IMPACT 2002+ A New Life Cycle Impact Assessment Methodology
Presemmg a New Meth6d
IMPACT 2002+: A New Life Cycle Impact Assessment Methodology
Olivier Jolliet*, Manuele Margni, Raphael Charles, S6bastien Humbert, J~r6me Payet, Gerald Rebitzer and
Ralph Rosenbaum
Industrial Ecology & Life Cycle Systems Group, GECOS, Swiss Federal Institute of Technology Lausanne (EPFL), CH-1015 Lausanne,
Switzerland
*Corresponding author (Olivier.lolliet@epfl.ch)
Abstract
The new IMPACT 2002+ life cycle impact assessment methodol-
ogy proposes a feasible implementation of a combined midpoint/
damage approach, linking all types of life cycle inventory results
(elementary flows and other interventions) via 14 midpoint cat-
egories ro four damage categories. For IMPACT 2002% new con-
cepts and methods have been developed, especially for the com-
parative assessment of human toxicity and ecotoxicity. Human
Damage Factors are calculated for carcinogens and non-carcino-
gens, employing intake fractions, best estimates of dose-response
slope factors, as well as severities. The transfer of contaminants
into the human food is no more based on consumption surveys,
but accounts for agricultural and livestock production levels. In-
door and outdoor air emissions can be compared and the inter-
mittent character of rainfall is considered. Both human toxicity
and ecoroxicity effect factors are based on mean responses rather
than on conservative assumptions. Other midpoint categories are
adapted from existing characterizing methods (Eco-indicator 99
and CML 2002). All midpoint scores are expressed in units of a
reference substance and related to the four damage categories
human health, ecosystem quality, climate change, and resources.
Normalization can be performed either at midpoint or at damage
level. The IMPACT 2002+ method presently provides characteri-
zation factors for almost 1500 different LCI-results, which can
be downloaded at http://www.epfl.ch/impact
Keywords:
Ecotoxicity; human toxicity; IMPACT 2002+; life
cycle impact assessment (LCIA); midpoint/damage approach
Introduction
Life cycle impact assessment (LCIA) methods aim to con-
nect, as far as possible, and desired, each life cycle inventory
(LCI) result (elementary flow or other intervention) to the
corresponding environmental impacts. According to ISO
14042, LCI results are classified into impact categories, each
with a category indicator. The category indicator can be lo-
cated at any point between the LCI results and the category
endpoints (where the environmental effect occurs) in the
cause-effect chain. Within this framework, two main schools
of methods have developed:
a) Classical impact assessment methods [e.g. CML (Guin&
et al. 2002) and EDIP (Hauschild and Wenzel 1998)] which
restrict quantitative modeling to relatively early stages in
the cause-effect chain to limit uncertainties and group LCI
results in so-called midpoint categories, according to themes.
Themes are common mechanisms (e.g. climate change) or
commonly accepted grouping (e.g. ecotoxicity).
b) Damage oriented methods such as Eco-indicator 99
(Goedkoop and Spriensma 2000) or EPS (Steen 1999), which
try to model the cause-effect chain up to the endpoint, or
damage, sometimes with high uncertainties.
Recently, the definition study of the SETAC/UNEP Life Cy-
cle Initiative suggested utilizing the advantages of both ap-
proaches by grouping similar category endpoints into a struc-
tured set of damage categories. In addition, the concept also
works with midpoint categories, each midpoint category
relating to one or several damage categories.
As shown in Fig. 1, LCI results with similar impact path-
ways (e.g. all elementary flows influencing stratospheric
ozone concentrations) are grouped into impact categories at
midpoint level, also called midpoint categories. A midpoint
indicator characterizes the elementary flows and other envi-
ronmental interventions that contribute to the same impact.
The term 'midpoint' expresses the fact that this point is lo-
cated somewhere on an intermediate position between the
Midpoint Damage
categories categories
Z Human toxicity
// Respiratory effects ~ Human Health
///Ionizing radiation
////
Ozonelayerdepletion
" "~ ....
/////
Photochemical oxidation
..... " ..... =
///S.~ Aquatic ecot oxicity ~~
Ecosystem QualRy
"~k~\ik~t" "~ Terrest dal ecotoxicity ~
~\ "~ Aquatic acidification
J//
\ \ \ \ ~ Aquatic eutrophication / / ?L~limate
?hange
"~'~ ~ Terrestrial acid/nutr // /,.,J ('eSuppo Systems)
~ ~
Land occupation
/ J
\ \ G,oba, wa--g
Non-renewable energy =
Resources
Mineral extraction
Fig.
1 : Overall scheme of the IMPACT 2002+ framework, linking LCI re-
sults via the midpoint categories to damage categories, based on Jolliet et
al. (2003a)
324
Int J LCA 8 (6) 324 - 330 (2003)
9 ecomed publishers, D-86899 Landsberg, Germany and Ft. Worth/]X * Tokyo ~ Mumbai 9 Seoul 9 Melbourne * Paris
A New Life Cycle Impact Assessment Methodology IMPACT 2002+
LCI results and the damage (or endpoint) on the impact
pathway. In consequence, a further step may allocate these
midpoint categories to one or more damage categories, the
latter representing quality changes of the environment. A dam-
age indicator result is the quantified representation of this
quality change. In practice, a damage indicator result is al-
ways a simplified model of a very complex reality, giving only
a coarse approximation to the quality status of the item. More
information on the general concept of such a methodological
LCIA framework can be found in Jolliet et al. (2003a).
The new IMPACT 2002+ LCIA methodology proposes a
feasible implementation of the aforementioned combined
midpoint/damage-oriented approach. Fig. 1 shows the overall
scheme of the IMPACT 2002+ framework, linking all types
of LCI results via 14 midpoint categories (human toxicity,
respiratory effects, ionizing radiation, ozone layer depletion,
photochemical oxidation, aquatic ecotoxicity, terrestrial eco-
toxicity, terrestrial acidification/nutrification, aquatic acidi-
fication, aquatic eutrophication, land occupation, global warm-
ing, non-renewable energy, mineral extraction) to four damage
categories (human health, ecosystem quality, climate change,
resources). An arrow symbolizes that a relevant impact path-
way is known or assumed to exist. Uncertain impact path-
ways between midpoint and damage levels that are not modeled
quantitatively are represented by dotted arrows.
In addition to this combined midpoint/damage structure, sev-
eral scientific challenges had to be tackled, especially in the
areas of human toxicological and ecotoxicological impacts:
9 How to adapt conventional regulatory-orientated risk
assessment methods, often based on conservative assump-
tions, in order to estimate cumulative chronic toxico-
logical risks and potential impacts in comparative appli-
cations such as LCA?
9 How to account in a generic but accurate way for non
linear functions, such as the intermittent character of rain-
fall or the differences between indoor and outdoor emis-
sions, which can generate large errors if neglected?
9
How to structure fate, exposure, and effect of chemicals in
a consistent way following impact pathways, looking at
production-based rather than subsistence-based exposures?
To address these challenges new concepts and methods for
the comparative assessment of human toxicity and
ecotoxicity were developed for the IMPACT 2002+ meth-
odology. For other categories, methods have been transferred
or adapted mainly from the Eco-indicator 99 (Goedkoop
and Spriensma 2000) and the CML 2002 (Guin6e et al. 2002)
methods. The following Sections of this paper discuss the
main assessment characteristics for midpoint and damage
categories, as well as related normalization factors, with a
focus on innovative features and performed adaptations.
1 IMPACT 2002+ at Midpoint Category Level
Midpoint characterization factors are based on equivalency
principles, i.e. midpoint characterization scores are expressed
in kg-equivalents of a substance compared to a reference
substance. Table I shows the reference substances and dam-
age units used in IMPACT 2002+. The principal scope is
common to all impact categories: overall long-term effects
are being considered through the use of infinite time hori-
zons (sometimes approximated by a 500 years horizon). In
general, the average impact has been modeled, avoiding the
use of conservative assumptions in determining effect fac-
tors. The updated midpoint characterization factors for the
number of substances indicated in Table 1 can be down-
loaded from the Internet at http://www.epfl.ch/impact.
Table 1: Number of LCI results covered, main sources for characterization factors, reference substances, and damage units used in IMPACT 2002+.
Sources are: [a] IMPACT 2002 (Pennington et al. 2003a, 2003b), [b] Eco-indicator 99 (Goedkoop and Spriensma 2000), [c] CML 2002 (Guin6e et al.
2002), and [d] ecoinvent (Frischknecht et al. 2003)
769 [a]
12 [b]
25 [b]
22 [b]
130 [b]
Human toxicity kg~ chloroethylene into air Human health
(carcinogens + non-carcinogens)
Respiratory (inorganics) kgeq PM2.5 into air Human health
Ionizing radiations Bqeq carbon-14 into air Human health
Ozone layer depletion Human health
Photochemical oxidation
[= Respiratory (organics) for human health]
393 [a] Aquatic ecotoxicity
393 [a] Terrestrial ecotoxicity
5 [b] Terrestrial acidification/nutrification
10 [c] Aquatic acidification
10 [c] AqUatic eutrophication
15 [b] Land occupation
38 [b] Global warming
9 [d] Non-renewable energy
20 [b]
Mineral extraction
kgeq CFC-11 into air
Kgeq ethylene into air
Human health
Ecosystem quality
kgeq triethylene glycol into water Ecosystem quality
kgeq triethylene glycol into water Ecosystem quality
DALY
MJ Total primary non-renewable
or kg~ crude oil (860 kg/m 3)
PDF * m 2 * yr
kgeq SO2 into air Ecosystem quality
kgeq SO2 into air Ecosystem quality
Under
development
kgeq PO43- into water Ecosystem quality
Under
de velopment
m2eq organic arable land.year Ecosystem quality PDF * m 2. yr
kgeq CO2 into air Climate change (kgeq CO2 into air)
(life support system)
Resources MJ
MJ additional energy
or kgeq iron (in ore)
Resources
Int J LCA 10 (6) 2003
325
IMPACT 2002+ A New Life Cycle Impact Assessment Methodology
1.1 Human toxicity (carcinogens and non-carcinogens)
Characterization factors for chronic toxicological effects on
human health, termed Human Toxicity Potentials (HTP) at
midpoint- and Human Damage Factors (HDF) at damage level,
provide estimates of the cumulative toxicological risk and
potential impacts associated with a specified mass (kg) of a
chemical emitted into the environment. These are determined
with the tool IMPACT 2002 (Impact Assessment of Chemical
Toxics) 1, which models risks and potential impacts per emis-
sion for several thousand chemicals (Pennington et al. 2003a,
2003b). Generic factors are calculated at a continental level
for Western Europe, whereas spatial differentiation for 50
watersheds and air cells in Europe is also enabled.
( I
Emissi~ in compartment m
] ]
89
Fate
J I Fraction
transferred ton
I Chf,?~ca[
factor}
c n Tinn I
L ,'"u~
L
j exposure
ncentration
i e~eS~ake I Dose taken in I
Potency
l - response l ' ....... ~, ...... ' ~' L
fEffoeC~r ,~ LI Potentiallaffected I IRisk~ J (D
....
response)
/ rl
fraction of species
[
I persons I
]
( ~ ] ("'~:~at~ "~ Damage on
f
Severity
~" I damage on I human health J
I
ecosystems
Intake
fraction
iF
Effect
factor
Fig. 2: General scheme of
the Impact pathway for human toxicity and
ecotoxicity
(Jolliet et al. 2003b)
Fig. 2 summarizes the different types of relevant informa-
tion regarding human toxicity: fate, which is composed of
transport in the environment, exposure, and the resulting
intake. This is then combined with an effect factor charac-
terizing the potential risks linked to the toxic intakes. Sever-
ity finally characterizes the relative magnitude of the dam-
age due to certain illnesses. The Human Damage Factor of
substance i (HDFi, in DALY (Disability Adjusted Life Years)
per
kgemitted)
is calculated as follows:
HDF i = iF l " EF i = iF I" 13i " Di (1)
The intake fraction (iF) is the fraction of mass of a chemical
released into the environment that is ultimately taken in by
the human population as a result of food contamination,
inhalation, or dermal exposure (Bennet et el. 2002a, 2002b),
in
kgintake
per
kgemitte d.
The effect factor (EF) is the product
of the dose-response slope factor (6, in risk of incidence per
kgintake) and of the severity (D, in DALY per incidence).
The intake fraction, therefore, accounts for a chemical's fate
in regards to multimedia and spatial transport as well as
human exposure associated with food production, water
supply, and inhalation. The complete fate and exposure as-
sessment enables the estimation of a chemical's mass (or
1 'IMPACT 2002' denotes the model
which focuses
on human toxicity and
ecotoxicity, while the complete LCIA methodology, with all impact cat-
egories, is termed 'IMPACT 2002+'
concentration) in the environmental media at a regional or
at a global scale using the same basic model. Per default,
characterization factors are calculated for emissions into a
Western European system nested in a global box. Special
attention was paid to air modeling and a new simple and
accurate method has been developed to account for the in-
termittent character of rainfall in a steady-state model. The
IMPACT 2002 model accounts for multiple exposure path-
ways that link a chemical's concentration in the atmosphere,
soil, surface water, or in vegetation to human uptake through
inhalation and ingestion. Ingestion pathways include drink-
ing water consumption, incidental soil ingestion, and intake
of contaminants from agricultural products (fruits, vegeta-
bles, grains, etc.), as well as from animal products, such as
beef-, pork-, and poultry-meat, eggs, fish, and milk. Com-
pared to conventional approaches, the transfer of contami-
nants into the human food is no more based on consump-
tion surveys, but accounts for agricultural and livestock
production levels that are eventually eaten by humans, in-
dependently from their living location. Latest developments
also include the calculation of pesticide residues in food due
to direct applications. The intake fraction concept also fa-
cilitates the comparison between indoor and outdoor emis-
sions, with the intake fraction for indoor air emissions be-
ing a direct function of the ventilation rate per inhabitant.
For the effect factor, IMPACT 2002 uses a new approach to
calculate the health effect metric for non-cancer toxicologi-
cal impacts. The selected measure is the EDa0 , the effect dose
inducing a 10 % response over background. It is derived from
the health-risk-assessment concept of benchmark dose to
estimate a default linear low-dose extrapolation, as detailed
by Crettaz et al. (2002) for cancer effects and by Pennington
et el. (2002) for non-cancer effects. One gets:
0.1 1.
~human
ED1---- ~ BW-LT h 9 N365
(2)
with:
J~human
EDt0
BW
LTh
Naes
Human health effect factor [risk of an incidence per kg cumu-
lative intake]
Benchmark dose resulting in 10% effect over background [mg/
kg/day]
Average body weight in considered population [kg/pers]
Average lifetime of humans in considered population in years
[yr]
Number of days per year [days/yr]
Preliminary ~ slope factors were calculated from bioassays
on animal data using best-estimate extrapolation factors from
TDh0, NO(A)EL, and LO(A)EL data 2. The DALY (Disability
Adjusted Life Years (Murray and Lopez 1996)) characterizes
severity, accounting for both mortality (Years of Life Lost (YLL)
due to premature death) and morbidity. Default DALY values
of 6.7 and 0.67 [years/incidence] are adopted for most carci-
nogenic and non-carcinogenic effects, respectively.
2 Toxic Dose 50%, No and Low Observed (Adverse) Effect Levels
326
Int J LCA 10 (6) 2003
A New Life Cycle Impact Assessment Methodology IMPACT 2002+
There is no real midpoint for human toxicity as intermedi-
ary parameters for fate and exposure like intake fraction
(see above) cannot be interpreted on their own. A real mid-
point could be the number of cases for the same illness.
However, as one or several substances cause a large number
of illnesses, risk of illnesses cannot be added up without
considering implicitly (equal severity), or preferably explic-
itly, their respective severity. The characterization factors at
midpoint are therefore simply obtained by dividing the Hu-
man Damage Factor of the considered substance by that of
the reference substance, which is chloroethylene (declared
human carcinogen with well defined fate data and a main
impact pathway by air inhalation):
HTPi=HDFi/HDFchl .... thyl ....
in kgeq chloroethylene into air per kg i
(3)
Expressing scores in kg-equivalent of a reference substance
facilitates communication and stresses that these characteri-
zation factors are mostly interesting for relative compari-
sons rather than for their absolute values.
1.2 Aquatic and terrestrial ecotoxicity
In many respects, impacts on aquatic ecosystems are treated
similar to human toxicity including both fate and effect, with
however some noticeable differences. First, one is generally
interested in effect at species level rather than on individu-
als. Second, the same fate model is applied as for human
toxicity, but the interface between fate and effect is at the
level of concentration (see Fig. 2). Fate enables to relate
emissions to the change in concentration in the pure aque-
ous phase of freshwater. Exposure is generally implicitly
taken into account in the effect factor that characterizes the
risks at species level, eventually leading to a Potentially Af-
fected Fraction (PAF) or Potentially Disappeared Fraction
(PDF) of species and to a preliminary indicator of damages
on ecosystems.
For aquatic freshwater ecosystems, the time- and space-in-
tegrated Potentially Affected Fraction of species per unit of
emission (APAF, in PAF-m3-year/kg) is therefore estimated
on the basis of a fate factor ( F. 0, in years) and an effect
factor (13, in PAF m3/kg) as follows:
APAI~i
= Vi mw" 0w" ~i,
in PAF.m3.year/kg
(4)
The
fate factor
itself is obtained by the multiplication of
two parameters that are calculated using the IMPACT 2002
model (Pennington et al. 2003a): F mw is the dimensionless
i
fraction of the emission of substance i in compartment m
transferred to freshwater. 0 w , in years, is the equivalent resi-
dence time of substance i in water, equal to the inverse of
the overall decay rate constant in water (k). It also corre~
sponds to the time- and space-integrated increase in concen-
tration in the aquatic freshwater per mass input of chemical
M released into the aquatic environment:
O=I/k=AC-V.At/M
(5)
tiC
(in kg/m 3) is the concentration increase in the volume of
water V (in m3), due to an emission flow of MAt (in kg/
year). This space integration differs from traditional regula-
tory-orientated risk assessment based on pure PEC/PNEC
approaches (predicted concentration divided by predicted
no effect concentration). Introducing the volume of water
that is polluted to a certain level accounts for the fact that
polluting all the lakes in Europe versus a small lake is not
equivalent in term of impacts. The aquatic ecotoxicological
characterization factors do not include an exposure compo-
nent to account for bio-magnification (additional exposure
due to contaminants in food, including suspended particulate
matter). Only bio-concentration is considered (direct trans-
fer of chemicals from the exposure medium to the species,
as observed in aquatic laboratory toxicity tests).
The risk-based
effect factor
(~3 i) is the change in the Poten-
tially Affected Fraction of species that experiences an increase
in stress for a change in contaminant concentration. As de-
scribed in the A_MI method for aquatic ecosystems (Payet et
al. 2003), the effect factor assesses the mean impact on spe-
cies, using the HC50, the mean hazardous concentration af-
fecting 50 % of the species present in the ecosystem:
13i
=
0.5/HcSOw
(in PAF m3/kg)
(6)
This HC50, in kg/m 3, is itself calculated as the geometric
mean of available EC50s 3 on individual species. For com-
parative assessments it is better suited than regulatory PNEC
approaches based on most sensitive species, since the latter
are too sensitive to the species eventually tested.
At midpoint level, the freshwater Aquatic Ecotoxicity Po-
tential (AEP i in kgeq triethylene glycol into water per kgi) is
derived by normalization to the reference substance, trie-
thylene glycol:
AEP i
=
APAFi
APAFtriethylene
glycol
(7)
Terrestrial ecotoxicity potentials are calculated in a similar
way. As data availability is limited, terrestrial HC50s are
mostly extrapolated from aquatic HC50 w with the method
proposed by Hauschild and Wenzel (1998), as a function of
the adsorption coefficient of the considered substance i (Kdi ,
in m3/kg), the soil density (fls, in kg/m3), and the dimen-
sionless volumetric water content of soil (fw):
HC50~ = HC50 w (KdiP s + fw)
(8)
1.3
Other midpoint category effects
The characterization factors for the midpoint categories res-
piratory effects, photochemical oxidation, ionizing radiation,
ozone layer depletion, terrestrial acidification/nutrification,
land use occupation, and mineral extraction are obtained from
3 Effect concentration, where 50% of the population of a species
are af-
fected
Int J LCA 10 (6) 2003 327
IMPACT 2002+ A New Life Cycle Impact Assessment Methodology
Eco-indicator 99 (Goedkoop and Spriensma 2000), adopting
the default egalitarian scenario and by normalization to a ref-
erence substance. For climate change, the latest IPCC Global
Warming Potentials (IPCC 2001) have been used with a 500
years time horizon to account for long term effects.
The characterization factors for aquatic acidification and
aquatic eutrophication are adapted from Hauschild and
Wenzel (1998), which also correspond to Guin& et al.
(2002). Aquatic eutrophication is divided into two classes,
respectively valid for P-limited and N-limited watersheds.
The values for P-limited watersheds are applied by default
as recent evidence shows that ultimately phosphorus is the
relevant compound in most cases. This can be explained by
the fact that cyano-bacteria in lakes and rivers are fixing the
atmospheric N when nitrates are limiting in the aquatic
media. Therefore, in the long term, increases in nitrate con-
centration will not influence the ecosystem's development,
whilst an increase in phosphate will always lead to an in-
creasing impact, except in particular areas as for example
estuarial ecosystems (Barroin 2003).
Characterization factors for non-renewable energy consump-
tion, in terms of the total primary energy extracted, are calcu-
lated with the upper heating value (Frischknecht et al. 2003).
2 Damage Categories
Damage characterization factors of any substance can be ob-
tained by multiplying the midpoint characterization potentials
with the damage characterization factors of the reference sub-
stances (Table 2). The present Section shortly details how these
damage characterization factors were determined.
Human Health. Human toxicity (carcinogenic and non-car-
cinogenic effects), respiratory effects (inorganics and organ-
ics), ionizing radiation, and ozone layer depletion all con-
tribute to human health damages. As for human toxicity
(see Eq. 3), all of these midpoint characterization factors
can be expressed straightforwardly in
[DALY/kgemission] 4.
4 Or
[/Bqernission ]
for the 'ionizing radiation' midpoint category
Ecosystem Quality. The midpoint categories terrestrial acidi-
fication, terrestrial nutrification, and land occupation were
directly taken from Eco-indicator 99 and their impact can
directly be determined as a Potentially Disappeared Frac-
tion over a certain area and during a certain time per kg of
emitted substance, expressed in [PDF-m2-year/kgemittea]. For
ecotoxicity, the midpoint assessment is based on the time-
and volume-integrated Potentially Affected Fraction of spe-
cies, expressed in terms of [PAF.m3.year/kg] (see Eq. 4). Four
different approaches are possible to convert PAF into PDF:
(1) Eco-indicator 99 (Goedkoop and Spriensma 2000), which
uses a direct extrapolation factor of ten between the NOEC
based PAF and PDF; (2) The second family of models fo-
cuses on the recovery potential of species exposed to chemi-
cals; (3) Tools based on an assessment of the probability of
the extinction of species under toxicant stress, which are
currently used in conservation biology; (4) An assessment
of change in genetic diversity. After a comparison of all these
options (Payer 2002), the simple extrapolation factor has
been employed for Impact 2002+. The variability among
the different methods was less than one order of magnitude
and the alternative methods cannot be easily applied in LCIA
approaches, since they are not compatible with the assump-
tion of time and space integration of impacts. For Impact
2002+, though, the factor has to be changed compared to
Eco-indicator 99, as the HC50 is based on EC50 instead of
NOEC, yielding a factor 0.5. This represents the assump-
tion that one half of the species affected over their level of
chronic EC50 will disappear due to the toxic stress, result-
ing in the following Aquatic Ecotoxicity Damage Factor
(AEDF, in [PDF.mZ.year/kg~mitte~]) for a substance i:
AEDF i = 0.5- APAF i / hW,
where h w is the mean depth of freshwater, in [m]
(9)
Extrapolation methods are presently under development for
damage factors characterizing impacts on ecosystem qual-
ity caused by aquatic acidification and aquatic eutrophi-
cation. In addition, photochemical oxidation and ozone de-
pletion also potentially contribute to the overall impacts on
Table 2:
Characterization damage factors of the various reference substances
Carcinogens 1.45E-06 DALY/kg chtoroethylene
Non-carcinogens 1.45E-06 DALY/kg chloroethylene
Respiratory inorganics 7.00E-04 DALY/kg PM2.5
Ozone layer 1.05E-03 DALY/kg CFC-11
Radiation 2.10E-10 DALY/Bq carbon-14
Respiratory organics 2.13E-06 DALY/kg ethylene
Aquatic ecotoxicity 8.86E-05 PDF.m2.yr/kg.tdethylene glycol
Terrestrial ecotoxicity 8.86E-05 PDF.m2.yr/kg.tdethylene glycol
Terrestrial acidification/nutr. 1.04 PDF-m2-yr/kg SO2
Land occupation 1.09 PDF.m2.yr/m2.organic arable land.yr
Global Warming 1 kg CO2/kg CO2
Mineral extraction 5.10E-02 MJ/kg iron
Non-renewable energy 45.6 MJ/kg crude oil
328 Int J LCA 10 (6) 2003