The first part of this paper series (Sonderegger et al.
2020) identified 27 existing methods to assess
impacts of mineral resource use. The wide variety of methods causes confusion among
LCA practitioners, and often the “wrong” method is used to answer the
“right” question. For instance, methods assessing the long-term
depletion of geological resource stocks (e.g., the abiotic depletion potential) are
often used by LCA practitioners who are actually interested in the short-term supply
risk of raw materials (Fraunhofer
2018). This paper builds on the description and categorization of
methods provided in Sonderegger et al. (
2020) by providing further guidance on the use of these
methods.
The first category of questions focuses on how the use of mineral
resources in a product system can affect the opportunities of future users to use
resources (termed the “inside-out” perspective), whereas the second
category focuses on how environmental and socioeconomic conditions can affect the
accessibility of mineral resources for a product system (termed the
“outside-in” perspective). For the first category, five individual
questions are related to physical depletion, resource quality, resource quality
change and its consequences, (economic) externalities due to overexploitation of
resources, and thermodynamics. For the second category, two questions were
identified, concerning the mid- and short-term supply of mineral resources.
Subsequently, the 27 methods were assigned to the question(s) they
address, and their capability to answer them was assessed based on (a) the modeling
approach, (b) the underlying data used, (c) the coverage of characterization factors
(CFs) as analyzed in the method review (Sonderegger et al.
2020), and (d) the degree to which existing
methods are compatible with this definition of the safeguard subject. Finally, the
most appropriate method(s) for the specific questions were recommended with a level
of recommendation ranging from “suggested,” “interim
recommended,” “recommended” to “strongly
recommended” (Frischknecht et al.
2016). An interpretation of these recommendation levels and more
detailed criteria can be found in the
supplementary
material. Limitations of recommended methods have been made
transparent to justify the level of recommendation and to propose methodological
improvements. Also methods published after the Pellston Workshop® in June
2018 (e.g., Bulle et al.
2019;
Vogtländer et al.
2019) could
not be considered for recommendation but have been included in the discussion if the
methodological concepts have been available to the task force (e.g., Huppertz et al.
2019). Since most method developers
contributed actively to this task force and partly participated in the Pellston
Workshop®, it is unavoidable that methods get recommended whose developers
were involved in the recommendation process. Further, recommendations were derived
based on transparent criteria and in a consensus finding process which involved all
participants of the Pellston Workshop®. The following subsection was written
by the members of the Pellston Workshop®, who are co-authoring this paper
together with other active members of the task force. To avoid different
understandings of the recommendations and rationales, the text below is only
slightly modified from the corresponding section in the Pellston Report (chapter 5.4
in (Life Cycle Initiative
2019)).
Table
2 provides more
information about the geographical resolution, the timeframe of impacts, the users
affected, and the number of CFs as related to the recommended methods. The CFs of
the recommended methods can be accessed via links to the method developers’
websites and publications provided in the
supplementary
material. As it can be seen, most methods focus on metals, and
only SOP
URR and CEENE provide a relevant number of CFs for
minerals and aggregates. A more comprehensive assessment of the recommended methods,
along with the remainder of the 27 methods reviewed, can be found in the
Supplementary Material to (Sonderegger
et al.
2020).
Table 2
Description of recommended methods in terms of geographical
resolution, timeframe, concerned users, and number of
characterization factors available
Geographical resolution/perspective | Global | Global | Global | Global | Global | Global | Country |
Timeframe of impacts | More than decades to hundreds of years | More than decades to hundreds of years | More than decades to hundreds of years | Current change | A few decades | Current accessibility | Current accessibility |
Users affected | Future users | Current users | Future users | Current users | Next few generations | Current users | Current users |
Number of CF for mineral resources (metals and
metalloids/non-metal elements/minerals and
aggregates) | 49 (44/5/0) | 75 (45/4/26) | 19 (19/0/0) | 65 (23/2/40) | 42 (39/3/0) | 49 (41/4/4) | 32 (21/4/7) |
Number of CF for energy carriers/other resources
(water, land use, biotic resources, intermediates,
etc.) | 4/0 | 0/0 | 4/0 | 4/12 | 4/0 | 4/7 | 1/13 |
In the following, the recommended methods are described and a rationale
for their recommendation is provided along with a discussion on limitations, which
explain the level of recommendation.
3.1 Question: How can I quantify the relative contribution of a product system
to the depletion of mineral resources?
Recommended method: ADP
ultimate reserves
(method from Guinée and Heijungs (
1995), CFs latest version at CML (
2016)
Level of recommendation: recommended
The ADP model relates annual extraction rates to a stock estimate.
As shown in Eq.
1, depletion is assessed
using the ratio of an extraction rate (
E) to
a stock estimate (
R), and this ratio is
multiplied by a factor of 1/R to account for differences in stock size (see
Guinée and Heijungs (
1995)
for a detailed explanation of modeling choices). Furthermore, the ADP is
normalized to antimony as a reference substance. Equation
1 shows the calculation of the ADP (which serves
as the CF for a resource
i relative to the
reference substance antimony (
ref)). For
ADP
ultimate reserves, the stock estimate
R is the ultimate reserves (also known as the
“crustal content”).
$$ {ADP}_i={CF}_i=\frac{E_i/{R}_i}{E_{ref}/{R}_{ref}}\ast \frac{1/{R}_i}{1/{R}_{ref}}=\frac{E_i/{R}_i^2}{E_{ref}/{R}_{ref}^2} $$
(1)
According to Guinée and Heijungs (
1995), the ultimately
extractable reserve is the only relevant stock
estimate with regard to depletion of natural stocks. However, given that it
depends on future technological developments, it can never be known. Therefore,
a proxy is needed, and “ultimate reserves” is considered a better
proxy than fluctuating stock estimates like “resources” or
“economic reserves” as defined by the US Geological Survey (USGS),
that provide a midterm perspective (a few decades). Alternatively, a simpler
model without extraction rates, such as those used in the EDIP and
LIME2
midpoint methods, could be used. However, these
methods do not provide CFs based on crustal content but economic reserves
(although they could be easily calculated). While we recommend using
ADP
ultimate reserves as the baseline method, we,
along with the method developers (van Oers et al.
2002), recommend using alternative depletion methods –
in addition to ADP
ultimate reserves – for
sensitivity analysis.
Regarding depletion of natural stocks, the ADP model is valid and
has also been recommended by other initiatives (EC-JRC
2011). However, the need to use a proxy for
the ultimately extractable reserves is a limitation. With regard to depletion of
total stocks (i.e., natural stocks in the earth’s crust and anthropogenic
stocks in the technosphere), further limitations should be acknowledged. The
method does not distinguish between the part of the resource extraction that is
occupied for current use (but can be available for other uses in the future) and
the part that is “dissipated” into a technically and/or
economically unrecoverable form (the concept of dissipation is further discussed
in section 5.3). By considering the ultimate reserves as a resource stock,
anthropogenic stocks are not explicitly taken into account. However, it can also
be argued that anthropogenic stocks are implicitly included, as there is no
deduction of already extracted resources from ultimate reserves. Further,
anthropogenic stocks can be occupied rendering them inaccessible during the life
time of the stocks. The AADP and AADP (update) models consider geological and
(estimated) anthropogenic stocks explicitly. However, besides uncertainties
involved in the determination of anthropogenic stocks, the use of extraction
rates in the numerator of the characterization model is considered an
inconsistency as extraction shifts mineral resources from geological to
anthropogenic stocks. Until the concept of dissipation is operationalized, the
ADP
ultimate reserves method could be interpreted as
the best available proxy for depletion of the total resource stock and therefore
is a recommended method. An update of the ADP method was published during the
processing of this paper (van Oers et al. (
2019)) but couldn't be considered by the task force.
A minority of the Pellston Workshop® participants and task
force members disagreed with the level of recommendation of
ADPultimate reserve. Since the method considers only
the extraction and stocks of mineral resources and neglects anthropogenic stocks
and dissipation rates, the minority argued that the recommendation level should
be “interim recommended” pending future methodological
development.
3.1.1 Question: How can I quantify the relative contribution of a product
system to changing mineral resource quality?
Recommended method: none
This question refers to modeling approaches that evaluate a
change in resource quality without considering any consequences of it. The
only suitable method identified – ore grade decline (Vieira et al.
2012) – is
operational only for copper and therefore is not recommended. Moreover,
methods answering the follow-up question (“How can I quantify the
consequences of the contribution of a product system to changing resource
quality?”) can be interpreted as proxy for the question posed here,
depending on modeling choices. For instance, the ore requirement indicator
(Swart and Dewulf
2013) and the
surplus ore potential (Vieira et al.
2017) methods quantify the amount of surplus ore required
to mine the same amount of metal – which can be considered a
consequence of a quality change.
3.1.2 Question: How can I quantify the relative consequences of the
contribution of a product system to changing mineral resource
quality?
Recommended method: SOP
URR (Ultimate Recoverable Resource) (Vieira
2018)
Level of recommendation: interim recommended
The surplus ore potential (SOP) (Vieira et al.
2017) method measures the average
additional ore required to produce the resource in the future, based on
resource grade-tonnage distributions and the assumption that higher grade
ores are preferentially extracted.
A log-logistic relationship between ore grades and cumulative
extraction is developed for each resource “x” based upon
fitting regression factors (
αx and
βx) to the observed (A
x;
kg
x) grade-tonnage distribution of deposits.
Prior to this procedure, an economic allocation of ore tonnage is performed
to account for potential co-production. An average CF is developed by
integrating along the product of resource extraction
(RE
x) and the inverse of the grade log-logistic
relationship (OM
x, the amount of ore mined per amount
of resource x) from cumulative resource extraction
(CRE
x) to the maximum resource extraction (MREx)
then dividing by total remaining extraction (R
x).
Therefore, the CF representing the average surplus ore potential of each
resource (SOP
x; kg
ore per
kg
x) can be expressed as:
$$ {SOP}_x=\frac{\int_{CRE_x, total}^{MRE_x}{OM}_x\left({RE}_x\right)\ d{RE}_x}{R_x} $$
(2)
$$ {OM}_x=\frac{1}{G_x}=\frac{1}{\mathit{\exp}\left({\alpha}_x\right){\left(\frac{A_{x, sample}-{CRE}_{x, sample}}{CRE_{x, sample}}\right)}^{\beta_x}} $$
(3)
As the total remaining extraction is unknown, it is
approximated by demonstrated economic reserves and ultimate recoverable
resources (URR, approximated as 0.01% of the resource within 3 km) to
provide two sets of characterization factors
(SOP
reserves and SOP
URR).
In the recommended version of the method (Vieira
2018), the set of CFs for 18 resources
based on the approach described above (Vieira et al.
2017) was extended to 75 resources
through the extrapolation of SOP values using a correlation between SOP and
resource prices.
Other methods were not recommended for the following reasons:
ReCiPe2016 endpoint is based on “surplus cost potential” (SCP)
and uses a mid-to-endpoint conversion factor based on copper, which may not
be applicable to all resources. The original SCP method (Vieira et al.
2016) and the ore
requirement indicator (ORI) method (Swart and Dewulf
2013) were not recommended as they are
based on regression data that were determined using mined ore tonnage and
mining cost data over a period characterized by very high growth in mineral
demand and mineral price increases that significantly distorted short-term
mineral markets. Hence, the CFs developed in those methods are highly
sensitive to the underlying time period, whereas
SOP
URR is based on grade-tonnage distributions
that are considered very robust for each deposit type. ReCiPe2008 (Goedkoop
et al.
2013) is based on data
for existing mines only and does not include data for undeveloped mineral
deposits known to be available. Eco-indicator 99 (Goedkoop and Spriensma
2001), Impact2002+ (Jolliet
et al.
2003), Stepwise2006
(Weidema et al.
2008; Weidema
2009), EPS 2000/2015 (Steen
1999,
2016), and thermodynamic rarity methods
(Valero and Valero
2014) are
not recommended because they do not model an ore grade decline (and its
consequences) based on extraction data but only consider an assumed change
in ore grades at a future point in time (see section 6.2 in Sonderegger et
al. (
2019)).
A key limitation of the SOP
URR method is
that it assumes mining from highest to lowest grade and does not explicitly
account for competing factors such as technological and economic
considerations (Sonderegger et al.
2020). However, the marginal gradient of the
grade-tonnage curves should provide a good relative assessment between
mineral resources, which is useful for LCA purposes. The extrapolation of
observed grade-tonnage data is also an assumption for the long-run future
and therefore impossible to prove or falsify. Therefore, the
SOP
URR method (Vieira
2018) is only “interim
recommended.” Considering the limitations discussed above, one task
force member representing the exploration and mining industry does not
support this recommendation and published a split view in parallel to this
work (Ericsson et al.
2019) in
which the validity of the impact pathway addressed by methods in this
category is challenged.
3.1.3 Question: How can I quantify the relative (economic) externalities of
mineral resource use?
Recommended method: LIME2
endpoint
(Itsubo and Inaba
2014)
Level of recommendation: interim recommended
The LIME2
endpoint method is based on El
Serafy’s user cost (El Serafy
1989). The user cost assesses the share of the economic
value of extracted resources that needs to be reinvested to maintain the
benefit obtained from the extraction of resources (Itsubo and Inaba
2014). The indicator of
LIME2
endpoint expresses the economic externality
of resource use in units of monetary value and is calculated as
follows:
$$ {\mathrm{CF}}_{\mathrm{LIME}2\mathrm{endpoint}}=\mathrm{R}\left\{1/{\left(1+\mathrm{i}\right)}^{\mathrm{N}}\right\}/\mathrm{P} $$
(4)
where R is annual profit of the target element; i is the
interest rate; N is ratio of economic reserves to production (years to
depletion); P is current annual production amount of the target
element.
The LIME2 method is recommended given that it incorporates
uncertainty data and was the only peer-reviewed method available in this
category at the time of the Pellston Workshop®. A few months later,
the future welfare loss method was published (Huppertz et al.
2019), which describes a complementary
impact pathway to the one modeled in LIME2. While LIME2 assesses the
potential externality of lost future income due to a hypothetical lack of
investment of earnings from the sale of finite resources, the Future Welfare
Loss method assesses the potential externality of lost hypothetical rents
due to current overconsumption of the resource.
The main limitations of the recommended LIME2endpoint method
are the uncertainty of determining the relevant interest rate, different
opinions on the applicability of the El Serafy’s method (which
estimates pricing failure in the market as a whole society) to a specific
mineral, and the limited number of CFs (19 for mineral resources and 4 for
energy carriers). The LIME method has three versions (LIME/LIME2/LIME3).
LIME2 is the updated version of the original LIME method, with the addition
of uncertainty analysis. LIME3, which was not yet published at the time of
the Pellston Workshop®, is an extended version of LIME2 with
country-specific (LIME and LIME2 provide generic CFs without consideration
of country-level differences in production and reserves).
3.1.4 Question: How can I quantify the relative impacts of mineral resource
use based on thermodynamics?
Recommended method: CEENE (Dewulf et al.
2007)
Level of recommendation: interim recommended
The exergy of a resource is the maximum amount of useful work
that can be obtained from it when it is brought to equilibrium with the
environment (reference state). As mineral resources differ from the
reference state with respect to their chemical composition and their
concentration, in principle they can produce work. Although most mineral
resources are not extracted from nature with the aim to directly produce
work, they still contain exergy. For example, the copper in a copper deposit
is much more concentrated and occurs in another chemical form (e.g.,
CuFeS2) than the copper dissolved in seawater
(the reference state for copper). This distinction with respect to
commonness makes a resource to be valuable in exergy terms.
The cumulative exergy extraction from the natural environment
(CEENE) method (Dewulf et al.
2007) aggregates the exergy embedded in extracted
resources (e.g., copper), measured as the exergy difference between a
resource as found in nature and the defined reference state in the natural
environment. Using the definition of Szargut et al. (
1988), the reference state is
represented by a reference compound that is considered to be the most
probable product of the interaction of the element with other common
compounds in the natural environment and that typically shows high chemical
stability (e.g., SiO2 for Si) (De Meester et al.
2006). For metals, CEENE calculates the
exergy value of the mineral species (e.g., CuFeS
2)
containing the target metal, making it independent of the ore grade.
The Pellston Workshop® participants recommend the CEENE
method over other thermodynamic accounting methods because it was originally
operationalized to LCA by proposing a more accurate exergy accounting method
than the one used in the Cumulative Exergy Demand (CExD) method. For
instance, in CExD the exergy values of metals are calculated from the whole
metal ore that enters the technosphere, whereas CEENE only regards the
metal-containing minerals of the ore (with the argument that the tailings
from the beneficiation are often not chemically altered when deposited).
While thermodynamic rarity (TR) offers an alternative reference state
(Thanatia) and as opposed to the other approaches considers ore grade in the
evaluation of resources, it is not mature enough when compared to Szargut et
al.’s (
1988) approach
(used in CEENE).
Another method with a thermodynamics-based approach is the
solar energy demand (SED), which is based on the energy approach (with a few
differences in the calculation approach) (Rugani et al.
2011). It considers the equivalent solar
energy that nature requires to provide a resource, which includes more
energy than what can be used out of this resource. Therefore, the method is
less relevant than CEENE with regard to the safeguard subject of mineral
resources.
As the focus of this work is on mineral resources, and the
overall (inside-out) concern is “changing opportunities of future
users to use resources,” the CEENE method is “interim
recommended.” A higher level of recommendation is not given because,
although the CEENE method allows quantifying the value of a resource in
exergy terms, the approach, as currently applied to mineral resources, does
not fully reflect their societal value as it leaves aside non-thermodynamic
aspects.
3.1.5 Question: How can I quantify the relative potential availability issues
for a product system related to physico-economic scarcity of mineral
resources?
Recommended method: ADPeconomic reserves
Level of recommendation: suggested
The model for calculation of ADP
economic reserves is the same as in Eq.
1, but economic reserves are used as the stock estimate
R. The
(economic) reserves are the part of known resources that is determined to be
economically extractable at a given point in time. The extraction-to-stock
ratio used in the model can be interpreted as a scarcity measure, and
accordingly the CFs of ADP
economic reserves provide a
measure of the pressure on the availability of primary mineral
resources.
Given that the extraction rates are considered important for
this midterm perspective (a few decades), a model excluding extraction rates
– as used in the EDIP and LIME2midpoint
methods – is not recommended here.
The exclusion of anthropogenic stocks is considered a major
limitation because these stocks can strongly influence the “resource
availability for a product system” (Schneider et al.
2011). Unlike the ADP
ultimate reserves method, anthropogenic stocks are not implicitly
included in the natural stock estimate of the ADP
economic reserves method. Previous attempts to include anthropogenic
stocks in the characterization model (e.g., the AADP method, (Schneider et al.
2015)) still face the
challenge of considering how much of this stock would become available
within the time horizon considered by the CFs.
Furthermore, the use of the economic reserves estimate is
problematic because historically it has actually grown in absolute terms,
and the extraction-to-economic-reserve ratios have been relatively stable,
indicating no increase in resource scarcity. Furthermore, economic reserve
estimates are highly uncertain for by-products. Finally, the method has not
been explicitly developed to address outside-in questions, and consequently
the results need to be interpreted carefully. For these reasons, the
ADPeconomic reserves method is only
“suggested.”
3.1.6 Question: How can I quantify the relative potential accessibility
issues for a product system related to short-term geopolitical and
socioeconomic aspects?
Recommended methods: ESSENZ (Bach et al.
2016) and GeoPolRisk (Gemechu et al.
2015; Helbig et al.
2016; Cimprich et al.
2017)
Levels of recommendation: interim recommended and suggested,
respectively
The ESSENZ method (Bach et al.
2016), which enhanced the preceding ESP method (Schneider
et al.
2014), quantifies eleven
geopolitical and socioeconomic accessibility constraints (country
concentration of reserves and mine production, price variation,
co-production, political stability, demand growth, feasibility of
exploration projects, company concentration, primary material use, mining
capacity, and trade barriers). Indicators for these categories are
determined and divided by a target value above which accessibility
constraints are assumed to occur. This distance-to-target (DtT) ratio is
normalized by the global production of the respective resource to reflect
the assumption that the accessibility constraints described above can be
more severe for resources produced in relatively small amounts. Finally, the
normalized DtT factors are scaled (to a range between 0 and
1.73 × 10
13 in each
category) to balance the influence of the LCI and the CFs on the LCIA result
and to ensure a similar range of CFs among the supply risk
categories.
The GeoPolRisk method weights the political stability of
upstream raw material producing countries by their import shares to
downstream product manufacturing countries (Gemechu et al.
2015; Helbig et al.
2016; Cimprich et al.
2017). It incorporates the country
concentration of production as a mediating factor in supply disruption
probability arising from political instability of trade partner countries.
The logic is that highly concentrated production of raw materials limits the
ability of importing countries to restructure trade flows in the event of a
disturbance (such as political unrest) that may lead to supply disruption.
Domestic production is assumed to be “risk-free” from a
geopolitical perspective. The method also incorporates a
“product-level importance” factor that effectively
“cancels out” the magnitude of inventory flows. The term
“inventory flows” is used to encompass both elementary and
intermediate flows – as the total supply risk associated with a
product system is a function of its entire supply chain (for further
explanation see (Cimprich et al.
2019)).
Comparing the two methods, the GeoPolRisk method allows the
consideration of the specific import structure of a particular country,
while ESSENZ takes a global perspective. Further, ESSENZ considers a broader
set of potential geopolitical and socioeconomic constraints and provides
more CFs for mineral resources. Considering the respective strengths of the
two approaches, the ESSENZ method is interim recommended to assess the
supply risk of multinational companies having locations all over the world.
The GeoPolRisk method is suggested to assess country-specific supply risks
arising from political instability of trade partners from which mineral
resources are imported. Both methods are usually applied outside an LCA
software because the elementary flows reported in LCI datasets do not
necessarily reflect the intermediate flows or the material composition of
products.
The ESSENZ and GeoPolRisk methods rely on the key assumption
that supply risk is a function of supply disruption probability and
vulnerability. They share the limitation of focusing on the supply risk of
primary resources only and either do not consider the country-specific
import situation (as in the ESSENZ method) or are limited concerning the
accessibility constraints considered (as in the GeoPolRisk method).