The methodology for supporting the interpretation of the results has been focused on resource-related and toxicity-related impact categories. A testing phase has been performed through sensitivity analyses in order to highlight where there are critical aspects which may undermine results and interpretation thereof.
We selected a case study in which LCA has been applied to assess a waste electrical and electronic equipment (WEEE) management system (Biganzoli et al.
2015). The case study has been run using International Reference Life Cycle Data System (ILCD) methodology (EC-JRC
2010), currently recommended for the EU context and for the PEF (EC
2013) as life cycle impact assessment method. Currently, the models used in several impact categories are subject to a process of revision based on testing the strengths and weaknesses of available models. Therefore, we tested several options of interest for the evaluation of benefits associated to recycling, selected based on the extent they are debated within circular economy context (e.g. resource recovery and toxicity-related impacts).
2.1 Rationale for defining the sensitivity analysis for the impact assessment phase
Firstly, we tested impact characterization models and characterization factors (CFs) for toxicity-related impact categories: human toxicity (cancer and non-cancer) and ecotoxicity. Toxicity-related impact categories are already considered relevant for the recycling processes (Lim and Schoenung
2010). However, there is an evolving debate on the robustness of these impact categories and specifically for what concern impacts due to metals (see e.g. Pizzol et al.
2011). In fact, some specific features of the metals (e.g. metal essentiality), as well as elements affecting their fate modelling (e.g. different conditions affecting their bioavailability), are not fully captured by currently available models applied in LCA. Those aspects are also affecting the USEtox model (Rosenbaum et al.
2008) currently recommended within the ILCD methodology. Different sets of CFs have been proposed for metals (such as Gandhi et al.
2011 and Dong et al.
2014) to improve the way in which metals are characterized and are now included in USEtox version 2. Additionally, long-term emissions are already identified as a relevant aspect to be taken into account in life cycle inventories and impact assessment, especially concerning emission of metals in specific contexts (e.g. mining and landfilling) (Bakas et al.
2015). The results of toxicity impact categories may vary for several orders of magnitude (e.g. the case study presented by Huijbregts et al.
2003 in which the toxic impact of metals is reported to differ more than six orders of magnitude depending on the time horizon chosen). A guideline for their treatment in the context of EU recommendation is needed based on the recent literature (e.g. Pettersen and Hertwich
2008; Hauschild et al.
2008; Clavreul et al.
2012), given the prominence of these emissions in the characterized results. Hence, we performed:
Secondly, we tested impact assessment models for the resource depletion impact category. Models currently available differ both in the modelling approach, in the perspective adopted for assessing the resources (Dewulf et al.
2015) and, as consequence, in the metric and factors adopted for the characterization (Mancini et al.
2015a). Beyond the potential benefit associated to a mass-based approach to recycling (e.g. kilogramme of material recycled), there is indeed the need of understanding to which extent the recycled materials are contributing to the resource depletion impact category.
Traditionally, LCIA models focusing on resource depletion have been based on different assumptions (Steen
2006), namely: (1) assuming mining cost being a limiting factor, (2) assuming increasing demand of energy due to extraction from low-grade sources, (3) assuming that scarcity is a major threat and (4) assuming that environmental impacts from mining and processing of mineral resources are the main problem. The characterization models express the available amount of a resource at a given point in time (e.g. ore deposits or fossil fuel reserves) or the future consequences (e.g. higher economic and/or energetic costs) of the extraction of a certain amount of a resource in the present. Furthermore, in business and policy context, there is an increasing need of assessing the so-called critical raw materials (CRMs), having a strategic economic role for certain sectors (Mancini et al.
2015b) and for which a set of CFs has been recently proposed (Mancini et al.
2016).
Therefore, we performed the sensitivity analysis on the resource-related impacts adopting different sets of CFs based on existing models, including a model for CRM’s.
The different sets of CFs based on existing models that we tested are mainly based on the review of Klinglmair et al. (
2014). We tested abiotic depletion potential (ADP) (CML
2012), which is focusing on potential resource depletion based on the ratio between resource consumption and availability (either considering ultimate reserves in the earth crust, known base reserves or economically viable reserves); ILCD model (EC-JRC
2010), which is an extended version of CML
2012 reserve based (based on CML algorithm, few other resources have been added by (Sala et al.
2012)); EDIP97 (Hauschild and Wenzel
1998), which is comparing the resource with the deposits economically exploitable, without accounting for current level of consumption; EPS2000 (Steen
1999), which is assessing the cost (as the society’s willingness to pay) of substituting a resource by an alternative for future generations affected by the current level of depletion; ReCiPe 2008 (Goedkoop et al.
2009), which is assessing the marginal increase of extraction cost per kilogramme of extracted resource, differentiating it by deposit and assuming a discount rate over an indefinite time span; the anthropogenic stock extended abiotic depletion potential (AADP, Schneider et al.
2011), which is accounting for the potential of resource recycling, assuming urban mining as an additional source of resources. The recent update of the AADP (2015) (Schneider et al.
2015) was also considered. It introduces the concepts of “ultimately extractable reserves” represented by the amount available in the upper earth’s crust that is ultimately recoverable.
Regarding CRMs, the sets of CFs recently developed by Mancini et al. (
2016) have been tested. This is of particular relevance for a case study related to WEEE giving the number of CRMs, both metals and materials, which currently lacks CFs in LCIA and which may represent a further benefit of recycling to be accounted for. The new CFs benefitted from the results of a workshop exploring the role of LCA for the accounting and the assessment of CRMs (Mancini et al.
2013). The characterization model is based on the use and adaptation to LCA of the supply risk indicators developed by the European Commission (EC
2014). Among those proposed by Mancini et al. (
2016), three sets of CFs are tested, based on different assumptions: (1) a baseline option, the supply risk factors as such—(SR); (2) an exponential function which magnifies the differences between the CRMs—(SR)^6 and (3) the ratio between supply risk and production data (SR/world mine production in 2011), which reflects the size of the market, giving more importance to the materials used in small amounts in products and applications, like, e.g. specialty metals that are often perceived as critical.
2.2 Case study on WEEE
The LCA of the WEEE management system implemented in the Lombardy region (Italy) and described by Biganzoli et al. (
2015) was taken as a case study. In that study, the assessment was carried out to quantify the mass balance of the WEEE management system in the Lombardy region in the year 2011 and to calculate its environmental benefits and burdens. All the five categories of WEEE (i.e. according to the Italian legislation, heaters and refrigerators—R1, large household appliances—R2, TV and monitors—R3, small household appliances—R4 and lighting equipment—R5) were included in the analysis. The main goal was the assessment of the environmental performance of the treatment of each WEEE category, with the aim to understand if the benefits arising from the material and energy recovery were offsetting the burdens due to the processing of the waste itself.
The present study focuses on the management of the small household appliances (R4). This WEEE category was specifically selected as it is the most heterogeneous one and includes various components (e.g. printed wiring boards, base metals, different types of plastics, etc.), none of which is predominant.
The functional unit (FU) was defined as the treatment of 1 t of R4. The system boundaries included all the treatment processes, from the moment the waste enters the first treatment plant to when it leaves the system as an emission (solid, liquid or gaseous) or as secondary raw material, following the “zero burden assumption” (Ekvall et al.
2007). They thus included the preprocessing at the first treatment plant and the subsequent treatment of the separated components in recycling/disposal of final plants. The collection and the transport to the first treatment plant were excluded from the system boundaries, as their impacts resulted negligible in the previous study (Biganzoli et al.
2015).
Due to the complexity of the treated waste, it was not possible to quantify the kind of materials entering the system based on their composition; consequently, it was not possible to quantify how much of a specific element (e.g. gold, silver, etc.) was lost during the recycling processes. Literature studies, such as the UNEP reports (UNEP
2011 and
2013), can complement the missing information, by providing typical compositions and recovery rates, but they were not applied in the specific case analysed, which focused only on the amount of materials recovered. The available information to date are mainly related to the typologies of materials which are recycled. For the specific case study, Table
A1 of the Electronic supplementary material shows the amount of each component separated in the first treatment plant, together with its destination.
Cases of multi-functionality were resolved by expanding the system boundaries to include avoided productions due to material and energy recovery from waste (EC-JRC
2010; Finnveden et al.
2009). Avoided primary material and energy productions were based on average technologies. Table
A2 of the Electronic supplementary material shows the amount of avoided primary materials and energy associated with the treatment of 1 t of R4.
For each process, a new module was designed in the SimaPro software, including the energy and material consumption, the direct emissions as well as the substituted materials and energy, with the same approach adopted in previous studies (Rigamonti et al.
2010; Rigamonti et al.
2013a,
b). Primary data were used in the modelling of the foreground system, in particular, for the mass and energy balances of the first treatment plant and of the treatment of some of the separated components. These data derived from the Italian regional database O.R.So. (Osservatorio Rifiuti Sovraregionale) and from field visits to the main treatment plants located in the Lombardy region. For some components, primary data were not available and so data from the literature and from the ecoinvent database version 2.2 (Swiss Centre for Life Cycle Inventories
2010) were used.
The life cycle impact assessment step was performed as explained in Sect.
2.1.