Before presenting and discussing the results of the criteria-based comparison, the six LCIA methods are described by summarizing the main ideas of their characterization models.
3.1.1 Description of methods
The abiotic depletion potential (ADP) of a resource was originally defined as a ratio of a resource’s annual production to the square of the resource’s crustal content, normalized to the same ratio of the reference resource antimony leading to the common unit of antimony equivalents (Guinée and Heijungs
1995). The ADP of a product system is calculated by multiplying all resource extractions reported in the LCI by the respective CFs and summing the results. Thus, the method uses physical scarcity of resources in the earth crust as an indicator to measure the impact of resource use. After a data update in 2002 (van Oers et al.
2002), the method was revised in 2009 by separating it into two categories: ADP for fossil fuels and ADP for elements. The ADP for elements was updated in 2016 (CML-IA
2016) and methodologically enhanced in 2019 using recent crustal content data and cumulated production from 1970 to 2015 instead of single-year production rates which can change considerably (and thus alter the CFs) over time (van Oers et al.
2019). In the case study presented below, ADP is used in its 2016 version (CML-IA
2016), which is also implemented in the product environmental footprint (EF 3.0
2022).
In contrast to assessing the extraction of resources by means of ADP, the environmental dissipation potential (EDP) aims at assessing the long-term dissipative losses of resources, i.e., the emission of resources to the environment (van Oers et al.
2020). The characterization model, which is used to determine the CFs, is similar to the one of the original ADP method (Guinée and Heijungs
1995) as it comprises a ratio of resource extraction in a reference year to the squared crustal content of the resource. However, this ratio is not normalized to the ratio of antimony (in ADP of elements) but to the ratio of copper, and, thus, EDP is expressed in copper equivalents. The central assumption in the characterization model is that all resources extracted in the reference year will be dissipated in the very long-term perspective. In contrast to ADP, the CFs are not multiplied by resource extractions of the product system but by the emission of resources to the environment. It should be noted that the authors of EDP also provide conceptual characterization models to measure impacts of technosphere hibernation and occupation in use (van Oers et al.
2020), which are not considered in this work as applicable CFs are not available yet.
Building up on the idea of defining emissions of resources to the environment as dissipative losses, the abiotic resource project (ARP) developed a classification model to differentiate metal emissions to the environment into dissipative and non-dissipative flows (Owsianiak et al.
2022). The basic assumption behind this concept is that metal emissions to the environment are only dissipative if both of the following two conditions apply. First, they “originate from a source with a concentration higher than a reference (concentration in upper continental crust) reflecting what is accessible for humans within the considered time span” (Owsianiak et al.
2022). For example, copper leached out from a rain pipe can be dissipative, as the copper originates from a copper mine extracting the metal from geologic reserves (with a concentration higher than the average copper concentration in the upper continental crust). In contrast, copper emissions from coal-fired power plants are not dissipative flows, as the copper contained in coal is not a geologic reserve but an impurity (with a concentration lower than the average copper concentration in the upper continental crust). As a second criteria, a metal dissipation is only considered dissipative, if “the current annual rate of total anthropogenic emissions results in a steady state concentration in the receiving environment that is below a reference concentration” (Owsianiak et al.
2022). Hence, metal emissions (even if originating from geologic reserves) are not dissipative if the concentration in the receiving compartment is above the metal’s concentration in the upper continental crust. The ARP method is not a characterization model but a classification model, which can be combined with other emission-based characterization models. In this study, it is combined with the EDP method described above. As current LCI databases do not specify whether metal emissions originate from ores, fossil fuel impurities, or other sources, the first criteria to identify non-dissipative emissions could not be applied in the case study presented in this paper. However, for the method evaluation shown below, the method is included in its original version including both criteria.
To avoid the uncertainty related to resource emissions used as proxy for dissipative losses in current LCI databases (unclosed mass balances in some datasets, origin of emissions from resources or impurities, concentrations in receiving compartments), Charpentier Poncelet et al. (
2021) follow a different approach. The authors developed two LCIA methods which use the concept of dissipation in their characterization models, but the resulting CFs are applied to the resource extraction and not resource emission inventory flows. The first method, ADR, assesses impacts of resource extraction based on their average dissipation over time, considering global average dissipation rates (ADR) which have been determined for each metal based on dynamic material flow analysis data. ADR depends on the function of resource dissipation over time and is calculated as the inverse of the total service time, which can be understood as the area below the dissipation function measured in kg ⋅ years per kg extracted. The second LCIA method, LPST, denotes the lost potential service time within a certain timespan, which is defined as the difference between the optimum service time (no dissipation, rectangular area of kg ⋅ years per kg extracted in a dissipation over time diagram) and the actual service time (area below the dissipation function, kg ⋅ years per kg extracted) within this time span (Charpentier Poncelet et al.
2021). In this work, a time horizon of 100 years is used, and the indicator is termed LPST100. It should be noted that these methods are based on the global average dissipation rates of resources, and the dissipation of resources extracted in the product system under study can be different. In the opinion of the author, this potential mismatch between the LCI and the LCIA levels can lead to counterintuitive results. For example, if a product system (in theory) does not have any dissipative losses, it will still show impact due to its resource extraction and the average (not product specific) dissipation rates of these resources. Vice versa, if a product is made from 100% recycled content, no resource extraction will be reported in the LCI and, thus, the LCIA result will be zero regardless of the amount of resources that gets emitted from the product system into the environment. Such results are nor “wrong,” but the difference between dissipation rates reported in the LCI (specific) and applied in the LCIA methods (average) should be kept in mind when interpreting the results.
After having proposed a new inventory scheme to clearly list dissipative resource flows in the LCI (Beylot et al.
2021), Ardente et al. (
2022) propose a price-based impact assessment method. Assuming that market prices reflect “the multiple, complex and varied functions and values held by mineral resources” (Ardente et al.
2022), the authors use resource prices averaged over a 50-year timespan as CFs to assess the impact of dissipative resource losses. In this work, the CFs are applied to the emission of resources to the environment as reported in the GaBi database, as the dissipation-specific inventories (Beylot et al.
2021) are not available for the analyzed metals yet.
3.1.2 Criteria-based comparison
The complete evaluation of the six LCIA methods described above against the 22 criteria shown in Table
1 is presented in a spreadsheet in the supplementary material
S1. In the following, the main findings and differences between the methods are presented and discussed.
Concerning the classification scheme according to which a working group of the UNEP Life Cycle Initiative recommended methods for different questions (Berger et al.
2020), all methods except for the price-based method are considered to address the question: “How can I quantify the relative contribution to the depletion of mineral resources?”. This is not surprising, as dissipation directly contributes to resource depletion, and the UNEP working group recommended the development of dissipation-based methods for this question. In contrast, the price-based method addresses the question: “How can I quantify the relative (economic) externalities of mineral resource use?”.
Concerning the time scale, all methods address the long-term impacts of resource dissipation except for the price-based method. The latter assesses the short-term impacts reflected by market prices, which is consistent with the previously proposed LCI approach (Beylot et al.
2021). This also takes a short-term perspective and considers resource flows into waste disposal facilities or non-functional recycling as dissipative. In addition to the long-term perspective, the authors of the EDP method (van Oers et al.
2002) also propose (not yet operational) concepts for the short- and medium-term perspectives. Also the LPST can be calculated for different time horizons. In this context, it should be noted that the terms short-, medium-, and long-term are neither clearly nor consistently defined. Often short-term is considered as < 5–10 years, mid-term around 25 years, and long-term > 100 years (Arvidsson et al.
2020; Schulze et al.
2020) or even > 500 years (Dewulf et al.
2021).
The characterization models of EDP, ARP, ADR and LPST consider emissions of resources into the environment as a form of dissipation. Additionally, the conceptual methods of the EDP authors as well as the dynamic material flow models underlying the ADR and LPST characterization models define emissions into the technosphere (e.g., landfill or non-functional recycling) as dissipative. None of the methods considers occupation in use or hibernation in the technosphere (e.g., unused products such as old smartphones not taken to recycling yet) a relevant form of dissipation.
The CFs of classical resource LCIA methods, such as ADP, are applied to (multiplied by) the resource extraction flows of the LCI. In contrast, the CFs of most dissipation-based methods are applied to the emission of resources (EDP and ARP) or to flows of dissipative resource losses from specific LCIs (price-based method). Two exceptions to this are the ADR and LPST methods, whose CFs are applied to the resource extraction flows as their characterization models describe the average dissipation rates per kg extracted resource.
The characterization models (classification model for ARP) and underlying main assumptions are described above. With regard to normalization, the analysis revealed that all methods except for ADR and LPST provide applicable normalization factors (inverse of global per-capita impacts). For the latter methods, normalization factors have been calculated by using extraction data from the ADP method, multiplying the resource extractions by their corresponding ADR and LPST100 CFs, and dividing it by the world population to obtain per capita impacts.
None of the method publications discusses the option of weighting the impact assessment results of resource dissipation to compare or aggregate them to other impacts. To illustrate the applicability and effects of weighting, the LCIA results of the theoretical product are normalized and weighted using the weighting set of the product environmental footprint (PEF) (EF 3.0
2022).
All LCIA methods are published in peer-reviewed scientific journals. The data quality of the characterization models is considered high for the extraction (ADP and EDP) and price-based models as global production, crustal content, and market prices of resources are well reported. The data quality of the other characterization models is considered medium due to uncertainties associated with the use of modelled data (fate models in ARP and dynamic material flow models in ADR and LPST). However, in both cases, it should be noted that central assumptions, such as crustal content being a proxy for ultimately extractable reserves, complete dissipation of current extraction in the long-term future, or market prices being a proxy for the value of resources are not less relevant than numeric data uncertainty.
In addition to data quality of the characterization models, the quality of the LCI data to which the CFs are connected is also important. In general, it can be said that the quality of resource extraction data needed for ADP, ADR, and LPST is higher than the quality of resource emission data, which is used as a proxy for dissipative flows into the environment. This is because resource extractions are comparatively easy to measure and well reported. In contrast, emissions of resources do not necessarily represent dissipative losses (as addressed by ARP), and the comparison of resource inputs (extraction) and outputs (emissions and product) often shows inconsistent mass balances.
Concerning the practical implementation of the LCIA methods, it can be said that all methods provide applicable CFs which are publicly available, with only ADR and LPST being not published open access. The number of CFs ranges from 18 for ADR and LPST to 108 for EDP. While ARP, ADR, and LPST cover mainly metals, the other methods also provide CFs for minerals. At this point (November 2023), the methods are not available in the LCA software with the exception of ADR and LPST being implemented in SimaPro. However, an older version of ADP (van Oers et al.
2002) is implemented in all of the above-mentioned LCA softwares. The effort for manually implementing the LCIA methods in the GaBi software is considered low for ADP, ADR, and LPST, as only new environmental quantities (impact categories) need to be created and CFs for the resource extraction flows need to be entered. The implementation of the other impact categories requires more effort, as the names of the elementary flows in the method publications (e.g., copper) need to be matched to a list of emissions in the software (e.g., copper [heavy metals to air], copper [heavy metals to freshwater], etc.).