Mineral resources in life cycle impact assessment: part II – recommendations on application-dependent use of existing methods and on future method development needs

Given the importance of mineral resources for society and the persistent debate about how mineral resource use should be addressed in life cycle assessment (LCA), a wide variety of impact assessment methods have been developed, each of which assesses different aspects of mineral resource use. Within the “global guidance for life cycle impact assessment (LCIA) indicators and methods” project of the Life Cycle Initiative hosted by UN Environment, a task force has been established to develop recommendations on the LCIA of mineral resource use. This “task force mineral resources” consisted of 62 members representing different countries and stakeholders (academia, the metals and mining industry, other industries, geological departments, consulting, and life cycle inventory (LCI) database providers). While some members followed the process passively, 23 contributed actively on a regular basis, out of which 22 (mainly from academia, among them many method developers) are co-authoring this paper.

As a first step, the task force described, discussed, and assessed 27 existing impact assessment methods. Based on this comprehensive review, which is published in part I of this paper series (Sonderegger et al. 2020), as well as earlier reviews and recommendations (e.g., EC-JRC 2011; Sonderegger et al. 2017), the task force provided initial conclusions regarding the use of existing methods and areas for future method development. In parallel, the task force articulated a precisely defined and agreed upon safeguard subject for mineral resources within the AoP natural resources, which defines what actually should be protected with respect to mineral resources in LCA. At the Pellston Workshop® held in Valencia in June 2018, eight task force members (5 from academia, 2 from consulting, 1 from the oil and gas industry) refined the definition of the safeguard subject and used the task force’s initial conclusions to derive recommendations on application-dependent use of existing methods and on future method development needs. This paper presents the final reflections and recommendations of the Pellston Workshop®.

The definition of the safeguard subject for mineral resources is described in section 2. In section 3, a set of impact assessment methods is recommended, addressing seven different questions that stakeholders may have with regard to mineral resource use. These methods are applied on an LCA case study of a European-manufactured electric vehicle in section 4. Section 5 provides recommendations for further improvement of the existing methods and new methodological developments.

Although the subject of mineral resource use has been addressed in life cycle impact assessment (LCIA) methods for more than 20 years (Guinée and Heijungs 1995) and more than 20 impact assessment methods have been developed during this time, the safeguard subject within the (AoP) “natural resources” is still debated (EC-JRC 2010; Mancini et al. 2013; Dewulf et al. 2015; Sonnemann et al. 2015; Sonderegger et al. 2017). Previous reflections on the safeguard subject range from (1) the asset (natural resources as such independent of their specific function), (2) the provisioning capacity (the ability of natural resources to provide functions for humans), and (3) global functions (additionally considering non-provisioning functions for humans and functions beyond human needs) to (4) the supply chain (from the provisioning capacity to products and services) and (5) human welfare (including perspectives 2–4) (Dewulf et al. 2015). Such different perspectives of “the problem” with respect to mineral resource use are reflected in the diverse set of impact assessment methods, which model different cause-effect chains (Sonderegger et al. 2020). To address this challenge, the task force used the outcome of a stakeholder survey and workshop conducted within the “Sustainable Management of Primary Raw Materials through a better approach in Life Cycle Sustainability Assessment” (SUPRIM) project (Schulze et al. 2020). The majority of survey respondents indicated that they consider the following:

After extensive discussions and several iterations within the task force and at the Pellston Workshop®, the safeguard subject was articulated as follows:

Within the area of protection “natural resources”, the safeguard subject for “mineral resources” is the potential to make use of the value that mineral resources can hold for humans in the technosphere. The damage is quantified as the reduction or loss of this potential caused by human activity.

This definition reflects the three components of the SUPRIM survey outcome. Further, it clarifies that mineral resources first “hold” a value which humans “make use of” in a second step. Accordingly, mineral resources were defined as follows:

Mineral resources are chemical elements (e.g., copper), minerals (e.g., gypsum), and aggregates (e.g., sand), as embedded in a natural or anthropogenic stock, that can hold value for humans to be made use of in the technosphere.

It should be noted that there are cases in which a mineral (e.g., chalcopyrite – CuFeS₂), the contained elements (Cu, Fe, and S – even if Fe ends up in the smelter slag for economic reasons), or both (the mineral and the metals) can be considered as “mineral resources” as all of them can hold a value for humans in the technosphere. The inclusion of both primary and secondary resources is not considered a contradiction to the AoP “natural resources” because all mineral resources – both primary and secondary – originate in nature. The degree to which existing methods are compatible with this definition of the safeguard subject is one aspect considered in the recommendation of methods.

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.

At the Pellston Workshop®, seven questions that stakeholders (policy, industry, consultants, NGOs, etc.) may have with regard to mineral resource use were formulated (Table 1) and grouped into two broad categories.

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 1 shows the two major categories of questions, the seven individual questions, the methods available to answer them, the recommended methods (bold), and the level of recommendation. In general, we recommend using the inside-out related questions within environmental LCA and the outside-in related questions within broader life cycle-based approaches, such as life cycle sustainability assessment (LCSA). However, it should be noted that this recommendation was strongly debated within the task force and at the Pellston Workshop®. A minority of the task force members and Pellston Workshop® participants argued that outside-in related questions and methods can be considered as part of environmental LCA.

The participants of the Pellston Workshop® did not intend to reach consensus on which of the inside-out related questions is most relevant to LCA. We suggest that the LCA practitioner considers the goal and scope of the LCA study to determine the relevance of the question to the assessment. There is also no recommendation on which of the outside-in related questions is more relevant to broader life cycle-based approaches. Thus, the level of recommendation denotes how well the recommended method can answer the respective question and should not be interpreted as an absolute judgment. To enable a comprehensive analysis of the various impacts of resource use on different aspects of the safeguard subject, a broad set of the recommended LCIA methods can be applied. If the practitioner simply selects the method with the highest recommendation level (ADP_{ultimate reserves}), he or she should be aware that the result is the answer to a specific question only and cannot be used as a proxy result for other questions (Table 1).

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).

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.

In order to illustrate the application of different methods, all 27 identified methods were tested on a case study of a European-manufactured electric vehicle (EV). The functional unit is defined as 1 km traveled. The life cycle inventory developed by Stolz et al. (2016), which comprises the extraction of 34 primary mineral resource elements, 37 primary mineral resource aggregates, and 4 energy carriers, has been used for this purpose.

Before presenting and discussing results, it should be noted that the development of a life cycle inventory is controversial with regard to mineral resources. The definition of elementary flows and the allocation of metals in multi-metal ores (e.g., copper-gold ore) can be accomplished in two different ways: either the metal content of the ores (e.g. Cu and Au) is considered the relevant elementary flows and allocated to the produced metals (e.g., copper and gold) based on physical mass balances and the remaining inputs and outputs (e.g.,, gangue and emissions) based on economic or other relationships (Fig. 1a), or the entire ore (e.g., containing Cu, Au, and gangue) is regarded as the elementary flow and allocated to the products using economic relationships (Fig. 1b).

While the task force members could not agree to a recommendation for one approach over the other, it should be noted that the choice of the allocation procedure can strongly influence the resulting LCI: In the first case, the LCI reflects the material composition of the product based on physical mass balances. In the second case, the LCI reflects the environmental interferences related to producing one metal, which often leads to the co-extraction of other metals. So the LCI can contain metals which are not physically present in the product.

Considering the relevance of multi-metal ore allocation for the LCI and the fact that it is handled differently in leading LCI databases, the two allocation approaches are further described in the supplementary material. In this case study, the first option (allocation according to physical mass balances and economic relationship) has been used to derive the LCI.

Figure 2 shows the LCIA contribution analysis for all the minerals included in the LCI of the EV life cycle determined by means of the seven recommended methods. Resources contributing more than 10% each to at least one impact category are presented individually, while the remaining resources are summarized in the category “other resources.” As the number of CFs differs between LCIA methods, and as the methods partly cover different elementary flows, care should be taken when interpreting the LCIA results to not confuse a null value with a missing CF. We refrained from reducing the LCI to the number of resources for which all methods provide CFs. While this would ease the interpretation, it would reduce the number of resources drastically and would not reflect the “real” result which LCA practitioner obtain when selecting one of the methods in an LCA software.

Before discussing the case study results in detail, it can be seen that the findings are highly method dependent and hardly any similarities regarding the contribution of resource uses to the total results can be observed. While this might appear confusing at first, such an outcome is logical because different methods describe different cause-effect chains (Sonderegger et al. 2020) and address different questions related to resource use (Table 1). So it is clear that, e.g., a method assessing the long-term depletion of geological stocks does not come to the same results as a method analyzing short-term supply risks.

Despite being used in a relatively small amount in the LCI, gold dominates the result for ADP_{ultimate reserves} due to its relatively low abundance in the earth’s crust. In contrast, the result of ADP_{economic reserves} is dominated by tantalum as the current economic reserves are under relatively high pressure due to current extraction rates. Even though the reserve and extraction data for tantalum can be considered uncertain, this indicates a potential mid-term technology-driven, physico-economic availability constraint. The different results from the two versions of the ADP method reveal the strong influence of the respective stock estimates (ultimate reserves vs. economic reserves) used in the characterization model (Eq. 1).

The use of copper makes a significant contribution for the inside-out related methods (13–31%) but makes a smaller contribution for the outside-in methods (0–5%). This result indicates that short-term availability constraints for the use of copper in electric vehicles are relatively small, though this current use may affect the opportunities of future users to use copper. Besides copper, nickel is another large contributor to the LCIA results when using the future efforts methods (SOP and LIME2) or the CEENE method.

Gravel causes a relatively high contribution to the LCIA result obtained by the CEENE method and a noticeable contribution to the result of the ESSENZ method, although the CFs for gravel are relatively small in both methods. The reason for this is the relatively large amount of gravel in the LCI which includes the construction of roads. The other LCIA methods do not provide CFs for gravel.

Cobalt and tantalum are the main contributors to the LCIA results when using the outside-in related methods ADP_{economic reserves} and ESSENZ – despite the different scopes and timeframes of these methods: mid-term physico-economic availability for ADP_{economic reserves} and short-term geopolitical and socioeconomic accessibility for ESSENZ. It should be noted that the GeoPolRisk method does not have CFs for these minerals and that the ESSENZ method comprises eleven different supply risk factors that are not intended to be aggregated into an “overall” CF (Bach et al. 2016); aggregation was performed in this case study for illustrative purposes only.

The differences in the LCIA results when using the GeoPolRisk and ESSENZ methods can be explained by the broader range of supply risk aspects considered in the ESSENZ method, the different coverage of inventory flows, the “canceling out” of mineral resource amounts in the GeoPolRisk method, and the spatial resolution of the CFs assessing the supply risk of European imports (GeoPolRisk) or global production (ESSENZ). Further discussion of the case study, results obtained by the supply risk methods is provided in a separate publication by Cimprich et al. (2019).

The impact assessment results for all 27 methods are shown in Figs. S4 and S5 in the supplementary material along with a more detailed comparison and discussion within the four method categories (depletion, future efforts, thermodynamic accounting, and supply risk) presented in Figs. S6–S13.

Based on the review of methods by Sonderegger et al. (2020) and on the findings of the case study presented above, we provide recommendations for future method developments. In the following subsections, we provide general recommendations applicable to all methods along with specific recommendations for each method category (depletion, future efforts, thermodynamic accounting, and supply risk). Finally, we provide recommendations to define the “dissipative resource use” and include it in the development of future characterization models.

The subject of mineral resource use has been a topic of persistent debate among LCIA method developers and a source of confusion for LCA practitioners given the variety of LCIA methods to choose from. Based on the review of 27 existing LCIA methods assessing the impacts of mineral resource use in LCA, accomplished in the first part of this paper series (Sonderegger et al. 2020), this paper provides recommendations for application-dependent use of existing methods in LCA studies and for future method development. As a starting point, the safeguard subject for mineral resources within the AoP natural resources has been defined. Accordingly, we formulated seven key questions regarding the consequences of mineral resource use (Table 1), which can be grouped into “inside-out” (i.e., current resource use changing the opportunities for future users to use resources) and “outside-in” related questions (i.e., potential resource availability issues for current resource use). Existing LCIA methods were assigned to these questions, and seven methods (ADP_{ultimate reserves}, SOP_URR, LIME2_endpoint, CEENE, ADP_{economic reserves}, ESSENZ, and GeoPolRisk) are recommended for use in current LCA studies at different levels of recommendation in relation to the questions they address. In general the levels of recommendation are relatively low (1 recommended, 4 interim recommended, 2 suggested) indicating the need for methodological enhancements across method categories. All 27 identified LCIA methods were tested on an LCA case study of a European-manufactured electric vehicle, and yielded divergent results due to their modeling of impact mechanisms that address different questions related to mineral resource use. Besides method-specific recommendations, we recommend that all methods increase the number of abiotic resources covered, regularly update their CFs, and consider the inclusion of secondary resources and anthropogenic stocks. Furthermore, the concept of dissipative resource use should be defined and integrated in future method developments.