Journal of Cleaner Production 13 (2005) 1337e1343

www.elsevier.com/locate/jclepro

The ecoinvent database system: a comprehensive

web-based LCA database
Rolf Frischknecht a,*, Gerald Rebitzer b
a
Swiss Centre for Life Cycle Inventories, c/o EMPA, Überlandstraße, 129, CH-8600 Dübendorf, Switzerland
b
Life Cycle Systems Group, Swiss Federal Institute of Technology Lausanne (EPFL), CH-1015 Lausanne, Switzerland
Accepted 3 May 2005
Available online 22 June 2005

Abstract

The paper describes the general structure of the ecoinvent database developed by the Swiss Centre for Life Cycle Inventories. The
database accommodates more than 2500 background processes often required in LCA case studies.
Quality guidelines, established in order to ensure coherent data acquisition and reporting across the various institutes involved,
are described. These include aspects such as the reporting of pollutants (e.g., heavy metals), or the nomenclature of processes and
elementary flows.
The data (exchange) format is also described. Processes are documented with the help of meta-information and flow data
(including both unit process raw data and aggregated LCI results). The structure of the data format corresponds to the ISO/TS
14048 data documentation format. Data exchange between project partner institutes and between the database and its customers
(database users) is based on XML-technology. Matrix inversion is used to calculate the cumulative LCA data using efficient
algorithms and making use of the fact that LCA matrices are usually sparse.
Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: Life cycle inventory (LCI); Ecoinvent; Data format; Matrix inversion; LCA database; EcoSpold; ISO/TS 14048

1. Introduction and background tool for e.g., Integrated Product Policy (IPP), Technol-
ogy Assessment or Design for Environment (DfE). In
Up to now, several public Life Cycle Assessment parallel with this increasing trend for LCA applications
(LCA) databases exist in Switzerland, partly covering the demand for high quality, reliable, transparent and
the same economic sectors [1e6]. However, life cycle consistent LCA data increases as well. Only a few
inventory data for a particular material or process publicly available LCI databases fulfill these criteria,
available from these databases often do not coincide and most of them from the 1990s and thus not up-to-date.
thus the outcome of an LCA is (also) dependent on the
database used and therefore the executing institute.
2. Goal of the ecoinvent 2000 project
Furthermore, the efforts required to maintain and
update comprehensive and high quality LCA-databases
Under the lead of EMPA, LCA-institutes in the ETH
are beyond the capacity of any individual organization.
domain (Swiss Federal Institutes of Technology (ETH)
At the same time, LCA is receiving more and more
Zürich and Lausanne, Paul Scherrer Institute (PSI)
attention in industry and authorities as one important
Villigen, Swiss Federal Institute for Environmental
Science and Technology (EAWAG), and Swiss Federal
* Corresponding author. Fax: C41 1 940 61 94. Laboratories for Materials Testing and Research
E-mail address: frischknecht@ecoinvent.ch (R. Frischknecht). (EMPA) St. Gallen and Dübendorf) as well as the

0959-6526/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jclepro.2005.05.002
1338 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343

LCA-department of the Agroscope FAL Reckenholz, 1. Central database,

the Swiss Federal Research Station for Agroecology and 2. Calculation routines,
Agriculture, in Zürich started close cooperation. To- 3. Editor,
gether with the Swiss Agency for the Environment, 4. Administration tool,
Forests and Landscape (SAEFL or BUWAL), the Swiss 5. Query tool,
Agency for Energy (BFE) and other Swiss Federal 6. Data (exchange) format,
Offices, the Swiss Centre for Life Cycle Inventories was 7. Local databases.
founded. The LCA-institutes launched the project
ecoinvent 2000, which resulted in a harmonised, revised Ad 1. The central database contains Life Cycle
and updated Swiss national LCI database. The database Inventory data on energy systems, transport systems,
comprises LCI data from the energy, transport, building waste treatment systems, chemicals, building materials,
materials, chemicals, paper and pulp, waste treatment etc., and Life Cycle Impact Assessment (LCIA) methods
and agricultural sectors reflecting the production and such as the Swiss Ecological Scarcity 1997 [8], Eco-
supply situation in the year 2000, based on the Swiss and indicator 99 [9], IMPACT 2002C [10] or the CML
European demand patterns. The major applicability is in characterisation scheme 2001 [11]. The database is
a European context, but selected data sets (such as oil located on a computer server and accessible via the
production in Nigeria or natural gas production Internet.
in Russia) have a broader international application. Ad 2. Data are supplied by the partner institutes as
The tasks within the project have been distributed non-terminated unit processes (i.e., they can and usually
according to the expert knowledge of the partners (see do contain exchanges from and into the technosphere as
Table 1). well as elementary flows). The computation of cumula-
A large network-based database and efficient calcu- tive inventory results is performed with powerful
lation routines are required for handling, storage, calculation routines related to the central database.
calculation and presentation of data and have been Unit process raw data as well as LCI results include
developed over the course of the project. The corre- (cumulative) uncertainty ranges.
sponding components are partly based on preceding Ad 3. The local database administrators of the
work performed at ETH Zürich [7]. participating institutes use the editor to create new data
sets and to change, complete or delete existing data sets.
The editor administrates the data set names (via a direct
link to the central database, where the index of data set
3. Basic structure of the ecoinvent database system
names is located), ensures the use of the actual list of
names when compiling new inventories and includes
The ecoinvent database system consists of the
a unit converter. The editor acts as the interface between
following components (see Fig. 1):
the local administrator and the central database and
generates files in the ecoinvent data format (named
Table 1 EcoSpold, see below).
Database content, responsible institutes and their partners in LCI data
compilation
Database content Responsible institute Partners
Energy supply PSI ESU-services Local
Local Administrator database
-Fuels
-Heat production
-Electricity production
Editor
Materials and waste EMPA Doka Life
-Plastics Cycle Assessments, Central
-Paper and board Chudacoff Administrator
-Basic chemicals Öko-science
-Detergents
-Waste treatment Administration Central Calculation
services tool Database routines
-Metals
-Wood
-Building materials
-Basic chemicals
Query tool
Transport services ETHZ UNS
Basic chemicals ETHZ ICB
Data consumer XY
Agricultural products FAL/FAT
and processes
Fig. 1. The basic structure of ecoinvent database system.
 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343 1339

Ad 4. The administration tool supports the integration dismantling) and the operation phase. Thereby the
of data sets delivered by the cooperating institutes into entire production means necessary to run, for instance,
the central database. It helps to verify the completeness of a fuel cell combined heat and power unit is evenly
data sets, calculates inventories and (normalised and spread over the expected life time production. For
weighted, if appropriate) category indicator results and instance, 1 kWh ‘‘electricity, at fuel cell’’ requires a tiny
supports the administration of ecoinvent users. share of 1 unit ‘‘fuel cell, 1.5 kWe’’.
Ad 5. The Query tool is the users’ interface to the Co-production processes such as the above-men-
database and is used to download data sets from the tioned fuel cell, which delivers both electricity and heat,
central database. It enables the search for individual are stored as such in the database (i.e., before allocating
processes, for processes of a certain economic sector inputs and outputs to its co-products). In order to be
(e.g., transport or energy sector) or for data from able to attribute a certain share of requirements and
a certain institute. General information (so-called meta- burdens to each of the co-products, allocation factors
information) on the processes (technology, age of data, are defined and stored separately. This procedure allows
geographic coverage, etc.) is accessible to everyone, for a flexible and easy handling (and change/adaptation,
whereas the quantitative LCI data is only accessible for if necessary) of allocation factors.
registered ecoinvent members (customers). All processes are defined by the four data fields
Ad 6. The data exchange format lists all data fields ‘‘name’’, ‘‘unit’’, ‘‘location’’ and whether it is an
that were available to describe a data set. It has evolved ‘‘infrastructure process’’ or not.1 The process name is
from the international SPOLD data exchange format created according to the following order and scheme
[12] and corresponds to the international technical (see also Table 2):
specification ISO/TS 14048 [13]. Some of the data fields
are mandatory, i.e. information must be provided. (1) Name of the good/service, production process or
Among other features, the data exchange format allows processed product, level of treatment (see below).
for specifying upper and lower estimates (or the 95% (2) Additional description (if available) in the following
standard deviation) as well as the probability distribu- fixed order: sum formula, site (or place of origin),
tion (e.g., lognormal) of inventory data. company, imports included or not, any further
Ad 7. Commercially available LCA-software such as description.
Emis, PEMS, Regis, SimaPro, TEAM and Umberto (3) Level of the value chain (e.g., ‘at plant’, ‘at regional
are used as local databases. These local databases storehouse’) or destination (especially for wastes,
are tailored for an implementation of ecoinvent data e.g., ‘to waste incineration’, ‘to landfill’).
v1.0 and its updates. It is recommended to use the
ecoinvent data (exchange) format for the purpose of For an unequivocal identification of elementary
data import. flows, slightly different data fields are required, namely,
In the following sections three aspects of the name, unit, category and subcategory. The names of
ecoinvent database system are highlighted. The first elementary flows are based on the structure developed
covers the quality guidelines for life cycle inventory by the SETAC working group on data quality and data
analysis performed within ecoinvent 2000 project. The availability [14,15]. Category and subcategory are used
second describes certain aspects of the web interface to describe the compartment (air, water, soil and
(Query tool), and the third focuses on the calculation resources) and its specification (e.g., for air: strato-
routines implemented in the ecoinvent database system. sphere, high population density, low population density,
protected area and unspecified). Additionally, long-term
emissions are distinguished for processes that are likely
4. Quality guidelines for ecoinvent data to emit during several thousands of years (such as
landfill sites, nuclear waste depositories and overbur-
The creation of one central life cycle assessment dens at mining and milling sites).
database requires a high degree of coordination and Pollutants are reported only once and on the level of
harmonisation. In this section, several harmonisation detail of the information source. This avoids double
issues are listed and discussed. Beside structural aspects counting and conserves the original information and
and naming conventions, content-related aspects have detail. For instance, benzene, reported as emitted to
been elaborated and unified. This guarantees a maximum air in a highly populated area is registered under
degree of consistency of the available process data. ‘‘benzene, air, high population density’’ but not under

4.1. Structural aspects and naming conventions 1

For infrastructure processes, the construction (and dismantling)
requirements and emissions are reported (as opposed to the other
The inventory of a unit process is divided into the processes where only operational requirements (incl. a demand for
needed production means (infrastructure; including a share of the infrastructure process) are reported).
1340 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343

Table 2 waste treatment processes. If no specific information is

Examples of process names used in ecoinvent data v1.0 available, standard waste treatment processes defined per
Name Location Unit Infrastructure material are applied. It is assumed that inert materials go
process to landfills, plastics and paper are incinerated and metals
Heat, natural gas, RER MJ no are recycled.
at industrial Allocation is a ubiquitous issue, which calls for
furnace O100 kW
Gypsum, at plant CH kg no
a harmonised approach. A cut-off approach is used for
Electricity, low voltage, UCTE kWh no recycled materials and for by-products (outputs with no
production UCTE, at grid economic value that are not sent to waste treatment but
Disposal, newspaper, CH kg no are used in other processes). No burdens/impacts and no
14.7% water, to municipal requirements of a proceeding process chain and of
incineration
Transport, airplane, freight GLO tkm no
a process are allocated to the recycled materials and by-
Nuclear power plant, DE unit yes products, respectively. On the other hand, no benefits
pressure water reactor 1000 MW are granted for any subsequent use of recycled material
or by-product. No fixed prescriptions are made for joint
product allocation (co-products) except that system
‘‘hydrocarbons, aromatic’’, nor under ‘‘NMVOC, non- expansion (especially the ‘‘avoided burden’’-concept) is
methane volatile organic compounds, unspecified ori- not recommended.
gin’’, nor under ‘‘VOC, volatile organic compounds, Fossil and renewable carbon are distinguished for
unspecified origin’’. Benzene might, however, also be CO2-, CH4- and CO-emissions. For renewable energy
included in the sum parameter ‘‘NMVOC, non-methane sources and materials, an equal amount of CO2 is
volatile organic compounds, unspecified origin’’, in case registered as a resource consumption according to the
a Benzene emission is reported as NMVOC in the data binding capacity of the corresponding crops. Carbon
sources used and the analyst is not able to identify the that is emitted as CO is considered when calculating
emitted hydrocarbon. CO2-emissions. On the other hand, CO is assigned
a global warming potential assuming its subsequent
conversion to CO2.
4.2. Content-related aspects Uncertainty of flow data is quantified on the level of
each individual input and output of the unit processes. If
While structural aspects and naming rules can in uncertainty is not known (because not stated in the
many cases be controlled by the software system, sources used or because not known by the company
a consistent implementation of content-related rules is providing the data) a standardised estimation procedure
less straightforward. Nevertheless, clear rules are re- is used. A data quality matrix has been developed based
quired in order to minimise differences caused by on the pedigree matrix published by Pedersen Weidema
individual, unsystematic choices of the LCI data and Wesnaes [16]. Scores from one to five are given for
collectors involved. reliability, completeness, temporal correlation, geo-
In ecoinvent, system boundaries are drawn based on graphical correlation, further technological correlation
expert knowledge and not based on fixed rules such as and sample size. Fixed uncertainty factors are attributed
mass or energy shares. If an emission of a pollutant is to each of the scores and an additional basic uncertainty
expected but no data are available, estimates are used in is attributed to categories of exchanges (such as
order to identify whether or not this pollutant may be electricity and thermal energy consumption, groups of
environmentally relevant. combustion emissions, waste treatment requirements
Electricity is supplied as high, medium and low and the like). In all cases a lognormal distribution is
voltage with increasing losses and requirements of assumed. With the help of these standardised uncer-
production means. Hence electricity demand of pro- tainty factors, the geo-variance (square of the geometric
cesses must be linked to the correct (or most likely) standard deviation) is determined for each individual
voltage. National supply mixes (as well as the export exchange in the unit processes.
mixes) are calculated based on the domestic production
plus the imports. In cases where the electricity mix
actually purchased deviates from the average supply mix 5. Data exchange and XML-technology
of a nation (or region) such specific mixes (or particular
power plant technologies) are used. 5.1. Data set documentation
In cases where the exact distances are unknown,
standard transport distances are applied for the pro- A process, its products and its life cycle inventory
duction phase of materials such as steel, cement, basic data are documented using the ecoinvent data format
chemicals, etc. A similar approach is utilized for (EcoSpold) with the following structure:
 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343 1341

For a flexible application of a data exchange between

Meta-information local LCA-software tools and the central database,
Process a data exchange format in XML-technology is used. For
ReferenceFunction, defining the product or service that purpose XML schemes are applied. Although
output to which all emissions and requirements are Document Type Definitions (DTD) are more wide-
referred. spread nowadays, they will most likely be substituted by
TimePeriod, defining the temporal validity of the schemes. On the one hand, schemes have a much higher
data set. performance and, on the other, they themselves use the
Geography, defining the geographical validity of XML language (as opposite to DTD, which uses its own
the data set. specific language).
Technology, describing the technology(ies) of the Schemes are used for validation and for documenta-
process. tion purposes. The scheme provides information on the
DataSetInformation, defining the kind of process general structure of an ecoinvent data set. Furthermore,
or product system, and the version number of the all elements of a scheme may be completed with docu-
data set. mentation information and comments.
Modelling and validation
Representativeness, defining the representativeness
of the data used. 6. Matrix-calculation with sparse matrices
Sources, listing the literature and publications used
in a data set. A unit process is the smallest portion of a product
Validations, listing the reviewers and their com- system for which data are collected when performing
ments. a life cycle assessment [17]. Unit processes are linked by
Administrative information inputs and outputs with
DataEntryBy, documenting the person in charge
of implementing the data set in the database. - the technosphere (technical systems or the economy)
DataGeneratorAndPublication, documenting the and
originator and the published source of the data set. - the biosphere (nature or the life-supporting eco-
Persons, listing complete addresses of all persons system).
mentioned in a data set.
Flow data The first link comprises exchanges with the economic
Exchanges, quantifying all inputs from technical system such as the purchase of intermediate goods
systems and nature to the process and all outputs (machinery, materials, ancillaries, etc.) or services (waste
from the process to nature and to other technical treatment, transportation, telecommunication). Direct
systems resource extraction (e.g., supply of river water) and
Allocations, describing and quantifying allocation direct emissions to air, water or soil are relations of the
procedures and factors, respectively, required for second kind.
multi-function processes. Accordingly, the directly caused energy and material
flows of a unit process form a vector that is divided
Customers (ecoinvent users) have the ability to check into a technosphere and a biosphere sector a and b,
the content of processes online (see excerpt in Fig. 2). This respectively. The vector includes m technosphere ex-
helps to judge whether or not the process is of interest and changes ai and n biosphere exchanges bj.
worthwhile to download the corresponding data set. 0 1
Links within the html-documents facilitate the navigation. a1
B. C
B C
B ai C
5.2. Role of XML-technology B C
B C

 B. C
a B am C
Once a data set is chosen for download, one or ZB C
B b1 C ð1Þ
b B C
several data sets are converted to one XML-file2 and B. C
saved on the local computer. XML schemes facilitate B C
B bj C
data exchange between different LCA-databases and B C
@. A
-software. It can easily be extended by LCA-software-
bn
specific requirements and upwards and downwards
compatibility poses no major problems.
The next step is to create a model of that part of the
world economic system which comprises all processes
2
XML: extended markup language. required to produce a certain good or for the delivery of
1342 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343

Fig. 2. Excerpt of online data set documentation in ecoinvent data format.

a certain service. Such a model will e among others e


include the energy supply sector, the transportation ) C
P Z ð3Þ
sector, the waste treatment sector and certain parts of D
the mining sector (minerals). All unit process vectors
together form a matrix with an economic part A and an contains information on the total (cumulative) require-
ecological part B. ments of economic entities (intermediate goods and
services) and about the total (cumulative) flows of
0 1 ecological entities (elementary flows, i.e. resource
a11 . a1l . extractions and emissions) of all unit processes the
B. . . .C matrix is composed of.
B C
B ai1 . ail .C For the numerical implementation of the matrix
B C
B .C inversion, direct methods are usually applied that make

 B. . . C
A B am1 . aml .C use of publicly available source code libraries (see [7] for
PZ ZB
B b11
C ð2Þ
B B . b1l .CC further details).
B. . . .C
B C
B bj1 . bjl .C
B C
@. . . .A 7. Outlook
bn1 . bnl .
Since Fall 2003 consistent, quality-controlled LCI
data with the reference year 2000 are available via the
By calculating the inverse of the economic part A of the Internet for many basic commodities and services (such
matrix P one gets the cumulative demand of interme- as energy supply, transportation and waste treatment
diate goods and services required to produce each of the services, building materials, wood products, chemicals
unit processes (economic part of the cumulative matrix and agricultural products) that are part of most LCI
called C). In order to be able to invert the matrix, it has process networks. The database content will be updated
to be square and non-singular. These two conditions are and extended every two to three years. On which
normally fulfilled when modelling real product systems. economic sectors to focus update and extension work is
The cumulative amounts of resource extractions and not yet decided.
emissions (cumulative ecological part D) are calculated The experiences gained concerning decentralised LCI
by multiplying the ecological part B of matrix P with the data acquisition using common quality guidelines and
computed cumulative economic part C. its feeding into a central database are very valuable for
The inventory matrix P), composed of C and D, future database developments. Variations of the data
 R. Frischknecht, G. Rebitzer / Journal of Cleaner Production 13 (2005) 1337e1343 1343

exchange format are possible thanks to the flexibility of Umwelt, Wald und Landschaft, Schriftenreihe Umwelt No. 250,
the XML-technology. Major LCA-software tool and 2. korrigierte und aktualisierte Auflage: Bern, Schweiz; 1998.
[7] Frischknecht R, Kolm P. Modellansatz und Algorithmus zur
LCA data suppliers are currently discussing the possi- Berechnung von Ökobilanzen im Rahmen der Datenbank
bility of creating a common data exchange format that ecoinvent. In: Schmidt M, Schorb A, editors. Stoffstromanalysen
would further enhance data exchange and by that in Ökobilanzen und Öko-Audits. Berlin/Heidelberg: Springer-
increase acceptance and efficiency of LCA. This may Verlag; 1995. p. 79e95.
happen either within the life cycle inventory programme [8] Brand G, Scheidegger A, Schwank O, Braunschweig A. Bewer-
tungin Ökobilanzen mit der Methode der ökologischen Knap-
of the UNEP SETAC Life Cycle Initiative or within pheit e Ökofaktoren 1997, Bundesamt für Umwelt, Wald und
COST Action 530. Landschaft (BUWAL) No. Schriftenreihe Umwelt 297. Bern; 1998.
[9] Goedkoop M, Spriensma R. The Eco-indicator 99: a damage
oriented method for life cycle impact assessment. 2nd revised ed.
In: Methodology Report, vol. 4. Amersfoort, The Netherlands:
References PRé Consultants; 2000. <http://www.pre.nl/eco-indicator99/>.
[10] Jolliet O, Margni M, Humbert S, Payet J, Rebitzer G,
[1] Frischknecht R, Hofstetter P, Knoepfel I, Dones R, Zollinger E. Rosenbaum R. IMPACT 2002C: a new life cycle impact
Ökoinventare für Energiesysteme. Grundlagen für den ökologischen assessment methodology. International Journal of Life Cycle
Vergleich von Energiesystemen und den Einbezug von Energie- Assessment 2003;8(6):324e30.
systemen in Ökobilanzen für die Schweiz, Gruppe Energie e [11] Guinèe JB, Gorrée M, Heijungs R, Huppes G, Kleijnde R, de
Stoffe e Umwelt (ESU). Auflage No. 1. Bern: Eidgenössische Koning A, et al, editors. Life cycle assessment: an operational
Technische Hochschule Zürich und Sektion Ganzheitliche guide to the ISO standards: Parts 1 and 2. Den Haag/Leiden, The
Systemanalysen, Paul Scherrer Institut Villigen, Bundesamt Netherlands: Ministry of Housing, Spatial Planning and Envi-
für Energie (Hrsg.); 1994. ronment (VROM) and Centre of Environmental Science (CML);
[2] Künniger T, Richter K. Ökologischer Vergleich von Freileitungs- 2001.
masten aus imprägniertem Holz, armiertem Beton und korro- [12] Weidema B. SPOLD ’99 format e an electronic data format for
sionsgeschütztem Stahl. Dübendorf: EMPA; 1995. exchange of LCI data (1999.06.24), SPOLD, !http://www.spold.
[3] Frischknecht R, Bollens U, Bosshart S, Ciot M, Ciseri L, Doka G, orgO; 1999.
et al. Ökoinventare von Energiesystemen: Grundlagen für den [13] International Organization for Standardization (ISO), Environ-
ökologischen Vergleich von Energiesystemen und den Einbezug von mental Management e Life Cycle Assessment e Data documen-
Energiesystemen in Ökobilanzen für die Schweiz, Gruppe Energie e tation format No. Technical Specification ISO/DTS 14048.
Stoffe e Umwelt (ESU). Auflage No. 3. Bern, CH: Eidgenössische Geneva; 2002.
Technische Hochschule Zürich und Sektion Ganzheitliche System- [14] de Beaufort-Langeveld ASH, Bretz R, van Hoof G, Hischier R,
analysen, Paul Scherrer Institut, Villigen, Bundesamt für Energie Jean P, Tanner T, et al. Code of life-cycle inventory practice
(Hrsg.); 1996. <http://www.energieforschung.ch>. (includes CD-ROM). SETAC; 2003.
[4] Habersatter K, Fecker I, Dall’Acqua S, Fawer M, Fallscheer F, [15] Hischier R, Baitz M, Bretz R, Frischknecht R, Jungbluth N,
Förster R, et al. Ökoinventare für Verpackungen, Bundesamt für Marheineke T, et al. Guidelines for consistent reporting of
Umwelt, Wald und Landschaft, Schriftenreihe Umwelt No. 250. exchanges from/to nature within life cycle inventories (LCI).
Bern, Schweiz; 1996. International Journal of Life Cycle Assessment 2001;6(4):192e8.
[5] Gaillard G, Crettaz P, Hausheer J. Umweltinventar der land- [16] Pedersen Weidema B, Wesnaes MS. Data quality management for
wirtschaftlichen Inputs im Pflanzenbau, Eidg. Tänikon: For- life cycle inventories e an example of using data quality
schungsanstalt für Agrarwirtschaft und Landtechnik (FAT); indicators. Journal of Cleaner Production 1996;4(3e4):167e74.
1997. FAT-Schriftenreihe No. 46. [17] International Organization for Standardization (ISO), Environ-
[6] Habersatter K, Fecker I, Dall’Acqua S, Fawer M, Fallscheer F, mental Management e Life Cycle Assessment e Principles and
Förster R, et al. Ökoinventare für Verpackungen, Bundesamt für framework No. European standard EN ISO 14040. Geneva; 1997.