriented one or several Eco-indicator 99 The damage function represents the relation between the im-
impact assessment scores (Goedkoop et al. 1998) pact and the damage to human health or to the ecosystem as
methods
impact assessment end-points. The main idea is to combine
Quantitative weighting single score Panel methods (Bengtsson different impact categories into three areas of damage:
2000, Giegrich et al. 1995)
9 Damage to mineral and fossil resources
Semi-quantitative ordinal ranking UBA method (Giegrich
9 Damage to ecosystem quality
weighting et al. 1995)
9 Damage to human health
Hierarchical ordinal ranking CAU method (Volkwein,
approaches Gihr and KI6pffer 1996). In order to get a single impact factor for each emission or
resource, a weighting across the three areas of damage is
performed.
1.1.1 Method of environmental scarcity / Ecopoints of BUWAL The Eco-indicator 99 concept provides three sets of weight-
The method of environmental scarcity has been developed ing factors according to assumptions concerning the time
at the Swiss federal environmental agency BUWAL in 1990 frame and other model choices. One set based on modest
assumptions is used as a default by the authors of the Eco-
(Brand et al. 1998) as one of the first valuation approaches.
indicator 99 ('Hierarchist' version).
It allows a comparative weighting and aggregation of vari-
ous environmental interventions by use of 'eco-factors' based The so-called weighted damage factors are substance spe-
on life cycle inventory results. No impact assessment ac- cific and can be used analogous to the impact assessment
cording to ISO 14042 is included, but the most important factors in midpoint impact assessment:
impact categories are covered by this method indirectly.
Basically, the method uses a distance-to-target approach by EI99 = ~., m~ 9 E 1 9 9 ~ (3)
i
comparing the actual pollution (current flows) in the inven-
tory with critical flows (critical loads) which are deduced with: EI99: Eco-indicator 99 score
from the scientifically supported goals of the Swiss environ- EI99i: weighted damage factor of substance i
ment policy (Brand et al. 1998).
It is worth noting that the subsequent steps of the E1 99
For emissions and energy sources weighting factors ('eco- method entail an increasing subjectivity, which thus rises
factors') are calculated for different paths into the environ- from the characterisation, to the damage modelling, and,
ment (air, surface water, ground water, soil) as follows: finally, to the weighting and normalisation.

274 Int J LCA 8 (5) 2003

LCA Methodology Modular Approach

1.1.3 Panel methods Impactassessment Valuationcdteriafor impact

results categories
In general, panel methods are based on judgements of ex-
perts or stakeholders. Individual ratings of the panel mem-
bers are compiled to a common assertion. Quantitative i I ii

weighting methods based on panel methods use a value ben- Valuation of impact
efit analysis approach: each impact category receives a categories
'weight'. After performing an impact assessment with indi- /
cator results for all impact categories, the impact categories
are evaluated using individual weighting factors (e.g. values
between 1 and 10). This evaluation of impact categories is, Overall importance of
of course, mainly based on subjective values and is performed impact categories ~
by an expert panel using questionnaires. These values can
be based on legislation, market research, or monetary valu-
ation (Schmidt and Sullivan 2002).
Ranking
An aggregation of all impact indicators can be performed
after their weighting (multiplying the indicator results and Fig. 1 : Overview of the UBA method
the weighting factor).
settings' as described in (Giegrich et al. 1995), climate change
1.1.4 UBA method is rated to be of very high importance, and resource deple-
The UBA method (UBA: German Federal Environmental tion to be of high importance. This valuation is permanent,
Agency) provides an ordinal ranking as result (Fig. 1). Analo- and does not depend on case study results.
gous to panel methods and quantitative weighting methods, On the other side, the impact indicator results per category
a (midpoint) impact assessment and a valuation of the im- are normalised, and then also converted into a five-stage
pact categories are performed separately (see Fig. 2). The scale using the highest value of all normalised impact indi-
main difference to panel and weighting methods is the semi- cators as reference. The result is called 'z value' and expresses
quantitative merging of both impact category indicator re- the quantitative importance of each impact category for each
sults and of the valuation result of each category in order to alternative compared within the valuation.
get an ordinal ranking.
Both elements, the importance of the impact categories and
The main ideas of the UBA method were presented in 1995 the normalised impact indicator results, are merged to an
by (Giegrich et al. 1995). The German case study on drinks overall importance of each impact category by using a sym-
packaging system developed and used the method for the metrical matrix (Example: 'high' as category importance and
first time (Schmitz, Oels and Tiedemann 1996). A slightly 'very high' as quantitative importance gives 'very high' as
modified version has been used in the update of the case overall importance).
study on drinks packaging systems (Plinke et al. 2000).
Finally, in the last and most important step, the impact cat-
To balance the environmental burdens related to two alter- egories can be directly compared with each other. The aim is
natives to be compared, the ecological importance of the to find an alternative that scores better in 'overall more im-
different impact categories has to be estimated. For this valu- portant' categories, and to eliminate categories where this al-
ation the following - not independent and not measurable ternative scores worse. In this process, categories of the same
(Hildenbrand 1999) - criteria are used: overall importance, but with different 'winners' (i.e. with dif-
9 Hazard potential: intensity of impacts for the safeguard objects ferent alternatives that score better in this category), can be
concerned eliminated against each other in order to get a clear ranking.
9 Temporal aspects: temporal range of the effects, reversibility or
irreversibility
9 Spatial coverage of the effects (from local to global) 1.1.5 CAU method
9 Preferences of the citizens (according to social science studies) This valuation method has been developed within a Ger-
9 Relation of current and previous environmental damage to envi-
ronmental quality objectives (qualitative distance-to-target-ap-
man research project (Fleischer et al. 2000). For a detailed
proach). description see (Volkwein, Gihr and K16pffer 1996).

For each impact category, these five criteria are used to re- As the main idea, the prioritisation of impact categories in-
ceive a semi-quantitative valuation of each impact category cluding normaiised indicator values is used. Similar to the UBA
using a five-stage scale (low importance to very high impor- method, impact categories are valuated using valuation crite-
tance}. This valuation is partly based on scientific findings. ria, but without providing a single value for each category.
Subjective elements cannot be avoided in the choice of crite- As the first step, the impact assessment results are normal-
ria, in the valuation of impact categories using the criteria, ised. The method is designed for performing a valuation of
and also in the overall valuation of impact categories. To a pair, i.e. of two alternatives. For these two alternatives the
give an example for the results obtained, with the 'default difference for each impact indicator is calculated.

lnt J LCA 8 (5) 2003 275

Modular Approach LCA Methodology

Within the next step the impact categories are valuated us- 14042 in order to evaluate the environmental impact of com-
ing five-stage scales. This valuation is usually the result of a ponents for railway applications. The decision makers and
panel discussion. The following criteria are used in the valu- experts from the railway industry needed clear, concrete and
ation (Volkwein, Gihr and K16pffer 1996), which correspond easy to communicate results such as 'alternative A is better
to three of the five criteria used for the UBA method: than B'. The results from the environmental assessment
9 time criterion: The regeneration time as the time needed for resto- would then be comparable to the result received from the
ration of a damage is used to express the time criterion Life Cycle Costing performed simultaneously in the project.
9 hazard criterion: regards the danger potential of impacts Before starting to develop a new method, a great variety of
9 space criterion: refers to the area related to the expected impact valuation methods commonly used in LCA were compared.
Applying the three criteria together with the normalised in- As a result, especially regarding the variety and number of
dicator results yields a 'hierarchy level' (from 1 to 5) for existing valuation methods, it seemed neither necessary nor
each impact category. Based on this hierarchy, a hypothesis efficient to develop a completely new method, thus "re-in-
is stated of the loser of both alternatives compared: The as- venting the wheel a second time" (Cliff 2001).
sumed loser is the alternative with the higher environmental All methods compared had their specific advantages and
burdens related to the most important, top hierarchy im- disadvantages. The most important result from the compari-
pact category (for an example see chapter 3). son was that a combination of different valuation methods
The following steps are used to verify the hypothesis: will diminish some of their individual disadvantages.
1. A pair wise comparison of impact category indicator re- An example: Comparing the CAU and UBA method dis-
sults from the assumed winner and loser, based on the played that both methods introduce subjective, evaluating
categories' hierarchy. Similar to the UBA method, the elements by ranking the environmental importance for the
aim is to eliminate categories where the assumed loser different impact categories. The CAU method distinguishes
did not lose. the three criteria temporal and local coverage and risk of
2. A similar comparison, with the difference that more than the potential impact described for each impact category.
one category where the loser did not lose can be com- Evaluating these three criteria separately, a subjective weight-
pared to a category where he did lose, provided that the ing of the importance of temporal versus geographical cov-
latter category is high enough in the hierarchy. Indicator erage or versus potential risk is avoided. But due to this
values of several categories are in fact aggregated. cautious behaviour, the CAU method frequently does not
Similar to the UBA method, the goal is to eliminate all cat- reach a clear ranking result.
egories where the assumed loser did not lose. If this is suc- On the other hand, by using the UBA method, a definite result
cessful, and there remain categories where the assumed loser is obtained more frequently, but also more subjectivity is in-
has lost (or the assumed winner has won, accordingly), the troduced in the valuation procedure, since this valuation
hypothesis is confirmed. Fig. 2 illustrates the process of elimi- method evaluates environmental importance by one single
nating categories against each other. indicator, and for obtaining this indicator, merges three crite-
ria (see chapter 2), assuming implicitly a similar importance
for each. This is of course a subjective consideration: why
should a high risk for a small group of individuals have the
same value than a lower risk for a larger group of individuals?
The central idea of the valuation approach proposed in this
paper is to get a definite ranking result (A better than B) by
using a valuation procedure which is as transparent as pos-
sible and which, in addition, fulfils the requirements for a
valuation method stated by ISO (see chapter 1). Therefore
two valuation methods which fulfil ISO requirements are
combined within an iterative approach, going from qualita-
tive, semi qualitative up to completely quantitative methods
until the definite result is reached. In other words, the level
Fig. 2: Elimination of impact categories (example) of subjectivity is increased until a clear valuation result is
achieved. A combiation of both methods is the CAU method
and the UBA method.
2 Proposal for a New Approach to Valuation
Both methods have been used in practice: For example, the
2.1 Background
CAU method as described in (Volkwein 1996) was applied
The new approach to valuation presented in the following manually for some examples within the euroMat project
originates in a research project called Bahnkreis (see TUB (Fleischer et al. 2000), where it showed its capability of pro-
2002, Siemens 2000 and Trebst 2001). Within this project, viding cautious and also clear results. The UBA method has
a midpoint assessment was performed according to ISO been used for nearly 5 years (Schmitz 1996).

276 Int J LCA 8 (5) 2003

LCA Methodology Modular Approach

The iterative approach described here is implemented in the tative importance than another, this category has a higher
software trainEE (GreenDeltaTC 2002), a software pack- level of hierarchy. Now the probable winner of both alter-
age especially developed for assessing the life cycle of com- natives is determined. It is the alternative for which a lower
plex products with a long and potentially complex life cy- impact is calculated for the impact category which has the
cle, such as trains and their components. Within trainEE, highest level of hierarchy. In step V4 this hypothesis is tested:
the method can be performed automatically 1. A previous Impact categories which favour the probable loser are elimi-
version has been used within the research project 'Bahnkreis' nated by impact categories which obtained a higher hierar-
(Siemens 2000), where applicability has been proven. chical level in V3 and favour the probable winner. This com-
parison is done in two attempts: First, in V4.1 only one to
one elimination are allowed: one high level impact category
2.2 Method description favouring the winner can only eliminate one category fa-
Fig. 3 resumes the method and describes the most impor- vouring the probable loser. If all impact categories which
tant calculation steps. It is presented in a rather condensed favour the loser can be eliminated in V4.1, the hypothesis is
way, for further details see (Volkwein 1996, Giegrich 1995). confirmed and the ranking is a level II result. In a second,
semi-quantitative attempt, in V4.2 a high level impact cat-
egory of the probable winner may eliminate two or more
2.2.1 The semi-quantitative CAU method (Block V) impact categories when its z criterion is higher than the sum
In case all categories favour the same alternative, no valua- of the values for the z criterion for the lower level impact
tion is necessary in order to obtain a clear result 'alternative categories favouring the probable loser. If all categories fa-
A is better than B', and in case both alternatives have the vouring the probable loser can be eliminated, a level III re-
same values in all categories, also a ranking is neither neces- sult is obtained. For more details see (Volkwein 1996).
sary nor possible. Both cases are checked in the first steps of
the method, V0 and V1. In the following, the LCIA results In case even after V4.2 no clear ranking of the alternatives is
are normalised by the global contributions to each impact possible, the more rigid UBA valuation method is used.
category. After that, the valuation itself starts: in the first
valuation step (V1, see Fig. 3) for each impact category a 2.2.2 The UBA method (Block G)
rating of importance (from 1 to 5) of impact is done, con-
sidering (i) temporal and (ii) local implications of the im- Within this method, instead of the semi quantitative valua-
pact and (iii) the severity of the hazard considered. A fourth tion matrix an 'interval judgement' is established, using only
criterion is the z value, given by the absolute difference of two values for the comparison of the alternatives A and B:
the normatised impact values of both alternatives for an the ecological importance of a given ecological impact cat-
impact category. It has no dimensions. These four values for egory expressed in only one value and its quantitative im-
each alternative constitute the semi quantitative valuation portance value (analogue to the z value calculated in V1 of
matrix, which is the starting point for the following com- the CAU method). The quantitative importance required by
parison of two different alternatives A and B. For the imple- the UBA method can be described by the normalised z-Value
mentation in the software tool, default values for the fac- taken from the CAU method using a five level scale. This is
tors i, ii and iii for these rating factors were developed, but done in step G1. The valuation criterion for ecological im-
the user is free to modify these values when be or she con- portance, required by the UBA method, is derived from quan-
siders it necessary. Then the winner for each impact cat- titative weighting factors for location, time and hazard used
egory (i.e. the alternative with contributes less than this spe- in the CAU method in step G2. In doing so, the original
cific impact category) is stated. Afterwards, a first com- UBA method is slightly altered (since the factors from the
parison of both alternatives is done in V2. When one alter- original UBA method are not used), for sake of consistency
native is the winner in all categories, a definite result is ob- within the whole valuation procedure. The factors are ag-
tained and the valuation is finished. This result is called a gregated, simply by calculating the arithmetical mean.
'Level I result'. If no clear decision can be reached at this Then the overall importance of each impact category (ag-
point, levels of hierarchy are assigned to each impact cat- gregating quantitative and environmental importance) is
egory in step V3. Therefore, not only the ranking of impor- assigned by a valuation matrix in G3. Finally, in G4, a com-
tance of the impact category is taken into account but also parison of alternatives A and B is done. Impact categories
the quantitative criterion z. If one category reaches higher with higher overall importance which favour one alterna-
values for the three criteria of importance and for the quanti- tive are used for eliminating impact categories with lower
overall importance which favour the other alternative. If all
impact categories favouring one alternative can be elimi-
nated, this alternative is the loser. A result obtained in G4
has the level IV.
The result within the original UBA method is text describing the result,
with all its underlying assumptions. So it does not provide a result easy When, even after this procedure, no clear ranking is possi-
to communicate, as required before. The software adapted the way to
present the results from the CAU method, reporting only the ranking and ble, both alternatives are considered to have equal environ-
the categories which where eliminated against each other. mental importance. Fig. 3 shows the iterative approach.

Int J LCA 8 (5) 2003 277

Modular A p p r o a c h LCA M e t h o d o l o g y

LCIAresults StartUBA
Startt. method

i 11 aggregation /
rules
_KNov=o.,o~) $
Yes necessary

NO / normalisatior* /
NormaJization • ~r,ac~ic.a~O~!
,;mes~
StartCaumethod
weightingfactors / I ic
x. .... ~t,me,Xh.~y'

I
No
I
No
,Lev,vi j

No
eSu,:Aanda
Cecaogicallyequal//
Yes~/~ ResultA: betterB"~
-\ (LevelII) J

No

J..u,,:...e,B
Ye. . \ {.evel.,~ j

No

V0 Are the alternatives A and B different in at least

one impact category?
Vl Qualitative rating of the impact categories by G1 Assignment of the normalised indicator values to
the parameters place of effect, time and risk. a five level scale for quantitative importance of the
indicator.
V2 First comparison of alternative A and B. Only if G2 Assignment of impact categories to a five level
all results favour the same scenario, a result is scale for environmental importance, using the
given. arithmetical mean of the values assigned in Vl for
each category.
V3 Assignment of levels of hierarchy between the G3 Calculation of the overall importance for each
impact categories of both alternatives consider- indicator using a combination matrix.
ing time, place, hazard and quantitative indi-
cator value. Designation of the probably better
scenario (hypothesis).
V4.1 + 2 Confirmation of the hypothesis eliminating low G4.1 + 2 Ranking of scenario A and B eliminating impact
level categories favouring the proposed loser by categories with a lower overall importance which
higher level categories favouring the proposed favour the supposed loser with higher level cate-
winner. A first attempt eliminates categories gories which favour the supposed winner. Similar
one by one (4.1) in a second attempt (4.2) to V4, two attempts are made. In G4.1 only one by
several impact categories can be eliminated by one elimination is possible, in G4.1 several impact
a higher one if the sum of their z values is categories can be eliminated by a higher one if
lower. The result is confirmed when all cate- the sum of their quantitative importance is lower.
gories which favour the supposed loser can be The result is confirmed when all categories which
eliminated. favour the supposed loser can be eliminated.

Fig. 3: Iterative approach of the valuation method

278 Int J LCA 8 (5) 2003

LCA Methodology Modular Approach

2.3 Discussion, from an abstract point be given in this case. Describing and documenting the subjec-
Combining valuation methods, and treating them as modules2, tivity introduced helps also to make sure users know what
has advantages and disadvantages. The most important ad- they do when performing an automated valuation procedure.
vantage of the iterative valuation method proposed is its flex-
ible level of subjectivity used in the valuation, since it does not 3 Method Application
establish a certain level of subjectivity considered acceptable The method described in the previous chapter will be dem-
by using one specific valuation method or another, but in- onstrated by the help of a virtual case study. The case study
creases the level of subjectivity stepwise up to a level where a
stems from the railway industry, being a Life Cycle Assess-
clear ranking between both alternatives compared is possible.
ment of two different electric motors m l and m2 for a met-
This enables the LCA practitioner him- or herself to judge
ropolitan train operating in the area of Frankfurt and Stutt-
whether the 'price of subjectivity' paid for getting a clear rank-
gart in Germany. Both motors differ in product design and
ing is too high or not. The type of LCA performed (e.g. inter-
in their materials composition. The interpretation for the
nal use, external use), the quality of data used in the study,
case study is performed by using trainEE, Fig. 4 shows the
possible requirements of different stakeholders on the trans-
form to start the interpretation.
parency and objectivity of the LCA result, and other specific
considerations will play a role in this decision.
The resuk of the proposed valuation method is an ordinal rank-
ing between two alternatives accompanied by the level where a
result was achieved. Since both methods (UBA and CAU) pro-
duce clear and reproducible results when performed by a com-
puter (and not by human beings), the iterative approach for a
combination described in detail above is also reproducible.
The level of additional effort for the LCA practitioner who will
use this methodology within the software is very low. It is re-
quired to either accept the default values for valuation or - if
this is considered necessary - to change them. Since a normali-
sation of indicator values also is required, an additional data
acquisition for global contributions to each impact category
might be necessary. But since normalisation is used frequently
in LCA today, gathering these normalisation factors is not con-
sidered to be specific to the valuation method described here.
A certain disadvantage of the method may show when a large
number of different alternatives are compared within one study.
Frequently, this is done by a pair wise comparison of each pos-
sible pair of alternatives, yielding a large number of pair wise
ranking results. It is then required to rank, in a second step, all
alternatives based on these pair wise comparisons. This rank- Fig. 4: Form to start the interpretation in trainEE
ing should also take into account the level of the result (I to IV
see above) from the pair wise comparisons. In some cases, it
may be difficult to establish a clear and reproducible order of The impact assessment leads to impact indicators for both
alternatives from a large field of alternatives compared. motors in eight categories. Fig. 5 shows these results as rela-
tive values, i.e. results for m l divided by results for m2, for
Some implications from the proposed valuation approach each impact category. It can be seen that in this case, motor
are worth being mentioned: ml scores better than m2 in the categories human toxicity
9 The iterative valuation approach can be extended to cover even potential, eco toxicity potential, and climate change potential.
more valuation methods. For example, as a third step, a valuation
method which obtains a single value (as the Ecoindicator 99 The upcoming paragraph describes in detail the outcomes
method does, in the Hierarchist, or other modes) can be added. of different steps of the interpretation procedure.
9 Different methods combined should contain consistent value
choices. For this reason, we use identical factors to assess the The first check ((V0), about whether m l and m2 are equal
environmental importance of impact categories in both methods, in each impact category) is passed, m l and m2 are not pari
slightly deviating from the original methods.
in the case study.
A certain subjectivity, and influence on the result, may be in-
troduced by the order in which those new methods are ex- Step (V2) (whether ml scores equal or better than m2 in
ecuted. Therefore, a clear description of the 'subjectivity inter- each category, or vice versa, i.e. whether there is a dominant
face' of each method used, and a consistency check of the solution or not) proves no dominant solution. Both results
subjective values inherent in the methods combined, should can also be verified by looking at Table 3: In three catego-
ries, ml wins, in other five, m l loses against m2.
aA module defined as 'a self-contained unit or item that is used in combi- As described in section 2.2, the remaining procedure tries to
nation with other units' (UELD 2002). (Jungbluth et al. 2000) present a
'modular LCA', where different parts of an LCA for food (transport, pack- achieve a clear winner and loser by worsening the results
aging, consumption, and so forth) are treated as modules. for the assumed winner, or by improving the results for the

Int J LCA 8 (5) 2003 279

Modular Approach LCA Methodology

Fig. 5: Indicator values of motor ml divided by those of motor m2 for eight impact categories, initial values and scenarios ccp+ and ccp-. Further
explanation see text

In the following step V4.1, the procedure is unsuccessful in

climate change ) hierarchylevel its try to strike out categories where the loser m2 has won
1
(see Fig. 5: aquatic eutrophication, ozone depletion poten-
tial, resource depletion, terrestric eutrophication, acidifica-
(res~ depleti~1 ( eco t~ ) I eutr~176
(aquatic) 2
tion) on a one-by-one basis. Only one of the categories (e.g.
acidification, by the help of eco toxicity) can be cleared.
I acidificati~ ) (~176176 (eutrOphicati~ I 3 In the next step V4.2, aggregating several categories where
I m l has lost does not change the picture. Table 2 shows the
I h,,ao,ox,oi,y starting values.
Only two categories (climate change and acidification) can
Fig. 6: Calculated hierarchy levels for the impact categories in the case be used for clearing the categories where the assumed win-
study, initial values
ner has lost, since human toxicity has the lowest hierarchy
level (and hence there is no other category with lower hier-
hypothetical loser, respectively, based on the hierarchy in archy's level which may be cleared by human toxicity). Eco
the importance of different impact categories, as described toxicity is used for clearing acidification, and ccp's indica-
in section 2.2. The calculated hierarchy level for each im- tor value is smaller than the sum of the inferior categories.
pact category is shown in Fig. 6. Climate change potential is As a consequence, it is not possible to get a clear result in
the top-level category, three categories get a hierarchy level this valuation step.
of 2 and 3, respectively.
Here, and also in the following steps, there are several differ-
Next, step (V3) finds that m2 has a higher, i.e. worse, value ent actions possible to perform (which category should be used
in climate change potential, which is the top in the catego- against which other category?), with possibly a different out-
ries' hierarchy. So, it proposes m2 as 'assumed loser'. Note come. To overcome that problem, the method is implemented
that it is vital within this step not to have several highest so that it uses a brute force approach, i.e. it tries all possible
level categories with different losers and winners. actions, until one proves successful. This is possible because

Table 2: The basis for the interpretation step V4.2: Superior categories where the winner has won, and inferior categories where the winner has lost,
together with differences delta z in their dimensionless z-values

a t ........ .... : ,uper!~ ~ ~l'~ i:!~!

climate change 1 2 resource depletion 1,06E-09 2,91E-10 m2 ml

climate change 1 2 eutrophication 1,06E-09 1,03E-09 m2 ml
(aquatic)
climate change 1 3 eutrophication 1,06E-09 7,95E-10 m2 ml
(terrestric)
climate change 1 2 ozone depletion 1,06E-09 1,14E-11 m2 ml
eco-toxicity 3 3 acidification 7,44E-10 6,84E-11 m2 ml

280 Int J LCA 8 (5) 2003

LCA Methodology Modular Approach

Table 3: calculated importancefor the impact categories according to the Values for both of these scenarios are shown in Fig. 5, indi-
UBA method cated as ccp+ and ccp-. Note that indicator values for the
iii
Calculated other categories remain unchanged compared to the initial
case study. This demonstrates clearly that the method is sen-
climate change m2 5 sitive to changes in important impact categories.
eco-toxicity m2 4
human toxicity m2 2 4 Conclusions and Outlook

eutrophication (aquatic) ml 4 This text presents a new method to perform a valuation of

impact category results in LCAs. The method consists of a
ozone depletion ml 3
combination of existing, unique valuation methods described
resource depletion ml 3 earlier in literature, which were broadly used in LCA case
eutrophication (terrestric) ml 2 studies in Germany, and which both are of a rather non-
aggregating nature, and were designed to be performed by
acidification ml 2
LCA experts.
The iterative nature of these methods and especially the com-
the number of impact categories (and hence the number of bination of these methods, helps in achieving a valuation
possible actions) is limited, and this makes sure that no possi- result for the LCA with not more subjective and aggregat-
ble solution is overlooked by the valuation method. ing elements than necessary. Subjective elements are clearly
separated from others.
The next step in the evaluation is a qualitative one, it is the
first part of the UBA method, see Fig. 3. Similar to the pre- The method is implemented in a computer software that
vious step, climate change potential gets the highest impor- runs on a PC. Besides constants used within the method (e.g.
tance of the impact categories, see also Table 3. Since motor normalisation factors), the software needs as input solely
m2 loses in this category, m2 is assumed to lose in compari- indicator values from the Impact Assessment, and it pro-
son to m l , it is taken as the 'assumed loser' (steps G2 and vides as output the valuation's result: Which of both prod-
G3 according to Fig. 3). ucts compared scores better in the Impact Assessment. To
better help the user to understand the result and to use it
In the following step G4.1, the algorithm is not successful in
appropriately, this statement is accompanied by the infor-
eliminating the categories where ml has lost. For example,
mation in which step of the algorithm the result was achieved,
using the category eco toxicity against aquatic eutrophi-
especially how many aggregating steps were applied, and
cation, and human toxicity against acidification, there re-
which subjective elements introduced.
mains climate change on the winner's side, and three cat-
egories on the side of the assumed loser, which evidently cannot The implementation makes the application of the method
be compensated on a one-by-one basis. That means that the easier by transferring routine work (building hierarchy lev-
assumption cannot be confirmed in this step. Table 3 shows els, eliminating different categories, and so forth) from LCA
the calculated importance (which replaces the 'hierarchy experts to a machine. With a given valuation of impact cat-
level' from the CAU method). egories, and given indicator results, the method ensures a
Finally, in the last step G4.2, the method provides a result: reproducible result, and prevents erroneous steps in a rather
Aggregating the categories where m2 has won (ODP, AP, complicated valuation procedure. It further helps largely in
resource depletion), all categories can be compensated by hiding the complexity of the method from the user, since for
categories where m l has won. the user, the method needs only a single step and is in that
respect comparable to, e.g., weighting methods.
--) according to the method, m2 scores worse than ml. The
assumption from the beginning could be confirmed, finally. The implementation also allows performing a Monte Carlo
simulation of the valuation method, which gives access to a
For the case study, the behaviour in the valuation strongly thorough and systematic analysis of the method's behav-
depends on the value in the category climate change. For iour with uncertain data, and it facilitates studying the meth-
example, varying the result for this category only in two od's sensitivity / robustness by introducing small changes in
scenarios 'ccp-' and 'ccp+' provides as results: indicator values (Ciroth 2002). Care has to be taken to make
9 'cop-': Reducing the climate change potential for ml (in relative sure users know what they do when performing an auto-
values: from 89% to 79% in relation to the loser m2), yields a clear mated valuation procedure.
result already in step V4.2. Also in this scenario, ml wins. The
algorithm does not need to go through the UBA method at all. The method is ready to be used as such; in addition, it offers
9 'ccp+': With a climate change potential of 92% (compared to 89% rich possibilities for future extensions:
in the initial value), there are two top level impact categories (ccp
1. Uncertainty: Some commercial LCA software packages
and aquatic eutrophication), and, since ml wins in ccp and m2 in
aquatic eutrophication, the application of the CAU method is not already provide an uncertainty assessment for the inven-
possible. The algorithm skips the CAU method, and gets a clear tory and impact assessment, and provide information on
result by performing the UBA method. Also, in that case, ml wins. the uncertainty in indicator values. It is desirable for the

lnt J LCA 8 (5) 2003 28 1

Modular Approach LCA Methodology

m e t h o d to be able to handle this information, i.e. to be Goedkoop M, Hofstetter P, Miiller-Wenk R, Spriemsma R (1998): The
able to explicitly deal with uncertainty. This w o u l d also Eco-Indicator 98 Explained. Int J LCA 3 (6) 352-360
GreenDeltaTC (2002): www.GreenDeltaTC.comlTCIsoftwarel
help to m a k e the m e t h o d more stable to uncertainties
trainEE.html
i n t r o d u c e d in i m p o r t a n t c a t e g o r i e s (such as c l i m a t e Hildenbrand J (1999): Vergleichende Darstellung yon Auswertungs-
change in the case study presented above). methoden in Okobilanzen. Diplomarbeit im Rahmen des Studien-
2. N o r m a l i s a t i o n factors: To ensure a w o r l d w i d e applica- gangs Technischer Umweltschutz im Fachbereich 6 - Fachgebiet
bility, especially with regard to the ongoing life cycle ini- Abfallvermeidung der Technischen Universit/it Berlin
tiative, the m e t h o d ' s value will increase if it provides nor- ISO -International Organisation for Standardisafion (2000): ISO 14042
malisation factors, and factors describing the importance - Environmental management - Life cycle assessment - Life cycle
impact assessment
of i m p a c t categories, on a w o r l d w i d e level, or, even bet- Jungbluth N et al. (2000): Jungbluth, N, Tietje, O, Scholz, R: Food
ter, on a level o f different regions in the w o r l d where Purchases: Impacts from the Consumers' Point of View Investigated
these factors matter. with a Modular LCA. IntJ LCA 5 (3) 134-142
3. T o o l b o x : The m e t h o d m a y be p a r t of a t o o l b o x of m o d u - Plinke E et al. (2000): Okobilanz fiir CJetr/inkeverpackungen II, Umweh-
lar valuation m e t h o d s , that are implemented in a soft- bundesamt (UBA):Texte 27/2000, Berlin
ware and m a y be c o m b i n e d by L C A users according to Pr6 Consultants (2000): The Eco-indicator 99 - A damage oriented
method for Life Cycle Impact Assessment. Manual for Designers
their needs a n d the goals of a specific case.
Schmidt WP, Sullivan J (2002): Weighting in Life Cycle Assessment in a
The software will also be m a d e available as a stand alone Global Context. IntJ LCA 7 (1) 5-10
solution, so t h a t it m a y be used in any LCA case study with Schmitz S, Oels HJ, Tiedemann A (1996): Life-cycleassessment for drinks
packaging systems. UmweItbundesamt (UBA): Texte 19/96, Berlin
m i d p o i n t i m p a c t assessment.
Sepp/il/iJ, H~im/il/iinenRP (2001): On the Meaning of the Distance-to-
Target Weighting Method and Normalisation in Life Cycle Impact
References Assessment. Int J LCA 6 (4) 211-218
Siemens, et al. (2000): Verbundprojekt Bahnkreis: nachhahiges
Bengtsson M (2000): Environmental Valuation and Life Cycle Assess- Wirtschaften am Beispiel yon Schienenfahrzeugen, 2000; F/Srder-
ment. G6teborg: Chalmers University of Technology kennzeichen BMBF 02PV21334, online publication of the Technical
Brand G, ScheideggerA, Schwank O, Braunschweig A (1998): Bewertung University Hannover; h t~://has22.tib.uni-hannover.de:8080/DB=1/
in Okobilanzen mit der Methode der 6kologiscben Knappheit - SET=2/TTL=3/SHW?FRST= 1
Okofaktoren 1997; BUWAL Schriftenreihe Umwelt Nr. 297, Bern Trebst, W, Fleischer G, Fischer W (2001): Interdiszipliniires Forschungs-
Ciroth A (2002): Monte Carlo Simulation in the interpretation step of a projekt 'Nachhaltiges Wirtschaften am BeispieI yon Schienenfahr-
Life Cycle Assessment, platform presentation SETAC Annual Meet- zeugen (BAHNKREIS)'. In: ETR 89 January/February 2001, Fach-
ing, Vienna, 14 May 2002 zeitschrift fiir die gesamte Bahntechnik, S. 29-37
Cliff R (2001): Industrial ecology and material chain management, key TUB (2002): http//itu301 ur.tu-berlin de/projekte/bahnkreis/engl/
lecture, l l t h Annual Meeting of SETACEurope, Madrid, 9 May 2001 bahnkreis.htm
Ecosite (2002) http://www.ecosite.co.uk/depart]sandm.html Udo de Haes HA, et al. (1999): Best Available Practice Regarding Im-
Finnveden G (1997): Valuation Methods Within LCA -Where are the pact Categories and Category Indicators in Life Cycle Impact As-
Values? IntJ LCA 2 (3) 163-169 sessment - Background Document for the Second Working Group
Finnveden G (2000): On the Limitations of Life Cycle Assessment and on Life Cycle Impact Assessment of SETAC Europe (WIA-2). Int J
Environmental Systems Analysis Tools in General. Int J LCA 5 (4) LCA 4 (2) 66-74
229-238 UELD (2002): Ultralingua English Language Dictionary, h ttp://
Fleischer G e t al. (2000): Eco-Design - Effiziente Entwicklung wxvw.uhralingua.net/dictionary/index.php3
nachhahiger Produkte mit euroMat. Berlin: Springer Verlag 2000 Volkwein S, Gihr R., Kl6pffer W (1996): The Valuation Step within
Giegrich J e t al. (1995): Bilanzbewertung in produktbezogenen LCA - Part II: A Formalized Method of Prioritization by Expert
Okobilanzen - Evaluation von Bewertungsmethoden, Perspektiven; Panels. IntJ LCA 1 (4) 182-192
Forschungsbericht ver6ffendicht in UBA-Texte 23/95: Methodik der
produktbezogenen Okobilanzen; Berlin Received: January 14th, 2003
Goedkoop M (1995): De Eco-Indicator 95. Final report; NOH report Accepted: August 27th, 2003
9523; PR~ Consultants; Amersfoort (NL); ISBN 90-72130-77-4 OnlineFirst: August 28th, 2003

Andreas Ciroth studied environmental engineering at the TU Ber- Ciroth A, HageIQken M, Sonnemann GW, Castells F, Fleischer G
lin. His master thesis dealt with an example of a screening LCA. (2002): Geographical and Technological Differences in Life Cy-
From 1998 to 2001 he worked as a research scientist at the TU cle Inventories Shown by the Use of Process Models for Waste
Berlin where he completed his dissertation on error propagation in Incinerators- Part I1:Technological and Geographical Differences.
2001. In August 2001, he founded GreenDeltaTC, a consultant com- Int J LCA 7 (6) 363-368
pany in the field of software development, modeling and decision
support, with emphasis on environmental questions. Starting from Ciroth A, HagelQken M, Sonnemann GW, Castells F, Fleischer G
1999, he shared the SETAC Working Group on Impact Assessment (2002): Geographical and Technological Differences in Life Cy-
(task group normalization and weighting) and Data Quality. Up to date, cle Inventories Shown by the Use of Process Models for Waste
he has published a limited number of papers on uncertainty assess- Incinerators- Part I. Technologicaland Geographical Differences.
ment, modeling, and LCA case studies. His scope of interest includes: Int J LCA 7 (5) 295-300
Systems analysis, modeling in the face of data gaps and data uncer-
tainties, model verification and validation, user specific software, and Ciroth A (2002): New LCA Theses - Error Calculation in Life Cycle
database design and development. Published in Int J LCA: Assessments. Int J LCA 7 (5) 310

282 Int J LCA 8 (5) 2003